A growing literature on income inequality and labor earnings has overlooked the contribution of disparities in hours worked. In new research with Lara Vivian, Daniele Checchi and Cecilia García-Peñalosa find that divergences in hours worked between low- and high-earning workers can explain a significant amount of the overall increase in income inequality in some high-income countries.

Over the past few decades, a vast literature has documented changing patterns in the distribution of labor earnings in high-income countries. The timing and the extent varies across countries, with some, such as the US and the UK, experiencing an increase in inequality in the 1980s before stabilizing. Others, such as Germany, have witnessed increased inequality only in recent years. These differences in levels and trends can be seen in Figure 1, where we depict the dynamics of an inequality index for labor earnings in four large economies. 

Figure 1

Figure 1 depicts the evolution of inequality in weekly labor earnings in four large industrial economies over the period 1995-2016. Inequality is measured by the mean log deviation (MLD), a statistical indicator that is nil in case of egalitarian distribution and is increasing in the dispersion, especially in the bottom part of the distribution.

Changes in the dispersion of hourly wages, that is, the pay received by an individual per hour of work, have been the major explanation for these differences in national earnings inequality levels. For example, inequality in hourly wages has been shown to be considerably larger in the US than in other countries and to have increased markedly in Germany in recent decades, in line with the differences observed in overall earnings inequality (Figure 1).

However, earnings are the product of hourly wages and hours of work. Yet, the behavior of the latter has received little attention in the literature. In a recent article co-authored with Lara Vivian, we explore the role that working time plays in explaining differences in earnings inequality across countries and over time. If all individuals in employment worked the same number of hours per week, then differences in earnings would be due exclusively to the dispersion of hourly wages. However, when this is not the case, understanding the dynamics of earnings requires examining differences in hours worked.

Figure 2 reports inequality in weekly hours worked in the four countries we consider in our study. The data indicate that hours inequality is low in France and the US and high in Germany and the UK. Moreover, while the UK has witnessed a decline in the dispersion of hours, Germany has seen it increase over our period of study (1995-2016). By 2016, our inequality measure for hours worked was twice as high in Germany and the UK as in France and the US.

Figure 2

Decomposing earnings inequality

Does more inequality in working time imply a more dispersed distribution of earnings? Not necessarily. The reason for this is that the impact of hours inequality depends on who works more. Consider a high-paid worker whose wage is twice as high as that of a low-paid worker. If both are employed the same number of hours, the earnings of the former will be twice as high as those of the latter. Now, suppose that the low-paid individual works twice as many hours as the high-paid one: hours inequality is now higher, yet the two workers will earn the same amount, implying no earnings inequality. That is, whether differences in hours worked tend to increase or decrease earnings inequality depends on the correlation between hours and wages. If those with lower hourly wages tend to work more, that is if the correlation is negative, then hours inequality will be an equalizing force. If the correlation is positive, they will be an unequalizing force that will tend to reinforce the dispersion of hourly wages and total earnings inequality.

To understand the importance of hours worked for earnings inequality, we perform a decomposition that allows us to compute which fraction of overall earnings dispersion is due to wages and which to hours, with the latter incorporating both the dispersion of hours and their correlation with wages. Figure 3 reports, for each country, the fraction of earnings inequality that is due to the dispersion of hourly wages (in blue) and that which is due to hours (in orange) for the first and last years in our sample. It indicates important differences across the four economies. We find that in the US and France, the overall contribution of hours in 2016 is moderate, with hours worked accounting for at most 40% of inequality in earnings. In contrast, hours play a more crucial role in the UK and in Germany at the end of the period, being responsible for almost half of the dispersion in earnings.

Figure 3

When we consider changes to the contribution of hours inequality to total earnings inequality over time, we observe considerable stability in the US and the UK. In contrast, the contribution of hours rose in France and Germany, increasing from 25 and 28% in 1995 to 40 and 48% in 2016, respectively. In France, the increase in hours inequality was offset by a decline in wage dispersion, leaving overall earnings inequality roughly unchanged. In Germany, hours were the main culprit for the 50% increase in the earnings inequality index that we observe in Figure 1.

The contribution of hours to earnings inequality rose in the continental economies, both because hours of work became more dispersed and because in both countries the wage-hours correlation went from being negative to being positive. That is, while in the 1990s, those with the lowest wages worked the hardest, thus creating an equalizing force, in recent years it has been those at the top of the distribution that worked the most. 

To illustrate this, we partition the data into five groups according to the individual’s position in the distribution of hourly wages. Figure 4 reports the average hours worked by individuals in each of these groups. In the US, we observe a (broadly) stable positive correlation, with those in the top wage quintiles working more than those in the bottom quintile. The case of Germany is striking. While in the mid-1990s those with low wages worked slightly more than those with high wages, over the next two decades this pattern reversed. The working time of those at the top increased and that of those in the two bottom quintiles fell. The result was not only greater dispersion of working time but also a correlation that went from being negative (and thus equalizing) to being positive (and therefore unequalizing).

Figure 4: Average hours worked by quintile of the distribution of wages

Explaining changes in working hours

Our results raise the question of what factors have driven the dynamics of working patterns. Changes in hours inequality and in the correlation with wages can be due to composition effects or to changes for a particular category of individuals. For example, women tend to work fewer hours than men and are more responsive to wage variations (namely, exhibit a greater elasticity of hours with respect to wages). Our results from individual hours regressions suggest that, although the increase in female employment has played a role, changes in the hour-wage elasticity for each category of workers have been important. In particular, this elasticity has increased for men in the two continental economies, raising the question of whether this is an optimal choice by workers (labor supply response), or if workers face constraints to work their desired number of hours (labor demand constraints).

Although identifying causal effects is not possible, we look at a number of correlates that move together with the elasticity of hours worked. We find that greater trade openness or output volatility are correlated with a higher elasticity, in line with the hypothesis that trade and uncertainty force firms to be more competitive and hence offer a shorter working week to the least productive (lowest-paid) workers. The weakening of labor market institutions over the period is also correlated with the increase in the elasticity, potentially because it makes it easier for firms to use zero-hours or fractioned-day contracts.

What are the policy implications?

In some countries, those at the bottom of the wage distribution are working today less than they did two decades ago. Our analysis implies that this increases earnings inequality but is silent about whether reduced working hours are the result of individual choices, with the increase in leisure offsetting the loss in relative income, or if low-wage workers are unable to work as much as they would like. Recent evidence for Germany seems to indicate that the latter is indeed the case.

In a context in which some workers are claiming not to be able to work as much as they wish to, it is important to turn towards the constraints that may prevent them from doing so. We have identified macroeconomic and institutional correlates that could potentially explain this. An extensive literature has examined the effects that increased trade openness and weaker labor market institutions have had on distribution through their impact on wages. Our analysis indicates that they also have consequences for hours worked by different groups, thus affecting the extent of hours and hence earnings inequality.

Lastly, a caveat is in order. Our entire analysis has focused on individuals with positive earnings, and amongst those we have seen different patterns across countries. For example, earnings inequality was stable in the US and France but increased considerably in Germany. These changes may hide the fact that individuals who were previously not employed have entered the labor market or vice-versa. In the case of Germany, the increase in earnings inequality has been accompanied by increased access of low-wage workers to employment, indicating that these individuals have moved from zero earnings to low-but-positive earnings. Policies seeking to reduce earnings inequality should hence not ignore the fact that selection into employment is another important aspect of the observed distribution.

The posts represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.