web stats

Different metrics all describe the same thing – labor conditions: Response to Nate in Seattle

We’ve taken a couple of months off from the IDEA effective unemployment rate – our simple straightforward measure that replicates the pre-2008 unemployment rate. (Find the original discussion of the IDEA effective unemployment rate here.) The description and rationale have not changed from the first edition in March 2014. Behind them lies the premise that the unemployment rate and the participation rate measure the same thing – labor conditions. We contend that other measures, such as unemployment claims, wages, hours, part-time employment and even productivity, also describe what is often called the labor “market.” If these metrics diverge it is a problem of measurement, not a schizophrenia in labor conditions.

Beginning in 2008, an easily observable and widely noted drop in workforce participation occurred. It was common to hear, “The improvement in the unemployment rate is largely due to people leaving the workforce.” In a healthy labor market, unemployment and participation vary inversely, both measuring increases in employment. Figure 1 displays both the long-term stability and the inverse relationship.

Figure 1

When labor conditions are good, when the unemployment rate goes down, the participation rate goes up. In the graph, this is when the yellow line is falling. In the current market, the unemployment rate goes down BECAUSE the participation rate is dropping.

It was a simple matter to take a long-term average of the participation rate, subtract the contemporaneous participation rate and add this difference to the official rates to come up with an effective measure that corresponds to pre-crisis numbers. Since we have lost over three percent of the labor force since 2008, the effective rate is then three percent higher than the official BLS number. (The latest comparison is 8.9 vs. 5.3 on the headline and 14.0 vs. 10.4 on the all-in U-6.)

Figure 2

Nate’s objection, raised on Twitter, was that no long-term stable participation rate as we described existed. We simply disagree. We have taken 66.2 as this number from 1992 to 2007. (See the data on the website.) Whether it is a tenth of a percent or half a percent higher or lower does not prevent the result from being substantially more useful in comparing to the pre-crisis rates.

Others have proposed that the drop in participation is a demographic event caused by boomers retiring in the natural course of their work lives. Again, one makes a mistake, even with retirement decisions, if one assumes they are insulated somehow from general labor conditions.

Retirement from some occupations is mandated by the nature of the work; that is, an older person simply cannot physically do it. On the other hand, there is no want of anecdotal evidence that people are delaying retirement because they cannot afford it. Their 401ks have become 201ks and their savings are simply not adequate. On the other hand, reports are widespread of older workers who lose their jobs and cannot find other employment, and so are forced into a modest retirement often funded by social security. Further, we find that employment since the crisis has actually grown among older workers, while it is still in net decline for others.

Figure 3

But it is simply not credible that a substantial increase in healthy retirement and the collapse of the economy in 2008 occurred simultaneously.

Similarly, weakness in wages and salaries is contrary to a presumed strength in employment. Although weakening earnings have been evident since the de-industrialization of the economy under Reagan, wage pressure ought to be evident if employment conditions are accurately depicted by the drop in official BLS rates.

Figure 4

Healthy labor conditions would show strong wage growth.

Productivity itself is a measure of labor conditions. While this is contrary to the conventional view, which without supporting evidence ascribes productivity primarily to exogenous technologies, some evidence shows that the source of productivity gains is healthy labor markets. One hypothesis is that with the approach of full employment, managers are motivated to look for efficiencies. These may come from new tools (i.e., technology), but they may also come from more efficient processes, or simply shifting workers to more productive tasks. The graph below shows the unmistakable correlation between the pre-crisis unemployment rate higher productivity. (The full paper is here. By “The Rule of Eight”, zero productivity growth implies a headline unemployment rate of 8.0.)

Figure 5

A proper metric would combine all aspects of labor conditions: wages, unemployment, participation, productivity, hours worked, full- or part-time, etc., etc. These are each a different side of the same animal. One cannot accept a series of measures which contradict each other, or if one does accept such, one must also accept confusion.

Confirmation bias is rampant in this discussion. Tepid employment growth is seen as “solid;” meager wages gains are labeled “the start of improvement;” zero productivity increase is blamed on workers; and so on. The reader ought to suspect confirmation bias from us as well, as we see an economy struggling under a load of private debt and no help from the source of growth – public and private investment. Our “stagnation” is another’s “stable, if modest, Improvement.”

There is obviously much more that could be said, but to the point at hand, the effective unemployment rate: In a situation of unhealthy labor conditions, as labor goes underground or people choose not to commodify their activity (“leisure “ in the textbooks) or simply cannot afford to work for the skimpy wage and make other arrangements. This reduction in participation measures exactly what the unemployment rate measures from a different angle, so it is entirely appropriate to combine the two, as does the IDEA effective unemployment rate. The fact that it is simple ought not to detract from its descriptive value.


As noted in the original article, the Fed combines different metrics for labor to assess full employment, one of its mandates. Its calculation may not be so simple as ours, but it has to be apparent that all metrics should be pulling in the same direction. If September sees a raise in interest rates, a decision of enormous importance to Wall Street, we suspect it will be driven not by an improving labor market, but by the evidence that five years at zero have produced little growth and much distortion.

Alan Taylor Harvey