fukui@econ.tohoku.ac.jp http://www.econ.tohoku.ac.jp/~fukui/site.htm 200 7
Cookbook-style
. (Inference) (Population) (Sample) f(x = θ = θ ) (up to parameter values) (estimation) 2
3 (multicolinearity) K+=N R 2 y X cos 2 θ + + + + = + = + = N NK K K K N N K e e e y y y e e X y 2 2 2 0 2... ),..., (, β β β β β
(asymptotic) (conditional homoskedasticity) (eogeneity) F χ 2 Estimator (consistency) OLS E( i e i ) = 0 (orthogonality/predeterminedness condition) estimator consistent GLS White estimator 4
(endogeneity) (serial correlation) (instrumental variables) orthogonality condition GMM AR Durbin-Watson invalid Cochrane-Orcutt (common factor ) 5
GMM Instrumental Variables + White estimator GMM orthogonality condition (over-identified) LS White estimator weight matri OLS GMM Just-identified OLS OLS GMM 6
(Statistical Significance) P(h(θ) X) P(X h(θ)) 5 h(θ) h(θ ) 95 significance level specification search (regression fishing) = correctly specified model 5 test Probability confidence Bayesian? (frequentist) confidence Bayesian diffuse prior 7
Significance power Fisher p-value power Neyman Pearson decision-theoretic approach power 00 θ = θ significance level Likelihood ratio significance level test ( likelihood ratio ) Neyman Pearson likelihood ratio test 8
P(X θ = 0) = 5% test 5% 95%. test Fisher objective epistemic Neyman-Pearson objective behavioral Bayesian inter-subjective epistemic behavioral 9
R 2 AIC R 2 Theil R 2 R 2 R 2 s 2 log(s 2 ) 2 2 N 2 K + log( s ) = log( σˆ ) log( σˆ ) + N K N Kullback-Leibler AIC log( σˆ K ) + 2 N 2 + inconsistent overfit AIC Scwartz consistent 2 K + log( σˆ ) + log( N ) N 0
Estimation R 2 R 2 nested models out-of-sample performance cross validation science art specification search
Hypothesis testing estimation Mehra Prescott equity premium puzzle significance estimates Calibration (simulation) Real business cycle Kydland Prescott hypothesis testing calibration non-parametric estimation bootstrap Conditional epectation significance digression estimates 2
Appendi. Econometrics Tetbooks. Introductory Amemiya (994): Written for theory-oriented beginners. Goldberger (998): Introductory version of Goldberger (99). (999): A Japanese tetbook comparable to Pindyck and Rubinfeld (997). Pindyck and Rubinfeld (997): Application-oriented. Targeted at advanced MBAs. 2. Basic (Used in introductory Ph.D. econometrics courses) Goldberger (99): Unconventional and stimulating in many respects. Greene (999): Benchmark? Comprehensive and widely used among applied researchers. Hansen (200): Lecture notes used at Wisconsin. Concisely and rigorously written. Poirier (995): Written from a Bayesian point of view. 3. Advanced Davidson and MacKinnon (993): Widely used in theory-oriented Ph.D. courses. Hamilton (994): The tet of time-series analysis. Hayashi (2000): Net generation s orthodoy unified under GMM. 3
Amemiya, T. 994. Introduction to Statistics and Econometrics. Cambridge, MA: Harvard University Press Davidson, R., and J. G. MacKinnon. 993. Estimation and Inference in Econometrics. Oford, U.K.: Oford University Press. Goldberger, A. S. 99. A Course in Econometrics. Cambridge, MA: Harvard University Press. Goldberger, A. S. 998. Introductory Econometrics. Cambridge, MA: Harvard University Press. Greene, W. H. 2000. Econometric Analysis. Upper Saddle River, NJ: Prentice Hall. Hamilton, J. D. 994. Time Series Analysis. Princeton, NJ: Princeton University Press. Hansen, B. E. 200. Lecture Notes for Economic Statistics and Econometrics II (http://www.ssc.wisc.edu/~bhansen/70/70.htm). University of Wisconsin, Madison, WI. Hayashi, F. 2000. Econometrics. Princeton, NJ: Princeton University Press. 999 Pindyck, R. S., and D. L. Rubinfeld. 997. Econometric Models and Economic Forecasts. New York, NY: McGraw Hill. Poirier, D. J. 995. Intermediate Statistics and Econometrics: A Comparative Approach. Cambridge, MA: MIT Press. 4