Sunteți pe pagina 1din 6

Econometrics STG 16,17 2010

ECONOMETRICS, ECONOMIC DATA AND PROBABILITY

Two of the existing definitions of econometrics:


1) Econometrics may be defined as the social science in which the tools of economic theory,
mathematics and statistical inference are applied to the analysis of economic phenomena.
A.S. Goldberger
2) Econometrics, the result of a certain outlook on the role of economics, consists of the
application of mathematical statistics to economic data to lend empirical support to the
models constructed by mathematical economics and to obtain numerical results.
P. A. Samuelson, T. C. Koopmans v J. R. N. Stone
Methodology of econometrics:
1) Development of hypothesis or theory;
2) Data collection;
3) Specifying mathematical model of theory;
4) Specifying econometric model of theory;
5) Estimating the parameters of the model;
6) Checking for model adequacy;
7) Testing the hypothesis derived from the model;
8) Using the model for prediction or forecasting.

ECONOMIC DATA
There exists two types of data:
1) Experimental obtained within the laboratory environment characterised by accuracy;
2) Observational collector cannot interfere with the data formation process characterised by
non-accuracy and non-intentional (?) mistakes done while collecting.
Economic data is observed data.It is collected in two basic dimensions:
1) Time-series data colleced on the same indicator of the same YEAR GDP (mln manats)
observation object over some time. For example.: GDP of 1996 2732.6
Azerbaijan for 1996-2008: 1997 3158.2
1998 3440.6
Time-series are usually notated by (t) subscript. Time series may be 1999 3775.0
2000 4718.2
annual, quarterly, monthly, weekly, dayly and even tick-by-tick. 2001 5315.6
2002 6062.5
USD 0,8035 2) Cross-sectional data collected on 2003 7146.5
EUR 1,1745
the same indicator of different observation 2004 8530.2
RUR 0,0267
objects for one period. For example.: 2005 12522.5
GBP 1,2874
AUD 0,7061 Exchange rates of 16 foreign currencies 2006 18037.1
2007 26815.1
AED 0,2188 against manat on September 30, 2009 (1
CNY 0,1177 2008 38005.1
DKK 0,1578
Foreign currency = ? AZN):
SEK 0,115
CHF 0,7771 Usually (i) subscript is used.
ILS 0,2135
CAD 0,7429
EGP 0,1461
NOK 0,1379
PLN 0,2773
SGD 0,5679

Apart from these, it is common to use some sort of combination of the two data dimensions in
econometrics. Most popular ones are pooled and panel data (to be exlained at the lecture).
For consistency of econometric research it is advised to have at least 25-30 observations.

1 Kamil lsgrov
Econometrics STG 16,17 2010

PROBABILITY
First, lets define some useful concepts:
Statistical or random experiment is any process of observation or measurement which has more
than one outcome and there is uncertainty about which outcome will actually materialize.
Example: throwing a die has 6 possible outcomes.
Sample space or population is the set of all possible outcomes of an experiment.
Example: consider two games on the same day:
1) Galatasaray Yuventus
2) Neftchi Dinamo
Home wins H
Draw D
Guest wins G
There are 9 possible outcomes for those two games (call it set L):
{HH, HD, HG, DH, DD, DG, GH, GD, GG} = L
Sample point each member of sample space. Example: outcome HH from previous example
Event is a particular colection of outcomes and a subset of population.
Example: Consider we name the event one home wins, the event A:
A = {HD, HG, DH, GH}, A L
If the occurence of one event prevents the occurence of another at the same time we name them
mutually exclusive events.
If all of the events have the same possibility to happen then they are equally likely events.
If the event covers all the possible outcomes of the experiment then it is called a collectively
exhaustive event.

First pair Second pair Outcomes Stochastic or random variables value is determined by the
H H 0 outcome of an experiment.
H D 2 Consider the pair of football games as in above example has
H G 3
D H 0
happened 15 times:
D D 2
D G 1 The numbers show the value of the variable.
G H 0
G D 2
G G 5
A discrete random variable takes only finite numbet of values (or countably infinite).
A continuous random variable takes any value in some interval of values.

Probability:
Classical or priory definition if an experiment can result in n mutually exclusive and equally
likely outcomes and m of them are favourable for the event A then: P(A) = m/n

Relative frequency approach Result, % Absolute frequency Relative frequency


Example: Econometrics exam results of 09 0 0
STGs 16, 17: 10 19 0 0
20 29 0 0
If the number of trials is sufficiently large, 30 39 0 0
we can treat the relative frequencies as 40 49 0 0
probabilities. 50 59 0 0
60 69 0 0
70 79 15 0,47
80 89 12 0,38
90 99 5 0,15
Total 32 1

2 Kamil lsgrov
Econometrics STG 16,17 2010

Properties of Probability:
1) For any event A:
0 P(A) 1
2) For mutually exclusive events:
P(A+B+C+...) = P(A) + P(B) + P(C) + ...
3) For mutually exclusive and collectively exhaustive events:
P(A+B+C+...) = P(A) + P(B) + P(C) + ... = 1
4) If the events are statistically independent:
P(ABC...) = P(A) P(B) P(C) ...
P(ABC...) joint probability (prob. of events occuring simultaneously):
5) If the events are not mutually exclusive:
P(A+B) = P(A) + P(B) P(AB)

Conditional probability the probability of event A in case event B has occured:


P( AB)
P( A B) = ; P( B) > 0
P( B)
Conditional probability is calculated for statistically dependent variables. In case of statistical
independence it takes the form:
P( AB) P( A) P( B)
P( A B) = = = P( A)
P( B) P( B)

Probability distribution of random variables:


Discrete random variables:
Consider X takes the values {xi , i = 1, n} . Its probability function (wwhich is called prob. mass
function - PMF) is:
f(X=xi) = P(X=xi), i = 1, n
= 0 if (xxi)
The following are true for the discrete random variables:
0 f(xi) 1
n

f (x ) = 1
i =1
i

Continuous random variables:


If is called prob. density function (PDF):
x1

P(x1 < X < x 2 ) = f(x)dx


x2

Cumulative Distribution function:


F(X) = P(Xxi), i = 1, n
Look at page 38

3 Kamil lsgrov
Econometrics STG 16,17 2010

Multivariate probability density functions


Example: Computer store sells PC and printers. The sales history for the last 200 days is as follows:
Number of computers sold (X)
0 1 2 3 4 Total
Number of 0 6 6 4 4 2 22
printers 1 4 10 12 4 2 32
sold (Y) 2 2 4 20 10 10 40
3 2 2 10 20 20 54
4 2 2 2 10 30 46
Total 16 24 48 48 64 200

Converting to relative frequency:


Number of computers sold (X)
0 1 2 3 4 Total
f(Y)
Number of 0 0.03 0.03 0.02 0.02 0.01 0.11
printers 1 0.02 0.05 0.06 0.02 0.01 0.16
sold (Y) 2 0.01 0.02 0.10 0.05 0.05 0.23
3 0.01 0.01 0.05 0.10 0.10 0.27
4 0.01 0.01 0.01 0.05 0.15 0.23
Total 0.08 0.12 0.24 0.24 0.32 1.00
f(X)

Relative frequencies here are examples of multivariate prob.-s.


f(xi,yi) = P(X=xi,Y=yi), i = 1, n
= 0, if Xxi, Yyi
Statistical independence see computer store example

Characteristics of probability distributions:


Probability distribution are characterised by their moments.
Expected value the first moment of probability distributions is defined as follows

Properties of expectedd value:


1. Expectation of const. is const. itself: E(a) = a
2. E(aX) = aE(X)
3. E(aX+b) = aE(X) + b
4. E(X + Y) = E(X) + E(Y)
5. E(XY) E(X)E(Y)
If X and Y are statistically independent it holds with equity sign.
6. E(X/Y) E(X)/E(Y)
7.
Expected value of multivariate probability distribution:

4 Kamil lsgrov
Econometrics STG 16,17 2010

Variance measures variables dispersion around its expected (mean) value:

Practical definition:

Square root of variance is standard deviation of the variable.

Properties of variance:
1. Variance of const. is zero: var(a) = 0
2.
3.
4. For statistically independent variables:

Coefficient of variation is defined as follows:

Small COV means constancy of variable around its mean.


Covariance is the measure of how two variables vary together.
If for X and Y v , covariance is defined the following way:

Properties of covariance:
1. For statistically independent variables covaiance is zero.
2.
3.

Correlation coefficient defines the how strongly the two variables are linearly related:

Skewness is the measure of asymmetry of prob.


distribution around its mean:

Kurtosis is the measure of flatness of prob distribution:

5 Kamil lsgrov
Econometrics STG 16,17 2010

NOTE: Everything above is defined in terms of discrete random variables. For continuous random

From population to sample

We usually do not observe the whole population, so we have to deal with a part of it which is called
a sample. Therefore all the characteristics of probability distribution we have seen above must
calculated for a sample. For more details see D. Gujarati (2006).

References:
1. D. Gujarati, Essentials of Econometrics, 3rd edition, McGrawHill, 2006
2. D. Gujarati, Basic Econometrics 4th edition, McGrawHill, 2004
3. Website of the Central Bank of Azerbaijan Republic, www.cbar.az

6 Kamil lsgrov

S-ar putea să vă placă și