Heterogeneity and the Public Wage Policy

Transcription

1 Heterogeneity and the Public Sector Wage Policy Pedro Gomes Universidad Carlos III Conference in honor of Christopher A. Pissarides June 2015

2 Stylized facts about public sector employment and wages Major components of labour market and gov. budget [OECD: public employment is 18% of total employment, public sector wage bill is more than 50% of government consumption expenditures] The public sector predominantly hires skilled workers [UK: 36 percent of college graduates, 16 percent of workers with lower qualifications]. On average, the public sector pays higher wages. [Katz Krueger (1991), Postel-Vinay and Turon (2007), Christofides and Michael (2013), Castro et al. (2013), Giordano et al. (2011), Dickson, Postel-Vinay and Turon (2014)]

3 Other stylized facts Heterogeneity: Wage compression across education groups: More educated workers have lower premium, less educated workers have higher premium. [Katz Krueger (1991), Postel-Vinay and Turon (2007), Christofides and Michael (2013), Castro et al. (2013), Giordano et al. (2011), Dickson, Postel-Vinay and Turon (2014)] Wage compression within education groups: Bottom quantiles have higher premium, top quantiles have lower or even negative premium [Poterba and Rueben (1994), Postel-Vinay and Turon (2007), Disney and Gosling (1998), Mueller (1998), Christofides and Michael (2013)]

4 Objective Build a quantitative macro model that incorporates these stylized facts and use it to evaluate a reform of public sector wages that strengths the link with private sector, across workers. Why this reform? Implicit wage policy in a frictionless labour market. Frictionless labour market no role for public sector wages. Labour market frictions the market tolerates different wages. High public sector wages induce queues for public sector jobs. Low public sector wages leads to recruitment and retaining problems.

5 Why look at heterogeneity? Most literature considers homogeneous workers Without frictions [Algan et al. (2002), Finn (1998), Ardagna (2007)] With frictions [Quadrini and Trigari (2007), Michaillat (2014), Gomes (2015), Afonso and Gomes (2014)] Wage heterogeneity [Bradley, Postel-Vinay, Turon (2015)] Heterogeneous skills [Domeij and Ljungqvist (2006)] Different public sector wage premiums imply different distortions. Current Euro Area crisis (one-size-fits-all policy might not be suitable).

6 Government wage bill and employment, 2008 Government wage bill Private wage bill and Government employment Private employment Government wage bill (% of private sector wage bill) Portugal Spain Ireland Cyprus Italy Greece Government employment (% of private sector employment)

7 Build an extension of the model of Gomes (2015) Search and matching frictions and worker heterogeneity (education and ability) Given a wage schedule, the government decides how many workers of each type to hire. Calibrate the model to the United Kingdom using LFS microdata. Main results Aligning the distribution of public sector wages with the private sector reduces steady-state unemployment rate by 1.9 percentage points, particularly of the low-ability unskilled. Reduces the distortions in the labour market. Gives the incentive for governments to hire more unskilled workers. Gives the ability for governments to hire more skilled workers. The reform can reduce inequality.

8 Model

9 Setting Population 1 Skilled h Unskilled µ ω h ω h ω µ ω µ High ability h [z h ] Low ability h [z h ] High ability µ [z µ ] Low ability µ [z µ ] Assumption 1: segmented markets Microfoundations

10 Labour market for each type Assumption 2: directed search v p,i t v g,i t 1 η p 1 η g l p,i t+1 = (1 λp,i )lt p,i + mt p,i m p,i t m g,i t l g,i t+1 = (1 λg,i )lt g,i +mt g,i η p η g u p,i t u g,i t Value of working in the two sectors (w p,i t+1, w g,i t+1, λp,i, λ g,i ) Probability of finding a job (v p,i t+1, v g,i t+1 ) Idiosyncratic preference for the public sector (γt i Γ)

11 Household Accumulate capital (K t) Choose consumption (c t) Search of unemployed members (s i t ) l p t, k t l g t Intermediate producers Post vacancies v p,i t [free entry in 4 submarkets] Matched firms rent capital [complement to skills] Nash bargaining for wages x t Final good producer Buys 4 inputs in competitive markets Produces final good Y t = F (x t) Government Exogenous wage schedule Produces services with workers ḡ = g(l g t ) Chooses vacancies to minimize costs Pay unemployment benefit χ b Buy intermediate goods ḡ int Collect taxes (income and lump-sum) Y t = c t + ḡ int + K t+1 (1 δ)k t + i GDP t = Y t + i ωi l g,i t w g,i t j ωi v j,i t κ j,i

12 Decentralised Equilibrium Definition Given a sequence of policies of public wages {wt g,i, i} t=o, unemployment benefits χ b, government services ḡ, intermediate purchases ḡ int and income tax τ and a set of initial conditions {K 0, l p,i 0, l g,i 0, i}; a decentralised equilibrium is a sequence of prices {r t, wt p,i allocations {K t+1, C t, kt, i vt p,i, vt g,i, st, i i} t=o such that:, p x,i t, i} t=o and 1-Representative household satisfies the Euler Equation. 2-Unemployed members of type i choose which sector to search. 3-Matched intermediate goods firms choose optimal capital for each type. 4-Free entry of intermediates goods firms. 5-Private sector wages are the outcome of Nash bargaining. 6-Wholesale representative firm maximizes profits. 7-Government minimizes the cost of producing services. 8-Lump-sum taxes balance the budget. 9-Intermediate goods, final good and capital markets clear.

13 Calibration

14 Calibration Country: United Kingdom Frequency: quarterly Main source: Labour Force Survey, microdata, F (x t ) = ( Ψ((x h t ) ϱ + (x h t ) ϱ ) ς ϱ + (1 Ψ)((xt µ ) ϱ + (x µ ) t ) ϱ ) ς 1 ς ϱ ( g(l g t+1 ) = Φ((ω hz hl g, h t+1 )ϱ + (ω h z h l g,h t+1 )ϱ ) ϱ ς + (1 Φ)((ω µ z µ g, µ lt+1 )ϱ + (ω µ z µ l g,µ ) t+1 )ϱ ) ϱ ς 1 ς

15 Share of educated workers Figure: Share of college graduates in labour force % Mean: 32% 1996q1 2000q1 2004q1 2008q1 Year [ω h = ω h = 0.16 and ω µ = ω µ = 0.34] [Robustness]

16 Government production and services Steady-state level of government services (ḡ) Importance of skill in production (Φ) Figure: Public employment by skill % of total employment q1 2000q1 2004q1 2008q1 Year College graduates Mean: 37.3% Mean: 16.7% Without college

17 Separation rates Figure: Separation rates % Mean: 1.2% Mean: 0.4% 1996q1 2000q1 2004q1 2008q1 Year Private sector Public sector % Mean: 1.7% Mean 0..6% 1996q1 2000q1 2004q1 2008q1 Year Private sector Public sector (a) College degree (b) Without college [λ p,h = 0.012, λ p,µ = 0.017, λ g,h = 0.004, λ p,µ = ]

18 Steady-state public sector wage premium Method: quantile regressions of log net wages Table: Estimation of public sector wage premium Education Percentile R-squared Estimated Premium College educated Obs: Without college degree Obs: Note: quantile regression of log net wages on several control variables and a dummy for public sector. Controls include: sex, industry and occupation dummies, status in previous quarter, tenure, age and its square, marital status, time and region dummies, average hours worked and its square. The sample from 1996 to [ w g, h w p, h w g,h = 1.016, w g, µ = 1.039, w p,h w p, µ = and w g,µ p,µ = 1.071] w [Robustness]

19 Labour market frictions Cost of posting vacancies and matching efficiency Table: Cost per hire and vacancy duration by sector and worker type Cost per hire ( ) Vacancy duration Type of worker Man. Serv. Public Man. Serv. Public Senior Managers - Directors Managers and professionals Administrative, Secretarial and Technical Services (costumer, personal and sales) Manual, craft workers Source: Chartered Institute of Personal Development, Recruitment, retention and turnover survey, 2008 (Survey of 800 organizations: Manufacturing, Services and Public sector). Vacancy duration in weeks. Matching elasticities: estimated by Gomes (2014), using JOLTS data. [κ p,h = 1.35, κ g,h = 0.90, κ p,µ = 0.14, κ g,µ = 0.13] [ζ g,h = 0.73, ζ p,h = 0.56, ζ g,µ = 0.99 and ζ p,µ = 0.98] [η p = 0.4 and η g = 0.15]

20 Flow value of unemployment Unemployment benefits (χ b = 0.21) replacement rate of the low ability unskilled worker is 60 percent of the net wage [Salomaki and Munzi (1999)]. Home production (χ u = 0.37) unemployment rate of unskill workers. Bargaining power of workers (b = 0.35) the overall unemployment rate. Figure: Unemployment rate % of labour force Mean: 7.3% Mean: 6% 1996q1 2000q1 2004q1 2008q1 Year College graduates All workers Without college Note: the flow value of unemployment is 40, and 77 percent of the net wage.

21 College premium The parameter of the private production function Ψ targets a college premium of 40 percent. Regress log net wages on a dummy for college education, on average hours and its square Estimated coefficient: R-squared=0.64 Observations=312070

22 Wage dispersion z h = z µ = 1: Normalization. z µ, z h measure of wage dispersion. Table: Estimation of inter-quantile wage residual Education R-squared Obs percentile residual difference Total Adjusted Adjusted (100%) (80%) (20%) College educated Without college degree Note: regression of the log of net wages on several control variables: sex, industry and occupation dummies, status in previous quarter, tenure, age and its square, marital status, time and region dummies, average hours worked and its square. The sample from 1996 to The fourth column reports the percentile difference of wage residuals. Is all wage dispersion due to unobserved heterogeneity? Benchmark: 80% of difference is due to unobserved heterogeneity. stness : 100% and 20%

23 Remaining parameters Technology parameters Elasticity of output w.r.t capital (α = 0.34) labour share of 61.8%. Elasticity of substitution 1 between skilled and unskilled input. [Robustness] 2 between high and low ability. [Robustness] Standard parameters Discount factor (β = 0.99). Risk aversion (σ = 2). Depreciation rate (δ = 0.02). Income tax ( τ = 0.2). Purchase of intermediate inputs (ḡ int = 0.034) gov. consumption is 20 % of GDP

24 Distribution of public sector preferences Benchmark: Γ has a uniform distribution [ν 1, ν 2 ] Google Trends: Indexes of of keyword searches Index of Jobs Compound index for the public sector. Other studies Keyword Relative importance in index nhs jobs 46% council jobs 32% jobs in nhs 5% gov jobs 4% public jobs 4% direct gov jobs 2% government jobs 2% army jobs 2% local government jobs 1% raf jobs 1%

25 Distribution of public sector preferences Figure: Google indexes Index w1 2006w1 2008w1 2010w1 2012w1 2014w1 Year Jobs Government Jobs Fraction Mean: 0.14 Std. Deviation: Correl with public private wage ratio: q1 2006q1 2008q1 2010q1 2012q1 2014q1 Year (a) Original indexes (b) Search in public sector Benchmark: s = 0.14 and ν 2 ν 1 = 2 w Robustness:

26 Results

27 S.S. Effects of skilled public sector wages Unemployment Rate Public sector employment Public employment: share of high ability 10 All workers 40 Skilled workers 65 Unskilled workers 8 % High ability skilled public wages relative to baseline % All workers Skilled workers Unskilled workers High ability skilled public wages relative to baseline % Skilled workers Unskilled workers High ability skilled public wages relative to baseline Private sector wages (relative to baseline) Skilled (high ability) Skilled (low ability) Unskilled (high ability) Unskilled (low ability) 24 Share of unemployed searching in public sector Government spending (% GDP) High ability skilled public wages relative to baseline % 16 8 Skilled (high ability) Skilled (low ability) Unskilled (high ability) Unskilled (low ability) High ability skilled public wages relative to baseline % of GDP Wage bill + Recruitment costs + Unemployment benefits High ability skilled public wages relative to baseline

28 S.S. Effects of unskilled public wages Unemployment Rate Public sector employment Public employment: share of high ability 10 All workers 40 Skilled workers 65 Unskilled workers 8 % 6 4 % % Low ability unskilled public wages relative to baseline All workers Skilled workers Unskilled workers Low ability unskilled public wages relative to baseline 50 Skilled workers Unskilled workers Low ability unskilled public wages relative to baseline Private sector wages (relative to baseline) Skilled (high ability) Skilled (low ability) Unskilled (high ability) Unskilled (low ability) 24 Share of unemployed searching in public sector Government spending (% GDP) Low ability skilled public wages relative to baseline % 16 8 Skilled (high ability) Skilled (low ability) Unskilled (high ability) Unskilled (low ability) % of GDP Low ability unskilled public wages relative to baseline Wage bill + Recruitment costs + Unemployment benefits Low ability unskilled public wages relative to baseline

29 Welfare effects Figure: Welfare effects of public sector wages adjustments 5 4 Welfare change from baseline Progressive Regressive 5 4 Welfare change from baseline Progressive Regressive 3 3 % % Public wages relative to baseline (a) Lump-sum taxes Public wages relative to baseline (b) Distortionary taxes

30 Steady-state effects of the reform Lump-Sum Taxes Distortionary Taxes Public-private wage premium Baseline 0% 0% Variables Unemployment rate Skilled High-ability Low-ability Unskilled High-ability Low-ability Consumption % +3.79% Welfare Gains % 3.11% Note: model simulations under the baseline calibration. given in percent of GDP.

31 Steady-state effects of the reform Lump-Sum Taxes Distortionary Taxes Public-private wage premium Baseline 0% 0% Variables Public employment Skilled High-ability Low-ability Unskilled High-ability Low-ability Government Wage bill recruitment costs unemployment benefits Income taxes Implied public [private] sector wage change Skilled (high-ability) - 0.5% [1.1%] 1.7% [3.4%] Skilled (low-ability) % [-1.4%] -3.1% [0.7%] Unskilled (high-ability) % [0.4%] -1.5% [2.2%] Unskilled (low-ability) % [-1.6%] -8.1% [-1.5%] Note: model simulations under Pedrothe Gomes, baseline UC3M calibration. Heterogeneity andgiven the Public inwage percent Policy of GDP.

32 Robustness Elasticity of substitution between abilities [both sectors, only public sector] (ρ = 0.8,ρ = 0.3) Elasticity of substitution between skills [both sectors, only public sector] (ς = 0.4,ς = 0.4) Search in public sector (s = 0.07,s = 0.21) Dispersion in preferences for public sector (ν 2 ν 1 = 3 w, ν 2 ν 1 = 0.2 w) Share of skilled in economy (ω h = ω h = , ω h = ω h = 0.20) Heterogeneity in ability ( w p, h / w p,h = 1.09, w p, µ / w p,µ = 1.08, w p, h / w p,h = 1.46, w p, µ / w p,µ = 1.42) Lower baseline premium ( w g, h / w p, h = 0.986, w g,h / w p,h = 1.009, w g, µ / w p, µ = 1.007, w g,µ / w p,µ = 1.041) No dispersion in premium ( w g, h / w p, h = 1.03, w g,h / w p,h = 1.03, w g, µ / w p, µ = 1.03, w g,µ / w p,µ = 1.03)

33 Robustness Lump-sum taxes Distortionary taxes Scenario Unemployment Consumption Welfare Unemployment Consumption Welfare rate rate Elasticity of substitution between abilities [both sectors] ς = pp 1.9% 1.4% -2.8pp 3.8% 3.1% ς = pp 1.9% 1.5% -2.6pp 3.7% 3.1% Elasticity of substitution between abilities [only public sector] ς g = pp 1.9% 1.4% -2.8pp 3.7% 3.0% ς g = pp 1.9% 1.5% -2.6pp 3.8% 3.1% Elasticity of substitution between skills [both sectors] ϱ = pp 2.3% 1.8% -3.2pp 4.5% 3.7% ϱ = pp 1.8% 1.3% -2.4pp 3.5% 2.9% Elasticity of substitution between skills [only public sector] ϱ g = pp 1.9% 1.5% -2.7pp 3.7% 3.0% ϱ g = pp 1.9% 1.5% -2.7pp 3.8% 3.1% Search in the public sector s = pp 1.3% 1.0% -2.0pp 3.0% 2.5% s = pp 2.1% 1.6% -2.9pp 4.0% 3.3% Note: model simulations under alternative calibrations. For each scenario the model was recalibrated according to Section 3. The table reports the steady-state change of implementing a zero public sector wage premium for all workers relative to baseline of: unemployment rate (percentage points), consumption (percent) and welfare (percent of consumption equivalent variation).

34 Robustness Lump-sum taxes Distortionary taxes Scenario Unemployment Consumption Welfare Unemployment Consumption Welfare rate rate Dispersion in preferences for public sector ν 2 ν 1 = 3 w -2.0pp 2.0% 1.5% -2.8pp 3.9% 3.2% ν 2 ν 1 = 0.2 w -1.4pp 1.5% 1.1% -2.2pp 3.2% 2.7% Share of skilled workers ω h = ω h = pp 2.2% 1.6% -3.3pp 4.4% 3.6% ω h = ω h = pp 1.7% 1.3% -2.2pp 3.2% 2.7% Heterogeneity in ability w p,ī = w p,i -2.2pp 2.6% 2.0% -3.2pp 5.0% 4.1% w p,ī = w p,i -1.9pp 1.8% 1.4% -2.6pp 3.5% 2.9% Lower average premium Baseline-3% -0.7pp 0.8% 0.6% -1.0pp 1.4% 1.1% No dispersion in premium Premium=3% -1.3pp 1.3% 0.9% -1.9pp 2.7% 2.2% Note: model simulations under alternative calibrations. For each scenario the model was recalibrated according to Section 3. The table reports the steady-state change of implementing a zero public sector wage premium for all workers relative to baseline of: unemployment rate (percentage points), consumption (percent) and welfare (percent of consumption equivalent variation).

35 How about inequality? Why is the public sector wage distribution so distorted? Why is it hard to defend cutting the lowest public sector wages? Key insight: The government has a redistributive role, but not all instruments have to be redistributive (Mirlees report). If the government wants to fight inequality, it should use the income tax system, or potentially, other regulatory policy (minimum wage). Using public sector wages does not solve the problem, and creates inefficiencies in the labour market. Ω i t = lt p,i Wt p,i + lt g,i Wt g,i + utu i t, i i. (1)

36 Effects of reform on inequality Alternative tax scenarios Public-private wage premium Baseline (1) (2) (3) Variables Taxation Capital tax rate Income tax rate Skilled (high-ability) Skilled (low-ability) Unskilled (high-ability) Unskilled (low-ability) Unemployment rate Consumption % +2.9% +4.4% Welfare Gains - 3.1% 2.2% 3.3% Labour market value of type Skilled (high-ability) % +4.6% +3.1% Skilled (low-ability) % +0.0% -1.1% Unskilled (high-ability) % +3.2% +1.0% Unskilled (low-ability) % +1.0% +10.3%

37 Conclusion I propose a reform of public sector wage: Aligning the distribution of public sector wages with the private sector reduces steady-state unemployment rate by 1.9 percentage points, particularly of the low-ability unskilled. Reduces the distortions in the labour market. Gives the incentive for governments to hire more unskilled workers. Gives the ability for governments to hire more skilled workers. With such substantial gains, why don t governments implement it? Worry about inequality. Political economy issues.

38 Additional results

39 Adverse selection with labour market friction Guerrieri, Shimer and Wright (2010), use hours worked as the self-selection mechanism (disutility of work). Michelacci and Suarez (2006), use the wage setting (wage posting attracts the low type and wage bargaining attract the high type) Fernandez-Blanco and Gomes (2013) use capital as the self-selection mechanism. Go Back

40 Households Consumption is pooled between the members of the household (Merz, 1995). Preferences are: E 0 t=0 β t [ c1 σ t 1 σ + χu u t ], Budget constraint: c t + K t+1 = (1 δ)k t + (1 τ t) r tk t + j i ω i w j,i t l j,i t + i ω i χ g u t + Π t, Optimality conditions: u c (c t ) = βe t [u c (c t+1 )(1 δ + r t+1 (1 τ t+1 )], (2) χ u t = χu u c (c t ) (3)

41 Households members: value functions W j,i t = (1 τ t )w j,i t + E t β t,t+1 [(1 λ j,i )W j,i t+1 + λj,i U i t+1], i, j, (4) U j,i t = χ u t + χ b + E t β t,t+1 [ft j,i W j,i j,i t+1 + (1 ft )Ut+1], i i, j, (5) Unemployed choose which sector to search U p,i t = U g,i t + γ i t, i. (6) γ i t: random variable with cumulative distribution Γ (idiosyncratic preference for the public sector). Shortcut: without it search is too responsive.

42 Households members γt i, : the cut-off point of the distribution given by; γ i, t = ft p,i E t β t,t+1 [W p,i t+1 Ui t+1] ft g,i E t β t,t+1 [W g,i t+1 Ui t+1], i. (7) The fraction on unemployed searching in the public sector s i t is: st i ug,i t ut i = 1 Γ(γ i, t ), i, (8) The ex-ante value of unemployment: Go Back Ut+1 i = st+1u i g,i t+1 + (1 si t+1)u p,i t+1, i, (9)

43 Intermediate good producers Large continuum of firms. Produce one of four types of intermediate goods x i t that is sold at price p x,i t. Pay a cost κ p,i to open vacancies v p,i t, in a given sub-market i. If the vacancy is filled, the firm is matched to a type-i worker, chooses capital, and produces f (a, z i, k i t) = az i (k i t) α. f (a, z i, kt) i is increasing and concave in all its arguments with a positive cross partial derivative of capital and skill (there is an optimal capital for each worker). Surplus is shared: wages (wt p,i ) are determined by Nash Bargaining.

44 Intermediate good producers V i t = κ p,i + E t β t,t+1 [qt p,i Jt+1 i + (1 qt p,i )Vt+1], i i. (10) J i t = max k i t [pt x,i f i (a t, z i, k i t) w p,i t r t k i t +E t β t,t+1 [(1 λ p,i )J i t+1], i. (11) Optimal capital: pt x,i fk i (a t, z i, kt i ) = r t, i. (12)

45 Intermediate good producers V i t = κ p,i + E t β t,t+1 [qt p,i Jt+1 i + (1 qt p,i )Vt+1], i i. (13) Jt i = pt x,i f i (a t, z i, kt i ) wt p,i r t kt i + E t β t,t+1 [(1 λ p,i )Jt+1, i i. (14) Optimal capital: pt x,i fk i (a t, z i, kt i ) = r t, i. (15)

46 Intermediate good producers Private sector vacancies satisfy the free entry condition: V t = 0, i. Nash wage bargaining between workers and firms: (W p,i t Ut) i = b(1 τ t l ) 1 bτt l (Wt p,i Ut i + Jt), i i. (16) Go Back

47 Wholesale firm Buys the 4 intermediate goods to produce a wholesale good max[f (x t ) x t i p x,i t x i t], (17) F x i = p x,i t, i. (18) F (x t ) = Ψ((x h t ) ϱ + (x h t ) ϱ ) ς ϱ + (1 Ψ)((xt µ ) ϱ + (x µ t ) ϱ ) ς ϱ }{{}}{{} Skilled Unskilled 1 ς Go Back

48 Government Needs to produce a minimum level of services ḡ. Wages w g,i are the exogenous policy variables. Chooses the vacancies of each type of worker at time t to minimize the total cost of providing the government services. min v g,i t i ω i κ i v g,i t + β t,t+1 [ i s.t. ḡ = g(l g t+1 ) l g,i t+1 = (1 λj,i )lt g,i Two opposite effects of public wages Wage bill effect. Recruitment effect. ω i w g,i t+1 l g,i t+1 + qt g,i vt g,i, i. ]

49 Government First-order conditions: ω i κ g,i q g,i t + E t β t,t+1 [ω i w g,i t+1 ] = ζ te t g i,t+1, i (19) Spending: wage bill, recruitment costs, unemployment benefits and exogenous purchases of goods. Revenue: income taxes (distort the wage bargaining and capital accumulation) and lump-sum taxes (balance the budget). ( ) τ t ω i l j,i t w j,i t + r tk t + T t j i }{{} Income taxes + Lump-sum = i ω i l g,i t w g,i t } {{ } Wage bill + i ω i v g,i t κ g,i }{{} Recruitment costs + i χ b ω i ut i + ḡ int } {{ } U Benefits, }{{} Int. goods (20) Go Back

50 Market Clearing Intermediate goods x i t = ω i l p,i t f i (a t, z i, k i t), i, (21) Final good Y t = F (x t ) = c t + ḡ int + K t+1 (1 δ)k t + i j ω i v j,i t κ j,i. (22) Capital K t = i ω i k i tl p,i t. (23)