The Declining Worker Power Hypothesis: Questions & Responses

In this post, Larry Summers & I collate questions people have asked about our paper, “The Declining Worker Power Hypothesis”, and respond to them (thanks to Tyler Cowen for the impetus for us to write this up!).

1. Could the decline of labor rents have been caused by “management measuring the marginal product of labor more precisely”, perhaps through increased use of employee monitoring technologies?

Yes, this is possible. Increased ability for managers to monitor employee activity would likely lead to increased pay dispersion within employees and, under certain conditions, can also cause an across-the-board decline in wages. This is partly what we mean when we allude to more “ruthless” management practices. These practices would tend to push workers’ pay closer to the minimum that a company can afford to pay while retaining each worker – that is, the worker’s outside option – which would be the marginal product of labor in a perfectly competitive labor market or the monopsonistically competitive level in a monopsonistic labor market.

2. Don’t some – highly paid – workers have more bargaining power than before? This might include managers and executives, or financial professionals.

Yes: for this reason, we note explicitly that we are measuring the decline in worker power and labor rents of the *majority* of workers. Our measure doesn’t capture the very highest paid. Some of the redistribution of labor rents may have been upwards from the majority of workers to a small segment at the top of the income distribution of very highly paid managers and executives (See Appendix Section C11). Our baseline analysis is for the nonfinancial corporate sector so excludes workers in finance, but we replicate similar results for the corporate sector including finance. Note, though, that our estimates of labor rents won’t include highly paid workers in finance, so it’s also possible that there was redistribution of labor rents upwards to highly paid workers in finance (See Appendix Section B2).

3. Has the labor share even fallen? Part 1: Labor’s share does not appear to have fallen nearly as much if you consider workers’ receipts of equity compensation, or if you consider the owners of passthrough firms (i.e. people who run and own their own businesses).

It’s unclear whether to consider equity-based compensation, or the compensation of the owners of passthrough firms, to be labor income or to be capital income. Regardless, our work focuses on the decline in worker power and in labor income for the *majority* of workers. The large increases in income in the form of equity-based compensation, or accruing to owners of passthrough firms, have been at the very top of the income distribution. So while the magnitude of the decline in the aggregate labor share may not be clear, it is very clear that the share of income which is labor income for – say –the bottom 90% of workers or even 99% of workers has fallen substantially.

4. Has the labor share even fallen? Part 2: How do we take into account housing (Rognlie), depreciation (Bridgman), or the imputation of mixed income?

We focus on the compensation share of income in the *corporate* sector (to be precise: the nonfinancial corporate sector in our main analysis, and the entire corporate sector in the Appendix). Focusing on the compensation share in the corporate sector excludes housing, and excludes issues of the imputation of mixed income. Furthermore, in our main analysis we focus on the labor share of value added net of depreciation. Since depreciation rates rose over the period, the decline in the gross labor share was bigger than the decline in the net labor share (See Appendix Section C6). Most of the analyses of the labor share in recent years have focused on the corporate sector for these reasons, and indeed Rognlie (2014), who raised the point about the role of housing, finds a decline in the labor share in the corporate sector since the 1980s and identifies a large role for what he calls “pure profits” (which could represent our channel of the redistribution of rents between labor and capital).

5. If corporate profits are so high, how is this consistent with the persistently low demand postulated by Summers’ “secular stagnation” hypothesis?

Secular stagnation as we think of it is the product of a rising gap between the desire to save and the desire to invest (which, in an IS-LM type framework, would push down the neutral real interest rate). Falling worker power redistributes income from lower and middle-income people to the rich. The rich have a higher propensity to save. Thus, falling worker power increases the desire to save relative to the desire to invest. Rising inequality has been posited by several authors as a contributor to the declining neutral real interest rate (see e.g. Smith and Rachel 2015). Under this view, secular stagnation is exemplified by low private return to capital investment – but, in a noncompetitive world, this may or may not be the same thing as an abnormally low profit rate or capital share.

6. Was the decline in worker power endogenous to changes in globalization and/or technology?

Certainly both technological developments and globalization, by influencing the “outside options” of firms, influenced the extent to which workers can exercise power. In addition, for some firms, increased global competition may have destroyed product market rents, reducing the degree to which even workers with some degree of rent-sharing power are able to increase their pay. However, there are strong indicators that at least some large part of the decline in worker power was not endogenous to these broader macro trends. First, different countries, with arguably similar exposure to globalization and technological change, saw very different trends in unionization rates – one measure of worker power – over the period (see e.g. Schmitt and Mitukiewicz 2012). Second, within the U.S., both tradable and non-tradable industries – i.e., with different exposures to globalization – saw similar proportional declines in unionization rates. Third, within manufacturing, the industries with the biggest declines in wage premia were not those with the biggest increases in import competition (see Section II.B). Fourth, there is a wealth of direct evidence of changes in policy, institutions, and norms, which would be expected to decrease worker power (and which did not stem from globalization or technological change), including the weakening of labor law and labor law enforcement, the breakdown of pattern bargaining, the expansion of right-to-work across several states, and the increase in shareholder activism.

7. The measure of inequality you focus on is the share of income going to labor vs. capital. What about other aspects of income inequality?

While we don’t focus on it in our paper, there is good evidence that declining worker power – particularly, the evidence focuses on the decline of unions – has increased income inequality in the U.S.. For recent evidence, see Farber et al (2018) and Fortin et al (2019), for example. We do estimate labor rents for college vs. non-college workers, and find that labor rents as a share of compensation fell much more sharply for non-college workers, both because the non-college unionization rate declined much faster, and because the large firm wage premium declined much more for non-college workers. We also carry out a back-of-the-envelope estimate as to the degree to which the decline in worker power might explain the rise in the income share of the top 1% (See Appendix Section C11).