Ability And Income Inequality

Usain Bolt [1]

Part I of a two part series looking at ability and income. Part I focuses on ability income payoffs by occupation and ability’s contribution to income inequality. Part II will examine the geographical distribution of ability and the degree to which ability income payoffs are distorted by cost of living differences by state.

Summary:

• Primary “breadwinners” (individual who earns the most in a household) in the top decile of household income earn significantly more than the average of their peers defined by age, education, occupation, sex and race. The converse is true for workers in the bottom decile of household income. 
• This relative payoff to “ability” has been relatively stable from 1980 to 2015. But this ability multiplier significantly amplified the increased payoff to ”skilled” employment (requiring education) that also occurred over this period. Thus, ability materially exacerbated the increase in household income inequality.

• While anyone’s genetic endowment can be improved, the mere presence of inherent ability is a significant driver of income inequality. The fact that ability exists is not an excuse for complacency about inequality, or for a lack of compassion. However, it does suggest there are limits to what can be achieved by social policies designed to eliminate income inequality.

Ability

The salary of professional athletes is a very visible example of the relative payoff to ability. In an economic context, we can attempt to infer ability by looking at a worker’s wage income relative to their peers. I looked at two sets of peers. The first set I defined as workers who are approximately the same age, have the same educational attainment, and are in the same occupation (“AEO peer” for age, education and occupation). Because of the likelihood of wage discrimination based on race and gender, I expanded the first peer definition set to also include workers of the same sex and race (“AEOSR peer” for age, education, occupation, sex and race).

This analysis used US Census microdata and American Consumer Survey data from 1980 and 2015 provided by IPUMS-USA, University of Minnesota.[2] These dates were chosen because they coincide with a significant increase in household income inequality. Age was categorized into decades (e.g., 20 year olds, 30 year olds, 40 year olds, etc.). Educational attainment was measured using variable “EDUCD.” All workers with less than a high school degree were in one category, but all other EDUCD categories were used “as is” (differentiating between those with Master’s, Doctoral Degrees etc.). Occupation used the granular “occ2010” variable. I used the IPUMS-USA “SEX” and “RACE” variables. The “RACE” variable has seven categories plus two additional categories for a mixture of two and three races respectively. I defined household “head” as the household member with the highest wage income (not the person identified as “person 1” in the Census).

Table 1 shows how much the top household income decile head out-earned their AEO and AEOSR peers in 2015. The median head’s wage income was 1.5 times larger than their AEO peers (average was 2 times larger). When gender and race discrimination are taken into consideration, this median apparent “ability” advantage drops to 1.4 times (and average drops to 1.9 times). Thus discrimination does indeed need to be considered when attempting to measure ability. From this point forward I use the AEOSR peer group as the relevant peer group.

Table 1: Top Decile Households in 2015: Individual Wage Income Ratio To Age/Education/Occupation and Age/Education/Occupation/Sex/Race Averages

Approximately 80% of top decile households are married. Spouses showed a modest median advantage of 1.1 times and average advantage of 1.2 times versus their AEOSR peers. There is no strong evidence that especially skilled household heads married especially skilled spouses. In fact, the data hints that blending reduced total household ability. For example the 3rd quartile combined ability is lower than the separate average of 3rd quartile heads and spouses.

The relative payoff to ability was similar in 1980 as shown in Table 2. Median ability payoff was 1.4 times AEOSR peers as in 2015, with a slightly lower average payoff of 1.7. If the distribution of “ability” is an inherent characteristic of human populations, it would be surprising to find large changes in relative payoffs over time. The greater difference in spouse AEO vs AEOSR ability ratios in 1980 versus 2015 suggest there was greater gender discrimination in 1980.

Table 2: Top Decile Households in 1980: Individual Wage Income Ratio To Age/Education/Occupation and Age/Education/Occupation/Sex/Race Averages

Yo Yo Ma

Payoffs To Ability By Occupation

Tables 3 and 4 show the top 25 most common occupations of household heads in the top household income decile in 2015 and 1980 respectively.[3]. The two tables are not strictly comparable because some occupation codes did not exist in 1980 (those highlighted in grey in Table 3). The last column of both tables show how over or under represented the occupation is in the top decile relative to the entire population.

The occupations highlighted in yellow in Table 3 became more prominent among top decile households from 1980 to 2015. The increasing share of computer related jobs is apparent.

While ability AEOSR payoffs by occupation are relatively stable in both periods (see Physicians/Surgeons and Lawyers), the data suggests that occupations have different potential payoffs. The largest “Wage/Avg Normal Wage” ratios can be interpreted to reveal which jobs show the most variation in ability payoff. Occupations related to sales (including real estate brokers) form a notable cluster of occupations with large ability payoffs. This does not seem to me to be a surprising result. People often talk about whether someone is temperamentally “suited” for sales, and good salespeople are highly valued for their ability to directly drive revenue. Indeed the pay of those that work on commission is directly tied to their ability.

Table 3: Top Decile Households in 2015: Top 25 Most Common Occupations (Normal Wage Is Adjusted By Age, Education, Occupation, Sex and Race) Occupations in Yellow have increased in importance; occupations in grey are new categories not present in 1980.

Occupation	Pct	Wage	Avg AEOSRwage	Wage/Avg Normal Wage	Pct/Universe Pct
Managers; nec (including Postmasters)	7.5%	$178,931	$97,659	1.99	4.89
Chief executives and legislators/public administration	4.7%	$241,431	$158,793	1.62	10.89
Physicians and Surgeons	4.2%	$290,286	$217,761	1.40	13.87
Lawyers; and judges; magistrates; and other judicial workers	3.7%	$219,076	$131,967	1.73	9.48
First-Line Supervisors of Sales Workers	3.2%	$175,778	$72,474	2.69	2.08
Registered Nurses	2.6%	$117,763	$66,479	1.85	2.28
Software Developers; Applications and Systems Software	2.3%	$154,408	$104,263	1.52	5.59
Financial Managers	2.2%	$213,724	$124,888	1.83	5.37
Accountants and Auditors	2.1%	$159,360	$77,974	2.14	3.07
Managers in Marketing; Advertising; and Public Relations	2.0%	$195,673	$114,978	1.82	5.39
Sales Representatives; Wholesale and Manufacturing	1.8%	$174,974	$89,548	2.10	3.72
Computer Scientists and Systems Analysts/Network systems Analysts/Web Developers	1.7%	$140,081	$82,509	1.77	3.32
Postsecondary Teachers	1.6%	$131,742	$73,866	1.91	2.79
Computer and Information Systems Managers	1.5%	$163,451	$113,890	1.48	7.37
Management Analysts	1.5%	$163,824	$83,300	2.07	4.74
General and Operations Managers	1.3%	$184,637	$100,472	1.94	4.14
Elementary and Middle School Teachers	1.2%	$77,988	$43,263	1.89	0.90
Education Administrators	1.1%	$126,318	$77,846	1.73	3.45
Engineers; nec	1.1%	$147,272	$99,014	1.57	5.42
Medical and Health Services Managers	1.0%	$170,332	$106,425	1.72	4.33
Personal Financial Advisors	1.0%	$227,120	$131,841	1.81	7.58
Real Estate Brokers and Sales Agents	0.9%	$155,844	$56,734	3.09	2.99
Retail Salespersons	0.9%	$139,723	$40,113	3.92	0.62
Sales Representatives; Services; All Other	0.9%	$201,768	$100,790	2.23	3.60
First-Line Supervisors of Office and Administrative Support Workers	0.9%	$128,771	$63,998	2.17	1.71

Table 4: Top Decile Households in 1980: Top 25 Most Common Occupations (Normal Wage Is Adjusted By Age, Education, Occupation, Sex and Race)

Occupation	Pct	Wage	Avg AEOSRwage	Wage/Avg Normal Wage	Pct/Universe Pct
Managers; nec (including Postmasters)	15.0%	$38,823	$24,699	1.72	5.40
First-Line Supervisors of Production and Operating Workers	3.2%	$30,789	$20,157	1.61	3.44
Physicians and Surgeons	2.6%	$45,162	$33,042	1.42	13.00
Sales Representatives; Wholesale and Manufacturing	2.5%	$36,453	$20,646	1.89	4.15
Managers in Marketing; Advertising; and Public Relations	2.4%	$40,084	$28,083	1.50	7.20
First-Line Supervisors of Sales Workers	2.2%	$30,173	$13,756	2.41	2.74
Lawyers; and judges; magistrates; and other judicial workers	2.1%	$31,109	$18,487	1.72	8.35
Driver/Sales Workers and Truck Drivers	2.1%	$27,547	$13,396	2.17	1.57
Elementary and Middle School Teachers	1.8%	$19,370	$13,400	1.57	1.42
First-Line Supervisors of Construction Trades and Extraction Workers	1.5%	$32,208	$17,258	1.94	3.85
Accountants and Auditors	1.5%	$27,807	$17,422	1.66	3.03
Real Estate Brokers and Sales Agents	1.4%	$29,616	$13,136	2.44	3.96
Farmers; Ranchers; and Other Agricultural Managers	1.3%	$14,724	$3,012	5.55	1.79
Postsecondary Teachers	1.2%	$29,778	$21,994	1.44	3.85
Insurance Sales Agents	1.2%	$35,378	$19,036	1.93	4.30
Secretaries and Administrative Assistants	1.2%	$15,332	$7,926	2.00	0.53
First-Line Supervisors of Office and Administrative Support Workers	1.1%	$28,129	$18,778	1.61	2.61
Financial Managers	1.1%	$36,864	$26,598	1.45	5.53
Education Administrators	1.0%	$29,677	$22,939	1.38	5.30
Electrical and Electronics Engineers	0.9%	$32,155	$25,750	1.29	6.14
Other production workers including semiconductor processors and cooling and freezing equipment operators	0.9%	$23,477	$12,184	2.11	0.80
Retail Salespersons	0.8%	$26,573	$11,362	2.61	1.11
Laborers and Freight; Stock; and Material Movers; Hand	0.8%	$21,541	$9,380	2.64	0.57
Bookkeeping; Accounting; and Auditing Clerks	0.8%	$15,584	$8,122	1.99	0.78
Registered Nurses	0.8%	$18,938	$11,163	1.76	1.19

"Babe" Ruth

Ability’s Effect On Income Inequality

Table 5 looks at average ability (defined as wage/AEOSR peer wage) of all wage earners in households grouped by household income. Unlike Tables 1 and 2, it is an average of household heads, spouses and others. This “blended” measure shows a monotonic decrease in “ability” by household income decile.

Table 5: Average Total Individual Wage Income Divided by Age, Education, Occupation, Sex and Race Average Income By Household Income Decile

Household Income Decile	1980	2015
1	0.28	0.24
2	0.55	0.47
3	0.72	0.61
4	0.81	0.70
5	0.89	0.77
6	0.94	0.82
7	1.01	0.88
8	1.07	0.94
9	1.14	1.01
10	1.58	1.40

The effect of these payoffs in 2015 on average wage income/earner by household income decile is shown in Figure 1. The reddish bars in Figure 1 show the actual average wage income/earner by household income decile in 2015. The aqua bars show what average wage income/earner would have been if there had been no differences in relative ability (i.e., each worker received their average AEOSR income). The aqua bars show that compositional differences between households due to age, education, occupation, sex and race are significant drivers of income inequality. However the reddish bars show how much those compositional differences are magnified by differences in ability.

Figure 1: 2015 Total Wage Income By Household Income Decile Versus “No Relative Ability” Mean Wage Income (Average By Age, Education, Occupation, Sex and Race)

Figure 2 shows the corresponding distribution in 1980.

Figure 2: 1980 Total Wage Income By Household Income Decile Versus “No Relative Ability” Mean Wage Income (Average By Age, Education, Occupation, Sex and Race)

In 1980, the “no relative ability” top decile wage income per earner was 1.8 times that of the bottom decile, while the actual top decile wage income per earner was 11.8 times the bottom decile. This difference demonstrates that ability plays a significant role in structural income inequality.

By 2015, the “no relative ability” top decile wage income/earner was 3.3 times that of the bottom decile. This increase in baseline (“no relative ability”) inequality reflects the effects of spousal income, the increased payoff to education and “skilled” employment, and the impact of global trade documented elsewhere.

If the increased ratio of “no ability” income is applied to the actual 1980 top to bottom wage income ratio (3.3/1.8 * 11.8), the expected actual top decile wage income/earner to bottom decile ratio in 2015 would be 21.6 times. That is slightly above the actual ratio of 19.6 in 2015, but it’s broadly consistent with the conclusion that underlying changes in education/skill payoffs since 1980 were amplified by a relatively stable payoff to ability shown in Tables 1 and 2.

Samuel Reshevsky - Child Chess Prodigy

Ability 

Ability is a very generic term. For a professional comedian, it might be an exceptional sense of humor. For an athlete it might include speed, or hand to eye coordination. For a salesperson, it might be an unusual ability to make people trust them. Innate ability, however, rarely leads to success by itself. It must be accompanied by hard work, discipline and motivation.  Ability in fact can be viewed partially as a pattern of behavior. People in the top income decile probably place a very high value on income and they are willing to work hard to achieve it, even if doing so crowds out other parts of their lives.

Certain traits related to success appear early. The Stanford “marshmallow” experiment seemed to demonstrate that traits in preschool children could be correlated with their later success. Jonathan Haidt in his book The Righteous Mind cites a study by David Perkins of student’s ability to generate arguments:

“Rather, the high school students who generate a lot of arguments are the ones who are more likely to go to college, and the college students who generate a lot of arguments are the ones who are more likely to go on to graduate school. Schools don’t teach people to reason thoroughly; they select the applicants with higher IQs, and people with higher IQs are able to generate more reasons.”[4]

IQ itself is not a good predictor of income. Dr. Freeman A. Hrabowski, III, President of the University of Maryland, Baltimore County, has stated that qualities like grit, determination and persistence are the ones he observes which make the difference between students’ success and failure.[5] But in general, the qualities college admission directors select for serve admitted students well in their subsequent careers, and would have served them well regardless of whether or not they went to college. Thus when looking at the apparent payoff to college education, perhaps almost half of the payoff may be attributable to the students inherent qualities, not their education.

While anyone’s genetic endowment can be improved, the mere presence of inherent ability is a significant driver of income inequality. The fact that ability exists is not an excuse for complacency about inequality, or for a lack of compassion. However, it does suggest there are limits to what can be achieved by social policies designed to eliminate income inequality.

Transparent and reproducible: All of the labeled Figures and Tables can be generated using the free, publicly-available R program and the R code available in “AbilityIncomeInequality.r” on github to analyze the publicly available data obtainable from the links in the article.

[1] Usain Bolt's net worth is estimated to be in excess of $60 MM.

[2] IPUMS-USA, University of Minnesota, www.ipums.org. The data is provided freely but is subject to their licensing restrictions.

[3] The “Wage/Avg. Normal Wage” column numbers are the observed mean of individual earner ratios (they cannot be calculated using the “Wage” and “Avg. AEOSR Wage” columns).

[4] Page 94 of The Righteous Mind: Why Good People Are Divided By Politics And Religion by Jonathan Haidt.

[5] At the Connecticut Forum on “The Future Of Higher Education” on December 1st, 2016.