Labor economics1. Too-high wages and unemployment Efficiency wages vs gift exchangeWhy are wages too high?Moral hazard in contracting: Theory and experimental evidenceCompetition does not drive wages down…firms choose high wage offer workers & expect reciprocity2. Crowding outBenabou-Tirole REStud 033.Critiques of standard agency modelBehavioral critiques4. Labor supplyCab driver “income targeting” (Camerer et al QJE 97)Cab driver instrumental variables (IV) showing experience effectFarber (JPE 04) hazard rate estimation: Do hrs worked or accumulated income predict quitting?Do they quit because of hours or $? Getting tired is a stronger regularity than targettingLabor economicsLabor economicsWhy is labor behaviorally interesting?Why is labor behaviorally interesting?Important in scaleImportant in scaleWhat people sell is themselves (identity, What people sell is themselves (identity, appreciation)appreciation)Natural social comparison with othersNatural social comparison with othersQuality assurance problem + room for rationalizationQuality assurance problem + room for rationalizationFirms’ problem is endogenous sorting & incentiveFirms’ problem is endogenous sorting & incentiveBehavioral effects in labor markets: Behavioral effects in labor markets: ““Gift exchange” and supra-marginal wagesGift exchange” and supra-marginal wagesCrowding outCrowding outCritique of the single-activity agency modelCritique of the single-activity agency modelLabor supply: CabsLabor supply: Cabs1. Too-high wages and unemployment1. Too-high wages and unemploymentEfficiency wages vs gift exchangeEfficiency wages vs gift exchangesupplydemandWage wQuantity QPrice P unemployment at wWhy are wages too high? Why are wages too high? Efficiency wages (Stiglitz et al)Efficiency wages (Stiglitz et al)Pay “too much” so workers have Pay “too much” so workers have something to lose if they shirksomething to lose if they shirkWhy don’t workers bid for jobs?Why don’t workers bid for jobs?Role for nepotism, social networks, “hiring Role for nepotism, social networks, “hiring bonusses” ($5k consulting firm “bounties”)bonusses” ($5k consulting firm “bounties”)““Gift exchange” (Akerlof-Yellen)Gift exchange” (Akerlof-Yellen)Pay “too much” so workers reciprocate Pay “too much” so workers reciprocate with high (uncontractible) effortwith high (uncontractible) effortConsistent with resistance to wage cuts Consistent with resistance to wage cuts (Bewley)(Bewley)Experimental evidence (Fehr et al, PJ Healy,…)Experimental evidence (Fehr et al, PJ Healy,…)Moral hazard in contracting: Moral hazard in contracting: Theory and experimental evidenceTheory and experimental evidenceFehr setup:Fehr setup:Firms offer wFirms offer wFirms earn 10e-wFirms earn 10e-wWorkers choose eWorkers choose eWorkers earn w-c(e)Workers earn w-c(e)No reputations (cf. No reputations (cf. PJ Healy)PJ Healy)Competition does not drive wages down…firms Competition does not drive wages down…firms choose high wage offer workers & expect reciprocitychoose high wage offer workers & expect reciprocity2. Crowding out2. Crowding outDo extrinsic ($) incentives crowd out Do extrinsic ($) incentives crowd out intrinsic motivation?intrinsic motivation?Do puzzles for $ or no-$. After $ removed, Do puzzles for $ or no-$. After $ removed, no-$ group does more puzzles (Deci et al)no-$ group does more puzzles (Deci et al)Female tennis players: Play for fun as kids…Female tennis players: Play for fun as kids………later on tour, quit after getting appearance later on tour, quit after getting appearance feefeeQ: Is it a “strike” or permanent decrease in Q: Is it a “strike” or permanent decrease in incentive? incentive?Benabou-Tirole REStud 03Benabou-Tirole REStud 03Workers infer task difficulty or skill from Workers infer task difficulty or skill from wage offer (“overjustification”, “self-wage offer (“overjustification”, “self-perception”, “looking glass self”)perception”, “looking glass self”)Worker exerts effort 0,1, cost is c in [cWorker exerts effort 0,1, cost is c in [c**,c*],c*]Worker gets signal Worker gets signal σσ correlated with c correlated with cSuccess pays V to agent, W to firmSuccess pays V to agent, W to firmΘΘ is probability of success given effort is probability of success given effortFirm offers bonus bFirm offers bonus bWorker exerts effort c(Worker exerts effort c(σσ,b),b)<<ΘΘ(V+b)(V+b) works if works if σσ>>σσ*(b)*(b)Prop 1: In equilibriumProp 1: In equilibriumBonus is short-term reinforcer: bBonus is short-term reinforcer: b11<b<b22  σσ*(b*(b11)>)>σσ*(b*(b22))Rewards are bad news: bRewards are bad news: b11<b<b22E[c|E[c|σσ11,b,b11] < E[c|] < E[c|σσ22,b,b22]]Empirical leverage: Negative effect occurs only Empirical leverage: Negative effect occurs only if firm knows more about task difficulty or if firm knows more about task difficulty or worker skill than the worker knowsworker skill than the worker knows3.Critiques of standard agency model3.Critiques of standard agency model Standard model (one activity)Standard model (one activity)Firms pay wage package w=f+b(e+Firms pay wage package w=f+b(e+θθ))Workers choose hidden effort eWorkers choose hidden effort eb is “piece rate”, b is “piece rate”, θθ is “luck” is “luck”Risk-neutral firms earn Risk-neutral firms earn ΠΠ(e)-w(e)-wRisk-averse workers earn w-c(e)-var(w)Risk-averse workers earn w-c(e)-var(w)Tradeoff: Tradeoff: ““High powered incentive” b iHigh powered incentive” b increases ncreases motivation…motivation………but creates bad variance in wagesbut creates bad variance in wagesBehavioral critiquesBehavioral critiquesWorkers don’t know c(e) (prefs constructed)Workers don’t know c(e) (prefs constructed)U(W-r) depends on reference pointU(W-r) depends on reference pointPrevious wages, wages of othersPrevious wages, wages of othersWorkers care about procedures or income sourceWorkers care about procedures or income sourcePsychic income: meaning and appreciationPsychic income: meaning and appreciationCrowding out of intrinsic motivatoinCrowding out of intrinsic motivatoinBiases in separating e and Biases in separating e and θθHindsight bias