Returns to Apprenticeship Training in Austria: Evidence from Failed Firms Josef Fersterer...
-
Upload
calvin-gilmore -
Category
Documents
-
view
213 -
download
0
Transcript of Returns to Apprenticeship Training in Austria: Evidence from Failed Firms Josef Fersterer...
Returns to Apprenticeship Training in Austria:
Evidence from Failed Firms
Josef Fersterer
Jörn-Steffen Pischke
Rudolf Winter-Ebmer
Role of the Apprenticeship System
• Apprenticeship plays major role in training non-college bound youths in Germany, Austria, and Switzerland
• Perceived as well functioning system which benefits all participants, often seen as model
• This implies high returns but suggestive evidence of important selection effects
Our Identification Strategy
• Use exit of training firms to construct an instrument for length of training– No comparison of apprentices with non-apprentices
• Exit of a firm may cut short apprenticeship for some apprentices– Use potential duration in failing firm as instrument for
total training
• Identifying assumption: Apprentices that joined 1 to 4 years before failure are similar
The Austrian Education System
Apprenticeship in the Dual System
• 40 percent of cohort complete an apprenticeship• Firm based training in a prescribed occupation• One day a week in state-run vocational school• Length 2 to 4 years but 88 percent of contracts for
3 or 3.5 years• Training firms have to be approved, external
exams for apprentices• Apprentices receive a wage (starting low but can
rise to 80 percent of skilled worker wage)
Dropouts in the Apprenticeship System
• About 10 to 18 of apprentices drop out of apprenticeship early
• About a third of apprentices had more than one apprenticeship contract
• Exam pass rate: 85 percent – 90 percent after repeat attempts
The Data
• Universe of employment records of the Austrian Social Security Administration 1975 – 1998
• Firm exit: employer identifier ceases to exist– not more than 70 percent of employees continue under
new firm id
• Estimation sample: Firms with less than 25 percent employment decline in six months before quarter of failure
Data on Individuals
• Identify individuals in failing firms
• Only use those who joined failing firm within 15 quarters before failure
• Only men
• Daily wage
• Neglect top-coding due to contribution maximum (sample is young)
Empirical Framework
• L* is latent apprenticeship duration, related to ability and different human capital levels
• Realized duration L = L* if not affected by firm failure
• K is time from entry into first training firm to failure
• Actual time in failing firm may differ because L* < K or apprentice switches firms
Identification• Key identifying assumption: L* K
– Declining quality in apprenticeship hires before failure would lead to upward bias
• Realized duration L = L(K), maybe the same, shorter, or longer than L*
• Assumption (existence of first stage): E(L(K)) < E(L*)
• Estimating equation
iiii xLw '
Table 1: Sample Characteristics(in percent)
Sample
firm size (beginning of
Estimation sample
joined failing firm within 15 quarters of
failure
joined failing firm 16 or
more quarters before failure
all apprentices
apprenticeship) (1) (2) (3) (4) 0-4 employees 32.7 21.7 17.3 8.5 5-19 employees 41.0 39.2 41.6 34.5 20-99 employees 19.4 26.6 26.1 29.8 100 and more 6.9 12.5 15.0 27.3
sample size 9,105 23,893 103,201 676,368
Table 2: Length of Apprenticeship by Potential Duration in Failed Firm
Eventual length of quarters to failure
Apprenticeship 1 to 3 4 to 10 11 to 15 16+
Shorter than time to failure 0.9 8.9 15.2 15.9
Within one quarter of failure 8.1 11.8 1.7 0.0
Longer than failure/incomplete 12.0 5.4 0.0 0.0
Complete (11 to 13 quarters) 61.4 57.3 68.5 65.0
Longer (14 to 15 quarters) 17.6 16.6 14.5 19.1
Mean duration (days) 1025 1025 1063 1073
No. of obs. per quarter 463 687 860
First Stage Relationship between Length of Apprenticeship and Quarters in Failing Firm
(all firms)
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
4 5 6 7 8 9 10 11 12 13 14 15
Potential Duration in Failing Firm (Quarters)
Impa
ct o
n Le
ngth
of
App
rent
ices
hip
(Yea
rs)
Table 5: Returns to Length of Apprenticeship Training
(all firms)
OLS OLS IV LIML (1) (2) (3) (4) Length of apprentice- 0.016 0.029 0.020 0.019 ship in years (0.002) (0.002) (0.013) (0.013) Controls Experience Age Age Age R2 0.51 0.50
Reasons for Small Returns Compared to Previous Estimates
• Comparison completers/dropouts rather than apprentices/non-apprentices
• Estimates of the impact of the latter part of training, maybe mostly on-the-job experience– But: estimates contain effect of obtaining credential
• Everybody in the sample changes employer, so no effects of firm specific human capital
• Firms in the sample are very small • Failing firms may have poorer training• Sample excludes self-employed• We observe returns when workers are relatively young
(average age is 26)
Conclusion
• Attempt to estimates effects of apprenticeship training on earnings free of selection bias
• Identification only seems to work in small firms
• Small estimates of the returns to apprentice-ship training but similar to OLS little selection in dropout behavior?