Mismatch Unemployment in the U.K.

Mismatch Unemployment in the U.K. Christina Patterson MIT Ayşegül Şahin Federal Reserve Bank of New York Giorgio Topa Federal Reserve Bank of New York, and IZA Gianluca Violante New York University, CEPR, and NBER Bank of England, June 2013 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 1 /41

Unemployment rate in the U.K. Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 2 /41

Unemployment rate in the U.K. 9 8 Unemployment Rate 7 6 5 4 2000 2002 2004 2006 2008 2010 2012 increased to above 8% during the recession remained persistently high Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 2 /41

The mismatch hypothesis Growing misalignment between distributions of job seekers and job openings across sectors (locations, industries, occupations) Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 3 /41

The mismatch hypothesis Growing misalignment between distributions of job seekers and job openings across sectors (locations, industries, occupations) A priori, plausible: Outward shift in the Beveridge curve Job losses concentrated in certain sectors Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 3 /41

Job destruction rates 0.012 Corporate Managers Science and Technology Professionals Skilled Construction Personal Service Process, Plant and Machine Operatives 0.01 Employment to Unemployment Rate 0.008 0.006 0.004 0.002 0 2007 2008 2009 2010 2011 2012 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 4 /41

The mismatch hypothesis Growing misalignment between distributions of job seekers and job openings across sectors (locations, industries, occupations) A priori, plausible: Outward shift in the Beveridge curve Job losses concentrated in certain sectors Questions: How much of the rise in unemployment is due to mismatch? Along which dimensions has mismatch worsened? Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 5 /41

Methodology analogous to misallocation literature Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 6 /41

Methodology analogous to misallocation literature Economy with I frictional labor markets indexed by {φ it,z it,δ it } {v it }: observed allocation of vacancies {u it }: observed allocation of unemployed {u it }: benchmark allocation under free mobility across markets (constrained first-best) Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 6 /41

Methodology analogous to misallocation literature Economy with I frictional labor markets indexed by {φ it,z it,δ it } {v it }: observed allocation of vacancies {u it }: observed allocation of unemployed {u it }: benchmark allocation under free mobility across markets (constrained first-best) Discrepancy between {u it } and {u it } lower aggregate job-finding rate additional (mismatch) unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 6 /41

Environment I distinct frictional labor markets New production opportunities v it arise exogenously Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 7 /41

Environment I distinct frictional labor markets Infinite cost-elasticity of vacancy creation Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 8 /41

Environment I distinct frictional labor markets Infinite cost-elasticity of vacancy creation CRS matching function: Φ t φ it m(u it,v it ) Worker-firm match subject to productivity shocks (Z t,z it ) Measure one of infinitely-lived agents with linear utility Agents can be employed (e) or unemployed (u) Unemployed: search directed toward one market only Employed: no OJS, exogenous separation shocks ( t,δ it ) Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 8 /41

Planner s allocation rule 1. Shocks {φ it,z it,δ it } are i.i.d. across sectors, orthogonal to each other, and follow positive martingales 2. No impediment to mobility of labor across markets The planner s allocation rule requires equalizing (across sectors): z it 1 β(1 t )(1 δ it ) φ itm uit ( vit u it ) productive and matching efficiency-weighted v/u ratios Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 9 /41

Planner s allocation rule 1. Shocks {φ it,z it,δ it } are i.i.d. across sectors, orthogonal to each other, and follow positive martingales 2. No impediment to mobility of labor across markets The planner s allocation rule requires equalizing (across sectors): z it 1 β(1 t )(1 δ it ) φ itm uit ( vit u it ) productive and matching efficiency-weighted v/u ratios Assume: m(u it,v it ) = v α it u1 α it Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 9 /41

Mismatch Index (special case: no heterogeneity) Observed aggregate hires: h t = Φ t I i=1 Planner s aggregate hires: h t = Φ t v α t u 1 α t v α it u1 α it Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 10 /41

Mismatch Index (special case: no heterogeneity) Observed aggregate hires: h t = Φ t I i=1 v α it u1 α it Planner s aggregate hires: h t = Φ t v α t u 1 α t Mismatch index: M t h t h t h t = 1 I ( vit v i=1 t u t ) α ( ) 1 α uit [0,1] measures the fraction of hires lost because of misallocation Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 10 /41

Mismatch Index (special case: no heterogeneity) Observed aggregate hires: h t = Φ t I i=1 v α it u1 α it Planner s aggregate hires: h t = Φ t v α t u 1 α t Mismatch index: M t h t h t h t = 1 I ( vit v i=1 t u t ) α ( ) 1 α uit [0,1] measures the fraction of hires lost because of misallocation Mismatch shifts the aggregate matching function: h t = (1 M t ) h t = (1 M t ) Φ t v α t u 1 α t Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 10 /41

Add heterogeneity inφ it M h φt h t h t h t = 1 I i=1 ( )( ) α ( ) 1 α φit vit uit φ t v t u t where φ t = [ I i=1 φ 1 α it ( vit v t ) ] α Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 11 /41

Add heterogeneity inφ it M h φt h t h t h t = 1 I i=1 ( )( ) α ( ) 1 α φit vit uit φ t v t u t where φ t = [ I i=1 φ 1 α it ( vit v t ) ] α With heterogenous (φ it,z it,δ it ) M h xt Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 11 /41

Counterfactual unemployment Observed unemployment: u t+1 = u t +s t (1 u t ) f t u t Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 12 /41

Counterfactual unemployment Observed unemployment: u t+1 = u t +s t (1 u t ) f t u t Aggregate job finding rate: 1. observed: f t = (1 M t ) Φ t ( ) α v t u t 2. no mismatch: f t = Φ t ( v t u t ) α = ft 1 1 M t }{{} Direct Effect ( ut u t ) α }{{} Feedback through u Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 12 /41

Counterfactual unemployment Observed unemployment: u t+1 = u t +s t (1 u t ) f t u t Aggregate job finding rate: 1. observed: f t = (1 M t ) Φ t ( ) α v t u t 2. no mismatch: f t = Φ t ( v t u t ) α = ft 1 1 M t }{{} Direct Effect ( ut u t ) α }{{} Feedback through u Counterfactual unemployment in absence of mismatch: u t+1 = u t +s t (1 u t) f t u t Mismatch unemployment: u t u t Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 12 /41

Data used: July 2006 - June 2012 Unemployed {u it }: Jobseeker s Allowance Claimant Counts Vacancies {v it }: Jobcentre Plus Vacancy Statistics Sought occupation Geographic location (Travel To Work Area) Productivity {z it }: Annual Survey of Hours and Earnings (ASHE) EU rates {δ it }: Labor Force Survey (LFS) Matching function parameters α,{φ it }: estimated Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 13 /41

Summary statistics Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 14 /41

Summary statistics No. Sectors Vacancies Claims 2-digit occupation 24 7712 23173 3-digit occupation 76 1510 6471 Travel To Work Areas 215 619 2288 TTWA x 2-digit 1059 118 314 Region 11 26920 113725 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 14 /41

Unemployment: Claimant Count vs. LFS Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 15 /41

Unemployment: Claimant Count vs. LFS Unemployed (Thousands) 2700 JobSeeker Allowance Claimant Count ONS Unemployment Count 2500 2300 2100 1900 1700 1500 1300 1100 900 700 500 2007 2008 2009 2010 2011 2012 Correlation = 0.98 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 15 /41

Vacancies: Jobcentre Plus vs. ONS vacancy survey Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 16 /41

Vacancies: Jobcentre Plus vs. ONS vacancy survey 700 600 JobCentre Plus Vacancy Count ONS Data Vacancies (Thousands) 500 400 300 200 100 2007 2008 2009 2010 2011 2012 Correlation = 0.92 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 16 /41

Claimant Count vs. LFS Age Labor Force Survey Claimant Count Pre-Recession Post-Recession Pre-Recession Post-Recession 16-24 0.42 0.39 0.42 0.35 25-49 0.43 0.45 0.52 0.57 50+ 0.14 0.15 0.07 0.08 Gender Male 0.57 0.59 0.73 0.71 Female 0.43 0.41 0.27 0.29 Duration under 6 months 0.60 0.52 0.47 0.54 6-12 months 0.16 0.19 0.21 0.21 12-24 months 0.13 0.16 0.16 0.14 24+ months 0.11 0.13 0.17 0.12 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 18 /41

Comparisons Claimants: more prime-age males relative to LFS. Interesting pattern re unemployment duration: fewer ST unemployed than in LFS, pre-recession; duration distribution matches LFS, post-recession. Jobcentre Plus vacancies: more concentrated in banking, finance and insurance. Under-represent manufacturing; distribution, hotels and restaurants; transport and communications. Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 19 /41

Jobcentre Plus vacancies vs. ONS Industry Vacancy Survey JobCentre Plus Pre-Rec Post-Rec Pre-Rec Post-Rec Energy and Water 0.01 0.01 0.01 0.01 Manufacturing 0.09 0.07 0.03 0.03 Construction 0.04 0.03 0.03 0.03 Distribution, Hotels & Restaurants 0.28 0.28 0.17 0.13 Transport and Communications 0.11 0.10 0.04 0.03 Banking, Finance & Insurance 0.23 0.21 0.57 0.53 Public Admin., Education & Health 0.21 0.24 0.12 0.20 Other Services 0.04 0.04 0.04 0.05 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 20 /41

Change in unemployment shares across occupations Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 21 /41

Change in unemployment shares across occupations 0.08 Corporate Managers Science and Technology Professionals Skilled Construction Customer Service Process, Plant and Machine Operatives 0.06 0.04 0.02 0 2007 2008 2009 2010 2011 2012 Skilled Construction, Corporate Managers particularly hit Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 21 /41

Change in vacancy shares across occupations Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 22 /41

Change in vacancy shares across occupations 0.1 Corporate Managers Science and Technology Professionals Skilled Construction Customer Service Process, Plant and Machine Operatives 0.08 0.06 0.04 0.02 0 2007 2008 2009 2010 2011 2012 Sharp decline in Construction, Customer Service, Manufacturing Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 22 /41

Correlation between(u, v) shares across occupations Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 23 /41

Correlation between(u, v) shares across occupations 1 ρ x 0.8 0.6 0.4 0.2 0 2007 2008 2009 2010 2011 2012 A decline in the correlation is a sign of worsening mismatch Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 23 /41

Unemployment outflow data Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 24 /41

Unemployment outflow data 0.5 0.45 Vacancy Outflow Claims Outflow Average 0.4 Measures of the Hires Rate 0.35 0.3 0.25 0.2 0.15 0.1 2007 2008 2009 2010 2011 2012 We use the average of the two series to estimate α and {φ i } Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 24 /41

Matching function estimation At the aggregate level: ln ( hit u it ) = const+γ QTT t +αln ( vit u t ) +ǫ t Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 25 /41

Matching function estimation At the aggregate level: ln ( hit u it ) = const+γ QTT t +αln ( vit u t ) +ǫ t At the sectoral level: ln ( hit u it ) = γ QTT t +χ t 8.03 lnφ pre i +χ t>8.03 lnφ post i +αln ( vit u it ) +ǫ t Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 25 /41

Estimation results Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 26 /41

Estimation results α Sample Size Aggregate φ i Fixed φ i Varying 0.559 72 (0.059) 0.472 1728 (0.006) 0.463 1728 (0.006) Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 26 /41

Estimation results α Sample Size Aggregate φ i Fixed φ i Varying 0.559 72 (0.059) 0.472 1728 (0.006) 0.463 1728 (0.006) Distribution of estimated φ i varies between 0.43 and 0.67. Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 26 /41

Occupational mismatch (2-digit) 0.15 M x 1.5 u u u u x Mismatch Index 0.1 0.05 Mismatch Unemployment 1 0.5 0 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 About 6 pct of hires lost in the recession bc of higher mismatch About 0.5 pct points increase in mismatch unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 27 /41

Occupational mismatch (3-digit) 0.2 M x 2.5 u u u u x 2 Mismatch Index 0.15 0.1 Mismatch Unemployment 1.5 1 0.5 0.05 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 About 7 pct of hires lost in the recession bc of higher mismatch About 0.75 pct points increase in mismatch unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 28 /41

Geographic mismatch (TTWA s) 0.15 M z 1.5 u u u u z Mismatch Index 0.1 0.05 Mismatch Unemployment 1 0.5 0 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 Negligible increase in mismatch unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 29 /41

Occupation X geography mismatch (2-digit x TTWA) 0.25 M 2-Digit M 2-Digit X TTWA 2.5 u u 2-Digit u u 2-Digit x TTWA 0.2 2 Mismatch Index 0.15 Mismatch Unemployment 1.5 1 0.1 0.5 0.05 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 About 6 pct of hires lost in the recession bc of higher mismatch About 0.6 pct points increase in mismatch unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 30 /41

Routine-Cognitive Occupations Notion of job polarization: more jobs in highest/lowest skill (non-routine) occupations; fewer opportunities in middle skill (routine) occupations. Some examples: (C, NR): Managers; professionals; culture, media, sports; (C, R): Office and administrative support occupations; (M, R): Skilled trades (metal, construction, printing,...); (M, NR): Sales and customer service occupations. Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 31 /41

Routine-Cognitive Occupations 0.15 M RC 1.5 u u RC u u Mismatch Index 0.1 0.05 Mismatch Unemployment 1 0.5 0 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 About 6 pct of hires lost in the recession bc of higher mismatch About 0.4 pct points increase in mismatch unemployment Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 32 /41

Summary of results Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 33 /41

Summary of results (u u ) (u u )/ u 2-digit occupation 0.53 ppts 18.9% 3-digit occupation 0.75 ppts 26.9% Travel To Work Areas 0.03 ppts 1.2% TTWA x 2-digit 0.59 ppts 21.2% Region 0.07 ppts 2.6% Routine-Cognitive 0.36 ppts 12.8% Contribution of mismatch unemployment highest for 3-digit occupations and occupations X geography Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 33 /41

Endogenous vacancy creation Equilibrium: 1. Free entry vacancy creation cost: K i (v it ) = κ ε it v1+ε it 1+ε 2. Hosios condition 3. Bargaining solution: Shaked-Sutton (1984), Acemoglu (1996) ( ) 1 α (κ it v it ) ε uit Z t z it = Φ t φ it α v it 1 β(1 t )(1 δ it ) Given ε, back out {κ it } which replicates data on {v it } Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 34 /41

Endogenous vacancy creation Equilibrium free-entry condition: ( ) 1 α (κ it v it ) ε uit Z t z it = Φ t φ it α v it 1 β(1 t )(1 δ it ) Planner s optimal vacancy creation: ( ) u (κ it vit) ε 1 α = Φ t φ it Z t z it it vit α 1 β(1 t )(1 δ it ) Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 35 /41

Endogenous vacancy creation Equilibrium free-entry condition: ( ) 1 α (κ it v it ) ε uit Z t z it = Φ t φ it α v it 1 β(1 t )(1 δ it ) Planner s optimal vacancy creation: ( ) u (κ it vit) ε 1 α = Φ t φ it Z t z it it vit α 1 β(1 t )(1 δ it ) Comparison: v it v it = ( ) 1 α u 1 α+ǫ it u it Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 35 /41

Endogenous vacancy creation Planner vs. observed job finding rate f t = f t 1 1 M t }{{} Direct Effect ( ut u t ) α }{{} Feedback through u ( v t v t ) α }{{} Feedback through v More misallocation fewer vacancies more mismatch unempl. Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 36 /41

Mismatch u with endogenous vacancies Total Vacancies (Thousands) 600 550 500 450 400 350 300 250 200 v v (ǫ = 0.5) v (ǫ = 1) v (ǫ = 2) Mismatch Unemployment Rate (ppts) 3 2.5 2 1.5 1 0.5 u u using v u u using v (ǫ = 0.5) u u using v (ǫ = 1) u u using v (ǫ = 2) 150 100 2006 2007 2008 2009 2010 2011 2012 2013 0 2006 2007 2008 2009 2010 2011 2012 2013 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 37 /41

Mismatch u with endogenous vacancies Total Vacancies (Thousands) 600 550 500 450 400 350 300 250 200 v v (ǫ = 0.5) v (ǫ = 1) v (ǫ = 2) Mismatch Unemployment Rate (ppts) 3 2.5 2 1.5 1 0.5 u u using v u u using v (ǫ = 0.5) u u using v (ǫ = 1) u u using v (ǫ = 2) 150 100 2006 2007 2008 2009 2010 2011 2012 2013 0 2006 2007 2008 2009 2010 2011 2012 2013 Quantitatively, the value of ε is critical With ε = 0.5, increase in mismatch u is twice as large as baseline Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 37 /41

Endogenous vacancies: sensitivity Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 38 /41

Endogenous vacancies: sensitivity 2-digit occupation (u u ) (u u )/ u Baseline 0.53 ppts 18.9% ǫ = 0.5 1.16 ppts 41.2% ǫ = 1 0.88 ppts 31.5% ǫ = 2 0.70 ppts 25.1% Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 38 /41

Comparison to the U.S. 0.2 M U.K. M U.S. 2.5 u u U.K. u u U.S. 2 Mismatch Index 0.15 0.1 Mismatch Unemployment 1.5 1 0.5 0.05 2007 2008 2009 2010 2011 2012 0 2007 2008 2009 2010 2011 2012 More gradual increase and decline of mismatch in the U.S. Mismatch has been increasing again in the U.K. since 2011 Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 40 /41

Conclusion and future work Contribution of mismatch to unemployment in the U.K. Mismatch explains about 1/3 of rise in unemployment Unlike the U.S., mismatch is on the rise again Direct measurement of skill mismatch: Distance in terms of skill content between {v it } and {u it } Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 41 /41

Conclusion and future work Contribution of mismatch to unemployment in the U.K. Mismatch explains about 1/3 of rise in unemployment Unlike the U.S., mismatch is on the rise again Direct measurement of skill mismatch: Distance in terms of skill content between {v it } and {u it } Structural equilibrium models Sources of mismatch? Skill specificity, wage rigidity, policy,... Patterson-Şahin-Topa-Violante, Mismatch Unemployment in the U.K. p. 41 /41