Optimal Welfare-to-Work Programs
Nicola Pavoni
(University College London, and IFS)
Gianluca Violante
(New York University, and CEPR)
Introduction
• Government expenditures on labor market policies across OECD
countries amount to 3% of GDP:
◮ Examples of policies: unemployment insurance, social
assistance, job-search monitoring, training, wage subsidies
Introduction
• Government expenditures on labor market policies across OECD
countries amount to 3% of GDP:
◮ Examples of policies: unemployment insurance, social
assistance, job-search monitoring, training, wage subsidies
• Most governments use a mix of policy instruments
• A Welfare-to-Work (WTW) program is a government expenditure
What do we do?
• We study WTW programs from a normative perspective
• An optimal Welfare-to-Work (WTW) program is a program that
maximizes the unemployed agent ex-ante utility, for a given level of government expenditures
• We want to characterize:
◮ optimal sequence of policies
◮ optimal level and time-path of unemployment benefits
What do we do?
• We study WTW programs from a normative perspective
• An optimal Welfare-to-Work (WTW) program is a program that
maximizes the unemployed agent ex-ante utility, for a given level of government expenditures
• We want to characterize:
◮ optimal sequence of policies
◮ optimal level and time-path of unemployment benefits
How do we do it?
• Dynamic principal-agent (government-worker) relationship with
moral-hazard (unobservable worker effort), following Shavell and
Weiss (1979), Hopenhayn and Nicolini (1997)
• We generalize their set-up in two dimensions:
1. introduce human capital dynamics
◮ wages and unemployment hazard depend on human capital
2. develop a richer economic environment
Economic environment I
• Agent’s intra-period utility u(c) − a
◮ Separable in consumption c and effort a
◮ u(·) increasing, concave, smooth, and u−1
has convex first derivative (Newman, 1995)
• Agent’s effort a ∈ {0, e}, i.e. two effort levels
• Workers endowed with human capital h ≥ 0, and the wage ω(h) is
Economic environment II
• Agent can be either employed or unemployed
• Employment is an absorbing state
• During unemployment:
◮ human capital depreciates: h′ = (1 − δ)h
◮ search with job-finding probability π(h, a), where
π(h, e) > π(h,0) = 0, and πh(h, e) > 0
Principal-Agent relationship
• Agent’s search effort a is private information
◮ Monitoring technology: upon payment of per-period
monitoring cost κ, principal can observe search effort a
• The risk-neutral principal offers a contract that specifies:
◮ recommendations on the search effort level a
◮ consumption for agent
Options of the contract as policies of the WTW program
• Combination of recommendations on effort, and use of monitoring
technology lead to 3 policy instruments:
◮ UI: Unemployment Insurance (high effort, no monitoring)
◮ JM: Job-search Monitoring (high effort, monitoring)
Options of the contract as policies of the WTW program
• Combination of recommendations on effort, and use of monitoring
technology lead to 3 policy instruments:
◮ UI: Unemployment Insurance (high effort, no monitoring)
◮ JM: Job-search Monitoring (high effort, monitoring)
◮ SA: Social Assistance (low effort, no monitoring)
Economic forces at work in the choice of policies
Economic forces at work in the choice of policies
• Returns to search (UI/JM vs SA): increasing in h
Economic forces at work in the choice of policies
• Returns to search (UI/JM vs SA): increasing in h
• Effort compensation cost (UI/JM vs SA): increasing in U
• Incentive cost (UI vs JM)
Us − Uf ≥ e
βπ(h) (IC)
◮ decreasing in h (for UI)
◮ increasing in U: if u−1
Characterization
• Proposition 1: SA is an absorbing policy, i.e. if it is chosen at any
Characterization
• Proposition 1: SA is an absorbing policy, i.e. if it is chosen at any
period t, choosing it thereafter is optimal
• Proposition 2: Human capital depreciation is necessary for policy
transition within an optimal WTW program, i.e. in absence of
Characterization
• Proposition 1: SA is an absorbing policy, i.e. if it is chosen at any
period t, choosing it thereafter is optimal
• Proposition 2: Human capital depreciation is necessary for policy
transition within an optimal WTW program, i.e. in absence of
human capital depreciation, every policy is absorbing.
• Proposition 3: With human capital depreciation, the optimal
Quantitative analysis
• Calibration of the model to U.S. labor market
• Compute optimal WTW program for the same level of expected
utility U0 promised by the current program
Quantitative analysis
• Calibration of the model to U.S. labor market
• Compute optimal WTW program for the same level of expected
utility U0 promised by the current program
◮ Simulate current program to compute U0
• Question I: How different are the features of the optimal program
Quantitative analysis
• Calibration of the model to U.S. labor market
• Compute optimal WTW program for the same level of expected
utility U0 promised by the current program
◮ Simulate current program to compute U0
• Question I: How different are the features of the optimal program
compared to the current one?
• Question II: How big is the budget saving for the government from
Calibration I: U.S. labor market
• Period: one month
• Focus on individuals aged 18-50, ≤ HS degree
• Preferences: log(c) − a
• Skill depreciation rate: 15% per year
Calibration II: U.S. WTW system
Calibration II: U.S. WTW system
• Phase I: 6 months of UI, with 60% replacement ratio
• Phase II: Up to 24 months of Temporary Assistance for Needy
Families ($740 per month) with enrollment into JM program:
Calibration II: U.S. WTW system
• Phase I: 6 months of UI, with 60% replacement ratio
• Phase II: Up to 24 months of Temporary Assistance for Needy
Families ($740 per month) with enrollment into JM program:
◮ κ = $478 per worker-per month
• Phase III: policies of social assistance (SA)
Calibration II: U.S. WTW system
• Phase I: 6 months of UI, with 60% replacement ratio
• Phase II: Up to 24 months of Temporary Assistance for Needy
Families ($740 per month) with enrollment into JM program:
◮ κ = $478 per worker-per month
• Phase III: policies of social assistance (SA)
◮ Food Stamps, with indefinite duration ($290 per month)
10 20 30 40 50 60 0 0.2 0.4 0.6 0.8 1 Months
Fraction of workers in each program
Program assignment
10 20 30 40 50 60 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Months
Fraction of initial wage
Payments (replacement ratio)
10 20 30 40 50 60 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Months
Fraction of current wage
Subsidy upon re−employment
0.4 0.6 0.8
1 Program assignment
0.5 0.6 0.7 0.8 0.9 1
Payments (replacement ratio)
0.15 0.2 0.25 0.3 0.35