Postgraduate course
2024/2025
Modelling heterogeneity: quantile regression and latent class analysis
Type:
Elective course
Area of studies:
Postgraduate Studies
Delivered by:
Department of Applied Economics
When:
2 year, 2 semester
Mode of studies:
offline
Open to:
students of all HSE University campuses
Instructors:
Galina Besstremyannaya
Language:
English
ECTS credits:
2
Course Syllabus
Abstract
The course is targeted at the study of advanced econometric methods for modelling observable and unobservable heterogeneity of various economic agents: individuals, households, firms, countries and regions. The methodology includes quantile regressions and models with a finite number of unobservable classes, which are incorporated in parametric efficiency analysis and policy evaluation techniques. A special attention will be given to applications in the corporate finance, industrial organization and public economics.One method of studying unobservable heterogeneity is the use of latent class approach in various regression models. Here, the researcher assumes that an observation belongs to one of the unobservable groups, which may be associated, for instance, with unverifiable behavioral characteristics of individuals or with immeasurable managerial practices by firms. Another method of tackling heterogeneity is the use of quantile regressions, in case of both cross-sectional and panel data models. Quantile regressions are widely used in evaluation of endogenous and exogenous policy reforms, as well as in the analysis of production and cost function.The analysis is applied to the conditional -th quantile of the dependent variable. Instead of extrapolating the results of the mean regression to the tails of the distribution of the dependent variable, quantile regression enables obtaining independent estimates for the impact of covariates in each conditional quantile of the dependent variable. Different values of the estimated coefficients for the explanatory variable obtained in regressions with different values of are interpreted as the presence of heterogeneous effect of this explanatory variable. For instance, quantile regression may be used for studying heterogeneous effect of policy reforms and macroeconomic shocks on production, or for evaluating heterogeneous effects of socio-demographic characteristics of consumers on their expenditure. Another merit of quantile regression is the applicability for efficiency analysis. High values of quantile index (e.g. 0.8, 0.9) may be taken as an approximation of the production possibility frontier, while in case of conditional quantile regression applied to cost function low values of quantile index (e.g. 0.1, 0.2) may serve an approximation for the best cost minimization trajectory. Sections will be devoted to the study of econometric packages, available for implementation of the analysis (R, Limdep, Stata) and to replications of the scientific papers/textbook chapters, using the datasets on firms and households.
Learning Objectives
- The purpose of this applied course in advanced econometrics is to study the theory and implementation of selected methods for modelling heterogeneity of economic agents in cross-section and panel data regression analysis. One method of studying unobservable heterogeneity is the use of a finite mixture (latent class) approach in various regression models. Here, the researcher assumes that an observation belongs to one of the unobservable groups, which may be associated, for instance, with unverifiable behavioral characteristics of individuals or with immeasurable managerial practicesof firms. Another method of analyzing heterogeneity is the use of quantile regression methodology in cross-sectional and panel data setting. Quantile regressions are widely used in evaluation of endogenous and exogenous policy reforms, as well as in the analysis of production and cost function.
Expected Learning Outcomes
- The students will learn the theoretical foundations and practical implementation of modelling heterogeneity of economic agents through finite mixture models and quantile regression (i.e. model selection, cross-validation, analysis of goodness of fit). The students will also learn how to choose an appropriate method for a given economic problem and how to motivate this choice in academic writing.
Course Contents
- Finite mixture models with cross section OLS regression
- Finite mixture models with panel data
- Finite mixture models and efficiency analysis
- Median and quantile regression
- Panel data quantile regression
- Quantile regression with short samples and smoothing techniques
- Instrumental variable quantile regression.
- Generalized finite mixture models
Assessment Elements
- ParticipationAn in-class empirical analysis econometric, packages related to the contents of homework assigment (dataset provided) or a report of an individual research prodject, implimented with datasets accummulated by the student
- Homework 1Homework is individual work only.
- Homework 2Homework is individual work only.
- EssayA critical review of an empirical paper chosen by the student
- ExamOption 1. An in-class empirical analysis in any econometric package, chosen by the student. The analysis is based on the contents of homework assignment (dataset provided, a modification of the previously analyzed code and/or of a method/algorithm will be required). Option 2. The results of the individual research project (implemented at home and submitted in the form of a short report at the time of exam). The project must deal with an application of either quantile regression models or finite mixture models to datasets accumulated by the student.
Interim Assessment
- 2024/2025 2nd semester0.1 * Essay + 0.7 * Exam + 0.05 * Homework 1 + 0.05 * Homework 2 + 0.1 * Participation
Bibliography
Recommended Core Bibliography
- Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics : Methods and Applications. New York, NY: Cambridge University Press. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsebk&AN=138992
- Econometric analysis of cross section and panel data, Wooldridge, J. M., 2010
- Greene, W. H. (2012). Econometric Analysis: International Edition : Global Edition (Vol. 7th ed., International ed). Boston: Pearson Education. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsebk&AN=1417839
- Joshua D. Angrist, & Jörn-Steffen Pischke. (2009). Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton University Press. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsrep&AN=edsrep.b.pup.pbooks.8769