## _wp15162

## Source details

**Canonical URL:** [_wp15162](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15162.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15162.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15162.pdf.json)

---

### Introduction
- Germany implemented the Hartz reforms in three steps between January 2003 and January 2005.
- Reforms: eased regulation on temporary work agencies; relaxed firing restrictions; restructured the federal employment agency; reshaped unemployment insurance to significantly reduce benefits for the long-term unemployed and tighten job search obligations.
- Unemployment rate trajectory:
  - peaked at almost 11 percent in 2005;
  - declined to five percent at the end of 2014.
- Empirical preview:
  - Normalized log earnings (1992–2002 normalization) show displaced workers’ earnings diverging sharply downward relative to continuously employed workers after 2003, especially in 2005 and remaining lower through 2010.
  - Regression evidence (controlling for worker observables and unobservables and time trends):
    - Before the reforms, the cost of displacement was about 20 percentage points.
    - After the reforms, the penalty increased by another 10 percentage points.
    - Including workers re-entering as part-time raises the penalty further.
  - Results robust to a number of methodological changes.
- Caveat: cannot reliably identify which specific element(s) of the Hartz package caused the observed effects because elements were complementary and introduced within a short period, with possible anticipatory effects and reaction lags.

### Background: German labor market dynamics and theory
- Long-term trends:
  - Unemployment rose from the mid-1970s, reversed around completion of Hartz reforms in 2005; decline paused only briefly during the 2009 downturn.
  - Period of the reforms coincided with prolonged negative growth in average wages (compensation per employee year-over-year growth).
- Theoretical perspectives:
  - Labor demand frictions (hiring/firing costs; insider-outsider dynamics).
  - Labor supply / reservation wage view (Ljunqvist and Sargent, 1998): skill loss from displacement and generous long-term benefits sustain high unemployment; less generous benefits allow wage adjustment and lower unemployment.
- Pre-reform benefit structure:
  - Unemployment benefits (UB): 60 percent of previous net earnings (67 percent for parents); duration up to 12 months for <45, up to 32 months for older workers.
  - Unemployment assistance (UA): 53 percent of previous net earnings (57 percent for parents); could be claimed indefinitely subject to means test.
  - Social assistance (SA): means-tested lump-sum; UB or UA below SA topped up.
- Post-reform structure:
  - UB I: relabeled UB; replacement rate unchanged; duration largely intact for many workers.
  - UB II: replaced UA and SA; means-tested lump-sum paying an amount similar to old SA — most former UA recipients experienced a drastic cut after short-term benefit exhaustion.
- Net replacement rate evidence referenced for 2001–2011 (chart values not reproduced).

### The Hartz reform package (Hartz I–IV) — dates and main measures
- Hartz I
  - Adoption: Dec 1, 2002; Effective: Jan 1, 2003.
  - Measures: new Personnel Service Agencies; vocational education support; deregulation of temporary work sector.
- Hartz II
  - Adoption: Dec 1, 2002; Effective: Jan 1, 2003 and Apr 1, 2003.
  - Measures: subsidy for one-person companies (Me-inc); introduction of mini and midi-jobs exempt from most social security taxes; threshold for layoff rules raised from five to ten workers.
- Hartz III
  - Adoption: Dec 1, 2003; Effective: Jan 1, 2004.
  - Measures: restructuring of the Federal Labor Office.
- Hartz IV
  - Adoption: Dec 1, 2003; Effective: Jan 1, 2005.
  - Measures: shortening of UB duration; merging UA and SA into UB II at lower social benefit levels; new definition of acceptable jobs with sanctions for refusal.
- Sources for table: Eichhorst and Marx (2011), Dlugosz and Wilke (2013).

### Data and empirical strategy
- Data source: Sample of Integrated Labor Market Biographies (SIAB)
  - 2 percent random sample from administrative social security records; coverage 1975–2010 (from 1992 for former East Germany).
  - Includes workers subject to social security contributions (excludes self-employed, civil servants, military).
  - Mini-jobs included starting 1999.
  - Overall coverage: approximately 80 percent of the German workforce.
- Advantages vs GSOEP:
  - SIAB: daily reporting, employer- and agency-reported data, ~100 times larger sample than GSOEP.
  - GSOEP: ~2,000 households per year; annual frequency limits precision for monthly transitions.
- Main data limitations and handling:
  - Earnings right-censored at social security contribution limit — imputed with Pareto tail methods.
  - Hours worked limited to full-time/part-time; primary focus on full-time employment; robustness checks address part-time returns.
  - Hartz reforms changed long-term unemployment beneficiary reporting in 2005–2006; analysis conditions only on employment or short-term unemployment to avoid that glitch.
  - Workers older than 62 excluded.
- Conversion and unique-observation rules:
  - Daily data converted to monthly by counting active days and associated earnings.
  - When multiple spells in a month: drop non-employment if employment active; if multiple employment spells, keep spell with highest pay; classify unemployment status by UB/UB I (short-term) and UA/UB II (long-term); define out of labor force if no valid observation.
- Earnings measure:
  - Real 2013 values using CPI; log of the value taken.
  - Earnings at month t = average log monthly real earnings from t to t+11.
  - Residuals clustered at the individual level due to serial correlation induced by averaging.
- Top-coding approach:
  - Assume Pareto tail using top decile of non-top coded earnings; assign conditional mean above threshold to top-coded values.
  - Alternatives tested: random draw from Pareto, keep top-coded value, drop top-coded observations — no material effect on baseline results.
- Definition of recently displaced (month t):
  - Full-time employed at t−13 for at least 36 months and short-term unemployed at t−12.
  - Alternative definitions and robustness variants explored (e.g., prior-work requirement of 1 year; reentry after 24 months; include part-time/vocational returns).

### Sample periods and descriptive statistics
- Sample last year: 2010.
- Pre-reform period: 1992–2002.
- Post-reform period: 2005–2009 (construction of earnings at t uses data from t to t+11, hence uses data from 2003 and 2010 as well).
- No attempt to separately identify impacts of each Hartz reform due to anticipation effects and phase-in provisions.
- Summary statistics (weighted by individual-months):
  - Full sample: 27 million individual-month observations for almost three hundred thousand individuals.
  - Before Hartz (1992–2002):
    - Displaced: Age 38.22; Non-displaced: Age 42.59.
    - Upper secondary (%) — Displaced 17.79; Non-displaced 15.39.
    - University (%) — Displaced 6.54; Non-displaced 9.85.
    - Average monthly earnings (logs) — Displaced 7.86; Non-displaced 8.18.
    - Fraction top coded (%) — Displaced 11.24; Non-displaced 22.97.
    - Individual-Months — Displaced 11,236; Non-displaced 17,893,823.
    - Individuals — Displaced 10,922; Non-displaced 240,475.
  - After Hartz (2005–2009):
    - Displaced: Age 40.69; Non-displaced: Age 43.93.
    - Upper secondary (%) — Displaced 14.55; Non-displaced 13.36.
    - University (%) — Displaced 10.85; Non-displaced 13.74.
    - Average monthly earnings (logs) — Displaced 7.78; Non-displaced 8.20.
    - Fraction top coded (%) — Displaced 11.19; Non-displaced 21.35.
    - Individual-Months — Displaced 5,209; Non-displaced 7,750,372.
    - Individuals — Displaced 5,182; Non-displaced 171,066.
  - Descriptive differences: displaced workers younger than non-displaced pre-reform by about four years; earnings gap increases from more than 30 log points pre-reform to 40 log points post-reform.

### Main empirical findings: earnings losses after displacement
- Key result: Re-entry earnings of displaced workers declined substantially after the Hartz reforms.
  - Before reforms: cost of displacement ≈ 20 percentage points (log points referenced).
  - After reforms: penalty increased by ≈ 10 percentage points (total ≈ 30 percentage points relative baseline).
  - Inclusion of part-time re-entrants enlarges the measured penalty.
- Interpretation: consistent with frictional search models (Mortensen and Pissarides, 1994; Ljunqvist and Sargent, 1998) — lower unemployment benefits reduce reservation wages and lower post-unemployment earnings.
- Robustness: results robust to multiple methodological changes.

### Regression framework and selected coefficient estimates
- Baseline regression:
  - earnings_{i,t} = X_{i,t}β + γ1 Disp_{i,t} + γ2 Hartz_t + γ3 (Disp_{i,t} * Hartz_t) + ε_{i,t}
  - Hartz_t = 1 for 2005–2009, 0 for 1992–2002.
  - X_{i,t} progressively includes education dummies, cubic in age, interactions, past earnings (t−13 to t−48 average), time trends, occupation/sector controls.
- Selected coefficient estimates (median-age, median-education worker at sample middle date; standard errors clustered):
  - Disp (displacement dummy):
    - Column (1): -.31666* (.00387)
    - Column (2): -.28135* (.00452)
    - Column (3): -.22213* (.02807)
    - Column (4): -.20959* (.02860)
  - Disp × Hartz:
    - Column (1): -.09645* (.00801)
    - Column (2): -.10378* (.00707)
    - Column (3): -.09703* (.01124)
    - Column (4): -.10211* (.01160)
  - Hartz:
    - Column (1): .01775* (.00097)
    - Column (2): -.02264* (.00086)
  - Disp × time:
    - Column (3): .00029* (.00008)
    - Column (4): .00031* (.00008)
  - Average past earnings:
    - Column (3): .99469* (.00060)
    - Column (4): .96612* (.00094)
- Fit and sample sizes:
  - R squared: Column (1) .0009; Column (2) .2610; Column (3) .8657; Column (4) .7624.
  - Individual-months: Columns (1)–(3) 25,660,640; Column (4) 19,901,846.
  - Individuals: Columns (1)–(3) 280,742; Column (4) 240,309.
- Interpretation:
  - Raw earnings differential: recently displaced paid markedly less; raw differential almost 32 log points prior to reforms and increases by a further 10 log points after reforms in uncontrolled specification.
  - Controlling for demographics and past earnings reduces magnitudes but additional loss after reforms remains about 10 log points.
  - Past earnings highly autocorrelated (coefficients .99469* and .96612*).
  - Separate time trend for displaced: earnings of recently displaced grow by .3 log points a year relative to non-displaced (.00029–.00031 per period interaction).

### Robustness checks (selected experiments and results)
- Experiments:
  1. All Employment Types (include part-time/vocational returns).
  2. 3-year Average of earnings.
  3. Prior employment 1-year (loosen requirement).
  4. Re-employment after 2 years.
  5. Controlling for occupation-year (120 occupations).
  6. Controlling for sector-year (nine sectors).
- Selected estimates (Disp and Disp × Hartz):
  - Disp:
    - (1) All Employment Types: -.22118* (.03002)
    - (2) 3-year Average: -.14556* (.02679)
    - (3) Prior employment 1-year: -.25924* (.02439)
    - (4) Re-employment after 2 years: -.14274* (.02874)
    - (5) Occupation-year: -.22013** (.02711)
    - (6) Sector-year: -.22583** (.02751)
  - Disp × Hartz:
    - (1) -.10120* (.01215)
    - (2) -.06730* (.01042)
    - (3) -.09358* (.00995)
    - (4) -.07130* (.01179)
    - (5) -.10008** (.01108)
    - (6) -.10167** (.01124)
  - R squared examples: (1) .8605; (2) .8723; (3) .8206; (4) .8356; (5) .8688; (6) .8650.
- Robustness interpretation:
  - Including part-time/vocational returns does not alter the estimated increase in the displacement penalty after reforms.
  - Averaging earnings over three years reduces estimated costs (from ~22.2 to ~14.5 log points) and the reform impact (from ~10 to ~7 log points), consistent with slow recovery from displacement.
  - Loosening prior-work requirement has marginal impact.
  - Controlling for occupation-year or sector-year has only marginal effect, suggesting switching into lower-paying occupations/sectors is not the main driver.

### Interpreting which elements of the Hartz reforms matter
- Plausible mechanisms:
  - Sharp reduction in long-term unemployment benefits and tighter job search/acceptance requirements → higher hazard of returning to work and larger displacement penalty via reduced reservation wages.
  - Liberalization of temporary agency work → employers may offer lower-paying jobs; temporary work accounted for a sizable fraction of new job creation in some post-Hartz years (Figure 6 referenced).
  - Reform of the employment agency may have improved matching efficiency, potentially coinciding with lower post-unemployment earnings.
- Data limitations prevent precise attribution:
  - SIAB does not identify temporary jobs or firm identifiers; cannot construct a mass-layoff subsample following Jacobson et al. (1993).
  - Therefore cannot disentangle which reform components drove earnings declines.

### Policy implications, limitations, and future directions
- Policy implications:
  - Hartz reforms strengthened incentives to take work (lower reservation wages via benefit cuts), contributing to lower unemployment.
  - Increased cost of displacement implies unemployment became more onerous for affected individuals due to lower subsequent earnings.
- Limitations:
  - No structural welfare analysis in this paper; welfare consequences require accounting for effects on post-unemployment earnings.
  - Cannot identify specific reform components responsible for observed effects because of complementary design and tight timing.
- Suggested extensions:
  - Study migration, probability of subsequent displacement, switching occupation/sector, earnings volatility, future earnings growth.
  - Exploit differential impacts across subpopulations to isolate causal effects and conduct a structural welfare evaluation.
- Additional supportive evidence:
  - Initial exploration suggests workers more affected by the reforms had larger increases in hazard rate of returning to work and larger post-unemployment earnings losses.

*Italic: Source — _wp15162 - 3. Is full-time employed at time t but at a different firm than at tെ13. (IMF PDF chapter/section).*

### References .............................................................................................................

### _wp15162 - References .............................................................................................................

### Introduction
- After a decade of high unemployment and weak growth leading up to the turn of the 21th century, Germany implemented the Hartz reforms in three steps between January 2003 and January 2005.
- Reforms eased regulation on temporary work agencies, relaxed firing restrictions, restructured the federal employment agency, and reshaped unemployment insurance to significantly reduce benefits for the long-term unemployed and tighten job search obligations.
- Unemployment rate trajectory:
  - peaked at almost 11 percent in 2005;
  - declined to five percent at the end of 2014, the lowest level since reunification.
- Empirical preview:
  - Normalized log earnings (1992–2002 normalization) show displaced workers’ earnings diverging sharply downward relative to continuously employed workers after 2003, especially in 2005 and remaining lower through 2010.
  - Regression evidence (controlling for worker observables and unobservables and time trends):
    - Before the reforms, the cost of displacement was about 20 percentage points.
    - After the reforms, the penalty increased by another 10 percentage points.
    - Including workers re-entering as part-time raises the penalty further.
  - Results robust to a number of methodological changes.
- Caveat: cannot reliably identify which specific element(s) of the Hartz package caused the observed effects because elements were complementary and introduced within a short period, with possible anticipatory effects and reaction lags.

### Background: German labor market dynamics and theory
- Long-term trends:
  - Germany experienced a long-term rise in unemployment from the mid-1970s, reversed around completion of Hartz reforms in 2005; decline paused only briefly during the 2009 downturn.
  - The period of the reforms coincided with a prolonged phase of negative growth in average wages (compensation per employee year-over-year growth).
- Theoretical perspectives discussed:
  - Labor demand frictions (e.g., high hiring/firing costs; insider-outsider dynamics).
  - Labor supply / reservation wage view (Ljunqvist and Sargent, 1998): skill loss from displacement and generous long-term benefits sustain high unemployment; implication: less generous benefits allow wage adjustment and lower unemployment.
- Pre-reform benefit structure (three layers):
  - Unemployment benefits (UB): benefit equal to 60 percent of previous net earnings (67 percent for parents with dependent children); for workers younger than 45 the benefit was limited to 12 months, older workers eligible for up to 32 months.
  - Unemployment assistance (UA): replacement rate of 53 percent of previous net earnings (57 percent for parents with dependent children); could be claimed indefinitely subject to means test and annual review.
  - Social assistance (SA): means-tested lump-sum transfer; least generous support; UB or UA benefits below SA level were topped up.
- Post-reform structure:
  - UB I: relabeled UB; replacement rate unchanged; duration largely intact for many workers (12 months), some older workers saw reduction.
  - UB II: replaced UA and SA; means-tested lump-sum benefit paying an amount similar to old SA — most workers who would have qualified for UA under the old system experienced a drastic cut in benefits after short-term benefit exhaustion.
- Net replacement rate evidence (Figure 5):
  - Time series presented for Germany short-term and long-term and OECD short-term and long-term (percent of previous earnings) for 2001–2011 (chart values not reproduced here).

### The Hartz reform package (Hartz I–IV) — dates and main measures (Table 1)
- Hartz I
  - Adoption of law: Dec 1, 2002
  - Effective date: Jan 1, 2003
  - Measures:
    - Setting up of new Personnel Service Agencies
    - Support for further vocational education from the German Federal Labor Agency
    - Deregulation of temporary work sector
- Hartz II
  - Adoption of law: Dec 1, 2002
  - Effective dates: Jan 1, 2003 and April 1, 2003
  - Measures:
    - Introduction of subsidy for one-person companies (Me-inc)
    - Introduction of low paid jobs (mini and midi-jobs) exempt from most social security taxes
    - Threshold size for firms subject to layoff rules raised from five to ten workers
- Hartz III
  - Adoption of law: Dec 1, 2003
  - Effective date: Jan 1, 2004
  - Measures:
    - Restructuring of the Federal Labor Office
- Hartz IV
  - Adoption of law: Dec 1, 2003
  - Effective date: Jan 1, 2005
  - Measures:
    - Shortening of the duration of unemployment benefits
    - Merging of unemployment assistance and social assistance, with benefit set at the lower level of social benefits (unemployment benefit II)
    - A new definition of acceptable jobs with sanctions for refusal of an acceptable job
- Sources for table: Eichhorst and Marx (2011), Dlugosz and Wilke (2013).

### Literature on Hartz and related empirical findings
- Macroeconomic search models calibrated and simulated for German economy (Krause and Uhlig, 2012; Krebs and Scheffel, 2013; Launov and Waelde, 2013): cuts in unemployment benefits significantly reduced unemployment.
- Aggregate matching function estimates (Fahr and Sunde, 2009; Klinger and Rothe, 2012; Hertweck and Sigrist, 2013): important positive effects of earlier reforms (Hartz I and III) on matching efficiency.
- Administrative-data evidence:
  - Dlugosz et al. (2014): considerable decline in transition rates from employment to unemployment after reforms, particularly for older workers.
  - Arent and Nagl (2011): test for structural break in wage equations; argue average wages fell after Hartz.
  - Giannelli et al. (2013): median wage of workers re-entering from unemployment declined in 1998–2010 post-Hartz years.
- Alternative explanation (Dustmann et al., 2014): German labor market success owed to threat of off-shoring and decentralized employer-union negotiations that limited wage growth and increased flexibility in wage bargaining and employment contracts, not primarily Hartz.

### Data and empirical strategy
- Data source: Sample of Integrated Labor Market Biographies (SIAB)
  - A 2 percent random sample from German administrative social security records (sampled to preserve panel structure).
  - Coverage: 1975 to 2010 (starting in 1992 for former East Germany).
  - Includes all workers subject to social security contributions (excludes self-employed, civil servants, military service).
  - People in “mini-jobs” (social security exempt jobs paying less than €400 a month) are included starting in 1999.
  - Overall coverage: approximately 80 percent of the German workforce.
- Advantages of SIAB over GSOEP:
  - SIAB: daily reporting of labor market transitions and income changes (minimizes misclassification of short spells), employer- and agency-reported data (reduces measurement error), much larger sample (roughly 100 times larger than GSOEP), enabling precise analysis conditional on covariates.
  - GSOEP: roughly 2,000 households per year; annual frequency and annual income variables limit precision for monthly transitions.
- Main data limitations and handling:
  - Earnings right-censored at social security contribution limit — imputed using standard approaches.
  - Hours worked limited to full-time/part-time — primary focus on full-time employment; robustness checks investigate sensitivity to these limitations.
- Sample selection decisions:
  - Focus on male workers aged 25 and older to avoid female labor force participation secular/lifecycle issues and timing of labor market entry.
  - Use only data from states in former West Germany for consistency.
  - Use data from 1988 to 1991 to condition on previous employment history and earnings; main analysis begins after reunification (explicit start year not provided in extract).

### Main empirical finding (earnings of workers returning from unemployment)
- Key result: Re-entry earnings of displaced workers declined substantially after the Hartz reforms.
  - Before reforms: cost of displacement ≈ 20 percentage points (in earnings).
  - After reforms: penalty increased by ≈ 10 percentage points (i.e., total post-reform penalty ≈ 30 percentage points relative baseline).
  - Inclusion of part-time re-entrants enlarges the measured penalty.
- Interpretation: findings are consistent with frictional labor-market models (Mortensen and Pissarides, 1994; Ljunqvist and Sargent, 1998) predicting that lower unemployment benefits reduce reservation wages and thus lower post-unemployment earnings.
- Robustness: results robust to a number of changes in methodology (details not reproduced in extract).

### Policy implications and unanswered questions
- Hartz reforms appear to have strengthened incentives for the unemployed to take up work (lower reservation wages via benefit cuts), contributing to lower unemployment.
- However, increased cost of displacement (larger re-entry earnings penalty) implies becoming unemployed became more onerous for individuals.
- Key policy question remains unresolved in extract: which specific element(s) of the Hartz package produced the observed effects cannot be reliably identified due to complementary design and tight timing of measures; anticipating effects and learning lags complicate timing-based tests.

*Source: _wp15162 - References (excerpt of IMF working paper PDF text).*

### 1992. The last year of the sample is 2010. We set the pre-reform period to 1992–2002 and

### _wp15162 - 1992. The last year of the sample is 2010. We set the pre-reform period to 1992–2002 and

### Data and sample periods
- Sample last year: 2010.
- Pre-reform period: 1992–2002.
- Post-reform period: 2005–2009 (because data from t to t+11 are used to construct earnings at t, this also uses data from 2003 and 2010).
- No attempt is made to separately identify the impact of each of the four Hartz reforms due to anticipation effects and phase-in provisions.

### SIAB data structure and limitations
- SIAB reports changes in employment status on a daily basis and average daily gross nominal labor income, calculated as the annual gross income paid by the employer divided by the number of days worked at that job.
- The number of days worked is based on reported start and end dates of the employment relationship.
- Hartz reforms changed how information on long-term unemployment benefit recipients was collected, producing a temporary glitch that impaired data on long-term unemployment beneficiaries during 2005–2006; these data essentially cannot be used for that period.
- To circumvent this limitation, the analysis conditions only on employment or short-term unemployment.
- Workers older than 62 are excluded from the sample since they are not covered by the SIAB.

### Conversion to monthly frequency and unique-observation rules
- For computational purposes, daily data are converted to monthly frequency by calculating the number of days a spell is active and the associated total earnings during the month.
- When multiple spells are active in the same month, a unique observation for an individual-month is defined by the following criteria:
  - 1. Drop non-employment spells if an employment spell is active.
  - 2. If one or more employment spells are active, define the unique observation as the spell paying the highest amount in that month.
  - 3. If no employment spell is active, define a worker as short-term unemployed if he/she receives UB before the reforms or UB I after the reforms, and as long-term unemployed if he or she receives UA before the reforms or UB II after the reforms (but did not receive UB/UB I).
  - 4. Define a worker as not in the labor force if there is no valid observation in that month.

### Earnings measure, real adjustment, and averaging
- Earnings are converted to real 2013 values using the CPI and the logarithm of this value is taken.
- The earnings measure in month t is assigned as the average log monthly real earnings from t to t+11.
  - Purpose: reduce the effect of any initial decline in earnings post displacement that is quickly recovered and reduce noise in the earnings measure.
  - Note: With earnings data averaged across overlapping time periods, regression residuals are serially correlated by construction; residual clustering is done at the level of the individual worker.

### Top-coding and imputation of right-censored earnings
- Earnings are subject to right-censoring at the social security contribution limit.
- Approach: assume the right tail of the distribution of log earnings follows a Pareto distribution and estimate its shape using the top decile of non-top coded earnings.
- Top-coded values are assigned the conditional mean above the top-coded threshold.
- Alternative experiments reported (no material effect on baseline results):
  - Assign top-coded values a random draw from the estimated Pareto distribution.
  - Keep top-coded observations at their top-coded value.
  - Drop all top-coded observations.
- Caveat: top coding remains an issue when analyzing the impact of the reforms within occupations or sectors.

### Education, geography, and agency reporting differences
- Unemployment spells are reported from a different agency than employment spells; unemployment reports do not contain information about education.
- Education is defined as constant over an individual's career and equal to the maximum reported education level.
- Geographic location of an unemployed worker is not available; a worker is defined to be in the East if any of his employment records stems from a state that was part of former East Germany.

### Definition of recently displaced (month t)
- A worker is defined as recently displaced in month t if he satisfies:
  - 1. Was full-time employed at time t−13 and had been so for at least 36 months.
  - 2. Was short-term unemployed at time t−12.

*Source: _wp15162 - 1992. The last year of the sample is 2010. We set the pre-reform period to 1992–2002 and (PDF chapter/section).*

### 3. Is full-time employed at time t but at a different firm than at tെ13.

### _wp15162 - 3. Is full-time employed at time t but at a different firm than at tെ13.

### Definition of "recently displaced" and sample selection
- Recently displaced at time t: workers who experienced unemployment for at least one month between t and t-12 and were back at work at t+13.
- Non-displaced at time t: worker has been employed full-time in the previous four years.
- Robustness variants:
  - Reentry examined after 24 months (alternative selection rule).
  - Loosen previous-work requirement from three years to one year.
  - Include returns to part-time or vocational training (excluding mini-jobs in baseline due to data availability from 1999).
- Rationale and limitations:
  - Restriction to workers with significant previous work history follows Jacobson et al. (1993) and increases likelihood separations were involuntary.
  - SIAB lacks reason for separation; receiving unemployment benefits does not fully rule out voluntary quits because reduced benefits may be available to quitters in Germany.
  - Excluding workers who re-enter at a different firm removes seasonal workers.
  - Excluding short separations (e.g., return within one month) reduces misclassification.
  - These misclassification sources bias results toward finding no impact of the reforms; thus estimates are lower bounds.
- Focus on full-time employees only because hours worked are not available; robustness documents an increase in the probability of returning to part-time employment after the reforms, but the proportion of part-time workers remains small.

### Regression framework to identify impact of the Hartz reforms
- Outcome variable:
  - earnings_{i,t} = average log monthly real earnings over month t to t+11.
- Key indicators:
  - Disp_{i,t} = 1 if worker is recently displaced (as defined).
  - Hartz_t = 1 if year is 2005–2009, and 0 if 1992–2002. (Footnote: to construct earnings at t we need data from t to t+11, so 2003 and 2010 are used as well.)
- Baseline regression:
  - earnings_{i,t} = X_{i,t}β + γ1 Disp_{i,t} + γ2 Hartz_t + γ3 (Disp_{i,t} * Hartz_t) + ε_{i,t}
  - X_{i,t} includes progressively richer controls; standard errors clustered at the individual level.
- Interpretation:
  - Expect γ1 negative (cost of displacement). Hypothesize γ3 negative (adverse effect larger after Hartz due to cut in long-term benefits and tighter job search requirements).
- Controls added progressively:
  - Constant only; then education dummies (three), cubic in age, interactions between age and education groups, interactions between education and displacement, interaction between age and displacement.
  - Past earnings: average log monthly real earnings over months t-13 to t-48 as control for unobservable differences.
  - Separate time trends by education group; linear interaction age × time; separate linear time trend for displaced; linear interaction between quarterly GDP growth and Disp; year and month dummies (Hartz dummy excluded due to collinearity).
  - Alternative specification: occupation or sector controls (120 occupations; nine sectors).

### Summary statistics (pre- and post-reforms)
- Sample:
  - 27 million individual-month observations for almost three hundred thousand individuals.
- Table 2 highlights (weighted by individual-months):
  - Before Hartz (1992–2002):
    - Displaced: Age 38.22
    - Non-displaced: Age 42.59
    - Upper secondary (%) — Displaced 17.79; Non-displaced 15.39
    - University (%) — Displaced 6.54; Non-displaced 9.85
    - Average monthly earnings (logs) — Displaced 7.86; Non-displaced 8.18
    - Fraction top coded (%) — Displaced 11.24; Non-displaced 22.97
    - Individual-Months — Displaced 11,236; Non-displaced 17,893,823
    - Individuals — Displaced 10,922; Non-displaced 240,475
  - After Hartz (2005–2009):
    - Displaced: Age 40.69
    - Non-displaced: Age 43.93
    - Upper secondary (%) — Displaced 14.55; Non-displaced 13.36
    - University (%) — Displaced 10.85; Non-displaced 13.74
    - Average monthly earnings (logs) — Displaced 7.78; Non-displaced 8.20
    - Fraction top coded (%) — Displaced 11.19; Non-displaced 21.35
    - Individual-Months — Displaced 5,209; Non-displaced 7,750,372
    - Individuals — Displaced 5,182; Non-displaced 171,066
- Key descriptive differences:
  - Displaced workers are younger than non-displaced by about four years before the reforms; about two years older after the reforms while the non-displaced are about one year older.
  - Non-displaced have higher fraction of university graduates (9.9 versus 6.5 percent) and both groups are better educated after the reforms.
  - Top-coded observations: more than 20 percent of non-displaced and 11 percent of displaced workers.
  - Earnings gap (log points): displaced are more than 30 log points lower than non-displaced before the reforms; difference widens to 40 log points after the reforms.

### Main regression results (Table 3) — key coefficients and fit
- General notes:
  - All standard errors clustered at the individual level.
  - Displayed estimates are for a median-age, median-education worker evaluated at the middle date of the sample at average GDP growth.
- Selected coefficient estimates and statistics (columns correspond to specifications (1)–(4)):
  - Hartz:
    - Column (1): .01775* (standard error .00097)
    - Column (2): -.02264* (standard error .00086)
  - Disp (displacement dummy):
    - Column (1): -.31666* (.00387)
    - Column (2): -.28135* (.00452)
    - Column (3): -.22213* (.02807)
    - Column (4): -.20959* (.02860)
  - Disp × Hartz (interaction):
    - Column (1): -.09645* (.00801)
    - Column (2): -.10378* (.00707)
    - Column (3): -.09703* (.01124)
    - Column (4): -.10211* (.01160)
  - Disp × time:
    - Column (3): .00029* (.00008)
    - Column (4): .00031* (.00008)
  - Disp × GDP growth:
    - Column (3): -.00250 (.00315)
    - Column (4): -.00114 (.00322)
  - Average past earnings:
    - Column (3): .99469* (.00060)
    - Column (4): .96612* (.00094)
- R-squared and sample sizes:
  - R squared:
    - Column (1): .0009
    - Column (2): .2610
    - Column (3): .8657
    - Column (4): .7624
  - Individual-months:
    - Columns (1)–(3): 25,660,640
    - Column (4): 19,901,846
  - Individuals:
    - Columns (1)–(3): 280,742
    - Column (4): 240,309
- Interpretation:
  - Raw earnings differential: recently displaced paid markedly less; raw differential almost 32 log points prior to the reforms and increases by a further 10 log points after the reforms in the uncontrolled specification.
  - Controlling for demographics and past earnings reduces magnitudes but the additional loss after reforms remains about 10 log points across specifications.
  - Past earnings are highly autocorrelated: coefficient on past earnings is .995 in one specification; in reported columns .99469* and .96612*.
  - Separate time trend for displaced: positive but small; on average earnings of recently displaced grow by .3 log points a year relative to non-displaced (reported as .00029–.00031 per period interaction).

### Robustness and additional tests (Table 4) — key findings
- Robustness exercises considered:
  1. Include workers who return to part-time or vocational training (All Employment Types).
  2. Average earnings over a three-year period (3-year Average).
  3. Loosen prior employment requirement to 1-year (Prior employment 1-year).
  4. Re-employment after 2 years (Re-employment after 2 years).
  5. Controls for occupation-year (Controlling for occupation-year; 120 occupations).
  6. Controls for sector-year (Controlling for sector-year; nine sectors).
- Selected coefficient estimates (Disp and Disp × Hartz) and fit:
  - Disp:
    - (1) All Employment Types: -.22118* (.03002)
    - (2) 3-year Average: -.14556* (.02679)
    - (3) Prior employment 1-year: -.25924* (.02439)
    - (4) Re-employment after 2 years: -.14274* (.02874)
    - (5) Occupation-year: -.22013** (.02711)
    - (6) Sector-year: -.22583** (.02751)
  - Disp × Hartz:
    - (1) -.10120* (.01215)
    - (2) -.06730* (.01042)
    - (3) -.09358* (.00995)
    - (4) -.07130* (.01179)
    - (5) -.10008** (.01108)
    - (6) -.10167** (.01124)
  - R squared:
    - (1) .8605
    - (2) .8723
    - (3) .8206
    - (4) .8356
    - (5) .8688
    - (6) .8650
  - Individual-months and individuals vary by specification (see table).
- Interpretation of robustness:
  - Including part-time/vocational returns does not affect estimated cost of displacement prior to reforms nor its increase after reforms.
  - Averaging earnings over three years reduces estimated cost of displacement from 22.2 log points (baseline) to 14.5 log points, and estimated impact of the reforms declines from 10 log points to seven log points; recovery from displacement is slow.
  - Loosening previous-work requirement to one year has only marginal impact on estimates.
  - Re-employment after 2 years reduces both the cost of displacement and its increment after Hartz but effects remain important.
  - Controlling for occupation-year or sector-year has only marginal effect on estimates, suggesting switching into lower-paying occupations/sectors is not the main driver.

### Interpreting which elements of the Hartz reforms matter
- Multiple reform components could explain results:
  - Sharp reduction in long-term unemployment benefits and tighter job search/acceptance requirements are consistent with higher hazard rate and larger displacement penalty through reduced reservation wages.
  - Liberalization of temporary agency work could allow employers to offer lower-paying jobs; temporary work accounted for a sizable fraction of new job creation in some post-Hartz sample years (Figure 6 referenced).
- Data limitations:
  - SIAB does not identify temporary jobs or firm identifiers; cannot isolate mass-layoff subsample following Jacobson et al. (1993).
  - Cannot disentangle precisely which reform components drove observed earnings declines.
- Qualitative note:
  - Reform of the employment agency might have improved matching efficiency; combined with curtailed surplus extraction by displaced workers, improved matching could coincide with lower post-unemployment earnings.

### Conclusion — substantive findings and implications
- German unemployment peaked at over 11 percent in early 2000s and fell to currently stand at five percent.
- Main empirical finding:
  - Using a difference-in-difference framework and rich controls, the paper estimates a 10 log point additional reduction in the earnings of displaced workers relative to similar workers who remain employed in the post-reform years.
- Interpretation:
  - Results are consistent with search-theoretic models where lowering unemployment benefits reduces reservation wages and hence post-unemployment earnings.
  - While Hartz reforms succeeded in reducing unemployment, they imposed significant costs on workers experiencing unemployment via reduced benefits and lower subsequent earnings.
- Limitations and future directions:
  - No structural welfare analysis in this paper; evaluating welfare consequences must account for effects on post-unemployment earnings.
  - Extensions: study migration, probability of subsequent displacement, switching occupation/sector, earnings volatility, future earnings growth, and exploit differential impacts across subpopulations to isolate causal effects.
- Additional supportive evidence:
  - An initial exploration suggests workers more affected by the reforms had larger increases in hazard rate of returning to work and larger post-unemployment earnings losses, supporting the Hartz reforms as the driving factor.

*Italic: Source — _wp15162 - 3. Is full-time employed at time t but at a different firm than at tെ13. (IMF PDF chapter/section).*

### References

### _wp15162 - References

### Labor market reforms and German labor market studies
- Arent, Stefan, and Wolfgang Nagl, 2011, Unemployment Benefit and Wages: The Impact of the Labor Market Reform in Germany on (Reservation) Wages, Ifo Working Paper No. 101.  
- Burda, Michael C., and Jennifer Hunt, 2011, What Explains the German Labor Market Miracle in the Great Recession?, NBER Working Paper No.17187.  
- Dlugosz, Stephan, Gesine Stephan, and Ralph A. Wilke, 2013, Fixing the Leak: Unemployment Incidence Before and After a Major Reform of Unemployment Benefits in Germany, German Economic Review, Vol. 15, pp. 329-352.  
- Dustmann, Christian, Bernd Fitzenberger, Uta Schönberg, and Alexandra Spitz-Oener, 2014, From Sick Man of Europe to Economic Superstar: Germany's Resurgent Economy. The Journal of Economic Perspectives, Vol. 28, pp. 167-188.  
- Fahr, Rene’ and Uwe  Sunde, 2009, Did the Hartz Reforms Speed‐Up the Matching Process? A Macro‐Evaluation Using Empirical Matching Functions, German Economic Review, Vol. 10, pp. 284-316.  
- Funk Kirkegaard, Jakob, 2014, Making Labor Market Reforms Work for Everyone: Lessons from Germany, Peterson Institute on International Economics Policy Brief 1-14.  
- Giannelli, Gianna G., Ursula Jaenichen, and Thomas Rothe, 2013, Doing Well in Reforming the Labour Market? Recent Trends in Job Stability and Wages in Germany, IZA Discussion Paper No. 7580.  
- Hertweck, Mattias S., and Oliver Sigrist, 2012, The Aggregate Effects of the Hartz Reforms in Germany, SOEP papers on Multidisciplinary Panel Data Research No. 532.  
- Klinger, Sabine, and Thomas Rothe, 2012, The Impact of Labour Market Reforms and Economic Performance on the Matching of the Short‐term and the Long‐Term Unemployed. Scottish Journal of Political Economy, Vol. 59, pp. 90-114.  
- Krause, Michael U., and Harald Uhlig, 2012, Transitions in the German Labour Market: Structure and Crisis", Journal of Monetary Economics, Vol. 59, pp. 64-79.  
- Krebs, Tom, and Scheffel, Martin, 2013, Macroeconomic Evaluation of Labor Market Reform in Germany. IMF Economic Review, Vol. 61, pp. 664-701.  
- Launov, Andrey, and Klaus Waelde, 2013, Estimating Incentive and Welfare Effects of Non-Stationary Unemployment Benefits, International Economic Review.  

### Job displacement, earnings losses, and long-term costs of unemployment
- Couch, Kenneth. A., and Dana W. Placzek, 2010, Earnings Losses of Displaced Workers Revisited, American Economic Review, Vol. 100, pp. 572-589.  
- Davis, Steven J., and Til Von Wachter, 2011, Recessions and the Costs of Job Loss, Brookings Papers on Economic Activity, Vol. 43, pp. 1-72.  
- Jacobson, Louis, Robert Lalonde, and Daniel Sullivan, 1993, Earnings Loss of Displaced Workers, American Economic Review, Vol. 83, pp. 685-709.  
- Schmieder, Johannes, F., Til von Wachter, and Stefan Bender, 2010, The Long-Term Impact of Job Displacement in Germany during the 1982 Recession on Earnings, Income, and Employment, IAB discussion paper No. 1.  
- Von Wachter, Til, Jae Song, and Joyce Manchester, 2011, Long-Term Earnings Losses due to Mass Layoffs During the 1982 Recession:  An Analysis Using Longitudinal Administrative Data from 1974 to 2008, mimeo, Columbia University.  

### Theoretical frameworks and seminal contributions on unemployment, matching, and insiders-outsiders
- Bentolila, Samuel, and Giuseppe Bertola, 1990, Firing Costs and Labor Demand: How Bad is Eurosclerosis?, Review of Economic Studies, Vol. 57, pp. 381-402.  
- Blanchard, Olivier J., William D. Nordhaus, and Edmund S. Phelps, 1997, The Medium Run, Brookings Papers on Economic Activity, Vol. 1997 No. 2,  pp.89-158.  
- Lindbeck, Assar, and David Snower, 1988, The Insider-Outsider Theory of Employment and  Unemployment, (Cambridge, Mass.: MIT  Press).  
- Ljunqvist, Lars, and Sargent Thomas J. Sargent, 1996, The European Unemployment Dilemma, Journal of Political Economy, Vol. 106, pp. 514-550.  
- Mortensen, Dale T., and Christopher A. Pissarides, 1994. Job Creation and Job Destruction in the Theory of Unemployment, Review of Economic Studies, Vol. 61, pp. 397-415.  
- Mortensen, Dale T., and Christopher. A. Pissarides, 1999, Job Reallocation, Employment Fluctuations, and Unemployment, in John B. Taylor and Michael Woodford (Eds.), Chapter 18 in Handbook of Macroeconomics, (Amsterdam: Elsevier).  

*Source: _wp15162 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2015/_wp15162.pdf_