Class Question 15, Part I Incidence of unemployment insurance

Transcription

1 Class Question 15, Part I Incidence of unemployment insurance Teresa Steininger EC426 - Public Economics February 2, 2018 eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

2 Paper Recap Change in unemployment insurance (UI) affects... the search strategy of unemployed (micro effect) equilibrium labour market conditions (incidence) In late 1980s UI programme (REBP) in Austria extends UI benefit duration for eligible 50+ in certain regions (52w to 209w) Authors identify the size of market externalities by looking at job search outcomes of non-eligible workers searching in the same labour market Findings: non-eligible workers have a higher job finding probability, lower unemployment duration, smaller risk of long-term unemployment UB duration labour market tightness eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

3 Class Questions Discuss Lalive, Landais and Zweimuller (AER, 2015) [1]: 1 Describe the empirical strategy to identify the incidence of UI on labor market tightness 2 Explain what random matching means. How can we identify the presence of random matching in the labor market? eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

4 Question 1: Identification Quasi-experimental setting: Restructuring of steel industry post 1986 UI extension in regions where steel industry important to mitigate LM consequences Eligibility requirements: 1 age last job + residence in a selected region 3 continuous work history 4 start of spell between 1998 and 1993 eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

5 Question 1: Identification Quasi-experimental setting: Difference-in-Difference set-up treated labour markets (M=1) untreated labour markets (M=0) eligible treated workers (T=1) non-eligible non-treated workers (T=0) Potential outcome yim T eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

6 Question 1: Identification Quasi-experimental setting: Difference-in-Difference strategy Effects of interest: average externality on non-treated in treated LMs: AE NT T Identifying assumption: unobserved differences between non-treated workers in M=1 vs M=0 remain fixed over time (i.e. parallel trends for non-eligible workers in REBP and non-rebp countries) E(y 0 i1 T = 0, M = 1) E(y 0 i0 T = 0, M = 0) = E(y 0 i1 y 0 i0 T = 0, M = 1) }{{} AE NT T + E(y 0 i0 T = 0, M = 1) E(y 0 i0 T = 0, M = 0) }{{} selection eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

7 Question 1: Empirical Implementation Main regression specification: Y it is the search outcome of interest H is an indicator of REBP-eligibility amounts to pooling of two dif-in-dif equations: one for REBP effect on eligible unemployed, one for non-eligible unemployed eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

8 Question 1: To-do list for implementation: Define treated group in REBP regions (T = 1, M = 1) exclude workers 54+ and women to minimize concern that REBP can be used as a partway to retirement Define untreated group in REBP regions (T = 0, M = 1) see Question 2 Define control labour markets in non-rebp regions (T = 1, 0, M = 0) workers with similar characteristics leave out regions which are too integrated (possible spillover effects) issue for parallel trend assumption: region-specific shocks solution: restrict sample to workers never observed in the steal industry LLZ show robustness of results to these sample restrictions eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

9 Question 2: Random matching Goal: measure extent of market externalities by estimating the effect on non-eligible unemployed in REBP regions searching in the same labour market Necessary condition: Eligible and non-eligible workers search in the same labour market Def: labour market is as a market where workers compete for the same vacancies The firm cannot distinguish between eligible and non-eligible workers Conditional on search effort, the firm will be randomly matched with an eligible or non-eligible worker. Labour supply is: n s = p n s a + (1 p) n s b eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

10 Question 2: Why is random matching necessary? If the firm cannot (ex-ante) distinguish between eligible and non-eligible workers vacancy posting will be a function of the expected profit of opening a vacancy. E(π) = p π a + (1 p) π b Changes in an outside option for one group affect overall vacancy posting, and not just for the group who experienced the UI extension Prediction of random matching: characteristics of a vacancy have little predictive power on the eligibility status of the worker who fills it eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

11 Question 2: Identification of Labour Markets Prediction of random matching: characteristics of a vacancy have little predictive power on the eligibility status of the worker Empricial strategy: 1 Regress P(worker eligible worker fills vacancy) on the characteristics of the vacancies. (logit model) 2 Run model for different categories of non-eligible workers. 3 Compare various measures of fit Good measure of fit: eligible workers fill vacancies with very different characteristics to non-eligible workers. Bad fit: cannot tell conditional on characteristics which group will fill the vacancy eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

12 Random Matching: Graphical Evidence Figure 1: Evaluating degree of competition for identical vacancies (p.3575) Goodness of fit declining in age of non-eligible workers. Similarly, Type-I errors particularly large for non-eligible 50+ in baseline sample include workers aged eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13

13 References R. Lalive, C. Landais, and J. Zweimüller. Market externalities of large unemployment insurance extension programs. American Economic Review, eresa Steininger (EC426 - Public Economics) Class 15 - slides to a & b February 2, / 13