ECONOMETRICS OF PANEL DATA Michele Cincera

Transcription

1 ECONOMETRICS OF PANEL DATA Michele Cincera (indicate Panel in the subject field!) A. THEORY 1. Introduction 2. One way Error Component Model 3. Test of hypotheses 4. Heteroskedasticity and serial correlation 5. Dynamic Panel Data Model 6. Limited Dependent Variables Models 7. Nonstationary Panels 8. Other topics B. ECONOMETRIC SOFTWARES * examples of panel data sets * benefits and limitations of panel data * pooled model * LSDV model * between model * fixed effects model * within model * random model * poolability of data * individual effects * Hausman test * heteroskedasticity * serial correlation * F.D. GMM model * system GMM model * Sargan test of over-identifying restrictions * negative binomial model * QGPML model * CML model * GMM model * panel unit root tests * panel cointegration tests * unbalanced panel data * measurement errors * selection bias * rotating panel * attrition * spatial panel * simultaneous equations TSP45 - STATA (Intercooled Stata 7.0 for Windows NT/98/95) GAUSS (for Windows NT/95 Version ) C. READING LIST C.1. Books Baltagi, Badi H., Econometric Analysis of Panel Data, John Wiley & Sons, 2nd edition, October 2001, 293p. 1

2 Wooldridge, Jeffrey M., Econometric Analysis of Cross Section and Panel Data, MIT Press, 1st edition, October 2001, 740p. Hsiao, Cheng, Analysis of Panel Data, Cambridge University Press, 2nd edition, December, 2002, 368p. C.2. Articles Cincera M., 1997, Patents, R&D and Technological Spillovers at the Firm Level: Some Evidence from Econometric Count Models for Panel Data, Journal of Applied Econometrics, 12, p Cincera M., 2002, Financing constraints, capital and R&D investment decisions of Belgian firms, National Bank of Belgium Working Paper, N 32. D. PRESENTATIONS AND AMENDMENTS date of presentation: to be announced! Name Topic Database NB datasets available at: 2

3 1. Introduction 1.1 Definition of Panel data Pooling of observations on cross section of households, firms, countries, over several time periods Surveying a number of households or individuals and following them over time Typically N = 30, 1000, and T = 2, 5, 10, Examples of panel data sets a) PSID: Panel Study of Income Dynamics (Institute for Social Research) b) NLS: National Longitudinal Surveys of Labour Market Experience (Center for Human Resource Research at Ohio State University and Census Bureau) c) HRS: Health and Retirement Study (Institute for Social Research at the University of Michigan) d) GSOEP: German Socio-Economic Panel (German Institute for Economic Research) e) Belgian Socioeconomic Panel (University of Antwerpen) f) SLID: Canadian Survey of Labour Income Dynamics (Statistics Canada) 3

4 g) French Household Panel h) Hungarian Household Panel (Social Research Informatics Center) i) BHPS: British Household Panel Survey (Institute for Social and Economic Research at the University of Essex) j) JPSC: Japanese Panel Survey on Consumers (Institute for Household Economy) k) ISEP: Dutch Socio-Economic Panel (Statistics Netherlands) l) RLMS: Russian Household-based Survey m) SHP: Swiss Household Panel n) PSELL: Luxembourg Panel Socio-Economique o) ECHP : European Community Household Panel (EUROSTAT) 4

5 Belgium Collected by: Years Available: Sample Description: Data Contents: Site Address: Canada Collected by: Years Available: Sample Description: Data Contents: Site Address: United States Collected by: Years Available: Sample Description: Data Contents: Site Address: Belgian Socio-economic Panel (SEP) Concept, data processing and analysis by Centre for Social Policy, University of Antwerp (UFSIA). Fieldwork by Dimarso (Gallup Belgium), Brussels 1985, 1988, 1992, 1997 Sample is representative of Belgian households and individuals in 1985, 3800 in1988, 3800 in 1992 (including new sample of 900 households), 4632 households in 1997 (including new sample of 2375 households) Demographic characteristics, activity status, employment situation, monthly (net) income of all household members, household wealth, housing situation, life-style indicators, subjective income evaluation. Survey of Labor Income Dynamics (SLID) Statistics Canada The Survey of Labor and Income Dynamics (SLID) is a longitudinal household survey designed to follow the same respondents for several years. The sample is approximately 35,000 households located throughout all ten provinces. The survey gathers information on the economic well-being of individuals and families, and the changes that influence this well-being. SLID also collects information about other related topics such as education and disabilities. Panel Study of Income Dynamics (PSID) Institute for Social Research at the University of Michigan The Panel Study of Income Dynamics (PSID), is a longitudinal study of a representative sample of U.S. individuals and the family units in which they reside. Sample size has grown from 4,800 in 1968 to 6,434 in Core topics: Income sources and amounts, poverty status, public assistance in the form of food or housing, other financial matters (e.g., taxes, inter-household transfers), family structure and demographic measures, labor market work, housework time, housing, geographic mobility, socio-economic background, and health. Other supplemental topics: housing and neighborhood characteristics, achievement motivation, estimating risk tolerance, child care, child support, and child development, job training and job acquisition, retirement plans, health, kinship, wealth, education, military combat experience, risk tolerance, immigration history, and time use. 5

6 United States Collected by: Years Available: Sample Description: Data Contents: Site Address: China Collected by: Years Available: Sample Description: Data Contents: Notes: National Longitudinal Survey (NLS) Census Bureau and the Center for Human Resource Research (CHRR) at Ohio State University The first set of surveys, initiated in 1966, consisted of four cohorts. These four groups are referred to as the "older men," "mature women," "young men," and "young women" cohorts of the NLS, and are known collectively as the "original cohorts." In 1979, a longitudinal study of a cohort of young men and women aged 14 to 22 was begun. This sample of youth was called the National Longitudinal Survey of Youth 1979 (NLSY79). In 1986, the NLSY79 was expanded to include surveys of the children born to women in that cohort and called the NLSY79 Children. In 1997, the NLS program was again expanded with a new cohort of young people aged 12 to 16 as of December 31, This new cohort is the National Longitudinal Survey of Youth 1997 (NLSY97). Surveys include data about a wide range of events such as schooling and career transitions, marriage and fertility, training investments, child-care usage, and drug and alcohol use. The depth and breadth of each survey allow for analysis of an expansive variety of topics such as the transition from school to work, job mobility, youth unemployment, educational attainment and the returns to education, welfare recipiency, the impact of training, and retirement decisions. Chinese Longitudinal Healthy Longevity Survey (CLHLS) National Archive of Computerized Data on Aging (NACDA) The largest sample size of centenarians and nonagenarians as compared to any other studies focusing on elders in the world, CLHLS was conducted among the most elderly population in counties and cities of 22 provences of China with the general goal of shedding light on the determinants of healthy human longevity and oldest-old mortality. Data are provided on the health, socioeconomic characteristics, family, lifestyle, and demographic profile of this aged population. NACDA seeks to contribute to the intellectual vitality of the gerontological sciences. Site Address: Source : 6

7 Cincera (2003) Financing constraints, fixed capital and R&D investment decisions of Belgian firms, in P. Butzen, and C. Fuss, Firms Investment and Finance Decisions: Theory and Empirical Methodology, Cheltenham, UK: Edwar Elgar. 7

8 8

9 9

10 1.3. benefits and limitations of panel data a) controlling for individual heterogeneity b) panel data give more informative data, more variability, less collinearity among the variables, more degrees of freedom and more efficiency c) panel data are better able to study the dynamics of adjustment d) panel data are better able to identify and measure effects that are simply not detectable in pure cross-section or pure time-series data e) panel data models allow one to construct and test more complicated behavioral models than purely cross-section or time series data f) panel data are usually gathered on micro units, like individuals, firms and households g) design and data collection problems h) distortions of measurement errors i) self-selectivity problems j) nonresponse k) attrition l) short time-series dimension 10