In December 2011 the Organization for Economic Cooperation and Development (OECD) published a volume, Divided We Stand: Why Inequality Keeps Rising, that examined trends in inequality in the OECD countries in the years since 1980 (OECD, 2011). The OECD's analysis found that growing inequality was a common trend among the OECD countries. It attributed this growth in part to institutional changes that had the effect of raising wage inequality, most notably declining labour tax rates and weaker employment protection legislation. It noted increases in secondary education were an important factor reducing inequality, along with increased labour force participation among women. And it argued that technology was the dominant force leading to higher wage inequality over this period. While the volume contained much useful analysis and data, its analysis of the dimensions and causes of inequality is incomplete.
There were three main ways in which the OECD analysis was inadequate. First, it relied heavily on the change in the ratio of the earnings of workers at the 90th percentile to the earnings of workers at the 10th percentile. This only captures part of the story. Workers at the 10th percentile of the earnings distribution were indeed losers over the last three decades in many OECD countries; however workers at the 90th percentile were not really winners. In most countries their wages just kept even with average wage growth in the economy, meaning that they were not the beneficiaries of any upward redistribution. The upward redistribution went to individuals further up on the income ladder.
The second inadequacy was the brevity of the discussion on the financial sector. Many of those who receive the highest incomes are located in the financial sector. Recent research also shows that a bloated financial sector can be a drag on growth by pulling resources away from productive sectors of the economy (Cecchetti and Kharroubbi, 2012). It is likely that the financial sector played an important role in the rise in inequality over the last three decades.
Finally, the OECD analysis was too quick to claim that technology was a major cause of the rise in inequality over this period. It actually found that a measure of the improvement trend in technology was not associated with a rise in inequality. Only a cyclical measure of technology was correlated with inequality; and cyclical changes in spending on technology cannot explain a decades-long trend of rising inequality.
We examine each of these issues in more detail below.
Figure 1a shows the difference between wage growth at each decile cut-off and average wage growth in the economy for six OECD countries. As shown, the lower eight deciles of the wage distribution all saw wage increases that were less than the average, with the biggest losses for those near the bottom of the distribution. Clearly those at the middle and the bottom of the wage distribution were not getting their share of the gains from growth over this period. However, workers at the 90th percentile of the wage distribution certainly were not big winners. In Australia - where the 90th percentile grew fastest relative to the mean - that wage only rose 0.27 percentage points faster than the average rate of wage growth over this period. This means that the OECD's analysis focused largely on the losers in this story of redistribution. The winners were further up on the income distribution.
Figure 2 shows average income growth at the 90th, 95th, 99th, 99.5th, 99.9th and (except Australia) 99.99th percentiles in comparison to the average rate of income growth from the World Top Incomes database (note that this series includes non-wage income, so these numbers are not strictly comparable to those in Figure 1). As can be seen, the incomes for the very highest income groups substantially outpaced the average rate of income growth over this period, with the difference being greater the closer you get to the top of the income distribution.
This point is important. The redistribution did not go from the middle and bottom of the distribution to those who were merely better off workers. It went primarily to those who were at or near the very top of the income distribution. This fact hugely influences how we should think about this upward redistribution and possible remedies.
The second inadequacy of the OECD analysis, its exclusion of finance from the discussion, is also a significant issue. We know that the financial sector has expanded hugely as a share of the economy in many OECD countries, especially in countries like the United States and United Kingdom which have seen some of the largest rises in inequality. In our analysis of the OECD data we find a strong association between the share of financial compensation in GDP and the ratio of wages for workers at the 90th percentile to workers at the 10th percentile.
While this analysis is far from conclusive, there are good reasons for believing that the financial sector has been a major contributor to the growth of inequality over this period. First and most importantly many of the highest incomes originate in the financial sector (Philippon and Reshef, 2009). The most successful managers of hedge funds can earn hundreds of millions or even billions of dollars a year. Even less successful but experienced traders often earn paychecks that are well into the millions, fifty or a hundred times the pay of a typical worker. By contrast, in 2010 the wage of the 90th percentile earner in the United States stood at only 2.37 times that of the 50th.
The earnings from the financial sector must come from somewhere. If the financial sector led to a more productive economy, then these high earners may be generating wealth comparable to their compensation. A recent study by Stephan Cecchetti and Enisse Kharroubbi, two researchers at the Bank of International Settlements suggests that a large financial sector does not contribute to growth, but rather is a drain on the economy. The paper examined the growth of 50 countries over the period from 1980 to 2009. It found a U-shaped relationship between the size of the financial sector and economic growth. The implication is that an underdeveloped financial sector impedes economic growth presumably because economies are not effectively allocating resources between sectors. However after the financial sector reaches a certain size relative to the overall economy, further expansion of the sector slows economic growth.
The paper then sought to examine how the financial sector could have this negative effect. It looked at the rate of productivity growth in 15 manufacturing industries in 30 wealthy countries. It found that a larger financial sector was associated with slower productivity growth in industries with large amounts of research and development spending. This would be consistent with a story where the financial sector was responsible for pulling people with highly developed mathematics skills away from other sectors; people who could have been developing computers and clean energy, for example, are instead developing algorithms to be one up on the competition.
Their study also found that a larger financial sector is associated with slower productivity growth in industries that are heavily dependent on external capital. This would be consistent with a scenario in which a larger financial sector ties up more capital in financial speculation, thereby making it harder for new firms to raise the necessary funds for investment.
A fuller analysis of inequality has to examine the role of the financial sector more closely. There is good circumstantial evidence to suggest that it is a major culprit, but more work needs to be done before the case can be closed
The report's final inadequacy, the rapidity at which the OECD accepted the argument that technology is an important factor in the rise in inequality over this period, is an issue because the evidence suggests otherwise. On this point there is an extensive literature, largely focused on the United States, which tries to blame technology for the rise in inequality over the last three decades (e.g. Goldin and Katz, 2008; Autor et al., 2005).
While many of the world's most prominent labour economists argue that this has been the case, there are a number of basic facts about the pattern of wage inequality that call this conclusion into question (e.g. Goldin and Katz, 2008).
First, the sharpest rise in the gap between the wages of college educated and non-college educated workers was in the early 1980s. This was well before computers and other information age technologies were having any major role in transforming the workplace or increasing productivity. Second, inequality has continued to rise in the years since 2000 even though college educated workers did not see real wage gains over this period. By education level, it has only been workers with advanced degrees who have seen real wage gains in the years since 2000. If technology is the one of the main forces that explain wage inequality thean the workers who are beneficiaries are consistently changing and now seem to comprise a very narrow group.
In the case of the OECD data, it is very hard to come up with a story whereby technology provides much of the explanation. While the OECD's simulation finds that technology explains two-thirds of the rise in inequality over this period, it is difficult to reconcile this conclusion with either their analysis or that of the authors of this paper. The OECD uses a measure of spending on research and development as a share of GDP as a proxy for technology. They find a cyclical relationship between this variable and the ratio of wages for workers at the 90th percentile and workers at the 10th percentile. This seems quite plausible. At a cyclical spending peak, workers with substantial technical skills, who are likely to be near the 90th percentile of the wage distribution will be in short supply. Therefore their wages will be bid up.
However, the OECD and the authors of this paper find no relationship between spending on technology over the longer term and the ratio of wages for workers at the 90th percentile to wages for workers at the 10th percentile. This result can be readily explained by a rising trend in the number of workers with technical skills. There is no obvious reason that the supply of such workers should not keep pace with the demand.
With no trend relationship between technology and inequality as measured by the 90/10 ratio, we find that technology explains none of the rise in inequality over the last 30 years. Whatever impact the cyclical component might have in raising inequality during the upturn of a technology cycle is offset by the opposite impact it has in the downturn, leaving zero net effect.
The technology point is important because it focuses on whether the rise in inequality was a part of the natural development of the economyendogenous development of the market over the last three decades or whether it was due to policy changes that more directly affect distribution. If we assume that technology is the culprit, thean inequality is a natural development that we may as a matter of policy decide to alleviate to a greater or lesser extent.
Figure 3 shows our calculations of the factors contributing to inequality as measured by the ratio of the wages of the 90th percentile worker to the 10th percentile worker.
While we find no role for technology, like the OECD we find that education, as measured by the share of the workforce with a secondary degree, has a substantial positive effect in reducing inequality. This is readily explained by the fact that the more educated workers there are, the less their wages will rise relative to those of less educated workers. Like the OECD, we also find that a change in institutional structures has been a major factor contributing to inequality. Among these institutional factors, employment protection legislation appears to be an important factor in the reduction of inequality. Some regressions found that greater labour union density lowers inequality, as does a higher minimum wage, however these results are not robust, while in the latter case reliable data is only available for a small subset of countries (the results of these regressions are available from the authors on request). It also appears that a lower tax rate on labour income is a factor that has increased inequality. While there is more work to be done on how these sorts of institutional factors may affect inequality, the OECD's research suggests that changes in institutional factors played an important role in the rise in inequality over the last three decades.
This analysis still leaves the rise in inequality over this period largely unexplained. While the weakening of institutional support for workers has led to more inequality, this was almost exactly offset by an increase in the number of educated workers. The net effect of these two developments on the ratio of wages for workers at the 90th percentile to the wages of workers at the 10th percentile was close to zero.
We believe that the causes of the rise in inequality are likely to be found elsewhere, most probably in the factors that have led to the sharp rises in income for those in the top five percent, and especially the top one percent of the income distribution. As noted above, it is likely that the financial sector is a big part of the picture, as the high incomes of those in the financial sector must come from elsewhere in the economy. If will be important to determine more precisely the extent to which the growth of the financial sector has come at the expense of the wages of those at the middle and bottom of the wage distribution. In effect, excessive income for the sector can be seen as a tax imposed on the rest of the economy, reducing real incomes for workers outside of the sector.
There are also other factors that are likely to lead to large economic rents for those at the top of the income distribution. Recent research on the pay of top executives at U.S. corporations has found little relationship between pay and any standard measure of performance (Bebchuk and Fried, 2004). Top executives in the United States receive compensation packages that are far above those of top executives in other countries or what CEOs in the United States earned 30 or 40 years ago. Insofar as this higher pay is not tied to productivity, it must be coming at the expense of others. Further research is necessary to determine the extent to which excessive pay for top corporate executives has contributed to inequality.
Finally, stronger patent protection is likely to have played a role in increasing inequality
. Patent rents have taken up an increasing share of GDP over the last three decades. This is especially true in the United States where spending on prescription drugs alone accounts for close to two percent of GDP. The vast majority of this spending is due to patent monopolies, since most drugs would be available at little cost in a free market.
To sum up, while the OECD report advances our knowledge about trends in inequality in wealthy countries over the last three decades, it leaves much unexplained. A substantial portion of the rise in inequality is clearly attributable to government policies such as weakened employment protection legislation and reduced tax rates on labour income. However much of the rise is still unexplained. Further research is required to determine the extent to which the rise in inequality is due to the development of technology and other processes that are largely endogenous to the economy, as opposed to conscious policy interventions that have had the effect of redistributing income upward.
Unequal income growth
Revenue growth of the very rich
The causes of income inequality