Inequality: first, do no harm.

• Author: Geloso, Vincent
• Position: Report

One aspect of the debate surrounding inequality is how problematic inequality really is. Inequality per se is presumably not a problem; rather, inequality is bad because of the problems critics claim it produces. For example, numerous authors (e.g., Ostry, Berg, and Tsangarides 2014) claim that inequality negatively affects economic growth, a claim disputed by others (e.g., Winship 2013). Some scholars argue that inequality has negative externalities that degrade social capital and health indicators (e.g., Wilkinson and Pickett 2009). Here, too, this claim is disputed by other authors (most notably Kahneman and Deaton 2010).

Whatever the merits of the various positions, the participants in this debate have not made important distinctions among how individuals perceive different forms of inequality. For example, we might be more concerned about forms of inequality that prevent people from satisfying their preferences and less concerned about forms of inequality that result from people actually satisfying those preferences. Although some philosophers (e.g., Tomasi 2012) and economists (e.g., Welch 1999) have attempted to make such distinctions, we hope to decompose inequalities more carefully into those that are socially beneficial (or at least neutral) and those that are socially harmful, especially to the least well-off.

Socially beneficial inequalities (what we call "good" inequalities) result from the satisfaction of individual economic preferences or demographic changes and have no perverse impact on economic growth. We argue that using policy to attempt to reduce such inequalities would produce a great deal of positive harm because they are desirable unintended consequences of economic progress that also improve the well-being of the least well-off or are neutral changes resulting from changes in family size, demography, and marriage patterns. Because the results of these inequalities are either good or neutral, and because they are unintended consequences of individual choice, they should at least get a prima facie assumption of not being policy relevant. By contrast, what we call "socially harmful" or "bad" inequalities are problematic because they result from limiting individual choice in ways that expand inequality by limiting overall growth and harming the least well-off. In this way, our criteria of social desirability are broadly Rawlsian (Rawls 1971) in that one key concern is whether inequalities benefit the least well-off. Our argument also parallels that of Tomasi (2012) and other recent literature arguing that inequalities created in largely free markets should be held to the Rawlsian difference principle and that they can meet that test.

We start our analysis by reviewing the extent of the rise in inequality since the 1970s and argue that although inequality has increased, various problems with measurement indicate that the extent of the growth in inequality is overestimated. If overall inequality is actually less than believed, we should be even more hesitant to adopt costly policies that are claimed to reduce inequality. Next we point out that a substantial share of the increase of inequality is explained by "good" inequalities. Then we explore the "bad" inequalities and how they result from government interventions that push down the lower end of the income distribution while pulling up the higher end. Although there are inequalities of birth or family upbringing, we argue that they are much costlier to combat than inequalities resulting from misguided government intervention and thus are far less policy relevant. Rather than combatting inequality per se, we should be looking to address the sources of inequality that generate undesirable unintended consequences. More specifically, we should focus on inequality growth that results from limiting the options of the least well-off and thereby hampering their ability to move up the income ladder. That is, inequality policy should first attempt to do no harm by removing policies that exacerbate inequality by harming the poor and not by penalizing rising inequality that contributes to economic growth and improves the condition of the least well-off.

Measuring Inequality

Indicators of inequality generally show a consistent upward trend starting in the 1970s (Galbraith 2012; Piketty 2014). The increase seems consistent across the Western countries, even though inequalities worldwide have been decreasing (Sala-i-Martin 2006). The Organization for Economic Cooperation and Development (OECD) reported that inequality (measured by the Gini coefficient for after-tax income) increased by 24 percent from 1980 to 2008 in the United States (2011, 24). In fact, the same report shows that, with a few minor exceptions, inequality in all OECD countries has increased since the mid-1980s (23).

Yet those claims are plagued with measurement problems with regards to (a) the price indices used to deflate real incomes and (b) the measurement of income. The problem of prices is probably larger because it involves issues across both time and space such that we overestimate inflation and fail to account for regional price disparities. Let us start with the latter. Our point here, we must emphasize, is not to review the entire literature and arrive at a conclusion about the "actual level" of inequality. Rather, our contention is merely that the increase in inequality has been more modest than generally believed.

Economic theory suggests that incomes tend to equalize in real terms across regions as factors of production move around. Part of this equalization will occur in noninflation-adjusted wages, but another part will occur through price changes. For example, a greater population moving into New York City from Iowa to take advantage of higher urban wages will increase land prices in New York, and lowered demand in Iowa will reduce land prices there. The result will be a convergence of real wages. That said, one has to be very careful with regional price indices because of endogeneity issues between incomes and baskets of measured prices, including the way income determines the basket of goods demanded by consumers and thereby determines the basket that government agencies construct to measure the cost of living. However, it is still relevant to see how important price disparities are across the United States (see figure 1), as one can see with prices in New York being 15 percent higher than the national price level and prices in South Dakota being close to 13 percent lower than the national price level.

There are huge gaps between metropolitan and non-metropolitan areas within states. For example, a report from the Bureau of Economic Analysis using data from 2006 indicated that prices in metropolitan areas of New York and California stood at 35.9 percent and 29.8 percent higher than the national price level, whereas prices in nonmetropolitan areas in those same states stood at 20.8 percent and 9 percent lower than the national price level (Aten and D'Souza 2008, 67). These gaps tell us that aggregating individuals together might lead to overestimating the inequalities among Americans. This is because deflating wages and incomes from cities such as New York and San Francisco by a national average or even a state average means that we are deflating their income by a lesser measure than we ought to and deflating the income of nonmetropolitan Americans by a larger measure than we ought to. Given that a large share of inequality in the United States is driven by a few key areas (Galbraith 2012, 144) where prices are well higher than the national average, this point is crucial.

The argument about regional disparities in prices is an argument about the price level, not about price trends. However, there are also problems with how to estimate changes in the cost of living in the United States. Since the 1990s, we have been aware that the Consumer Price Index (CPI) overstates annual inflation by roughly 1.1 percentage points (Boskin 2005). Other authors have offered a wide range of other estimates of CPI biases, but all are positive. One bias is a substitution bias, which means that the CPI does not properly capture changes in relative prices that induce households to shift their consumption to different outlets or substitute goods. The other significant biases are ones linked to quality. Increases in prices linked with increases in quality should not be considered an increase in the cost of living, and treating them as such biases the CPI upward. Significant efforts have been made to estimate biases from quality and substitution and their impact on estimates of inequality (Broda and Weinstein 2008a, 2008b; Broda, Leibtag, and Weinstein 2009). According to this work, the use of the CPI for all urban consumers indicates no real wage growth for workers at...