Persistent Poverty

This week we will be focusing on explanations of persistent poverty.

What is “persistent” poverty?

The description of persistent poverty is used to distinguish long-term poverty spells from those that are more transient in nature. In fact, most Americans will experience a poverty spell in their lifetime, so it is important to understand the difference between a relatively short incidence of poverty, due to job loss, sickness or some other life incident that results in a short term loss of income compared to a longer-term incidence of poverty that is not as easily recoverable. That being said, poverty can also be described as chronic or long-term within an individual’s life and not necessarily be “persistent” if their children grow up to earn a higher income as adults. Generally when people talk about the “persistently poor” they imagine poverty that persists from generation to generation. Of course, chronic and persistent poverty are related when the chronic poverty of a parent leads to lower outcomes for their children.

Persistent poverty is one primary concern of those who study inequality. The fear is that high inequality may in effect cause persistent poverty if class stratification becomes so divergent that it becomes extraordinarily difficult to enter from the “underclass” to the “upperclass.” Whether inequality causes this type of immobility will be explored later as we get into studies of intergenerational mobility.

Who experiences persistent poverty?

Policy discussions of ending poverty often focus on poor children, who can hardly be blamed for being born into a situation that places them at a significant disadvantage. In the long-term Panel Study of Income Dynamics (PSID), just under 10% of children had family incomes below the official poverty line for more than half of the period between 1979 and 1994. (The Economic Costs of Poverty in the United States: Subsequent Effects of Children Growing Up Poor, 12)

A study by Mark Rank and Thomas Hirschel of The Likelihood of Poverty across the American Adult Life Span found that by age 65 more than half of adult Americans will have experienced at least a year of poverty, and that most spells last for only a few years. They also found that the probability and average length of poverty is significantly different by race:

“One quarter of white Americans experienced poverty at some point during the 13-year [study] period compared with two-thirds of black Americans. Furthermore, 67 percent of white Americans who experienced poverty were poor for three years or less, whereas the figure for black Americans was only 30 percent. Consequently, black Americans were more likely to be touched by poverty and more likely to be exposed to poverty for substantially longer periods” (Rank, Hirschel 202)

Even more striking, they found that by age 75, Black Americans had over a 90% chance of experiencing poverty, while for White Americans that probability was just over 50%. This extreme gap can be seen clearly in the graph below.


[IMAGE SOURCE: Rank and Hirschl, 211]

Unsurprisingly, another study by a researcher at the Census Bureau found that people who experience chronic poverty (defined in her study as people who experienced one or two spells of poverty totalling 90% of a four year period) are more likely to be “Black, Hispanic, under the age of 18 or over the age of 65, and without a college education.” They are also more likely to be female and live with a female head of household. In addition, approximately one third had a disability that limited work. (Edwards, 9)

Geography also plays a role in persistent poverty. The Economic Research Service at the US Department of Agriculture defines persistent poverty at the county level, with those counties who have 20% or higher poverty for 30 years as persistent poverty counties. The vast majority of these 353 counties are located in the south and 85% are nonmetro areas, as you can see on this map:


[IMAGE SOURCE: USDA, Economic Research Service]

Why does poverty persist?

This question is really the heart of the matter, because any effort to reduce poverty or its persistence requires answering “why” first. There are a number of theories posited to explain poverty, and recent research on neighborhoods and child brain development has added significantly to the knowledge base. Here I will outline a few of the major theories.

Cultural Explanations

Cultural explanations are popular for explaining poverty because people who are poor seem to operate differently than people who are not poor. While of course the experience of poverty will affect people’s behavior, this explanation sometimes stems from harmful stereotypes that poor people are lazy, entitled, and morally bankrupt.

In “Culture” and the Intergenerational Transmission of Poverty: The Prevention Paradox, Ludwig and Mayer explore the notion that a traditional family structure, parental work and religion could reduce poverty, an idea sometimes perpetuated by policymakers. They first remind readers than many adults who experience poverty were actually raised in this “ideal” type of situation that policymakers promote as the cure to poverty and emphasize the limitations of cultural change as policy. Then, even accepting this as truth, they find that “If eighth graders lived with their biological parents, at least one parent worked, and the child attended religious services, poverty in their generation would fall by 22 percent, assuming that the effects of these indicators of culture on children’s poverty status are entirely causal.” (187) Hardly a silver bullet.

There is merit however in explanations that revolve around the concept of social exclusion and networking gaps. In Understanding Persistent Poverty: Social Class Context in Rural Communities, Sociologist Cynthia Duncan explains how class cultural divides can cause challenges when people who grew up poor try entering the working world. She emphasizes research by others that demonstrate how “participation requires talking, thinking, and acting like those who manage America’s institutions,” and how when poor people do not have the same cultural ‘toolkit’ as those who grew up nonpoor, they find it more difficult to enter mainstream society and network for opportunities. (108)

Political Economy or Institutional Explanations

Institutional explanations are often seen as the opposite of cultural explanations, because while the impetus for change under cultural theories rests largely on the poor themselves, the impetus for change under institutional explanations rests on the state. Institutional explanations do not mean easy solutions however, as they include the likes of institutional racism and historical and political power differentials. Considering the racial and geographic dynamics described above, historical explanations are more than appropriate to consider.

Cynthia Duncan did research in rural Texas, Mississippi and Appalachia and illustrates in her paper how the political economy of these communities is structured in a way that maintains stratification:

“Accounts of rural life in the deep South describe rigidly stratified communities. The poor are completely dependent on the wealthy, who maintain their privilege through control over labor and most social institutions that affect the poor… Given this social structure, the poor have no opportunities to improve their status, and their limited prospects engender a kind of fatalism that eschews achievement and individual mobility. The ‘cultural tool kits’ they develop do not equip them for mobility, and everyday interaction reminds them of their place at the bottom.” (Understanding Persistent Poverty: Social Class Context in Rural Communities, 112)

Neighborhood or “Place” Explanations

Institutional and historical explanations also tie in with neighborhood explanations that have become more dominant recently thanks to work by Harvard economist Raj Chetty. (If you are interested, you can watch a recording of a presentation by him at a Center on Budget and Policy Priorities event here: The Role of Neighborhoods in Persistent Poverty).

The idea is that the place in which one grows up has a significant impact on future potential. Presumably, areas with higher densities of poverty have less social capital and opportunities than areas with lower poverty.

An experiment called Moving to Opportunity in the US studied the effect of neighborhoods by moving poor people out of high poverty neighborhoods and into low poverty neighborhoods using randomized housing vouchers. It was only fairly recently that the longer term impacts of the study were evaluated by Chetty and others, who found that those who moved to a low poverty area before 13 years of age earned on average $3,477 more per year in their mid-twenties compared to those who did not. (The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment, 1)

“We estimate that moving a child out of public housing to a low-poverty area when young (at age 8 on average) using an MTO-type experimental voucher will increase the child’s total lifetime earnings by about $302,000.” (Chetty et al, 5)

In Understanding Persistent Poverty: Social Class Context in Rural Communities, Cynthia Duncan describes how a community with a large middle class may be less stratified and so more inclusive, with better opportunities for networking and increased social capital. However, the MTO experiment found negative effects on those who moved after the age of 13 and no effect on the adults who moved. This indicates that neighborhood has significant effects on development, which brings us to the final explanation.

Developmental Explanations

How poverty affects the brain has been a sensitive research subject for many years and is conducted delicately so as not to cause policymakers to get the wrong idea about the correlation between poverty and cognition. However, there is new evidence to suggest that poverty has a significant impact on brain development in early childhood, especially among children under 5 years of age. (The Long Reach of Early Childhood Poverty, 25).

In The Long Reach of Early Childhood Poverty, Greg Duncan and Katherine Magnuson outline the long term effects of child poverty, including the completion of schooling, income and work hours, while also highlighting that poverty during adolescence may also have future effects on health and birth outside of marriage. To illustrate these effects, the authors presented the following charts:


[SOURCE: Duncan and Magnuson, 27]

Looking Forward

Recent research has led to some major breakthroughs, especially about the importance of targeting poverty assistance at young children. The Moving to Opportunity experiment has demonstrated positive neighborhood effects on young children while scientists have discovered a link between brain development and poverty that has similar implications.

The persistence of poverty is an issue that we will continue to explore from the angle of intergenerational mobility. Knowing that many Americans experience poverty firsthand is important when considering downward mobility, while understanding some of the causes of intergenerational and persistent poverty will help explain some of the immobility at the bottom.

Next week we will explore the top one percent as a contrast to the description of persistent poverty, before we begin a four week segment on intergenerational mobility.

As always, you can leave me a message through the Contact Me page, and you can find the reading list here:inequality-social-mobility-and-public-policy-readings-schedule

Inequality: Deunionization

“From 1973 to 2007, private sector union membership in the United States declined from 34 to 8 percent for men and from 16 to 6 percent for women. During this period, inequality in hourly wages increased by over 40 percent.” (Western and Rosenfield, Unions, Norms, and the Rise in U.S. Wage Inequality, 513)

This week we focus on the impact of deunionization on wage inequality, complementing the trends in skill premiums and job polarization described last week. 

Unions increase wages. That much is clear from the literature, with the union wage impact estimated to range between 10-20%. However, the impact is not the same everywhere, for everyone, in every industry. So before we begin analyzing the impact of deunionization on wage inequality, I thought I would lay out what researchers have found about the union effect:

  1. The union effect generally falls during periods of inflation and increases during recessions.
  2. Unionization raises wages the most for young people and those with little tenure.
  3. Unionization has a much bigger impact on blue collar jobs (19%) than white collar (4%).
  4. The union impact varies based on the size of the company and the percent unionized, with smaller companies and greater proportions increasing the union effect.

(Freeman, Medoff Chapter 3)

Spillover Effects

Unions are generally a force for equality both in the workplace and larger community. In addition to quantitative analyses of the union impact, researchers describe something of a “moral economy” that is promoted by unions, which spills over into the economy by changing norms and expectations of wage equity. (Western, Rosenfield 514)

For example, companies regularly use wage surveys to estimate competition in the region and adjust accordingly, raising wages for non-unionized firms in proximity to unionized firms.  Freeman and Medoff also find that  approximately 20% of unionized firms pay their non-union employees in line with union agreements, and that workplace rules tend to spill over, evidenced by the erosion of merit pay and emphasis on seniority in layoff decisions. Finally, they found that within companies where blue-collar workers are unionized, white-collar employees benefit from improvements in fringe benefits, though their wages are seemingly unaffected (Freeman, Medoff Chapter 10)

Impact on Inequality

Researchers have long been concerned that social benefits of the union effect are offset by increased unemployment and depressed wages among non-unionized employees, resulting in a net increase in inequality by creating a gap between covered and non-covered employees.

Richard Freeman and James L Medoff consider this at length in the book “What Do Unions Do?” and discover that although unions do raise wages at the expense of others, there are three other benefits that offset it in favor of reducing inequality:

“…the increase in inequality induced by monopoly wage effects is dwarfed by three other trade union effects on wages that reduce inequality: union wage policies lower inequality of wages within establishments; union wage policies favor equal pay for equal work across establishments; and union wage gains for blue-collar labor reduce inequality between white-collar and blue-collar workers.” (Freeman, Medoff 78)

Freeman and Medoff’s book was published in 1984, so while it provides a good description of the benefits of unionization, it predates much of the significant decline in union membership described in the opening quote.

According to Nicole Fortin and Thomas Lemieux in Institutional Changes and Rising Wage Inequality: Is There a Linkage?, unionization declined slowly in the 60s and 70s and fell especially fast in the 80s (at the same time wage inequality increased) before slowing again in the 90s. (77) They found that deunionization has had a significant impact on wage inequality for men, but not women.  This is because unionization had a stronger equalizing effect for men, while most women who were unionized were relatively skilled, tempering the equalizing effect. (80)

The equalizing effect of unions can be seen in the following graphs, where the distribution of union wages are more concentrated around the average while nonunion wages are more spread out across the range. It is also clear when moving from 1979 to 1988, that as the proportion of union wages decreases, the overall spread in wages increases (though for women the minimum wage has a clearly more dominant role):


[Source:  Institutional Changes and Rising Wage Inequality: Is There a Linkage?, 85]

Fortin and Lemiuex conclude that deunionization had no effect on women but that for men, “the variance in log wages would have increased by 21.3 percent less if the rate of unionization had remained at its 1979 level.” (89)

In contrast, In Unions, Norms, and the Rise in U.S. Wage Inequality, Western and Rosenfield expanded the analysis of deunionization on inequality to include the spillover effect of unions on nonunion wages. After adjusting for these spillover effects, they conclude that deunionization can explain 20% of the increase in inequality among women and 33% for men.

Western and Rosenfield also find that most of the increase in inequality in the 90s stems from increases in “within-group” inequality. In other words, the variance of wages within groups of individuals with similar characteristics increased over this time period more than the variance of wages between groups of individuals with different characteristics. Similarly, they find that deunionization impacted mostly within-group inequality.

Impact on Working Poverty

Having established a link between deunionization and increased wage inequality, David Brady, Regina S. Baker and Ryan Finnigan explore the impact of deunionization on working poverty in When Unionization Disappears: State-Level Unionization and Working Poverty in the United States. Working poverty is a description used for officially poor households who have at least one worker (10.4% of families in 2004). (873) While the authors recognize that most union workers are earning above median wages and thus that most working poor are not unionized, they determine that the spillover effects have an impact on low wage workers in a way that does impact working poverty.

The broad relationship between unionization and working poverty can be seen in the following graph. States that have higher rates of unionization have fewer working poor as a percent of their population compared to states with low rates of unionization.


[Source: When Unionization Disappears: State-Level Unionization and Working Poverty in the United States,880]

Using a two-way fixed effect logit model of working poverty, the authors find that increases in state-level unionization decreases working poverty significantly, outweighing even the economic performance or social policies of the state. More specifically they find that “for a standard deviation increase in state-level unionization, relative working poverty in non-union households is expected to decline by a factor of 1.3” (888)

The authors also find that at the individual household level, the characteristics of education, part-time employment and having multiple earners still remain the most important predictors of poverty. However, state-level unionization arguably rival other individual-level characteristics including race or being a single mother. (885) This result has significant implications for future poverty research.

Looking Forward

There is some concern that unions have lost their bite as membership has declined. In Economic Impacts of New Unionization on Private Sector Employers: 1984-2001, John DiNardo and David S. Lee use a regression discontinuity approach to compare manufacturing plant wages that became unionized after barely winning an election to manufacturing plant wages where unions barely lost. Among these newly created unions, they found no statistically significant difference in wages, concluding that new unions have not been successful at increasing average wages. (DiNardo,Lee 1431)

The research described herein indicate that unionization, especially when incorporating spillover effects, may explain some of the increase in wage inequality over the 80s and 90s. It also may help explain differing rates of working poverty in the US. How this research will fit into policy decisions in the future is yet to be seen.

Next week we will be shifting our focus a bit. While the first four weeks have been focused on trends in inequality, we will be focusing on persistent poverty next week, followed by a focus on the top one percent. At that point we will shift into an analysis of intergenerational mobility.

As always, feel free to send me a message through the contact page. The reading list for each week can be found here: inequality-social-mobility-and-public-policy-readings-schedule

Inequality: Wages

“What primarily characterizes the United States at the moment is a record level of inequality of income from labor (probably higher than in any other society at any time in the past, anywhere in the world, including societies in which skill disparities were extremely large).” Piketty 2014, 265

This week we explore explanations for wage income inequality in the United States. Next week we will explore the role of deunionization in more detail. 

So far we have discussed inequality with a clear underlying assumption that inequality is bad, and equality is good. But this is not exactly the case, and in fact there is significant value to inequality as an incentive to work hard, build new skills, and increase educational attainment. In the United States especially, where higher education is often financed, education would be a net drain on lifetime wealth in the absence of inequality, or the expectation that education will pay off in the long run.

The real worry then, is two-fold: first that individuals at the bottom are struggling, which causes a number of problems for society including increased crime, and second that mobility up and down the economic ladder slows down or halts because those born in families at the bottom of the ladder can’t afford to or do not have the necessary knowledge to invest in skills and education, and those born in families at the top are significantly and perhaps unfairly advantaged (family income and parent educational attainment being the biggest predictive factors in a child’s future).

We already know from week one that income inequality in the United States has increased over the last 30 years, to the point that Piketty made the bold claim (quoted at the beginning of this post) in his book, Capital in the 21st Century. But what are the reasons behind this increase? And what are the implications?

A number of mechanisms have been suggested by economists, and I use the word suggested because it is notoriously difficult to establish causal relationships in public policy issues this large, especially when dealing with limited data. Usually the best we can do is explore trends, establish correlations and theorize based on the picture we find. Some of the main theories include:

Increased Education/Skill Premium

One explanation of increased wage inequality is that the premium for high-skilled jobs has increased relative to low-skilled jobs. In other words, the value that the market places on jobs that require higher education has increased over time at the same time the value that the market places on jobs that require no education has decreased, slowly but surely widening the gap.

There is evidence of this phenomenon. The earnings gap between those with a high school education and those with a college education has increased considerably over the past thirty plus years. In dollar terms, the earnings gap in 1979 was $17,411 between the median high school educated and college-educated individual. In 2012, it was twice that at $34, 969. (Autor, 844) This trend has been worse for men than women, with male high school dropouts seeing mostly negative changes in real wage levels since 1990:


[Source: Skills, Education and the Rise of Earnings Inequality among the “other 99 percent”, 849)

This wage premium seems to explain about 50-70 percent of the increase in wage inequality over this time period, based on a number of studies. The most robust explanation for the wage premium has to do with supply and demand, which we explored a bit last week. (Autor)

A subset of this theory involves the concept of “hypermeritocracy” and the rise of supersalaries. The examination above does not explain the extreme increase in salaries among the top 1%, but it is worth mentioning because the top percentile has collected over 10% of total US labor income since the late 90s (but only 5% in 1970):


[Source: Capital in the 21st Century, 292]

It is also worth mentioning because although many place their focus on the extreme incomes of superstars like actors and NFL players, most of these increases can be traced back to the very high salaries for managers and top executives (who make up about two-thirds of individuals in top 0.1 percent).  (Piketty 264-66, 298-303)

Lifecycle trends

Another explanation of wage inequality is that of lifecycle trends. The lifecycle theory of inequality makes a lot of sense, and provides another reminder that inequality is not necessarily a bad thing. As a natural and desired part of a lifecyle, income increases as an individual enters the working world, moves up, earns raises, saves, and then declines as the individual enters retirement and begins living off savings.

While this does explain some of the baseline inequality, it does not do a good job of explaining the increase in inequality, and in fact there is evidence that “the concentration of wealth is actually nearly as great within each age cohort as it is for the population as a whole,” which essentially eliminates lifecycle trends as a causal mechanism (Piketty, 244). In addition, for the lifecycle trend to explain increased inequality it would also mean that an individual was highly mobile: moving into jobs with significantly higher skill and pay levels at their peak. And this simply isn’t the case for the vast majority of workers. (Piketty, 299)

Job polarization

The job polarization theory boils down to this:

“The structure of job opportunities in the United States has sharply polarized over the past two decades, with expanding job opportunities in both high skill, high wage occupations and low skill, low-wage occupations, coupled with contracting opportunities in middle-wage, middle-skill white-collar and blue-collar jobs.”
(The Polarization of Job Opportunities in the U.S. Labor Market, Autor 1)

If true, the relative polarization of job opportunities, combined with the increase in the education/skill wage premium and the reduction in real value of the minimum wage, would explain the increase in wage inequality. Job polarization can be seen in the following chart, where ‘middle skilled’ jobs, classified here as office and administrative jobs, production and craft, and operators/laborers, have seen reduced employment since 1979, while jobs in high and low skilled jobs have seen increases in employment:


[Source: The Polarization of Job Opportunities in the U.S. Labor Market, Autor 8]

Interestingly, there are significant differences by gender. Decreases among men in middle skilled positions shifted relatively equally to high and low skilled positions, whereas women shifted largely to higher-skilled positions. [This is partly a result of education trends]:


[Source: The Polarization of Job Opportunities in the U.S. Labor Market, Autor 9]

Some of these occupation shifts can be explained by increases in the proportion of Americans with higher education generally, especially among women. Other explanations include technology substitution of routine-task jobs and the relative increase in demand of high-skilled jobs that use skill-biased technology as discussed last week. (Autor)

Another explanation that is frequently cited, and rose to the fore last election cycle, is the hollowing out of middle-income jobs due to increased international trade and offshoring. However, Piketty makes the point that this is not a very robust explanation. In fact, “the United States’ internal imbalances are four times larger than its global [trade] imbalances. This suggests that the place to look for the solutions of certain problems may be more within the United States than in China or other Countries.” (Piketty 297-8) That said, for a subset of individuals who were directly affect by trade, this explanation holds significant weight. It is also important to consider the fact that offshoring and trade has been largely made possible by improvements in technology.

The job polarization theory is convincing, but as with all theories there are critics. In Assessing the Job Polarization Explanation of Growing Wage Inequalityauthors Mishel, Schmitt and Shierholz point out that the contraction of middle income jobs has a trend that begins prior to the increase in wage inequality, which eliminates the possibility of a cause-effect mechanism. They also point out occupational explanations do little to explain wage trends in the 2000s, and generally dismiss it as too narrow of an explanation that does not include the full picture.

Devalued Minimum Wage and Deunionization

In addition to the explanations above, generally noted in these studies is the reduced value of minimum wage and deunionization, but as I will be exploring those in more depth later, for now just know that they are considered to play into these trends, at least for the middle and lower end of the skill/income range.

Clearly this is a complicated issue that must take a wide scope of analysis to unravel, but it is helpful to know some of the mechanisms behind these changes in order to develop effective policy solutions. There is plenty of room for future research.

Next week we will explore the role of unions in the US and how they may or may not explain trends in inequality. The reading list can be found here: inequality-social-mobility-and-public-policy-readings-schedule.

As always, feel free to leave comments or send a message through the Contact page.


Inequality: Technology and Capital

This post reviews the role of capital accumulation in inequality and how technology, as both a substitute and supplement to labor, has had a noteworthy effect on wage trends and inequality. 

Thomas Piketty’s book, Capital in the Twenty-First Century, brought rising inequality to the attention of the media and policymakers and contributed much to the discussion, in addition to drawing a lot of criticism. His book discusses the increasing role of capital in inequality and warns that it will only get worse without significant intervention.

Piketty defines capital as “the sum total of nonhuman assets that can be owned and exchanged on some market.”(48) This excludes human capital but includes what he calls “immaterial” capital, or what I would call identifiable intangibles, including assets like patents and the value of brand and reputation that is wrapped up in stock prices. (49)

Piketty makes a number of points about trends in capital including the following:

  1. The makeup of capital over the long term has shifted largely from being land-based in the eighteenth century, to now mostly financial and industrial capital. (42)
  2. While capital can be owned publicly or privately, most capital is private. (46)
  3. Globalization has increased cross-country ownership, such that many countries earn income from capital located in other countries. However, this point shouldn’t be too alarming because most countries end up having a national income within 1% of domestic output. (44)
    1. Note: This is not to say that global inequality does not exist. On the contrary, domestic output and per capita GDP is highly unequal, though it has been converging since the mid twentieth century. (60) Moreover, “the global income distribution is more unequal than the output distribution, because the countries with the highest per capita output are also more likely to own part of the capital of other countries with a lower level of per capita output.” (67-68)

The first chapter of his book is most important because it is where he lays out “The First Fundamental Law of Capitalism” which forms the basis for his main argument (52). He explains here that capital’s share of national income (as measured annually) is a function of the capital/income ratio (national capital stock divided by national income) times the rate of return on capital.

α = r x β

α = capital’s share of national income
r = the rate of return on capital
β = the capital/income ratio (52)

This relationship matters for inequality because, holding all else equal, as the capital/income ratio increases (as trends suggest will occur), capital’s share of national income increases. And because capital income is a greater share of individual income at the top, that accumulation of wealth increases inequality. Piketty’s major premise in his book is that inequality will increase as long as the rate of return on capital is greater than the rate of income growth: r > g. This assumption that the rate of return on capital will exceed growth for a sustained period of time is the main point of contention among critics. For one exploration, see Is Piketty’s “Second Law of Capitalism” Fundamental?.

In After Piketty?, economist A.B. Atkinson responds to Piketty’s book by considering a number of policies that might work to reduce inequality. I will only highlight two here, but he emphasizes that responses to rising inequality should come from many policy areas, not just those related to tax and spending.

“Inequality is as much a matter for the Minister responsible for science as for the Minister responsible for social protection.” (620) 

Atkinson discusses in part how technology can act as both a substitute for labor and a supplement to labor. He explains that as capital increases relative to labor, wages rise and the rate of return on capital decreases. But eventually, capital (technology) can reach a point of substitution, whereby capital income increases but labor does not benefit. And as discussed above, that impacts inequality because capital income is concentrated at the top, while middle-income folks depend primarily on wage income.

Atkinson argues however that the role of technology is not out of our control, and that investments in technology that increase the productivity of workers could be encouraged by government to balance investments in technology that act to replace labor. At the same time, he argues that bargaining power between capital and labor need also improve, a topic that we will be covering in two weeks.

Unlike Atkinson, in Separating Efficiency and Equality, Automation and Piketty’s Theory of Increasing Capital Share, Yew-Kwang Ng argues for more general tax and spend policies rather than issue specific policies. He also tempers the argument that increased technology will exacerbate inequality. He reminds us that there was a great fear of mass unemployment in the 1970s due to the rise of the personal computer, a fear that was not realized. He argues that to the extent technology complements labor and raises output, it increases the demand for labor, preventing massive unemployment. The problem, he conjectures, will be more with a lack of skilled labor than a lack of jobs.

Which brings us to the next reading on the list, The Race Between Education and Technology. [Full disclosure: I couldn’t access the book in time for this blog so I settled with reading this 2007 working paper version].  The paper discusses how the relative supply of educated individuals serves to affect the wage premium of education, and at the same time how technology is skill-biased.

“The two most important forces in the framework concern the change in the relative supply of more-educated workers, which has mainly occurred through changes in schooling, and the changes in the relative demand of more-educated workers, which has been driven by skill-biased technological change.” (6)

The authors explain that in the late twentieth century much of the increase in wage inequality can be explained by the increase in the education wage premium (an increase of 23% between 1980 and 2005), but that this was mainly a result of slower growth in the supply of college workers since 1980. At the same time, technology advances tend to be skill-biased, such that they increase the relative demand for more skilled workers over the unskilled. In other words, demand for skilled workers exceeded its supply over this time period, increasing the wage premium of education and subsequent inequality. “The race had been lost to technology.” (28)

So, what do we know?

  1. The relative growth of capital in a country, and the rate of return to that capital, matter when considering inequality. This is because capital income is a greater share of individual income at the top.
  2. There is a great deal of criticism around Piketty’s assertion that r > g .
  3. Technology can act as a substitute and supplement to labor. How technology impacts capital income inequality will depend on the elasticity of substitution and how technology serves to complement labor.
  4. The way in which technology impacts wage inequality depends not only on the substitution/supplement aspect, but also on the relative supply of skilled labor for skill-biased technology.
  5. In general, the main takeaway for me is that capital accumulation, increases in technology, and the relative skilled labor supply all serve to affect vertical inequality, as those who are more privileged (educated, skilled, capital income-earning) reap most of the benefits of these trends, while those less privileged will be left behind without intervention by policymakers.

Next week, we will be focusing specifically on wage inequality. The associated readings can be found linked to my first blog post.

Inequality: Trends Over Time

This first week’s readings explore the structure and nature of inequality in the US over time, and its persistence.

Inequality in the US has been increasing since the early 1970s, but what was the trend before that time? In 2003, economists Piketty and Saez published a study of  the long-term trends in inequality in the United States using tax return data. Their analysis concluded that top income shares (one measure of inequality) dropped sharply during the periods of WWI and WWII, and then largely stagnated until the early 70s when top income shares began to rise again, especially for the top 1 percent.


[Image from Piketty and Saez (2003) page 12]

Also interesting is their discovery that the make up of top incomes has seen a long-term shift from favoring capital to favoring wage income. (The authors note that this decreased share of capital income is a result of decreased concentration, not a relative loss of capital income shares in the larger US economy) This shift represents partially changes to the tax code over time including trends in progressive taxation, the 1986 Tax Reform Act and an increase in retirement dividends that are not reported as such. It also maps with the development of stock options and the increase in average CEO compensation since the 1970s. However, more recent increases in capital income has resulted in some researchers to criticize their conclusion in a permanent shift in income compensation premature.

Piketty and Saez later joined with Anthony Atkinson in 2011 to conduct an analysis of top incomes in twenty-two different countries to compare trends across the globe. They found that most countries experienced a drop in top income shares over the first half of the twentieth century during the wars, which then stagnated over a period similar to the United States. However, they find that the last thirty some years have seen a divergence across countries, with top income shares growing in the US, UK, Canada, Australia, Ireland, New Zealand, China and India, while not growing in France, Netherlands, Japan, Germany or Switzerland. They also conclude recent increases in other countries are also due to increases in top wage incomes.


[Images from Piketty and Saez (2011) pages 41-42]

A variety of explanations for this divergence were considered by the authors as avenues for future research, including differences in progressive taxation, macroeconomic shifts and political changes.

In context of the broader picture, the Congressional Budget Office in 2011 published an analysis of Trends in the Distribution of Household Income Between 1979 and 2007, in which they found similar trends as Piketty and Saez in increased inequality and increased concentration of labor income. The CBO report also examined the impact of taxes and transfers on inequality. They found that while government taxes and transfers do reduce inequality, “the equalizing effect of transfers and taxes on household income was smaller in 2007 than it had been in 1979.” In other words, while government policies consistently reduced inequality over this time period, those policies weren’t as effective at reducing inequality in 2007 as they were in 1979.


[Images from CBO Report (2011) page 20]

So what does this mean in real terms? Using the Gini index as a measure of inequality, CBO calculated that in 1979, government taxes and transfers reduced the average difference in income between pairs of households from a difference of $34,000 to $22,600. In 2007, the average pre-tax and transfer difference of $66,000 was only reduced to $48,900:


The report highlights two main explanations for the reduced redistributive power of government policies. First, transfers to the elderly population through Social Security and Medicare have increased significantly, which are not limited to low-income households. Second, less-progressive payroll tax rates increased while more-progressive income tax rates decreased. In future weeks I will explore the role of taxes and transfers on inequality in more detail, but for now it is enough to highlight the major trends.

There are a variety of challenges to the evidence provided by tax data, namely that it does not allow for the identification of important factors like race and gender roles, does not include the entire population, and the size of each tax unit has decreased over time. The authors acknowledge and address these concerns in more detail than I will explore here. However, analyses using other sources of data including census survey data have identified similar trends, if not to the same degree. [See Atkinson, Piketty and Saez (2011) pages 29-34 for more details.]

Another question that regularly arises is the role of income mobility in larger trends. The tax unit data points (households) within each income bracket in 1979 are not the exact same as those in 2007, for example. People may go up and down income percentile groups as a result of increased education or skills, sudden illness, household structure, and lifecycle trends. However, Debacker et. al in Rising Inequality: Transitory or Persistent? use microlevel tax data in an attempt to separate the transitory and persistent components:

Persistent: Differences in permanent characteristics such as education and unobserved ability, as well as long-term shifts as a result of “chronic illness or the permanent loss of a high-paying job.”

Transitory: “temporary illness or transitory unemployment”

The authors conclude that the most of the changes to inequality can be explained by the persistent components. This has implications for future policy and research, suggesting that macro changes to the economy such as technological increases and cultural shifts in compensation policy may better explain trends in inequality than income mobility explanations.

So, what do we know?

  1. Income inequality dropped sharply over the first half of the 20th century before rebounding in the last 30 some years.
  2. Increases in the top 1 percent are especially high.
  3. This more recent increase in inequality in the US shares trends with some developed countries across the globe, but not all, complicating global macroeconomic explanations.
  4. Increases in wage income at the top (rather than capital income) explains the majority of the increase in inequality
  5. US tax and transfer policy reduces inequality but did not do as well in 2007 as 1979
  6. Most of these changes are persistent in nature, meaning  big picture trends should explain the increase in inequality more than individual mobility across income brackets.

Next week I’ll be exploring the role of technology and capital in inequality. Feel free to leave comments or contact me with questions through the Contact page.

You can find the reading list I will be following throughout this series linked to my first blog post.

Why Blog?

This blog is the main deliverable for a self-designed independent study at Indiana University School of Public and Environmental Affairs on inequality, social mobility and public policy. The reading list for each week was determined based on common readings and authors found in syllabi for related courses at some of the most prestigious schools in the country as well as recommendations by the sponsoring professor, Dr. Maureen Pirog.

Each week I will blog on the weekly topic as outlined in the reading list, available here: inequality-social-mobility-and-public-policy-readings-schedule

You are welcome to read and learn along with me.