ECONOMETRICS

1
ECONOMETRICS

• Spring semester 2006

2
Introduction

• Historical review
• Basic ideas and approaches in econometrics
• Stochastic models
• Statistical inference

3
Historical review

• Econometrics is a young science. It developed
  during the previous century and primarily after
  World War II.
• Econometric Society 1933 Econometrics is the
  unification of statistics economic theory and
  mathematics. (Frisch 1933)

4
Historical review

• Empirical economic studies have though a longer
  history for instance
• 1676 Sir William Petty The political arithmetic
  studies of peoples ability to pay taxes and of
  the possibilities to recruit military personnel.
• 1699 Gregory King Studies of the price
  variation of corn as a result of weather changes

5
Historical review. The turn of the century 1900

• Engels law Income elasticity of food lt 1
• (Ernst Engel 1821-1896)
• Henry L. Moore studies of business cycles
  including the first attempt to estimate demand
  functions using multiple regression. (Economic
  Cycles Their Laws and Cause MacMillan New York
  1914)
• E.Ph. Makeprang Price theory first
  interdependent system of demand and supply (1906)

6
Historical review. Between the wars

• Cobb C.W. Douglas P. H. (1928) A theory of
  production AER Douglas P.H. (1948) Are there
  laws of production AER
• Sevall Wright (1934) on interdependent systems
  (The Method of Path Coefficients in Annals of
  Mathematical Statistics)
• Henry Schultz on the demand for private
  consumption (Statistical Laws of Demand and
  Supply 1928. The Theory and Measurement of
  Demand 1938 University of Chicago Press)
• Cowles Commission started in Colorado Springs in
  the 1930s and then moved to Chicago (Jacob
  Marschak Tjallings Koopmans Leonid Hurvicz
  Lawrence Klein T.W. Andersson developed LIML)

7
Historical review. Between the wars

• Ragnar Frisch (Econometric society 1933)
• Trygve Haavelmo (1944) The probability approach
  in econometrics Supplement to Econometrica
• Jan Tinbergen Macro models (1937) An Econometric
  Approach to Business Cycle Problems (1939)
  Statistical Testing of Business Cycle Theories.

8
Historical review

• The scientific development of Econometrics was
  made possible by
• The development of the theory for statistical
  inference (R.A. Fisher Jerzy Neyman Egon
  Pearson)
• The increased supply of data
• The development of fast computers.

9
Historical review. The post war period

• Large econometric macro models for business cycle
  analysis and for the analysis of economic growth
  (Tinbergen Klein Klein-Goldberger Brookings
  Centraal Plan Bureau in the Haag.) were not very
  successful and were later replaced by time-series
  models and methods.
• Methods for ID-systems Herman Wold Henri Theil.
  
• Critique of the macro economic applications T.
  Sargan David Hendry and the rational
  expectations hypothesis and Lucas critique.
• Time-series econometrics to a large extent driven
  by applications primarily in macro and in the
  financial markets (stationarity unit-root
  co-integration)

10
Historical review. The post war period

• Micro econometrics (cross-sections repeated
  cross-sections panel data discrete choice
  unobserved heterogeneity self selection) J.
  Heckman and D. McFadden
• Computer intensive methods (robust methods
  sample re-use methods simulation-based
  estimation)
• Experimental economics (new field for old
  methods)

11
Basic ideas and approaches in Econometrics.

• A central concept The Data Generating Process
  (DGP).
• Economic model
• DGP Econometric model
• Measurement instrument

12
Basic ideas and approaches in Econometrics.

• The ceteris paribus assumption
• Specification errors
• Diagnostic tests

13
The DGP is a random process

• Simple random experiment Throw a coin
• Random model
• Sample space Stochastic variable Probability
• Head X1 P(X1)1/2
• Tail X0 P(X0)1/2

14
A market model Structural form

• Demand function
• q1Q1(py)e1
• Supply function
• q2Q2(px) e2
• Market clearance
• q1q2

15
P
q
16
Interpretation of e1 and e2

• e1 and e2 represent (economic) behavior other
  than picked up by y and x
• e1 and e2 are random variables with zero means
• e1 and e2 are assumed uncorrelated with y and x
• e1 and e2 can also represent measurement errors
  in q.

17
A market model The reduced form

• p P(yxe1e2)
• q Q(yxe1e2)
• Marginal distribution of p and q
• F(pq)
• Conditional distribution of p and q
• F(pqyx)

18
Common econometric problems

• Model formulation and data collection
• Estimation
• Diagnostic testing
• Specification testing
• Prediction