May not be redistributed commercially without the author's permission.



Arnold Kling, "Arguing in My Spare Time", No. 14

May 20, 1998

"What is truly startling is not how fast and small computers have become but how stupid they remain." -- Paul Krugman. reprinted in "The Accidental Theorist"

I came across this quote after I wrote the previous essay. Krugman is a leading exponent of the view that recent U.S. economic performance reflects a slowdown in the growth of labor compensation rather than an increase in productivity.

The previous essay argued that the unusually sharp reduction in the growth rate of unit labor costs in recent years has been due more to curbing wage rates than increasing productivity. This assertion (which required little more than a recitation of the official statistics for the U.S. economy), along with the colorful "Junta" metaphor, carried with it an implication that there is something unfair and/or unstable about the current economic regime.

The stability of the current regime can be analyzed in two respects. One is political stability--is there likely to be widespread dissatisfaction? The other is economic stability--is wage behavior consistent with technology and markets?


We can start with some interesting facts from the Federal Reserve's 1995 Survey of Consumer Finances. The Fed survey is the best data available on the distribution of wealth in this country, even though it is not as up-to-date as other data.

Three observations from the Fed survey:

1. Wealth in the form of small business equity has been declining. This appears to run contrary to press reports that our economic boom is being fueled by small business and entrepeneurship. I like to put it, "For every successful new technology start-up, there is a mom-and-pop store being put out of business by Wal-Mart." This finding has little direct pertinence to the topic at hand, but it nonetheless is useful knowledge to have.

2. Adjusted for inflation, the median and mean level of income fell between 1989 and 1995. This is consistent with the view that wages have been suppressed. However, keep in mind that since 1995 the unemployment rate has fallen about 1.5 percentage points, which probably has helped to raise the mean and median income. See below.

3. The percentage of people who own stock (including direct holdings and mutual funds) has increased, to over 40 percent in 1995. The percentage of household wealth in the stock market had risen considerably, and it must have increased even further since the survey, as stock prices are up another 50 percent since then.

The ownership of stock still is highly concentrated. Once you get beyond 5 percent of the population, stock market wealth is rather limited. Still, it is more dispersed than it used to be.

The Effect of Lower Unemployment

A sustained drop in the unemployment rate affects many people. That is because of the high transition rate into and out of unemployment. Suppose that the unemployment rate is cut by one percentage point for one year. If the typical spell of unemployment is two months, then this means that approximately 6 percent of the population that would have experienced unemployment some time during the year at the higher unemployment rate will instead work the entire year at the lower unemployment rate.

For the past two years, the unemployment rate has averaged at least 2 percentage points below that in 1991-1992. Therefore, it is reasonable to conjecture that at least 20 percent of the population has experienced less unemployment in the past two years than they did 6 years ago. These people would not view themselves as worse off just because their hourly rate of pay has declined in real terms.

Overall Satisfaction

If we only looked at wage rates, and ignored unemployment rates and the stock market, then my guess is that the majority of people would have experienced a decline in well-being over the past several years. However, with about 20 percent of people better off because of less unemployment, and perhaps another 20 percent better off because of strong stock market performance, the decline in inflation-adjusted hourly wage rates has been offset in many cases.

Overall, my guess is that over the past several years about 40 percent of the population is much better off. This includes people who have experienced less unemployment, people who have bucked the general trend and received strong wage increases, and people with significant holdings of stock. I would conjecture that about 50 percent of the population is only slightly worse off to slightly better off, and about 10 percent is definitely worse off. Thus, satisfaction with economic performance should be fairly high.


In a competitive economy, how can companies pay people less than they are worth? If you are not paid your market value at one firm, you can solve the problem by moving elsewhere.

My guess is that the key to understanding the dynamics of compensation lies in the area of transferable vs. non-transferable skills. Transferable skills (the economic jargon would be "general human capital") are skills that can be used in many different firms. Non-transferable skills ("specific human capital") are skills that can be used primarily at one particular firm.

Non-transferable skills include understanding the processes and computers systems that are peculiar to a particular firm. Thus, if two companies merge and one set of processes and systems is phased out, some non-transferable skills become obsolete.

It is very difficult for a firm to avoid paying market prices for transferable skills. If pay is too low, people with those skills will take advantage of opportunities to move elsewhere.

On the other hand, non-transferable skills make the employee more valuable within the company than at some other firm. There is a range of indeterminacy in the extent to which this value differential is captured by the firm rather than by the employee.

The firm must provide at least some reward for non-transferable skills. Otherwise the workers will have no incentive to acquire such skills. In fact, the common tendency to make the "management track" attractive to technical workers probably reflects this need to reward non-transferable skills. However, once non-transferable skills have been acquired, the firm has the ability, if it chooses, to pay very little for those skills.

Generic software and bargaining power

The relative bargaining power of the firm and its employees with non-transferable skills probably depends on the substitutability between transferable and non-transferable skills in generating output. For example, if the cost of using generic software to replace a proprietary firm-specific system is high, then people with specific understanding of the proprietary system will be in a strong bargaining position. if the generic software provides a modest-cost substitute, then the people with the non-transferable knowledge of the proprietary system are in a weak position.

A theme of these essays is that the most important economic trend in software is the reduction in the cost of generic software relative to proprietary, firm-specific systems. This trend would tend to weaken the bargaining position of workers with non-transferable skills, and thus would help to account for the shift from wages to profits that has occurred.

Some analyses of differential wage trends are consistent with the view that transferability of skills is important. Reports indicate that the pay differential between college-educated and non-college-educated workers has increased in recent years. At least one study also found that employees who work with personal computers are paid more than other-wise comparable employees. It is likely to be the case that people with higher levels of education have a more transferable skills. Also, people who work with personal computers are more likely to have transferable computer skills than people who use firm-specific processes to perform the same work.

The hypothesis here is that software innovation does indeed account for some of the reduction in labor costs in recent years. However, a relatively small proportion of the cost reduction reflects visible improvements in productivity. Rather, the ability to substitute generic software for firm-specific systems has reduced the negotiating leverage of workers who have non-transferable skills tied to those systems, thereby lowering labor costs.

There are other interesting issues to consider concerning the hypothesis that generic software has become more substitutable for firm-specific software. For instance, how does this trend affect industry structure (large firms vs. small firms)? These topics are left for future essays.