Documente Academic
Documente Profesional
Documente Cultură
Reenues could be oer-estimated i schools are unable to raise ees regularly ,and there is eidence that parents may not pay on
time or ully, or underestimated i schools are able to raise unds through other means ,school material ees etc., that are not reported
or i schools ail to report non-teaching sta expenditures.
8
Schools with better inrastructure and in more central locations charge higher ees, but these dierences arise due to dierences
across illages and disappear once we include illage ixed-eects in the regression.
39
111T @HA1'= AH@"@ HG @ALHH=1DM
2.16.
2.1.
2.18.
1be co.t to .ociet, of educating a cbita iv gorervvevt .cboot. i. at tea.t aovbte tbe co.t of eavcativg a cbita iv a rirate
.cboot. 1be.e co.t .arivg. ari.e frov torer rage. for teacber. iv tbe rirate .ector.
1be .ociat co.t of .cbootivg vv.t ivctvae tbe totat co.t of rai.ivg vove, ava .evaivg vove,. Len i priate schools
are relatiely cheap, education in the public sector still seems relatiely ree`. loweer, while the parent may
not be directly paying ees in the public sector, society does hae to pay this cost indirectly as their taxes go
towards paying the costs o creating and running public schools. Moreoer, to the extent that the tax burden
does not all more on the rich, the poor,middle-class and salaried may be paying a high cost or such
proision. 1hereore, i the oerall costs or proiding education are lower in priate schools, a releant
question is whether taxpayers can get more out o their money taking their public dollars and paying or
children to attend priate schools-the idea behind oucher systems.
1bere are targe aifferevce. iv tbe co.t of eavcativg a cbita iv vbtic ava rirate .cboot.. 1he per-child annual
expenditure in the median public school o Rs.2039 is twice that o the median priate school o Rs.1012.
Multiariate regression analysis shows that these dierences remain just as large een with controls or
parental wealth and education and illage wealth and literacy. Moreoer, the dierences remain een when
comparing goernment and priate schools in the .ave illage.
9
lurthermore, since the administratie cost o
running goernment schools is not included in this analysis, this estimate is a torerbovva o the true cost
dierence between public and priate schooling.
"#*Q< YTYP ?&%Z#7< -:$99Q- #&< $#QS #- <VR<(-%Z< #- O9Z<&(0<(7 -:$99Q-
"97#Q A9-7 R<& @7.)<(7
I<)%#( ;i
7$
?<&:<(7%Q< hi
7$
?<&:<(7%Q<
M9Z<&(0<(7 @:$99Q 2039.2 963. 4006.1
@:$99Q "UR<
?&%Z#7< @:$99Q 1012.1 435.8 2248.8
?99& 1608.8 69.3 3898.2
I%))Q< 1558.834 593.5 3446.2 6%QQ#O< 1(:90<
C%:$ 1602.2 436.8 3316.9
2.19.
Prirate .cboot. co.t te.. vot becav.e tbe, bare ferer teacber. er .tvaevt, bvt becav.e tbe, a, teacber. a tot te... 1he
median goernment school spends almost three times as much per student on teacher`s salaries as the median
priate school. Since more than three-ourths o a school`s expenditure is spent on teaching costs ,almost 90
9 It is also likely that this is actually an underestimate o the true dierence: Since goernment schools rarely report rental costs o
buildings, it is likely that we are underestimating these costs at least relatie to priate schools that do not own their own premises.
According to the Punjab Lducation loundation ,2006,, the cost o educating a child in a goernment school is Rs.6000 a year, which
is more than 3 times the amount reported here.
40
percent i salaries o non-teaching sta are included,, the large dierences in per-student teaching costs drie
up the oerall cost dierence between public and priate schools.
10
1his dierence is een more remarkable
gien that priate schools hae almost twice as many teachers per students as goernment schools. Priate
schools cost less not because they hae ewer teachers per student, but because they pay teachers a lot less.
2.20.
2.21.
1be .ocietat co.t of tearvivg i. vvcb torer iv rirate .cboot.. Putting together the results on learning and
costs shows that the quality-adjusted cost in goernment
schools is three times higher than in priate schools. ligure
2.3 shows the cost per percentage correct` on the test.
1his assumes children in Class III hae been through three
years o schooling in the same type o school, and then
diide the costs accumulated oer three years by the
percentage obtained in the test or eery child. 1he results
are shown or the three subjects tested, and separately or
public and priate schools. lor eery subject tested, the
cost per percentage in the goernment sector is at least
three times higher than the priate sector. 1he societal
cost o learning` is much lower in priate schools.
G%O.&< YT]P "$< :9-7 R<& R<&:<(7#O<
:9&&<:7 %- ] 7%0<- $%O$<& %( R.*Q%:
:90R#&<) 79 R&%Z#7< -:$99Q-
260
184
259
75
68
82
0
50
100
150
200
250
C
o
s
t
f
o
r
E
v
e
r
y
P
e
r
c
e
n
t
C
o
r
r
e
c
t
(
i
n
R
s
.
)
Government Private
Cost of Schooling
English Math Urdu
16T @ALHH= G'A1=1"1>@
!itb tbe ecetiov of a fer rer, oort, eqviea gorervvevt .cboot., vo.t gorervvevt ava rirate .cboot. bare aaeqvate
cta..roov. ava btac/boara. bvt botb tac/ etra facititie. .vcb a. tibrarie., aaeqvate toitet. ava .ort. eqvivevt. Prirate .cboot.
borerer, ao offer better .cboot facititie. acro.. att categorie. covarea to gorervvevt .cboot..
School inrastructure is alued or its expected input into the educational production unction-it is
belieed that children cannot learn well unless they hae unctioning classrooms and blackboards and
reasonable student-teacher ratios. It also is likely to proide direct intrinsic alue. Boundary walls, ans, and
toilets make classrooms more comortable and sae or children and, it is assumed, more conducie to
learning. 1he ollowing discussion reiews the basic inrastructure and student-teacher ratios in public and
priate schools, extra` acilities, and dierences across public and priate schools.
10 1hese costs may be oer-stated to the extent that schools do not report rental costs o buildings they own ,in the case o priate
schools, or do not hae to pay rent on ,in the case o goernment schools,. loweer, een i we examine ,priate, schools that do
pay rental, we still ind that teaching costs are the largest share o expenditure.
41
@:$99Q @%3< #() @7.)<(7 "<#:$<& C#7%9-
2.22.
2.23.
2.24.
2.25.
1able 2.3 reports the mean, the
median, the 10
th
percentile and the 90
th
percentile or a number o indicators
related to school size and student
teacher ratios. 1he typical school in the
LLAPS sample has 120 children, 5
teachers, and classes. 1he median
student-teacher ratio is 2, which is below the benchmark igure o 40 suggested by the goernment.
Neertheless, the student-teacher ratio is a problem in schools in the 90 percentile-the student-teacher ratio
or the top 10
th
percentile o schools is only 15, compared to close to 50 or the 90
th
percentile. \e return to
this issue urther below.
"#*Q< YT]P @7.)<(7 "<#:$<& C#7%9- #&< Q#&O<QU
0#(#O<#*Q<TTT*.7 7$<&< #&< -90< -:$99Q- 7$#7 #&< (97
Mean Median
10
th
percentile
90
th
percentile
Students 163.64 119 46 320
1eachers 5.96 5 2 12
Classes .45 5 11
Students Per
1eacher
30.36 2 14.5 49.5
J#-%: R$U-%:#Q S#:%Q%7%<-
1be t,icat .cboot i. rea.ovabt, eqviea iv terv. of cta..roov. ava btac/boara.. Close to 96 percent o the
schools in the sample hae classrooms and 86 percent o these rooms are permanent structures. 1hese
classrooms typically accommodate 33 children. Len without data on the physical dimensions o these rooms,
this inormation alone shows that school acilities are better than generally belieed. Similarly, 95 percent o
all schools hae a blackboard with 26.5 students to a blackboard. loweer, as with student-teacher ratios, the
worst 10 percent o schools hae an aerage o 8 children to a classroom and 60 students to a blackboard.
1be ictvre ror.ev., borerer, rbev cov.iaerivg tbe araitabitit, of toitet. ava .eativg arravgevevt.. Slightly more
than a one-quarter o schools hae no toilet acilities. And those that do seem inadequate, with almost 4
children per toilet. Only 60 percent o schools hae desks aailable and almost one-third only use loors or
mats. Parents requently expressed a preerence or schools with better seating acilities.
'))%7%9(#Q G#:%Q%7%<- %( ?.*Q%: #() ?&%Z#7< @:$99Q-
.art frov tbe.e ba.ic facititie., .cboot. offer tittte et.e iv terv. of ivfra.trvctvre. Only one-third o all schools
oer av, o the ollowing extra acilities - a library, a computer acility, sports equipment, and transport
acilities. Schools are roughly equally split in terms o which o these our acilities they oer. One may not
expect computers to be common in illages, but the lack o sports equipment is surprising. 1he aerage
illage child probably lacks such equipment at home and so the only exposure to this type o equipment may
be at school. 1he lack o transport acilities suggests that distance will play an important role in school choice
and we will examine this in later sections.
42
77.3%
10.3%
7.4%
1.4%
1.4%
2.3%
Class Breaks Assemblies
Extracurriculars Prayer Other
School Activities
77.3%
10.3%
7.4%
1.4%
1.4%
2.3%
Class Breaks Assemblies
Extracurriculars Prayer Other
School Activities
2.26.
2.2.
2.28.
ac/ of etectricit,. It is also worth pointing out that one-third o the schools we sureyed lacked
electricity. 1his may sound minor but most o illages tend to be in airly hot locations with temperatures
rising oer 40
o
C during the school term ,schools are closed or summer recess during the hottest months,.
Since children are likely to learn better in a comortable enironment, this is a serious impediment. It is
common on hotter days to see children sitting under the shade o trees rather than inside stiling classrooms.
A broader issue is whether schools oer an attractie enironment or primary school aged children
,typically between 6 and 12 years o age, both physically and mentally. 1o the extent that children can be
excited about learning through the use o libraries, or just haing a good time at school through organized
sports, it seems that most schools in our sample oer no such beneits. 1his aspect o LLAPS schools is
examined in Box 2.1 below.
Box 2.1: Are schools too boring?
Do schools with ew acilities oer and aboe the basic necessities aect what children do in school lor many
children in high income countries, schools are a un place to go to, especially at the primary leel, not because o
the time spent in classes, but because o the extra-curricular and sports actiities during the school day. 1he
LLAPS project constructed detailed time charts o a day at school to
look at this issue.
G%O.&< YTcP @:$99Q ':7%Z%7%<-
lor the ast majority o students, the typical school day lasts 5'
hours. ligure 2.4 shows how the dierent actiities. More than three-
ourths o the day is spent in ormal classes and the rest o the time is
diided into break and assembly times.
Less than 5 percent o the day ,30 minutes, inoles non-academic
actiities such as superised sports, music,art classes, and extra-
curricular actiities such as drama,debates. 1he ast majority o
schools do not een oer such actiities ,65 or 8 percent,. But most
schools hae a 30-minute assembly period. Low-cost enrichment
programs could add another important element to learning and
educational quality.
/%SS<&<(:<- %( @:$99Q 1(S&#-7&.:7.&<
\hile most schools report adequate classrooms and blackboards, there is a small but signiicant
raction that lack een basic acilities. lor example, 10 percent o schools hae too ew ,or no, classrooms.
Most o these are goernment schools. Priate schools report better acilities oerall, and the aerage
inrastructure indicator or goernment schools is dragged down by schools where there is almost nothing.
43
'Z<&#O< #() )%-7&%*.7%9(#Q )%SS<&<(:<-
2.29.
2.30.
ligure 2.5 presents the ai.tribvtiov o student-
teacher ratios across goernment and priate schools
using a smoothed histogram. 1he typical priate school
has almost one-hal the student-teacher ratio o the
typical goernment school. As with the results on
learning outcomes, large dierences exist ritbiv
goernment schools. 1he dierence between the top 10
percent and the bottom 10 percent in priate schools is
small ,9 students per teacher compared to 33 students per
teacher, relatie to that in the goernment sector-1
students per teacher in the top 10 percent and : students
per teacher in the bottom 10 percent. Out o a total o
823 schools, there are 6 schools ,9 percent, with more
than 50 students per teacher and almost all ,5, o these are goernment schools.
G%O.&< YT^P @7.)<(7 "<#:$<& C#7%9-
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
G
i
v
e
n
R
a
t
i
o
0 20 40 60 80
Student-Teacher Ratio
Private Government
n Government and Private Schools
Student Teacher Ratios
Not only do Government
Schools have higher
average student teacher
ratios, but there are also
many more Government
schools with very high
student teacher ratios.
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
G
i
v
e
n
R
a
t
i
o
0 20 40 60 80
Student-Teacher Ratio
Private Government
n Government and Private Schools
Student Teacher Ratios
Not only do Government
Schools have higher
average student teacher
ratios, but there are also
many more Government
schools with very high
student teacher ratios.
1be re.vtt. for facititie. acro.. vbtic ava rirate .cboot. are .ivitar: rirate .cboot. reort better ivfra.trvctvre ov
arerage rbite vbtic .cboot figvre. are ./erea b, a vvvber of rer, oort, eqviea gorervvevt .cboot.. 1able 2.4 presents
simple mean comparisons o basic and additional acilities in public and priate schools. 1he latter come out
looking better with more classrooms, toilets, blackboards, boundary walls and access to electricity. Combining
these dierent types o inrastructure into two inrastructure indexes` ,one or basic acilities and one or
extra acilities, yields similar results.
11
1here is a 1.2 standard-deiation dierence in the basic acilities index
between the aerage priate and goernment school and a 1.4 standard-deiation dierence or the extra
acilities index. \hile the best goernment schools are comparable to the best priate ones, there are a
number o goernment schools that are truly dismal. 1he 10 percent o all schools that hae ew or no
classrooms are almost all goernment schools. Similarly 80 percent o the 65 schools that hae no or
inadequate ,more than 100 students,board, blackboard acilities are goernment schools. As with other
inrastructure, it is also goernment schools that typically lack toilet acilities.
11
In order to undertake comparison in acilities across type o school we construct an inrastructure index using principal component
analysis ,PCA,. PCA allows us to combine a ariety o related measures into a single index. 1he basic acilities index is constructed
using our indicators: ,i, classrooms per student, ,ii, black-boards per student ,iii, toilets per student, and ,i, sitting arrangements ,like
desks,. 1he basic acilities index is the irst principal component and by construction has a mean o 0 and standard deiation 1.
ligher alues o this index mean better acilities. lor instance, a basic acilities index o -1 means that the school has 50 students to a
classroom, 40 to a blackboard, 250 to a toilet and a 40 percent chance o haing desks. In comparison an index alue o 1 implies that
the school has 20 students to a classroom, 18 to a blackboard, 65 to a toilet and a 90 percent chance o haing desks.
44
"#*Q< YTcP 'Z<&#O< @:$99Q G#:%Q%7%<- [%( Yii]\
M9Z<&(0<(7 @:$99Q- ?&%Z#7< @:$99Q-
D.0*<& 9S AQ#--&990- 3.58 4.64
D.0*<& 9S @<0%_R<&0#(<(7 :Q#--&990- 0.41 1.08
D.0*<& 9S "9%Q<7- 0.26 0.1
D.0*<& 9S JQ#:W*9#&)- 5.04 6.65
?<&:<(7#O< 9S @:$99Q- 5%7$ # =%*&#&U 16 23
?<&:<(7#O< 9S @:$99Q- 5%7$ K#QQ- 6 98
?<&:<(7#O< 9S @:$99Q- 5%7$ G#(- 44 91
?<&:<(7#O< 9S @:$99Q- 5%7$ >Q<:7&%:%7U 50 96
2.31.
2.32.
.. ritb tbe tocatiov atterv. of vbtic ava rirate .cboot., rittage cbaracteri.tic. ao vot etaiv tbe qvatit, of .cboot
ivfra.trvctvre. 1here is hardly any dierence in student-teacher ratios or the aailability o inrastructure across
dierent types o illages. Multiariate regression analysis shows that whether the school is public or priate
is ar more important in explaining these dierences than illage wealth, literacy, or een the parental
education and wealth o students attending these schools.
!bite otiticatecovov, .torie. arirev b, aifferevce. acro..
rittage. are vot cov.i.tevt ritb tbe aata, gorervvevt .cboot. iv eriberat
.etttevevt. geveratt, tac/ etra facititie. ti/e tibrarie., covvter., .ort
eqvivevt, ava trav.ort. ligure 2.6 diides all goernment
schools in the LLAPS sample into those that hae no other
school ,23 percent o all goernment schools,, one other
school ,25 percent,, two other schools ,23 percent,, 3-5 other
schools ,23 percent, and 6 or more schools ,5 percent, within
a 5-minute walking distance. 1he number o other schools
within 5 minutes is a good indicator o whether the school is
in a main cluster or on the outskirts. 1he ertical axis plots
the inrastructure index o the school. \hat is clear rom the
igure is that goernment schools on the outskirts are worse
than those within schooling clusters with a dierence o more than 1 standard-deiation in the extra
inrastructure index. In a multiariate regression context, controlling or the illage that the school is in using
illage-leel ixed-eects, the dierence is hardly reduced and remains highly signiicant. Interestingly, the
G%O.&< YTeP M9Z<&(0<(7 >V7&#
1(S&#-7&.:7.&< %- *<77<& %( :Q.-7<&-
-.5
0
.5
1
n
f
r
a
s
t
r
u
c
t
u
r
e
n
d
e
x
O
n
ly
S
c
h
o
o
l
1
O
th
e
r
S
c
h
o
o
l
2
O
th
e
r
S
c
h
o
o
ls
3
-
5
O
th
e
r
S
c
h
o
o
ls
6
o
r
M
o
r
e
O
t
h
e
r
S
c
h
o
o
ls
n Response to Competition
nfrastructure in Government Schools
45
same ariable is vot corretatea with inrastructure or priate schools. 1his inding suggests that the community
that the goernment school is located in may not be the issue, or that priate schools are better able to
maintain and look ater their inrastructure compared to their goernment counterparts een when located in
the same settlement.
2.33. 1be fact tbat aifferevce. iv ba.ic facititie. are arirev vore b, .cboot t,e ratber tbav rittage reattb,titerac, .vgge.t. tbat
robtev. ritb ivaaeqvate facititie. iv gorervvevt .cboot. va, vot be ivtractabte. I priate schools are able to retain
teachers and proide basic acilities een in less desirable enironments, goernment schools can do so as
well. 1he problem appears to be the ast dierences in goernment school inrastructure ritbiv the same
illage, with those schools located in central clusters aring better. In 2003, the goernment recognized that a
signiicant raction o public schools lacked basic acilities, and one o the ocuses o the Punjab Lducation
Sector Reorm Program was to proide missing acilities` in goernment schools.
J9VP YTY j@90<9(< S&90 7$< Z%QQ#O< -79Q< 9.& 5#7<& R.0Rk
\heneer we see goernment schools without desks, without blackboards and without electric ans, our
irst instinct is to blame the goernment or not proiding these resources. loweer, proiding
resources to schools may not be enough. 1hey still need to be maintained and looked ater, and our
experience oten shows that this does not happen.
One school that I encountered during my ield work had the resources necessary to improe the
learning enironment, but was unable to use them. I was working in a illage in Rahim \ar Khan and
when I isited the goernment girl`s elementary school, the head-teacher came out to meet us. She
seemed quite experienced, with more than 15 years in the teaching proession.
1he school building appeared to be newly constructed, with six class rooms and a separate oice or the
head-teacher. Despite the relatiely high quality inrastructure, there were no electric ans and children
were sitting in temperatures exceeding 45 degrees. \hen I asked the head-teacher i the school had
electricity, she told me that they did hae a connection. \ere electric ans not gien to the school
despite the electric connection 1he head-teacher replied:
1be gorervvevt aia gire v. fav. for tbe cta..roov. bvt tbe aoor. of tbe cta..roov. ava office are vot .trovg ava are ea.it,
bro/ev. f .oveove .teat. tbe fav., cavvot covev.ate tbe gorervvevt for tbe to... .treaa, tbe rater vv ba. beev
.totev trice. !e ;tbe .cboot teacber. ava v,.etf) baa to bv, tbev agaiv frov ovr orv oc/et.. o re bare revorea tbe
fav. ava /ee tbev toc/ea .earatet,`.
!b, aov`t ,ov ivrotre tbe covvvvit,, .o tbat tbe, ta/e care of tbe re.ovrce. iv tbe .cboot. 1bi. va, be a gooa iaea .ivce
tbe, are re.evt iv tbe rittage erev at vigbt.` I asked.
She replied, oveove frov tbe rittage .tote tbe ratervv. 1be, ao vot bare av, ivtere.t iv tbe .cboot ava tbeir
attitvae torara. tbe .cboot i. rer, vegatire. !e reqve.tea tbe eavcatiov aeartvevt for a fvtt tive gate/eeer; ovt, if re
are girev a gate /eeer, ritt ovr .tvaevt. be abte to v.e tbe fav. avrivg tbe .vvver.`
-Contributed by Iran Ahmed
46
2.34. 1be aata .bor tbat ivfra.trvctvre iv gorervvevt .cboot. ba. ivrorea betreev 200 ava 200:, bvt tbe ivrorevevt.
iv rirate .cboot. bare beev erev greater. 1able 2.5 uses the 2005 round o the LLAPS sureys and combines it with
data on school inrastructure rom 2003. 1here hae been dramatic improements in goernment schools in
terms o the number o semi-permanent classrooms, aailability o toilets and aailability o libraries. lor
many types o acilities though, the improements in priate sector schools hae been een higher. lor
instance, the number o semi-permanent classrooms in goernment schools increased by 86 percent,
compared to 12 percent in priate schools. \hile the goernment sector has shown greater improements
or certain types o acilities ,such as toilets,, the gap between the goernment and priate sector still remains
large and signiicant.
"#*Q< YT^P A$#(O<- %( @:$99Q 1(S&#-7&.:7.&<
Public
2003
Public
2005
Priate
2003
Priate
2005
Percent
Increase in
Goernment
Percent
Increase in
Priate
No. Classrooms 3.58 3.4 4.64 4.0 -3 -12
No. Semi-permanent
classrooms
0.41 0. 1.08 2.46 86 12
No. 1oilets 0.26 0.36 0.1 0.1 38 0
No. Blackboards 5.04 5.19 6.65 .02 3 6
las Library 0.16 0.26 0.23 0.49 65 109
las \all 0.6 0.63 0.98 0.92 -5 -6
las lans 0.44 0.48 0.91 0.94 8 4
las Llectricity 0.50 0.55 0.96 0.95 10 0
6T @ALHH=1DM I'C4>"P ?H">D"1'= AHDA>CD@
2.35. 1bere i. a ri.ivg covcerv iv Pa/i.tav tbat tbe re.evce of rirate .cboot. i. teaaivg to eavcatiovat aartbeia.
According to this argument, Lnglish medium priate schools oer a dierent leel o education-but at a
price. Consequently, they are used only by the ery rich while the rest are orced to use ,lower quality,
goernment schools. 1he increasing tie-up between Lnglish skills and wages means that priate schools will
oer a way or the rich to become een richer. Compounding the problem o wealth stratiication is that o
social stratiication-internecine and clan-based politics has ed its way into the educational systems and
student bodies in schools are coming to represent these diides in the community. At a ery basic leel, these
concerns about dierent mediums o instruction are true-priate schools tend to use Lnglish ar more as a
medium o instruction and almost neer teach in the local language ,see the box,. \et, the data suggest a
more nuanced picture than the commonly heard stories o erfect segmentation.
4
53%
5%
1%
42%
61%
5%
23%
11%
Government Private
Urdu English
Both Urdu and English Urdu and Local Language
Language of nstruction
J9V YT]P =#(O.#O< 9S %(-7&.:7%9( _ >(OQ%-$ Z<&-.- B&).X
More than hal the schools in the LLAPS sample use only Urdu, which is not the local
language in any o the illages in Punjab, as their medium o instruction. Around 30 percent
either partly ,in combination with Urdu, or primarily use a local language as well, whereas
the remaining 15 percent partly use Lnglish. \hile we ind little dierences across schools
o dierent types ,goernment or priate, or in dierent illages ,rich,poor, leel o
education, in terms o the length o the class day or time spent on ormal teaching, but there
are large dierences across schools in terms o language o instruction. ligure 2. shows that
goernment schools are much more
likely to use the local language
,Punjabi, Saraiki, lindko, Pastho, in
addition to Urdu as their language o
instruction. Priate schools are
relatiely more likely to use Lnglish in
addition to Urdu in the classroom, and
also slightly more likely to exclusiely
use Urdu. In comparison there are little
dierences in language o instruction
across illages with dierent literacy
leels, although richer illages are
slightly more likely to use Lnglish ,with
Urdu, in the classroom.
Dierences across school types do not
necessarily lead to educational
apartheid to the extent that all types o
parents choose what type o school they preer. It is not the case that the poor or less
educated are destined only or one approach to teaching.
12
G%O.&< YTfP =#(O.#O< 9S 1(-7&.:7%9(
48
12
For more on language oI instruction in Pakistan, see Rahman (1997).
2.36.
2.3.
1be fir.t vvavce .vgge.tea b, tbe aata i. tbat of
geograbicat .egregatiov of rirate .cboot.. ligure 2.8 shows
greater clustering o priate schools and their
location in richer settlements. Schools are again
diided ,as in ligure 2.6, into dierent categories
depending on the number o schools within a 5-
minute walking distance-those with no other
schools, those with one other school, two other
schools, 3-5 other schools and 6 or more schools. It
then plots two dierent ariables. 1he lighter bar
shows the percentage o all schools in the releant
category that are priate. lor instance, o all schools
that do not hae any other schools within a 5-minute
walking distance, ewer than 20 percent are priate
and the rest are goernment. 1he percentage o priate schools increases with the number o other schools in
close proximity-among all schools with 6 or more schools nearby, 60 percent are priate and 40 percent are
public. 1he darker bar shows the aerage wealth o students attending schools by the same clustering
categories. Students in schools located ar rom any other school are relatiely poor, with aerage wealth
leels 0.5 standard-deiations below the mean, and student wealth increases with the number o other schools
in the icinity.
G%O.&< YT`P ?&%Z#7< @:$99Q AQ.-7<&%(O
-.5
0
.5
1
%
s
c
h
o
o
l
s
p
r
i
v
a
t
e
/
s
t
u
d
e
n
t
-
w
e
a
l
t
h
i
n
s
c
h
o
o
l
N
o
n
e
1
m
o
r
e
2
m
o
r
e
3
-
5
m
o
r
e
6
o
r
m
o
r
e
(By Schooling Clusters)
Private Schools and Student Wealth
Standardized Student Wealth
% Schools Private
Number of SchooIs within a 5-minute waIking distance
Students in these schools are
in the poorest 30%
Students in these schools are
in the richest 30%
-.5
0
.5
1
%
s
c
h
o
o
l
s
p
r
i
v
a
t
e
/
s
t
u
d
e
n
t
-
w
e
a
l
t
h
i
n
s
c
h
o
o
l
N
o
n
e
1
m
o
r
e
2
m
o
r
e
3
-
5
m
o
r
e
6
o
r
m
o
r
e
(By Schooling Clusters)
Private Schools and Student Wealth
Standardized Student Wealth
% Schools Private
Number of SchooIs within a 5-minute waIking distance
Students in these schools are
in the poorest 30%
Students in these schools are
in the richest 30%
Re.earcb ov eavcatiov iv Pa/i.tav a. rett a. tbi. fietaror/
.vgge.t. tbe cboice of .cboot va, be affectea b, tbe ai.tavce of tbe .cboot to tbe
bov.ebota-a fivaivg covfirvea b, aata re.evtea iv tbe cbater ov
bov.ebota.. ligure 2.9 shows what this implies or the student
population in a school, in terms o their geographical locations.
lere, the graph shows the percentage o children who come to
the school rom dierent distances, as reported by the school
head-teacher. 1he oerwhelming majority lie less than a 15-
minute walk rom the school they attend, and close to 40 percent
come rom houses within 5 minutes o the school they attend.
1he patterns are similar or public and priate schools with 38
and 42 percent o all students, respectiely, liing within a 5-
minute distance o the school. Interestingly, in richer and more literate illages, the percentage o children
G%O.&< YThP I9-7 -7.)<(7- :90< S&90
5%7$%( ^ 0%(.7<- 9S 7$< -:$99Q
33.8%
39.9%
26.3%
Less than 5 Minutes 5-15 Minutes
More than 15 Minutes
Walking Time
Proximity of Students to School
49
who come rom within 5 minutes ivcrea.e. by 3- percentage points, suggesting that children in poorer illages
trael fartber to school. 1o the extent that distance to school is an important actor in school choice, the act
that priate schools generally do not locate in peripheral settlements suggest that the poor are geograbicatt,
segregated. Proiding incenties or priate schools to moe to peripheral areas will enhance accessibility or
the poor.
2.38. 1be .ecova vvavce i. tbat, ae.ite tbi. geograbicat .egregatiov, oor ava ittiterate arevt. ao at.o .eva tbeir cbitarev to
rirate .cboot.-t,icatt,, .vcb .cboot. bare cbitarev frov att /iva. of bac/grovva.. ligure 2.10 shows the distribution o
household wealth and education or students in public and priate schools. 1he construction o these
measures is described in the technical box below. 1he wealth index has a mean o 0 and standard-deiation 1,
and the education index measures the percentage o children in the school who come rom households where
at least one parent reports higher than elementary education. Both indices show that the parents o children
in priate schools are wealthier and more educated-these children tend to come rom amilies with wealth
indices 1.2 standard deiations higher and 60 percent o priate school students hae at least one parent with
more than primary education compared to 45 percent or those in public schools. Neertheless, there is
substantial orerta between the two distributions: Priate schools are not composed only o students rom
wealthy and educated amilies, and goernment schools also hae students rom rich and educated
backgrounds. lor example, in 13 percent o priate schools, more than 50 percent o the student body comes
rom households where the parents report no education.
G%O.&< YT;iP A$%Q)&<( %( ?&%Z#7< -:$99Q- :90< S&90 &%:$<& #() 09&< <).:#7<) *#:WO&9.()-g*.7 (97
#QQ &%:$ :$%Q)&<( #&< %( R&%Z#7< -:$99Q- #() (97 #QQ R99& :$%Q)&<( %( O9Z<&(0<(7 -:$99Q-
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
S
t
u
d
e
n
t
s
o
f
G
i
v
e
n
W
e
a
l
t
h
-4 -2 0 2 4
Average Wealth ndex of Students
Private Government
n Government and Private Schools
Wealth of Students
On average,
Private School
Students are
Wealthier than
Government
Students.
But not all private
students are richer
than government
students. n a large
share of schools
the students are
indistinguishable.
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
S
t
u
d
e
n
t
s
o
f
G
i
v
e
n
W
e
a
l
t
h
-4 -2 0 2 4
Average Wealth ndex of Students
Private Government
n Government and Private Schools
Wealth of Students
On average,
Private School
Students are
Wealthier than
Government
Students.
But not all private
students are richer
than government
students. n a large
share of schools
the students are
indistinguishable.
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
G
i
v
e
n
S
t
u
d
e
n
t
M
a
k
e
u
p
0 20 40 60 80 100
Percentage of Students in School Who Have a
Parent With Some Secondary Education
Private Government
Parent Education and Child School Choice
On average,
Private students
have more
educated
parents.
But not all Private
students have
more educated
parents than
Government
Students. n a
large share of
schools the
students are
indistinguishable.
S
h
a
r
e
o
f
S
c
h
o
o
l
s
w
i
t
h
G
i
v
e
n
S
t
u
d
e
n
t
M
a
k
e
u
p
0 20 40 60 80 100
Percentage of Students in School Who Have a
Parent With Some Secondary Education
Private Government
Parent Education and Child School Choice
On average,
Private students
have more
educated
parents.
But not all Private
students have
more educated
parents than
Government
Students. n a
large share of
schools the
students are
indistinguishable.
50
2.39. 1be tbira vvavce i. tbe tac/ of .egvevtatiov atovg ca.te tive.. A school caste segregation index was
constructed to estimate the probability that two students chosen rom the .ave school will be rom a dierent
caste. An index alue o 1.0 or this intra-school ractionalization index indicates that the student body in the
school is mixed, since the probability o two students being rom dierent castes is 100 percent. Conersely, a
alue o 0.0 indicates perect segregation ,the probability o two students in the school being rom dierent
castes is 0 percent,. 1he irst two rows o 1able 2.6 below show that i we look across schools, the probability
that randomly selected children rom a girev school will be rom dierent castes is around 0.52-0.53. 1his
suggests that there isn`t that much caste segregation ritbiv schools--i there are two castes in the school,
students are equally likely to belong to one o them. loweer, we may still see segregation across schools i
these two castes are not present in the other schools. In order to check or segregation acro.. schools we
construct a similar index but now look at the probability that two students randomly picked rom aifferevt
schools will be rom a dierent caste - an ivter-school ragmentation index. In this case an index alue o 1
means complete segregation. 1he last three rows o the table show that this measure is 0.8. 1his suggests
that while there clearly isn`t ull segregation across schools, there is some degree o segregation across schools
as this index is higher than then intra-school ragmentation index. Also worth noting is that the social
stratiication obsered in priate schools is much the same as in goernment schools.
1able 2.6: School Segregation
Intra-School lractionalization Mean Caste Segmentation Index
\ithin Indiidual Goernment Schools 0.52
\ithin Indiidual Priate Schools 0.53
Inter-School lractionalization Mean Caste Segmentation Index
Between Dierent Goernment Schools in the Same Village 0.8
Between Dierent Priate Schools in the Same Village 0.5
Between Goernment Schools and Priate Schools in the Same Village 0.3
2.40. Pvbtic ava rirate .cboot. are tbv. vot erfectt, .egvevtea b, eitber arevtat reattb,titerac,, or b, ca.te ;aat). 1he
obsered segmentation is drien by a combination o price actors-the poor are less likely to be able to
aord the ees o priate schools-and distance actors, because priate schools are located urther rom the
poor. It is encouraging that apart rom these two eects, we do not ind priate schools systematically
discriminating against children rom any particular background. Chapter 5 discusses how the price segregation
may be addressed through goernment policy. 1he segregation arising rom location patterns may be harder
to address-but it is worth pointing out that the best way o addressing such segregation is by encouraging
more priate schools to open, rather than trying to close or regulate those already in existence.
51
J9V YTcP @9:%9<:9(90%: %()%:<-
K<#Q7$ 1()<V
1his index is constructed rom data collected in a short surey administered in eery school to a randomly
selected group o 10 students in Class III ,or all, i less than 10,. Using ownership o dierent assets as
reported by the students, we construct a single wealth index through principal component analysis. 1he
index ,with a mean o 0 and standard deiation o 1, is the irst principal component and has a higher
alue or wealthier students.
>).:#7%9( 1()<V
1his index is constructed using the raction o children in a school with at least one parent who has some
non-zero leel o schooling ,similar results obtain i we look at higher leels o parental education,.
8##7 G&#O0<(7#7%9( 1()<V
1his index is a measure o the heterogeneity o castes within and between schools. A alue o 1 means
that all students are rom dierent castes, while a alue o 0 means that eeryone is rom the same caste.
A urther explanation o the index is presented in paragraph 2.38.
61T /1@AB@@1HD
2.41.
2.42.
1bi. cbater bigbtigbt. ba.ic cbaracteri.tic. of vbtic ava rirate .cboot. iv tbe .P ro;ect rittage. ava bor tbe.e
cbaracteri.tic. aiffer betreev .cboot. iv cevtrat ctv.ter. ava .cboot. tocatea ov .etttevevt eriberie.. It also examines claims
that priate schools lead to social stratiication, and inds little eidence o strong segregation along socio-
economic or caste lines. Priate schools are almost always clustered in the main settlement while goernment
schools are present in the settlements as well at the illage periphery. 1hree implications ollow rom these
location patterns. lirst, priate school ees and proits are airly low. 1he schools that report higher reenues
and proits also hae a higher quality o learning. 1he cost to educate a child in the goernment sector is
twice that in the priate sector. lactoring in dierences in learning implies that the cost-per-percent in public
schools is three times that in priate schools. Second, basic inrastructure in public and priate schools
appears to be in place. Priate schools do a somewhat better job o proiding basic inrastructure than
goernment schools, which hae some o the worst inrastructure. 1hird, there is eidence o segregation
arising rom pricing and location patterns, but no eidence o discrimination against illiterate or poor parents,
or children rom dierent castes. 1he discussion below suggests how these characteristics can be used to
inorm a debate on what should be done about schools in the proince.
1(S&#-7&.:7.&<a *#-%:+ #))_9(-+ #() <V7&#-
LLAPS project illages by and large report unctional classrooms, reasonable access to blackboards
and manageable student-teacher ratios. 1he problem appears when we look at additional acilities, such as
52
electricity, ans and toilets and een more so when we look at libraries and basic sports equipment. 1hree
important questions present a ramework or the wider debate.
2.43.
2.44.
2.45.
ir.t, rbat .bovta tbe ivfra.trvctvre ac/age for a .cboot too/ ti/e. lollowing the diision in the chapter,
one could think o the basics` ,classrooms and blackboards, and extras` such as ans, toilets, libraries and
transportation. 1here appears to be a widespread consensus that both the basic and some extra acilities ,like
bathrooms, should be aailable in eery school. \et extras` make schools a un place or children, and
educators oten suggest that simple additions that make learning un` also make children want to come to
school more regularly. \hat this really hinges on is whether we think o the decision to attend a school as a
parental choice, or partly, a choice o the child. 1he chapter on households shows that parents make ery
conscious choices about which child to push` through school, and which child to exert less eort on. 1he
act that schools are boring means that those children whose parents do not push them to attend school will
likely not attend. In our surey, the reasons parents gae or children not attending school were sometimes
related to the child`s perceptions o school, rather than the opinions o the parents. Reasons or children
remaining out o school included reasons such as the child was araid o teacher` or does not want to go to
school`.
riaevce .bor. tbat .ivte tbivg. tbat va/e .cboot. vore attractire teaa to targe gaiv. iv attevaavce, erev ritbovt a
.vb.tavtiat cbavge iv tbe bevefitco.t ratio for arevt.. lor instance, an experimental school meal program in
Rajasthan carried out by the NGO Sea Mandir, dramatically increased enrollment although the cost o the
eeding program was much smaller than the wage losses incurred by parents when their children stopped
working to attend school. In a detailed ealuation, Raallion and \odon ,2000, ind similar eects in
Bangladesh-school meal programs led to large gains in attendance, but no decline in child labor. Recent
studies by the International Labor Organization ,ILO, ind that a large group o children are idle`, in the
sense that they are neither in school, nor engaged in child-labor and the chapter on households shows similar
patterns or Pakistan. A recent ealuation o a program run by the organization Child Resources International
in Islamabad shows that child-centered` teaching approaches help to improe attendance and test-scores
,Naseer and Patnam ,200,,. Among primary-age children, the hours saed by not going to school or out-o-
school children are not spent working-they are spent playing. 1he act that most out-o-school children are
idle` suggests that increasing enrollment may not be that hard since it does not need to address the trade-o
between schooling and wages. Making schools child-attractors` by putting in extra acilities, such as small
libraries could improe enrollments substantially.
ecova, rbat i. roce.. b, rbicb aeci.iov. are vaae to revorate or vt iv tace ver ivfra.trvctvre iv vbtic .cboot..
Under the Punjab Lducation Reorm Program considerable unds were spent to proide and repair missing
53
acilities`. 1he situation has improed or many schools since 2003, but improements in priate school
inrastructure were een greater during the same period, without any inancial assistance rom the goernment.
Not surprisingly, in recent interiews head-teachers expressed concern that there is no clear mechanism or
petitioning the education authorities to ulill an inrastructure need. \hen such petitions are made, there is
oten no response. Gien that the condition o inrastructure aries dramatically across schools, there is a
clear need or a process to meet needs in a timely and targeted ashion, rather than by blanket proision. In
addition, the current stock o inrastructure requires maintenance. Communities are oten implicated in the
improper use o school inrastructure. Maintenance is a critical issue that requires imaginatie solutions.
6%QQ#O< Q<Z<Q -:$99Q Q9:#7%9(-P 5$9 )<:%)<-X
J9V YT^P =9:#7%(O # -:$99QX K$9 )<:%)<-X
In one o the sample illages, there are two goernment girls` primary schools. One school was built in
193 and the other was built in 198. 1he goernment girls` primary school is a little bit urther rom the
main settlement, but still less than a 15-minute walk. 1here are three teachers and 35 students.
\hen we isited this school and talked to the teacher about the small number o students, she told us that
this school was the irst school in the illage but the Nazim approed a new goernment girls` primary
school in the main illage settlement instead o upgrading this old school. Parents now preer to send
their children to the newer school, especially when they are liing in the main settlement.
1he teacher told me that their school was in the ields and there is no boundary wall, sometimes they ind
dangerous animals in the school and two days ago, when she opened a cupboard there was a snake in it.
1he goernment is spending almost Rs.1500 eery month on teachers` salaries, which means it costs
Rs.6000 per child per year.
I don`t understand why the goernment did not check beore opening the new school i there is already a
school near the settlement` or in the settlement`, and i there was, whether they should close or
upgrade it when they built the new school.
-Contributed by Kashi
2.46.
!bo .bovta aeciae rbere to tocate a .cboot. 1he decision o where to place a goernment school becomes
ar more complicated once priate schools enter the picture and we recognize that there are multiple
settlements with diering population characteristics in the same illage and there is a strong relationship
between distance and enrollment. 1he problem o access to schooling is illustrated in illage ater illage by
the large ariation ritbiv the same illage- some schools hae 2 teachers and 200 children and others hae 8
teachers and 90 students. 1his huge ariation relects large cost dierences in educating children in the
goernment sector, which can be as low as Rs.400 per child per year to as high as Rs.8000 per child per year.
54
2.4.
2.48.
2.49.
2.50.
atavcivg tbe veea. of a eriberat rittage .etttevevt ritb a oor ovtatiov ava a vaiv rittage .etttevevt ritb a targe
vvvber of rirate .cboot. i. aifficvtt. Goernment schools in peripheral settlements signiicantly improe equity in
access to schooling-the children attending these schools are usually rom poorer amilies and are at risk o
bypassing education entirely without access to these schools. In main settlements, goernment schools
compete directly with priate schools and there are seeral such schools with ew students. Access here is not
an issue. 1he real question is what would happen to enrollments i these schools were shut down 1he type
o data that is collected by the proince cannot be used to inorm such a decision since school codes used in
the Lducational Management Inormation System ,LMIS, do not allow easy mapping o schools to illages.
As incredible as this may sound, there is currently no easy way to identiy the number o schools in eery
rittage, which o course precludes going below this leel to the settlement. 1he single exception is the recently
completed school census, which details all schools in the country, both public and priate, and identiies them
by the illage they are in. Lnsuring access to these data would support a greater understanding .
.v ivortavt fir.t .te rovta be to vtt togetber aata ov tbe tocatiov. of .cboot. ava va/e tbev vbtic .o tbat
re.earcber. ava tavver. cav cove v ritb otevtiat .otvtiov. ritb fvtt acce.. to tbe reteravt ivforvatiov. laing such data
would make it easier to set up a process or setting up or closing down a school and deciding how teachers
could be relocated within schools in the same illage. A central allocation o teachers to speciic schools in a
illage may be less eicient than allowing Village Lducation Committees to make these decisions taking into
account their speciic enironment. Again, within the Pakistani context and the existing structure o power
relations this may or may not work.
@$9.Q) R&%Z#7< -:$99Q S<<- *< &<O.Q#7<)X
Regulating priate school ees is a topic o regular discussion among policymakers. In some areas
such as Islamabad, measures hae been taken to ensure that priate schools are charging appropriate` ees.
1he argument or regulating prices can stem rom one o two economic rationales: ,a, either the irm ,priate
school, being regulated is a monopoly with excessie proits or ,b, that lack o inormation on the part o
consumers leads to prices that do not relect quality.
1be re.vtt. bigbtigbtea bere frov tbe .cbootivg evrirovvevt iv Pvv;ab rittage. .vgge.t tbat ;v.tificatiov for eitber of tbe.e
ratiovate. i. rea/. Priate schools are oerwhelmingly located in school clusters and thereore behae in a
highly competitie manner. In act, the only potential monopolies in these enironments are goernment
schools located outside the main settlement where children hae no other choice. Consequently, priate
school ees are relatiely low and proits roughly correspond to the monthly wage o a priate school teacher.
1his is precisely the opportunity cost o the priate school`s head-teacher. linally the considerable ariation
in priate school ees suggests that ees respond to quality. A school with aerage test scores ,across all three
55
tests, charges Rs.1000 per year, a school with test scores 2 standard-deiations aboe the mean charges
Rs.l800 per year. Some priate schools are more expensie because they are better.
2.51.
2.52.
2.53.
v .vcb av eavcatiovat var/et attevt. to covtrot or regvtate rice. ritt re.vtt iv covev.ativg actiov b, rirate
.cboot.. lor instance, capping the prices that priate schools can charge implies that those proiding quality
leels aboe that justiied by the price will either lower their quality or shut down. Undoubtedly, there may be
some priate schools where proits are not excessiely high, or prices are aboe those justiied by the quality
o learning, but cracking down on a ew such schools could well impose a heay regulatory burden on the
rest. Instead, gien that there may still be a concern that parents are not adequately inormed about school
quality, an alternate approach would be to oster competition between schools through standardized testing
and the proision o report cards. 1he LLAPS project initiated such a scheme and the initial results o the
randomized interention appear quite positie: Proiding inormation on the relatie perormance o schools
to parents induced greater competition between schools and raised their inestments, leading to
improements in learning outcomes or the aerage child, and een more so or those who started o at
lower leels o learning.
@9:%#Q :9-7- #() R&%Z#7< -:$99Q-
Consider the ollowing arguments:
Corervvevt. .bovta get ovt of tbe rori.iov of eavcatiov ava focv. ovt, ov it. fivavcivg. igvre 2.11 i. tbe
rea.ov-if tbe co.terercevt iv vbtic .cboot. i. tbree tive. a. bigb a. iv rirate .cboot., it i. vvcb cbeaer
to tet tbe vove, fottor tbe cbita b, roriaivg rovcber. ava tettivg cbitarev aeciae rbere to go.
And,
Prirate .cboot. cav verer roriae eavcatiov to tbe oor, becav.e tbe, are ivtere.tea ovt, iv va/ivg rofit..
oo/ at igvre 2.. t i. ctear tbat rirate .cboot. tocate ovt, iv tbe ricber .etttevevt. of tbe rittage, ava rbere
gorervvevt .cboot. are atreaa, re.evt. f gorervvevt .cboot. are .bvt aorv, bor ritt oor cbitarev iv eriberat
.etttevevt. get acce.. to .cbootivg, rbicb i. tbeir cov.titvtiovat rigbt.
1his chapter shows that both statements are correct. Costs o learning are ar lower in priate
schools and their ees are set in a competitie ashion since they are located in schooling clusters. 1he
downside o these location patterns is that priate schools are seldom ound in peripheral settlements where
households are poorer. Despite criticisms, the goernment is deliering education in an equitable manner.
Not only are basic inrastructure and student-teacher ratios similar in rich and poor illages, but goernment
schools are oten the only access to education or children outside the main settlements. Data rom the
National Lducation Census ,2005, conirm that these patterns hold or the entire country-there are more
goernment schools per capita in poorer compared to richer illages.
56
2.54.
2.55.
1be correct otic, ;or vbticrirate artver.bi) rovta .eriov.t, cov.iaer tbe .atiat ai.tribvtiov of tbe ovtatiov iv
a rittage ava tbe acce.. to .cbootivg. lor instance, ouchers may leae a lot o children out o school i they are
insuicient to coer the cost o educating children in peripheral areas. Alternatiely, where priate schools are
already operational, ouchers may make a lot o sense gien the dramatic dierences in cost-per-percent
across public and priate schools.
!bat i. at.o ctear i. tbat tbe vaiv rea.ov for tbe.e targe co.t aifferevce. betreev vbtic ava rirate .cboot. i. tbe
covev.atiov of teacber.. 1he wage-bill in priate schools is less than one- quarter o that in goernment schools,
controlling costs thus has eerything to do with limiting the wage bill and restructuring teacher compensation.
1he next chapter turns to this diicult topic.
5
Chapter 3: Do 1eachers 1each
3.1 vrorivg teacber qvatit, ava erforvavce ba. beev cevtrat to vo.t eavcatiovat reforv effort. iv tbe rorta. A decade
o research on learning outcomes conirms that teachers are the most important school-based input: a school
with leaking roos, no textbooks, and uninoled parents can still produce good learning outcomes or
students i the teacher is motiated and committed.
3.2
3.3
3.4
. cto.e too/ at teacber cbaracteri.tic. ava covev.atiov rereat. aravatic aifferevce. betreev gorervvevt ava rirate
.cboot.: A teacher in a public school is absent one-ith o the time and has students that perorm ery poorly
but still earns 5 times more than a teacher in a priate school who is present nearly eery day and has students
that perorm ery well. One implication may be that the educational system would bevefit i this goernment
teacher were to stay at home, pocket 85 percent o his salary and use the other 15 percent to pay a teacher in
the priate sector to take his,her place. 1his chapter shows that such compensation schemes may still hae
beneits i used appropriately. 1he chapter analyzes why and how goernment and priate compensation
schemes are so dierent and suggest ways to build on the strengths o both to improe the oerall quality o
education.
Covarivg bor tbe rirate ava vbtic .ector. rerara. teacber. offer. iv.igbt ivto rbat factor. vatter iv tbe
roavctiov of eavcatiov. In a healthy education marketplace priate sector wages relect productiity ,more
eectie teachers are paid higher salaries, and outside opportunities ,good teachers hae multiple job
opportunities,. 1he report examines teacher demographic and educational characteristics in both sectors,
incenties in the orm o compensation policy and the system o monitoring, as well as the eect o these
incenties on teacher eort as measured by absenteeism and student outcomes as measured by test scores.
1he ocus o the LLAPS surey on illages with both priate and public schools is o particular
alue, since we are able to compare teacher demographic proiles, characteristics and compensation
dierences within the same illage and can appropriately correct or geographical eects in the analysis.
58
3.5
3.6
1be aata .vgge.t tbat rirate ava vbtic .cboot. ei.t iv aifferevt ava .earate .,.tev.. On most dimensions o
interest-teacher proiles, competence, eort and rewards-a series o the public-priate distinctions
dominate the landscape:
" 1eacber .etectiov ava birivg ractice. aiffer. 1he goernment sector hires teachers based on education and
training qualiications. Contrary to iews that richer illages or schools with richer children receie
the best teachers, the data show goernment teachers are eqvitabt, distributed across illages and
schools. 1he priate sector hires primarily locally born and resident young women, most o whom do
not hae a post-secondary degree or ormal teacher training.
" 1eacber. iv rirate .cboot. are aia a fractiov of tbe .atar, of vbtic .cboot teacber. earv. Ater controlling or
obsered dierences, goernment teachers are paid between 3 and 4 tive. as much as their priate
counterparts. But the story is not just restricted to leel dierences in salary, the reward and penalty
structure o teacher pay is radically dierent in the two sectors as well.
" Corervvevt .atarie. are target, aetervivea b, eerievce, eavcatiov ava traivivg. Priate sector compensation
schemes are more complex-not only do they reward teachers or eort and perormance, they also
respond to the ovt.iae oortvvitie. aailable to the teacher. 1hose who are likely to be paid higher in
jobs outside the teaching sector get higher salaries
" Ovce birea, accovvtabitit, for teacber. iv gorervvevt .cboot. i. tivitea. Job retention is used to leerage teacher
perormance once they are hired, but teacher turnoer is ery low and most o it happens as a result
o routine transers, not irings or absenteeism. 1urnoer in the priate sector is high and
requent-close to 25 percent o teachers are replaced eery year. 1his requent turnoer may hurt
the priate sector, but it also allows constant replenishment o the teaching workorce and pruning o
non-perorming teachers.
1be gorervvevt .,.tev rerara. ivvt. ava tbe rirate .ector rerara. erforvavce. 1he priate sector responds to
local conditions, the goernment sector does not. Combining these two systems presents an opportunity or
public-priate partnerships to enhance educational outcomes. 1here may be better solutions een under the
current compensation scheme ollowed in goernment schools. A series o acts and indings based on
special purpose data collected through the LLAPS project is used to rame the subsequent discussion.
59
J9V ]T;P K$#7 5< W(95 #*9.7 7<#:$<&-
Studies on teacher perormance in the United States and countries around the world systematically show that
teachers matter.
13
lor most other schooling inputs ,inrastructure, student-teacher ratios, direct unding to
schools, proision o textbooks and others,, careul empirical work inds little consistent eidence o a link
with student achieement. In contrast, there are some teachers whose children learn systematically more,
others whose children learn systematically less. loweer, while it has been easier to show that teachers
matter`, it has been ery diicult to pinpoint rbat it i. about teachers that matter-typically, experience
matters only in the irst year o teaching and qualiications and training hae ery little eect. 1hese indings
hae led researchers to hypothesize that intrinsic motiation and aptitude or teaching are the most important
determinants o teacher eectieness.
1he indings rom the United States and other rich countries may or may not hold or countries like Pakistan.
1he problem is that because educational systems in the richer countries already ensure that only highly
educated and trained teachers join the workorce and already hae well-designed monitoring mechanisms to
ensure that teachers show up or work, the dierences between dierent teachers are small.
1he dierences among teachers qualiications and work ethic in Pakistan and South-Asia is generally large-
there may be teachers with only secondary education and others with a masters degree, there are some
teachers who don`t show up or work and some who are present eery day.
1he literature on teachers in South Asia is small. Kingdon and 1eal ,2005, compare teacher incenties and
student outcomes in 30 priate and goernment schools ,12 teachers, in one Indian study. Murgai and
Pritchett ,2006, cite secondary data and use household surey data to look at teacher wages in both sectors,
arriing at conclusions ery similar to those presented here. In Pakistan, the work on teachers has typically
been conducted in education departments. Recent experiments` rom India on the importance o teacher
incenties and monitoring is discussed later in the report. loweer, while this strand o research has led to
many interesting studies on teacher deelopment and proessionalism, there is no eidence to link teaching
inputs and student outcomes in Pakistan.
14
Part o the problem is data. Publicly aailable data sets exist on public schools such as proincial Lducational
Management Inormation Systems ,LMIS, and one-o data sets on priate education by the lederal Bureau
o Statistics ,lBS,. But most research on education comes rom household data sets ,Pakistan Integrated
lousehold Sureys conducted by the lBS, which hae little usable inormation on teachers. Labor orce
sureys hae not been exploited in any depth.
loweer, the lack o data has not hampered the introduction o reorms` and changes in the teaching sector
that hae yet to be ealuated or their impact on student outcomes. Lxamples include creating a cadre o
Lnglish language teachers, the introduction o contract teachers, and large-scale, proince-wide teacher
training programs conducted by the Uniersity o Lducation. Numerous smaller programs conducted by
arious NGOs and donor initiaties that hae not been ormally analyzed. Understanding how goernment
teachers unction, what reorms are necessary, and where to go rom here requires a detailed study o the
existing system o teaching linked to student outcomes.
13
See loxby ,1996,, Glewwe, Ilias and Kremer ,2003,,Vegas ,2005, and reerences therein.
14
lor an example o Aga Khan Uniersity`s work, see
http:,,www.aku.edu,ied,raps,policydialogue,dialoguekeyedu,discussionreports,report1,index.shtml
60
1T G'A"@ 'D/ G1D/1DM@
"$< O<9O&#R$%: #() )<09O&#R$ : R&9S%Q<- 9S R.*Q%: #() R&%Z#7< 7<#:$<&- -$95 7$#7 7$<U #&< )( %
S&90 759 :90RQ<7<QU )%SS<&<(7 O&9.R-T H( R#R<&+ 7$< O9Z<&(0<(7 -<:79& %- 0.:$ *<77<& R9-%7%9(<) 79
)<Q%Z<& b.#Q%7U <).:#7%9( 7$#( 7$< R&%Z#7< -<:79&+ #() %( # $%O$QU <b.%7#*Q< 0#((<&T /9<- %7 #:7.#QQU
)9 -9X '() %S (97+ 5$#7 #&< 5< 79 Q<#&( S&90 7$< &<Q#7%Z< R<&S9&0#(:< )%SS<&<(:<- %( Q<#&(%(O
9.7:90<- *<75<<( 7$< R.*Q%: #() R&%Z#7< -<:79&-X
"$< 9Z<&#QQ )<09O&#R$%: R&9S%Q< 9S 7<#:$<&- )%SS<&- )�#7%:#QQU #:&9-- 7$< O9Z<&(0<(7 #() R&%Z#7<
-<:79&-T "<#:$<&- %( R&%Z#7< -:$99Q- #&< 09&< Q%W<QU 79 *< S<0#Q<+ U9.(O<&+ .(0#&&%<)+ #() S&90 7$<
Q9:#Q #&<#T
3. 1eacber. iv rirate .cboot.
are reaovivavtt, fevate, ,ovvger,
vvvarriea, ava frov tbe tocat area.
ligure 3.1 summarizes the
dramatic dierences in the
demographic proiles o
teachers in the goernment and
priate sector. 1he top let
corner shows that the gender
distribution o teachers-6
percent o priate school
teachers are emale compared
to only 43 percent in the
goernment sector. And the
teachers in priate schools are a
lot younger. 1he igure on the
top right corner shows that the
age distribution among priate
school teachers is highly
concentrated around 21, while the age distribution o goernment school teachers is more dispersed with an
aerage close to 40.
G%O.&< ]T;P ?&%Z#7< -:$99Q 7<#:$<&- #&< R&<)90%(#(7QU S<0#Q<+
U9.(O<&+ .(0#&&%<) #() S&90 7$< Q9:#Q #&<#g
43
57
76
24
0
20
40
60
80
P
e
r
c
e
n
t
a
g
e
o
f
T
e
a
c
h
e
r
s
o
f
G
i
v
e
n
G
e
n
d
e
r
Government Private
Teacher Gender
Female Male
Private School teachers are much
younger than Government teachers.
S
h
a
r
e
o
f
T
e
a
c
h
e
r
s
o
f
E
a
c
h
A
g
e
20 40 60 80
Teacher Age (Yrs.)
Government Private
Distribution of Teacher Ages
14.8
13.9
82.8
54.1
0
20
40
60
80
%
o
f
T
e
a
c
h
e
r
W
h
o
H
a
v
e
N
e
v
e
r
B
e
e
n
M
a
r
r
i
e
d
Government Private
Female Male Female Male
Teacher Marital Status
81
70
44
60
0
20
40
60
80
%
o
f
T
e
a
c
h
e
r
s
C
o
m
m
u
t
i
n
g
Government Private
Female Male Female Male
Teachers Who Don't Live in the Village Where They Teach
Commuting Teachers
61
3.8 As a consequence o the younger age proile, most teachers in priate schools, especially women, are
unmarried ,bottom let,. lurthermore, in the priate sector there is a large dierence in the percentage o
unmarried emale ersus male teachers ,83 percent s. 54 percent,, but in the goernment sector there is none
,both are just below 15 percent,.
J9V ]TYP K$#7 )#7# #&< &<b.%&<) S9& 7$%- S%&-7 Q99W #7 7$< 7<#:$%(O -<:79&X
1he LLAPS project deeloped questionnaires to collect detailed inormation on teachers in the sample
schools. 1he surey results proide a unique, detailed look at teacher proiles, competence, eort and reward
structures in Punjab. 1he teacher roster questionnaire collected basic inormation on salary, education,
training, experience and basic demographic inormation on eery single teacher in the sample schools. 1his
gies us data on 4,88 teachers, o which 4,825 are in goernment and priate schools and 2,826 are emale.
15
1his teacher roster captures the maximum breadth and diersity o the teacher inormation across
goernment and priate schools, as well as rich and poor illages.
3.9
3.10
1he detailed teacher questionnaire was
administered to the class teacher o the grade
tested in the LLAPS project ,typically one teacher
per school, and proides more extensie
inormation than the roster-or example these
teachers were administered tests as well 1he
teacher data is linked with student test scores on
specially designed tests or Class III children and
to the socioeconomic proiles o the children in
the classrooms.
Corervvevt ava rirate .cboot teacber. aiffer iv
rbere tbe, cove frov ava bor far tbe, traret to ror/.
According to data on the teacher`s illage o birth
and their current residence, a greater raction o
both emales and males in priate schools ,56 and 40 percent, respectiely, were born in the illage where
they currently work than in goernment schools ,19 and 30 percent or emales and males, respectiely,
,ligure 3.1, bottom right,. Priate school teachers who may not hae been born in the illage appear to
requently become part o the local community. ligure 3.2 looks how ar teachers lie rom the school they
G%O.&< ]TYP ?&%Z#7< -:$99Q 7<#:$<&- Q%Z< %( 7$<
Q9:#Q #&<#
35 35
25
5
64
27
7
1
0
20
40
60
%
T
e
a
c
h
e
r
s
w
i
t
h
C
o
m
m
u
t
e
o
f
G
i
v
e
n
L
e
n
g
t
h
Government Private
Distance of Teacher Commute
Less than 15 Minutes 15-30 Minutes
30 minutes-1 Hour More than 1 Hour
15
51 , 2, are in NGO schools. 1hey are excluded rom the discussion.
62
teach in-65 percent o priate school teachers lie within 15 minutes o the school while only 36 percent o
the goernment school teachers do so.
3.11
3.12
3.13
3.14
otb vea.vre. of tocat birivg ;rittage of birtb ava cvrrevt re.iaevce), ivcrea.e aravaticatt, ritb rittage titerac,. 1he
percentage o local teachers by place o birth increases rom 28 percent or the least literate illages to 48
percent or the most literate. 1he dierence is een larger i we look only at priate schools: 39 percent o
teachers are o local origin in the least literate illages while 66 percent are o local origin in the most literate
illage. 1he same pattern also emerges when we look at the distance traeled to school by teachers. 1he
percentage o teachers who lie within 15 minutes o the school increases rom 38 to 60 percent moing
rom the least to the most literate illages.
v .vvvar, tbe v.e of tocat teacber. iv rirate .cboot. ivcrea.e. aravaticatt, a. tbe .vt, iv tbe rittage-vea.vrea
tbrovgb rittage titerac,-ivcrea.e.. 1he aerage teacher in a priate school is a young, unmarried emale teacher
who is ery likely born in the illage where she is teaching and lies close to school. 1eachers in goernment
schools are older, married males who were neither born locally nor lie locally. lurthermore, or priate
schools, the use o locally resident teachers increases dramatically as the supply in the illage-measured
through illage literacy-increases, there is some eidence o an increase in goernment schools as well, but
the dierence is smaller.
H( 7$< *#-%- 9S 9*-<&Z<) :$#&#:7<&%-7%:-+ O9Z<&(0<(7 -:$99Q 7<#:$<&- Q99W 0.:$ *<77<& b.#Q%S%<)T
"$<U #&< 09&< <).:#7<)+ *<77<& 7&#%(<)+ #() $#Z< 09&< <VR<&%<(:<T "$<U #&< #Q-9 R#%) # Q97 09&<
7$#( 7$<%& :9.(7<&R#&7- %( 7$< R&%Z#7< -<:79&T
Corervvevt teacber. are vore eavcatea. 1he goernment sector ollows a strict and consistent hiring policy
or teachers. A large raction o goernment teachers is highly educated-19 percent hae a master`s degree
,MA, and another 26 percent hold a bachelor`s degree ,BA,. In contrast, only 4 percent o priate school
teachers hold a master`s degree, and 19 percent report a bachelor`s degree ,ligure 3.3, top let,. 1he male-
emale dierential in educational attainment in the goernment sector is also less pronounced than that in the
priate sector: 38 percent o emale teachers hold at least a BA compared to 51 percent or male teachers, the
equialent numbers or the priate sector are 1 and 39 percent.
1be aifferevce. iv eavcatiov ritbiv tbe gorervvevt .ector .tev target, frov aifferevt age cobort.. Oer time, the
educational qualiications required or joining the goernment teaching cadre has increased, so that younger
and newly hired teachers are more educated-in the latest wae o teachers hired on a contractual basis, more
than 93 percent reported a bachelor`s or higher degree. 1he lowest educated teachers in the goernment
63
sector are usually the oldest-matriculates or instance, report a median age o 40 compared to 35 years or
those with a bachelor`s degree.
3.15
3.16
3.1
1be rirate .cboot age
eavcatiov rofite i. qvite aifferevt
ritb age ava eavcatiovat
ovtcove. o.itiret, corretatea.
1he median age o a
matriculate is 22 years and
or more adanced degrees,
age increases rom 23 years
or completed secondary
schooling ,lA, to 25 years
or a bachelor`s and 29 or a
master`s degree.
Corervvevt teacber.
are better traivea. 1eacher
training is a goernment
requirement, only 6 percent
o goernment teachers
report no training`. 1he
bulk ,43 percent, holds a
Primary 1eaching Certiicate
,P1C, while the remaining 50 percent are diided between C1 ,that typically goes with lA, and the higher-
leel Bachelor o Lducation degree. More than one-ith o all goernment teachers in our sample report a
Bachelor o Lducation or higher degree. Lducational attainment among priate school teachers aries widely,
more than 0 percent report no training at all and only 14 percent report a Bachelor o Lducation or higher
degree ,ligure 3.3, top let,.
G%O.&< ]T]P M9Z<&(0<(7 7<#:$<&- #&< 09&< <).:#7<)+ *<77<& 7&#%(<)+ 09&<
<VR<&%<(:<)+ #() *<77<& R#%)
35.5
19.3
26.2
19.0
41.4
36.1
18.8
3.7
0
10
20
30
40
%
o
f
T
e
a
c
h
e
r
s
w
i
t
h
G
i
v
e
n
D
e
g
r
e
e
Government Private
Teacher Educational Qualifications
Matriculation and Below FA/FSc
BA/BSc MA/MSc
6.4
43.8
22.2
27.6
71.5
14.6
7.7
6.3
0
20
40
60
80
%
o
f
T
e
a
c
h
e
r
s
w
i
t
h
G
i
v
e
n
L
e
v
e
l
o
f
T
r
a
i
n
i
n
g
Government Private
Teacher Training
No training PTC/SV/JV
CT B.Ed and Above
6.4
5.2
88.4
21.2
39.0 39.8
0
20
40
60
80
%
o
f
T
e
a
c
h
e
r
s
w
i
t
h
G
i
v
e
n
E
x
p
e
r
i
e
n
c
e
Government Private
Teacher Experience
Less than 1 Year Teaching
1-3 Years Teaching
More than 3 Years Teaching
6185
5299
1250
1619
0
2,000
4,000
6,000
S
a
l
a
r
y
i
n
R
u
p
e
e
s
Government Private
Teacher Compensation
Actual Wages Adjusted Wage
Corervvevt teacber. are vore eerievcea. Not surprisingly gien the older age proile o goernment
teachers, 88 percent report three or more years o total teaching experience, with no dierence in experience
proiles between men and women.
16
1he situation in priate schools is quite the opposite: 21 percent were in
16
Our question on experience was categorical with three categories--1 year o experience, 1-3 years o experience and 3 years o
experience.
64
their irst-year o teaching eer and only 40 percent report three or more years o experience in the teaching
sector. lurthermore, women are generally less experienced than men-42 percent o the teachers who are
emale report 1-3 years o experience compared to 21 percent o men.
3.18
Corervvevt teacber. are better aia. linally, goernment school teachers are much better paid. 1he
bottom right corner o ligure 3.3 plots aerage
wages o goernment and priate school teachers.
1o control or the objection o comparing apples
and oranges-goernment teachers are paid more
simply because they are more educated and better
trained-salaries in the public and priate sector
were regressed on teacher characteristics to control
or potential dierences in education, qualiication,
and age. 1he igure shows both the unadjusted
wages, which is just the aerage in the two sectors,
and the adjusted` wages, which are wages in the
two sectors controlling or obsered characteristics.
It is immediately clear that priate teachers earn a lot
less than their goernment counterparts. 1he
unadjusted wages in the priate sector o Rs.1250 a
month are almost 5 times less than the goernment sector wage o Rs.6185. Although some o this dierence
can be attributed to dierences in teacher proiles, controlling or these dierences ,but retaining the
assumption that compensation .cbeve. are the same in both sectors, makes little dierence-teachers with
identical proiles are paid 3-4 times as much in the goernment compared to the priate sector. In particular,
a 25 year-old emale with a bachelor`s degree, 1-3 years o experience, and a 2-year teacher training course
residing locally ,thus controlling or age, gender, education, experience, training and residence, would earn
Rs.5299 in the public sector, but only Rs.1619 in the priate sector.
1
G%O.&< ]TcP "<#:$<& <).:#7%9(+ 7&#%(%(O+ 5#O<-
#() -7.)<(7_7<#:$<& %9- #&< Z<&U -%0%Q#& %( &%:$
#() R99& Z%QQ#O<-
25
29
26
0
20
40
60
80
100
P
e
r
c
e
n
t
a
g
e
o
f
T
e
a
c
h
e
r
s
Poor Middle Rich
Teachers with B.A.
20
16
21
Poor Middle Rich
Teachers with M.A.
90
96 97
Poor Middle Rich
Teachers with Any Training
Teacher Quality
5987
6356 6304
0
2,000
4,000
6,000
A
v
e
r
a
g
e
S
a
l
a
r
y
i
n
R
s
.
Poor Middle Rich
Teacher Compensation
39
34
35
0
10
20
30
40
S
t
u
d
e
n
t
/
T
e
a
c
h
e
r
R
a
t
i
o
Poor Middle Rich
Student/Teacher Ratios
Across Richer and Poorer Villages
Government Teacher Distribution
1
1he adjusted wages assumes that returns to characteristics are identical in the public and priate sector. \e return to this technical
issue below.
65
M9Z<&(0<(7 7<#:$<&- #&< <Z<(QU )%-7&%*.7<) #:&9-- R99& #() &%:$ Z%QQ#O<- #() R99& #() &%:$ -:$99Q-
3.19
3.20
3.21
1bere i. vo .,.tevatic aifferevce iv tbe tacevevt of gorervvevt teacber. iv rittage. of aifferevt .ocioecovovic teret.. A
political economy story, ot-repeated in the Pakistan case, suggests that less-educated teachers are placed
systematically in poorer illages that
hae less inluence in proincial
education departments. Len within
illages, it could be that schools with
richer children hae more
bargaining` power and hence get the
best teachers.
ligure 3.4 shows there is little
eidence that better teachers are
allocated to richer illages or schools,
at least on the basis o obsered
qualiications. ligure 3.4 uses the
LLAPS population census o all
illages in the sample to classiy
illages as rich, middle, or poor in
terms o household wealth-the same
classiication used to look at
enrollment in the introduction and the chapter on learning. 1he top panel o the igure looks at teacher
education and training in these three types o illages, it shows the proportion o teachers in the public sector
who hold a bachelor`s degree, a master`s degree, and who report some training. 1he bottom panel looks at
teacher wages and student-teacher ratios-the latter because it could be that the goernment allocates ewer
teachers to poorer illages. ligure 3.5 replicates the top and bottom panels o ligure 3.4, using school instead
o illage wealth-the school wealth measures are deried rom the asset index o 10 randomly selected
students in Class III o eery school.
G%O.&< ]T^P "<#:$<& <).:#7%9(+ 7&#%(%(O+ 5#O<- #() -7.)<(7_
7<#:$<& %9- #&< Z<&U -%0%Q#& %( &%:$ #() R99& -:$99Q-
29
24 24
0
20
40
60
80
100
P
e
r
c
e
n
t
a
g
e
o
f
T
e
a
c
h
e
r
s
Poor Middle Rich
Teachers with B.A.
18
21
17
Poor Middle Rich
Teachers with M.A.
92
95 95
Poor Middle Rich
Teachers with Any Training
Teacher Quality
6196 6271
6012
0
2,000
4,000
6,000
A
v
e
r
a
g
e
S
a
l
a
r
y
i
n
R
s
.
Poor Middle Rich
Teacher Compensation
36
38
35
0
10
20
30
40
S
t
u
d
e
n
t
/
T
e
a
c
h
e
r
R
a
t
i
o
Poor Middle Rich
Student/Teacher Ratios
Across School with Richer and Poorer Students
Government Teacher Distribution
Both igures show that, at least in the Pakistani context, illage and school wealth are not correlated
with teacher education leels, teacher training, teachers` wages or student-teacher ratios. Neither are they
correlated with teachers` wages or student-teacher ratios. Disaggregating teachers by gender or looking at
ariation in education within each sector does not change these results. Although these data are only or the
sample o illages with a priate school and thereore do not say much about smaller illages, there is
66
currently little eidence that more educated or better-trained goernment school teachers are sent to wealthier
illages or schools.
3.22
3.23
3.24
On paper, the goernment sector looks much better positioned to delier quality education than the
priate sector, and in a highly equitable manner. Goernment schools are staed by well-educated and trained
teachers, with lots o experience, the goernment also ensures an equitable distribution o teachers across
illages and schools. In contrast, the priate sector looks quite bad-teachers are inexperienced, less-
educated, and almost without exception, hae little teacher training. 1hey are also paid less, and turnoer is
requent. Schools must constantly recruit new teachers. 1he inputs approach to education would argue that
the goernment sector is doing eerything right and the priate sector is, in comparison, is doing less well and
has seere staing issues.
vt rirate .cboot. are .igvificavtt, ovterforvivg gorervvevt .cboot.. Inputs are important in that they are the
instruments through which preerred outcomes are achieed, but ultimately it is the outcomes that matter,
and here the rirate .ector is doing much better. ligure 3.6 reproduces the learning gaps across public and
priate schools rom the learning chapter, and also includes measures o parental satisaction with teachers in
goernment and priate schools.
18
As the igure .bor.,
priate schools are
signiicantly outperorming
goernment schools and that
ery little o this dierence is
attributable to dierences in
household wealth, parental
education, the child`s age or
the child`s gender.
lurthermore, parents rate
goernment schools ar below priate schools-on aerage, only 45 percent thought that the teaching skills
o goernment school teachers was aboe aerage or excellent. Parental satisaction with priate schools is
signiicantly higher at 60 percent. \hen asked about the oerall quality o the teacher, the results were similar.
G%O.&< ]TeP =<#&(%(O 9.7:90<- #&< S#& *<77<& %( R&%Z#7< -:$99Q-+ #()
R#&<(7- 7$%(W 7$#7 R&%Z#7< -:$99Q 7<#:$<&- #&< *<77<&
76
101
149
73
92
143
0
50
100
150
T
e
s
t
S
c
o
r
e
G
a
p
Actual Gap Gap Controlling for Differences in Students
Math Urdu English Math Urdu English
Amount Private School Students Out Score Gov't Students
Differences in Child Test Scores
0
20
40
60
P
e
r
c
e
n
t
a
g
e
R
e
p
o
r
t
i
n
g
A
b
o
v
e
o
r
B
e
t
t
e
r
Government School Private School
Parental Satisfaction with Public and Private Schools
Teaching Skills Overall Quality
18
1o look at parental satisaction with dierent types o schools, in the accompanying household surey parents were asked to report
their leels o satisaction with dierent types o schools in their illage. In particular, we asked parents to tell us what they thought o
the teaching skills o their children`s teacher as well as the oerall quality o the school. 1he igure shows the raction reporting that
teaching skills and oerall quality were aboe aerage` or excellent`.
67
3.25
3.26
3.2
3.28
v terv. of tbe eavcatiov roavctiov fvvctiov, ove veea. to accovvt for teacber effort iv aaaitiov to teacber t,e ava
covev.atiov. I test scores are worse in goernment schools and parents are less satisied, what accounts or
this \hen asked or more detailed iews parents do not ault goernment teachers or low education or
poor qualiications, rather they say that their children`s goernment teachers are not motiated`, do not
care about the children` or are almost neer there`. It may be that teacher eort matters as much, i not
more, than teacher competence in the production o learning. In addition, we also need to think about
teacher compensation: Gien that eort is an important component o teacher quality, do we ind dierences
in how public and priate schools reward eort monetarily 1owards such an analysis, the last part o this
section proides a detailed decomposition o goernment and priate teacher salaries, and in particular the
relatie contribution o dierent teacher attributes in the compensation package.
M9Z<&(0<(7 7<#:$<&- <V<&7 Q<-- <SS9&7 7$#( 7$<%& R&%Z#7< :9.(7<&R#&7-T "$< )%SS<&<(:<- O<7 Q#&O<& #-
7<#:$<&- *<:90< 09&< <VR<&%<(:<) #() S9& 590<( 5$9 Q%Z< S#&7$<& S&90 7$< -:$99QT
1o measure eort, we recorded the number o days absent in the last one month or each teacher as
reported by the head-teacher.
19
In addition, or 800 detailed teacher interiews, teachers were asked to rate
themseles on absenteeism. 1his was urther broken down into absenteeism arising rom emergencies,
personal reasons, or oicial business. 1hese results almost surely underestimate the extent o true`
absenteeism in the system, and it is ery likely that they underplay the dierence between the goernment and
priate sectors. lurthermore, goernment head-teachers may hae had reasons to beliee that high
absenteeism recorded in the surey could result in some kind o oicial backlash ,although all surey results
are stripped o identiiers and teachers were inormed that all responses were anonymous,, priate head-
teachers do not ace this incentie to underreport.
.b.evteei.v i. cov.iaerabt, bigber iv gorervvevt .cboot. tbav iv rirate .cboot.. In goernment schools the
absentee rate is 3.2 days per month s. 1.8 days per month or priate schools ,this translates to 15 and 8
percent rates o absenteeism, respectiely,. It`s possible the high degree o teacher accountability in the
priate sector accounts or this dierence.
.b.evce rate. for vore rer.v. te.. eerievcea teacber.. 1here is no dierence in absenteeism between public
and priate teachers ,1.9 days a month, with less than one year o experience. As shown in ligure 3.,
19
Lort is hard to measure in large-scale sureys without detailed classroom obseration o teacher`s going about their daily routine.
A plausible indicator o eort is teacher absenteeism ,see or instance, recent work by Chaudhury and Dilip ,2006, or an analysis o
absenteeism in India and Ghuman and Lloyd 200, or absenteeism in 12 illages in N\lP and Punjab,.` It`s likely that teachers who
are absent rom class are less eectie. Preious studies report teacher absenteeism using an audit approach, that is, by arriing
unannounced at the school and taking a roll-call o all present teachers. Since we needed to interiew teachers and they needed to be
present at time o testing we could not rely on random checks, and used head-teachers` reports o teacher absenteeism as the primary
measure. Das, Dercon, Habyarimana, and Krishnan (2005) discuss the diIIerent measures and their relative merits.
68
howeer, the story is dierent or more experienced teachers. Goernment school teachers with more than
three years o experience are absent 3.4 days a month while teacher absenteeism in priate schools remains
unchanged or those who are more experienced.
J9V ]T]P /9<-(d7 7$< O9Z<&(0<(7 #Q&<#)U $#Z< # 09(%79&%(O -U-7<0 %( RQ#:<X
1he goernment does hae an elaborate system o monitoring and management in place at the district leel,
with District Lducation Oicers aided by deputies and assistants. 1he schools report that they are isited
regularly by the district education sta, in act, 66 percent o all schools were isited by a school inspector in
the last three months and i anything poorer illages were isited more oten by the school inspectors than
richer illages. 1here is some weak eidence that schools that are isited more hae lower absenteeism but
that is not robust to the deinition o the time period, and runs into complicated issues o whether the isits
teaa to lower absenteeism, or whether inspectors just choose to isit schools where absenteeism is lower to
begin with.
'*-<(7<<%-0 #() '.)%7%(O C<O.Q#&%7U
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
Percentage
oI Schools
Mean Number oI
Monthly Absences
0-1 Month Ago 43.20 2.49
2-3 Months Ago 23.12 2.94
4-6 Months Ago 11.56 2.68
7-12 Months Ago 9.53 3.70
~ 1 Year Ago 11.97 2.22
Never 0.61 1.67
3.29
.b.evteei.v aifferevtiat. betreev fevate. ava vate.. As ligure 3. shows, emales are absent 3.9 days a
month in the goernment sector compared to 2. days or males. Again, there is no dierence in absenteeism
between men and women in priate schools ,1.82 days a month on aerage,. In the Pakistani enironment
where emales cannot moe about reely and transportation woes are requent, the literature has oten cited
transport problems and other responsibilities in the household as contributing to the increased absenteeism
o emale teachers. I this is the case teachers who lie urther rom work should hae higher absenteeism
rates and these problems should aect the goernment sector more, since only one-third o emale teachers
in goernment schools lie within 15 minutes o the school, compared to 66 percent o those in the priate
sector. In act, emale teachers in goernment schools who lie more than hal-an-hour away rom home are
absent 43 percent more ,3.6 days s. 2.56 days, than emale teachers in goernment schools who lie less
than 15 minutes away.
20
All these results also hold in a multiple regression ramework ater controlling or
teacher education, training, gender, experience, illage origin, school type, and illage ixed eects.
20
1he distance to school ariable is rom the teacher questionnaire so has a smaller sample size.
69
3.30 .b.evteei.v rate. va, be vvcb bigber iv tbe gorervvevt .ector becav.e of aaaitiovat re.ov.ibititie.. Part o this
higher absenteeism clearly
has to do with
accountability issues in the
goernment sector, but
part o it may also be due
to the non-teaching duties
that goernment school
teachers are oten asked to
undertake. Only 3 percent
o priate school teachers
were absent or work-
related reasons in the last
month, compared to 26 percent in the goernment sector. \hile accountability is a serious issue in the
goernment sector, oicially sanctioned absenteeism rom class is equally serious rom the students` point o
iew.
G%O.&< ]TfP G<0#Q< 7<#:$<&- #() 09&< <VR<&%<(:< 7<#:$<&- #&< 09&<
#*-<(7 %( 7$< O9Z<&(0<(7 -<:79&a*.7 (97 %( 7$< R&%Z#7< -<:79&
1.9
2.1
3.4
1.9
1.8
1.8
0
1
2
3
4
D
a
y
s
A
b
s
e
n
t
L
a
s
t
M
o
n
t
h
Government Private
< 1 year 1-3 years >3 years < 1 year 1-3 years >3 years
By Teaching Experience
Teacher Absenteeism
3.9
2.7
1.8
1.8
0
1
2
3
4
D
a
y
s
A
b
s
e
n
t
L
a
s
t
M
o
n
t
h
Government Private
Female Male Female Male
By Teacher Gender
Teacher Absenteeism
J9V ]TcP '))%7%9(#Q C<-R9(-%*%Q%7%<- S9& M9Z<&(0<(7 "<#:$<&-
One big problem is the extra responsibilities that goernment teachers ace oer and aboe their
regular teaching duties. I came across the ollowing example in our ield-work. \hen I went to a
Goernment Primary School in one o our sample mauza ,illage, there was only one teacher present.
\hen I spoke with him, he told me that there are only two teachers in the school and the other one
had been called by the Deputy District Lducational Oicer ,DLO,. 1he teacher complained that there
is lot o extra work gien to them by the goernment. Sometimes it is a meeting in the LDO`s oice,
sometimes the Deputy DLO calls them, sometime they are on election duty, others on examinations
duty and sometimes een on sureys conducted by the goernment.
^or tett ve, bor cav teacb tbe rbote .cboot. rrote to tbe gorervvevt .a,ivg tbat tbere re veea
vore teacber. bvt tbere ra. vo ret,. triea to tat/ to tbe C ava .O, bvt tbe, .aia tbat baa
girev tbe rrittev aticatiov ava .bovta vor rait becav.e gorervvevt ror/ cavvot be aove
qvic/t,. ^or ,ov tett ve-atreaa, tbe tro teacber. are vot evovgb ava ritb att tbe.e etra
re.ov.ibititie. re cavvot gire roer attevtiov to tbe cbitarev. Yov bare to a./ tbe Corervvevt
tbat to gire v. ove vore teacber ava if tbe, cav`t gire v. a teacber tbev tbe, .bovta vot gire v. etra
re.ov.ibititie.. Ovt, tbev re cav gire roer attevtiov to tbe .cboot or tbe cbitarev.
ovrce: Kashi
70
11T MH6>CDI>D" 'D/ ?C16'"> @>A"HC AHI?>D@'"1HDP 1D?B"@ HC
HB"AHI>@X
3.31
3.32
3.33
ovr iece. of aata bare beev brovgbt togetber for tbe fir.t tive iv tbe .P .vrre, to eavive teacber erforvavce
ava it. retatiov.bi to covev.atiov iv vbtic ava rirate .cboot.. Student test-scores, teacher absenteeism, teacher test-
scores and teacher`s salaries in the public and priate sector can be tiv/ea: to show, or instance, whether there
is a correlation between student test scores and teacher salaries or absenteeism and teacher salaries and i so,
whether this correlation diers between the public and priate sectors. \e highlight each o these our
correlations next and then discuss the oerall compensation schemes in the two sectors.
@#Q#&%<- S9& O9Z<&(0<(7 -<:79& 7<#:$<&- )9 (97 Z#&U 5%7$ <SS9&7 9& 9.7:90<-T 1( 7$< R&%Z#7< -<:79&+
7$<U )9T
ligure 3.8 shows the deiation rom
aerage salary or teachers based on absenteeism in
public and priate schools. As absenteeism
increases in the priate sector, salary goes down-
teachers with low absenteeism earn more ,close to
Rs.100 aboe the aerage salary, while those with
high absenteeism earn less. 1his relationship is
reersed in the goernment sector, where those
with low absenteeism report salaries Rs.400 betor
the mean in their sector, and those with the highest
absences receie salaries Rs.600 abore tbe veav.
Otaer teacber. are ab.evt vore ava tbe, are at.o
aia tbe vo.t. 1hese results hold in a multiariate
regression controlling or the age, gender, education, training and residence o the teacher, as well as the
geographical location o the illage ,illage ixed eects,. In priate schools, a teacher who is absent 5 days a
month is paid 5 percent less, in goernment schools a teacher is paid 3 percent more! One potential
explanation or these results may be that the teachers with higher absenteeism in the goernment sector may
hold more senior posts` such as a head or acting head. ligher posts carry higher salary terms, and these may
also be the teachers who tend to be more absent, whether because o oicial duties or other reasons.
G%O.&< ]T`P I9&< #*-<(7 7<#:$<&- #&< R#%) Q<-- %(
7$< R&%Z#7<+ *.7 (97 %( 7$< O9Z<&(0<(7 -<:79&
-400
-200
0
200
400
600
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
0 10 20 30
Days Absent per Month
Private Schools Government Schools
Teacher Absenteeism and Compensation
While in Private Schools,
absenteeism is penalized.
n Government Schools, the most absent
teachers are the highest earners.
-400
-200
0
200
400
600
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
0 10 20 30
Days Absent per Month
Private Schools Government Schools
Teacher Absenteeism and Compensation
While in Private Schools,
absenteeism is penalized.
n Government Schools, the most absent
teachers are the highest earners.
71
3.34
3.35
1he LLAPS project tested Class III teachers
on the curriculum they are supposed to teach in
Mathematics, Lnglish, and Urdu. ligure 3.9 shows
the relationship between teacher test scores and
salaries. As in the preious igure, the ertical axis
represents deiations rom the aerage salary in the
sector, the horizontal axis in this case represents the
percentage o questions the teacher answered
correctly on the test. As beore, better scores on the
test or priate school teachers were associated with
higher salaries-a teacher who scored 60 percent on
the test receies Rs.600 less than the priate sector
mean, while a teacher scoring 95 percent receies
Rs.200 vore than the mean. In the goernment
sector, teachers who scored 60 percent receied Rs.200 more than the mean. 1he only saing grace is that
teachers who scored highly were not penalized as much as in the absenteeism igure, reporting salaries only
Rs.100 below the mean. 1aken together, these data imply that the least competent teacher in the goernment
sector earns the same salary as the most competent teacher in the priate sector.
G%O.&< ]ThP L%O$<& -:9&%(O 7<#:$<&- <#&( 09&< %(
R&%Z#7< -:$99Q-
-600
-400
-200
0
200
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
700 800 900 1000 1100
Teacher Knowledge Test Scores
Private Schools Government Schools
Teacher Test Scores and Compensation
Again, the results hold in a multiariate regression context, albeit with a ew dierences in unctional
orm. Across the test-score range, a priate school teacher in the top 20 percent reports a salary that is 28
percent higher than one in the bottom 20 percent. loweer, priate schools are particularly responsie to
poor and exceptional teacher test scores, with little dierence in compensation or aerage teachers. Moing
rom the third to the ourth quintile o test scores does not increase salary, but moing rom the irst to the
second quintile o test scores increases salaries by 8 percent and rom the ourth to the ith quintile by 9
percent. In the goernment sector, the relationship between test scores and compensation is generally weak
and somewhat supportie o the representation in ligure 3.9 aboe.
72
3.36
3.3
3.38
ligure 3.10 presents a third association, in
this case between student and teacher test scores.
As beore, the ertical axis is deiation rom
aerage salary in the sector. 1he horizontal axis is
the aerage test score o the students taught by the
teacher. In the priate sector, teachers o students
with higher test scores are paid more. In the
goernment sector, there is no relationship
between student scores and teacher compensation,
at worst, those at the bottom o the distribution
are paid somewhat more.
1hese three igures highlight what is
already airly well known about goernment
compensation schemes-that they reward
experience and education in teachers ,see or instance, Vegas ,2005, or a discussion o Latin America or
Kingdon, Geeta and 1eal ,2004,, and Murgai and Pritchett ,2006, or India,. It also adds new inormation
about compensation in the priate sector. lor teachers in priate schools, eort ,as measured through lower
absenteeism,, competence ,as measured through their own test-scores,, and student achieement ,as
measured through children`s test-scores, are all rewarded with higher salaries. 1hus, although the goernment
system o education proides teachers who are more experienced, more educated and better trained, it seems
the lack o accountability and mechanisms or rewarding better perormers may lead to lower eort.
Compensation schemes in the priate sector seem to reward competence and eort, in the goernment
sector, better perormers can actually end up receiing torer salaries. 1his hints at the next and inal step,
which looks at the oerall compensation schemes in the two sectors and a discussion on how to structure a
debate about teacher recruitment, deployment, and compensation.
G%O.&< ]T;iP "<#:$<&- 5$9-< :$%Q)&<( R<&S9&0
*<77<& <#&( $%O$<& -#Q#&%<- %( 7$< R&%Z#7<+ *.7 (97
%( 7$< O9Z<&(0<(7 -<:79&
-200
-100
0
100
200
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
0 200 400 600 800
Student Test Scores
Private Schools Government Schools
And Student Test Scores
Teacher Compensation
Private School
teachers whose
students perform well
earn more.
While Government Salaries
do not reflect student
performance.
-200
-100
0
100
200
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
0 200 400 600 800
Student Test Scores
Private Schools Government Schools
And Student Test Scores
Teacher Compensation
-200
-100
0
100
200
S
a
l
a
r
y
B
o
n
u
s
o
r
P
e
n
a
l
t
y
(
i
n
R
s
)
0 200 400 600 800
Student Test Scores
Private Schools Government Schools
And Student Test Scores
Teacher Compensation
Private School
teachers whose
students perform well
earn more.
While Government Salaries
do not reflect student
performance.
"$< O9Z<&(0<(7 -U-7<0 &<5#&)- 7<#:$<&- )%SS<&<(7QU S&90 7$< R&%Z#7< -<:79&T
Ov arerage, gorervvevt teacber. are aia fire tive. a. vvcb a. tbeir rirate .ector covvterart.. 1his inding
echoing indings rom many other countries. 1o re-emphasize the point, we reproduce the unadjusted and
adjusted salaries o goernment and priate school teachers, where the adjusted salaries control or education,
experience, gender, qualiications, and training. 1he adjusted salary gap is somewhat smaller, but still
represents a three-old adantage or the goernment sector.
73
3.39
3.40
3.41
One problem is the adjusted wage calculation in the
two sectors assumes that the goernment and priate
sectors reward the same teacher characteristics in the same
way. 1hat is, a teacher with training receies the same
additional wage in the goernment sector as she does in the
priate sector. In act, as the relationship between eort and
wages suggests, the compensation unction in the two
sectors looks quite dierent. 1o examine how the dierent
sectors reward dierent characteristics, we regress ,log,
wages on teacher characteristics separately or public and
priate schools. Seeral noteworthy contrasts are
summarized in ligure 3.12.
G%O.&< ]T;;P M9Z<&(0<(7 7<#:$<&- <#&( #
Q97 09&<g
6185
5299
1250
1619
0
2,000
4,000
6,000
S
a
l
a
r
y
i
n
R
u
p
e
e
s
Government Private
Teacher Compensation
Actual Wages Adjusted Wage
atarie. iv tbe gorervvevt .ector are target, aetervivea b, eerievce, traivivg ava eavcatiov. In the priate sector,
education matters, but the premium to experience and teacher training is much smaller. lor example, in
goernment schools a 50 year-old teacher earns 0 percent more than a 25 year-old, in the priate sector, the
premium decreases dramatically to only 20 percent. 1he rewards to teacher training are equally dramatic. In
the goernment sector, the minimum leel o teacher training ,P1C, increases salaries by 88 percent, while in
the priate sector, such a teacher would earn only percent more than one without any training. In contrast,
education matters equally in both sectors, and perhaps een somewhat more in priate schools.
Prirate .cboot. a, tbeir teacber. accoraivg to bor vvcb
tbe, covta earv et.erbere rbite tbe gorervvevt .ector va/e. vo
aa;v.tvevt for attervatire evto,vevt oortvvitie.. As a result,
priate schools pay women and local teachers
considerably less. lemales in goernment schools actually
earn a little more than men ,3 percent, while in priate
schools they earn 36 percent less. In general, employment
opportunities or women are limited and because it is
diicult or them to trael outside the illage they lie in,
they hae ewer outside opportunities and they are paid
less in the priate sector. Local teachers are actually paid
less under both systems, but the dierence is much larger
in priate schools-the penalty is 5 percent in
G%O.&< ]T;YP ?.*Q%:m?&%Z#7< K#O< ?&<0%.0-
#() ?<(#Q7%<-
2.0
-28.1
-20.5
113.8
2.0
-5.8
-50
0
50
100
150
W
a
g
e
P
e
n
a
l
t
y
o
r
P
r
e
m
i
u
m
(
i
n
R
s
.
)
Private Government
Wage Premiums and Penalties for Different Characteristics
Differences in Teacher Compensation
Teacher with Some Training
Female Teacher
Local Teacher
74
goernment compared to 24 percent in priate schools. Again, residence in the illage means that trael costs
are lower and the teacher ,most o whom are women, enjoys the saety and conenience o working where
she lies. She is willing to accept a lower salary than a teacher who has to commute rom a neighboring
illage.
3.42
3.43
3.44
Gien these dierences in how the two sectors reward teachers, the pay dierential between the
teachers in the goernment and priate sectors becomes clearer. ligure 3.13 illustrates the compensation
outcome o moing a teacher with certain characteristics rom the public sector and to the priate sector by
using the estimated coeicients rom Column 1 in 1able 5. 1he igure plots the aerage pay o a teacher in
the public sector ,the irst bar,, a teacher in the priate sector ,the second bar, and a teacher in the priate
sector i he,she had the same characteristics as teachers in the public sector. 1he dierence between the irst
and the third bar is the dierence in salaries arising rom dierential returns in the two sectors, the dierence
between the second and the third bar is the dierence due to characteristics.
. teacber`. .atar, rovta aecrea.e
aravaticatt, if .be rere to vore frov tbe
vbtic .ector to tbe rirate .ector. 1he
aerage salary or the public sector
teacher alls rom Rs.5620 to Rs.165
or three reasons: the priate sector
does not alue teacher training ,which
the public sector does,, it does not
compensate experience to the same
leel as the public sector, and it pays
emale teachers a lot less. Only the
remaining dierence between the
salary that aerage public school
teacher would receie and the aerage
salary in the priate sector, Rs.1084 s.
Rs.165, comes rom the act the
aerage public school teacher is better trained and educated than the aerage priate school teacher.
G%O.&< ]T;]P 6<&U Q%77Q< 9S 7$< 5#O< )%SS<&<(:< *<75<<(
O9Z<&(0<(7 #() R&%Z#7< -:$99Q- %- *<:#.-< 9S )%SS<&<(:<- %(
:$#&#:7<&%-7%:-
5620
1084
1765
0
2,000
4,000
6,000
S
a
l
a
r
y
(
i
n
R
s
.
)
Government
Teacher in a
Government
School
Private Teacher in
a Private School
Government
Teacher in a
Private School
Between Government and Private School Teachers
Decomposition of Wage Gap
Wage Difference Explained by
Differences in Qualifications
Wage Difference Not Explained
by Differences in Qualifications
5620
1084
1765
0
2,000
4,000
6,000
S
a
l
a
r
y
(
i
n
R
s
.
)
Government
Teacher in a
Government
School
Private Teacher in
a Private School
Government
Teacher in a
Private School
Between Government and Private School Teachers
Decomposition of Wage Gap
Wage Difference Explained by
Differences in Qualifications
Wage Difference Not Explained
by Differences in Qualifications
1be.e fact. .vgge.t tbat tbe gorervvevt .ector care. vore abovt ob.errea teacber ivvt., .vcb a. eavcatiov ava traivivg,
rberea. tbe rirate .ector care. vore abovt teacber effort ava .tvaevt ovtcove.. Priate sector teachers are paid according
to local job market conditions-those with better options are paid more. 1he labor market or goernment
75
teachers responds to dierent signals, particularly the need to reward eeryone equally depending on
education, qualiications and training, but little else. 1he lexibility to adjust wages depending on local
conditions or teacher eort in the priate sector is absent rom goernment sector compensation schemes.
\e are not saying att goernment teachers are bad` and att priate school teachers are good`. As the
chapter on learning pointed out, the top goernment and priate schools perorm at the same leel. 1he
problem is the bottom 20 percent o goernment schools where little to no learning goes on at all. Part o the
problem is that goernment compensation schemes do not distinguish between good and bad teachers. In
act, compensation when linked to experience only results in more absenteeism, less-educated teachers ,recall
that older teachers are also less educated,, and poorer learning outcomes or students.
111T /1@AB@@1HD
3.45
3.46
3.4
1he goernment sector works well in some regards, ery poorly in others. It hires educated and well-
trained teachers and pays them well. It also allocates them airly across both rich and poor illages and
schools with rich and poor students. On the other hand, it has no mechanism or penalizing non-perorming
teachers and rewarding exceptional ones. 1he current mechanistic compensation scheme results in the highest
compensation being directed towards the worst teachers, in terms o education, training, or eort. 1eachers
are also burdened with duties outside the puriew o their main responsibilities.
1he priate sector compensates teachers or outcomes, and high turnoer permits retention based on
merit, but priate sector schools locate only in areas where teachers are aailable at a low wage. It is
exceedingly diicult or priate schools to arise in areas without educated women or in the rural areas with
secondary schools, where urther education is required. 1eachers receie wages according to their aailable
local opportunities-they are paid less in illages where there is higher literacy ,and thereore greater supply,,
and emales with these characteristics earn the least.
Oerall some priate sector characteristics may be desirable` in goernment compensation schemes,
others, such as gender discrimination in wages clearly iolate goernment rules and may be undesirable.`
Suggestions or how can Pakistani can best take adantage o these two ery dierent sectors and
compensation schemes requires either rethinking the stage o education and the geographical locations the
goernment sector should ocus on or a large, systematic reorm o the goernment compensation system.
76
C<7$%(W%(O 7<#:$<& &<:&.%70<(7+ )<RQ9U0<(7+ #() &<5#&)-
3.48
3.49
1he debate around teachers in goernment schools seems to hinge on two related issues: the
attributes o teachers and the institutional ramework that contribute to better student outcomes, and the
aailability o teachers or schools catering to dierent leels o education in each illage ,primary, middle, and
secondary,. 1eacher and institutional attributes can be broadly separated into three categories: bara to ob.erre
teacber cbaracteri.tic. such as motiation, which can emerge only oer time, ea., to ob.erre characteristics such as
educational qualiications, experience and training and, the institutional ramework embodied in ivcevtire. such
as the teacher salaries and bonuses. Research in the United States has tried to separate the inluence o the
irst two types o characteristics ,motiation and qualiication,, gien that most o this research is or public
school teachers, it has made less progress on the impact o incenties. 1his research inds that characteristics
like motiation and a loe o teaching are far more important in explaining the ariation in student learning
compared to educational qualiications, experience, and training. Lxperience or instance, matters only in the
irst year. In short, in systems with the same set o incenties, teachers appear to be borv, not made.
\hen there are many potential teachers to hire rom, it makes sense to try and recruit and retain the
best applicants and eentually ire those who do not perorm. loweer, in geographical areas or leels o
schooling where potential applicants are ew, this is no longer true. liring a teacher makes sense only i you
can replace him,her with another teacher who is better. In areas o limited supply, a teacher who is absent
three days a week may still be a good hire compared to the alternatie o being teacher-less. 1he key point is
that these dierent actors ivteract-proiding incenties or teachers mean that those with higher motiation
,and thus better outcomes, are more likely to apply, thus increasing the quality o the teacher pool. In act, in
the United States studies hae argued that the quality o the teacher pool has suered because o teacher
unionization, which pays a ixed wage regardless o the motiation o the teacher ,see loxby, 1996,. Some
options and their potential interactions in terms o student outcomes are discussed next.
77
'( <V#0RQ<P "$< R9R.Q#& 5%-)90 7$#7 j%(:&<#-%(O 7$< <).:#7%9(#Q b.#Q%S%:#7%9(- #() 7&#%(%(O 9S
(<5 <(7&#(7- %- # 0.-7k :9.Q) Q<#) 79 59&-< <).:#7%9(#Q 9.7:90<-
3.50
3.51
3.52
1his proposed policy option
assumes ea., to ob.erre teacher
characteristics drie student learning.
Do they ligure 3.14 plots the
Lnglish test scores ,results are ery
similar or Mathematics and Urdu,
o children against the Lnglish test
score o their teacher. 1he red line
shows the releant relationship or
public schools and the blue line or
priate schools. linally, the two red
dots show the test scores o teachers
with a secondary education ersus
those with a bachelor`s degree-the
teachers with secondary education
scored 80 percent on the test
compared to 90 percent or the
latter.
1he small gap on the
ertical axis suggests that hiring a teacher with a bachelor`s degree instead o secondary education in the
public sector results in a gain o roughly 1-2 percentage points on student test-scores. Contrast this with the
dramatic dierence o 19 percentage points between teachers with secondary education in the public ersus
the priate sector. 1he igure suggests that increases rom additional education are small and that eort, not
education accounts or the dierence, at least at the primary leel.
G%O.&< ]T;cP '7 7$< R&%0#&U Q<Z<Q+ %7- 7<#:$<& <SS9&7 7$#7 0#77<&-+
(97 7<#:$<& <).:#7%9(
400
450
500
550
600
650
C
h
i
l
d
E
n
g
l
i
s
h
K
n
o
w
l
e
d
g
e
T
e
s
t
S
c
o
r
e
s
700 800 900 1000 1100
Teacher English Knowledge Test Scores
Private Schools Government Schools
Teacher Education and Child Learning
Increase in chiId test scores
from getting the same
secondary educated
government schooI teacher to
behave Iike a private schooI
teacher.
Increase in chiId test
scores from repIacing a
teacher with a secondary
education with a teacher
with a BA.
Average Teacher with
Secondary Education
Average Teacher
with a BA
400
450
500
550
600
650
C
h
i
l
d
E
n
g
l
i
s
h
K
n
o
w
l
e
d
g
e
T
e
s
t
S
c
o
r
e
s
700 800 900 1000 1100
Teacher English Knowledge Test Scores
Private Schools Government Schools
Teacher Education and Child Learning
Increase in chiId test scores
from getting the same
secondary educated
government schooI teacher to
behave Iike a private schooI
teacher.
Increase in chiId test
scores from repIacing a
teacher with a secondary
education with a teacher
with a BA.
Average Teacher with
Secondary Education
Average Teacher
with a BA
Gien the eidence gien aboe, such a policy could cause particular problems in areas with limited
aailability. 1o the extent that the aerage illage does not hae indiiduals educated beyond the secondary
leel ,the median illage in Punjab had 8 secondary-school educated women in 1998,, teachers holding a
bachelor`s degree will hae to be brought in rom outside the illage. As we hae seen, absenteeism ivcrea.e.
when teachers are not local hires, so oerall eort is likely to decrease. Moreoer there is an oerall skill-
78
shortage in teaching the Pakistani economy. ligher goernment wages may attract a teacher with a bachelor`s
degree, but not necessarily a better teacher.
3.53
3.54
3.55
Increasing teacher training will probably not hurt in the same way as increasing the educational
requirements or new teachers, but good training is expensie. In act, it is too expensie an inestment or
rural priate schools. 1he Ali Institute, or instance, runs a teacher training course o three weeks that costs
Rs.3000 per teacher. It`s possible this amount could be better spent elsewhere.
'( #Q7<&(#7< 9R7%9( %- 79 :&<#7< #( <(#*Q%(O <(Z%&9(0<(7 S9& *<77<& Q<#&(%(OT "$&<< R97<(7%#Q S<#7.&<-
9S -.:$ # -U-7<0 59.Q) %(Z9QZ< [#\ )<:&<#-%(O 7$< #))%7%9(#Q ).7%<- S9& 7<#:$<&-n [*\ %0RQ<0<(7%(O #
R&9*#7%9(#&U #--<--0<(7 R<&%9) *<S9&< 7<#:$<&- #&< &<O.Q#&%3<) [R<&$#R- :90*%(<) 5%7$ %(:<(7%Z<
R#U0<(7- S9& 7<#:$<&-\+ #()n [:\ #QQ9:#7%(O 7<#:$<&- 79 7$9-< O<9O&#R$%:#Q #&<#- #() -:$99Q%(O Q<Z<Q-
5$<&< 7$< R&%Z#7< -<:79& %- (97 # R&<-<(:<+ R<&$#R- %( :90*%(#7%9( 5%7$ )<Z9QZ%(O 7<#:$<& $%&%(O #()
:90R<(-#7%9( 79 7$< )%-7&%:7 Q<Z<QT
et teacber. teacb. 1he goernment needs to decide what it wants its teachers to do. Burdening them
with additional duties and time consuming tasks detract rom the job o teaching. In a typical illage setting
the teacher may be the only representatie o the state and may be called upon to take a census or electoral
rolls. Indeed, work-related absences or goernment teachers ,though sel-reported and uneriied, are quite
common: 26 percent o teachers in the public sector reported some work-related absence in the last month
compared to barely 3 percent in the priate sector. 1he reasons ranged rom attending workshops and
meetings, collecting salaries, being on examination duty in other schools, and administering polio
accinations. 1eachers absent due to emergencies was
closer in both sectors-32 percent in the goernment and
26 percent in the priate sector. 1he incenties on teaching
are being conused with proiding related goernment
serices.
Cov.iaer a robatiovar, erioa to iaevtif, ava retaiv gooa
teacber.. 1he only way to igure out who has talent as a
teacher who does not is to obsere them or some period
o time. A reasonable amount o turnoer should be
expected. \et, the public sector is a settled` sector with
ery little turnoer. More than 0 percent o goernment
teachers in the LLAPS sample report hae more than three
years experience at the school they are teaching compared to 29 percent in priate schools. In 1 percent o
G%O.&< ]T;^P "<#:$<& ".&(9Z<& #()
'*-<(7<<%-0
3.3
4.7
2.9
4.0
1.7
2.7
1.5
1.8
0
1
2
3
4
5
A
v
e
r
a
g
e
N
u
m
b
e
r
o
f
D
a
y
s
A
b
s
e
n
t
L
a
s
t
M
o
n
t
h
Government Private
Bad School Good School Bad School Good School
Teacher Turnover and Absenteeism
Retained Teachers Non-Retained Teachers
79
priate schools, a teacher let in the preious year compared to 24 percent in the goernment. Oerall, in the
priate sector, 530 teachers let in the year preceding the surey and 690 new teachers were hired out o a
total o 2,16 present in the surey year. In the goernment sector, 11 let, and 356 new ones were hired out
o a total o 2,652 teachers. ligure 3.15 suggests that turnoer, both in the public and priate sector, may be
good in that it gets rid o teachers who exert less eort. As the igure shows, whether we look at initially high
or low perorming schools, teachers who were not retained the ollowing year were those who were more
absent than the others.
3.56
3.5
3.58
A middle ground thereore might be a probationary period o two to three years during which the
teacher is kept on a temporary contract. At the end o this period, the teacher may be conerted to a
permanent contract through a clearing process. 1he process or retention could eole. Indeed, the Punjab
goernment reeze on regular hiring o teachers in 2002 led to an increase in teachers hired on temporary
contracts-although this was a response to a iscal issue tied in to pension payments rather than as a means
o improing accountability. 1hese contract teachers may work, but only i ,a, the goernment does not
renew contracts or poorly perorming teachers and ,b, ultimate political pressures do not lead to their
regularization` ,there are already agitations towards this step,.
1he data on contract teacher perormance rom the LLAPS study shows that contract teachers are
absent less oten, are more competent, and are paid less. 1his analysis is a bit misleading because contract
teachers are also a considerably younger than the aerage teacher in public schools. A correct comparison
would control or age, but this cannot be done in a simple regression context because there is no ariation-
all contract teachers are young, all public school teachers are older. In the uture a more sophisticated analysis
using regression discontinuity designs will be used.
!bat abovt reraraivg teacber. for greater effort. \hile ew disagree with the oerall premise o rewarding
teachers or eort, the question o how best to do so is a subject o much debate. It is diicult to
institutionalize a mechanism or rewarding eort - ater all, how is eort measured Current literature
suggests it will be necessary to deelop policies that combine dierent means o measuring eort, rom
inputs` like attendance to outcomes` like child test scores, and possibly een the process` by which
teachers teach. In recent small-scale randomized ealuations, rewarding teachers based on inputs or outcomes
hae both been shown to work, see Dulo and lanna`s experiment o using cameras and compensation
linked to days present or an example o the irst and Muralidharan and Sundararaman ,2006, on
compensating teachers based on improements in test scores or the second. Rewarding teachers strictly on
objectie ovtcove. is diicult. 1he objection is somewhat technical. 1est scores display strong mean
reersion-which means, that in any gien year, the worst perormers at the beginning will show the
80
strongest gains and the best perormers the weakest. 1hus, teachers who start with a poor class will get
rewarded irrespectie o what they do, those who start with a strong class will not get rewarded irrespectie o
what they do. 1he set o teachers who can change their compensation depending on their eort may be ery
small so that the scheme turns out to be ery costly. Once teachers understand this process, gaming the
system combined with outright cheating in tests will become perasie ,the experiment aboe was tried only
or one year,. Because o similar problems, eery state in the United States has at some time tried and
discarded this strategy. An alternatie, adanced by Murgai and Pritchett ,2006, is to reward teachers not on
eriiable outcomes, but on eriiable processes. Under this scheme, the hiring and retention o teachers
would be decentralized to the illage leel. \et, this also comes with its own problems. Decentralization may
or may not work in the Pakistani context. It would depend on issues o elite capture` and how democratic
and participatie the ultimate process turns out to be. In the state o Uttar Pradesh in India, or instance, a
recent surey reealed that een vevber. o the Village Lducation Committee` are not aware o their
membership! 1he attractieness o this proposal lies in it promise o giing power to the ultimate receiers o
a serice, but there is little eidence on whether such a scheme could or would work.
3.59 .ttocate gorervvevt teacber. rbere tbe, are veeaea tbe vo.t. One o the most powerul tools or equity that
the goernment holds in its hands is the right to transer teachers to illages and schools where they are
needed most. 1he inherent dierences in salaries between the priate and public sector is explored in the
chapter on schooling. It will show that educating a child in a goernment school requires trice the resources
needed to educate that same child in a priate school when one is aailable. It makes little sense, i we are
concerned about the right to education o the poor in Pakistan, or the public sector to covete with the
priate sector when both choices are aailable. Matters are made worse because the goernment sector
competes not only or enrollment, but also or teachers in a limited labor market. Priate school teachers
desire jobs as public school teachers. \hen asked, one such teacher said: .va rovta ,ov vot ta/e a ;ob rbere ,ov
are aia 1 tive. a. vvcb ava ao vot bare to ror/.` \et, in many illages there are no priate schools. In areas or
sectors where the supply o educated potential teachers is low, it will be diicult or priate schools to hire
and retain good teachers. 1he data show that priate schools are likely to unction better as literacy increases
and that locally hired and resident teachers hae lower absenteeism. As illage literacy increases, the
percentage o local teachers increases-in priate schools, 39 percent are local in the least literate illages
while 66 are local in the most literate ,or goernment schools it increases rom 28 to 48 percent,. Similarly,
the percentage o teachers who lie within 15 minutes o the school goes up rom 38 percent to 60 percent
when we go rom the least literate to the most literate category. Priate schools will ovt, unction in sectors
and geographical areas where supply o teachers is suicient.
81
3.60
3.61
3.62
Gien these supply issues one possibility may be to deelop 2 dierent cadres` o teachers: a
proincial cadre that can be transerred across districts and a district cadre that can be transerred across
illages. 1hese cadres would proide education where the priate sector is reluctant to go. 1hey would also
justiy part o their high salaries because o the inherent diiculties o working in the places where they are
posted. Although problems o accountability and incenties will remain, at least these teachers will be
proiding some education where none was preiously aailable.
\hile some deolution or education has shited rom proinces to districts, teacher compensation
and posting is still a proincial responsibility. As we hae seen aboe, this does not permit dierent districts
to take adantage o local conditions-where there are many educated indiiduals, wages should be lower.
Allowing districts to come up with their own hiring, retention, and compensation policies would take the
deolution process urther and help strengthen local accountability mechanisms. At the same time, recovr.e to
a centralized cadre ,at a ixed cost per teacher, would ensure that they hae access to trained personnel i
needs cannot be met locally.
O roughly 12 million employees in the goernment workorce, 3 million are teachers and they hae
strong unions, as do teachers around the world. Moreoer, as a hangoer rom the British era, where teachers
were the only educated indiiduals or miles on end, illage teachers may be asked to man election booths,
draw up oter lists, and work as part and parcel o the political system. A democratic debate on this issue
must go beyond eorts to covete with the priate sector, and ocus instead on proiding somewhat lower
quality education in areas where the priate sector is absent, and gradually withdraw proision ,but not
inancing, as the priate sector takes oer. \idespread systemic reorm is needed beore the current cadre o
teachers retires since it will take close to 25 years to reresh at least 50 percent o the teaching workorce. 1he
question is whether Pakistan can aord to mortgage the uture o 2 million children eery year or the next 25
years to the desires o 3 million teachers.
82
Chapter 4: Parents and their Children
4.1 . it trve tbat arevt., rbo va, bare tor eavcatiovat attaivvevt tbev.etre., vv.t be ca;otea ava ivcevtiriea ivto
.evaivg tbeir cbitarev to .cboot. 1he media and goernmental policy suggest that the reality or children, een
when enrolled, is that their ability to participate and learn is seerely hampered by multiple demands on their
time, either through housework or paid child-labor. 1he girl-child is less likely to be enrolled in school in the
irst place, and when she is her learning suers because o the added burden o work at home and low
parental attention, both in terms o money spent and time gien. Popular wisdom is that the emergence o
rural priate schools may hae made matters worse still. Illiterate parents, unable to gauge the quality o
priate schools, it is conjectured, are easily ooled by unscrupulous priate school operators into paying
unnecessarily high ees. 1hese perceptions hae clear implications or educational policy. 1he goernment
must work hard to get children into school, compensate or the inability o parents to und or spend time
with their children, legislate against child-labor and regulate the priate sector. Leaing illiterate parents to
end or themseles is not the best educational inestment in the next generation o citizens. \hat do the
data say about parental participation in the education process
4.2 Particiatiov i. ovt, art of tbe .tor,. Researchers in Pakistan and elsewhere bare actiely explored the link
between households and schools. loweer, the ocus has been on actors that lead to higher enrollment. In
the Pakistani context, gien the importance o the distance to school and its interplay with gender, there is a
considerable literature on the impact o distance rom school on attendance, more so or girls than or boys.
21
As the chapter on learning shows, once children get to school ensuring learning outcomes is an entirely
separate issue. Upping enrollment is undoubtedly an important irst-step, but it is now time to think about
how to bolster their learning and expand our understanding about the role o households in supporting the
learning process.
4.3 1be rote of tbe bov.ebota iv rovotivg better eavcatiovat ovtcove. i. eqvatt, ivortavt. 1his chapter presents
detailed inormation on the choices that parents make regarding their children`s schools, the time and money
they spend on their children, the daily actiities that a child engages in, and reiterates the critical constraints
that distance to school has or enrollment, especially among girls. \hile some indings rom the surey data
accord well with some o the commonly held belies mentioned aboe, other indings suggest a closer
ealuation o the Pakistani rural household and its role in educating their children is in order. As a teacher
21
Alderman, Behrman, Khan, Ross and Sabot ,1995,, lolmes, Jessica. ,2003,, Lloyd, Mete and Sathar ,2005,,
Sawada and Lokshin ,2001,.
83
said to one o us recently 1be cbita i. iv .cboot bovr. a aa, ava at bove 1. f arevt. ao vot a, attevtiov, rbat cav
tbe cbita tearv.` 1his chapter looks at how parents both enhance and hamper their children`s educational
outcomes. 1he results show that households are ery dierent in what they want or their children, and that
dierent children within the same household are treated ery dierently. 1he chapter concludes with a
discussion o how these indings can enhance the educational policy debate.
1T G'A"@ 'D/ G1D/1DM@
4.4 . t,icat arevt robabt, goe. tbrovgb tbe fottorivg tbovgbt roce.. rbev cov.iaerivg tbe eavcativg tbeir cbitarev.
1hey look or an appropriate school, gather inormation about dierent schools in their illage in terms o
distance, ees, quality and other attributes important or them, decide which school to send their child to, i
they decide to enroll their child at all, and then determine how much time and money to spend on their
child`s education and how much time a child will deote to school work, household chores, and other
actiities. 1hese decisions are not taken one ater the other in the ashion presented here, but are intrinsically
inter-related and play out oer time. lor instance, a household may decide against enrolling a child under any
circumstances, in which case they may not look at the dierent schools in their illage. Similarly, the
household may decide which school to send their child to depending on their assessment o the inputs in
time and money they are likely to spend on their child. 1his model is not meant to sere as an explanation`
o the data presented here, but rather to proide a context or the speciic acts and indings that ollow.
?#&<(7- ?<&:<R7%9(- 9S F.#Q%7UP A$%Q)&<(+ "<#:$<&- #() @:$99Q-
4.5 Covtrar, to ovtar betief, arevt. /vor a tot abovt bor tbeir cbitarev are erforvivg, bor gooa tbeir teacber. are,
ava bor gooa tbe .cboot. iv tbeir rittage. are. 1he LLAPS household surey asked the ollowing questions o athers
and mothers, separately:
" Rate each o your children on a scale o 1 ,ery poor, to 5 ,excellent, or intelligence and eort ,how
hard-working are they,.
" Rate each o your children`s teacber. on perormance and attendance ,regularity, using a similar scale.
" Rate each o the .cboot. in the illage. lor eery school in the illage parents were asked whether they
had heard o the school, and i yes, how good they thought the school was.
4.6 1be .P .vrre, ae.igv attor. v. to vatcb tbe.e arevtat ercetiov. to ob;ectire ovtcove.. Because the LLAPS
surey tested a large number o children or whom household sureys were also completed, we can match the
responses o the households to the test scores o more than 800 children in Class 3. \e are also able to
84
match parental assessments o the child`s teachers and schools to the test scores o children taught by the
teacher and to the test scores o the Class 3 children in eery school.
4. ov.ebota. are rettivforvea abovt tbe erforvavce of tbeir cbitarev, tbeir cbitarev`. teacber., ava tbeir cbitarev`.
.cboot.. 1he ollowing our igures demonstrating the extent o household inormation, all ollow the same
pattern. 1he horizontal axis plots what the household said. 1he ertical axis plots the test scores in the
independently administered test, whose results had not been disclosed to households at the time o the
surey. Depending on the igure, these are the test-scores o their child, other children taught by the child`s
teacher, or children in the school that the parents ranked.
4.8 Re.vtt 1: Cbitarev erceirea a. vore ivtettigevt
b, tbeir arevt. .core bigber ov te.t.. ligure 4.1 plots
household reports o child intelligence against
the child`s actual test score or illiterate and
literate mothers ,the results or athers are
similar,. \hen households say that a child is
ery intelligent, he,she reports test scores that
are much higher than when the household says
that a child is not intelligent. 1he dierences are
large and signiicant-a child who is perceied
as less intelligent reports test scores close to 0.
standard deiations lower than a child perceied
to be intelligent. 1be re.vtt. ao vot aeeva ov rbetber
tbe arevt. are titerate. Illiterate mothers are as
good as literate mothers in iguring out which o
their children is intelligent and hard-working and which is not. linally, actors other than learning are not
associated with parental assessments o child intelligence: lor eery additional standard deiation increase in
the child`s aerage test score there is a 13 percentage point increase in the probability that the mother
perceies her child as highly intelligent, and a 10 percentage point increase in the probability that she
perceies her child as perorming highly`. It is particularly interesting to note that neither age nor gender are
statistically signiicant, suggesting that mothers do not discriminate among older or younger children or
among boys and girls.
G%O.&< cT;P A$%Q)&<( ?<&:<%Z<) #- 1(7<QQ%O<(7 @:9&<
L%O$<& 9( "<-7-
-100 -50 0 50
Combined Subjects Knowledge Score
L
i
t
e
r
a
t
e
l
l
i
t
e
r
a
t
e
Above Average
Average
Below Average
Above Average
Average
Below Average
For Literate and lliterate Mothers
Mother's Perception of ntelligence and Child Test Scores
85
4.9 Re.vtt 2: !bev arevt. .a, a .cboot i. gooa, it v.vatt, i.. ligure 4.2 shows a similar pattern in way
households` perceie school perormance. lere, we plot or both athers and mothers the actual Lnglish
score o tested children in the school against what parents told us they thought the quality o the school was
,poor`, aerage` or aboe aerage`,. 1he results are also separated by mothers and athers who can read
and those who cannot. ,1he choice to show results or Lnglish test scores is based on the expectation that
parents will be less able to assess the quality o teaching in Lnglish, which may or may not be true., \hen
households eel that a school is aboe aerage, the Lnglish test scores o children in that school are much
higher than that o schools that households eel are poor. Again, illiterate parents are as good at judging
school perormance as literate ones. And women are good as men in iguring out which schools perorm well.
Figure 4.2: Parental Perceptions of School Quality are Accurate
-20 0 20 40
Combined Subjects Knowledge Score
L
i
t
e
r
a
t
e
l
l
i
t
e
r
a
t
e
Above Average
Average
Below Average
Above Average
Average
Below Average
For Literate and lliterate Mothers
Mother's Rankings and School Test Scores
-40 -20 0 20 40
Combined Subjects Knowledge Score
L
i
t
e
r
a
t
e
l
l
i
t
e
r
a
t
e
Above Average
Average
Below Average
Above Average
Average
Below Average
For Literate and lliterate Fathers
Father's Rankings and School Test Scores
J9V cT;P '&< $9.-<$9Q) #--<--0<(7- 9S -:$99Q- *#-<) 9( 9*-<&Z#*Q< :$#&#:7<&%-7%:- $<&
7$#( 7<-7 -:9&<-X
Part o the result aboe could be drien by parents judging schools on attributes otber than test scores
but still correlated with the perormance o children in the school. lor instance, i parents ranked all
priate schools as good and all public schools as bad, similar correlations would arise. As it turns out,
the relationship between household ranking and test scores o children in the school remain
signiicant after covtrottivg for att ob.errabte cbaracteri.tic. of tbe .cboot.
Parents are indeed more likely to rate a school highly i it is priate ,26 to 31 percentage points,. In
addition, obsered characteristics o schools increase the probability that parents rate the school
highly, in particularly, the number o teachers increases the probability by 1 percentage point and
better inrastructure by 1.2 percentage points. loweer, test scores continue to matter most: a one-
standard deiation in the combined subject test scores increases the probability that a school is ranked
highly by more than 2.6 percentage points, twice as high as any other obserable characteristic other
than whether the school is public or priate. 1hese results tell us not only that households place alue
on test scores and rate schools where children are perorming better higher, but at.o that literacy and
the gender o the parent do not matter or the ability to judge schools and that households are not
taken in` by obserable characteristics o schools. Schools may hae better inrastructure and more
teachers, but at the end, it`s the quality o instruction that inluences household perceptions.
86
4.10 Re.vtt : Motber. /vor be.t. ligure 4.3
relates households` perceptions o teacher
attendance ,regularity, measured at the school as
part o the teacher questionnaires and roster
exercise ,see Chapter 3,. Again maternal
assessment o the teachers` regularity tracks the
actual state o aairs well. Literate mothers do a
slightly better job ,especially in iguring out
which teachers are not regular, than literate
athers. Combining this with results 1 and 2,
mothers are more aware o school conditions
despite, as we will see next, haing isited ar
ewer schools than athers.
4.11 Re.vtt 1: v a t,icat rittage .ove .cboot. are
vv/vorv to arevt.. ligure 4.4 is the ovt, set o results that resonate, at irst glance, with the common wisdom
that households are relatiely ignorant, and illiterate and less wealthy households know less. Mothers hae
beara of only 60 percent o the schools in their illage, and athers 0 percent, and illiterate mothers and
athers hae heard o een ewer schools. Seeral salient points emerge through a multiariate regression
analysis: both mothers and athers are ar more likely to report haing heard o a school that is public rather
than priate. 1he public adantage` translates into a 22 percentage point increase or athers, 1 percentage
points or mothers, resulting in a 20.6 percentage point increase or household responses that combine both
the parents` inormation. At the leel o the school, the size o the school as measured by the number o male
and emale teachers correlates strongly with the probability o being recognized ,or male, emale, and
household regressions,, while eery additional year that a school has been in the illage increases the
probability o recognition by 0.2 to 0.3 percentage points ,males and emales, respectiely,. At the leel o the
household, being rich, haing educated parents and being close to a school increase the probability that any
school is recognized by parents. lor eery kilometer that the school is urther rom the house, the probability
o hearing about it decreases by percentage points or athers and 9 or mothers, presumably relecting the
limited mobility o women within the illage.
Figure 4.3: Mother Knows Best
0
2
4
6
8
lliterate Literate lliterate Literate
Fathers Mothers
Not Regular Regular Very Regular
D
a
y
s
A
b
s
e
n
t
i
n
D
a
t
a
C
o
l
l
e
c
t
e
d
f
r
o
m
S
c
h
o
o
l
s
By Parental Literacy
Parental Assessment and Teacher Absenteeism
8
G%O.&< cTcP >Z<( %( # Z%QQ#O<+ R<9RQ< $#Z< (97 $<#&) 9S # Q#&O< (.0*<& 9S -:$99Q-
0 .2 .4 .6
Percent Schools Heard Of
Literate
lliterate
For Literate and lliterate Mothers
How Many Schools Have Mothers Heard Of
0 .2 .4 .6 .
Percent Schools Heard Of
8
Literate
lliterate
For Literate and lliterate Fathers
How Many Schools Have Fathers Heard Of
4.12 1hese results suggest that ignorance about schools, to the extent that it hampers the ability o
households to choose appropriately or their children, is limited ov tbe arerage to household`s knowledge o the
existence o certain schools. Len here, it is easy to imagine scenarios where this ignorance has no impact on
actual outcomes. lor instance, i the schools that the household has not heard o are those which would
neer be chosen ,perhaps they are too ar,, lack o knowledge does not imply that households are making
uninormed decisions. 1he aerage household is thus reasonably accurate in their perceptions o how well
their child is doing, how well their child`s teacher is doing, and how well dierent schools in the illage are
doing. 1he common wisdom that mothers and illiterate parents know less is limited, at most, to their
awareness o certain schools.
"$< )<:%-%9( 79 <(&9QQ # :$%Q)
4.13 1be bov.ebota .vrre, .borea tbat betreev tbe age. of : to 1:, ;v.t orer ovetbira of tbe cbitarev iv tbe .avte ;:
ercevt) rere vot evrottea iv av, .cboot. 1he data are consistent with the patterns obsered in the Pakistan
Integrated lousehold Surey ,PIlS,. In particular, girls are less likely to be enrolled than boys, and there is a
amiliar pattern o delayed enrollment, whereby enrollment irst increases with age, peaks between the ages o
9 and 10, and subsequently declines. 1here is a hint o a greater decline among girls compared to boys
between the ages o 10 and 15, but the additional decline or emales is neither large nor statistically
signiicant. Close to 30 percent o the enrolled children are in priate schools ,2.3 percent,, a number not
ery dierent rom that in the PSLM or rural areas o the proince. linally, as expected, enrollment is higher
among wealthier households and household with more educated parents. 1hese acts are well known and the
LLAPS data are similar to those obtained rom larger sureys.
88
4.14 1be .P .vrre, ivrore. ov P evrottvevt aata iv tbree ra,.. lirst, the LLAPS surey measured
distances rom households to schools using global positioning systems ,GPS,. lor eery child in the sureyed
household, there is data on the distance not only to the school he,she attends, but to erer, o..ibte .cboot iv tbe
rittage. 1his is an improement oer the PIlS where the questionnaire assumes a single school in the illage
and proides a single number or distance between home and school or eery household-a strategy that can
yield erroneous results when there are 10-12 schools in eery illage. Second, the surey recorded the speciic
name o the school so that the association between distance and enrollment can be linked to inormation
about the school the child attends. 1hird, as described in the preious section, the surey included questions
or parents on their perceptions o the child`s intelligence, work ethic, and school perormance. 1hese
additional bits o inormation lead to some surprising results. Relating enrollment to parental perceptions o
intelligence yields important insights into how parents allocate educational inestments among children in the
same household. And new indings ,relatie to what is well known rom analysis o standard household
surey data, emerge when these new distance measures and inormation on parental perceptions o the child`s
intelligence are used to better understand enrollment and school choice decisions.
22
4.15 Di.tavce ta,. a va;or rote iv tbe aeci.iov to evrott a cbita, ava vore .o for girt.. ligure 4.5 presents the
relationship between enrollment and the distance to the closest eligible school that the child can attend,
dierentiated by gender. 1he let axis presents a histogram o the distance measure or the children in the
sample. 1he right axis shows the enrollment or boys and girls as a unction o the distance measure.
23
1he
distance measure was computed by using GPS coordinates to measure the distance between eery household-
school pair ,close to 50,000 such pairs in the data,, by using inormation on whether a school is co-
educational ,all priate school, or single-sex ,all goernment schools are either or only boys or only girls, to
assess whether a child is etigibte to attend the school ,or instance, girls are not eligible to attend a boys` school,
een i it is much closer than the alternatie,, and by combining the eligibility and distance data to compute
the distance to the closest eligible school.
22
Parental perceptions o the child`s intelligence could be sel-reinorcing, in that, they inest in the child they tbiv/ is more intelligent,
and the child then perorms better, making the parental belies sel-ulilling. Neertheless, these perceptions do present an added
dimension to understanding intra-household allocations across children. lor instance, parents who inest more in children they tbiv/
are more intelligent still suggests that parents are not ollowing a policy o compensating or weaker children in their educational
inestment decisions.
23
1he associations are computed in a non-parametric ashion using locally weighted linear regressions. Since these non-parametric
measures tend to be highly sensitie to areas with ew obserations, the itted cures are trimmed` at the 95
th
percentile o the
distance measure. 1hat is, the associations are plotted using the ull sample data, but omit obserations in the top 5 percent rom the
graph.
89
4.16 .cro.. tbe tbree ai.trict., :0
ercevt of cbitarev are ritbiv 200 veter. of
tbe cto.e.t etigibte .cboot ava tbe arerage
ai.tavce of a cbita frov bi.,ber cto.e.t etigibte
.cboot i. :0 veter.. 1he histogram in
ligure 4.5 shows the distribution o
distance rom school or children in
the LLAPS surey. 1he ast majority
o children are within 500 meters o
the closest eligible school. In seeral
illages we walked this distance with
boys and girls between the ages o 8
and 12, and in no case did it take us
more than 10 minutes. Children lie
closest to schools in laisalabad
,median 150 meters and mean 220
meters, and urthest in Rahim \ar
Khan. Len in Rahim \ar Khan
though, 50 percent o all children lie within 400 meters rom the closest school ,the aerage distance is 550
meters,. At irst glance, low enrollments arising because children lie ar rom school does not appear be an
issue.
Figure 4.5: I9-7 -:$99Q- #&< 5%7$%( ^ii 0<7<&- 9S 7$<
$9.-<$9Q)gU<7 <(&9QQ0<(7 S#QQ- )�#7%:#QQU 5%7$ )%-7#(:<
#() 09&< -9 S9& O%&Q-
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0
5
10
15
P
e
r
c
e
n
t
o
f
A
l
l
S
c
h
o
o
l
s
0 .5 1 1.5
Distance to Closest School (KM)
Percent of all schools Girls Boys
Distance and Enrollment
4.1 rev ritb tbe retatiret, .vatt ai.tavce. to .cboot, tbe ai.tavceevrottvevt retatiov.bi i. .trovg, e.eciatt, for girt.: tbe
ai.tavce evatt, accovvt. for 0 ercevt of tbe gevaerga iv evrottvevt.. 1he relationship between enrollment and
distance is strongly negatie and more or less linear. Across boys and girls, enrollment declines by 8-10
percentage points or eery 500 meters that a school is urther rom the house, and the relationship is as
strong moing rom 0 to 100 meters as rom 800 to 900 meters. As expected, the relationship is ,much,
stronger or girls than or boys, in act, the drop in enrollment as the distance to the closest eligible school
increases is 3-4 times as much as that or boys-4 percentage points or eery 500 meters or boys and 11-16
percentage points or girls. In a multiariate regression context that controls or age, household expenditures
,a measure o income,, education and the intelligence o the child, the distance penalty` or boys is urther
reduced to 1.5-3 percentage points, or girls, the eect o distance is still large at 9-11 percentage-points or
eery 500 meters. \hile household and child characteristics, such as household expenditures and the child`s
age, all hae independent eects on enrollment they do not alter the size o the basic gender gap o 15-16
90
percentage points in enrollment in the LLAPS data. In sharp contrast, allowing or the distance-enrollment
relationship to dier across boys and girls reduces the pure eect o gender on enrollment to 5 percentage
points-a dramatic decline o 60 percent.
4.18 Ove ra, to a..e.. rbat oticie. vigbt ivcrea.e fevate evrottvevt i. to a./ rbetber tbere are av, bov.ebota or cbita
cbaracteri.tic. tbat reavce tbe ai.tavceevatt,. lor instance, i girls rom richer households are able to attend
schools arther away this is an indication that with more money, parents will be able to pay or transportation
costs. \e examine the eects o two household characteristics ,the maximum education o an adult emale in
the household and the wealth o the household, and two child characteristics ,age and intelligence, on the
distance penalty. 1he analysis examines a policy question: in the absence o new school construction, how
many households can be induced to send their children, particularly girls, to schools that are arther away
4.19 1be evrottvevtai.tavce
retatiov.bi for aifferevt t,e. of cbitarev
ava aifferevt t,e. of bov.ebota. i. v.ea
to aaare.. tbe otic, qve.tiov. ligure
4.6 presents the results or girls.
As in ligure 4.5, the relationship
between enrollment and distance
to the closest school or girls is
plotted, but now or dierent
sub-groups. 1he top-let graph
looks at the relationship or girls
perceied to be aboe aerage`
in intelligence and aerage or
below aerage intelligence`, the
top-right graph looks at girls aged
5-12 and those aged 13-15 ,post-
menarche,, the bottom let by
whether the mother has any
schooling or not and the bottom
right by household wealth ,poor or non-poor is deined such that 50 percent o all households all into each
o the two categories,. Lach graph coneys the teret eect o the categorization ,the extent to which, say,
intelligent girls are more likely to be enrolled than those who are less intelligent, and the graaievt eect o the
categorization ,the extent to which, say, intelligent girls are more likely to be sent to schools arther away,.
G%O.&< cTeP H(QU <).:#7<) #).Q7 S<0#Q<- &<).:< 7$< j)%-7#(:<_
R<(#Q7Uk S9& O%&Q-
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Above Average Average or Below
By ntelligence
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Ages 5-12 Ages 12-15
By Age
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
No Schooling Some Schooling
By Mothers Education
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Poor Non-Poor
By Household Wealth
Distance and Enrollment For Girls
91
4.20 or girt., ovt, tbe re.evce of eavcatea aavtt fevate. iv tbe bov.ebota reavce. tbe ai.tavce evatt,. Girls liing in
households with educated emales or in richer households are all more likely to be in school, as are those who
are more intelligent and those in the younger age groups. 1hese dierences are large and signiicant,
accounting or an increase in enrollment o between 15-20 percentage points ,the dramatic eects o
intelligence on enrollment will be discussed urther on,. loweer, apart rom the presence o an educated
emale in the household, vove o the other characteristics reduce the distance-penalty or girls. A priori, one
might hae thought richer households could and would obtain appropriate transport ,or chaperone serices,
or that post-pubescent girls will be less likely to trael arther, but there is .cavt eriaevce tbat ai.tavce ta,. a targer
rote for girt. frov oorer favitie. or for teevage girt. covarea to otber.. Girls in households where there is an adult
emale with .ove education ,30 percent o all households, are enrolled een when the school is arther away,
and the result is airly large and signiicant in a regression context.
4.21 v tbe ca.e of bo,., botb
eavcatea aavtt fevate. ava arevtat
ercetiov of tbe cbita`. ivtettigevce
reavce tbe ai.tavce evatt,. ligure
4. shows the same set o
relationships or boys. As
beore, child intelligence,
younger ages, adult emale
education and greater wealth are
all signiicantly associated with
higher enrollment, albeit with
two notable dierences. lirst,
consistent with ligure 4.1, the
oerall gradients are less
steep-distance matters less or
boys. Second, adult emale
education plays less o a role in
reducing the distance eect, but
parents are more willing to send
boys they perceie to be more intelligent to schools arther away. As beore, household wealth and the age o
the child has little eect on the enrollment-distance gradient.
G%O.&< cTfP D97$%(O &<).:<- 7$< j)%-7#(:<_R<(#Q7Uk S9& *9U-
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Above Average Average or Below
By ntelligence
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Ages 5-12 Ages 12-15
By Age
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
No Schooling Some Schooling
By Mothers Education
.2
.4
.6
.8
1
%
E
n
r
o
l
l
e
d
0 .5 1 1.5
Distance to Closest School
Poor Non-Poor
By Household Wealth
Distance and Enrollment For Boys
92
"#*Q< cT;P L9.-<$9Q) #() A$%Q) A$#&#:7<&%-7%:-+ >(&9QQ0<(7+ #() 7$< /%-7#(:< M&#)%<(7
J9U- M%&Q-
=%W<Q%$99) 9S
>(&9QQ0<(7
>SS<:7 9S /%-7#(:<
9( >(&9QQ0<(7
=%W<Q%$99) 9S
>(&9QQ0<(7
>SS<:7 9S /%-7#(:<
9( >(&9QQ0<(7
A$%Q) 1(7<QQ%O<(:< Increases
Reduces the
distance penalty
Increases No Change
A$%Q) %( 7<<(#O<
U<#&-
Decreases No Change Decreases No Change
L9.-<$9Q)
K<#Q7$
Increases No Change Increases No Change
>).:#7<) ').Q7
G<0#Q< %(
L9.-<$9Q)
Increases No Change Increases
Reduces the
distance penalty
4.22 Cbitarev vore ti/et, to be evrottea if tbe, cove frov ricber bov.ebota., are vore ivtettigevt ;accoraivg to tbeir arevt.),
ava are reteev.. 1he enrollment analysis is summarized in 1able 4.1. In essence, children rom richer
households, more intelligent children, and those in the pre-teen years are all more likely to be enrolled,
whether they are boys or girls. loweer, in the case o girls, child attributes do not play a role in decreasing
the detrimental eects o distance, the only statistically signiicant result is that the distance-penalty is lower
or girls in households with an educated emale. lor boys, child intelligence plays an important role in
attenuating the distance penalty.
"$< -:$99Q :$9%:< )<:%-%9(
4.23 1be qve.tiov of .cboot cboice at tbe bov.ebota teret i. tiv/ea to vvaer.tavaivg bor otic, cbavge. vigbt affect evrottvevt
iv aifferevt t,e. of .cboot.. 1he next section looks at the public-priate school choice and, in particular, the
eects o distance, household income and expenditure on the choice o a priate school. It then examines the
eidence on how parents are choosing schools. 1wo main indings emerge. lirst, poorer students attend
public schools partly because they are closer. Second, dierent households care about dierent things in
choosing schools. lor instance, despite the dramatic eects o distance on enrollment, 50 percent o enrolled
boys and girls ao vot go to tbe cto.e.t etigibte .cboot. School choice depends on whether households are quality
conscious, distance conscious, or price conscious.
"#*Q< cTYP A$%Q)&<( :Q9-<& 79 R&%Z#7< -:$99Q- #&< S&90 &%:$<& #() 09&<
<).:#7<) $9.-<$9Q)-
AQ9-<-7 >Q%O%*Q< @:$99Q %-P
lousehold Characteristics
?&%Z#7< ?.*Q%:
l all Children 53.14 46.86
Mother's Lducation ,\ears, 1.40 0.828
lather's Lducation ,\ears, 3.06 2.93
Per Capita Lxpenditure 8.933 659.394
PCA \ealth -0.199 -0.404
Median distance to closest eligible school 0.16 0.2
93
4.24 Chapter 2 on the schooling enironment showed that priate schools tend to locate in richer
settlements and public schools are also ound in peripheral areas o the illage, where households are poorer.
1able 4.2 conirms this typology using household data-children who are closest to a priate school tend to
come rom richer and more educated households, and trael shorter distances to school.
4.25 ligure 4.8 shows how location patterns relate to the wealth segmentation obsered in the data,
whereby children in priate schools were 1.2 standard-deiations richer on aerage than those in public
schools. It plots the percentage o children enrolled in priate schools against ,log, monthly household
expenditure or two groups o children-those whose closest eligible school is public, and those whose
closest eligible school is priate. 1he histogram gies some sense o the distribution o ,log, household
expenditures.
4.26 Pvbticrirate evrottvevt aeci.iov. .bor a ctear
atterv. lirst, there is a dramatic increase o 10
percentage points ,30 percent, in the probability o
using a priate school regarate.. of reattb when the
priate school is closer than a public school. A child
in a household with monthly expenditures o
Rs.3000 whose closest school is priate is as likely to
enroll in a priate school as a child in a household
with monthly expenditures o Rs.15000 whose
closest school is public. Second, the likelihood o
enrolling in a priate school as household income
increases is unaected by whether the closest school
is public or priate. 1his implies that een i eery
child in the illage had equal access to a priate
school, there would still be a considerable dierence
in the household incomes o children in public and priate schools. Predictions rom a regression model
suggest that i priate schools were the closest option or att children, enrollment in these schools would
increase rom 31 to 34 percent or the rich and 24 to 29 percent or the poor. Part o the percentage point
gap between the rich and the poor is because o the speciic location choices o priate schools, but een i
priate schools were to locate in all settlements, the gap would remain at 5 percentage points.
G%O.&< cT`P /%-7#(:< 79 # R&%Z#7< -:$99Q $#- #-
0.:$ 79 )9 5%7$ 7$<%& .-< #- 7$< 5<#Q7$ 9S 7$<
$9.-<$9Q)
.1
.2
.3
.4
.5
%
E
n
r
o
l
l
e
d
i
n
P
r
i
v
a
t
e
S
c
h
o
o
l
0
.2
.4
.6
.8
D
e
n
s
i
t
y
o
f
L
o
g
E
x
p
e
n
d
i
t
u
r
e
1096 2980 8103 22026
Monthly Household Expenditure on Log Scale (Rs.)
Density of Expenditure Distribution
Closest School Public
Closest School Private
Distance, Wealth and Public-Private Choices
4.2 1be ai.tavce ava evrottvevt avat,.i. .bor. tbat tbe ai.tavce to .cboot vatter. rbev arevt. cboo.e rbetber to evrott a
cbita or rbetber to .eva a cbita to a vbtic or rirate .cboot. One might think that parents always enroll their children
94
in the closest school gien the strong relationship between distance and enrollment ,a 500-meter increase in
this distance leads to a 20 percentage point decline or girls,. \et the results so ar say little about the school
that the parent actvatt, chooses-while children are less likely to attend school i it is urther away, this does
not necessarily imply that all evrottea children go to the closest eligible school.
4.28 Otber .etectiov factor.. In act, only 36 percent o all enrolled boys and 38 percent o all enrolled girls
attend the closest eligible school. O the remainder, 52 percent choose to bypass a nearby priate school to go
to a public school that is urther away, 15 percent bypass a nearby goernment school to go to a more distant
priate school, 25 percent choose a most distant goernment school than the one that is closest to their
house and 12 percent bypass one priate school to go to another. Neither is bypassing a purely male-child
phenomenon. \hile boys are more likely to bypass the closest school, 60 percent o all enrolled girls also
trael beyond the closest eligible school to one that is urther away. 1he additional distances they trael are
not large in themseles, ranging rom 150 meters ,priate or priate, to 500 meters ,public or public,, but
these do attain greater signiicance in light o the enrollment-distance relationship discussed aboe.
24
1hese
results may be reconciled in a number o ways.
4.29 Data ov .cboot b,a..ivg .vgge.t. tbree t,e. of bov.ebota.. A potential reconciliation is that the child whose
enrollment status is aected by the construction o a nearby school is ery dierent rom the one who
bypasses a school that is next door to go to one that is urther away. In act, the only child whose enrollment
changes ater school construction nearby i. the child who is sent to the closest school. 1he data on bypassing
suggests three types o households in the sampled illages ,see Annexure 1, 1able A4.1,:
" Distance conscious households are characterized by a combination o ewer enrolled children and
enrollment in the closest school or those who attend.
" Quality conscious households bypass a nearby public school to go to a priate school urther away.
1hese households tend to be rich with more educated parents, and the schools they inally select
report learning leels close to one standard-deiation higher than the school closest to their house.
" Price conscious households bypass a nearby priate school to go to a public school urther away. As
one may expect, these households tend to be poorer, less educated, and children are enrolled in
schools where the learning leels are much lower than the school close to their house.
24
I all households and children were identical this inding is puzzling. Consider a household with a school 500 meters away rom
home. 1he results suggest is that i the goernment constructs another school 200 meters rom the household, enrollment would
increase, but .ove of tbe vert, evrottea cbitarev rovta go to tbe .cboot tbat i. :00 veter. ara,. A natural question is: why were these children,
whose choice is not aected by the construction o the closer school, not in school in the irst place
95
4.30 1be re.evce of tbe.e aifferevt t,e. of bov.ebota. iv a t,icat rittage ;rbicb cavvot be iaevtifiea a riori) va/e. it rer,
bara to .a, vvcb abovt .cboot cboice ritbovt fvrtber a..vvtiov.. lor instance, does bypassing mean that households
are searching or quality Not necessarily. f the household is bypassing a public school to go to a priate
school urther away, it is searching or quality. But, i it bypasses a nearby priate school to go to a public
school that is urther away, it is probably searching or a lower cost. \hat happens in the aggregate data
depends on the proportions o each type o household.
4.31 vteractiov. betreev bov.ebota. or betreev cbitarev ritbiv tbe .ave bov.ebota covta at.o etaiv art of tbe aata ov
.cboot cboice. As an example, a nearby school may allow children rom a cluster o households to leae together,
some o whom may then attend the closest school, while others may trael a little urther, or een accompany
their companions to the school urther away beore returning. Such eects are likely larger i there are
particular obstacles that hae to be crossed-a main road or example, or an empty ield, or a stream.
4.32 1hese new data add seeral elements to the debate on distance and enrollment in Pakistan. 1o
summarize the new indings:
" Distance and enrollment are strongly and negatiely related, and more so or girls. 1here is little that
mitigates the distance penalty.
" 1he poor use public schools more partly because priate schools are located closer to richer
households, but een i location patterns were equitable, ,not surprisingly, there would still be a gap
in the use o priate schools by the rich and the poor.
" Moing beyond distance-enrollment relationships to the choice o speciic schools reeals dierse
household preerences with regard to quality, cost, and distance. Dierent households want dierent
things or dierent children. liguring out who these dierent households are or een their
proportions in the data is new territory with the potential to oer insights with important policy
implications.
L9.-<$9Q)- >).:#7%9(#Q 1(Z<-70<(7- %( 7$<%& :$%Q)&<(P I9(<U
4.33 1be fivat ai.cv..iov cevter. ov tbe ovtar ri.aov abovt Pa/i.tavi rvrat bov.ebota.; iv articvtar arevtat
ivre.tvevt. of tive ava vove, ov tbeir cbitarev`. eavcatiov. It is commonly belieed that households do not hae the
money to spend on their children`s education, it is also belieed that children are burdened by multiple
demands on their time, and that learning outcomes suer as a result. It is also commonly belieed that these
problems are ar worse or girls compared to boys. low much o this is true
96
4.34 ov.ebota. .eva .vb.tavtiat .vv. of vove, ov .cbootgoivg cbitarev. louseholds with children enrolled in
public schools spend Rs.155 eery month and households with children enrolled in priate schools spend
Rs.231 eery month. 1he median monthly expenditures at the household leel are about Rs.400, which
implies that a household with our children enrolled in a priate school would be spending close to 20
percent o its budget on schooling expenditures. It also implies, together with the results rom the schooling
chapter on expenditures by schools that close to 50 percent o the total spending on education in public
schools is incurred by households as out-o-pocket expenditures.
4.35 or tbe vove, i. .evt aeeva. ov rbetber tbe cbita i. evrottea iv a vbtic or a rirate .cboot. ligure 4.9 shows
the breakdown o schooling expenditures across public and priate schools. As expected, school and
admission ees are the dominant expenditure category in priate schools. 1he next most costly outlay is in
public schools, or uniorms and shoes` and pocket money`. In act, pocket money is the single biggest
expenditure category or students in public school-so large, that in absolute amounts, it is only Rs.10 less
than school-ees paid by priate school-going children.
4.36 f tbe co.t of vviforv. ava oc/et vove, i.
ectvaea frov .cboot eevaitvre., tbe arerage
avovvt .evt ov eavcatiov aro. aravaticatt,. I
expenditures on pocket money and uniorms
would hae been incurred regardless o
enrollment, it is wrongly counted as
spending associated with school enrollment.
It could well be that pocket money, which
parents say is used to purchase drinks or
snacks, would hae been spent as ood
expenses in any case, and represents a pure
substitution rom spending on the child at
home or at school. Similarly, i spending on
uniorms means that less is spent on other
clothing or the child, this is a clothing
expenditure rather than expenditure
associated with schooling. Lxcluding both o these categories drops the expenditure on schooling
dramatically to Rs.50 a month, with ree textbooks, this becomes only Rs.30 a month or public and Rs.100 a
month or priate school-going children. \ith our children in priate schools, this comes to less than 10
G%O.&< cThP L9.-<$9Q) @R<()%(O 9( :$%Q)&<( %( R.*Q%:
#() R&%Z#7< -:$99Q-
Public Private
Monthly School Fees Transport
Private Tuition Pocket Money
Annual School Fees Uniforms and Shoes
Textbooks Notebooks and school supplies
Other Expenditures
Education Expenditures by School Type
9
percent o the household`s monthly expenditures, with our children in public schools, it is less than 2
percent.
4.3 .rerage .evaivg ov cbitarev va./. cov.iaerabte rariatiov acro.. bov.ebota. ava cbitarev. Richer households and
households with educated parents spend signiicantly more and elder children incur greater educational
expenditures. 1his is not unexpected since educational spending is almost certainly a normal good. \hat is
surprising is the strong eect o child intelligence on educational spending. ligure 4.10 shows educational
expenditures or children disaggregated by parental perception o intelligence ,ery poor, poor, aerage,
aboe aerage and highly aboe aerage, and by enrollment in public,priate schools. Children perceied as
more intelligent are our times more likely to be enrolled in priate schools. 1he dierences are much smaller
or those enrolled in public schools, but still represent a two-old increase oer the same range. In contrast to
this intelligence` eect, gender discrimination in educational spending is small, accounting or at most a 5
percent dierence in spending ,ligure 4.11,. Putting the two together, households, on aerage spend a tot vore
on girls perceied to be intelligent than on boys who are not ,Rs.224 ersus Rs.180 per month,. 1o the extent
that parental perceptions o intelligence can be iewed as objectie`, this implies that een at the young ages
o 5-15, parents start supporting children who are doing better with more inestments, and cutting back on
the children they eel are not perorming.
L9.-<$9Q)- >).:#7%9(#Q 1(Z<-70<(7- %(
7$<%& :$%Q)&<(P A$%Q) "%0< 'QQ9:#7%9(
Cirt. ao vore ror/ at bove, bvt tbe, at.o .eva
ve .tva,ivg, rbite bo,. ta,. Cbitarev vot
attevaivg .cboot ao vore bov.eror/ ;girt.) ava aia
tabor ;bo,.) ava iv tbe teevage ,ear. ;e.eciatt, for
girt.) tbe.e are cto.e to fvtttive ;ob.. or tbe
rivar,age grov tbovgb, ovtof.cboot cbitarev
.eva vo.t of tbeir tive ta,ivg ;ava .teeivg).
vore ti
4.38 otb bo,. ava girt. iv .cboot. .eva te..
tbav ove bovr er aa, ov bov.eror/ or aia ror/.
1he aerage child in the sample spends 10
hours a day sleeping, 5 hours attending
school, 3 hours playing, 1 hour each on
housework and homework rom school. 1he
G%O.&< cT;iP ?#&<(7- -R<() 09&< 9( :$%Q)&<( 7$<U 7$%(W
#&< 09&< %(7<QQ%O<(7
0 10 20 3
Educational Spending Gap (%)
0
The Gaps in Child Educaitonal Spending
0 10 20 3
Educational Spending Gap (%)
0
The Gaps in Child Educaitonal Spending
The Gender Gap
The InteIIigence Gap
The Non-Poor/Poor Gap
The Literate/IIIiterate Mother Gap
The Gender Gap
The InteIIigence Gap
The Non-Poor/Poor Gap
The Literate/IIIiterate Mother Gap
98
remaining time is spent on prayer and religious actiities ,1 hour,, working or wages ,15 minutes,, preparing
or school ,45 minutes, and entertainment such as listening to the radio or watching teleision. 1his is ery
close to what children around the world do, certainly the time spent either on housework or on paid work
does not seem excessie.
"#*Q< cT]P ' /#U %( 7$< =%S<P L95 A$%Q)&<( @R<() 7$<%& /#Ua# 0%(.7< *U 0%(.7< #::9.(7
D9(_>(&9QQ<) ?.*Q%: ?&%Z#7<
@Q<<RmC<-7 661.9 59.0 603.3
?Q#U%(O 26.1 159.1 144.
L9.-<59&W 194.5 2.9 20.0
?#%) K9&W 5.0 2.8 1.5
@:$99Q 0 358.5 355.0
".%7%9( 0 16.3 25.6
@:$99Q 59&W 9.7-%)< @:$99Q 0 4.5 9.4
I<)%# >(7<&7#%(0<(7 31.1 35.2 45.1
C<Q%O%9.- >).:#7%9(m ?&#U<& 63.5 6.4 .8
?&<R#%9( S9& @:$99Q 0 58.8 5.4
H7$<& 12.6 23. 21.8
4.39 Privar, .cbootage cbitarev, ov arerage, ao vot .eva tbeir tive iv aia ror/ erev rbev tbe, are ovt of .cboot. An
immediate question is the dierence in time allocation between children who are going to school and those
who are not. 1he aerage school-going child spends 350 minutes, or roughly 6 hours in school and 1 hour, on
school work. 1he 6 hours or the child who is not enrolled is spent sleeping and playing ,2' hours,, working
or pay ,1 hour,, and housework ,3 hours, ,1able 4.3,. Children, on aerage, are not working all the time
whether they attend school or not. Despite the 6 hours that they spend in school eery day, the enrolled child
still gets 2 ' hours a day to play, spends a ' an hour on housework and less than 2 minutes on paid work.
Children o primary school age ,less than 12 years old, who do not attend school play more than 4 hours a
day, substantially more than those in school and spend minutes on aerage on paid work and housework.
1he notion that school-age children are being exploited or child labor` is not consistent with the data.
99
4.40 or otaer cbitarev, articvtart, girt., cbita tabor va, be av i..ve. ligures 4.11 plots the minutes spent eery
day on paid work and the minutes spent on the
combination o paid and housework or girls and
boys, separately. 1he graph on the let is or
children who are in school, that on the right or
those who do not attend school. 1he
abbreiations l\ stands or housework` and
P\ or paid work`. lor children in school, paid
work or both girls and boys is minimal and
probably relects one-o actiities during
particular seasons. It neer exceeds 10 minutes a
day, een or older children ,15 years old,. \hen
housework is added to paid work, girls come out looking much worse o than boys, and the dierence
between the two increases with age. At 10 years old, girls and boys are similar in their time allocation patterns,
but by the time they are 15, girls spend twice as much time as boys on these two categories, but the total time
spent neer exceeds 2 hours. lor children not attending school, the gradient o time spent on these two
categories becomes steeper with age or both boys and girls, with boys spending more time on paid and girls
more time on housework. By the time she reaches her early teen years, the girl-child who does not attend
school spends close to 300 minutes a day on housework and by the time she is 15, housework has become a
ull-time job ,8.5 hours a day,. lor these girls, the trade-o between child work and school is clear.
G%O.&< cT;;P
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
5 10 15
Age
HW+PW Girls HW+PW Boys
PW by Girls PW by Boys
For Children enrolled in School
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
5 10 15
Age
HW+PW Girls HW+PW Boys
PW by Girls PW by Boys
For Children not enrolled in School
Time Spent on Paid Work and House Work by Gender
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
5 10 15
Age
HW+PW Girls HW+PW Boys
PW by Girls PW by Boys
For Children enrolled in School
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
5 10 15
Age
HW+PW Girls HW+PW Boys
PW by Girls PW by Boys
For Children enrolled in School
0
1
0
0
2
0
0
3
0
0
4
0
0
5
0
0
5 10 15
Age
HW+PW Girls HW+PW Boys
PW by Girls PW by Boys
For Children not enrolled in School
Time Spent on Paid Work and House Work by Gender
4.41 .ttbovgb evrottea girt. .eva vore tive ov
bov.e ava aia ror/ tbav bo,. vore bov.eror/ for
girt. aoe. vot veav tbat tbe, are .evaivg te.. tive ov
.cbootretatea actiritie.. ligure 4.12 graphs the
time girls and boys o dierent ages spend on
schoolwork at home ,S\, and learning
actiities, in general, which includes not only
homework, but also time spent in tuition and
media actiities. 1he graph shows that girls
spend more time on learning actiities. In act,
most o the additional time spent doing
housework or girls comes at the expense o
playtime. In essence, boys and girls are both in
school or roughly the same time, but when they come home, boys are allowed to play more while girls are
G%O.&< cT;YP M%&Q- )9 09&< $9.-<59&Wg#() 09&<
-:$99Q 59&W
5
0
1
0
0
1
5
0
2
0
0
5 10 15
Age
Time Spent on Learning by Girls Time Spent on Learning by Boys
Time Spent on SW by Girls Time Spent on SW by Boys
Source: LEAPS Household Survey
For Children enrolled in School
Time Spent on Learning and on School Work by Gender
5
0
1
0
0
1
5
0
2
0
0
5 10 15
Age
Time Spent on Learning by Girls Time Spent on Learning by Boys
Time Spent on SW by Girls Time Spent on SW by Boys
Source: LEAPS Household Survey
For Children enrolled in School
Time Spent on Learning and on School Work by Gender
100
asked to do more housework. 1he dierences are not large, een with this additional housework, the aerage
girl-child spends 2 hours a day playing.
11T /1@AB@@1HD
4.42 1be .P bov.ebota .vrre, ava .vb.eqvevt avat,.i. aaa. av ivortavt ver aivev.iov to tbe eavcatiovat aebate.
As this study makes clear, the question acing parents in many illages is no longer should I enroll my child
in the single school in the illage` but rbicb school should I send my children to` 1he complexity o this
decision must be taken into account by policymakers i uture educational policies are to be as eectie as
possible. Some parents care about quality, some about cost, and some about distance, and i policies ail to
take this into account they will likely aid one type o households but not others.
4.43 1bree /e, a,vavic. everge tbat are v.efvt for tbe otic, aebate iv tbi. vore covte evrirovvevt. lirst, distance is
important or enrollment, especially or girls. Second, parents are airly well-inormed and make substantial
inestments in their children`s education. 1hird, through their choices o whether to enroll a child, through
the choice o school ,priate or public, and inally through the amount they chose to spend, households pick
winners` and try to carry them through. In the ollowing section, we examine the implications these three
dynamics hae or two key policy issues currently under debate, a third important policy issue-the regulation
o schools-is discussed in the epilogue.
?97<(7%#Q ?9Q%:U ;P 10R&9Z%(O <).:#7%9( S9& 7$< O%&Q_:$%Q)
4.44 1be gevaer ga iv eavcatiovat attaivvevt i. targe, bvt a. tbe fivaivg. iv tbi. cbater .bor, tbe gevaer ga i. atvo.t
evtiret, re.trictea to evrottvevt. 1he gap in education expenditures is small ,not more than 5 percent, and
households spend a lot more on intelligent girls than on less intelligent boys. It is true that girls enrolled in
school work more at home than boys, but the aerage working hours are still less than an hour per day and
girls at.o do more schoolwork at home compared to boys.
4.45 1be covvovt, beta rier tbat ai.tavce to .cboot affect. girt. vore tbav bo,. i. .vortea b, tbe aata ava accovvt. for 0
ercevt of tbe gevaer ga iv evrottvevt. And that is precisely the problem. 1he relationship is so strong and so
negatie, that it really does not make much sense iewed in physical terms. Lnrollment drops by 20
percentage points or girls or eery 500 meters increase in distance rom home to school. \hen we walked
with out-o-school children rom households in our sample, it took us 10 minutes to walk these 500 meters.
\ith 12 schools in eery illage and with 50 percent o our households within 500 meters o a school, the
policy option o building more schools makes no sense. No goernment can ensure the aailability o a
school within 100 meters o eery household, such a policy is neither easible nor cost-eectie. lurthermore,
101
household wealth, the child`s age and the child`s intelligent all hae an independent eect on enrollment, but
vove o them mitigate the distance penalty`, except or the presence o an educated emale, which is outside
the puriew o any short-term policies. It appears then that this distance eect has little to with monetary
needs o the amily or whether the child has reached her teenage years. A workable policy is not yet eident.
4.46 1be Pvv;ab gorervvevt`. .tieva rograv, rbereb, bov.ebota. are girev R..200 er vovtb for erer, evrottea girt
betreev Cta.. : ava .igvat. tbe gorervvevt`. covvitvevt to eavcatiov, ,et tbe co.t. va, far ovtreigb tbe bevefit.. An
ealuation ,Chaudhury and Parjuli 2006, shows the program increased enrollment by 10 percent. Out o 110
girls now in school, 100 would hae gone to school with or without the stipend and 10 additional girls
changed status rom non-enrollment to enrollment because o the stipend. 1he price o targeting these 10
girls is the stipend paid to all 100. 1he cost er aaaitiovat girtcbita evrottea is the total cost o the stipend diided
by 10-a staggering >400 per marginal child. 1he stipend cannot distinguish between children who would be
going to school anyway, or children who change their decision as a result o the stipend, or children who
would not go to school with or without the stipend. 1he cost to target the second group is unsustainably
large.
4.4 Differevce. acro.. bov.ebota. are eriaevt iv tbe .igvificavt fractiov of cbitarev rbo b,a.. tbe cto.e.t .cboot to go to a
.cboot tbat i. fvrtber ara,. Although males are slightly more likely to bypass, 60 percent o all enrolled girls are
bypassing as well. 1he key to improing emale enrollment is to understand better the relationship between
enrollment and distance. \hy is it that some households and children are able to send their kids to schools
urther away, while others take their children out as the distance to school increases \ould this change i
chaperones or elder siblings or neighborhood children walk to school together 1his requires urther
research.
?97<(7%#Q ?9Q%:U YP 10R&9Z%(O -:$99Q%(O 9.7:90<- 7$&9.O$ 09&< -:$99Q %(R.7-
4.48 1be goat. of oticie. to ivcrea.e .cboot ivvt. veea to be aebatea tbrovgb a bov.ebota tev.. 1he goernment
recently instituted a policy o proiding ree textbooks. 1his may bring in children or whom the costs o
textbooks were prohibitie and who were thereore not enrolled. It may also hae improed learning among
the children who were already enrolled but did not hae access to textbooks. 1he debate here centers around
two issues. lirst, among the out-o-school children, was the cost o textbooks really prohibitie 1he data are
unclear, and much depends on whether we think o pocket money and spending on uniorms as substitutes
or household expenditures on all children. Second, giing out textbooks or ree means that households who
were earlier buying textbooks or their children will stop buying them so that the total number o textbooks a
child has access to remains unaected by the policy. \hat is the speciic aim o this policy I all children
were enrolled, ree textbooks or instance, are a pure ivcove subsidy since parents sae the equialent amount
102
by not buying textbooks on their own. \hile this might be a desirable eect o such a program, we should
not expect improements in learning as a consequence. \e are asking, in essence, that a household lens` be
applied in thinking o potential educational policy.
4.49 . bov.ebota tev. .vgge.t. a rote for eavcatiovat oticie. tbat .vort cbitarev rbo receire ferer ivvt. at bove. Right
now some children are heaily supported by households and others are let by the wayside. Children
perceied as intelligent by their parents are more likely to be enrolled, more likely to be in priate schools, and
hae three times as much money spent on them. In high-income countries, higher inestments in more
productie` children typically kick in around the college-going years. But or children in primary school,
>12,500 per child is spent on learning-disabled children in the United States compared to >6,500 spent on
regular` children, with an implied spending ratio o close to 2:1. In Pakistan, by the time children are in
Class 3, parents hae already picked winners` perceied to be more intelligent and are spending three times
as much on these children. Public money or education should include these ulnerable` children. In India,
or instance, there is some eidence that programs designed to help poorly perorming children ,Banerjee and
Cole, Dulo, and Linden 2005, yield positie eects. Maybe it is time to think o similar programs or
Pakistan.
4.50 t va, at.o be tive for a riaer ai.cv..iov abovt tbe oreratt goat. of tbe eavcatiovat .,.tev. South Asian systems
were traditionally ormulated or stringent selection o talented students: the aerage or poorly perorming
child was relegated to trades or liestyles that did not require ormal schooling. 1he system was designed to
bring out the best possible children through tests and screening at eery possible occasion. 1he data show
that households are mirroring this system-certain public sector jobs, or instance, are made aailable only to
children who hae passed Class 10, not surprisingly, returns to education in Pakistan are covre at low leels o
education. I a amily is aiming or such a job or one o its children, it will be spent considerable resources to
ensure that the required qualiication is obtained, i the child can get to Class 9, but no urther, the returns are
signiicantly lower. 1hese kinds o non-linear` returns or sheepskin` eects, as they are known in the
literature, may explain non-compensatory behaior among households at a young age. 1o the extent that the
desire or the aerage child to perorm well seems to be more widely held, it is time now to think about
policies that screen poorly perorming children and take measures to help them speciically, without spending
large amounts on the aerage child.
4.51 !baterer tbe.e erevtvat oticie. va, be, eacb ae.erre. to be rigorov.t, eratvatea. During the course o our
sureys, a emale teacher oered the ollowing wisdom: 1be aifferevce`, she said, betreev Pa/i.tav ava ]aav i.
tbat iv ]aav tbe, tbiv/ for :0 ,ear. ava tbev va/e oticie., iv Pa/i.tav re va/e tbe otic, fir.t ava ao tbe tbiv/ivg tater, if
at att.` 1he existing eidence oers policy makers clear choices in some areas-teacher`s perormance, or
103
instance, would certainly improe i they were held accountable or their actions. loweer, dierences across
households and children make predictions about policy eects harder-the ultimate eects o these policies
will depend on the proportion o households with dierent types o preerences and returns to education.
1his proportion will certainly dier across regions ,what works or Punjab may not work or Sindh, and
plausibly across illages. In other areas more needs to be understood about the nature o dierences across
households and children that aect household educational choices. Lxperimenting with and ealuating
policies that are in the implementation process is critical. 1hese ealuations will at least alidate how useul
they were in the region they were implemented. 1hey will be een more inormatie i applied using the
household lens` so the extent to which these policies interact with the decisions that households make can
be better understood.
4.52 .v aarerti.ivg eecvtire ovce revar/ea tbat be /vor. tbat batf tbe vove, .evt ov aarerti.ivg i. ra.tea, tbe robtev
i. tbat be aoe. vot /vor rbicb batf. 1his is particularly true in this case. Policies now need to identiy the vargivat
rather than the arerage child so that they do not spend a lot o money on arerage children to beneit those who
are marginal`. In these situations, it is oten best to enable households to make better decisions without
orcing a particular set o actions on them, how the goernment can act as a better enabler requires discussion
and debate.
104
Chapter 5: 1houghts or Discussion and Debate
5.1 1be .tate roriae. eqvitabte acce.. to eavcatiov at tbe rittage teret iv Pvv;ab. Article 3 ,b, & ,c, o the
Constitution o Pakistan ,193, airms that the State shall remoe illiteracy and proide ree and
compulsory secondary education within the minimum possible period.` During the 1990s enrollment rose
rapidly in Punjab and other proinces. Based on the sample in the LLAPS study, which coers at least 50
percent o the population o Punjab at the time o the study, the State is indeed proiding access to
education. Lery illage in the sample has multiple ree public schools, and aerage learning is similar across
illages, independent o illage-leel wealth or learning. Data rom the National Lducation Census ,2005,
conirm that there are more goernment schools per student in poorer illages in all o rural Pakistan, thus
extending the indings rom the LLAPS sample illages..
5.2 iterac, goat. oftev vvvet. \hat Pakistan needs to debate is whether any attempt is being made to
address the airmation that the State shall remoe illiteracy`. It also needs to debate the implications o the
rapid rise o well-perorming priate schools or achieing this goal. 1his chapter examines ie issues related
to the quality o education and uture education policy that are requently debated in Pakistan. 1he goal o
this chapter is to argue that eidence can yield important insights or policy and proide a ramework or
urther debate.
1--.< ;P "$< =<#&(%(O 'O<()#P K$#7 )9<- %7 %0RQUX
5.3 vcrea.ivg tearvivg reqvire. brivgivg vore cbitarev ivto .cboot. One easily identiied group o children at risk
o being let out o the educational system are those who lie ar` rom school and are thereore less likely
to be enrolled. 1his study agrees with a number o other studies in Pakistan that the distance to school has a
big impact on enrollment, particularly or girls. 1he problem is that this eect is too big or a potential
supply-side response o constructing more schools to bring all out-o-school children into schools. It would
be expensie to construct, maintain, and sta schools within 100 meters o eery household.
5.4 1be gorervvevt veea. to actiret, itot ivterrevtiov. to .ee bor tbe.e ai.tavt effect. cav be orercove. 1he problem
with the results thus ar is that there are no clear household or child characteristics that mitigate the harmul
eects o distance. An idea that has been adanced is the use o chaperones` who bring children to
school-anecdotally, it has been suggested that when children can walk to school together or with a
responsible adult, they are more likely to enroll. It is exceedingly diicult to assess this type o phenomenon
with these data. 1his is partly because there are too ew households to construct indicators o what
105
neighbors` are doing and partly because such interaction eects ,or instance, using the number o siblings
as an appropriate indicator, are hard to interpret. It is diicult to dissociate household eects rom the
independent eects o haing someone to go to school with. I chaperones` are indeed the solution, this
should not be ery hard to address. Gien the diiculties in assessing the impact obserationally, this sort o
interention requires rigorous ealuation using randomized strategies.
5.5 Cbitarev rbo ao vot atteva .cboot ror/ for a, te.. tbav tro bovr. a aa,. 1he data are also notable in
showing that the alternatie to school is vot work. Primary age children who are not in school spend 93
minutes a day ,an hour-and-a-hal, working at home and working or a wage. Getting these children into
school does not require compensating` parents or a loss o income, but it may require schools that are
more interesting or children and schools where learning is o a higher quality. 1he appropriate inrastructure
and pedagogic package` to get children who are disinterested and not pushed by their parents into school
requires the combined knowledge o educators, head-teachers and parents themseles.
5.6 !bite ivrorivg evrottvevt i. criticat torara ev.vrivg tbat cbitarev tearv, it i. iv.vfficievt. Children are unable to
read simple sentences in Urdu and add and subtract by the time they are in Class 3. 1hat greater enrollment
does not imply more learning demonstrates the importance o a holistic approach toward the educational
rights o the Pakistani citizen. More children may be attending school in rich and highly literate illages
,Chapter 1,, but the aerage child`s test scores in Class 3 remain largely unchanged.
5. Ovce cbitarev are iv .cboot, effort. to vea.vre erforvavce vv.t fottor. Unortunately, there is no magic recipe
or improing learning. Despite many years o research in high income countries, there is little consensus
,apart rom the role o teachers discussed below, on how learning can be improed. 1here is considerable
agreement, howeer, that collecting inormation on learning is the critical irst step or the broader learning
agenda`. Beore we een think o how learning may be engendered, a system or tracking learning in dierent
schools oer time based on tests conducted by a reliable and impartial authority will important to ensure that
citizens know how well their children are perorming in school.
5.8 vforvatiov ov evrottvevt ava tearvivg ovtcove. veea to becove art of tbe ovtar ai.covr.e ov tbe .tate of
Pa/i.tavi eavcatiov. 1he widespread use o a test at the end o Class 5 in Punjab proince in 2006 is an
important irst step. lurther work is required to ensure that citizens are able to monitor and hold the state
accountable or its perormance in guaranteeing the right to education. Lnsuring exam results are
standardized and replicated eery year in a reliable manner is only part o the challenge, access to comparable
data or all schools is also critical. In recent years, Punjab has collected annual data on all its public schools,
the lederal Bureau o Statistics has enumerated and collected inormation on att schools in the country, and
106
Punjab has tested children in Class 5. \et, these data are not publicly aailable and their accuracy has yet to be
subjected to independent analysis. Researchers, academics, multilateral institutions ,who are oten ivticatea in
the withholding o data and inormation, and those interested in the Pakistani education system need to
ensure the general public has ready access to this inormation. \hile enrollment numbers are currently
released on an annual basis, similar summary statistics on academic achieement need to be made aailable as
well to ensure school quality becomes part o the popular discourse on the state o Pakistani education. In the
United States, school perormance in learning is posted on the internet. 1his may be a worthy goal to aspire
to.
5.9 1be covari.ov of vbtic ava rirate .cboot. vaae bere .trovgt, .vgge.t tbat better tearvivg i. vot ovt of reacb of tbe
Pa/i.tavi .,.tev. Len within the same illages, priate schools are doing much better in all subjects than
public schools. lurthermore, the dierences between schools are ar greater than the dierences across
households or illages, and are equally strong ater accounting or parental wealth and education. Based on
the results o this study, where public and priate schools in the same community are compared, the best
place to ocus attention is on the condition o schools and not the attitude o parents toward their children`s
schooling.
1--.< YP @$9.Q) 7$< M9Z<&(0<(7 C<O.Q#7< -:$99Q-X
5.10 v.vrivg .tavaara. of qvatit,. Popular policy pronouncements and the discourse surrounding priate
schools oten reole around regulation o quality and ,at arying points in time, prices. But what is such
regulation meant to do One purpose o regulation is to ensure that proiders delier a minimum standard o
perormance. Using perormance and test scores as a yardstick, it is goernment, not priate schools that
need better oersight. 1he bulk o the poorest-perorming schools are goernment schools, where there is
little accountability or mechanism or complaints-these schools are not meeting any minimum quality
standard. In contrast, parents unhappy with a priate school can simply withdraw their child, and this exit
option shows up clearly in the data-children in the worst priate schools hae ar higher test-scores than
those in the worst public schools ,Chapter 1,.
5.11 v.vrivg covetitire ricivg. A second purpose o regulation is to ensure that priate schools do not
charge excessie` ees. Such pricing ineiciencies arise rom monopolistic behaior. lirms that act as
monopolies maximize proits by charging prices aboe what is socially optimal, some consumers who would
like to buy the product rom the monopolist cannot because this implies decreasing the price or the evtire
market. 1ypically, eery country has an authority that looks at monopolistic or restrictie trade practices and
adocates alternaties. 1he chapter on schools showed that priate schools are not behaing as monopolies.
1hese schools are most oten located in schooling clusters` with other priate and goernment schools.
10
Direct competition rom other schools keeps their prices low-indeed, the aerage proit at a rural priate
school in Punjab is approximately equal to the salary o a male priate school teacher. 1his is precisely the
opportunity cost o the priate school entrepreneur. Goernment schools tend to be located both in
schooling clusters and in stand-alone locations outside the main settlement. Gien the strong relationship
between distance and enrollment, these schools may be the ovt, option or a large set o children. lrom
chapter 2, leels o inrastructure in these schools are lower, suggesting that they do not command the same
resources that goernment schools in denser and richer settlements do. I there is any eidence that some
schools may be monopolies` it is these goernment schools rather than the majority o priate schools.
5.12 v.vrivg tbat rice. reftect qvatit,. 1he third common rationale or regulation is that consumers are
unable to ealuate the quality o the product they receie. In the case o priate schools, parents cannot tell
the quality o education imparted in school and are thus able to leece` parents by charging high ees or low
quality. It is worth recognizing that this rationale or regulation does not imply that poor quality priate
schools should not exist. In act, poor quality need vot be a problem as long as it is completely embedded in
the price o the serice and as long as the preerences o the child are ully relected in the decisions o the
parent. A Mehran is cheaper than a BM\, yet ew would argue that there is a need to regulate Mehran`s out
o the market. Similarly, in the schooling case, i worse perorming and cheaper priate schools are regulated
out o existence, there will be a large number o households who will stop sending their children to school
,because the alternatie goernment school is too ar away or een lower quality, or be orced to opt or an
een lower quality goernment school. 1he rationale or regulation arises rom the discrepancy between price
and quality, rather than lower quality itsel.
5.13 Moreorer, tbi. i. a ratiovate for roriaivg vore ivforvatiov, vot vece..arit, regvtatiov. It becomes a rationale or
regulation only i the cost o proiding inormation about quality is prohibitiely high, leaing regulation as an
only alternatie. 1he results suggest that the market is already working well in some ways. 1he aerage
parent`s ranking ,whether the parent is literate or not, o schools coincides with the ranking rom
independently administered tests, suggesting parents are already aware o the quality o schools in their
illages. Quality is also relected in prices - higher ees are clearly associated with higher quality. 1hat said,
proiding more inormation could help increase awareness about the number o schools in the illage and
their relatie quality and lead to better outcomes through increased competition. I eery illage had posters
and a page o test scores or dierent schools and their addresses in the illage ,perhaps with comparisons to
other illages or their own district,, parents might make better school choices. Indeed, a randomized
ealuation o precisely such an experiment is underway. Preliminary results suggest that the experiment led to
better learning outcomes among the initially poor perormers. I conirmed, such an experiment could then
be reealuated within the context o the Punjab goernment`s large-scale testing exercise in Class 5.
108
5.14 1be.e tbree ratiovate. for regvtatiov-ev.vrivg a .tavaara of qvatit,, ev.vrivg covetitire ricivg ava ev.vrivg
tbat rice. reftect qvatit, .vgge.t tbat vbtic .ector .cboot. are better cavaiaate. for regvtatiov ov att tbree covvt. tbav rirate
.ector .cboot.. But goernment schools are already regulated! It seems parent`s ability to choose schools is
operating better than goernment regulation to delier basic minimum educational goals. Pakistan needs to
rethink the costs and beneits o regulating priate schools, gien these indings rom the data.
5.15 Regvtatiov ba. to fir.t focv. ov tbe ror.terforvivg gorervvevt .cboot.. As reported, the gap between the best
and worst goernment schools is 10 times the gap between children rom rich and poor amilies and 15 times
the gap between children with literate and illiterate parents. 1he utter ailure o the lowest perorming
goernment schools suggests an obious target or any uture education reorm program. Although eorts to
improe mediocre and slightly aboe aerage schools may yield positie results, these schools are not the
driing orce behind low child test scores. It is the worst goernment schools that drie down aerage test
scores-dropping the 50 worst goernment schools is equialent to raising the other 436 goernment
schools` scores by 25 knowledge points each. 1he debate about regulation has to fir.t ocus on ery poorly
perorming goernment schools. Once theses schools are on track, attention can then be ocused on the
extension o regulation to other types o schools.
1--.<- ]P @$%S7%(O 7$< S9:.- S&90 %(R.7- 79 9.7:90<-
5.16 Refravivg tbe eavcatiov otic, aebate ivrotre. .biftivg tbe focv. frov ivvt. to ovtcove.. Most popular writing on
Pakistani education starts with the poor state o school inrastructure, high pupil-teacher ratios, and low
budgetary support. 1he implication is that more money will sole all these problems. 1his would be an easy
solution and one that all unding agencies would be happy to support. Unortunately, as a number o studies
worldwide show these inputs cannot explain large dierences in learning between schools. As conirmed in
chapter 1, priate schools outperorm public schools regardless o their inrastructure or pupil-teacher ratio.
In act, the correlation between inrastructure and test scores or student-teacher ratios and test scores is airly
small to begin with.
5.1 1be avovvt of re.ovrce. reqvirea to eavcate a cbita iv tbe rirate .ector i. ovebatf tbe avovvt reqvirea b, tbe vbtic
.ector. 1ake the case o Urdu. Priate school students grasp 38 percent o the content in the test, goernment
school students grasp 25 percent. By the end o Class 3, assuming that children hae been in school or three
years, the spending on a goernment school student is Rs.6000, on a priate school student Rs.3000.
1hereore, it costs Rs.9 per Urdu-percent to educate a child in the priate sector and Rs.240 or 3 times as
much in the goernment sector. 1hese numbers do not change much i we account or dierences in
household characteristics. Priva facie, the argument that improing outcomes requires ar greater resources
does not appear to be accurate. Although there are more subtle arguments based on the tendency or priate
109
schools to open in richer, more educated illages that support a case or greater spending ,more on this
below,, the subject is clearly a matter or public debate. 1he automatic reaction to poor educational results is
to throw more money at it beore looking at whether something else is going on.
5.18 1be .ovrce of tbe vbticrirate .cboot tearvivg ga i. tbe abitit, of rirate .cboot aavivi.trator. to aaat to tbe
.ecific circvv.tavce. of tbeir .cboot. 1he key question o course, is what this something else might be. In a number
o discussions the irst question is always So what do priate schools do dierently that goernment schools
do not` 1he question and what is implied is that there must be some dierence in av ivvt or .et of ivvt. that
can be proided by the goernment to bring them to the same leel o learning as priate schools.
5.19 De.ite etev.ire effort. to i.otate ob.errabte factor. tbat covta etaiv tbe riratevbtic ga, tbe aata cottectea b, tbe
.P .vrre, cavvot covctv.iret, etaiv rb, rirate .cboot. ovt erforv vbtic .cboot.. Using the set o children in the
household surey who could be matched to test scores in the school-based assessment ,about 900 in all, a
kitchen-sink` regression was used to explain learning as a unction o child characteristics ,age, gender,
health status, household spending on education, intelligence, work ethic,,, amily characteristics ,aailability o
books and other media at home, parental education and income, time spent with the child on studies,, school
characteristics ,inrastructure, location, student-teacher ratio, and teacher characteristics ,absenteeism, age,
education, gender, test scores, training,. \hile some o these were independently correlated with test scores,
the priate-public gap could only be reduced by 30 percent at vo.t. 1his in sharp contrast to richer countries,
where the gap is sharply reduced the moment a basic set o household characteristics is included in the
regression.
5.20 lollow-up ieldwork was done to interiew three head-teachers in priate schools to get at this
elusie input. All agreed that it was important to bring motiated teachers onboard and ensure that they were
ulilling their teaching duties, but they diered on eerything else. In the irst interiew, the head-teacher elt
that one o his teachers was weak in mathematics, so he had arranged or her, together with other teachers
rom neighboring illages, to go or urther training in laizabad, 60 kilometers away rom the illage. 1he
second head-teacher elt that the lack o a boundary wall was distracting students ,the school was next to a
road,, so he spent unds on building such a wall and his impression was that children could now ocus more
as a result o this construction. In the third school, children rom one settlement were requently absent since
they had to cross a small orest to reach the school. 1he head-teacher decided to send a teacher eery
morning to this settlement to chaperone the children to the school, and attendance increased dramatically.
Much like the intrepid teacher dealing with 94 children in chapter 1, priate schools are probably doing better
not because o a set o inputs` that are o higher quality than those proided by the goernment, but
because they hae the ability and administratie lexibility to ix the weakest link in the chain.
110
5.21 biftivg frov ivvt. to ovtcove. reqvire. fteibitit,. One approach is the planning approach` that tries
one by one, to check the impact o each input on outcomes, this has also been the approach o most
educational research in low-income countries. A second approach admits that dierent places and dierent
children hae dierent needs and a central planner is inherently unable to align inputs perectly or eery
single child. loweer, it also recognizes agents at the local leel cav do this better-the lexible head-teachers
in priate schools or instance. 1he lex-approach` suggests that instead o trying to ix eery input
optimally, the planner ixes the .,.tev so that those who know more and are able to respond better to
indiidual needs prosper, while those who are inlexible and proide low-quality inputs are selected out o the
system. 1he priate system does this automatically. As we saw in the chapter on households-priate schools
that cannot proide the learning leels parents expect ultimately shut down or charge lower ees. A debate
around whether a more lexible approach` is better or een easible in the goernment system also needs to
tackle the question o educational perormance.
1--.< cP G%V%(O %(-7%7.7%9(-P ?.*Q%:_?&%Z#7< ?#&7(<&-$%R-X
5.22 v covvtrie. rbere rirate .cbootivg otiov i. riae.reaa, otic, otiov. iv eavcatiov bare rerotrea arovva vbtic
rirate artver.bi.. Such partnerships largely inole goernment inancing and priate deliery o education-
thus combining the lexibility o the priate system with unds rom the goernment. Lxamples include grant-
in aid schools ,UK, India, and charter schools in the US which largely inole block grants,unding to
priate schools. 1he other model is inancing amilies directly through ouchers to each school-going child.
Voucher systems de-link the inancing o education rom its proision by ensuring that money ollows the
child`. 1hat is, parents are gien a ixed sum eery month by the goernment and they use this to pay the ees
o any school they elect to send their child to, i they choose a ree public school, the money is sent back to
the goernment. 1his has been tried in Colombia, Chile, Sweden and the U.S., among other countries. low
well these arious arrangements hae worked is highly debated and depends on country circumstances.
5.23 .ttbovgb tbe targe co.terercevtage aifferevce betreev vbtic ava rirate .cboot. .vgge.t. a .trovg ;v.tificatiov for
rovcber., tbere are a fer trovbtivg i..ve. to cov.iaer. Priate schools are oerwhelmingly located in the main
settlements, which are richer. lor these main settlements, there is indeed a strong case or a oucher system.
But, under a oucher system, will priate schools locate in peripheral settlements, and at what price 1he
perormance and prices o priate schools are closely tied to their ability to ind good teachers. I such
teachers are hard to ind, priate schools will just increase their ees to cater to the larger demand under a
oucher system, i they are not allowed top-up` ees oer and aboe the oucher price, it`s likely they will
decrease quality. In act, there has been no increase in perormance as a result o the oucher systems used in
111
Chile oer the past 20 years. As one obserer noted, priate schools tend to become like public schools, and
in direct proportion to the public subsidy they receie.
5.24 vcb i..ve. of erforvavce ava rice va, vot be robtev. iv tbe targer ava aev.er .etttevevt. rbere tbe rirate
.cbootivg ctv.ter. atreaa, ei.t. But this leaes the poorer and disadantaged children in peripheral settlements
undersered. 1he current goernment system, counter to what is commonly belieed, proides education
equitably across illages, and in most cases ,with the exception o inrastructure,, within them as well
,Chapters 2 and 3,.
5.25 1bat tbe gorervvevt .ector proides eqvitabte
eavcatiovat oortvvitie. i. criticat, becav.e it i. tbe ovt,
art of tbe eavcatiovat .,.tev tbat aoe. .o. Priate
schools, by irtue o their prices and their
location choices do not-poor households
cannot pay een the relatiely small ees, and lie
arther rom such schools. As ligure 5.1 shows,
households do not treat children in an equitable
manner either. 1he igure shows what parents
spend per month on children`s education,
separated across the children they perceie as
intelligent ersus those they perceie are not.
Children who are not perceied as intelligent hae three strikes against them-they are 20 percentage points
less likely to be enrolled in school, they are more likely to be enrolled in public schools and, een when
enrolled, households inest less time and money on them. 1he problem does not seem to be one o
prioritizing expenditures in a low resource setting-the gap in spending by child intelligence actually ivcrea.e.
with income. As it turns out, most o the money spent on the education o children thought to be o aerage
or below` intelligence comes rom the public sector. Len with a oucher system, it will take a while beore
supply o priate schooling picks up-meanwhile, i Pakistan is to take seriously the idea o Lducation or
All`, there is really no option but to improe the perormance o goernment schools.
G%O.&< ^T;P ?#&<(7- -R<() 09&< 9( %(7<QQ%O<(7
:$%Q)&<(g#() 7$< )%SS<&<(:< %(:&<#-<- 5%7$
%(:90<
0
1
0
0
2
0
0
3
0
0
4
0
0
M
o
n
t
h
l
y
E
d
u
c
a
t
i
o
n
a
l
E
x
p
e
n
d
i
t
u
r
e
s
p
e
r
C
h
i
l
d
0 5000 10000 15000 20000
Monthly Household Expenditures
Above average intelligence Average or Below ntelligence
EducationaI Expenditure, HousehoId Income and InteIIigence
0
1
0
0
2
0
0
3
0
0
4
0
0
M
o
n
t
h
l
y
E
d
u
c
a
t
i
o
n
a
l
E
x
p
e
n
d
i
t
u
r
e
s
p
e
r
C
h
i
l
d
0 5000 10000 15000 20000
Monthly Household Expenditures
Above average intelligence Average or Below ntelligence
EducationaI Expenditure, HousehoId Income and InteIIigence
5.26 !bicberer ra, tbe aebate goe., a carefvt eratvatiov of rbat a rovcber .,.tev actvatt, ritt ao i. ritat. A proper
ealuation program would hae at least 3 components. lirst, it would last at least 5 years, since it will be
critical to see whether the priate sector is able to respond to greater inancing by increasing supply. Second,
it would consider the rittage as the appropriate unit or assessing test-score responses. Vouchers may lead to
greater social stratiication, i such stratiication means that children learn less rom each other, it may hae a
112
detrimental eect on learning. 1hird, it would look not only at the eect on the aerage` child but also the
eect on disadantaged children ,whether because o their location or their backgrounds,.
1--.< ^P G%V%(O %(-7%7.7%9(-P A#( M9Z<&(0<(7 @:$99Q- 10R&9Z<X
5.2 rev ritb greater rirate .ector ivrotrevevt, fiivg gorervvevt .cboot. i. .titt ivortavt, goat. 1he biggest issue in
the goernment sector seems to be what to do about teachers-the rest is tinkering at the margins. 1he
chapter on teachers oers a airly comprehensie ramework or debate, and is briely summarized here.
5.28 1ro tbivg. abovt teacber. vatter vo.t-tbeir ivtriv.ic votiratiov ava tore of teacbivg ava tbe ivcevtire. tbat tbe,
face. \hat matters te.. is their ormal educational qualiications, so long as they are aboe secondary ,or
primary school children, and they hae some training. Although a number o studies in low-income countries
argue that schools should design systems that proide teachers with better incenties, this might be ery hard
to implement and not sustainable in the long-run. It is particularly diicult in an enironment where
goernment teachers ulill non-teaching responsibilities such as manning election booths, drawing up oter
lists, working as part and parcel o a political system ,chapter 3,, and perorming administratie duties such as
administering polio accinations. Len i teachers were to be relieed o non-teaching responsibilities and
better monitoring o teacher attendance could encourage less absenteeism, more subtle incentie schemes,
such as pay based on test scores o the children in their classes, are probably out-o-reach at the moment. 1he
goal o getting the right teachers into the public sector and ensuring that top perormers are retained while
those who do not perorm are gradually let-go remains a daunting problem.
5.29 .t tbe vovevt gooa gorervvevt teacber. receire tbe .ave covev.atiov a. tbo.e tbat ao vo ror/ ava tbere i. vo
roce.. for .beaaivg teacber. rbo .bir/. As Pakistan grows and other job options become aailable, the mechanical
wage unction in the public sector may orce the best teachers ,who would arguably receie higher salaries
elsewhere, to leae, and the worst ,who would arguably receie lower salaries outside, to stay-similar
patterns hae emerged in the teaching workorce in the United States oer the last two decades, and are
already eident in urban Lahore and Karachi.
5.30 1be gorervvevt cav affora to vore teacber. arovva. In rethinking how goernment teachers-who
currently earn 5 times as much as their priate sector counterparts-should be compensated, two broad
issues arise. lirst, gien that this higher salary buys` the goernment the right to transer teachers, perhaps
the goernment should think about deploying teachers to geographical areas and to educational sectors ,the
secondary sector or instance, where the priate sector presence is low. Second, deoling teacher
compensation to the district, along with the creation o a district cadre o teachers in combination with the
113
proincial cadre, could allow salary scales based on the structure o the local market to deelop. 1his would
also support the ongoing deolution process.
5.31 1be otitic. of teacber reforv reqvire. tbe .earatiov of teacber.` aevava. frov roter.` aevava.. \heneer teacher
reorm is discussed, the irst reaction is that it is politically diicult`. \et what does politically diicult`
mean-does the aerage oter vot support teacher reorm In the last year o the LLAPS study, we asked
parents a number o questions about what they wanted rom the goernment and rom their schools. Parents
graded hypothetical schools with dierent characteristics on a scale o 1 ,ery bad, to 5 ,ery good,. 1he irst
school had a roof tbat verer tea/., a ver bovvaar, ratt ava ae./., bvt teacber. rbo rere freqvevtt, ab.evt ava vot
votiratea ava vo free tetboo/. or .cboot .vtie.`. 1he second had a roof tbat tea/., a bro/ev bovvaar, ratt, teacber. rbo
rere freqvevtt, ab.evt ava vot votiratea bvt gare free tetboo/. ava .cboot .vtie.`. 1he third had a roof tbat tea/., a
bro/ev bovvaar, ratt ava ae./., vo free tetboo/. or .cboot .vtie. bvt teacber. rbo rere atra,. re.evt ava bigbt,
votiratea`.
114
5.32 1be va;orit, of arevt. ;igvre :.2) tbovgbt
tbat .cboot. ritbovt aeaicatea teacber. ;bvt ritb rer,
gooa ivfra.trvctvre or free .cboot .vtie.) rere baa or
rer, baa. Close to 80 percent thought those
schools with poor inrastructure and no ree
school supplies but with dedicated teachers are
good or ery good. 1he indings were mirrored
in a separate question where 62 percent o men
and 68 percent o women reported dedicated
teachers` as their top priority in schools with
good acilities` coming a distant second with
13 percent ,men, and 8 percent ,women,. Our
third question asked what the top priority
demand rom the goernment was. Not
surprisingly, 50 percent o men and women
reported jobs` as their top priority. loweer, 20 percent o men and 25 percent o the women reported
their top priority was secondary schools in their illage-ahead o roads, 24-hour electricity, 24-hour water,
and greater security.
G%O.&< ^TYP ?#&<(7- Z#Q.< -:$99Q- 5%7$ )<)%:#7<)
7<#:$<&- Z<&U $%O$QUg
0
20
40
60
80
P
e
r
c
e
n
t
% Reporting Bad/Very Bad % Reporting Good/Very Good
Men Women Men Women
How Parents Rank Schools
(a) Great infrastructure (b) poor teachers (c) no free school su pplies
(a) Poor infrastructure (b) poor teachers (c) free school suppl ies
(a) Poor infrastructure (b) great teachers (c) no free school su pplies
100
0
20
40
60
80
P
e
r
c
e
n
t
% Reporting Bad/Very Bad % Reporting Good/Very Good
Men Women Men Women
Parents Speak: How good is a school with.
(a) Great infrastructure (b) poor teachers (c) no textbooks/notebooks
(a) Poor infrastructure (b) poor teachers (c) free textbooks/notebooks
(a) Poor infrastructure (b) great teachers (c) no textbooks/notebooks
100
0
20
40
60
80
P
e
r
c
e
n
t
% Reporting Bad/Very Bad % Reporting Good/Very Good
Men Women Men Women
How Parents Rank Schools
(a) Great infrastructure (b) poor teachers (c) no free school su pplies
(a) Poor infrastructure (b) poor teachers (c) free school suppl ies
(a) Poor infrastructure (b) great teachers (c) no free school su pplies
100
0
20
40
60
80
P
e
r
c
e
n
t
% Reporting Bad/Very Bad % Reporting Good/Very Good
Men Women Men Women
Parents Speak: How good is a school with.
(a) Great infrastructure (b) poor teachers (c) no textbooks/notebooks
(a) Poor infrastructure (b) poor teachers (c) free textbooks/notebooks
(a) Poor infrastructure (b) great teachers (c) no textbooks/notebooks
100
5.33 ivce arevt. forv tbe bvt/ o roter. iv av, etectiov, ivcrea.ivg teacber accovvtabitit, ava roriaivg .ecovaar, .cboot.
i. a otiticatt, fea.ibte otiov. And it is not as though all teachers would lose rom such reorms. 1he problem
with the current system is that all teachers are treated the same way-regardless o whether they are highly
motiated and hard-working or not. Reorming teacher compensation will beneit teachers who are working
around the clock in diicult circumstances to ensure that children learn. 1he losers are non-perorming
teachers. Pakistan needs to decide whether it can mortgage the uture o millions o children a year to the
demands o a raction o teachers who are not perorming.
115
J%*Q%9O&#R$U
Action Aid Pakistan. 1999. Comparatie Analysis o Public, Priate and NGO Schools.`
Andrabi, 1ahir, Jishnu Das and Asim Khwaja. 2006. A Dime a Day: 1he Possibilities and Limits o Priate
Schooling in Pakistan.` \orld Bank Policy Research \orking Paper 4066, \ashington, DC.
Andrabi, 1ahir, Jishnu Das and Asim Khwaja. 200. Students 1oday, 1eachers 1omorrow Identiying
constraints on the proision o education.` Processed. \orld Bank, \ashington DC. Aailable
at:http:,,ksghome.harard.edu,~akhwaja,papers,teachersstudentssept0.pd
Andrabi, 1ahir, Jishnu Das, Asim Khwaja, and 1ristan Zajonc. 200. Do Value-Added Lstimates Add
Value Accounting or Learning Dynamics.` Processed. \orld Bank, \ashington DC.
Alderman, larold, Jere Behrman, Shahrukh Khan, Daid Ross, and Richard Sabot. 1995. Public Schooling
Lxpenditures in Rural Pakistan: Liciently 1argeting Girls and a Lagging Region.` In D. an de
\alle and K. Nead, eds., Pvbtic evaivg ava tbe Poor: 1beor, ava riaevce. Baltimore, Md.: Johns
lopkins Press.
Alderman, larold, Jere Behrman, Shahrukh Khan, Daid Ross and Richard Sabot. 1996. "Decomposing the
regional gap in cognitie skills in rural Pakistan." ]ovrvat of ..iav covovic., ol. ,1,: 49-6.
Ari, Charagh, Saima Chaudhary and Uzma Gilani. 1999. Learning Achieement o Grades 3 and 5 Children
in Rural Primary Schools,` Punjab Curriculum Deelopment and Research Centre. October.
Balochistan Bureau o Curriculum and Lxtension Centre. 1999. Learning Achieement o Grade 3 and 5
Children in Rural Primary Schools o District Pishin, Balochistan,` December.
Banerjee, Abhijit, Shawn Allen Cole, Lsther Dulo and Leigh L. Linden. 2005. "Remedying Lducation:
Lidence rom 1wo Randomized Lxperiments in India," NBLR \orking Paper No. \11904.
December.
Banerjee, Abhijit and Lsther Dulo. 2006. "Addressing Absence," ]ovrvat of covovic Per.ectire., ol. 20,1,:
11-132. \inter.
Behrman, Jere, Shahrukh Khan, Daid Ross and Richard Sabot. 199 "School quality and cognitie
achieement production: A case study or rural Pakistan," covovic. of avcatiov Rerier, ol. 16,2,:
12-142.
Bracey, Gerald \. 2000 1he 1IMSS linal \ear` Study and Report: A Critique.` avcatiovat Re.earcber, ol.
29,4,: 4-10.
Chaudhury, Nazmul and Dilip Parajuli. 2006. Conditional Cash 1ransers and lemale Schooling: 1he
Impact o the lemale School Stipend Program on Public School Lnrollments in Punjab, Pakistan.`
\orld Bank Policy Research \orking Paper No. 4102, \ashington, DC.
Cheng, \in Cheong. 199 Monitoring School Lectieness: Conceptual and Practical Possibilities and
Dilemmas in Deeloping a lramework.` Paper presented at the Asia-Paciic Lconomic Cooperation
Lducation lorum on School-Based Indicators o Lectieness ,Quilin, China, April 13-16,.
116
Das, Jishnu, Stean Dercon, James labyarimana and Pramila Krishnan. 2005. 1eacher Shocks and Student
Learning: Lidence rom Zambia` \orld Bank \orking Papers No. \PS3602, \ashington, DC.
Dawn, 2006. Lducation reorms 2nd phase begins` 1be vtervet aitiov, Noember 14.
Aailable at: http:,,www.dawn.com,2006,11,14,nat13.htm
Dulo, Lsther and Rema lanna. 2005. "Monitoring \orks: Getting 1eachers to Come to School" CLPR
Discussion Paper No. 5426. December.
Aailable at SSRN: http:,,ssrn.com,abstract~895584
LDI. 1999. Basic Skills Assessment - A Pakistan Case Study.` 1heme 3 at LDI Seminar helt at Bhurban
,lebruary -10,.
Gazdar, laris. 2000. State, community and uniersal education: A political economy o public schooling in
rural Pakistan` Asia Research Centre, London School o Lconomics, October.
Ghuman, Sharon and Cynthia Lloyd. 200. Poerty, Gender, and \outh,` Population Council \orking
Paper No. 1.
Glewwe, Paul, Michael Kremer and Nauman Ilias. 2003. 1eacher Incenties` NBLR \orking Paper No.
\961. Aailable at SSRN: http:,,ssrn.com,abstract~403642
laque, Manzoorul, Shari Kalhoro and Shoukat Saeed. 2000. Baseline Surey o Learning.` Sindh Bureau
o Curriculum Deelopment and Lxtension \ing. May.
lolmes, Jessica. 2003. Measuring the Determinants o School Completion in Pakistan: Analysis o
Censoring and Selection Bias.` avcatiov covovic., ol. 9,3,:23-44.
loodbhoy, Perez. 1998. avcatiov ava tbe tate: ift, Year. of Pa/i.tav, Oxord Uniersity Press,
Karachi: Oxord.
lowie, Sarah J. and 1jeerd Plomp. 2001. Lnglish Language Proiciency and Other lactors Inluencing
Mathematics Achieement at Junior Secondary Leel in South Arica.` Paper presented at the
Annual Meeting o the American Lducational Research Association ,Seattle, \A, April 10-14,.
loxby, Caroline. 1996. low 1eachers' Unions Aect Lducation Production,` Qvartert, ]ovrvat of covovic.,
ol. 111,3,.
IAL. 2001a. 1IMSS Assessment lrameworks and Speciications 2003.`
http:,,timss.bc.edu,timss2003i,t03pd,103_Al_Booklull.pd
IAL. 2001b. lrameworks and Speciications or PIRLS Assessment 2001.`
http:,,timss.bc.edu,pirls2001i,pd,PIRLS_rame2.pd
Lloyd, Cynthia, Cem Mete and Zeba Sathar. 2005. 1he Lect o Gender Dierences in Primary School
Access, 1ype, and Quality on the Decision to Lnroll in Rural Pakistan,` covovic Deretovevt ava
Cvttvrat Cbavge, ol. 53,3,: 685-10.
Kain, John, Lric lanushek and Steen Rikin. 2005. 1eachers, Schools, and Academic Achieement.`
covovetrica, ol 3,2,: 41-458. March.
11
Kardar, Shahid. 1995. Demand or Lducation Among Low and Middle Income louseholds in Karachi and
Lahore.` Mimeo. 1he \orld Bank, \ashington, DC.
Khan, Shahrukh, Sajid Kazmi, and Zainab Lati. 2005. A Comparatie Institutional Analysis o Goernment,
NGO and Priate Rural Primary Schooling in Pakistan.` 1be vroeav ]ovrvat of Deretovevt Re.earcb,
ol. 1,2,: 199 - 223.
Khan, labib, Seyed Shah, Mirza Ahmad, Nasir Amin, lumala Khalid and Ahmad Malik. 1999. Measuring
Learning Achieement at Primary Leel in Pakistan` Pakistan Ministry o Lducation ,Academy o
Lducational Planning and Management,, Islamabad.
Kingdon, Geeta and lrancis 1eal, 2004. "Does perormance related pay or teachers improe student
perormance Some eidence rom India," Deretovevt ava Cov ,.tev. 0409009, Lcon\PA.
Kizilbash, lamid. 199. \orking with Rural Primary 1eachers: linal Report on Improing the Quality o
Rural Primary School 1eachers in Pakistan.` Ali Institute o Lducation, Lahore, Pakistan.
Mirza, Munawar and Abdul lameed. 1994. Dierential Achieement o Primary Age Students and the Cost
Lectieness by School 1ype.` Primary and Non-lormal Lducation \ing, Ministry o Lducation.
Multi-Donor Support Unit or the Social Action Programme. 1995. Determinants o Primary Students
Achieement - National Surey Results.` Islamabad, July.
Muralidharan, Karthik and Venkatesh Sundararaman. 2006. 1eacher Incenties in Deeloping Countries:
Lxperimental Lidence rom India.`
Aailable at: http:,,www.people.as.harard.edu,~muralidh,
Murgai, Rinku and Lant Pritchett, 2006. 1eacher Compensation: Can Decentralization to Local Bodies 1ake
India rom the Perect Storm through 1roubled \aters to Clear Sailing` Mimeo. 1he \orld Bank,
\ashington, DC.
NALP. ,undated, Mathematics lramework or the 1996 and 2000 National Assessment o Lducational
Progress.` http:,,nces.ed.go,transer.asplocation~www.nagb.org,pubs,math96-2000.pd
NALP. ,undated, Reading lramework or the National Assessment o Lducational Progress: 1992-2000,.`
http:,,nces.ed.go,transer.asplocation~www.nagb.org,pubs,read92-2000.pd
NALP. ,undated, \riting lramework and Speciications or the 1998 National Assessment o Lducational
Progress.` http:,,nces.ed.go,transer.asplocation~www.nagb.org,pubs,writing.pd
Naseer, larooq and Manasa Patnam. 200. Are child centred classrooms eectie Impact o the CRI
program on student learning outcomes in Pakistan.` Mimeo. Lahore Uniersity o Management
Sciences.
N.\.l.P. Bureau o Curriculum Deelopment and Lxtension Serices. 1999. Learning Achieement o
Grade 3 and 5 Children in Rural Primary Schools.` December.
PAKIS1AN. ,undated, Strategic lrameword or National Lducation Assessment 2001`
Pakistan Country Gender Assessment: Bridging the Gap: Opportunities and Challenges. 2005. Lnironment
& Social Deelopment Sector Unit, South Asia Region, \orld Bank. October.
118
Perez, Muhammad. 1995. Basic Competencies o Children in Pakistan.` UNICLl: Islamabad, March.
Punjab Lducation loundation. ,undated, Promoting Quality Lducation 1hrough Public Priate
Partnership` Aailable at: siteresources.worldbank.org,IN1INDIA,Resources,Kardar.ppt
Punjab Literacy \atch. 1999. Leels o Pupil Achieement in Primary Schools o Punjab: A Sample Study.`
UNLSCO. October.
Rahman, 1ariq. 199. avgvage ava otitic. iv Pa/i.tav. Oxord Uniersity Press, USA. Noember 13.
Rahman, 1ariq. 2005. Deviev. of .tiev !orta.: . tva, of avcatiov, veqvatit, ava Potariatiov iv Pa/i.tav.
Oxord Uniersity Press, USA. January 20.
Raallion, Martin and Quentin \odon. 2000. Does Child Labor Displace Schooling Lidence on
Behaioral Responses to an Lnrolment Subsidy.` covovic ]ovrvat, ol. 110: C158-C16.
Sathar, Zeba and Cynthia Lloyd. 1994. \ho Gets Primary Schooling in Pakistan Inequalities among and
within amilies.` Pa/i.tav Deretovevt Rerier, ol. 33,2,: 103-34.
Sathar, Zeba, Cynthia Lloyd, and Minhaj ul laque. 2000. Inestment in Children`s Lducation and lamily-
Building Behaior in Pakistan: lindings rom Rural N\lP and Punjab.` Islamabad Population
Council.
Sanders, \illiam. 1998. 1evve..ee 1atve.aaea ...e..vevt ,.tev Online. Aailable at :
http:,,www.shearonorschools.com,1VAAS_index.html
Sanders, \illiam and June Riers. 1996. Cumulatie and Residual Lects o 1eachers on luture Student
Academic Achieement`, |virer.it, of 1evve..ee 1atve.aaea Re.earcb ava ...e..vevt Cevter Re.earcb
Progre.. Reort.
Sawada, Yasayuki and Michael Lokshin. 2001. lousehold Schooling Decisions in Rural Pakistan.` Stanord
Uniersity, Lconomics Department, Stanord, CA.
Shah, Mubarik lussain. 1984. National Achieement 1ests ,Protocol Analysis,.` Primary and Non-lormal
Lducation \ing, Ministry o Lducation. Islamabad.
1irole, Jean. 1994. "1he Internal Organization o Goernment," Ofora covovic Paer., Oxord Uniersity
Press, ol. 46,1,: 1-29. January.
UNLSCO 2001 Learning Achieement in Primary Schools o Pakistan: A Quest or Quality Lducation.`
UNLSCO, Islamabad, in collaboration with the Ministry o Lducation Curriculum \ing. March.
Vegas, Lmiliana, ed. 2005 vcevtire. to vrore 1eacbivg: e..ov. frov ativ .verica, \orld Bank Publications,
\ashington, DC. August.
119
'((<V<-
'((<V ;P @#0RQ%(O ?&9:<).&< 9S 7$< =>'?@ R&92<:7
1. 1his annexure describes the sampling procedure or the LLAPS project. Samples were drawn or
illages, schools, and households and each are described in turn.
A9(-7&.:7%9( 9S 7$< Z%QQ#O< #() -:$99Q -#0RQ<
2. 1he illage and school sample was constructed in two steps.
! Step 1: A preliminary sample o 125 illages, with 41-42 illages in each o the 3 districts,
was randomly drawn rom a list rame o illages with at least one priate school ,as reported
in the priate school census inormation lBS, 2000,.
Mauza Border
Out of Sample School Sample School Household Circle
G%O.&< ';T;P @#0RQ%(O 9S @:$99Q- %( =>'?@ Z%QQ#O<-
! Step 2: 1he team conducted a surey o all primary schools ,public and priate, in the illage
and within a 30-minute walk o the illage boundary. A choice set` in each illage is deined
as schools in the illage and within a easible` walking distance o the illage. In Attock and
laisalabad this distance was determined to be within a 15-minute walk o the illage whereas
120
in Rahim \ar Khan, which is less densely populated, this distance was determined to be
within 30 minutes o the illage.
25
1he igure aboe shows how this exercise was conducted
oer a 6-month period. 1he two schools in red are not in the sample, since ,a, they are
outside the illage boundary and ,b, they are not within 15-minute walking distance o any
household in the illage. All the schools in blue are in the choice set`, in particular, two
schools outside the illage would still be chosen because they are within 15 minutes o .ove
household in the illage. All illages without a priate school or with greater than 24 public
and priate schools in their choice set were excluded, proiding the inal sample o 112
illages.
3. 1able A1.1 uses 1998 census data to compare the surey sample to rural Punjab, the three districts,
and illages in the three districts with a priate school. 1here are seeral obserations. lirst, illage
population, literacy, and access to inrastructure and schools in the three districts are ery similar to that o
rural Punjab as a whole ,Columns 1 and 2,. Second, in accordance with preious obserations on the location
patterns o priate schools, illages in these three districts with a priate school tend to be larger, more
literate, and with more access to inrastructure ,Columns 1 and 3,. linally, the LLAPS project illages look
ery much like other rural illages in these three districts with a priate school, although they are smaller, both
in population and in geographical land area-the last ollows rom our systematic elimination o illages with
more than 25 schools.
25
Lnrollment data based on a census o all households in the preliminary sample support this inding. \hile in Attock and laisalabad
91 percent and 94 percent o grade 3 enrollment is in schools in the illage or within 15 minutes, in Rahim \ar Khan the percentage
enrollment or the same distance is signiicantly less at 85 percent. By including schools within a 30-minute walk o the illage or
Rahim \ar Khan comparable enrollment coerage o 91 percent is obtained.
121
"#*Q< ';T;P @#0RQ< @.00#&U @7#7%-7%:-
"#*Q< ';T;P 6%QQ#O< A$#&#:7<&%-7%:-
@7#7%-7%:-
[;\
[Y\
'"4+
G@/+
CE4
[]\
'"4 G@/+
CE4 5%7$
?&%Z#7< @:$99Q
[c\
;Yi @#0RQ<
6%QQ#O<-
YieY Yee^ cff; c;Y^
.e;veav) 16 129 228
6%QQ#O<
?9R.Q#7%9(
^ 24531 2491 611
I<#( ]f ]c c^ cY
.e;veav) 0 0 0 1
?<&:<(7 ').Q7
=%7<&#:U
^ 2491 611 110
I<#( eT`` fT;^ fT;^
.e;veav) 0.01 0.02 0.04 0.10 L9.-<$9Q) @%3<
^ 24501 2486 611 110
;e]h ;^`; ;`ec ;`h;
.e;veav) 19 30 93
6%QQ#O< '&<#
[':&<-\
^ 2403 2115 481
I<#( iTi;h iTiY; iTiY^ iTiY;
.e;veav) 0.000 0.001 0.001 0.003
G&#:7%9( L9.-<-
5%7$ ?97#*Q<
K#7<& ^ 1905 59 103
I<#( iTihf iTi`e iT;i]
.e;veav) 0.000 0.001 0.001 0.004
G&#:7%9( L9.-<-
5%7$ >Q<:7&%:%7U
^ 21091 2141 60 109
iT^Y; iT^i] iTecY iTecf
.e;veav) 0.002
G&#:7%9( ?#::#
[A<0<(7\ L9.-<-
C.&#Q
?.(2#*
I<#(
49
110
24531
fT;h
I<#(
19
4
1689
iT;;i
I<#(
0.006 0.010 0.022
^ 23260 2409 611 110
I<#( iT]h iTc] ;The ;Te`
.e;veav) 0.01 0.02 0.0 0.11
D.0*<& 9S
?&%Z#7< @:$99Q- %(
6%QQ#O< ^ 25836 20 612
0<#( YT]] YTf; ]Tc] YTh^
.e;veav) 0.01 0.05 0.12 0.20
D.0*<& 9S
M9Z<&(0<(7
@:$99Q- %( 6%QQ#O< ^ 19501 2111 604 109
110
122
1able A1.2 presents a more recent comparison o enrollments in the three districts o the study and the rest
o Punjab, based on PSLM data. 1he table shows the net enrollment rates or children aged 6-10 and the
percentage in priate schools or the three districts o the study. 1hese data conirm that the stratiication into
North, Central, and South relects existing education data-Net enrollments are ar higher in the north and
the center ,Attock and laisalabad, than in Rahim \ar Khan ,South,. It also conirms that the center o
Punjab, with 42 percent o children enrolled in priate schools ,26 percent in rural laisalabad, is the heartland
o priate schooling, ollowed by Attock and laisalabad. As beore, the choice o the three ai.trict. is airly
representatie o the proince as a whole.
"#*Q< ';TYP >(&9QQ0<(7 *U /%-7&%:7
Urban Rural 1otal
District Indicator
M l 1otal M l 1otal M l 1otal
in Priate Schools 38 42 40 13 12 12 1 1 1
Attock
NLR, 6-10 85 83 84 9 82 80 80 82 81
in Priate Schools 62 60 61 2 26 26 41 42 42
laisalabad
NLR, 6 81 9 80 1 61 66 5 68 2
in Priate Schools 42 45 43 1 13 15 22 21 22 Rahim \ar
Khan NLR, 6 6 68 2 49 39 44 54 44 49
in Priate Schools 55 52 53 23 22 23 31 32 31
Rest o Punjab
NLR, 6 6 6 6 58 63 69 63 66
in Priate Schools 55 52 54 22 22 22 32 32 32
1otal
NLR, 6 6 66 58 62 69 63 66
A9(-7&.:7%9( 9S 7$< $9.-<$9Q) -#0RQ<
4. In each sample illage a total o 16 households were sureyed. louseholds were picked randomly
rom two strata with 12 households in the irst and 4 in the second strata. 1he list rame or sampling
households was obtained rom a census o all households in the illage with at least one child between the
ages o 5 and 15. 1he two populations sampled were:
! Strata 1: louseholds with at least one child eligible or and enrolled in grade 3
! Strata 2: louseholds with at least one child eligible or but not enrolled
5. 1he ocus on grade 3 was in order to maximize the potential match between children that were tested
in the schools ,grade 3 only, and the children included in the household surey.
123
6. In practice, it was not possible to obtain
these two strata since eligibility or grade 3` is not
a readily deined concept in the enironment. \hile
age cut-os could be drawn, as igure A3 shows,
there was a large degree o ariation in reported age
or children enrolled in grade 3, with a signiicant
majority between the ages o 8 and 10. Moreoer,
households that enrolled their children earlier or
later in grade 3 were dierent: \ounger children in
grade 3 are rom wealthier and more educated
households. Neertheless, comparisons between
households with an out-o-school child between the
ages o 5-15 with households reporting an out-o-
school child between the ages o 8-10 showed no
dierences in household expenditure, land, or the
head o household`s education.
G%O.&< ';T;P 'O< /%-7&%*.7%9( 9S A$%Q)&<( %(
AQ#-- ]
. Gien the concerns with ollowing a wider age group, it was inally decided to adopt the 8-10 cut-o
and reweight based on the proportions in the census population. 1he household sample cannot be used, or
example, to compute ,say, aerage poerty in these illages-it is ,at most, representatie o households with
a child eligible or grade 3.
124
'((<V Y P A$%Q)&<( K$9 A$#(O< @:$99Q-
"#*Q< 'YT;P "$< @7#U<&- #() 7$< JUR#--<&-
M9<- 79 AQ9-<-7 >Q%O%*Q< JUR#--<-
D97
>(&9QQ<)
?.*Q%:
@:$99Q
?&%Z#7<
@:$99Q
?.*Q%:
S9&
R.*Q%:
?&%Z#7<
S9&
?.*Q%:
?.*Q%:
S9&
?&%Z#7<
?&%Z#7<
S9&
?&%Z#7<
"97#Q
D.0*<& 9S
H*-<&Z#7%9(-
1,369 82 414 668 1,443 386 421 5,528
l 9S #QQ :$%Q)&<( 24.6 14.96 .49 12.08 26.1 6.98 .62 100
L9.-<$9Q)
A$#&#:7<&%-7%:-
Mother's
Lducation
,\ears,
0.415 0.936 2.319 0.96 1.469 2.10 2.02 1.141
lather's
Lducation
,\ears,
1.99 3.21 4.44 3.455 3.961 5.23 5.8 3.36
Per Capita
Lxpenditure
613.322 691.69 819.031 60.34 80.996 94.086 903.212 22.836
lousehold
\ealth
-0.915 -0.113 0.165 -0.301 -0.021 0.58 0.605 -0.296
A$%Q)
A$#&#:7<&%-7%:-
Age 10.183 9.513 9.199 9.938 9.958 9.3 9.436 9.83
Male 0.393 0.564 0.46 0.664 0.5 0.66 0.53 0.51
Aboe Aerage
Intelligence
0.131 0.216 0.389 0.329 0.315 0.345 0.391 0.249
Below Aerage
Intelligence
0.221 0.126 0.09 0.059 0.098 0.062 0.042 0.135
Birth Order 3.229 3.506 3.098 3.353 3.261 3.044 3.222 3.266
1otal
Lducational
Lxpenditures
0.461 130.603 234.628 212.649 10.592 341.344 436.19 138.991
@:$99Q
A$#&#:7<&%-7%:-
Dierence in
Lnglish 1est
Scores
0.12 -1.031 1.068 0.024 -0.28
Dierence in
Urdu 1est scores
0.083 -0.624 0.838 -0.01 -0.163
Additional
Distance
1raeled
,Median,
0.503 0.213 0.191 0.148 0.095
125
"<:$(%:#Q '((<V<-
"<-7 A9(-7&.:7%9(+ A$#&#:7<&%-7%:- #() '(#QU-%-
Ye
A1.1
A1.2
1o construct the LLAPS test, a easibility study was conducted in June,July, 2002 to build the
knowledge and capacities to conduct the more extensie LLAPS test. 1his appendix proides background
inormation or the test instrument, the rationale behind the test design, a description o the test content, a
ormal alidation o each indiidual test and notes on issues that arose during the test`s administration and
preliminary data analysis. It is organized as ollows. Section A2 summarizes assessments o primary education
undertaken in Pakistan oer the past two decades, till 2003. 1he summary is based on published studies and
publicly aailable documents, as such, it may miss small-scale studies that are hard to access or recent tests
whose details hae not been publicly released. Section A3 proides an oeriew o the assessment instrument,
discussing both the test design and content. It then coers the procedural and implementation issues aced
during the administration o the pilot test instrument and a dierse range o concerns related to the
interpretation o the results. 1his section includes a summary o the actions that were taken in the inal
LLAPS test to address these issues and concerns. Section A4 presents an assessment o the testing
instrument using Item-Response 1heory to examine the alidity o each question ,henceorth item, as well as
the precision o each test and the set o item parameters.
1T ?C>61HB@ @"B/1>@
Since 1984, at least 19 assessments o primary education in Pakistan hae been conducted. 1hese
eorts gained momentum ater the \orld Declaration on Lducation or All ,LlA, in 1990 and again with
the joint UNLSCO-UNICLl global initiatie or Monitoring Learning Achieement in 1996. 1he studies
hae been both national and proincial in scope and ocused on arious competencies,content areas.
UNLSCO ,2001, proides an excellent summary o preious assessment work in Pakistan and the summaries
below draw rom this document.
26
This document is based on the test Ieasibility report prepared by the authors oI this report together with Duriya Farooqi, who
was at Harvard University at the time oI the report preparation.
126
>#&QU @7.)%<- & ;h`i-
A1.3
A1.4
A1.5
A1.6
A1.
1he irst two major assessments o primary education in Pakistan were the \orld Bank`s Primary
Lducation Project in 1984 and the BRIDGLS project o the larard Institute o International Deelopment
in 1988-89.
1he Primary Lducation Project study compared Science and Mathematics achieement o 3,300
students o grades 4 and 5 in a representatie sample o project and non-project schools in Punjab, Sindh and
the North-\estern lrontier Proince ,N\lP,. Shah ,1984, summarizes the results. In all three proinces,
girls scored higher in sciences while boys scored higher in mathematics. loweer, achieement or all groups
was low and based on these results the author suggests that schools should primarily ocus on ensuring that
students acquire basic competencies rather than increasing peripheral luxuries in the curriculum. Interestingly,
the correlation between proessional teacher qualiications and test outcomes was airly weak, with a
signiicant positie correlation only in one proince.
1he BRIDGLS project collected student achieement data rom a random sample o about 11,000
students in grades 4 and 5 and 1,000 teachers rom 500 schools using the same instrument as the Primary
Lducation Project study described aboe. 1he test showed a decline in both Science and Mathematics scores
between 1984 and 1989. 1he study also collected inormation on student and teacher background
characteristics and classroom practices, in an eort to relate these teacher characteristics to student outcomes.
=#7<& @7.)%<- o ;hhi-
1esting actiity in Pakistan increased signiicantly during the 1990s as a result o the Lducation lor
All ,LlA, declaration and the associated UNLSCO-UNICLl initiatie. 1hese studies ocused on a broad
ariety o topics. 1he results are briely oeriewed below.
Mirza and lameed ,1994, explored the eectieness o arious school types. 1hese types were: 1,
mosque school, 2, two teacher primary school, 3, ie teacher primary school, 4, primary section with middle
school and 5, primary section with high school. 1hree test instruments were administered. 1he irst two
coered Mathematics, Science, Social Studies and Dinyat ,the study o the practical laws and ideology o
Islam,. 1he third attempted to assess students` behaior as measured by classroom participation, motiation,
cooperation and socialization, discipline, cleanliness, careulness and regularity and punctuality. 1he sample
consisted o 15,991 students in grades 3 and 5 rom 42 schools selected rom the our proinces o
Balochistan, M.\.l.P., Punjab and Sindh. Mosque schools had the lowest gross achieement and gross
behaior scores. Primary sections o middle schools perormed best on the achieement score and primary
12
sections o high schools ranked irst or behaior. lie teacher primary schools were the most cost-eectie
in terms o cognitie achieement. loweer oerall, the correlation between achieement scores and per-
student cost was extremely weak.
A1.8
A1.9
Perez ,1995, attempted to determine the degree to which students possess basic competencies. 1he
testing instrument was a semi-structured questionnaire coering lie skills, reading competence, writing
competence, counting and arithmetic, mental arithmetic and loly Qur`an. 1he sample consisted o a
representatie sample ,multi-stage, systematic-random sampling design, o 1,241 rural and 1,341 urban 11-12
year olds. One o the unique characteristics o Perez`s study was the use o a household rather than school
based sampling rame. 1hus, children rom the appropriate age group were sampled rom households
irrespectie o whether they were currently enrolled in school or not-as a result, the test outcomes are a
snapshot o learning achieement or the entire age-speciic population, rather than a selected school-going
group. Perhaps as a result o this unique sampling rame, only 20. o children were competent at leels
considered basic. Competency was lowest or letter writing whereas numerical skills, arithmetic, rote-reading
and writing rom dictation were deemed acceptable. As a result, Perez ,1995, concludes that Pakistani
schools should shit away rom teaching only rote-memorization-based skills.
2
1he national surey conducted by the Multi-Donor Support Unit or the Social Action Program in
1995 aimed at determining the critical ariables impacting academic achieement at the primary school leel.
Basic inormation was collected on 52 schools throughout the country. Academic achieement in
mathematics, general knowledge and comprehension was tested or 914 teachers and 11,563 students in grade
5 ,although data rom the Sindh proince was later discarded,. 1he test instruments were based on textbook
materials or grades 3 and 4. 1he surey ound student perormance to be satisactory. 1here were three
interesting correlations reported rom this study:
" 1eacher qualiications improed student scores ,teacher perormance aried across proinces rom a
high o 91 in Punjab to a low o in the Northern Areas ,lANA,. 1his eect was particularly
pronounced in rural areas where trained teachers accounted or a 12 percent increase in student
scores.
" Students taught in mixes settings or by emale teachers generally outperormed their peers.
" Priate school consistently scored better than goernment schools.
27
Since the document does not provide detailed breakdowns by the schooling status oI the child, we are unclear Ior the basis oI
this statement.
128
A1.10
A1.11
A1.12
A1.13
A1.14
1he NGO Action Aid Pakistan conducted an assessment o Mathematics, Urdu and General
knowledge on 965 students in 50 schools sampled rom six districts in the our major proinces and AJK. In
addition, the study included ocal group sessions as well and interiews with community leaders. Priate
schools perormed signiicantly better than NGO and goernment schools in all categories. 1he dierence
between NGO and goernment schools was negligible. 1hese results matched the opinions expressed in
ocal groups and interiews.
Khan et al ,1999, assessed learning achieement in Science, Mathematics, and Language ,Urdu, or
grade 4 students. 1he test instruments were based directly on the curriculum. In total, the sample consisted
o 145 schools and 294 students rom 28 districts ,he sample was not random,. 1he results mirrored earlier
studies. Girls perormed better in Science and Urdu whereas boys were better in Mathematics. Urban
students scored higher on all three sections than rural students and children in Sindh ranked irst, ollowed by
Punjab and N\lP and, signiicantly lower, lA1A, lANA, AJK and Balochistan.
Ari et al ,1999, explored the actors expounded by head teachers, teachers, learning coordinators,
parents and students or low educational outcomes. Additionally an achieement test in Social Studies,
Mathematics and Urdu was gien to 200 students. Generally head teachers criticized the standard o teaching,
absenteeism o teachers, and lack o support proided to students at home. 1eachers predominantly blamed
low perormance o the lack o adequate acilities and physical resources. Learning coordinators also ocuses
on the lack o physical resources in addition to teacher absenteeism and, poor school administration and non-
cooperation between teachers and the community.
Studies by laque et al ,2000,, the Bureau o Curriculum Deelopment and Lxtension Serices,
N.\.l.P, ,1999, and the Bureau o Curriculum and Lxtension Centre, Balochistan, ,2000, hae similarly
explored the actors responsible or poor perormances. Questionnaires in these studies were submitted to
head teachers, teacher and learning coordinators along with achieement tests or students. Complaints were
similar to those ound by Ari et al ,1999,.
Punjab Literacy \atch ,1999, tested students rom 31 schools ,16 boys public schools, 14 girls public
schools and one co-educational priate school,, 822 boys and 549 girls, to determine competency in
Mathematics and Urdu. lor the choice o schools, District Lducational Oicers rom two districts in each o
Northern, Central and Southern Punjab were asked to select equal numbers o good, aerage and weak
schools. 1he test instruments were designed based on the complete common curriculum. Students did well
on simple and mechanical mathematical problems and tasks like joining letters into words, making sentences
and comprehension. In contrast, students perormed poorly on geometry, ractions, decimals, and problems
129
inoling thinking or application o knowledge as well as grammar, letter writing and arrangements in
alphabetical order.
A1.15
A1.16
A1.1
1his array o tests and outcomes is summarized in the table in Annexure 1. A surprising act that
emerges rom the summary below is that although a large number o assessments hae been undertaken in
Pakistan, there has been little ,i at all, any, coordination between the dierent testing bodies, and as a result,
there is no systematic reporting o inormation. lor instance, UNLSCO ,2000, in their document on
Assessing Learning Achieement speciies that a good surey` should include at the ery minimal, careul
documentation o the sampling methodology ,with standard errors,, as well as instrument construction and
alidation. As our summary below shows, only in a ery ew cases are we able to learn these essential test-
characteristics rom aailable documentation.
28
As a result, there is almost no comparability o these dierent
tests- while they each proide an indiidual snapshot that can be used to understand ariation ritbiv the
testing sample, they cannot be used to urther our understanding o changes in achieement oer time ,or
instance, we hae no idea whether students score higher in one Urdu test compared to another because o
learning or because o dierences in the test instrument,.
11T "L> =>'?@ ">@" 1D@"CBI>D"
Our description o the test instruments as well as the testing enironment is organized as ollows: \e
start with an oeriew o the aims o the test instruments as well as the general principles used in the oerall
selection o test items. \e then outline the basic structure o the three tests, with speciic emphasis on the
content domain and a discussion o our instrument in the context o testing rameworks that hae been
deeloped by other organizations.
K$#7 -$9.Q) 5< 7<-7X
Schools sere multiple purposes and teach a dierse range o subjects, skills as well as morals and
attitudes ,see Box A.1,. As such, measuring their output is diicult and controersial. 1he test instrument
deeloped here attempts to roughly quantiy a small set o a school`s output. Accordingly, the test only
includes sections on Lnglish, Urdu, and Mathematics. Although ar rom measuring the total output` o a
school, or een the educational deelopment o students, such an instrument is still useul or numerous
purposes.
28
\e are currently in the process o trying to contact each indiidual author to update this reiew, but in a number o cases,
instruments hae been discarded and published documents remain the only source o inormation.
130
J9V 'T;P I.Q7%RQ< @:$99Q >SS<:7%Z<(<--
Schools sere a multiplicity o unctions inoling indiidual, institutional, community, national and
international leels. As a consequence, assessing quality or eectieness is both diicult and
controersial. lor example, one school may deelop students with a strong sense o ciic responsibility
whereas another might excel at producing talented scientists. Since any assessment instrument
implicitly supports a conceptualization o eectieness, determining which roles to measure is an
important consideration during the test construction stage. Cheng ,199, classiies the potential school
unctions into ie types: tecbvicat,ecovovic fvvctiov., bvvav,.ociat fvvctiov., otiticat fvvctiov., cvttvrat fvvctiov.,
and eavcatiovat fvvctiov.. In turn, these unctions create ie corresponding measures o school
eectieness.
"<:$(%:#Qm>:9(90%: @:$99Q >SS<:7%Z<(<--P 1he degree to which schools contribute to the
technical or economic deelopments. 1his ranges rom the indiidual leel-skills and job training-
all the way to the international leel-the high quality orces needed or eicient economic cooperation
and competitions.
L.0#(m@9:%#Q @:$99Q >SS<:7%Z<(<--P 1he extent to which schools contribute to human and social
deelopment throughout society. Schools proide such things by promoting psychological
deelopment or indiiduals, nurturing human relationships within the institution, supporting the social
needs o the community, acilitating social integration in society and ostering international social
cooperation and riendship.
?9Q%7%:#Q @:$99Q >SS<:7%Z<(<--P 1he ability o schools to contribute to political deelopments at
dierent leels o society. 1his includes promoting ciic responsibility and attitudes within the
citizenship, proiding a enue or political discourse, sering the political needs o the community,
encouraging democracy throughout society, and securing peace and stability at the international leel.
A.Q7.&#Q @:$99Q >SS<:7%Z<(<--P 1he degree to which schools contribute to cultural transmission and
deelopment o society. At the indiidual leel, this consists o the socialization with alues, norms
and belies. At higher leels, schools sere as an epicenter or cultural transmission, reproduction and
understanding.
>).:#7%9(#Q @:$99Q >SS<:7%Z<(<--P 1he extent to which schools contribute to the deelopment and
maintenance o education. 1his type o eectieness reers to the role o schools in teaching students
how to learn to learn rather than simply the educational attainment o pupils. Similarly, at higher leels
the ability reers to the sel-reinorcing power o schools-schools beget more support and resources
or educational systems. lor example, schools increase the supply o teachers.
C#7%9(#Q S9& "<-7%(O B&).+ >(OQ%-$ #() I#7$<0#7%:-
A1.18
1he inclusion o Urdu and Mathematics-the staple subjects o preious assessments-was made or
primarily three reasons. lirst, literacy and Mathematics are two important competencies supplied by primary
schools. Second, literacy and Mathematics are generally more standardized than other subjects and hence
lend themseles to greater cross-school comparability.
29
1hird, most subjects besides Urdu and Mathematics
29
lor example, due to the nature o the subject matter, the progression o social studies and lie sciences can easily dierge
signiicantly rom curricular guidelines. Indeed, een the curricular guidelines or social studies indicate a substantial degree o
lexibility, stating much o the learning o pupils in the early years should be based on direct experience and practical actiities,
achieed as ar as possible through the exploration o their immediate enironment.`
131
are based almost exclusiely on rote-memorization at the primary school leel. lence, the chosen subjects
hae the additional beneit o testing reasoning and logical analysis as well as critical and complex thinking.
A1.19
A1.20
A1.21
A1.22
1he inclusion o Lnglish separates this assessment rom most preious studies. 1he primary ground
or its inclusion is that it can be considered a core competency, which is now part o the oicial curriculum o
the country. 1here are also more subtle reasons. lor instance, there may be a speciic connection between
Lnglish achieement and priate school ees. 1his suggestion is motiated by the prealence o schools
adertising Lnglish language instruction. Lnglish instruction may also sere as an important signaling
mechanism indicating quality or both employers and parents. lor these reasons, the assessment o Lnglish
achieement is particularly interesting.
/<Z<Q9R0<(7 #() @<Q<:7%9( 9S "<-7 17<0- _ HZ<&#QQ
1he principles and rameworks prealent in the learning assessment literature hae been useul as
guiding principles or designing the test but they hae not been treated as rigid structures that the test must
conorm to. Many o the rameworks are ormulated in the context o deeloped countries and thus are not
entirely well suited or settings in which the national distribution o achieement is dramatically dispersed
,een across the same grade,. In the design o the pre-test, we chose not to administer a criterion-reerenced
test since an explicit goal in the uture or such tests would be to relate test outcomes to educational inputs
,such as teacher quality, that we beliee are important. As such, it was important that the test should measure
learning with high precision leels at att teret. of /vorteage- while a criterion-reerenced test would distinguish
sharply between students who meet ,do not meet, the speciied criteria, it may not yield any inormation o
those below ,or aboe, the critical leel.
1he use o a norm-reerenced test creates special needs in the case o Pakistan: although the test will
be administered to Pakistani children who hae all completed the third grade, there is wide ariation in
learning across schools and proinces, and it is expected that many o the children tested may not be amiliar
een with the content o the irst grade curriculum. 1his places special demands on test construction een i a
norm-reerenced test is used.
Speciically, to use a norm-reerenced test properly, we need to irst know what the lower and upper
bounds o learning in the grades to be tested are, this is particularly important in the case o alue-added
assessment, where the test has to try and coer children at all leels o knowledge in the population. Because
o the huge dierences in learning across schools, it is important to ascertain careully the questions that can
be used in a norm-reerenced test. Prior to the design and administration o the inal LLAPS test, an
extensie pilot was used to identiy lower and upper limits o learning in the population and proide an
132
analysis o the alidity and reliability o the instrument used. 1he data rom this phase was then used to reine
the inal test used in the LLAPS project.
A1.23
%
A1.24
1he choice and structure o content or Lnglish, Urdu, and Math was based on an attempt to
optimize on the ollowing:
" J&<#)7$ 9S :9(7<(7: 1he test should coer the general range o content taught to children by the
time they reach ith grade.
" C#(O< 9S )%SS%:.Q7U: 1he range o diiculty should be aried across skills ,questions, as well as
within each skill ,question, to better capture ariation in achieement.
" /%-7& *.7%9( 9S #*%Q%7U 7UR<: 1he test should call upon the dierent cognitie abilities releant or
understanding the content in question ,or example, in Mathematics: conceptual, procedural and
problem soling,.
" 6#&%#7%9( %( 7UR< 9S b.<-7%9(-: Depending on the content being tested, there should be some
ariation in type o questions, including multiple-choice questions, short answers and long answers.
Lach type connotes a dierent leel o prerequisite skill as well.
In addition to these general principles, other rules ollowed or wording and compiling each question
included the ollowing criteria:
" >#-U 79 .()<&-7#(): lormats that are amiliar to children should be used and understanding the
question should not require an ability greater than that which is being tested by the question.
" >#-U 79 #)0%(%-7<&: Questions that required additional materials can be discriminatory. Such
questions should only be considered i they are easible to administer properly.
" B(*%#-<) %(-7&.0<(7 with respect to socio-economic status: Reerences to items that children
rom particular socio-economic backgrounds will hae no exposure to can create bias. Such
reerences should be aoided.
A1.25 1he initial ersion o the test ollowed the curricular standards or Grade 3 and 4 closely. loweer, it
was quickly noted that the perormance o children was considerably below what the curriculum stated it
should be, thus urther alidating the choice o a norm-reerenced test rather than a criterion-reerenced test
based on the curriculum. lor the irst week o pilot testing ,6 schools, parts o the test were reconstructed
ater each test to minimize the problems discussed aboe and improe the precision o ability,learning
estimates across the range o students. Although all the content o the inal ersion was checked or
consistency with the Pakistan`s curricular standards and some material was added based on emphasis in the
133
curriculum, the instrument is designed to test basic competencies rather than comprehensie knowledge o
speciic curricular items. 1he content or the portions o the test speciic to Grade 3 and 4 ,the medium and
diicult sections, ollows rameworks prealent in the learning assessment literature ,see Annexure 1,.
/<Z<Q9R0<(7 #() @<Q<:7%9( 9S "<-7 17<0-
|rav ava vgti.b
A1.26
A1.2
Literacy rameworks used by other assessments ,see Annexure 1, categorize the dierent purposes o
reading and writing that should be assessed. loweer, these rameworks assume that the pupil will hae the
basic ability to do .ove reading and writing. \e hae not made this assumption while constructing the
instrument. lence both the Lnglish and Urdu sections begin with the alphabet and progress through the
basic elements o writing: word construction, grammar, ocabulary, sentence construction, and conclude with
a reading comprehension and essay exercise.
1he instruments included in the Urdu and Lnglish sections were adapted rom a ariety o sources
30
.
1he two sections coer a comparable range o content and diiculty: Both sections begin with alphabet
recognition and end with an essay question and the tests maintain consistency in content, structure and
intellectual demands. 1able I below summarizes the content areas o the Urdu and Lnglish sections.
30
Many questions in the Urdu section were adapted Irom instruments included in Kizilbash (1997). Other than tests and exams
administered by teachers in schools, we were unable to Iind suitable instruments Ior the English component and in addition to
designing questions ourselves, web-based resources Ior educationists were used to Iill this gap. Based on Dr. Catherine Snow`s
(Harvard School oI Education) suggestion a cloze passage (every IiIth word or so blanked out) was included in the English
section (source: www.tut-world.org) and Urdu section.
134
"#*Q< 'T;P A9(7<(7 '&<#- S9& B&). #() >(OQ%-$
A9(7<(7 '&<#- B&). >(OQ%-$
1,e Q. 1,e Q.
'QR$#*<7-
\ritten: Complete chronological
order o alphabets
1
Verbal: \rite alphabets read aloud
\ritten: Complete chronological
order o alphabets
1
3
K9&) C<:9O(%7%9(
\ritten: Match words with
pictures
Y
Verbal: \rite words read aloud
\ritten: Match words with
pictures
2
c
K9&) A9(-7&.:7%9(
Break words into alphabets
Join alphabets to orm a word
3
4
\rite words read aloud
Complete word or each picture
Create words rom gien alphabets
2
5
9
M�#&
Match words with antonyms
\rite plural or singular words
lill blanks or gender agreement
Cloze passage
e
8
h
Math words with Antonyms
lill blank words in sentences
Close passage
e
f
`
69:#*.Q#&U
lill blank word in sentence
^
lill blank word in sentence
Create words rom gien alphabets
f
9
@<(7<(:<
A9(-7&.:7%9(
Use words in sentences 10 Use word in sentences 10
A90R&<$<(-%9( 11 11
>--#U 12 12
Note: Multiple-choice questions are indicated in bold. Some questions are listed or more than one
content area.
A1.28 Note howeer, that the starting items o the Lnglish section are easier than the Urdu since
competency in Lnglish may be lower: lor instance, the irst question in Lnglish requires recognizing 3
alphabets read aloud and writing them. No knowledge o alphabetical order or the ability to recognize other
alphabets is required. In Urdu on the other hand, the irst question inoles illing the right Urdu alphabet in
the blank and students need to know both how to write the missing alphabet as well as recognize other
alphabets and know the alphabetical order. Apart rom this dierence in diiculty o the early items, the
standard test ormat between the two languages allows the student to amiliarize hersel with the test in Urdu,
beore proceeding to the section on Lnglish. linally, while diiculty increases with each section o the test,
there is also a range o diiculty within each test section. 1hus or example, the section on ocabulary
includes both easy and diicult words to aid in discriminating between dierent students.
135
Matbevatic.
A1.29
A1.30
lor the Mathematics test, we hae adapted rameworks used by other assessments that outline the
content domains to be assessed. 1he ie major domains that are identiied under these outlines are
31
:
" Number sense, properties, and operations ,40,,
" Measurement ,20,,
" Geometry and spatial sense ,15,,
" Data analysis, statistics, and probability ,15,, and
" Algebra and unctions ,15,.
\hile we hae ollowed a similar pattern, greater emphasis is placed in our math test on the irst
major domain- Number sense, properties and operations. During the initial ielding o the test instrument in
6 schools, it was noted that perormance in math was lower than expected, and the test was not proiding
adequate inormation or children at the lower end o the knowledge distribution. 1hus, additional items were
added coering the irst domain to obtain iner partitions o knowledge or this subset o test-takers. 1he
content and cognitie demand o the inal test is summarized below:
"#*Q< 'TYP A9(7<(7 #() A9O(%7%Z< /90#%( S9& I#7$<0#7%:- 5%7$ F.<-7%9( >V#0RQ<-
A9(7<(7 /90#%(- ,example Qs rom test, A9O(%7%Z< /90#%(- ,example Qs rom test,
a, Number Operations ,1-5,
b, Measurement ,15,
c, Geometry ,22,
d, Algebra ,11 and 20,
e, Data Analysis ,21,
1. Conceptual Understanding ,4 and 11,
2. Procedural Knowledge ,5, 10, 13,
3. Problem Soling ,9, 12, 14,
A1.31
1he more adanced content domains such as algebra hae only been conceptually tested, whereas
more rudimentary elements such as addition, subtraction, multiplication and diision hae been tested both
procedurally and through problem soling. As with the Lnglish and Urdu tests, the range o diiculty aries
within math skills tested ,e.g. Addition: single digit, two digit, three digit with carry, decimal with carry, as well
as across math skills tested ,counting to percentages and ractions,. In addition to better capturing ariation in
achieement, the test design could also be useul in identiying particular stumbling blocks` or students with
31
See Appendix 1 Ior a detailed description
136
regard to particular skills ,or example, diiculty with diision`, or particular leels o diiculty ,or example,
diiculty with 2 digit diision,multiplication,. 1he order o questions and progression o content is based on
the order in which particular math skills are taught to students in Pakistani schools ,as indicated by the state
curriculum in 2002, and is summarized in the table below:
"#*Q< 'T]P A9(7<(7 '&<# #() C#(O< S9& I#7$<0#7%:- 17<0-
A9(7<(7 '&<# C#(O< 9S -W%QQ- 7<-7<) Fp
A9.(7%(O
Count objects, compare numbers, complete chronological
order o numbers, addition o objects, translate numbers in
words, tell time, retriee count rom word prob.
;,Y,3,4,6,,8,9
'))%7%9( 1 digit no carry- 3 digit with carry, word problem 5, 9, 10, 12
@.*7&#:7%9( 1 digit - 3 digit with carry, word problem 5, 9, 10, ;c
I.Q7%RQ%:#7%9( 1 digit by 1 digit - 3 digit by 2 digit with carry, word prob. 5, 13, ;c, 18,19
/%Z%-%9( 1 digit by 1 digit- 3 digit by 2 digit, word prob., LCM, lCl 5, 9, 13
/<:%0#Q- Addition, subtraction 10
G&#:7%9(-
Read chart, conersion to mixed ractions, addition o
ractions, subtraction o ractions
;^, 16, 1
/#7# '(#QU-%-
Read Bar chart, read chart in ractions, read chart in
percentages
;^, Y;
/<).:7%Z< Complete Sequence, weight comparison 11, Yi
Note: Multiple-choice questions are indicated in bold. Some questions are listed or more than one
content area.
A1.32
A1.33
1he careul design o the test instruments based on the pre-assessment in the irst six schools
combined with a large number o interiews with teachers and children has allowed us to estimate with airly
high precision, the knowledge o children at all leels o learning. loweer, seeral problems were noted in
the design and implementation o the tests in the pilot which were later corrected. Some o these are speciic
to the educational enironment in Pakistan and we outline these briely below or uture testing exercises,
these problems need to be urther discussed in the context o the national assessment exercise and we hope
that the documentation here ensures that uture tests better accommodate these issues in the design and
administration o the instrument.
Procedural/Implementation Issues and Other Concerns
1he irst set o problems that we outline arises rom the multiplicity o natie languages currently
used in Pakistan. Speciically, we address the issue o the language that should be used to proide instructions
in dierent tests and the implication that restricting our tests to Urdu and Lnglish has or the potential uses
o such assessments. Lastly, we briely note some problems that arose in the ormatting o test questions.
13
1be vterretatiov of iterac, core.
A1.34
A1.35
A1.36
A1.3
A1.38
1he exclusie use o Urdu in our test instrument places limits on the interpretation o literacy scores.
lunctional literacy, in terms o the ability o indiiduals to participate in society, may dierge signiicantly
rom Urdu literacy i Urdu is used primarily as a second-language` in the region considered. 1his limits the
use o the literacy score in certain types o analysis. lor example, using Urdu literacy as a proxy or human
capital may underestimate human capital in areas where Urdu is not the primary language. 1his would suggest
the expansion o the test-instrument to include testing in the ernacular, particularly, Saraiki, Pashto and
Punjabi or Punjab proince.
On the other hand, this does not signiicantly aect the alue o Urdu literacy scores or other
potential questions. Since Urdu literacy is a core competency in Pakistan`s curricular standards, Urdu literacy
scores proide aluable inormation on educational attainment in dierent schools. 1hus, the decision to test
in the ernacular as opposed to ,or in addition to, testing in Urdu must be based on the aim o the test
instrument-i the main aim o the test instrument is to assess fvvctiovat titerac,, ernacular testing would be
essential, but i the primary ocus is on tearvivg iv .cboot. ,or alue-added learning,, restriction to testing in
Urdu would yield signiicant insights.
iterac, ia. iv Matbevati'!
Pakistan`s linguistic ractionalization also has implications or the language used in proiding
instructions or the test instrument. lor our pilot, we chose Urdu and Lnglish primarily due to high uniorm
exposure rom early childhood and to ensure compatibility with the Pakistan Lducation Ministry in their
Strategic lramework or National Lducation Assessment 2001.
Despite the act that our tests were administered in areas where Urdu is vot a second language, the
use o two languages or instructions raised important issues in the Math assessment as a result o the
interaction between language skills and math skills in the design o the instrument. 1his interaction was noted
in two dierent areas o the test: the medium o instruction, and the design o questions that required
conceptual translations between language and math.
Meaivv of v.trvctiov
In the schools or the pilot test, which included top priate schools in urban areas, all language
statements ,including instructions, in the mathematics section were either in Urdu or Lnglish depending on
138
the language o instruction. 1he need or separate languages arises due to the presence o dierent language
mediums` or school instruction ,note though, that this is an issue only i the sample includes urban schools,
or the rural sample o the LLAPS project, all instructions were in Urdu,. Speciically, in Lnglish-medium`
schools exposure to mathematical terminology may be only in Lnglish, and the use o only Urdu terminology
would lead to problems in comparisons across schools: lor instance, children in Lnglish ,Urdu, medium
schools may be amiliar with the Lnglish ,Urdu, bvt vot the Urdu ,Lnglish, concept o the lowest common
multiple` o two numbers.
A1.39
A1.40
A1.41
Although an eort was made to limit unneeded erbiage in the mathematics section o the test,
signiicant portions still required a basic leel o literacy. 1he choice to include written instructions with
mathematical terminology and some word problems is intentional, and is intended to test the ability o the
student to mathematically interpret common ,erbal, situations.
loweer, this prerequisite o basic literacy, although reasonable, may bias the use o test scores as a
measure o students` skills and competencies in mathematics, i language skills are poor
32
. luture tests need to
careully consider the implications o using erbal statements in mathematics exams, especially in the context
o students who may not hae attained any degree o literacy by the time the test is administered. 1his
problem is urther compounded in regions where the ernacular language is dierent rom Urdu-in these
cases, test designers need to assess the degree to which the use o ernacular instructions would help in
ensuring that the test instrument is alid as a measure o learning in mathematics.
1ocabvtar, et.
A second, perhaps subtler problem was the requirement o aifferevt teret. of rocabvtar, in the translation
o language to mathematical statements, depending on the medium o instruction. One item that led to such a
problem was the translation o numbers in words to numbers in numerals. 1he choice o one particular
number, 65` raises a direct issue, since in Lnglish, to write numbers in words till 100` requires
memorization o each number till the 20` as well as the speciic numbers 30`, 40` etc. In Urdu howeer,
this requirement requires the memorization o att vvvber. titt 100: and it is common to ind students with
the same mathematical skills who dier along this particular dimension. Particular care needs to be taken that
when instructions and items are translated, the .ie of tbe reqvirea rocabvtar, .et is the same in all languages used
or test administration.
32
For example, Howie and Plomp (2001) Iind a relationship between the number oI Iirst language speakers in a class and pupils`
achievement in mathematics in their analysis oI South AIrican school level TIMSS data. Similarly, The SAP National Survey
(1995) Iound similar literacy eIIects when testing teacher skills. Both male and Iemale teachers Irom urban and rural areas
scored worse on narrative questions as compared to numerical questions.
139
A1.42
A1.43
A1.44
1hus, the testing o Math in Pakistan raises special problems through the interaction o the subject
matter with language skills. Our recommendations or uture tests in this regard are to assess with some care
,potentially through the deelopment o a more extensie item-bank, the use o the medium o the test
instrument, with particular emphasis on the second issue raised aboe-the teret. o language skills required
by the math instrument should ideally be independent o the particular language used.
Qve.tiov ava .v.rer orvat
During the initial testing rounds, there was some concern regarding the ormat o test questions.
1eachers elt that questions might be unair, since students were not used to the particular question-answer
ormat used. As a result, eorts were made to ormat questions in collaboration with the teachers, in a simple
and straightorward manner. 1o assess the importance o ormatting in test scores, a simple experiment was
carried out, where students in a large school were randomly diided into two groups. lor the irst group, the
test was administered with no guidance and ollowing standard guidelines. lor the second group, an
additional instructor was detailed to proide assistance or problematic questions`. 1he comparison o scores
o the irst and second groups shows no signiicant dierence, leading us to conclude that the ormat o the
test was not related to test scores o students
33
. loweer, we eel that the ormatting o test questions
requires greater attention in uture assessments. Prior to the design o such an assessment, it would be
important to assemble a database o tests currently used in dierent schools to check or dierences in the
use o common test ormats. 1he inal ormat o the test instrument could then account or these dierences,
potentially through the choice o a subset o ormats that are common to all schools.
Assessment of the Instruments using Item Response Theory
1his section assesses the content o the test, and statistically examines the alidity o the test or
examining arious issues regarding learning achieement in Pakistan. lor this section o the document, we
rely on methods deried rom Item-Response 1heory to examine the alidity o each question ,henceorth
item, as well as the precision o the test taken in its entirety. As explained preiously, this test was speciically
designed to proide inormation on the ability,knowledge o children rom all points o the distribution-
33
One particular Iormatting problem arises with the popular `matching` questions used in tests in Pakistan. This item requires
students to `match` two words Irom diIIerent columns, Ior instance:
Q. Match the opposites
Good Smart
Stupid Bad
This sort oI question has two problems: First, our preliminary results indicated that students were conIused by this style oI
question, maniIested by dramatically lower perIormance compared to levels seen Ior the same question Iormatted in alternative
ways. Second, the answers Ior each match depends on the availability oI remaining unmatched words. Since each match is not
independent oI the others, the inIormational content provided by every new match is reduced. This dependence between two test
answers would then invalidate standard test response assessments.
140
how precise is our test then, in distinguishing between dierent ability,knowledge
34
leels lor a brie
introduction to Item Response theory ,to the extent needed to understand this section,, reer to Annexure 2.
or rett ao tbe te.t. e.tivate .tvaevt abitit,,/vorteage
A1.45
ligures A.1, A.2 and A.3 show the
characteristics o the tests and the distribution o
student scores. Lach igure is ormatted in the same
manner: a histogram is drawn or the distribution o
test-scores in the releant subject and, the
histogram is oerlaid with the 95 conidence
interals or the estimate o the score. Oerall, the
standard errors o the knowledge estimates are low.
Comparable tests or other countries ,Sweden,
1ogo and Zambia hae been compared, are
characterized by much higher at the lower and
upper ends o the ability distribution ,almost double
that o our tests,.
G%O.&< 'T;P 4(95Q<)O< @:9&<- #() @7#()#&) >&&9&- S9&
>(OQ%-$
-
5
0
0
0
5
0
0
1
0
0
0
1
5
0
0
S
t
a
n
d
a
r
d
E
r
r
o
r
B
o
u
n
d
s
0
2
4
6
8
1
0
P
e
r
c
e
n
t
0 200 400 600 800 1000
Knowledge Scores
Score Histogram Test Score
5% Confidence nterval 95% Confidence nterval
Test Standard Errors: English
34
Through this document, we use the words ability and knowledge interchangeably, although the terms have very diIIerent
meanings in economics and Item Response. While in the Iormer, ability is a measure oI the student`s learning, in economics
ability is an intrinsic attribute oI the individual that remains Iixed over time. We attempt to satisIy both strands by using
`ability/knowledge` but our Iunctional deIinition oI either is a measure oI the student`s level oI learning.
141
G%O.&< 'TYP 4(95Q<)O< @:9&<- #() @7#()#&)
>&&9&- S9& I#7$<0#7%:-
G%O.&< 'T]P 4(95Q<)O< @:9&<- #() @7#()#&)
>&&9&- S9& B&).
-
5
0
0
0
5
0
0
1
0
0
0
S
t
a
n
d
a
r
d
E
r
r
o
r
B
o
u
n
d
s
0
2
4
6
8
P
e
r
c
e
n
t
0 200 400 600 800 1000
Knowledge Scores
Score Histogram Test Score
5% Confidence nterval 95% Confidence nterval
Test Standard Errors: Mathematics
-
5
0
0
0
5
0
0
1
0
0
0
1
5
0
0
S
t
a
n
d
a
r
d
E
r
r
o
r
B
o
u
n
d
s
0
2
4
6
8
1
0
P
e
r
c
e
n
t
0 200 400 600 800 1000
Knowledge Scores
Score Histogram Test Score
5% Confidence nterval 95% Confidence nterval
Test Standard Errors: Urdu
A1.46
A1.4
Neertheless, the graphs show that our estimates around the middle o the distribution are ar better
than at the ends o the distribution-this is a standard issue with all tests, since items designed or proiding
inormation at the extremes o the distribution also add to inormation or the mean, but not necessarily the
other way around. 1he problem is compounded in our case because the distribution o knowledge scores
appears to be a mixture consisting o a hump` o students who perorm extremely poorly combined with a
standard normal distribution. 1his hump is the group o students who know so little that een the LLAPS
test, which starts with the basics, does not yield much inormation about their knowledge-we know they are
doing badly, but not how badly they are doing. Clearly, it is possible to rectiy this or Urdu by starting the
test with asking children to write alphabets, rather than ill in the order. Similarly or Mathematics, the test
could start with asking children to write down numbers next to a collection o objects. lor Lnglish, we don`t
see any alternatie at all, since the test started with the easiest possible question that can be used without
inoling superisors more integrally-something that usually causes biases and greater error due to
superisor-speciic eects.
111T AHDA=B@1HD
1his document proides a detailed description o the testing enironment in Pakistan as well as
documentation and analysis o a pre-pilot test carried out in priate and public schools in Pakistan. 1he
analysis o this test shows that while there are some areas or improement, on the whole the test has
perormed extremely well in its ability to distinguish between students o dierent caliber. Neertheless,
aluable lessons were learnt, relating to the issues o testing in a linguistically ractionalized region and these
should be careully noted, especially with regards to the upcoming national assessment program.
142
'((<V.&< ;P "<-7 @.00#&U "#*Q<-
E<#&
9S
"<-7
@.*2<:7-
"<-7<)
@#0RQ<
@%3<
[-:$99Q-\
@#0RQ%(O
"<-7
H.7:90<-
I<#( l
[-7#()#&)
<&&9&\
"<-7
'Z#%Q#*Q<X
"<-7
6#Q%)#7%9(
/9:.0<(7-
'Z#Q#*Q<X
@9.&:<
1984
Science
Mathematics
3,300
Representatie.
Grades 4 and 5.
No. No. Shah ,1984,
1988-
9
Science
Mathematics
11,000
,500,
Grades 4 and 5. No. No.
BRIDGLS
project at
larard
Institute or
International
Deelopment
,1989,
1994
Mathematics
Science
Social Studies
Dinyat
15,991
,42,
Grades 3 and 5. No. No.
Mirza and
lameed ,1994,
1995
Lie Skills
Knowledge
Rote Reading
Reading with
Comprehension
\riting rom
Dictation
\riting Letter
Numeracy and
Arithmetic
Mental
Arithmetic
Reading o
loly Qur`an
2582
Multi-stage,
systematic-
random sample
design. All
children ages 11-
12.
26.1
63.
26.8
61.
18.1
69.6
6.
44.2
No. No. Perez ,1995,
1995
Mathematics
General
Knowledge
Comprehension
11,563
,52,
Grade 5.
45.6
4.4
69.1
No. No. MSU ,1995,
1999
Mathematics
Urdu
General
Knowledge
965
,50,
Grade 4.
60
1
5
No. No.
Action Aid
Pakistan ,1999,
1999
Science
Mathematics
Urdu
294
,145,
Grade 4.
Sample not
proportionate to
unierse.
2
58
2
No. No.
Khan et al
,1999,
1999
Mathematics
Science
Social Studies
Urdu
200
,20,
Grade 3 and 5. No. No. Ari et al ,1999,
143
E<#&
9S
"<-7
@.*2<:7-
"<-7<)
@#0RQ<
@%3<
[-:$99Q-\
@#0RQ%(O
"<-7
H.7:90<-
I<#( l
[-7#()#&)
<&&9&\
"<-7
'Z#%Q#*Q<X
"<-7
6#Q%)#7%9(
/9:.0<(7-
'Z#Q#*Q<X
@9.&:<
1999
Mathematics
Science
Social Studies
Urdu
160
,20,
Sample 10 male,
10 emale
schools ,80
students each,.
Grades 3-5.
No. No.
Research 1eam
o Bureau o
Curriculum
Deelopment
and Lxtension
Serices,
N\lP ,1999,
1999
Mathematics
Urdu
131
,31,
DLOs selected
equal numbers
o good, aerage
and weak
schools.
No. No.
Punjab Literacy
\atch ,1999,
2000
Sindhi
Mathematics
Science
Social Studies
Islamiyat
300
,20,
Randomized
sample 10 male,
10 emale
schools ,150
students each,.
Grades 3-5.
No. No.
laque et al
,2000,
2000
Mathematics
Science
Social Studies
Urdu
801
,20,
10 male, 10
emale schools.
Grades 3-5.
No. No.
Research 1eam
o Bureau o
Curriculum and
Lxtension
Centre,
Balochistan
,2000,
@9.&:<P UNLSCO ,2001,
144
H7$<& '--<--0<(7- "#*Q<
'.7$9&[-\ m C<-<#&:$
H&O#(%3#7%9(
"%7Q< /#7< @R9(-9&%(O H&O#(%3#7%9(
Institute o Lducation and
Research Uniersity o Punjab
,Not known, 1996
Primary Lducation Project
lrom LDI ,1999,
North \est Lducational
Assessment Programme
,Not known, 1996
PLDP
,N\lP,
lrom LDI ,1999,
Bureau o Curriculum and
Lxtension \ing, JAMSlORO,
Sindh
,Not known, 199
SPLDP
lrom LDI ,1999,
Primary Lducation Directorate,
QUL11A, Balochistan
,Not known, 1998
BPLDP
lrom LDI ,1999,
Bureau o Curriculum and
Lxtension \ing, JAMSlORO,
Sindh
,Not known, 1998
SPLDP
lrom LDI ,1999,
1est Deelopment Centre,
Lducation Department,
LAlORL
,Not known, 1999
Punjab Middle Schooling
Project
lrom LDI ,1999,
Directorate o Lducation, Gilgit
Baseline Achieement o Class 4
Students in Northern Areas,
Pakistan
2000 NALP
@9.&:<P UNLSCO ,2001,
145
'((<V.&< Y
GC'I>KHC4@ GHC =1">C'AE
D#7%9(#Q '--<--0<(7 9S >).:#7%9(#Q ?&9O&<--P
According to the ^.P Reaaivg raveror/, deeloped by the National Assessment Goerning Board ,NAGB,,
NALP assesses three purposes or reading. In addition to reading or dierent purposes, NALP reading
comprehension questions are deeloped to engage the dierent approaches that readers take in the process o
trying to understand what is being read.
"$&<< )%SS<&<(7 R.&R9-<- S9& &<#)%(O 5<&<
#--<--<)P
" C<#)%(O S9& Q%7<&#&U <VR<&%<(:<P Readers
explore human emotions and eents by reading
noels, short stories, poems, plays, and essays.
" C<#)%(O 79 O#%( %(S9&0#7%9(P Readers gain
inormation by reading such materials as
magazines, newspapers, textbooks, encyclopedias,
and books.
" C<#)%(O 79 R<&S9&0 # 7#-WP Readers apply what
they learn rom reading such materials as bus or
train schedules, directions or repairs, games,
classroom procedures, tax orms, maps, etc.
@7.)<(7- 5<&< #--<--<) 9( S9.& )%SS<&<(7 -7#(:<-
9& 5#U- 9S &<-R9()%(O 79 5$#7 %- &<#)P
" G9&0%(O #( %(%7%#Q .()<&-7#()%(OP what is the
oerall meaning and purpose o what is read
" /<Z<Q9R%(O #( %(7<&R&<7#7%9(P what meaning do
the relationships among the dierent parts o the
text hae
" ?<&-9(#Q &<SQ<:7%9( #() &<-R9(-<P how does
what is read relate to or compare with the reader's
knowledge and experience
" A&%7%:#Q -7#(:<P how does what is read
communicate inormation or express ideas
According to the ^.P !ritivg raveror/, deeloped by the National Assessment Goerning Board ,NAGB,, the
NALP writing assessment should hae the ollowing objecties:
" Students should write or a ariety o purposes: narratie, inormatie, and persuasie.
" ^arratire rritivg inoles the production o stories or personal essays. It encourages writers to use
their creatiity and powers o obseration to deelop stories that can capture a reader's imagination.
" vforvatire rritivg communicates inormation to the reader to share knowledge or to coney
messages, instructions, and ideas. 1he inormatie topics in the 1998 writing assessment required
students to write on speciied subjects in a ariety o ormats, such as reports, reiews, and letters.
" Per.va.ire rritivg seeks to inluence the reader to take some action or bring about change. It may
contain actual inormation, such as reasons, examples, or comparisons, howeer, its main purpose
is not to inorm, but to persuade. 1he persuasie topics in the 1998 writing assessment asked
students to write letters to riends, newspaper editors, or prospectie employers, to reute
arguments, or to take sides in a debate.
" Students should write on a ariety o tasks and or many dierent audiences.
" Students should write rom a ariety o stimulus materials.
" Students should generate, drat, reise, and edit ideas and orms o expression in their writing.
?&9O&<-- %( 1(7<&(#7%9(#Q C<#)%(O =%7<&#:U @7.)UP
1he PR Reaaivg iterac, raveror/ or 2001 was deeloped through an international collaboration inoling the
PIRLS Reading Deeloping Group and National Research Coordinators rom oer 40 countries. 1he ramework
ocuses on two aspects o reading literacy: purposes or reading and processes o comprehension.
1he PRILS ramework outlines two purposes or reading:
" Acquire and Use Inormation ,50,, and
" Literary Lxperience ,50,.
It also includes our processes o comprehension:
" locus on and Retriee Lxplicitly States Inormation ,20,,
" Make Straightorward Inerences ,30,,
" Interpret and Integrate Ideas and Inormation ,30,, and
" Lxamine and Laluate Content, Language and 1extual Llements ,20,.
146
GC'I>KHC4@ GHC I'"L>I'"1A@
D#7%9(#Q '--<--0<(7 9S >).:#7%9(#Q ?&9O&<--:
1he NALP mathematics assessment uses a ramework inluenced by the National Council o
1eachers o Mathematics ,NC1M, Curriculum and Laluation Standards or School Mathematics.
1he ^.P Matbevatic. raveror/ describes ie broad strands o mathematics content with arying
target assessment times ,in parentheses,, as ollows:
" Number sense, properties, and operations ,40,,
" Measurement ,20,,
" Geometry and spatial sense ,15,,
" Data analysis, statistics, and probability ,15,, and
" Algebra and unctions ,15,.
It includes three types o mathematical abilities as ollows:
" Conceptual understanding,
" Procedural knowledge, and
" Problem soling.
And it includes mathematical power as ollows:
" Reasoning,
" Connections, and
" Communication.
"&<()- %( 1(7<&(#7%9(#Q I#7$<0#7%:- #() @:%<(:< @7.)UP
1he 1IMSS mathematics assessment uses a similar ramework. It contains two dimensions-content
and cognitie domains. 1he deelopment o this ramework included widespread participation and
reiew by educators around the world. 1he close relationship with the NALP domains illustrates the
considerable consensus on appropriate elements.
1he 1MM Matbevatic. raveror/ or 2003 outlines ie mathematical content domains:
" Number ,40,,
" Algebra ,15,,
" Measurement ,20,,
" Geometry ,15,, and
" Data ,10,.
It also include our mathematical cognitie domains:
" Knowing lacts and Procedures ,20,,
" Using Concepts ,20,,
" Soling Routine Problems ,20,, and
" Reasoning ,20,.
14
'((<V.&< ]P >V#0RQ< F.<-7%9(- S&90 /%SS<&<(7 "<-7-
Note that a single qve.tiov can be multiple items, since each part o the question is treated as a separate item.
Question Number: 1 ,Items 1-3,: Lasiest Questions in Lnglish
1ask Content: Verbally recognize and write alphabet
lormat: Short answer - Question to be read aloud
Question:
Listen to your teacher careully, and then write the letter.
Answer Key a, B
b, G
c, Q
Scoring Guide Record 1 i the Lnglish alphabet read aloud is written correctly
Record 0 otherwise ,including partially correct and blank answers,
Question Number: 2 ,Items 4-6,
1ask Content: Match words with pictures
lormat: Multiple choice
Question Instructions 1ranslation:
1here are three words gien in ront o each picture. Mark the word matching the picture.
Answer Key:
Scoring Guide: 1~Only the correct word ticked or the picture
0~Incorrect word ticked or the picture or multiple words ticked or the
picture
148
Question Number: 10 ,Items 4-51,: Diicult Questions in Lnglish
1ask Content: Using gien word in sentence construction
lormat: Short answer
Question:
Use the ollowing words in sentences:
school
doctor ______
______________________
beautiul
deep
play
Answer Key
Scoring Guide lor each
part
Record 1 i
any
sentence
written
using the
word has
been
constructed
Record 0 i
149
>V#0RQ<- S&90 7$< B&). "<-7
Question Number: 8 ,Items 34-40,: Diicult Questions in Urdu
1ask Content: Recognize Masculine,leminine gender in nouns
lormat: Short answer
Question Instructions 1ranslation:
\rite masculine` in ront o the masculine word and eminine` in ront o the eminine word
Answer Key:
Scoring Guide: 1~ Correct masculine,eminine illed in the blank
0~all other responses
Comments:
1his question was drated dierently rom the way students are used to seeing this problem in their exams. In
their exams, they are used to conerting masculine nouns to eminine and ice ersa. \e wanted to gie some
abstract nouns that did not hae an opposite gender counterpart. 1his question turns out to be 0.:$ harder
than all other questions in the test.
150
>V#0RQ< F.<-7%9(- S&90 7$< I#7$ "<-7
Question Number: 1 ,Item 1, : An easy question in Mathematics
1ask Content: Recognize more ersus less
lormat: Multiple choice ,question read aloud,
Question:
Circle the box that has more objects
Answer Key: Box with 8 objects
Scoring Guide: Record 1 i ONL\ the correct box is marked
Record 0 otherwise
Question Number: 5 ,Items 8-12,: 1he irst two questions are easy, the rest are diicult
1ask Content: One digit addition, subtraction and multiplication
lormat: Short Answer
Question:
Sole
Answer Key:
A. 10
B. 5
C. 2
D. 20
L. 10
Scoring Guide:
lor each part:
Record 1 i the correct answer is illed in
Record 0 otherwise
4
6
8
- 3
9
9
9
4
x 5
2
20
x
151
Question Number: 2 ,Items 2-3,: A medium diiculty question in Mathematics
1ask Content: Count objects and identiy corresponding number
lormat: Multiple Choice ,question read aloud,
Question:
Circle the number that matches the number o objects.
Answer Key: A. 2
B. 5
Scoring Guide: lor each part
Record 1 i ONL\ the correct number is ticked, circled or underlined
Record 0 otherwise
1, 2, 3, 4,
1, 2, 3, 4, A
B
152
'((<V.&< cP 17<0 C<-R9(-< "$<9&U
1T 1">I AL'C'A">C1@"1A@
Item response theory seeks to model test-responses to answer our types o questions:
1. low likely is a person with a gien leel o knowledge to answer a particular item correctly
2. \hat are the estimates o ability,knowledge o the indiidual test-takers
3. \ith what precision does a single item allow us to estimate this ability,knowledge
4. \ith what precision does the evtire te.t allow us to do the same
\hile the statistical tools or analyzing tests using item-response theory are airly complex, the basic concepts
are readily understood in the ollowing diagram:
igvre .1.1: tevRe.ov.e 1beor, De.critiov of a ivgte 1e.t tev
P
r
o
b
a
b
i
l
i
t
y
o
I
C
o
r
r
e
c
t
A
n
s
w
e
r
0
1
0.
Ability/Knowledge as standard deviations Irom mean
DiIIiculty: The level
of ability at which
the probability of a
correct answer is 0.5
Guessing Parameter: The
probability with which a person
with no knowledge would obtain
a correct answer
Discrimination: The maximum
slope of the Test Characteristic
Curve: a measure of how well the
item differentiates between
different groups
Test Characteristic
Curve: The
probability with which
a person oI a speciIied
knowledge obtains a
correct answer
In using Item-Response 1heory methods to model responses to items, the logistic amily o distributions is
used to structure the probability o a correct response as a unction o the knowledge o the student. Lach
item is associated with an tevCbaracteri.tic Cvrre ,in red aboe, that shows how this probability changes with
153
the knowledge o the student. lurther, or each question, there are three related parameters that ully deine
this Item Characteristic Cure:
1. /%SS%:.Q7U: As suggested by the term, the more diicult` a question, the lower the probability that a
student o a gien ability will obtain a correct answer. In terms o the diagram aboe, increasing the
diiculty o the item will shit the item-characteristic cure to the right- a student at the same leel o
knowledge will hae a lower probability o obtaining a correct answer ,technically, the diiculty o
the question is the ability,knowledge required to obtain a correct answer with a 50 probability,.
2. /%-:&%0%(#7%9(: \hile the diiculty o an item tells us something about the probability with which a
person will answer the question correctly, it does not tell us much about the ability o the item to
distinguish ,or discriminate`, between indiiduals o diering ability,knowledge. 1his ability to
discriminate is described by the )%-:&%0%(#7%9( o the item, which measures the slope o the
characteristic cure at the point o inlection ,equialent to the maximum slope, - the greater the
slope at this point, the greater the ability o the item to discriminate between students aboe or below
tbe oivt at rbicb tbe .toe i. vea.vrea. 1he discrimination parameter lends itsel easily to an intuitie
ormulation: suppose that both students o ability 1 and students o ability 2 had an equal probability
o answering an item correctly- in this case, the item would be unable to discriminate between these
two types o students, and, drawn between these two ability types, the slope o the characteristic
cure would be 0. loweer, i it were the case that att students o ability 1 had a 0 probability o
answering the question correctly, and all students o ability 2 had a probability 1 o answering the
question correctly, then the item would ai.crivivate erfectt, between these two groups, as in the
diagram below.
3. M.<--%(O ?#�<7<&: 1he inal parameter associated with an item is a guessing parameter` that is
an estimate o the probability with which a student with no knowledge o the item can obtain a
correct response. One way to think about this parameter is in the case o multiple choice questions-
i there are our options, random-guessing would imply that the probability o a correct answer is at
least 0.25. Note howeer, that the guessing parameter in this case may actually be bigber than 0.25, i
there is reason to aor one sort o answer or the question- the notion o the educated guess`.
154
G%O.&< 'cTYP 17<0 5%7$ R<&S<:7 )%-:&%0%(#7%9( #7 iTY^.
P
r
o
b
a
b
i
l
i
t
y
o
I
C
o
r
r
e
c
t
A
n
s
w
e
r
Ability/Knowledge as standard deviations Irom mean
1
0.25
At the end o this appendix, we will proide the estimates o these parameters or each item in the tests- the
primary aim o doing so is to aid in the deelopment o an item-bank, and moe towards a systematic
standard-setting or tests in Pakistan that would generate comparability across time and countries.
Once the three parameters discussed aboe are estimated or each item, the alues o these parameters are
then used to estimate the ability,knowledge o each student who takes the test
35
. 1hese estimates are
essentially the itev re.ov.e .catea estimates o student perormance, and take into account not only the total
number o questions that a student has answered correctly but also the t,e. of qve.tiov. that the student
answers- two indiiduals with otherwise identical test responses will score dierently i one answers a more
diicult question correct than the other. 1his is the /vorteage .core used in Chapter 1, with a mean o 500 and
standard-deiation 150.
11T ">@" AL'C'A">C1@"1A@
\hile the three parameters aboe proide a description o eery item in the test, we are also interested in
oreratt test characteristics- particularly the ability o the test to be able to accurately assess the knowledge o
students that we estimated at all points in the distribution. 1wo additional concepts are used to present this
35
\hat is potentially conusing here is the circularity o using student responses to calculate item parameters and using the same
parameters to calculate student ability. 1his problem is aoided by assuming a airly innocuous assumption regarding either the
distribution o student parameters in the irst stage ,that o normality o the distribution,, or using the empirical distribution o raw
scores. 1his distributional assumption is used to integrate-out` student-abilities in the conditional likelihood maximization.
155
inormation. 1he irst is the itevivforvatiov, which shows or each item the avovvt of ivforvatiov the item
reeals about the knowledge o students at arious points in the distribution. Intuitiely, a the perectly
discriminating item in ligure A5.1 would yield a lot o inormation regarding students o knowledge 0.25, but
no inormation about those either below or aboe this leel o knowledge. 1he inormation or each item can
then be aggregated to proide te.tivforvatiov, which as the term suggests is a measure o how inormatie the
test, taken in its entirety, is or estimating the distribution o knowledge. 1he .tavaara error o the test- which is
the error in estimating knowledge- is then a unction o the inerse o the test inormation. Ultimately, the
plot o standard error s. knowledge,ability will tell us the accuracy with which our test is able to distinguish
between dierent students in the knowledge distribution, as in ligures 1, 2 and 3 in the main text aboe.
111T MHH/D>@@ HG G1"
\e may also be concerned about the assumptions o Item-Response 1heory or the modeling o our test-
responses. Lach o the items in the pilot was put through a battery o standard goodness-o-it tests, briely,
these compare the predicted answers or children in dierent parts o the knowledge distribution with the
actual answers in the test. Items where there is a large dierence between the predicted answers rom the
structural modeling and the actual answers were re-examined and the inal LLAPS test items are selected
accordingly. 1he test statistic used to assess goodness-o-it was the chi2 goodness-o-it statistics that
compare the obsered proportion o correct responses to predicted responses obtained rom the logistic
model. linally, the unidimensionality assumption was tested or eery test using the eigenalues ollowing
rom actor analysis. lor all 3 tests, the irst eigenalue was at least 8 times as high as the second, with ery
little dierence between the second and consequent eigenalues. 1his suggests that the tests assessed a single
dimension, which we call knowledge.
156
"#*Q< '^T;T 17<0 A$#&#:7<&%-7%:- 9S 'QQ 17<0- %( =>'?@ 7<-7
@.*2<:7
17<0
D.0*<&
/%-:&%0%(#7%9(
@7#()#&) >&&9&
9S
/%-:&%0%(#7%9(
>-7%0#7<
/%SS%:.Q7U
@7#()#&)
>&&9&
[/%SS%:.Q7U\
M.<--%(O
?#�<7<&
@7#()#&)
>&&9&-
[M.<--%(O\
LNGLISl 1 0.8465 0.0242 -2.94356 0.09194 0.01902 0.0182
LNGLISl 2 1.01656 0.02814 -2.2258 0.05834 0.0093 0.0064
LNGLISl 3 1.0329 0.02944 -1.55684 0.04555 0.00508 0.00489
LNGLISl 4 1.1849 0.03839 0.2202 0.02431 0.0118 0.0031
LNGLISl 5 2.44615 0.08562 1.02168 0.0152 0.0143 0.00364
LNGLISl 6 2.18622 0.03028 0.3163 0.00 0.02048 0.00338
LNGLISl 2.28913 0.03125 1.0531 0.00655 0.0029 0.00105
LNGLISl 8 1.9548 0.0302 -0.58533 0.01644 0.09535 0.0016
LNGLISl 9 2.0463 0.024 -0.2045 0.00808 0.00188 0.00154
LNGLISl 10 1.83451 0.02111 0.09919 0.00698 0.00055 0.00053
LNGLISl 11 1.9506 0.02338 0.638 0.00585 0.0006 0.0006
LNGLISl 12 2.95612 0.04398 -0.0898 0.0096 0.12234 0.00541
LNGLISl 13 2.341 0.04396 -0.3852 0.01344 0.19515 0.0036
LNGLISl 14 2.49802 0.016 0.1366 0.01531 0.06532 0.0018
LNGLISl 15 2.0135 0.0644 0.21391 0.0253 0.20314 0.01142
LNGLISl 16 2.55253 0.03342 0.35012 0.0063 0.02286 0.0026
LNGLISl 1 3.56596 0.10329 0.2021 0.01001 0.03523 0.0041
LNGLISl 18 2.64652 0.03535 -0.21606 0.00992 0.098 0.00503
LNGLISl 19 2.42036 0.033 0.85513 0.00586 0.01226 0.0013
LNGLISl 20 2.3628 0.0358 1.280 0.00801 0.00545 0.00105
LNGLISl 21 3.2903 0.09482 1.1685 0.01345 0.0008 0.0006
LNGLISl 22 2.80555 0.04943 1.429 0.01162 0.0008 0.0003
LNGLISl 23 3.09896 0.08096 0.80115 0.01032 0.00512 0.0015
LNGLISl 24 1.3822 0.0694 1.12909 0.0282 0.06303 0.00816
LNGLISl 25 1.82085 0.0843 0.981 0.02056 0.09488 0.00834
LNGLISl 26 2.0919 0.13359 1.41058 0.0313 0.1322 0.00685
LNGLISl 2 0.8991 0.0344 1.342 0.03046 0.1248 0.008
LNGLISl 28 1.30099 0.06906 1.2889 0.03418 0.065 0.00828
LNGLISl 29 2.3985 0.05819 1.00925 0.0062 0.13459 0.0036
LNGLISl 30 3.11145 0.0135 1.15553 0.0046 0.14291 0.00318
LNGLISl 31 1.963 0.1426 1.6032 0.04612 0.1660 0.0088
LNGLISl 32 1.3996 0.045 1.30964 0.0336 0.06353 0.0081
LNGLISl 33 1.81619 0.08141 2.0113 0.05312 0.00193 0.00109
LNGLISl 34 1.315 0.06865 2.053 0.06515 0.00391 0.00225
LNGLISl 35 1.63033 0.0804 2.36093 0.0652 0.0003 0.00068
LNGLISl 36 1.35085 0.06586 2.1446 0.0014 0.00204 0.00135
LNGLISl 3 1.6085 0.09191 2.4102 0.0846 0.00102 0.00085
LNGLISl 38 1.358 0.022 2.163 0.0424 0.0089 0.002
LNGLISl 39 1.49866 0.08458 2.56355 0.10001 0.0008 0.0003
LNGLISl 40 2.845 0.0351 1.31538 0.0068 0.00048 0.0003
LNGLISl 41 3.3128 0.04969 1.49526 0.0028 0.0004 0.0002
LNGLISl 42 3.98209 0.16845 1.64156 0.02148 0.00039 0.00031
LNGLISl 43 3.8690 0.11584 2.22199 0.0208 0.00036 0.00014
LNGLISl 44 3.3282 0.06953 1.9192 0.0134 0.000 0.00023
LNGLISl 45 2.82208 0.06344 1.0348 0.00 0.142 0.00353
15
@.*2<:7
17<0
D.0*<&
/%-:&%0%(#7%9(
@7#()#&) >&&9&
9S
/%-:&%0%(#7%9(
>-7%0#7<
/%SS%:.Q7U
@7#()#&)
>&&9&
[/%SS%:.Q7U\
M.<--%(O
?#�<7<&
@7#()#&)
>&&9&-
[M.<--%(O\
LNGLISl 46 2.01364 0.055 1.45825 0.01313 0.1383 0.00385
LNGLISl 4 1.4454 0.1024 1.59618 0.03881 0.0581 0.00563
LNGLISl 48 2.13839 0.05204 1.40805 0.01099 0.08311 0.00316
LNGLISl 49 0.3241 0.0885 3.0603 0.21544 0.05011 0.01066
LNGLISl 50 0.9615 0.05635 3.4611 0.1341 0.0515 0.0051
LNGLISl 51 0.91312 0.10883 2.8426 0.1833 0.060 0.00951
LNGLISl 52 1.9259 0.13123 2.49325 0.06568 0.13568 0.00385
LNGLISl 53 1.33264 0.08266 2.5169 0.06413 0.10142 0.0043
LNGLISl 54 1.31211 0.02221 -1.5909 0.03004 0.00344 0.00331
LNGLISl 55 1.15058 0.01929 -2.23263 0.0345 0.00381 0.0036
LNGLISl 56 1.23421 0.02056 -1.34249 0.02546 0.00296 0.00284
LNGLISl 5 2.921 0.04335 1.1469 0.0065 0.00269 0.00109
LNGLISl 58 2.64309 0.029 2.08824 0.02124 0.0068 0.00093
LNGLISl 59 2.08092 0.05216 0.44153 0.0166 0.2516 0.0082
LNGLISl 60 2.20655 0.0463 0.1502 0.01052 0.11341 0.00545
LNGLISl 61 1.6666 0.0802 2.8111 0.06421 0.004 0.00103
LNGLISl 62 2.9544 0.05508 1.44412 0.00859 0.00548 0.00105
LNGLISl 63 3.80363 0.08226 1.59432 0.00886 0.00643 0.00084
LNGLISl 64 0.9386 0.0485 2.28309 0.0504 0.08203 0.00685
LNGLISl 65 1.81441 0.09319 1.988 0.03099 0.18088 0.0043
LNGLISl 66 2.84361 0.08629 1.4169 0.0109 0.105 0.0038
LNGLISl 6 1.2006 0.03856 0.9918 0.01161 0.0235 0.00486
LNGLISl 68 3.2898 0.11681 1.40864 0.01104 0.2623 0.00394
LNGLISl 69 1.19493 0.05442 2.0839 0.0368 0.08933 0.0056
LNGLISl 0 1.3582 0.0286 1.60 0.02069 0.0015 0.0010
LNGLISl 1 0.9631 0.0299 2.603 0.0528 0.00056 0.00053
LNGLISl 2 3.1366 0.139 2.68832 0.0494 0.00019 0.00012
LNGLISl 3 2.21485 0.0801 1.0811 0.0131 0.1485 0.00388
LNGLISl 4 1.61226 0.0344 1.82 0.02056 0.00135 0.00091
LNGLISl 5 1.43914 0.04303 2.38562 0.0384 0.00215 0.0012
LNGLISl 6 1.119 0.02695 2.11242 0.03218 0.00109 0.00095
LNGLISl 1.5341 0.03889 2.24496 0.03205 0.00106 0.00063
LNGLISl 8 1.54403 0.0431 2.4818 0.0402 0.00096 0.00058
MA1l 1 1.42515 0.02808 0.09966 0.0224 0.25289 0.0083
MA1l 2 1.19113 0.02482 0.43699 0.02049 0.1388 0.0011
MA1l 3 1.24143 0.05143 0.36555 0.0326 0.0823 0.01246
MA1l 4 1.91901 0.0632 -0.03201 0.03112 0.22109 0.01406
MA1l 5 1.90589 0.0691 -0.0201 0.022 0.15145 0.0128
MA1l 6 1.0105 0.06114 0.0333 0.02835 0.13801 0.01239
MA1l 1.0103 0.0562 0.00232 0.0254 0.09536 0.01124
MA1l 8 1.04263 0.03236 -2.25381 0.06321 0.00539 0.00642
MA1l 9 1.0139 0.01469 -2.6046 0.03422 0.00193 0.0023
MA1l 10 0.86984 0.03331 -0.1333 0.06233 0.02035 0.0126
MA1l 11 1.14338 0.01543 -1.21202 0.01828 0.00099 0.00118
MA1l 12 1.2105 0.02383 -0.06065 0.01099 0.0361 0.00408
MA1l 13 1.5452 0.0214 0.62235 0.00863 0.0156 0.00246
158
159
@.*2<:7
17<0
D.0*<&
/%-:&%0%(#7%9(
@7#()#&) >&&9&
9S
/%-:&%0%(#7%9(
>-7%0#7<
/%SS%:.Q7U
@7#()#&)
>&&9&
[/%SS%:.Q7U\
M.<--%(O
?#�<7<&
@7#()#&)
>&&9&-
[M.<--%(O\
MA1l 14 0.0685 0.53962 0.01129 0.0122
MA1l 15 2.26011 0.9444 0.0061 0.00094 0.00049
MA1l 16 2.68 0.04094 0.008 0.06943 0.00391
1 3.02693 0.08842 0.1621 0.0332 0.0058
MA1l 1.61985 0.02424 1.04086 0.00898 0.00213
MA1l 19 0.0329 0.486 0.002 0.04108
MA1l 20 1.9643 0.60991 0.0046 0.02406 0.00238
2.52653 0.00328
0.02691
0.11893
MA1l 0.01192
18 0.01491
2.341 0.0029
0.02583
MA1l 21 2.55883 0.06824 0.6406 0.01138 0.0098 0.00291
MA1l 22 2.34508 0.0328 0.23124 0.00844 0.06059 0.003
MA1l 23 1.43691 0.0298 2.0696 0.02108 0.006 0.00123
MA1l 24 1.0816 0.01662 -1.92198 0.03221 0.00691 0.0066
MA1l 25 1.44163 0.0108 -0.00624 0.0088 0.00111 0.0012
MA1l 26 1.41641 0.01991 0.9088 0.00924 0.00361 0.0014
MA1l 2 1.29891 0.0200 -0.6345 0.02095 0.03609 0.00696
MA1l 28 1.3688 0.0144 0.8141 0.008 0.00084 0.00086
MA1l 29 1.89958 0.0155 1.0142 0.01962 0.02289 0.00434
MA1l 30 1.589 0.02415 0.210 0.00891 0.02052 0.00316
MA1l 31 1.81638 0.0249 1.5293 0.01125 0.0004 0.00034
MA1l 32 2.1563 0.02816 1.16462 0.004 0.00155 0.000
MA1l 33 2.34891 0.04659 2.2253 0.019 0.00048 0.00026
MA1l 34 1.31864 0.03 2.1323 0.02564 0.0260 0.00256
MA1l 35 1.84941 0.0923 1.89224 0.0506 0.00366 0.0016
MA1l 36 1.26991 0.06825 3.6522 0.11012 0.0048 0.00116
MA1l 3 0.48565 0.01804 4.80166 0.15506 0.0006 0.0009
MA1l 38 2.2043 0.09484 3.03462 0.055 0.00236 0.00044
MA1l 39 2.346 0.06326 2.5098 0.0211 0.00132 0.0003
MA1l 40 1.9629 0.03682 2.19913 0.02003 0.00033 0.00025
MA1l 41 4.2532 0.39364 2.1631 0.0631 0.00115 0.00043
MA1l 42 1.01899 0.0318 2.26842 0.03319 0.0329 0.00343
MA1l 43 1.14854 0.03159 2.002 0.0254 0.02159 0.00295
MA1l 44 2.16982 0.05019 2.34116 0.0230 0.00053 0.00029
MA1l 45 1.1019 0.04443 2.9999 0.06093 0.00 0.00169
MA1l 46 1.36128 0.13894 3.50545 0.10419 0.15838 0.00324
MA1l 4 1.09843 0.03313 0.02696 0.0306 0.0031
MA1l 48 2.03642 0.04543 2.2286 0.02124 0.00048
MA1l 49 1.4029 0.0285 0.064 0.00548
MA1l 50 2.66689 0.04199 0.01091 0.0015
51 0.96635 0.016 -2.15189 0.03551 0.0015
MA1l 52 1.32306 0.02612 -0.1515 0.00832
MA1l 53 2.48566 1.4366 0.00886 0.00029 0.00022
MA1l 0.03224 0.8195 0.00666 0.00388 0.00111
55 1.94855 0.0306 1.23284 0.00893 0.00133
URDU 1 0.0533 0.01188 0.00348 0.00585
URDU 2 -2.01605 0.03363 0.00234 0.00405
3 1.445 0.02401 -0.9555 0.02353 0.00918
URDU 4 1.49811 0.02496 -0.896 0.0086
2.01294
0.0010
0.108 0.01496
1.429 0.05118
MA1l 0.0018
0.02293 0.0868
0.03802
54 2.35439
MA1l 0.0016
-0.48468 0.02681
0.511 0.01119
URDU 0.006
0.02356 0.01631
@.*2<:7
17<0
D.0*<&
/%-:&%0%(#7%9(
@7#()#&) >&&9&
9S
/%-:&%0%(#7%9(
>-7%0#7<
/%SS%:.Q7U
@7#()#&)
>&&9&
[/%SS%:.Q7U\
M.<--%(O
?#�<7<&
@7#()#&)
>&&9&-
[M.<--%(O\
URDU 5 0.894 0.01251 -0.1691 0.01356 0.00024 0.00042
URDU 6 1.91426 0.0221 0.2353 0.00985 0.04888 0.0040
URDU 1.163 0.0218 -0.0835 0.0135 0.054 0.0055
URDU 8 1.35248 0.04045 1.0865 0.0238 0.00052 0.00089
URDU 9 1.065 0.02226 1.33149 0.0092 0.00019 0.0003
URDU 10 1.8933 0.03539 -0.32069 0.01935 0.23334 0.0095
URDU 11 1.695 0.06203 0.93024 0.01804 0.02001 0.0046
URDU 12 0.0294 0.69002 0.002 0.02414 0.00258
URDU 13 2.63928 0.0385 1.5329 0.00826 0.00141 0.00049
URDU 14 2.6922 0.10044 1.564 0.02595 0.00081 0.000
URDU 15 2.8015 0.09065 1.1134 0.0144 0.00463 0.00198
URDU 16 1.89853 0.02584 0.4833 0.0086 0.02258 0.00281
URDU 1 1.53 0.02482 -0.0624 0.01193 0.04459 0.00449
URDU 18 1.90119 0.05459 0.40861 0.0146 0.01421 0.00493
URDU 19 1.52044 0.0204 0.52841 0.01259 0.0448 0.00509
URDU 20 1.50545 0.02569 0.63492 0.01122 0.0419 0.00441
URDU 21 1.33112 0.05156 0.44 0.0245 0.03699 0.0091
URDU 22 0.88529 0.01532 1.61349 0.02013 0.00053 0.0009
URDU 23 1.3688 0.0321 1.58964 0.01646 0.0514 0.00428
URDU 24 1.01095 0.0211 1.3659 0.01562 0.0049 0.00328
URDU 25 1.91648 0.1333 2.45201 0.10099 0.00159 0.0008
URDU 26 1.42993 0.021 1.85932 0.01 0.00486 0.00125
URDU 2 2.3881 0.0885 1.4495 0.02386 0.00303 0.00133
URDU 28 1.60492 0.022 1.468 0.01164 0.00909 0.00161
URDU 29 2.3016 0.05688 2.429 0.02662 0.0003 0.00032
URDU 30 2.10989 0.0621 2.9621 0.0492 0.00012 0.00014
URDU 31 3.0161 0.084 0.596 0.00956 0.008 0.00244
URDU 32 2.92059 0.0363 1.26442 0.00584 0.00096 0.00036
URDU 33 3.2541 0.08443 0.8492 0.00964 0.0025 0.00122
URDU 34 3.23898 0.03884 1.02563 0.0042 0.00133 0.00049
URDU 35 3.3535 0.08618 0.6658 0.00903 0.00904 0.00208
URDU 36 1.83694 0.0252 0.8959 0.0045 0.01904 0.00205
URDU 3 2.42224 0.03339 0.84645 0.00614 0.0221 0.00201
URDU 38 2.9612 0.08249 0.61594 0.0105 0.02452 0.003
URDU 39 2.59485 0.0898 0.136 0.01218 0.03611 0.00432
URDU 40 2.96539 0.08103 0.60028 0.01036 0.0212 0.00346
URDU 41 1.82036 0.083 1.59509 0.0344 0.0004 0.00225
URDU 42 1.16611 0.05119 1.64384 0.04536 0.00255 0.0029
URDU 43 2.4634 0.03191 0.9362 0.00581 0.01224 0.00143
URDU 44 2.91288 0.08849 0.9285 0.01161 0.0114 0.00241
URDU 45 2.425 0.03939 1.48508 0.0064 0.0018 0.0005
URDU 46 2.8229 0.112 1.35101 0.01934 0.0114 0.0019
URDU 4 2.4541 0.1222 1.092 0.03844 0.00604 0.0014
URDU 48 2.82462 0.12885 1.139 0.03245 0.00306 0.00099
URDU 49 2.58916 0.045 1.4105 0.0085 0.00643 0.00113
URDU 50 1.82212 0.0309 1.49618 0.01118 0.0023 0.00103
2.02298
160
161
@.*2<:7
17<0
D.0*<&
/%-:&%0%(#7%9(
@7#()#&) >&&9&
9S
/%-:&%0%(#7%9(
>-7%0#7<
/%SS%:.Q7U
@7#()#&)
>&&9&
[/%SS%:.Q7U\
M.<--%(O
?#�<7<&
@7#()#&)
>&&9&-
[M.<--%(O\
URDU 51 1.6813 0.02982 1.2544 0.01026 0.0109 0.0020
URDU 52 2.44388 0.04405 1.2856 0.009 0.02154 0.00198
URDU 53 0.81583 0.01358 -1.4591 0.0308 0.0013 0.003
URDU 54 1.81262 0.03682 2.16209 0.0215 0.00014 0.00021
URDU 55 2.32155 0.0362 1.2062 0.0056 0.00563 0.0014
URDU 56 2.3213 0.03484 1.291 0.0084 0.0015 0.00091
URDU 5 2.69515 0.04892 1.85505 0.0123 0.00028 0.00024
URDU 58 2.30 0.03815 0.255 0.01005 0.0582 0.0046
URDU 59 0.54593 0.01325 1.4528 0.02992 0.00055 0.00095
URDU 60 1.82683 0.05686 2.65564 0.03935 0.00163 0.000
URDU 61 2.54493 0.054 2.43263 0.026 0.00158 0.0004
URDU 62 2.86968 0.03989 1.1495 0.00603 0.00235 0.0001
URDU 63 2.8052 0.04269 1.4539 0.0053 0.00068 0.00032
URDU 64 2.9919 0.05011 1.68636 0.00908 0.00012 0.00012
URDU 65 2.00863 0.042 1.495 0.0138 0.0093 0.00159
URDU 66 2.21346 0.0423 1.3198 0.00903 0.03045 0.00249
URDU 6 1.881 0.03311 1.3096 0.01002 0.0093 0.0019
URDU 68 2.5945 0.046 1.0948 0.01059 0.00249 0.000