Matt Steinglass at the Economist has replied to my recent piece on how liberals often ignore labor markets by outlining how we can have more unionization without less employment.  His argument goes like this:  unions capture profits and increase the labor share of national income. This increases aggregate demand, which fuels growth and leads to higher employment. I’m going to address this argument in two posts, since the reply will be lengthy.

The first question I want to address is “if unions increase wages by capturing profits, would it increase economic growth?” I think this argument suffers from what I’d like to call the fallacy of permanent Keynesianism. It’s true that there is slack in the economy right now, and that increasing consumption and therefore aggregate demand will increase economic growth. But the level of unionization in the economy is a long-term structural and institutional issue, not a short-term countercyclical one.  Attempting to increase consumption like this will come at the expense of savings, which in the long-run means lower investment, a lower capital stock, and therefore lower economic growth. In short, more consupmtion does not necessarily mean more economic growth. Consider, as a simple example, the Golden Rule of savings in a Solow growth model.

In fact, in a symposium commemorating the 25th anniversary review of his celebrated book “What do unions do?”, labor economist and union defender Richard Freeman makes the complete opposite argument as Steinglass.  He argues that the negative direct impact of of unions on economic growth (which, as discussed below, he acknowledges) may be offset by an increase in workers’ savings that result from labor contracts with larger pensions.

But even if it were true that more consumption always meant more economic growth, I do not agree with Matt’s contention that unionization would increase the labor share of national income. The following graph shows the ratio of labor compensation to corporate profit from the BEA’s NIPA tables. While Matt is right that this ratio is at a historical low, notice that the pattern bears no relationship to the level of unionization in the economy, shown in the graph below it.

Labor’s share of national income actually fares much better when you use the definition used by Robert Gordon and Ian Dew-Becker in their paper on inequality.

Here the denominator is GNP minus consumption of fixed capital, minus indirect business taxes. While the number has fallen recently, it is well above historical lows, and well above what it was during the heydey of unionism. Again, the important thing is there is nothing to indicate that the decline in unionization has affected labor’s share. Gordon and Dew-Becker conclude in their paper:

“Thus,to a first approximation, we conclude that the increase in American inequality after the mid-1960s has little to do with labor’s share in domestic income.  What has happened is a sharp increase in skewness within labor compensation.”

However, even if unionization doesn’t allow labor to capture profits in the aggregate, the empirical evidence (some of which is summarized usefully here by Barry Hirsch, more can be found in the symposium discussed above) does suggest that it happens within unionized firms. But is this a good thing as Matt believes? The problem is that measured profits, when they are rents that unions can potentially capture at all, can be either pure rents or they can be quasi-rents. In the long-run, competition eats away at pure rents, and quasi-rents that represent normal returns to long-lived physical and non-tangible capital are necessary and important. Allowing unions to ex-post grab quasi-rents to long-lived capital incentivizes firms away from making those investments in the first place. In fact the empirical evidence on this topic shows that unionized firms have less profit, less investment, and less R&D spending.

While Matt seems to disagree, the notion that unionized firms suffer from lower employment is actually not a very controvertial claim among labor economists.  For instance, in his aforementioned book, Richard Freeman agrees that union firms not only have lower rates of R&D, investment, as discussed above, but that they have lower employment growth. He argues that this may not negatively impact economic growth if the decreases in union firms are offset by increased investment, R&D, and employment growth by non-union firms. But this is exactly the problem with arguing for a more unionized economy: the only way it isn’t damaging is if there are nonunion firms to take up the slack and grow. In the long-run, this suggests a steady decline of unionization is inevitable.

There is another major disconnect between the way liberal writers like Matt and liberal labor economists like Freeman write about unions. Typically, the former praise the so-called “monopoloy face” of unions, whereas the latter usually recognize that unions ability to raise wages above market level is a downside of unions. Labor economists who support unions tend to do so for what economists call the “voice face” of unionism. Indeed, I think this positive aspect of unions, whereby they communicate with owners and managers the desires of the workers, is underestimated by many conservative writers. The “voice face” of unions can lead unionization to have a positive impact on firm productivity. Granted, there are a lot of issues here, like the possibility of alternative ways of providing workers voice that don’t risk a “monopoly face”, and the fact that the empirical impact of unions on productivity is ambiguous whereas the wage impact is not. But suffice it to say that the aspect of unions liberal writers praise is often recognized by prominant liberal labor economists as a problem.

Next I’ll discuss why unionization has fallen in the first place, why high unions could co-exist with low unemployment in the 50s and 60s, and why both of these things tell us that high unionization is undesirable today.