Friday, October 4, 2013

NBER Recessions vs. Actual Recessions? (part 2 of 2)

(continued from part 1)

Employment Recessions. The end of the discussion of part 1 of this blog post suggests another way that the econopundits and people differ. Our commentators usually only care about the flow of money through the market economy (corrected for the impact of inflation, of course). In terms of the analogy, they care about the health of the "tiger." That’s what’s measured by GDP: nonmarket goods and service and nonmarket costs are not counted as part of GDP (with one minor exception). One reason why the econopundits have this focus is that they care a lot about stock prices (perhaps because they own stock), while the stock market’s speculative ups and downs are encouraged by GDP fluctuations. They are also more likely to be served well by the market than are people who are living from paycheck to paycheck.

But for most people, there’s a more important issue than GDP: to quote an old labor leader, for most people what’s counts is “jobs, jobs, jobs.” That is, though money flowing through the economy helps create jobs, what’s crucial is the general availability of job vacancies. We must ask: is the flow of money fast enough to lower the unemployment rate, to make the labor-market situation better for the vast majority? Or with a recession, is the flow of money so slow that the availability of jobs sags and unemployment soars? (That is, what about the health of the people who are clinging to the tiger's back?)

It’s this conflict of perspectives that’s behind the seemingly oxymoronic phrase “jobless recovery.” In this situation, the “economy” is recovering in the sense that real GDP is rising (with more money flowing) but jobs aren’t being created quickly enough to provide employment to new job-seekers and those who have lost their jobs. Thus, despite rising GDP, unemployment rates rise!

This contradiction – and the possibility of a jobless recovery – arises because of what economists call “Okun’s Law,” named after the late economist Arthur Okun. It’s really just a rule of thumb based on studying the real world, not a law. A law is a regularity which always works the same way, such the law of gravity in physics. There are no laws like that in economics. Thus, Okun's rule of thumb says that the slow growth of the U.S. economy since the 2009 should imply rising official unemployment rates (U3), but exactly the opposite has happened.

However despite this seeming contradiction, Okun's law captures the nature of a real problem. This idea goes beyond the common-sense idea that producing more real GDP means that more jobs are available, so that unemployment rates fall. It says that in order to prevent unemployment rates from rising the year-to-year growth rate of real GDP must exceed approximately 3 percent per year.

Why is it that real GDP must grow faster than 3 percent per year to get unemployment rates to fall? Partly it’s because new workers keep entering (or reentering) the labor force, seeking jobs and adding to the potential pool of unemployed workers. In addition, the normal growth of workers’ productivity – their ability to produce output during an hour of paid work – means that if the demand for products doesn’t rise fast enough, bosses may find some or even all of their existing employees to be “redundant” and so lay them off.

That is, if the economy – as measured by flows of money through markets – is growing at 4 or 5% per year (or even 3.5% per year), unemployment rates fall significantly and in a sustained way. This kind of true recovery is exactly what the doctor ordered for the current U.S. economy, since unemployment rates are so high. (It’s true that inflation may result from a true recovery, but that doesn’t make it any less of a recovery. Rather, it tells us that the tiger can eat too much.)

On the other hand (if Harry Truman hasn’t sawed off the economist’s other arm yet), if the economy is growing at only 1 or 2% per year, unemployment rates rise. That situation might be a jobless recovery. But there’s a second possibility: it might be a case of that mysterious creature called a growth recession. In this case, real GDP slows its growth without actually falling (a negative growth rate), so that unemployment rates rise.

There’s a third situation where we see rising unemployment despite growing real GDP. This occurs before an NBER recession occurs: if real GDP growth slows (causing rising unemployment as collateral damage), it can lead businesses to start retrenching and cut their new fixed investment spending. This in turn can lead to an NBER recession (an actual sustained fall of real GDP) to follow. This might be called a prelude employment recession (though it would be helpful if someone could suggest a better name).

Major causes of prelude recessions include a private-sector slowdown (as in classic stories of business cycle) and efforts by the Federal Reserve or other policymakers to attain a “soft landing.” (This refers to an effort to engineer a gradual reduction of real GDP growth and slow the fall of unemployment rates (or even raised them) in order to keep inflation from getting worse.)

A prelude recession might also happen due to the government’s budget sequester and the current partisan-driven “shutdown.” Both reduce government spending and hiring, which can reverberate through the economy causing the real GDP growth to slow. This process might snowball as both consumers and businesses cut spending, again reducing the availability of jobs and income. Thus an NBER recession can result. It's also possible that all we're going to have is a growth recession, but that's not what the doctor ordered when we still haven't recovered from the Great Recession.

The Measures. With these three kinds of non-NBER recessions in mind, I measured a “recession” as involving two or more back-to-back quarters of rising unemployment rates. These may be called “employment recessions” since employment recedes as unemployment rises. I use the unemployment rate, since talking about thousands or millions of unemployed workers misses the fact that the labor force (those both willing and able to work for pay) steadily increases over time. I use quarters (rather than months or years) in order to parallel the journalist’s version of the NBER definition of a recession. Just as with an NBER downturn or a Household Income Recession, I omit the period of stagnation that occurs in the aftermath of a recession. However, note that Household Income Recessions typically last longer than Employment Recessions.

The Bureau of Labor Statistics produces several other measures of “labor market slack,” but here I’m going to use what econopundits think of as the “official” unemployment rate (U3). This is partly due to the fact that the BLS didn’t start reporting other measures until relatively recently. It’s also the number that receives the most attention in the press even though it leaves out problems such as involuntary part-time workers, people who are driven out of job-seeking by bad prospects, and long-term unemployment. I doubt that using other measures will change my results, but we shall see. (Being fundamentally lazy, I’ll let someone else do this work.)

In any event, I doubt that there is a “correct” gauge of the starts and stops of recessions. If anything, I’d prefer the Household Income Recession measure of my previous blog. But it’s time to get to the results. But the point is not to present a total alternative to the NBER’s recession as much as to look at the economy from a different perspective.

I used quarterly data from 1948 to the present as provided by the BLS and massaged by the Federal Reserve Bank of St. Louis, to get quarterly numbers. I also got the NBER dates from the St. Louis Fed. First, we see three recessions that the NBER missed completely. They are growth recessions, since unemployment rose due to slowing real GDP growth without a full-scale GDP downturn happening. They occurred in 1951, 1959, and 1976. By sheer coincidence each of these occurred in the third and fourth quarters of the year (and yes the number were seasonally adjusted). Only the middle one (1959) really deserves serious attention, however, since the increases in the unemployment rate during the other two growth recessions were minor (i.e., one tenth of a percentage point). Of course, these growth-rate dips are “minor” only from an economist’s perspective. For those people in involved, the situation could easily been dire, since so many of us have a hard time doing well unless the economy is truly booming.

Next, we see the infamous jobless recoveries. They appear in the table below, using both my dates and those of the NBER, where “q” refers to the quarter of a year. Ignoring the growth recessions, all of the recessions I found except the second Volcker recession (1981q4-1982q4) ended after the trough quarter of the corresponding NBER recession. For Volcker #2, it’s cold comfort that unemployment stopped rising, since that one attained the highest unemployment rate the U.S. had seen since the Great Depression of the 1930s. Anyway, here’s my list, with the ways in which the two measures differ highlighted in boldface. The list includes the second Volcker recession as #8.

            dates of cyclical peaks and following troughs
                 NBER Recession  ||  Employment Recession
  1. 1949q1–1949q3  ||  1949q1–1949q4
  2. 1953q3–1954q2  ||  1953q3–1954q3
  3. 1957q4–1958q1  ||  1957q2–1958q2 
  4. 1960q2–1961q1  ||  1960q2–1961q2
  5. 1970q1–1970q4  ||  1970q1–1971q1
  6. 1974q1–1975q1  ||  1974q1–1975q2
  7. 1980q1–1980q2  ||  1979q3–1980q3
  8. 1981q4–1982q4  ||  1981q4–1982q4
  9. 1990q3–1991q1  ||  1990q3–1992q2
  10. 2001q2–2001q4  ||  2001q12002q2
  11. 2008q1–2009q2  ||  2007q3–2009q4
(I have to figure out how to format this to make this table look decent.)

On the other hand, all of the employment recessions ended a quarter or more after the NBER recession ended. That is, the jobs situation (as measured by the official unemployment rate) continued to get worse even though the speed of the money flow through markets started rising. This joblessness was significantly worse during the 1990q3-1992q2 recession (#9), which ended fully five quarters after the NBER declared the recession over. It was this event which gave birth to the phrase “jobless recovery” while also helping to push President Bush #1 out of office (to be replaced by Clinton #1).

The lack of job creation after the NBER recovery began got worse, with recessions #10 and #11. The allegedly “mild” recession of 2001 (which I date as continuing all the way to 2002q2) ended two quarters after the NBER date. The same applies to the 2007q3-2009q4 “Great” one. Jobless recoveries seem to becoming the rule rather than the exception.

What about those prelude recessions? The first Volcker recession (#7) started two quarters earlier than in the NBER’s log. That is, the recession was much worse for working people than would be indicated by only looking at fluctuations of real GDP or NBER dates. The 2001 employment recession (#10) started one quarter “early” (compared to the NBER measure). I don’t know why this happened. Suffice it to say that the “Clinton boom” wasn’t as good as advertised.

Finally, the Great Recession (#11), which I date as being from 2007Q3 to the end of 2009, began one quarter earlier than the NBER measure. The fading job market was actually noted by the NBER committee that determines dates for business-cycle peaks and troughs, so that they stressed employment numbers much more than the usual real GDP measure in their dating. (They also dated the peak before the storm as during late 2007, but that’s lost when you use quarterly data.)

Conclusion. It's hard to draw a simple conclusion from these data. But two general conclusions are obvious. First, we shouldn't take NBER recessions as somehow reflecting the "gospel truth." The popular view that the "recession isn't over" actually says more than the NBER studies. Second, how we measure a peak month or quarter that begins a "recession" and the trough month or quarter that ends it depends on what our purposes are.  The use of an unemployment measure, for example, illuminates the phenomenon of a "growth recession" (and similar) and the conflict between what's good for the market economy (measured by GDP) and what's good for working people (measured by employment rates).

Jim Devine

No comments: