Elisabetta Trevisan Discussion by Ernesto Villanueva (Bank of Spain)

1. Job security and New Restrictive Permanent Contracts: Are Spanish Workers More Worried of Losing their Job? Elisabetta Trevisan Discussion by Ernesto Villanueva (Bank of Spain)

2. 1. Question of the paper • QN: How is worker’s satisfaction with job security affected by changes in firing costs? • Spanish labor market costs: • 2 contracts FT (fire-at-will), and PC (45 days per year) • 1998: (perm) contracts workers < 30 or >45 reduced firing costs (treated group) • 2-step procedure • Identify “comparable control group” using matching • D-in-D treated vs. Control. • Several treatment groups: PC and FT

3. 1. Findings • Cohabitation as a good for parents (Manacorda and Moretti, 2005) • Credit markets (Martins and Villanueva, 2006) • Ruiz-Castillo and Martínez-Granado (2002) • De la Rica and Iza, 2004: decision to get married. Use fixed-term contracts. • Subjective measures of employment risk (Becker et al., 2004)

4. 2. Contribution • Work examining real impact of legal firing costs (Angrist, Autor) • Work looking at subjective perceptions (Clark) or at expectations. • This paper: How do workers perceive increased risk of losing jobs as brought about by legal reforms? • Broad implications, possibly beyond labor econ. • Labor market: investment in job-specific skills • Health economics: stress • Macro: consumption and saving.

5. Concerns • 1. What is exactly being measured? • 2. What can this strategy (D-in-D) measure? • 3. Why is it measured this way? • Why D-in-D + matching? • Why FT as a treated group?

6. 3. What does the paper want to measure? • Option 1: Perception of job insecurity? • QN: Has the subjective probability of losing the job increased? (interpersonally comparable, unlike satisfaction). • Spanish EFF has some information • Option 2: A predictor of wellbeing or of fulfilment of expectations? • Question on how to compare across individuals • Make sure we are not picking “something else”.

7. 2.What does this strategy capture? • D-in-D captures all effects of a reform with GE effects • Wage increases (Lazear)? • More jobs available for all? For the treated? • The reform moves other aspects of job satisfaction (beyond job security) • Subjectivity bias may change reports of all job-related satisfaction components. • ECHP contains satisfaction w/job and w/other aspects (some unrelated, like commuting) • Recommend exploring overall satisfaction w/job and other dimensions.

8. 3. The empirical strategy (i) • D-in-D plus propensity-score matching • Heckman et al. (98) • Matching required to “balance covariates” • D: deterministic function of age (used to match) • Compare person 28 w/ children and person 32, no children, same P. Really comparable? • Same story for employment experience. • Could start by plotting group-specific trends in satisfaction (checking if they are parallel).

9. 3. Which treatment group? • Version 1: workers with a PC already. • Directly affected in their current job. • Version 2: workers with FT. • Affected if contract upgraded or in new job. • The latter case involves the impact estimated many other margins.

10. Other concerns (FE) • Discrepancy between results that include fixed-effects and those that do not. • Given substantial heterogeneity in formation of satisfaction, FE advisable (Hamermesh, 99) • But FE, if truly constant, should be uncorrelated with legal changes. • Is the sample composition changing? • Attrition among FT contracts who find new job?

11. Other concerns (ii) • How are standard errors computed? • Choice of bandwidth of kernel/locllinreg used when propensity-score matching? • Sensitivity of choices?