Why Productivity Increases Have Not Translated Into Higher Median Incomes

Written by Sig Silber

Random Thoughts from the High Desert

To bor or not To bor?

Actually I got that backwards:

It should be Robot or not Robot?

Here is another take on the issue.

And the problem is not going away.

Supposedly this Interactive Risk Assessor . will help a person determine their future.

I have a different approach (okay, maybe actually be the same approach with a slightly different slant) but my interest is definitely a bit different.

But before I get to that,  here is the Levy Institute Report on Destabilizing our Economy,which I think I may have commented on before. It contains this extremely important graphic.

I do not necessarily agree with their interpretation of the cause but the effect is fairly clear.

I do not necessarily agree that this is a transfer of labor from the 90% to the 10%. Of course you get a more extreme graphic if you focus on the top 1% and you get something a little but not much less extreme if you focus on the top 20% and you get the same pattern but again even less extreme if you focus on the top 40%.

So I conclude that it is not a transfer of labor from the bottom to the top it is IMO the shifting of the demand for labor of the bottom 60% otherwise known as the bottom three quintiles. Dividing into five groups of 20% of households is a good way of looking at things.

I do not think like Trump that it is primarily due to open borders. I also do not think like the Right that it is primarily due to the burden of government regulations. Nor do I think like the Left that it is inadequate regulation or lack of mandated levels of compensation. Some of the aforementioned issues have an impact but I do not think they are the major driver of secular decline or stagnation of household incomes. If they were the problem it would be relatively easy to make the changes necessary for those issues to become less of a problem. 

How Should We Look at Things.

There are a number of questions that we want to consider. Two of the most important relate to the two processes by which technology changes the demand for labor:

1. Is technology creating more demand for labor than it replaces with respect to existing products and services?.

This question can be divided into two different impacts

1. The labor content (and perhaps the cost) might be reduced and

2. The utility might be increased.

Both results should increase the size of the market and create additional demand for labor. So there is both a reduction in demand for labor per unit of product or service produced and an increase in demand for labor related to more units being demanded.

1. Historically, technology has created demand for new products and services. With respect to new products and services, it is generally an increase in the demand for labor except if the new products and services supersede and thus reduce the demand for other existing products and services

Process "A" may increase or decrease the demand for labor. Process "B" almost always increases the demand for labor.

So we are interested in the trend in Process "A" with respect to the ratio of labor replaced compared to labor added. And we are interested in the ratio of Process "B" to Process "A".

We then have a second set of questions. If we conclude that "A" and "B" ceteris paribus is reducing the demand for labor relative to the size of the population, we know that the price of labor will decline. In almost all cases the price of labor is higher when more is demanded and lower when less is demanded. One can look at this in the aggregate or attempt to break it down by categories of labor. For our purposes, that may or may not be necessary. It is a refinement useful if we were attempting to create a forecasting model to attempt to pin down the rate of the transformation. But it is not necessary to have a general discussion.

We might expect that if the labor component of producing goods and services declines, their price will decline. If that happens and to the extent it happens, what is lost to labor relative to lower wages is gained back at least partially, by lower prices for things workers purchase. Thus it is not clear that reducing the demand for labor and thus the price of labor will result in a decline in real wages for those who remain employed.

The data suggests that this is indeed the case. Real wages have not declined overall but they have not increased either. So we have the mystery of where did the productivity gains go. Recently in the U.S. profit margins have been unusually high. But the have now come down. This seems like a business cycle phenomena and not a long-term trend. It is unlikely for higher profit margins to be a long term trend in a competitive market. Sorry Trump, your idea of restricting trade makes as much sense as Sander's ideas that Communism raises standards of living. They are both ideas that are not sound from an economics perspective.

They may have been absorbed by the providers of capital. But it is not clear that the capital to labor ratio has increased. And if it has, the value of labor should be higher than before not lower.

Another possibility is that the gains from productivity have been exported. There may be a tendency for that in a highly competitive world market. But that would imply that there has been more productivity in what we export than what we import.

The gains from productivity may have been taxed and redistributed to lower income workers and non-workers as transfer payments. This can shift the demand curve towards lower wages. It gets further complicated to the extent that transfer statements result from increased National Debt.

There is another set of questions which relate to the distribution of wages. Although wages have probably never fit a normal distribution, it seems reasonably clear that recently there has been an increase in the size of the work force that is receiving wages much above the median and much below the median with a reduction in the size of the work force receiving near median wages. Curious or not so curiously, the middle of the distribution is where Process "A" above appears to have had the largest impact and the high end of the distribution is where Process "B" above has had the largest impact.

There is also the question of our ability to tease out the impact of technology on the median wage. At such low levels of productivity growth and the Hedonic adjustments, it may not be possible to detect and accurately quantify changes in the median income.

I created this table to see if it sheds any light on the situation.

I think it does. I think it suggests that Process "A" has become more dominant than Process "B". I think it shows a trend from early in the period to now of:

• Inventions that often tended to require a unit per worker and thus the labor cost of producing the labor reduction impact was significant. An improved lathe might be a good example of this. A worker with a better lathe is more productive but the lathe itself required a lot of labor to manufacture.

• Inventions that created new demand for labor in geographic areas opened to development by technology. This was very important in the second phase of the Industrial Revolution. The Interstate Highway System is a good example of this.

• Semiconductors is a good example of a technology where labor is required to produce the manufacturing devices but the variable labor per unit produced is low. So this is a change from technologies that have both a high labor content to produce and a high ability to save labor to technologies that share the labor cost to produce among a large number of units that save labor. This impacts both aspects of Process A to be on the one hand, more effective at reducing labor in the production of goods and also more effective at potentially reducing the cost and expanding the market for products and services. I say "potentially" reducing the cost to allow for the possibility that profit margins expand as discussed elsewhere.

• Software carries this trend to another level because of the World Market for products and services based on software. The labor input to produce the software is large but the variable labor content of the service is minimal. Anything to do with the Internet is a good place to look for this to be important.

Conclusions

It seems as if we have more questions than answers. It also seems to me that the investigations of this subject are not well directed to the questions which require answers. Most of what I read seems to be ideological based rather than engineering/economics based. .