Documente Academic
Documente Profesional
Documente Cultură
Robert E. HALL
National Bureau of Economic Research, Stanjord, CA 94305, USA
In this paper. a theory of the natural or equilibrium rate of unemployment is built around a
theory of the duration of employment. Evidence is presented that most unemployed workers
became unemployed because their previous jobs came to an end; only a mtnority are on
temporary layoff or have just enttred the labor force. Thus. high-unemployment labor markets
are generally ones where jobs are brief and there is a large flow of newly jobless workers. The
model of the duration of employment posits that employment arrangements are the efficient
outcome of the balancing of workers’ and employers’ interests about the length of jobs. Full
equilibrium in the labor market also requires that the rate at which unemployed workers find
new jobs be elficient. The factors influencing the resulting natural unemployment rate are
discussed. Under plausible assumptions, the natural rate is independent of the supply or demand
for labor. Only the costs of recruiting, the costs of turnover to employers, the efficiency of
matching jobs and workers, 2nd the cost of unemployment to workers are likely to influence the
natural rate of unemployment strongly. Since tlrese are probably stable over time, the paper
concludes that fluctuations in the natural unempiayment rate are unlikely to contribute much to
fluctuations in the observed unemployment rate.
1. Introduction
Milton Friedman’s (1968) famous definition of the natural unemployment
rate runs as follows:
*This research was supported by r:le National Science Foundation and IS part of the NBER’s
program of research on Economic Fluctuations. The paper has not undergone the review
accorded official NBER publications: in particular, it has not been submitted for approval by
the Board of Drrectors. The author,is grateful to Robert Barre for helpful comments.
154 R.E. Hall. Natural unemployment rate
‘In th.is respect, the paper’s objectives arc much the silrnc a.s Lucas and Prescott’s (1974).
Their concern is primarily with creating an apparatus for a complete farmal drszny,tion of the
stochastli: equilibrium of the labor market: in this respect thtir model is rdthcr more
microeconomic than the one presented here. They are not specifically concerned with the
duration cf employment, but their model embodies the grl!erA principle that workers laiive their
jobs whenever the expectedreturns to search elsewlrereexceed ths evpec~cdmums to rem~inmg
on the job.
RX. Hall. Natwai unemployment rate 1%
2. A model sf unemployment
The efficient duration of employment depends on the cost of recruiting to
the employer and on the cost of finditlg new jobs to the worker. Tight
markets where jobs arc cCtsy to furd make workers more receptive to shorter
jc!hs and tugher separation rates. but inrpcrsc higher recruiting costs on
em&>! ers 50 ~rnploycr~) fr-!vor lor:b:cr _K~x. Though the analysis of the
eflicicrtt durGrri)n of cmplopment dp$es for almos: a+~ specification of the
operation t>f the labor market, it seems useful to carry on the discussiorl
within ir articular moilci where 11 IS possible to be completely clear about
the mechailics of unemployment and its role in the economy.
156 R.E. Hall. Natural unemployment rate
1 -f=(l- l/U)J
=[(l- l/U)-u]-J [‘.
.f= 1 -emJiU.
Since job offers are made at random, some job seekers may receive more
than one offer in one period, and employers must generally make more olfers
than the number of jobs they hope to fill. Of the J offers made, vf are
accepted. The number of offers needed to yield an expectation of one
acceptance is p = J/tJf. But J/U is functionally related to $: J,lU = - log( I
-.I’). so p is just a function off(see lig. l),
0.5 10
Fig. 1
U-(1 -f)~‘+sE,
or
b?,‘E= s .f.
Now the unemployment ratr defined ad the fraction of the labor force that is
unemployed throughout the period, is
s
= _- .-__.
s +_f/( 1 -$) -
The quantity f/(1 -$) is the numb&r of _%bs found by all job-seekers for e2.ch
person who remains unemployed throughout the period.
The next section of the paper examines the determinants of the separation
1% R.E. Hall. Narural mvn~!armcnr rare
rate, s, and the subsequent section does the same for the job-finding rate,J
Then the two are combined to form a model of the equilibrium or natural
unemployment rate.
I separation
rata
Fig. 4
is the efficiency condition expressed by the tangency in the diagrams) but not
permit one party to take advantage of the other, This problem has been
discussed extensively in the rather different context of employment contracts
where employers insure workers against fluctuations in demand. The present
discussion will not attempt much of an answer to the problem, but rather
will pursue the implications of contracting over duration in cases where both
parties follow the rules after the contract is made. The simplest rule is just to
R.E. Hall. Natural atnempbyment rate 161
There is another function of S that gives rhe wage at the point of tangency,
but it \\ill IW be needed in what follows.
Fig. 6
Under some further assumptions about costs and preferences, much more
can be said about the eflkient job-finding rate. Suppose that firms have well-
defined job slots, which need to be filled some of the time and not others.
The firm’s ability to avoid unnecessary staffing is limited by the separation
rate it has promised; in the limit of zero separations, jobs must always be
occupied if they are ever filled. Let A(s) be the probability that a job will be
tilled, as a function of the separation rate.” Then the average wage cost
i
associated with the job is WA(S).The hiring rate needed to keep the job filled
is s&s), which requires a flow of offers of p(j)sA(s). Suppose each oiler costs
the firm rw (r is the fraction of a period’s wage that an offer costs). Then the
firm is interested in minimizing the cost function,
C(w,s,f) = Gt(s)(rsp(_f)+ I ) .
.P=(l --u)W
‘The function E.(s) is the oukome of the oplunnl sraflinp policy given rhc scpararlon KM. It
will depend on the stochasric properties of demand and olhrr cansidcrallons not rrcated
explicitI} here.
R.E. Hall. Natural unemploymnt rate 163
Then the efkient job-finding rate minimizes cost subject to the constraint of
achieving a given effective income, y,
subject to
or
recwting cost. I’. the clficicnt jclb-l’llPdii1g riltc 1%a\most 1lnchuiIgcd ilCrO\Z it
5. Equilibrium unemployment
Equilibrium in the labor market requires first that the number of people
looking for work balance the number of people who want to work, net of
those who are in the process of finding work, and second that the terms of
er.+4oyment be efficient. Conditions in the market are measured by the wage
rate, w, the separation rate, s, and the job-finding rate, J The demand for
labor is a function of these three variables: L!(rc,s,f). Similarly, the number
of workers attracted tc the market is a function of the same variables,
L”(w,s,f). However, a fraction, a, of the workers will be looking for work at
any one instant, so the appropriate equilibrium condition is
LD(w,s,f)=(l -u)L”(w,s,f),
or
f ‘
dP(~d,f~=-f)S+f mw,f).
Give>.)a wage level, w, the utility function g(w,s,/) and the two equations
can be solved to get the efficient separation rate and job-finding -rate as
functions of w,
There are good reasons to believe that neither s nor $ is very sensitive to w -
R.E. Hall. Natural unempl jyment rate 165
or in a diagram.
Equilibrium occurs at the intersection of the demand schedule I!‘(w) and the
net supply of labor, (1 - rc~w))1&\(w~. The level of employment. E. is the
horizontal coordinate of this point, and the labor force L. is the point on the
labor supply schedule, E(w), at the equilibrmm wage.
Under arild assumptions about the smoothness of 111~various functtons. an
equilibrium is guaranteed to exist m the labor market. Beyond this rather
general statement, not much can be said about tile equilibrium without
introducing further assumpttons. 1 hc separabtlity wndttions d~scusscd III the
previous section turn out to give some much more definite results. In the first
place. when both costs and utility ;frc proportronai to the wage. the ct‘ficlent
wparatton rate at-113job-findtng rate are independent of the level of tiorker
satisfaction. The eficient separation rate, ~(1’). is clbtained from
166 R.E. Hall. Natural unemployment rate
and depends only on f, the occupancy rate, I(s), and the recruiting rate, I:
The efficient job-finding rate, 9(s), is obtained from
and depends only on s and the recruiting cost, r, as discussed earlier. The
intersection of the two schedules determines the efficient levels of s andf, and
so determines the efficient unemployment rate, quite independently of the
equilibrium levels of employment and she labor force. Under the separability
assumptions, the natural unemployment rate is unaffected by shifts in either
the labor supply or labor demand functions.
It is possible (but tedious) to demonstrate the following propositions about
4(f) and $6):
(1) The efficient separation rate, 4(f), is an increasing function of the job-
finding rate,f - jobs are briefer if they are easier to find.
(2) The efficient separation rate is a decreasing function of the recruiting
cost, r.
(3) The efficient separation rate increases if the occupancy rate, A(s), shifts
so as to lower the marginal cost of a higher separation rate.
(4) The effkient job-finding rate, $(s), is an increasing function of the
separation rate - if jobs last longer, they should not be as easy to Iind.
(5) The etlicient job-finding rate is a decreasing function of the recruiting
cost.
(6) The efficient job-finding rate is unaffected by shifts in the occupancy
rate, A(s).
AS shown in the previous section, the efficient job-finding rate, e(s), is very
insensitive to s. It appears that the efficient separation rate, Qi($). is similarly
insensitive to$, though this depends on the curvature of the occupancy rate
function, J(s). If so. one schedule is nearly horizontal and the other is nearly
vertical, as shown in fig. 8. The ekient combination is at the mtersection of
the two schedules, f * and s*, but ,f* is pretty much determined by $(s) and
s* by 4(f). An increase in recruiting costs shifts +(I) to the left and e(s)
downward, so both f* and s* decline. Turnover declines, but the effect on
the natural unemployment rate is ambiguous. A change in the occupancy
rate that shifts 4(f) to the right has no effect on $(s), so almost all the effect
is to increase the separation rate. The natural unemployment rate increases.
Evidence from cross sections has tended to show that most differences in
167
6. Concluding remarks
In the model of this paper, the natural unemployment rate is the outcome
of eficient employment arrangements. When the labor market is in equilib-
rium and unemployment is at its natural level, no alternative combination of
wages, job duration, and jab-finding rates could make workers better ofI
without also increasing costs to employers. If the matching processembodied
in the model is the best that can be dane to bring workers and employers
together, then the natural rate just described is also the social optimum.
There are no externalities in the model, and no good case for government
intervention in the labor market. It appears that the socially optimal,
natural unemployment rate is quite low. The only influence limiting the
tightness of the labor market is the cost of congestion in recruiting that tight
markets impose on employers. This is important only as the job-finding rate
approaches one.
How strong are the economic forces that push the labor market toward its
equilibrium? A full answer requires a theory of the market in disequilibrium,
which exceeds the ambitions of this paper. A few comments do seem
168 RX. Hnll. Narurol rrrwrnploymerrr rcltv
References
Feldstein. Martin, 1975, The importance of temporary layoffs: An empirical analysis. Brookings
Papers on Economic Activity 3, 725 -744.
R.E. Hall. Narural unemployment rate 163
Friedman, Milton. 1968. The role of monetary policy. American Economic Reklew 58, March, I
IS.
Hall, Robert E.. 1972, Turnover in the labor force, Erookings Papers on Economic Actlklty 3.
709- 756.
Lucas, Jr., Robert E. and Edward C‘. Prescott. 1974, Equihbrium search and unemployment,
Journal of Economic Theory 7, Feb., t8% 209.
Pencavel. John H., 1932, Wages, specific training, and labor turnover rn U.S. manufacturmg
industries. International Economrc Retlew 13. Frb.. 53 64.