» Staff


Home People Research Intranet About Seminars News Computing Prospectus Library
You are in: Home » Staff

Staff

University Senior Lecturer
Fellow of Clare College

Tel:	44-(0) 1223 335260
Email:	mw217@econ.cam.ac.uk
Interests:	Microeconometrics, Estimation and Testing of Labour Supply Models. Latent Variables Models. Macroeconometrics, Testing of Non-Linear Threshold Models. Growth Models. Non-Tested Hypothesis Testing and Simulation-Based Inference.

Econometrics: Part IIA - Paper 3

Microeconometrics and Panel Data

The following is more a less a course outline for the first term's work.

Introduction to Microeconometrics
GaussMarkov: A Reexamination in the light of Attributing Causality from Observational Data
Friedman's Critique of OLS Estimation of the Consumption Function
Circumventing the Confounding Effects of Endogeneity: Instrumental Variable Estimation
Discrete Dependent Variables: An Introduction to Binary Response Models
Panel Data Models

To date much of the emphasis in econometrics has been upon the behaviour of aggregates of economic agents, whether one is focussing upon an aggregation of firms in the case of analysing trends in industrial production, or aggregation over all consumers in an economy when we examine broad trends in consumption.

These aggregates represent the sum total of a large number of decisions made by individual consumers and firms. In these series of lectures we examine how the use of Microeconometrics can assist in unraveling some of the determinants of economic behaviour when the observational unit is founded at the micro level. As we will see a focus on individual decision making, and the use of attendant micro data generates a number of interesting problems which to date you have not considered. For example, in many applications where the analysis involves the consumption behaviour of individuals we will observe certain individuals consuming zero quantities; similarly, as a result of either supply or demand constraints, some individuals will record zero hours of work. This distinction between behaviour at the extensive versus intensive margin is important, but often neglected in traditional Marshallian demand analysis where the focus is on the determinants of movement along a given supply or demand curve.

A central theme for the first 3 to 4 topics will be on the notion of causality and how econometrics overcomes the severe problems for inference following directly from the lack of experimental data. Econometricians, in general, work with observational data, that is data that is the result of decisions made by agents such that ex post it is difficult to attribute causality to an observed effect. Following on from the much quoted mantra correlation does not imply causality, we examine a number of ways where causal processes can be confounded by intervening processes.

In the fifth topic we consider how econometric methods may be used to address policy issues where the observed data is binary. Questions such as what are the determinants of labour market participation decisions, what are the determinants of fertiliser use by farmers in the Phillipines, or why do people prefer cars over buses as a principle mode of transport, are characterised by a yes/no or zero/one (binary)\ structure. As such the use of the standard OLS model is not satisfactory. To date you will have encountered dummy variables as discrete right-hand-side covariates. Here we introduce methods which will allow us to model these phenomenon by explicitly accounting for the nature of the data.

In the final topic we combine cross-sectional information with time-series data, and explore how panel data methods can allow us to discriminate between a more complex set of behaviours.

A hard copy of the Course Outline is available.