1. 1 Comment on Zabel/Schwartz/Donald: An Analysis of the Impact of SSP on Wages Alexander Spermann Mannheim 28 October 2006

2. 2 Why is this paper important? Time-restricted employee or employer subsidies are a promising instrument of Active Labour Market Policy Wage progression: Main channel to self sufficiency Card et al. 2001: Limits to wage progression

3. 3 Few empirical papers to time-restricted subsidies so far: e.g. Michalopoulos et al. 2005 on SSP (JPubEc): evidence from social experiments with the Canadian SSP e.g. Orszag/Snower 2003 (LE): numerical simulation on time-restricted wage subsidies (Benefit-Transfer Program) e.g. Spermann/Strotmann 2005 (ZEW DP): quasi- experiment with the Targeted Negative Income Tax (TNIT) in Mannheim; now implemented in Germany This paper focuses exclusively on wage progression.

4. 4 How does the Canadian SSP work? Source: Blank, Card and Robins (2000)

5. 5 What is done in the paper?

6. 6 1. Decomposing the program group The program group includes participants (randomly assigned) of the SSP. The take-up group comprises those participants who took up a supplemented full-time job. The incentivized group is the part of the take-up group that would not have worked full-time without the program. The non-incentivized group would have worked anyway.

7. 7

8. 8 Problem: The subsets of the take-up group cannot be directly observed. Solution for identification: Propensity Score Matching 2. Identification

9. 9 Result of Matching: First step: Non-incentivized program group: PS- Matching to link CG (465 members) and PG (465 members) via predicted wage probabilities Second step: Remaining 389 member of the take- up group constitute the incentivized program group Question 1: Does the selection-on-observables assumption hold?

10. 10 incentivized non-incentivized 465 389

11. 11 Simple comparison of median wages for non-incentivized program group reveals positive effect (month 15-52) Same procedure for incentivized program and control group. But: Simple comparison of median wages is not meaningful, because the incentivized control group did not work during the first 13 months Question 2: Why exactly?

12. 12 3. Measuring the impact of SSP on wages Problem: Partial observability of starting and ending wages w i is not observed for all those who are working in time s Question 3: Why don‘t you motivate this crucial point more clearly?

13. 13 Solution: Estimation of a model of individual wages -Observed labor experience -Other covariates

14. 14 Absolute wage progression: Mean returns for program and control groups in % Relative wage progression (=TOT): Difference of these means in ppt

15. 15 Three treatment effects: TOT of SSP TOT of non-incentivized program group TOT of incentivized program group

16. 16 Problems to overcome –Sample selection –Endogeneity of experience –Unobserved individual effects Appropriate Solutions –Heckman Two step –IV –FE due to Hausman test

17. 17 4. Results observed experience in months estimated absolute wage progression in % relative wage progression (=TOT) in ppt PG11.66.4 Δ1.9 CG8.24.5 TOT of SSP for British Columbia by month 52:

18. 18 TOT of non-incentivized program group: + TOT of non-incentivized program group after the end of the supplement period: + TOT of incentivized program group: ++ (8.9 ppt)

19. 19 5. Discussion 1. What’s new? Card et al. 2001 apply Matching as well 2. Policy Conclusion Can’t we be more optimistic? What should be done to improve wage progression?

20. 20 3. Lessons for Germany - Low Wage Progression in Germany (Rhein et al. 2005) - Inspiration for TNIT evaluation in Germany