Do Salaried Workers Perform Better than Piece Rate Workers?

Transcription

1 Do Salaried Workers Perform Better than Piece Rate Workers? Alex Bryson (NIESR, CEP, London) Babatunde Buraimo (University of Central Lancashire) Rob Simmons (Lancaster University Management School) Aalto University, Helsinki, 5 th May 2010

2 Motivation Literature indicating piece rate workers are more productive than fixed rate workers Effect due to incentives and worker sorting But salary contracts have their own incentive effects which have not been explored very much in the empirical literature We can test what the productivity effects of piece rate->salary are using panel data on football referees in England

3 Our contribution Identify productivity effects of switch from piece rate to salary. Exploit panel data on individual workers in an occupation where some but not all of the contracts on offer switch from piece rate to salary. We observe workers over time pre- and post- the switch so can account for fixed unobservable worker attributes (eg. ability) affecting worker probabilities of being offered a salaried contract. We observe workers entering/leaving the population of workers in the occupation so can examine the extent to which any productivity effects are driven by worker sorting.

4 Empirical Literature on Piece vs. Fixed Rate Switch from fixed rate to piece rate increases labour productivity Lazear (2000) Safelite: half incentive, half worker sorting Shearer (2004) tree-planters; randomisation Switch from piece rate to fixed rate/salary: Freeman and Kleiner (2005): shoe manufacturing Productivity fell but profits rose Fernie and Metcalf (1999): jockeys who switched to salary-like retainer perform more poorly than those continuing to be paid piece rate Yet piece rate in decline and most are paid salary Perverse incentives; opportunities to game ; costs of buying out worker risk aversion; rise of team production; costs of additional worker monitoring and system maintenance

5 Incentive effects of salaries despite the fact that there is no immediate relation between pay and performance, a salaried worker is likely to have incentives to exert effort because good performance will improve future contracts. Such reputational concerns imply that effort exertion can occur without explicit pay-for-performance contracts Prendergast (1999: 10) Long-term salary contracts: deferred payments; promotions Short-term salary contracts: contract renewal; income security

6 English Football Referees Prior to 2001/02 season all paid piece rate per match From 2001/02 Premiership referees paid a salary plus piece rate match fee Salary contracts reviewed after 2 seasons (later 1) Income from salaries much higher than piece rate contracts and real value increases over time Salaried referees required to undertake regular offthe-job training All Premiership and Championship referees eligible if promoted to the Select List Elsewhere, referees in Spain s La Liga get a salary, but referees in Germany still paid by match fee

7 Number of referees each season by tier and N salaried contracts Season Tier 1 only Tier 2 only Tiers 1 and 2 All referees Referees on salaried contracts

8 Mean number of games per season by tier and type of contract Tier 1 Tier 2 Season Contract Noncontract Contract Noncontract

9 Salaries and Tier 1 match fees for referees after 2001 Season Salary Match fee 2001/02 33, /06 50, /08 53, /09 57,

10 Mechanisms by which salary may elicit improved performance Career concerns -> incentive Efficiency wage effect (higher wages) Quality effect through training Worker sorting

11 Referee performance Multiple tasks: subjective assessment by experts with peer review for each match Assessment by managers and media Poor performance readily observable and sanctions immediate Dropped Reputational damage Decisions influenced by fan pressure Controls for local rivals, crowd size

12 Data (1) 11,169 of the 11,184 matches played in tiers 1 and 2 between 1997/98 and 2008/09 All 168 referees officiating those games 9 only tier only tier 2 45 both tier 1 and tier 2 All tier 1 games officiated by salaried referees from 2001/02 with a few exceptions But 15% of tier 2 games officiated by salaried referees

13 Data (2) At referee level we have an unbalanced panel Officiate between 1 and 312 matches Mean = 66 matches 69 referees observed pre- and post-salaried contracts Includes 37 referees on salary at some point after 2001/02 Dependent variable = cards given per match Good referees show fewer cards (Frick et al., 2008); increasing number of cards indicates loss of control by the referee

14 Estimation (1) Y = CONSTANT + β 1 TIME +β 2 TIER 1 + β 3 (TIER 1 POST01/02) + β 4 LOGATTEND + β 5 SCORE + β 6 SCORESQ + β 7 DERBY + β 8 MONTHSTOGO + β 9 HOMEPOINTS + β 10 AWAYPOINTS + ε Unit: football match; DV=cards per match OLS with tier 1*post01/02 capturing introduction of salaries Identifying assumption: interaction captures causal impact of salary contracts on referee performance (rather than contemporaneous changes)

15 Estimation (2) Controls account for fact that referee allocation to games is non-random Attendance; derby ; club ranking prior to the game Closeness of game can affect incentive to foul Months to end of season Non-random allocation of salaries to referees Perhaps salary contract simply reflects talent Therefore we introduce referee fixed effects Run on unbalanced v balanced panel of referees to establish impact of worker sorting

16 OLS and referee fixed effects models for N first yellow cards per game VARIABLES OLS OLS FE FE TIER *** 0.460*** 0.381*** 0.233** (10.16) (6.25) (4.36) (2.49) POST00/ *** 0.226** (3.24) (2.39) (0.20) (0.27) TIER 1 POST00/ *** *** *** *** (7.14) (6.56) (5.24) (4.63) LOGATTEND 0.220*** 0.227*** (4.82) (5.07) SCORE ** * (2.01) (1.68) SCORESQ *** *** (7.33) (7.29) DERBY 0.716*** 0.724*** (8.17) (8.44) MONTHSTOGO 0.067*** 0.067*** (9.95) (10.21) HOMEPOINTS ** ** (2.35) (2.17) AWAYPOINTS 0.140*** 0.157*** (3.81) (4.36) CONSTANT 2.620*** *** (38.55) (0.59) (38.13) (1.09) Year Dummies YES YES YES YES Observations Adjusted R-squared

17 OLS and referee fixed effects models for the number of yellow and red cards per game VARIABLES OLS OLS FE FE TIER *** 0.484*** 0.375*** 0.230** (9.85) (6.16) (4.62) (2.29) POST00/ *** 0.268*** (3.43) (2.64) (0.10) (1.18) TIER 1 POST00/ *** *** *** *** (7.16) (6.64) (5.32) (4.75) LOGATTEND 0.219*** 0.225*** (4.49) (4.70) SCORE (1.56) (1.24) SCORESQ *** *** (6.55) (6.47) DERBY 0.737*** 0.739*** (7.86) (8.05) MONTHSTOGO 0.073*** 0.073*** (10.09) (10.35) HOMEPOINTS ** ** (2.32) (2.15) AWAYPOINTS 0.160*** 0.176*** (4.06) (4.58) CONSTANT 2.753*** *** (37.89) (0.74) (37.74) (1.32) Year Dummies YES YES YES YES Observations Adjusted R-squared

18 OLS and referee fixed effects models for the number of yellow and red cards per game, confined to referees observed in the pre- and post-salary contract eras VARIABLES OLS OLS FE FE TIER *** 0.375*** 0.365*** 0.236** (7.30) (4.23) (3.84) (2.28) POST00/ (1.01) (0.57) (0.17) (0.79) TIER 1 POST00/ *** *** *** *** (4.94) (4.22) (5.19) (4.59) LOGATTEND 0.196*** 0.209*** (3.45) (3.76) SCORE * (1.86) (1.64) SCORESQ *** *** (5.74) (5.77) DERBY 0.831*** 0.823*** (8.18) (8.28) MONTHSTOGO 0.080*** 0.080*** (9.79) (10.06) HOMEPOINTS (1.06) (1.16) AWAYPOINTS 0.173*** 0.191*** (3.92) (4.44) CONSTANT 2.893*** *** (33.72) (1.08) (35.23) (1.29) Year Dummies YES YES YES YES Observations Adjusted R-squared

19 Salary effects over time Interacted tier 1 with years since 2000/01 to see whether this is informative with respect to alternative mechanisms Coefficient immediately negative, significant and grows in first few years, then stabilises and becomes ns in last 2 years. Competing explanations: Competitive shock Piece rate referees catch up Not consistent with income effect rising over time Training regime effects not sustained

20 Conclusion Worker performance improves with salary Runs counter to finding in literature on piece rates perhaps because first paper on professionalisation Some of the impact due to non-random allocation of contracts to workers and to worker sorting Remainder due to salary contracts themselves which can be interpreted as career concerns, training or income