Evaluating U.S. Unemployment Data
Introduction
An economic concept known as Okun’s Law—which explains the relationship between rates of unemployment and GDP growth—has held true for all US economic downturns and recoveries since 1948. However, we discovered a major discrepancy for the recovery period after the Great Recession of 2008, where Okun’s Law failed to explain the drop in the US unemployment rate as reported by the US government. This means the US labor market might not be as strong as numbers suggest, and considering central banks use Okun’s Law to help set monetary policy, this discrepancy could have major consequences.
So here’s a look at the current US labor market from Okun’s perspective, derived from our independent research.
Background
In 1962, Arthur Okun set out to answer this question: How well does an economy function under the condition of full employment?
Okun found an empirical regularity:1 a negative short-run relationship between unemployment growth and deviation of output from potential. In other words, Okun identified how economic growth (GDP) affects employment: generally, when GDP grows, unemployment drops. That relationship became known as Okun’s Law.
Okun’s Law is important to macroeconomic policy. Central banks use it to predict gaps between actual and potential output and to adjust policy accordingly.
- For instance, a positive gap—actual economic output exceeds potential output—and low unemployment suggests the economy is overheating. The US Federal Reserve might then try to rebalance spending and borrowing by increasing interest rates and limiting the amount of available credit. (Higher interest rates encourage companies and individuals to save more and spend less.)
- Conversely, if actual output is lower than potential output and unemployment is high, the Fed might want to reduce interest rates and increase the availability of “cheap money,” which would encourage spending and stimulate the economy.
We saw that second example in the Great Recession. But as mentioned in the introduction, our research found that Okun’s Law—which has held true for all US economic downturns and recoveries since 1948—doesn’t apply to the US’ recovery from the Great Recession, because it failed to explain the drop in unemployment as reported by the US government. Here’s our test in more detail.
“IMF research found that almost all deviations from Okun’s Law are modest and short-lived. Otherwise, Okun’s relationship is strong and stable, even in modern conditions.”
Okun in Brief
In his original research, Okun found that in the United States every percentage point drop in the unemployment rate from the preceding quarter corresponded to a roughly 3.2 percent increase in economic output. Some researchers have challenged and revised these numbers, but recent IMF research2 found that almost all deviations from Okun’s Law are modest and short-lived. Otherwise, Okun’s relationship is strong and stable, even in modern conditions.
Estimating Okun’s Law for the United States
We derived our estimates using the “changes” version of Okun’s Law.
For the curious, here’s a technical explanation of our methodology.
Equation (1): △Ut=α+β△Yt+ϖt
where:
- ΔUt is the change in seasonally adjusted unemployment rate from the previous period;
- ΔYt is the percent change in real seasonally-adjusted GDP from the previous period;
- ϖ is the error of approximation;
- β is the so-called Okun’s coefficient, representing the change in the unemployment rate for every percent change in GDP.
We estimated Okun’s law with annual and quarterly data using the ordinary least squares (OLS) technique. In the case of annual data, our specification coincides with equation (1). In the case of quarterly data, as suggested in [2], we included two lags of output terms. We did this to reflect the idea that companies and individuals need more than one quarter to adjust employment and labor supply in response to changes in output.
We’ve summarized these estimates in Table 1 below:
Note: Table 1 reports standard errors in parentheses.
For the period from 1948 to 2010, our results were nearly identical to the IMF research, except that for the period after 2010, Okun’s relationship failed to explain changes in the unemployment rate. When we tried to estimate Okun’s equation for the 2011-2018 period, we got statistically insignificant results.
And, as shown in Chart 1, the changes in the unemployment rate during the 2011-2018 period were much lower than what Okun’s Law would project.
What’s more, Okun’s Law suggests unemployment rates should have increased over the last eight years. To the contrary, US Bureau of Labor Statistics (BLS) reports show unemployment rates began to decline in 2011.
“Since the Great Recession, the share of the U.S. working-age population with a job or actively looking for one compared to the total working-age population has declined substantially”
Explaining the Breakdown of Okun’s Law
Okun’s predictions (Chart 1) can be explained based on changes in labor productivity, output, or labor force statistics. Let’s examine each in turn.
Labor productivity
All else being equal, the gap between predicted and actual data shows that from 2011 to 2018 the US economy needed more labor than it did in previous periods in order to sustain the same growth. In other words, during this period unemployment was a sign of decreased labor productivity.
This slowdown in labor productivity growth was observed in all major developed economies after the Great Recession (Chart 2), so we set it aside to examine other potential explanations of deviation from Okun’s law: output growth or labor force statistics.
Growth in economic output
Jon Williams at Shadow Government Statistics produces alternative estimates of GDP growth, inflation, and unemployment. According to Williams, US GDP growth has actually fallen for almost two decades.
- It’s interesting to note for this purposes of this article that Williams’ estimates of GDP growth for the 2011-2018 post-recession period correspond with unemployment predictions derived from Okun’s Law.
- But, if you look back further, Williams’ GDP estimates for the 2000-2010 period do not match actual GDP growth rates during the 2000-2010 period. Yet, according to our estimates, Okun’s Law does fit actual 2000-2010 unemployment figures without use of Williams’ alternative GDP growth rates, steering us away from the hypothesis that US GDP growth from 2011 to 2018 should be lower than official government estimates.
Most notably, though, we found that Okun’s Law would indeed apply for the 2011-2018 recovery period if the BLS unemployment rate estimates had been higher, leading us to our third hypothesis around labor force statistics.
Labor force statistics
From the second half of 2010 through November 2018, the official US unemployment rate dropped from 9.4 percent to 3.7 percent, which stoked speculation about labor shortages in the United States. But the BLS numbers don’t actually mean more people newly employed—they also show more people were removing themselves from the workforce. Since the Great Recession, the share of the US working-age population with a job or actively looking for one (which is how BLS defines labor force participation) compared to the total working-age population has declined substantially (see Chart 3).
Between the Great Recession and 2018, the official BLS labor force participation rate dropped from 66 percent to less than 63 percent of the working-age population. This accounts for the nearly 16 million additional working-age Americans who in that time dropped out of the labor force and did not ‘have or actively seek a job in the last four weeks’ (Chart 5). If those 16 million people had continued looking for jobs—even if unemployed—the BLS unemployment rate would have been nearly ten percent by the end of 2018.
The BLS definition of labor force participation therefore masks the real US jobless rate, and we conclude that the sharp decline in workforce participation is the primary factor behind the breakdown of Okun’s Law.
“The US government’s definition of the labor force participation rate masks the real U.S. unemployment rate. “
Conclusion
Okun’s Law, which held true through all US economic downturns and recoveries since 1948, fails to explain deviations in the unemployment rate during and after the recovery from the Great Recession. Specifically, Okun’s method predicts that the US unemployment rate would have increased from 2011 to 2018, while official US government statistics for that period show the opposite.
Why? The US government’s definition of the labor force participation rate masks the real US unemployment rate. This explains the discrepancies between the official unemployment rate and the rate Okun’s Law would predict.
We conclude that the US labor market isn’t as strong as government numbers say it is. Further, as the BLS continues to report a steadily declining unemployment rate, the Federal Reserve’s recent interest rate hikes might have been premature.
Looking for more data? Explore our curated dashboards on the topic in Knoema:
References
- Okun, Arthur M. 1962. “Potential GNP: Its Measurement and Significance.” Reprinted as
Cowles Foundation Paper 190. Retrieved from https://milescorak.files.wordpress.com/2016/01/okun-potential-gnp-its-measurement-and-significance-p0190.pdf - Laurence Ball, Daniel Leigh, Prakash Loungani. 2013. “Okun’s Law: Fit at 50?” IMF Working Paper.
Measuring Technological Impact On Employment
Introduction An economic concept known as Okun’s Law—which explains the relationship between rates of unemployment and GDP growth—has held true for all US economic downturns and recoveries since 1948. However, we discovered a major discrepancy for the recovery period after the Great Recession of 2008, where Okun’s Law failed to explain the drop in the …
Evidence for Chinese Economic Recession
Introduction An economic concept known as Okun’s Law—which explains the relationship between rates of unemployment and GDP growth—has held true for all US economic downturns and recoveries since 1948. However, we discovered a major discrepancy for the recovery period after the Great Recession of 2008, where Okun’s Law failed to explain the drop in the …