Presentation is loading. Please wait.

Presentation is loading. Please wait.

The Value of Bosses Kathryn L. Shaw SOLE, May 5, 2012.

Similar presentations


Presentation on theme: "The Value of Bosses Kathryn L. Shaw SOLE, May 5, 2012."— Presentation transcript:

1 The Value of Bosses Kathryn L. Shaw SOLE, May 5, 2012

2 Productivity Why are some workers more productive than others? Why are some plants more productive than others? 2

3 HR Practices Incentives (selection) Productivity Incentives (behavioral) Teams Peer Effects Information Technology 3

4 HR Practices Incentives (selection) Productivity Incentives (behavioral) Teams Peer Effects Information Technology Lazear (2000) Ichniowski, Shaw, Prennushi (1997) Bartel, Ichniowski, Shaw (2007) Bandiera, Barankay, Rasul (2009) Hamilton, Nickerson, Owan (2003) Mas and Moretti (2009) 4

5 5 http://www.nytimes.com/2011/03/13/business/13hire.html?pagewanted=1&_r=2 Googles Quest to Build a Better Boss NY Times, March 12, 2011 The starting point was that our best managers have teams that perform better, are retained better, are happier they do everything better, Mr. Bock says. So the biggest controllable factor that we could see was the quality of the manager, and how they sort of made things happen. The question we then asked was: What if every manager was that good? And then you start saying: Well, what makes them that good? And how do you do it?

6 6 Most research tends to focus on these guys.. And we still have limited evidence on the impact of CEOs on productivity.

7 7 What about these guys… A typical boss is a supervisor in retail trade.

8 The Importance of Bosses A digression on retail trade 8

9 Retail trade was growing…manufacturing declining. Source: County Business Patterns 9

10 The dip in the recession… 10

11 Sales jobs fell… 11

12 Supervisors remain… at over 1 million. 12 There is about one supervisor for every 7 sales workers.

13 Retail Key Point 1: there are a lot of bosses. Key Point 2: not all jobs are bad jobs; the variance of pay is high. 13

14 Pay in retail is highly dispersed… 14 Not all jobs are minimum wage jobs.

15 This is especially true for supervisors… 15 Supervisors earn $48,000 at the 75 th percentile; they earn $78,000 at the 90 th percentile. Average pay is $34,000.

16 Supervisors come from all educational backgrounds; returns to education are high. PayEmployment Share Less than High School$16,2369% High School$28,79939% Some College$32,42731% Bachelors Degree$56,00318% Post-BA$89,8383% 100% Source: CPS Data Retail Supervisors Pay and Employment, 2010 Returns to firm size are also high. 16

17 17 –One extreme: Bosses have little effect on worker productivity and bosses get their jobs through internal politics –Another extreme: Workers are substitutable and output is determined by the quality of the supervisor Some bosses are earning a lot; are they worth it? New data allow us to estimate boss effects. Do Bosses Matter?

18 The Value of Bosses Edward P. Lazear Kathryn L. Shaw Christopher T. Stanton 18

19 Questions How much do bosses influence workers productivity? –What is the marginal product of a boss compared to a worker? –What is the variance in bosses productivity? Do some bosses raise worker output more than others? Why are some bosses more productive than others? –Do bosses teach or motivate? –Which bosses should be assigned to which workers? 19

20 DATA 20

21 21 Technology-based Service Jobs A technology-based service job is one in which the company uses some form of advanced IT system to record every transaction and how long it takes Examples: Skilled –insurance-claims processing –computer-based test grading –technical call centers –in-house IT specialists –technical repair workers –some retail sales Examples: Less skilled –airline gate agents –telemarketers –some cashiers

22 Summary Statistics 22 VariableObsMeanStd. Dev. Output Per Hour 5,729,508 10.26 3.16 Uptime 4,870,6100.960.03 Output Per Hour * Uptime 4,870,61010.013.00 Tenure 5,729,508648.91609.83 Worker as Unit of Analysis Number of Unique Bosses Per Worker 23,8783.992.78 Team as Unit of Analysis Daily Team Size 633,8189.044.54 Boss as Unit of Analysis Number of Unique Workers Per Boss 1,94049.1535.41

23 HOW MUCH DO BOSSES MATTER? 23

24 Estimation Estimation of boss effects δ j from q ijt = X it β + α i + δ j + ε ijt. X it contains month dummies, day of week dummies, and a fifth order tenure polynomial 24

25 25 Standard Deviations of Boss And Worker Effects Dependent Variable: Output-per-hour OLS Worker Fixed Effects Worker and Boss Fixed Effects R-squared 0.061 0.237 0.243 Standard Deviation of Worker Effects 1.52 1.45 Standard Deviation of Boss Effects *Avg Team Size (9.04) 4.61 Number of observations 5,729,508 Number of workers 23,878 Number of bosses 1,940

26 The Variance of Boss Fixed Effects 26

27 Is There Non-Random Assignment of Workers to Bosses? Very little. 27 On any given day, 98.4 % of bosses work with both stars and laggards. On any given day, 89% of bosses work with both new and old workers.

28 Do the worst bosses leave? Picking a random day, the probability that a boss is present 1 year later is regressed on measures of boss quality: Bad bosses leave 64% more often. 28 Table 6: The Probability that a Boss is Retained for 1 Year Bottom 10% of Boss Effects-0.236-0.239 (.082)*** Top 10% of Boss Effects-0.044 (.050) Constant0.645 (.021)***(.022)*** R-squared0.018 Number of Observations 1,444 Notes: The data are repeated cross sections of bosses, with their respected estimated boss fixed effect, on January 10 of each year. The dependent variable is an indicator that the boss is present in the data 1 year in the future. Year fixed effects are not displayed. The distribution of boss effects uses each unique boss as the unit of analysis. Standard errors are in parentheses.

29 The Value of a Boss Suppose that the bottom 10% of bosses are about as good as the top 10% of workers. Setting the 10 th percentile of bosses at 12 units/hour, the average boss produces about 18 units/hour. The average boss is 76% more productive than the average worker. 29

30 WHY DO BOSSES MATTER? TEACHING AND MOTIVATION 30

31 Teaching and Motivation 31

32 HETEROGENEITY IN BOSS EFFECTS: MATCHING STAR WORKERS TO STAR BOSSES? 32

33 33 Theory Output is q = H * E with where S is boss skill in teaching or motivating.

34 34 Assignment of Bosses and Workers: The Best Workers with the Best Bosses?

35 CONCLUSION 35

36 36 Summary of Results How much? –The marginal product of a boss is 76% greater than a typical worker, consistent with compensation ratios. –The variance in boss effects is large. Replacing the lowest decile boss with the highest decile boss improves team productivity by as much as adding one team member (9 member team). Why? –Bosses teach and motivate, but teaching is the main role. –Comparative Advantage: Star bosses increase the output of Star workers the most.

37 Conclusion Key Point 3: To understand why some workers are more productive than other workers, study the boss-worker relationship. 37

38 BACKUP MATERIAL 38

39 Actual Variation or Sampling Error? 39 Consider three bosses and their fixed effects: δ 1 estimated using 1000 worker days of data δ 2 estimated using 30 worker days of data δ 3 estimated using 300 worker days of data Fixes: Maximum likelihood to purge sampling variation: assumes the estimates are normally distributed, and separates sampling variation from true variation. Weighting by boss-worker days gives a similar estimate. Random effects estimates via REML recovers the variance directly.

40 40

41 41 Regressions of output-per-hour with lags Lag Type14 Day Avg.14 Day Average1 Day Number of Lags1112 Worker TenureNoYes R-squared0.25990.26170.25160.2558 Coefficient on the first lag0.4020.3750.1050.0968 Coefficient on the second lag0.0749 Standard Deviation of Worker Fixed Effects Weighted by frequency0.790.831.191.09 Standard Deviation of Boss Effects Multiplied by Average Team Size (9.04) Weighted by frequency2.262.123.072.85 NPV of a Standard Deviation of Boss Effects for an Average Team 3.783.393.433.16

42 42 How do boss effects compare to peer effects? Boss effects are large. Peer Effects Estimated from q ijt = X it β + α i + δ j + ξ p ijt + ε ijt where peer effect, p ijt, is specified in several ways: 1.Mean contemporaneous productivity of co-workers on team – very upward biased by daily demand effects. 2.Mean fixed effects of co-workers on team, estimated via two-step non-linear least squares – yields negligible peer effects. 3.Mean of peers first few months of output as a proxy for the peers current output. Results show peer effects not economically significant In contrast with the large boss effects, peer effects are close to zero.

43 43

44 Teaching and Motivation 44

45 45 Teaching and the Fadeout of Boss Effects Teaching ( λ) 0.78 Monthly Rate of Decay ( γ ) 0.87 Amount of Boss Effect Remaining After 1 Year ( γ ^12)* λ 0.13 Standard Deviation of Worker Fixed Effects Weighted by frequency1.28 Standard Deviation of Boss Effects Multiplied by Average Team Size (9.04) Weighted by frequency3.51 Number of Workers1679 Number of Bosses155 Number of Observations 391,730

46 46 Assignment of Bosses and Workers: The Best Workers with the Best Bosses?

47 The Importance of CEOs? CEO pay in the 500 biggest companies in the U.S. 47

48 The Importance of CEOs? CEO-to-worker compensation ratio, with options granted and options realized,1965–2011 Note: "Options granted" compensation series includes salary, bonus, restricted stock grants, options granted, and long-term incentive payouts for CEOs at the top 350 firms ranked by sales. "Options exercised" compensation series includes salary, bonus, restricted stock grants, options exercised, and long-term incentive payouts for CEOs at the top 350 firms ranked by sales. 48

49 49 Robustness and Non-Random Assignment to Bosses Precision of the boss effect estimates –Bias if worker effects do not capture all ability –Worker trends or transitory shocks that are correlated with boss assignment Bias in estimation of the distribution of boss effects –Omitted observations (treatment of the treated)


Download ppt "The Value of Bosses Kathryn L. Shaw SOLE, May 5, 2012."

Similar presentations


Ads by Google