Zero Returns to Compulsory Schooling in Germany: Evidence and Interpretation

1
Zero Returns to Compulsory Schooling in Germany
Evidence and Interpretation

• Jörn-Steffen Pischke (LSE)
• Till von Wachter (Columbia University)

2
Why another paper about compulsory schooling
laws?

• IV estimates of returns to schooling exist for a
  variety of countries, using compulsory schooling
  laws as instruments
• German states extended basic secondary schooling
  from 8 to 9 grades between 1948 and 1970
  research design very similar to Acemoglu and
  Angrist (2000), Oreopoulos (2003)
• Returns to compulsory schooling around 10 15
  percent in other countries, zero in Germany

3
Previous literature

• Angrist and Krueger (1991) use season of birth
  for the US, 7.5 percent return
• Harmon and Walker (1995) UK, 15 percent return
• Acemoglu and Angrist (2000) US, 10 percent
  return
• Oreopoulos (2003) US, Canada, UK, 10 15 percent
  return
• Meghir and Palme (2003) Sweden, 18 percent
  return
• Aakvik, Salvanes, and Vaage (2003) Norway, 10
  percent return
• Grenet (2004), Oosterbeek and Webbing (2004)
  small returns for France and Netherlands

4
Overview

• The German school system
• Data
• Empirical approach and results
• Why these differences to the previous literature?

5
The German school system
University
Other Schools
Apprenticeship (2 to 4 years)
Middle Track (6 years)
Optional Grade 10
Academic Track (9 years)
Basic Track (4 years or 5 years)
Primary School (4 years)
6
Figure 1Shares of students by secondary track
Academic Track
Middle Track
Basic Track
7
Introduction of 9th grade in basic track
8
Why a 9th School Year?

• Early 1950s Labor market arguments
• From mid-1950s Educational arguments
• Maturity of school leavers
• Basic skills
• Social and political competencies
• Curricula difficult to glean purpose
• Combination of basic skills, political education,
  and occupational orientation

9
Data Qualification and Career Survey

• Samples working Germans age 15-65, about 25000
  workers in each wave
• Four cross-sections 1979, 1985/86, 1991/92,
  1998/99
• Earnings Gross monthly wage reported in 13 - 22
  brackets (use bracket midpoints)
• Lots of information on schooling and training

10
Years of Education

• German data record secondary degree (track) but
  no highest grade attended variable
• Typical approach convert usual number of years
  taken for a particular educational program
  (secondary or post-secondary)
• 8th or 9th grade in basic track has to be
  assigned on the basis of year of birth and state
• Only state of residence available
• Length of schooling Secondary school graduation
  year year of birth 6

11
Micro Census

• Annual household survey of 1 percent of the
  population
• Use 1991, 1993, and 1995-2001 surveys
• Earnings measure Net income in 18 or 24 brackets
  (use bracket midpoints)
• Information on secondary track and post-secondary
  qualifications

12
Social Security Data(IAB Employee Sample)

• 1 percent sample of social security records
• Excludes self-employed, civil servants, marginal
  jobs
• Annual panel data from 1975 to 1995
• Earnings measure gross pay subject to social
  security contributions (censored on top)
• Education information only know academic versus
  lower track, state of workplace in 1975
• Cell level data, 12566 cells based on 2.4 million
  micro records

13
Sample

• 1930 1960 birth cohorts
• 10 West German states (excluding Berlin)

14
Measurement Framework

• Wage equation
• Schooling equation (first stage)
• Reduced form

15
Table 2 First Stage Regressions
16
Table 2 Attends Basic Track
17
Table 3 First Stage RegressionsQualification
and Career Survey
18
Figure 2Length of School Basic Track Students
19
Figure 3 Average Number of Grades Attended by
School Leavers Four States
20
Table 4 Log Wage Regressions
21
Table 5 Reduced Form Regressions
22
Table 5 Reduced Form Regressions Basic Track
23
Implied IV Estimateswith Squared State Trends
24
Interpretation Why the difference between
Germany and other countries?

• Measurement issues measurement error, mobility,
  etc.
• The German wage setting system is rigid, and
  therefore returns to skill are low
• True returns to schooling are low because of the
  role of the apprenticeship system
• The returns to 9th grade schooling are low in
  Germany because key skills relevant in the labor
  market are learned earlier

25
Table 6 Reduced Form Regressions Employment and
Self-Employment Micro Census
26
Interpretation Why the difference between
Germany and other countries?

• Measurement issues measurement error, mobility,
  etc.
• The German wage setting system is rigid, and
  therefore returns to skill are low
• True returns to schooling are low because of the
  role of the apprenticeship system
• The returns to 9th grade schooling are low in
  Germany because key skills relevant in the labor
  market are learned earlier

27
Table 7 Reduced Form Regressions Basic Track
Only Micro Census
28
Interpretation Why the difference between
Germany and other countries?

• Measurement issues measurement error, mobility,
  etc.
• The German wage setting system is rigid, and
  therefore returns to skill are low
• True returns to schooling are low because of the
  role of the apprenticeship system
• The returns to 9th grade schooling are low in
  Germany because key skills relevant in the labor
  market are learned earlier

29
Basic skills are learned earlier in Germany

• Drop-outs/basic school leavers only use basic
  skills in the labor market, these skills are
  still being learned in the US and UK at age 14
  16, but not in Germany
• Other skills are learned on the job (particularly
  in apprenticeship in Germany)
• This story means returns in Germany must be high
  to some grades, just not grade 9.

30
Table 8 Primary and Secondary School Skills Used
on the Job
31
Table 9 Quantitative Test Scores IALS 1994-98
32
Table 10 Scores on the FIMS 1964 Age 13,
Standardized scores (mean 0, std 1)
33
Summary

• Large returns to compulsory schooling in many
  countries
• Evidence suggests that the returns to the 9th
  grade for basic track students in Germany are
  zero or small
• Different results most likely due to differences
  in when learning takes place