Firming Up Inequality - PDF Free Download

Firming Up Inequality Jae Song Social Security Administration Fatih Guvenen Minnesota, FRB Mpls, NBER David Price Stanford Nicholas Bloom Stanford and NBER Till von Wachter UCLA and NBER February 22, 2016

Motivation I US income inequality has been rising for almost four decades I Much research on inequality between groups defined by observable characteristics: education/skill, occupation, age, gender, race, and so on. Main conclusion: substantial rise in within-group inequality. I This paper: study the employer/firm as an observable worker characteristic: Between firms (e.g., top firms are paying better) Within firms (e.g., higher executive pay relative to average pay) Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 1 / 93

This Paper Two questions: 1. How much of the rise in income inequality is between firms and how much is within firms? For bottom 99% : Almost all of it between firms: Almost all of it between firms For the top 1% : Almost all of it between firms up to 99.8th percentile: Almost all of it between firms up to 99.8th percentile 2. Why has inequality risen so much between firms? Large rise in sorting between firms and workers sorting between firms and workers Large rise in Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 2 / 93

Outline I The Social Security Administration (SSA) database I Non-parametric results on inequality The bottom 99% Robustness (region, industry, gender, age, measures) The top 1% I More formal econometric approach I Why is this happening? The changing structure of firms Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 3 / 93

The Data

Data: SSA Master Earnings File I Universe of all W-2s from 1978 to 2013. I For each job: SSN, EIN, and total compensation Total compensation includes: wages, salaries, tips, restricted stock grants, exercised stock options, severance payments, and many other types of income considered remuneration for labor services by the IRS. I For each worker: age, sex, place of birth, date of death I For each EIN: 4-digit SIC (industry) code, location We define Firm = EIN (same as used by Bureau of Labor Statistics) I No top-coding; no survey response error Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 4 / 93

Building a US Matched Employer-Employee Dataset I MEF: Universe of US workers =) Universe of U.S. firms I Individuals assigned to firm where they earn most of their annual income. I Baseline: Firms with 20+ employees. Workers at those firms. Exclude government and education. Covers 1.1 million firms (about 18% of total) and 103 million workers (73% of total) and $5.4tn in wages (80% of total) Results robust to sample selection (All firms & all sectors) & worker assignment to firms. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 5 / 93

Firm Size Distribution: EIN vs. Census Firm (Log) Number of Firms 0 5 10 15 SSA Firms Census Firms 1 4 5 9 10 19 20 49 50 99 100 499 499 1000 1000+ Employees Notes: Natural log of the number of firms in each size category are shown. Census figures count the number of employees at a point in time, while the SSA numbers count the number of FTEs over the course of a year. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 6 / 93

Total Payroll Total Income Over Time Total Income (Trillions of 2012 Dollars) 3 4 5 6 7 SSA total NIPA 1980 1990 2000 2010 Year Notes: SSA data includes all entries in the MEF. All data are adjusted for inflation using the PCE price index. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 7 / 93

Total Employment Total Employment (Millions) 100 120 140 160 SSA individuals Total Employment Over Time CPS Total Employment 1980 1990 2000 2010 Year Notes: SSA data includes all entries in the MEF. Current Population Survey (CPS) total employment shows the yearly average of the monthly employment numbers in the CPS. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 8 / 93

Number of Firms Total Firms Over Time Number of Firms (Millions) 4.5 5 5.5 6 6.5 SSA firms Census Firms 1980 1990 2000 2010 Year Notes: SSA data includes all entries in the MEF. Census firms shows the total number of firms reported by the Census Bureau s Statistics of U.S. Businesses data set. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 9 / 93

EMPIRICAL RESULTS

Basic Variance Decomposition I w ij t : log income of worker i at firm j I Simple decomposition: w ij t w j t {z} Firm avg. wage + h w ij t w j t i {z } Worker wage rel. to firm avg. ) var i (w ij t {z } ) var j (w j t {z } ) + Total dispersion Between-firm dispersion JX P j var i (w ij t i 2 j). {z } Within-firm j dispersion j=1 P j : employment share of firm j Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 10 / 93

Total Wage Inequality Variance of Log(Wage).2.4.6.8 1 Total Variance 1980 1990 2000 2010 Year Note: Firms with less than 10,000 FTE employees Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 11 / 93

Total vs. Between-Firm Wage Inequality Variance of Log(Wage).2.4.6.8 1 Total Variance Between Firm 1980 1990 2000 2010 Year Note: Firms with less than 10,000 FTE employees Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 12 / 93

Total, Between- and Within-Firm Inequality Variance of Log(Wage).2.4.6.8 1 Total Variance Within Firm Between Firm 1980 1990 2000 2010 Year Note: Firms with less than 10,000 FTE employees Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 14 / 93

Large Firms Only (10,000+ FTE) Variance of Log(Wage).2.4.6.8 1 Total Variance Within Firm Between Firm 1980 1990 2000 2010 Year Note: Firms with more than 10,000 FTE employees Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 15 / 93

A GRAPHICAL FRAMEWORK

Empirical Framework 0.4 Year = 1982 0.35 0.3 Density 0.25 0.2 0.15 0.1 0.05 0 10000 20000 30000 40000 50000 60000 70000 80000 90000 100000 Income Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 17 / 93

Empirical Framework 0.4 Year = 1982 0.35 0.3 Density 0.25 0.2 0.15 0.1 0.05 0 P10 P50 P90 Income Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 18 / 93

Empirical Framework 0.4 Year = 1982 0.3 Density 0.2 0.1 0 P10 P50 P90 Income 0.4 Year = 2012 0.3 Density 0.2 0.1 0 P10 P50 P90 Income Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 19 / 93

Example: No Rise in Inequality Earnings Change: 1982 to 2012 1 1.2 1.4 1.6 1.8 Individuals 0 20 40 60 80 100 Percentiles of Earnings Distribution Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 20 / 93

Example: Rise in Inequality Between Top and Rest Earnings Change: 1982 to 2012 1 1.2 1.4 1.6 1.8 Individuals 0 20 40 60 80 100 Percentiles of Earnings Distribution Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 21 / 93

Example: Rise in Inequality Everywhere Earnings Change: 1982 to 2012 1 1.2 1.4 1.6 1.8 Individuals 0 20 40 60 80 100 Percentiles of Earnings Distribution Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 22 / 93

RESULTS: BOTTOM 99%

Wage Inequality: By Percentile Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage 50%ile 1981: $32k 2013: $36k top %ile 1981: $280k 2013: $540k 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 24 / 93

Calculating Average Log Employer Pay I Take the employers of workers who are in the same percentile bin of income distribution. I Then compute the average of log pay of each employer in this group. I Then compute the average of average log pay across all employers in the group Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 25 / 93

Wage Inequality: Between Firms Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm 50%ile 1981: $30k 2013: $35k top %ile 1981: $49k 2013: $83k 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 26 / 93

Wage Inequality: Within Firms Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 50%ile 1981: 1.06 2013: 1.03 top %ile 1981: 5.7 2013: 6.5 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 27 / 93

ROBUSTNESS

Wage Inequality: Within Firms Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 29 / 93

Many Measures of Firm Pay Log Change, 1981 2013.2 0.2.4.6.8 Avg of Log Wages at Firm (Difference) 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 30 / 93

Many Measures of Firm Pay Log Change, 1981 2013.2 0.2.4.6.8 Avg of Log Wages at Firm (Difference) Firm Log Average (Leaving Out Indv) (Difference) 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 31 / 93

Many Measures of Firm Pay Log Change, 1981 2013.2 0.2.4.6.8 Avg of Log Wages at Firm (Difference) Firm Log Average (Leaving Out Indv) (Difference) Firm Average Wage (Difference) 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 32 / 93

Many Measures of Firm Pay Log Change, 1981 2013.2 0.2.4.6.8 Avg of Log Wages at Firm (Difference) Firm Log Average (Leaving Out Indv) (Difference) Firm Average Wage (Difference) 50th %ile at Firm (Difference) 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 33 / 93

Robustness: Std Dev. Log Wage Log Change, 1982 2012 0.2.4.6.8 Std Dev of Log Wage at Firm Indv Total Wage 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 34 / 93

Robustness: Frac. Going to Bottom 95% Log Change, 1982 2012 0.2.4.6.8 Frac of Wages to Bottom 95% Indv Total Wage 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 35 / 93

Individual Industries

Wage Inequality: Controlling for (4-Digit SIC) Industry Log Change, 1981 2013 0.2.4.6 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Within Industry Note: Sample contains workers in firms with 20+ full-time equivalent employees. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 37 / 93

Fama-French Industries: Beer and Liquor Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Beer & Liquor Note: Sample contains an average of 65,660 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 38 / 93

Fama-French Industries: Candy and Soda Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 193,000 workers in 1981 and 2013. Candy & Soda Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 39 / 93

Fama-French Industries: Pharmaceuticals Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Pharmaceutical Products Note: Sample contains an average of 140,650 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 40 / 93

Fama-French Industries: Chemicals Log Change, 1981 2013 0.5 1 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 644,660 workers in 1981 and 2013. Chemicals Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 41 / 93

Fama-French Industries: Defense Log Change, 1981 2013.2 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 74,350 workers in 1981 and 2013. Defense Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 42 / 93

Fama-French Industries: Recreation Log Change, 1981 2013.2 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Recreation Note: Sample contains an average of 142,200 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 43 / 93

Fama-French Industries: Utilities Log Change, 1981 2013.2 0.2.4.6 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Utilities Note: Sample contains an average of 703,320 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 44 / 93

Fama-French Industries: Consumer Goods Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 1,699,270 workers in 1981 and 2013. Consumer Goods Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 45 / 93

Fama-French Industries: Communication Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Communication Note: Sample contains an average of 951,920 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 46 / 93

Fama-French Industries: Computers Log Change, 1981 2013.5 0.5 1 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Computers Note: Sample contains an average of 197,520 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 47 / 93

Fama-French Industries: Electronic Equipment Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Electronic Equipment Note: Sample contains an average of 407,150 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 48 / 93

Fama-French Industries: Agriculture Log Change, 1981 2013.2 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 931,380 workers in 1981 and 2013. Agriculture Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 49 / 93

Fama-French Industries: Insurance Log Change, 1981 2013 0.2.4.6.8 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Note: Sample contains an average of 1,452,050 workers in 1981 and 2013. Insurance Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 50 / 93

Fama-French Industries: Trading Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Trading Note: Sample contains an average of 1,240,390 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 51 / 93

Exceptions

Fama-French Industries: Healthcare Log Change, 1981 2013.5 0.5 1 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Healthcare Note: Sample contains an average of 7,667,800 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 53 / 93

Fama-French Industries: Banking Log Change, 1981 2013.5 0.5 1 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Banking Note: Sample contains an average of 2,013,760 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 54 / 93

Fama-French Industries: Apparel Log Change, 1981 2013.5 0.5 1 1.5 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Apparel Note: Sample contains an average of 606,320 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 55 / 93

Fama-French Industries: Hotels & Restaurants Log Change, 1981 2013.1 0.1.2.3 Indv Total Wage Avg of Log Wages at Firm Indv Wage/Firm Average 0 20 40 60 80 100 Percentile of Indv Total Wage Restaurants, Hotels, Motels Note: Sample contains an average of 2,610,400 workers in 1981 and 2013. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 56 / 93

Subgroups: Bottom 99 pct I By Industry: HERE I By Region: HERE I By Firm Size: HERE I By Sex: HERE I By Age: HERE Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 57 / 93

RESULTS: TOP 1%

Rise in Top 1% Inequality Log Change, 1982 2012 0.5 1 1.5 Indv Total Wage 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 59 / 93

Rise in Top 1% Inequality: Largely Between Firms Log Change, 1982 2012 0.5 1 1.5 Firm Average Wage Indv Total Wage 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 60 / 93

Rise in Top 1% Inequality: Largely Between Firms Log Change, 1982 2012 0.5 1 1.5 Firm Average Wage Indv Total Wage Indv Wage/Firm Average 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 61 / 93

CAUTION

Firm Size: 20 10, 000 FTE (Top 1%) Log Change, 1982 2012.5 0.5 1 1.5 Indv Total Wage Firm Average Wage Indv Wage/Firm Average 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 63 / 93

Firm Size: 10, 000+ FTE (Top 1%) Log Change, 1982 2012.5 0.5 1 1.5 Indv Total Wage Firm Average Wage Indv Wage/Firm Average 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 64 / 93

Recap: Between- vs. Within Rise in Inequality: Fraction Within-Firm 1 0.5 0-0.5-1 100 80 100 60 60 80 40 40 Firm Size Percentiles 20 20 0 Income Percentiles Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 65 / 93

Bottom 99%: Almost All Between Firms Rise in Inequality: Fraction Within-Firm 1 0.5 0-0.5-1 100 80 100 60 60 80 40 40 Firm Size Percentiles 20 20 0 Income Percentiles Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 66 / 93

Rise in Within-Firm: Top 0.5% of Firms Rise in Inequality: Fraction Within-Firm 1 0.5 0-0.5-1 100 80 100 60 60 80 40 40 Firm Size Percentiles 20 20 0 Income Percentiles Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 67 / 93

Why Are Large Firms Different? 1. Top End Figure: Sensitivity to S&P Returns, By Employee Rank and Firm Size Regression Coefficient 0.1.2.3.4 100 1k 500 100 50 25 10 5 4 3 2 1 Rank Within Firm log(wage) vs log(s&p 500) w/ controls, Aggregated by Geometric Mean, Winsorized at Max in Execucomp Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 68 / 93

Why Are Large Firms Different? 1. Top End Figure: Sensitivity to S&P Returns, By Employee Rank and Firm Size Regression Coefficient 0.1.2.3.4 100 1k 1k 5k 500 100 50 25 10 5 4 3 2 1 Rank Within Firm log(wage) vs log(s&p 500) w/ controls, Aggregated by Geometric Mean, Winsorized at Max in Execucomp Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 69 / 93

Why Are Large Firms Different? 1. Top End Figure: Sensitivity to S&P Returns, By Employee Rank and Firm Size Regression Coefficient 0.1.2.3.4 100 1k 1k 5k 5k 10k 500 100 50 25 10 5 4 3 2 1 Rank Within Firm log(wage) vs log(s&p 500) w/ controls, Aggregated by Geometric Mean, Winsorized at Max in Execucomp Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 70 / 93

Why Are Large Firms Different? 1. Top End Figure: Sensitivity to S&P Returns, By Employee Rank and Firm Size Regression Coefficient 0.1.2.3.4 100 1k 1k 5k 5k 10k 10k+ 500 100 50 25 10 5 4 3 2 1 Rank Within Firm log(wage) vs log(s&p 500) w/ controls, Aggregated by Geometric Mean, Winsorized at Max in Execucomp Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 71 / 93

Why Are Large Firms Different? 2. Bottom End Figure: Change in Wage Percentiles By Firm Size 10th %ile 10th %ile Log Change in Indv Pctls Since 1981.4.2 0.2.4 25th %ile 50th %ile 75th %ile 90th %ile 1980 1990 2000 2010 Year 1000 Firm Size < 2000 Log Change in Indv Pctls Since 1981.4.2 0.2.4 25th %ile 50th %ile 75th %ile 90th %ile 1980 1990 2000 2010 Year 10,000 Firm Size Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 72 / 93

) Major Change in Firm Size Pay Relation Number of Employees (Thousands).26.5 1 2 5 6.9 1981 2013 0 20 40 60 80 100 Percentile of Indv Total Wage Median of Firm Size, 20 Firm Size Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 73 / 93

What is the Role of CEO Pay in Rising Inequality? As for wages and salaries... all the big gains are going to a tiny group of individuals holding strategic positions in corporate suites. Paul Krugman (NY Times, 02/23/2015) The primary reason for increased income inequality in recent decades is the rise of the supermanager. Piketty (2013, p. 315) I Policy: Dodd-Frank act (Section 953(b)): companies to report the ratio of top executives compensation to average wage in the firm. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 74 / 93

Rise in Inequality: Baseline Log Change, 1982 2012 0.2.4.6.8 Indv Total Wage 0 20 40 60 80 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 75 / 93

Rise in Inequality Without Top Executives Log Change, 1982 2012 0.2.4.6.8 Indv Total Wage Indv Total Wage (Non Top 1 Employees) 0 20 40 60 80 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 76 / 93

Rise in Inequality Without Top Executives Log Change, 1982 2012 0.2.4.6.8 Indv Total Wage Indv Total Wage (Non Top 1 Employees) Indv Total Wage (Non Top 5 Employees) 0 20 40 60 80 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 77 / 93

Rise in Inequality Without Top Execs: 1000+ FTE Log Change, 1982 2012.2 0.2.4.6 Indv Total Wage Indv Total Wage (Non Top 1 Employees) Indv Total Wage (Non Top 5 Employees) 0 20 40 60 80 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 78 / 93

Top 1% Inequality Without Top Executives: Baseline Log Change, 1982 2012.6.8 1 1.2 1.4 1.6 Indv Total Wage Indv Total Wage (Non Top 1 Employees) Indv Total Wage (Non Top 5 Employees) 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Note: Excluding top 5 individuals reduces the sample size from 76,251 to 73,620 in 1982 ( 3.45%) and from 119,155 to 115,602 in 2012 ( 2.97%). Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 79 / 93

Top 1% Inequality Without Top Executives: 1000+ FTE Log Change, 1982 2012.6.8 1 1.2 1.4 1.6 Indv Total Wage Indv Total Wage (Non Top 1 Employees) Indv Total Wage (Non Top 5 Employees) 99 99.2 99.4 99.6 99.8 100 Percentile of Indv Total Wage Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 80 / 93

Why Don t Executives Matter (Much)? I US Wages and Salaries: $6.9 Trillion I Wage income share of top 1 percent: 12% (Guvenen, Kaplan, and Song (2014)) 12% of $6.9 Tr = $828 Billion I Average annual compensation of S&P500 CEOs: $22 million Total income: $22 million 500 = $11 Billion I $11B Large firm CEOs account for: $828B of top 1 percent. = 1.3% of the total compensation I Bottom line: Top executives control too small a share of the top incomes to make a dent. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 81 / 93

Subgroups: Top 1 pct I By Industry: HERE I By Region: HERE I By Firm Size: HERE I By Sex: HERE I By Age: HERE Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 82 / 93

A More Formal Econometric Approach

What We Have Done So Far I A simple decomposition: w ij t = w j t + h w ij t w j t var i (w ij t )= var j(w j t {z } ) + Between-firm dispersion I Our main conclusion: i JX P j var i (w ij t i 2 j). {z } Within-firm jdispersion j=1 large increase in between-firm dispersion little change in within-firm dispersion, except at the top end for very large firms I Q: Can we go deeper into between and within-firm components? Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 84 / 93

AKM+ Decomposition I Consider this model for wages: w ij t = {z} i + j {z} Worker FE Firm FE I Estimating (1) from US population: 150 million worker FEs and 6 million firm FEs. + Xt i {z} + " i t (1) Time var. char. I Set X i t 0 for a moment. Average firm wage: w j t = j + j I Key decomposition: var i (w ij t )= var j( j )+var j ( j )+cov( i, j ) {z } Between-firm dispersion + X P j (var i ( i i 2 j)+var i (" i t i 2 j)) j {z } Within-firm dispersion Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 85 / 93

AKM Decomposition, Cont d Change in: w ij t = i + j + Xt i + " i t Baseline Drop mega firms Drop large firms Between-Firm varj( j ) 035.6 000042.6 000052.5 Components + varj( j ) 6.6 1.2 4.9 of Variance + 2 cov( i, j ) 31.4 33.0 31.9 + 2 cov( i + j, X i b) 8.2 10.2 12.3 = Between-firm var. 69.1 87.6 102.1 Note: Mega firms: 10,000+ male employees. Large firms: 1,000+ employees. Within-Firm vari( i + X i b i 2 j) 40.0 29.4 21.5 Components + vari(" i t i 2 j) 9.2 16.1 22.3 of Variance = Within-firm var. 30.9 12.4 2.1 Total in var(w ij t ) 100 100 Change in: = i + j + Xt i + " i t Baseline Drop mega firms Drop large firms Between-Firm var j ( j ) 035.6 000042.6 000052.5 Components + var j ( j ) 6.6 1.2 4.9 of Variance + 2 cov( i, j ) 31.4 33.0 31.9 w ij t + 2 cov( i + j, X i b) 8.2 10.2 12.3 = Between-firm var. 69.1 87.6 102.1 Within-Firm var i ( i + X i b i 2 j) 40.0 29.4 21.5 Components + var i (" i t i 2 j) 9.2 16.1 22.3 of Variance Song, Price, Guvenen, Bloom, von Wachter = Within-firmFirming var. Up Inequality30.9 12.4 2.1 86 / 93

Increasing Sorting Joint Worker and Firm Fixed Effect Distribution Interval 1: 1980 1986.025 Proportion of Observations.02.015.01.005 0 1 2 3 4 5 6 7 8 9 10 Firm Fixed Effect Decile Worker Effect Decile 1 Worker Effect Decile 2 Worker Effect Decile 3 Worker Effect Decile 4 Worker Effect Decile 5 Worker Effect Decile 6 Worker Effect Decile 7 Worker Effect Decile 8 Worker Effect Decile 9 Worker Effect Decile 10 Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 87 / 93

Increasing Sorting Joint Worker and Firm Fixed Effect Distribution Interval 5: 2007 2013.025 Proportion of Observations.02.015.01.005 0 1 2 3 4 5 6 7 8 9 10 Firm Fixed Effect Decile Worker Effect Decile 1 Worker Effect Decile 2 Worker Effect Decile 3 Worker Effect Decile 4 Worker Effect Decile 5 Worker Effect Decile 6 Worker Effect Decile 7 Worker Effect Decile 8 Worker Effect Decile 9 Worker Effect Decile 10 Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 88 / 93

Increasing Sorting.01 Change in Joint Worker and Firm Fixed Effect Distribution from Interval 1 to 5 Proportion of Observations.005 0.005.01 1 2 3 4 5 6 7 8 9 10 Firm Fixed Effect Decile Worker Effect Decile 1 Worker Effect Decile 2 Worker Effect Decile 3 Worker Effect Decile 4 Worker Effect Decile 5 Worker Effect Decile 6 Worker Effect Decile 7 Worker Effect Decile 8 Worker Effect Decile 9 Worker Effect Decile 10 Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 89 / 93

Related Evidence I (US, 1977 1992) Rise in between-establishment dispersion of average wage in manufacturing tracks rise in wage dispersion across workers (Dunne et al (2004)). I (US: 1992 2007) Rise in between-establishment inequality is 2/3 of rise in overall wage inequality (Barth et al (2014)). I Very similar results for UK (1984 2001), Faggio, et al (2007) Germany (1985 2009), Card et al (2013) Brazil (1986 1995), Helpman et al (2015) Sweden (1986 2008), Håkanson et al (2015)) I So, whatever the driving force(s) are, they seem global. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 90 / 93

Further Thoughts I Why are worker FEs getting (i) more dispersed across firms, and (ii) more systematically related to firm FEs (sorting)? I In our estimation, correlation between j and j goes from 0.12 up to 0.52 (by 0.40) over the period. Hakanson et al (2015): increasing sorting by cognitive and noncognitive skills in Sweden due to stronger complementarities between worker skills and technology. Handwerker and Spletzer (2015): Increasing occupational segregation in the US. Increased domestic outsourcing: Dube and Kaplan (2010), Berlingieri (2014), and Goldschmidt and Schmieder (2015) Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 91 / 93

Conclusions I Rising in income inequality is almost entirely between firms. Within-firm inequality flat. True for very fine industry groups, across regions, and across firm size categories. Only exception: Very large firms. Within dispersion increased both at very top end and bottom end. I Rise in between inequality, not due to firm effects, but due to rising dispersion of worker FEs and increased sorting. I Evidence points to major changes in firms organization. Song, Price, Guvenen, Bloom, von Wachter Firming Up Inequality 92 / 93