Everyone agrees that evidence-based policy is likely to be more constructive and effective than that based on dogma or fancy. The problem, for those concerned with social or economic policy, is that we seldom have the luxury of being able to undertake controlled experiments of the type conducted by natural scientists. Instead, we have to draw our inferences from the analysis of non-experimental data, and that is the function of econometrics. This introductory course is intended to serve two constituencies:
Professionals - Each year the course is attended by many professionals who have found that the acquisition of econometric skills would be valuable in their work. Included in this category are PhD students, typically in disciplines other than economics, who are including a serious empirical component in their dissertations.
Undergraduate students - Many participants are college students from other universities. Those from the US ought to be able to negotiate credit worth at least one semester since the teaching is at the same standard as that for EC220, the regular-year LSE course taken by economics majors, and the course is distinctly more ambitious in both coverage and depth than the typical one-semester introductory econometrics course in the US.
The first 7 days (21 teaching hours) are devoted to the Classical Linear Regression model, including extensions to logarithmic regressions and the use of dummy variables, treatments of variable (mis-)specification, the testing of linear restrictions, and models appropriate for heteroscedastic data. The remaining 5 days (15 hours) are devoted to common problems arising in the practical use of econometrics, such as measurement error and the fitting of models with simultaneous equations, and to the fitting of dynamic models, including autoregressive models, with time series data.
The material gives emphasis to the analysis of the finite sample and asymptotic properties of least squares and instrumental variables estimators, and the accompanying implications for statistical inference, under different assumptions concerning the data generation process. The derivation of asymptotic results is coupled with the use of simulation methods to establish finite-sample properties. The course contains many proofs in simple contexts.
The course uses college algebra supplemented by the differential calculus at a basic undergraduate level when it is appropriate and useful. It does not use matrix algebra.
Examples of simple applications in economics are used throughout. Participants use Stata to fit educational attainment and wage equation models with cross-sectional data and EViews to fit demand functions with time series data. Technical support is provided.
In addition to its technical content, the course emphasizes the development of intuitive understanding. The aim is that participants should at all times understand why the material is useful and necessary.
C. Dougherty, Introduction to Econometrics, Oxford University Press, (4th edition) 2011.
*A more detailed reading list will be supplied prior to the start of the programme
**Course content, faculty and dates may be subject to change without prior notice