Management and Inequality Nick Bloom Stanford Scott Ohlmacher

Management and Inequality Nick Bloom (Stanford) Scott Ohlmacher (Census) Cristina Tello-Trillo (Census) ASSA January 4 th 2019 Disclaimer: Any opinions and conclusions expressed herein are those of the author and do not necessarily represent the views of the US Census Bureau. All results have been reviewed to ensure that no confidential information is disclosed.

Long history of work on management in economics e. g. Walker (1887)

Francis Walker (1840 -1897) was the founding President of the AEA Walker ran the 1870 and 1880 Census, claiming management was the major source of performance differences across US firms in Walker (1887) But he had no management data – this was pretty much pure speculation

So the US Census ran the Management and Organizational Practices Survey (MOPS) in 2010 and 2015 (and in preparation for 2020)

Initial work on the MOPS management data looked at plant performance, e. g. Source: Bloom, Brynjolfsson, Foster, Jarmin, Patnaik, Saporta. Eksten & Van Reenen (forthcoming AER) 5

What about Management & Inequality? Many claim that aggressive management practices only enrich CEOs and managers - presumably raising inequality Maybe the rise of more structured management (private equity, multinationals etc) is driving the rise in inequality? 6

Data – management and worker earnings Management and Inequality Management and Earnings Volatility 7

Management & Organizational Practices Survey 2010 It was delivered to ~50, 000 manufacturing plants in 2011 (asking about 2010) and in 2016 (asking about 2015) This was quick and easy to fill out - and mandatory - so 74% of plants responded. In 2010: covering 5. 6 m employees (>50% of US manufacturing employment)

MOPS contacts were mostly senior managers

MOPS asks about performance monitoring e. g.

Examples of monitoring– manufacturing

Example of no performance metrics: Textile Plant

Examples of monitoring: hotels (from a prior ASSA)

MOPS also asks about incentives e. g.

Examples of incentives - performance reviews 15

Each of the 16 questions is assigned a value from 0 (least structured) to 1 (most structured) 8 Monitoring 4 Bonus 16 Management 8 Incentives 2 Promotions 2 Reassignment/ Dismissal 16

Overall management score displays a wide spread Note: The management score is the average of the scores for each of the 16 questions

Longitudinal Employer-Household Dynamics (LEHD) § Linked employer-employee quarterly wage data for all workers in state unemployment insurance records § Use workers with quarterly earnings at least full-time federal minimum wage ($3, 800) around 2010 (2009 Q 4 -2011 Q 1) § Use firm-state (SEIN) manufacturing with 20+ employees 18

Data Management and Inequality Management and Earnings Volatility 19

Correlation of management and within firm inequality is…. Decreasing in Structured Management (binscatter) 20

Correlation of management and within firm inequality is strongly negative Decreasing in Structured Management (binscatter) 21

Maybe this is all due to industry, regional, size, age or some other variation? 22

No – the Management and within firm inequality correlation is very robust Dependent Variable Management Log(Emp) Log(Capital/Emp) Log(VA/Emp) Share of Employees w/ a Bachelor's Degree Firm Age Log(Firm Employment) Observations (Firm-State) Number of Firms (Clusters) Fixed Effects Log(90 th Percentile) - Log(10 th Percentile) (1) (2) (3) -0. 1447*** -0. 1066*** -0. 057*** (0. 0185) (0. 0192) (0. 019) -0. 0312*** -0. 013*** (0. 0026) (0. 003) -0. 0207*** -0. 016*** (0. 0032) (0. 003) 0. 0084** 0. 015*** (0. 0038) (0. 004) 0. 2027*** 0. 201*** (0. 0203) (0. 020) 0. 001 (0. 000) -0. 022*** (0. 002) 17, 000 11, 000 23 Industry, State

This negative management & within-firm inequality correlation driven by the greater rise in lower half earnings at firms with more structured management 24

The 90 -10 Earnings Differential is Strongly Decreasing in the 8 Monitoring questions 25

The 90 -10 Earnings Differential is Weakly Increasing in 8 Incentives questions 26

Within incentives bonuses and reassignment (or dismissal) the most linked to inequality Dependent Variable Monitoring Incentives Bonuses Promotions Reassignment/Dismissal Observations (Firm-State) Number of Firms (Clusters) Fixed Effects Log(90 th Percentile) - Log(10 th Percentile) (1) (2) -0. 146*** -0. 143*** (0. 018) 0. 049*** (0. 014) 0. 035*** (0. 009) -0. 018* (0. 010) 0. 020*** (0. 007) 17, 000 11, 000 Industry, State Note: Includes controls for log(SEIN employment), log(parent firm employment), log(capital/employment), 27 log(VA/emp), employee share with a degree and firm age.

Robustness: look at longer-run pay for workers in the firm 2009 -2011, finding similar results Dependent Variable Management Monitoring Incentives Bonuses Promotions Reassignment/Dismissal Observations (Firm-State) Num Firms (Clusters) Fixed Effects Log(90 th) - Log(10 th) Percentile (1) (2) (3) -0. 071*** (0. 022) -0. 179*** -0. 176*** (0. 020) 0. 058*** (0. 016) 0. 038*** (0. 009) -0. 008 (0. 011) 0. 016** (0. 007) 14, 500 10, 000 Industry, State Note: Includes controls for log(SEIN employment), log(parent firm employment), log(capital/employment), 28 log(VA/emp), employee share with a degree and firm age. Uses earnings from 2009 Q 1 to 2011 Q 4.

More generally find a weak negative link between performance and inequality Dependent Variable Log(Firm Employment) Log(Shipments/Emp) Log(Profit/Shipments) Log(90 th Percentile) - Log(10 th Percentile) (1) (2) (3) -0. 027*** (0. 002) -0. 011*** (0. 004) -0. 024*** (0. 007) Largest Plant TFP Observations (Firm-State) Number of Firms (Clusters) Fixed Effects 17, 000 11, 000 Industry, State

Data Management and Inequality Management and Earnings Volatility 30

Well known inequality exists within and between firms (and is increasing in both) – motivating this paper Source: Song, Bloom, Guvenen, Price and Von Wachter (2019, QJE)

Less well known: US earnings volatility is falling LEHD data, (Abowd and Mc. Kinney, 2019) SSA data, (Bloom, Guvenen, Pistaferri, Sabelhaus, Salgado & Song, 2018)

So what about management and earnings volatility – maybe good management reduces inequality but increase volatility? Measure variance of the four quarters of 2010 earnings growth for each employee, then average at the SEIN (firm-state) level 33

Earnings volatility small positive correlation with management (negative for monitoring and positive for incentives) Dependent Variable Management Monitoring Incentives Bonuses Promotions Reassignment/Dismissal Observations (Firm-State) Num Firms (Clusters) Fixed Effects Variance in Log(Quarterly Worker Earnings) (1) (2) (3) 0. 005** (0. 002) -0. 012*** -0. 011*** (0. 002) 0. 011*** (0. 001) 0. 015*** (0. 001) -0. 004*** (0. 002) -0. 001 (0. 001) 17, 000 11, 000 Industry, State Note: Includes controls for log(SEIN employment), log(parent firm employment), log(capital/employment), 34 log(VA/emp), employee share with a degree and firm age.

One mechanism is simply 4 th quarter bonuses Dependent Variable Management Monitoring & Targeting Incentives Bonuses Promotions Reassignment/Dismissal Obs (Firm-State) Num Firms (Clusters) Fixed Effects Firm-State Mean of (Log Q 4 Earnings - Average Log Earnings for Q 1 -Q 3) (1) (2) (3) 0. 020** (0. 008) -0. 021*** -0. 019** (0. 008) (0. 007) 0. 028*** (0. 006) 0. 031*** (0. 004) -0. 003 (0. 003) 17, 000 11, 000 Industry, State Note: Includes controls for log(SEIN employment), log(parent firm employment), log(capital/employment), log(VA/emp), employee share with a degree and firm age.

But not just individual 4 th quarter bonuses as results are similar in the 3 year panel 2009 -2011 Dependent Variable Management Monitoring & Targeting Incentives Bonuses Promotions Reassignment/Dismissal Obs (Firm-State) Num Firms (Clusters) Fixed Effects Average Variance in Log(Quarterly Worker Earnings) (1) (2) (3) 0. 007*** (0. 002) -0. 010*** -0. 009*** (0. 002) 0. 012*** (0. 001) 0. 013*** (0. 001) -0. 003** (0. 001) -0. 000 (0. 001) 14, 500 10, 000 Industry, State Note: Includes controls for log(SEIN employment), log(parent firm employment), log(capital/employment), 36 log(VA/emp), employee share with a degree and firm age.

Conclusions 1) Structured management practices (and better firm performance) are correlated with lower within-firm inequality 2) Offsetting effects: § Monitoring is correlated with less within firm inequality (and lower volatility) § Incentives - particularly bonuses & firing - correlated more within firm inequality (and higher volatility) Next: (A) panel data (2015 MOPS), and (B) some causality…. 37

Thank you 38

Performance and Inequality Dependent Variable Log(Firm Employment) Log(90 th Percentile) - Log(10 th Percentile) -0. 014*** (0. 003) Log(Emp) -0. 022*** (0. 002) -0. 034*** Average Annual Employment (0. 011) Growth, 2005 -2010 (Winsorized) Log(Capital/Emp) -0. 013*** (0. 003) 0. 201*** Share of Employees w/ a Bachelor's Degree (0. 020) Firm Age 0. 000 (0. 000) Observations (Firm-State) 17, 000 Number of Firms (Clusters) 11, 000 Fixed Effects Industry, State

Monitoring Question Examples Return

Targeting Question Examples Return

Bonus Question Examples Return

Promotion Questions Return

Reassignment & Dismissal Question Example Return

Establishment-Level Results from Bloom et al. (2013) Dependent Variable Management Log(Emp) Log(Capital/Emp) Share of Employees w/ a Bachelor's Degree Observations (Firm-State) Number of Firms (Clusters) Fixed Effects Log(VA/Emp) (1) (2) 1. 272*** 0. 498*** (0. 05) (0. 037) -0. 035*** (0. 006) 0. 179*** (0. 007) 0. 418*** (0. 041) 32, 000 18, 000 None Industry Log(Profit/ Shipments) (3) 0. 058*** (0. 01) 0. 001 (0. 002) 0. 01*** (0. 002) 0. 004 (0. 011) 32, 000 18, 000 Industry Return

Structured Management Strongly Correlated with Performance (Bloom et al. 2019) Dependent Variable Management Log(Emp) Log(Capital/Emp) Share of Employees w/ a Bachelor's Degree Observations (Firm-State) Number of Firms (Clusters) Fixed Effects Log(TFP of Log(Shipments Log(Profit/ Log(VA/Emp) Largest Plant) /Emp) Shipments) (1) (2) (3) (4) (5) 1. 281*** 0. 620*** 0. 075*** 0. 691*** 0. 064*** (0. 052) (0. 044) (0. 029) (0. 038) (0. 022) 0. 012 -0. 005 0. 004** (0. 008) (0. 007) (0. 002) 0. 002*** -0. 000 (0. 001) (0. 000) 0. 673*** 0. 637*** -0. 024 (0. 052) (0. 045) (0. 044) 17, 000 17, 000 11, 000 11, 000 None Industry, State

Descriptive Statistics Log(90 th Percentile) - Log(10 th Percentile) Mean 0. 975 Standard Deviation 0. 305 25 th 75 th Percentile 0. 761 1. 152 Log(90 th Percentile) - Log(50 th Percentile) 0. 617 0. 244 0. 446 0. 748 Log(50 th Percentile) - Log(10 th Percentile) 0. 359 0. 141 0. 257 0. 439 Average Variance in Log(Quarterly Worker Earnings) 0. 033 0. 032 Management Score 0. 658 0. 136 0. 581 0. 757 Monitoring & Targeting Score 0. 698 0. 153 0. 604 0. 813 Incentives Score 0. 607 0. 185 0. 500 0. 739 Bonuses Score 0. 413 0. 285 Promotions Score 0. 858 0. 257 Reassignment/Dismissal Score 0. 632 0. 347 Log(Emp) 4. 882 1. 065 47

Again, relationship particularly driven Dependent Variable Monitoring & Targeting Incentives Log(90 th Percentile) - Log(50 th Percentile) (1) -0. 094*** (0. 015) 0. 042*** (0. 012) Log(50 th Percentile) - Log(10 th Percentile) (2) -0. 053*** (0. 008) 0. 007 (0. 006) -0. 014*** (0. 002) -0. 019*** (0. 003) 0. 009*** (0. 003) 0. 087*** (0. 016) 0. 000* (0. 000) -0. 020*** (0. 002) 17, 000 11, 000 Industry, State 0. 002* (0. 001) 0. 005*** (0. 001) 0. 006*** (0. 002) 0. 116*** (0. 010) 0. 000 (0. 000) -0. 000 (0. 001) 17, 000 11, 000 Industry, State Bonuses Promotions Reassignment/Dismissal Log(Emp) Log(Capital/Emp) Log(VA/Emp) Share of Employees w/ a Bachelor's Degree Firm Age Log(Firm Employment) Observations (Firm-State) Number of Firms (Clusters) Fixed Effects 48

Linking LEHD & MOPS § Aggregate MOPS (& ASM) to the firm-state (SEIN) level § Employment-weighted mean of management scores § Sum of shipments, employment, etc. 49