Labour Economics: The short and long-term career effects of graduating in a recession

The short and long-‐term career effects of gradua4ng in a recession

Daniel Song Kenneth Ho Wu Shaoyi Zhong Xian

WHAT IMPACT DOES GRADUATING DURING A RECESSION HAVE ON OUR TOTAL LIFETIME EARNINGS?

QUESTION:

•  Male college graduates aged 20 and above •  Canadian University students and graduates from 1976-‐1995 – Course of study ( degree type, major, date of graduaRon)

– Career informaRon (annual earnings, province of residence and receipt of unemployment benefits)

•  Individual tax records and firms’ payroll informaRon from 1982-‐1999

Sampling Data

Introduc4on | AlternaRve ExplanaRons | Model | Empirical Findings | Recovery | Conclusion | Singapore

Overview


Difference across cohorts

• Graduate during recession versus those that do not

Difference within cohorts

• Advantaged graduates versus less advantage graduates •  College type •  Program of study •  Length of study

•  Lower returns on investment into higher educaRon

•  Lower quality firms and fewer job opportuniRes •  Poor wage adjustments within firms •  Poor knowledge about workers’ producRvity by potenRal employers

•  Poor iniRal firm placement •  Few opportuniRes for human capital accumulaRon

How does recession affect earnings?


ALTERNATIVE EXPLANATIONS FOR PERSISTENCE OF INITIAL LABOUR MARKET CONDITIONS

AlternaRve ExplanaRons

•  Temporary worsening of labour condiRons leads to longer search for jobs

•  Lower skilled workers may search less intensively aeer receiving lower quality jobs in recessions

ReducRon of Job Search Intensity

•  Employers gradually learn about worker quality, increasing recovery Rme

•  Faster learning about higher skilled workers can explain differenRal speeds of recovery

Rate of Employer Learning

IntroducRon | Alterna4ve Explana4ons | Model | Empirical Findings | Recovery | Conclusion | Singapore

AlternaRve ExplanaRons

• Students graduaRng in a recession spend more Rme looking for a job and less Rme acquiring job-‐specific skills

Human Capital AccumulaRon

•  IniRal labour market condiRons set the starRng wage • Persistence arises if renegoRaRon is imperfect • Permanence arises if renegoRaRon is absent

RenegoRaRon

IntroducRon | Alterna4ve Explana4ons | Model | Empirical Findings | Recovery | Conclusion | Singapore

PROPOSED JOB SEARCH MODEL

•  High and low ability workers –  Higher ability workers receive beger and/or more frequent job offers

•  Mobility costs are posiRvely related to age and tenure •  Aeer an iniRal one-‐period decline in wage offer distribuRon, high-‐skilled workers recover more quickly by moving between jobs due to more transferable skills

•  Smaller search intensity of low-‐skilled workers implies slower mobility to higher-‐wage firms and more accumulaRon of firm-‐specific capital (lock-‐in effect)

Proposed Job Search Model

IntroducRon | AlternaRve ExplanaRons | Model | Empirical Findings | Recovery | Conclusion | Singapore

PredicRons

Larger recessions = more lasRng increases in inequality and mismatch • The larger the iniRal shock, the more likely that age-‐related slowdown in search will occur before the iniRal effect dissipates, especially for lower-‐skilled college graduates

Higher inequality in pre-‐exisRng earnings = greater persistence in inequality due to iniRal shocks • Persistence due to age-‐related costs increases with the dispersion in firms’ average wages


Model

ycrt =α +βeURcr0 +φt +θr +γe + χc +ucrt


Model


Earnings


Model


Unemployment Rate


Earnings

Model


Unemployment Rate

First region of residence

Year of gradua4on

Year of poten4al labour market experience

Calendar year


Earnings

•  Current state of regional labour markets conRnue to affect earnings of more experienced workers

•  Long-‐term effect of first unemployment rate + weighted sum of effects of unemployment rates faced in career

•  Captures average change in earnings from graduaRng in a recession, given regular evoluRon of regional unemployment rate faced aeerwards

Dynamic Effects


•  Need to isolate effect of labour market condiRons at entry, net of subsequent effects on earnings from exposure to a possibly long recession

•  Allow us to disRnguish impact of labour market condiRons at entry as compared to entering the labour market at mid-‐career

Dynamic Effects


Dynamic Model

logwcrt = φt +θr + χc +γe +βe,01UR01 +βe,23UR23 +...+ucrt

UR01 ≡ (URcr0 +URcr11) / 2, UR23 ≡ (URcr2 2 +URcr33) / 2

Effect on earnings in experience year (e) from the unemployment rate at experience year 0-‐1, 2-‐3, 4-‐5


EMPIRICAL RESULTS AND FINDINGS

Average Log Earning Profile

Recession

Recession

Convergence


Strong CorrelaRon Between StarRng Wages and IniRal Unemployment Rate CondiRons

High Employment

Mean

•  Persists Into Higher Experience Years

•  Slowly Fades Over Time


Empirical Table

•  Strong iniAal effect ~2% •  Fade AFer About Five Years ~0.4%


Regressing Log Annual Earnings on Regional Unemployment Rates

Convergence AFer 10 Years

•  Similarity Between NaAonal & Regional

•  Can Exclude Unobserved Cohort Effects

NaAonal

Regional


Controlling for Dynamic Effects Imply that important effect of iniAal unemployment rates is driven by the “first shock”

Dynamic

IniAal

Small Difference


Stronger Effects on Labor Market Entrants

Large Difference

Experienced Entrants


Effects on Employment

Fades Within 3 Years



•  Effects are numerically small •  Temporary unemployment shocks has no persistent effects

on employment or parAcipaAon


•  Losses in employment could depress wages by reducing accumulaRon of labor market experience





Findings: • Loss in experience due to recession is not very large

• Does not explain the persistent effect of iniRal labor market condiRons

Heterogeneity in the Effect Linear regression model to predict log earnings based on college aZended, program of graduaAon and years of study

Short-‐Lived

Persistent


Graduates with lower predicted wages are more adversely affected by higher unemployment rates

• Year 1: 15% lower from 5% increase in UR • 7.5% aeer 10 years Bogom • Year 1: 7.5% lower from 5% increase in UR • 2% aeer 4 years Top


Overcall Cost of Recessions

•  Present Value -‐ 5% Interest Rate

•  Actual (Absolute) Loss is

Greater for Most Advantages Workers

Most Advantaged Workers

Least Advantaged Workers

4 to 5 Times Difference


RECOVERY

Labour Mobility across jobs and industries

Let PVUS = Firm A PVPR = Firm B M = cost of switching job

RHS = Gain LHS = Cost


Channels for recovery through first employers and job mobility


High-skilled workers close the gap faster than lower-skill workers


Sensitivity analysis: Labour market entrants respond strongly to market conditions

Robust specificaRon checks

• NaRonal vs Provincial unemployment variaRon • High unemployment rates reduce job mobility amongst experience workers (Shimer, 2005)

Unemployment increases workers’ propensity to switch industries • Cyclical downgrading toward low-‐wage industries •  Job mobility plays significant role in recovery process


Magnitude of job mobility

40-‐50% of recovery aeer iniRal earnings • Control for cohort’s average firm quality and unemployment rates in cell-‐level regression

• Decomposing the persistent effect of iniRal unemployment rate • Differences and changes in firm quality • Persistent unemployment rates • Other factors


Reduction in firm quality and market conditions explain earnings losses

Blue: Base line model Red: remaining effect of iniRal unemployment rates • 40-‐50% explained by reducRons in firm quality

Purple: addiRon of current unemployment rate • Long term effect of iniRal unemployment faded IntroducRon | AlternaRve ExplanaRons | Model | Empirical Findings | Recovery | Conclusion | Singapore

Discussion

• Hetrogeneous workers search for jobs at beger firms • Recovery slowed due to accumulaRon of specific capital • SecRon 1: Adjustment may differ due to workers’ skill level • Low-‐skilled workers more likely to be affected • Permanent earnings differences • Consistent with neoclassical model of gradual sorRng (Gibbons 2005)

• First job placement highly correlated to worker’s ability • Findings are consistent with nature of catch up changes once workers enter high producRvity firm


Conclusion


Across cohorts

•  IniRal loss : 9% •  Aeer 10 years: 0%

Within cohorts

•  Least advantaged graduates: 8% loss of earnings (cumulated over first 10 years)

•  Middle graduates: 4% loss •  Top graduates: 2% loss and below

•  Labor market entrants are more adversely affected by the recession than employed workers •  Mobility to beger firms main way to recover

•  Excluded female graduates •  Assumes that firm characterisRcs are permanent throughout

•  Regression model does not have external validity – Difference in seqng – Difference in populaRon

LimitaRons


•  Recession in 2008, due to sRcky wages, effects only felt in

2009 as shown through a decline in employment rate and fall in salary.

Singapore

2007 2008 2009 2010 2011p 2007 2008 2009 2010 2011p 2007 2008 2009 2010 2011p

Universities 89.8 87.3 84.7 84.8 86.4 4.7 4.0 5.8 5.9 5.0 94.5 91.3 90.6 90.7 91.4 2,750 2,850 2,700 2,900 3,000

Polytechnics

Fresh Graduates 75.3 68.5 62.8 68.5 67.0 17.7 21.3 25.7 23.0 25.1 93.0 89.8 88.5 91.5 92.1 1,700 1,800 1,700 1,800 1,850

Post-NS Graduates 82.3 76.0 71.1 81.3 80.1 10.4 11.8 16.6 14.1 14.6 92.7 87.8 87.7 95.4 94.7 2,000 2,000 2,000 2,000 2,100Institute of TechnicalEducation (ITE)

Fresh Graduates 71.1 61.9 57.6 57.8 63.5 21.8 26.3 23.5 26.9 20.6 92.9 88.3 81.1 84.7 84.1 1,217 1,300 1,200 1,291 1,300

Post-NS Graduates 79.8 76.4 78.4 78.3 79.7 14.2 14.4 10.0 12.0 10.0 93.9 90.8 88.4 90.4 89.6 1,400 1,600 1,500 1,600 1,600

Employment Rate (%)

Full–time Permanent Part–time/Temporary Total

KEY STATISTICS ON EMPLOYMENT OUTCOME OF GRADUATES FROM INSTITUTIONS OF HIGHER LEARNING, 2007–2011

Institutions

TABLE 9

Median Monthly Gross Starting Salary ($) of Graduates in Full-time Permanent

Employment

2007 2008 2009 2010 2011p

Data extracted from MOM


•  Job mobility declines with age

Singapore

Data extracted (Xing & Yang, 2003)


•  Out of the 117 respondents searching for other employment, 60.68% was due to low salary expectaRons

•  Job mobility is crucial for helping workers get beger wages

Singapore



•  Job mobility is important for recovery yet many remain in current firm therefore the effects experience by Singapore graduates tend to be greater (more persistent losses in earnings)

Singapore



•  Our group feels that since Singapore is an open market, we are extremely suscepRble to recessions.

•  Conversely, Singapore tend to be able to recover quickly and experience strong economic growth, therefore wages pick up quickly as well.

•  We cannot determine the persistent effects on earnings losses due to a recession as there is insufficient data.

Singapore


The short and long-‐term career effects of gradua4ng in a recession

Daniel Song Kenneth Ho Wu Shaoyi Zhong Xian

Labour Economics: The short and long-term career effects of graduating in a recession

Technology