Is Okun's Law Really Broken?

“Okun’s law” is a much-loved rule of thumb — it links increases in the unemployment rate with decreases in output.? The red dots in the chart below illustrate?Brad DeLong‘s version of this rule, which relates the change in output over the past eight quarters with the change in the unemployment rate.? Most of these dots lie pretty close to the dashed line, which suggests a stable relationship.? Based on this rule, and the relatively mild decline in measured output, you might have expected the unemployment rate to have risen by 3.5 percentage points over the past two years, to about 8 percent.? But the dots at the top left show what actually happened-unemployment rose by something closer to 5.5 percentage points, and the unemployment rate is closer to 10 percent.? That’s a big difference.? And it has led?many commentators to ask whether Okun’s Law is broken.

But perhaps the problem isn’t Okun’s Law.?? Perhaps the problem is how we measure output growth.? In fact, there are two measures of output growth-the usual measure, which adds up total spending in the economy, and the alternative, which adds up total income.? In theory, the two should be exactly the same. In practice, they have been very different during this recession.? The blue dots show recent changes in this alternative measure of output.? These GDI numbers suggest that output growth actually declined much more sharply than had been widely understood.? Based on this alternative measure, the recent sharp rise in unemployment is no mystery at all. (Indeed, the 2008 data suggest that the real mystery may be why it didn’t rise faster, earlier.)


There’s a simpler way to show all this: Let’s map out Okun’s Law using this alternative income-based measure of output.? That’s what is shown in the figure below.? The rise in unemployment seems quite consistent with these alternative output data. If anything, the puzzle now appears to be why unemployment didn’t rise by more in early 2008, given the very weak state of this income-based measure of output.


What’s good news for Okun’s law, though, is bad news for the economy.? This alternative measure of output growth suggests that the recession may have been deeper, and longer-lasting than previously thought, although data for the fourth quarter aren’t yet available. While many economists believe the recession ended in the second quarter of 2009, this income-based measure of output kept shrinking in the third quarter, too.? And while the expenditure-based measure is back to its level from the third quarter of 2006, the income-based measure suggests that output is still 3.5 percent below that level.? That’s a pretty big hole to dig out of.

Which is the right measure of output to focus on?? It’s still an open question, but some?interesting recent research by the Fed’s?Jeremy Nalewaik suggests that we should be thinking harder about the income-based measure. And Jeremy has promised further new results, which he’ll present at the?forthcoming Brookings Panel.


The chart shown does not have much data (8 or 9 points) with negative changes in the 8 quarter GDP. Also, a significant portion of the negative data is from the current recession. Okun's law may be appropriate for between 0-14% positive. Extrapolating into the negative GDP region obviously has been ineffective.


Does the output growth numbers take into consideration things like unemployment benefits, the under-employed, and things of that nature. My thought is that if we include the under-employed, which may be a much larger number than the unemployed, how do the results change?


As it is well known Okuns law is quite imperfect.One of the problems is the lack of measurement of productivity.


Does it take into account all of us that became self-employed after the dotcom crash and are now unemployed because contracting and consulting has dried up?

I know dozens of people who are now unemployed and underemployed that got by during the mid 00's freelancing. They won't show up on any unemployment statistics though.


You say the income is higher than the expenditures, and they should be the same. So are you saying people are spending *abnormally* less than they are earning, even for a recession? What might be the reasons behind that? And if that is the case, aren't we due for a pretty significant resurgence when people start spending all this money they've been keeping back?

Dr J

Savings? Savings rate went from near zero to 6% during the recession, may have something to do with income based vs. expenditure based measures - nice to see you try to do some economics on the blog instead of promoting libertarian/corpo-aristocratic ideology


Firms also will retain employees (inputs) above the level of production in anticipation of recovery. It is more costly to the firm to lay off employees only to have to hire and train a new staff when recovery occurs, than it is to retain staff. Thus employment is misaligned with production.

Eric M. Jones

In the space of one hundred and seventy-six years the Lower Mississippi has shortened itself two hundred and forty-two miles. That is an average of a trifle over one mile and a third per year. Therefore, any calm person, who is not blind or idiotic, can see that in the Old Oolitic Silurian Period, just a million years ago next November, the Lower Mississippi River was upwards of one million three hundred thousand miles long, and stuck out over the Gulf of Mexico like a fishing-rod. And by the same token any person can see that seven hundred and forty-two years from now the Lower Mississippi will be only a mile and three-quarters long, and Cairo and New Orleans will have joined their streets together, and be plodding comfortably along under a single mayor and a mutual board of aldermen. There is something fascinating about science. One gets such wholesale returns of conjecture out of such a trifling investment of fact.
- Mark Twain (Life on the Mississippi)



What is the difference in how total expenditures v total income is measured. Why are they not the same?


Productivity goes way up when your export jobs to slave-wage countries. Expentitures here don't go back into our economy anymore. The multiplier effect is zero now.

David Hart

Also, there is considerable problems in measuring unemployment in the US because of limited unemployment benefits in the US. Most people find it quite difficult to admit that they are unemployed and will do so only if there is some benefit (i.e. unemployment benefits). If the US instituted a small benefit such as a very small utility, gas or public transit ($5 perhaps) rebate that would automatically be removed once a person was employed, we could get a much better handle on current unemployment statistics and compare them with other countries unemployment statistics.


It looks to me like there is a time lag in the unemployment: when the curve is moving up and to the left, it is below the dotted line and when it is moving down and to the right it is above it. (This is just a guess: I can't tell which direction the to high/right loops are moving in.)

Try replotting with various time lags: e.g. a 2 quarters lag would plot a point for GDP change 2006Q1-2008Q1 vs unemployment change 2006Q3-2008Q3. Perhaps one of these time offsets will give a much tighter correlation. (It would also be more useful: it would let you predict the future instead of predicting the present.) (Also: are the GDP figures adjusted for inflation? What effect would it have if you changed this?)

Interestingly, the dotted line is above/right of the (0,0) point, indicating some combination of increasing population and increasing GDP per capita (which in turn could be increased efficiency and/or inflation.) I think it would be better to plot GDP per capita (or maybe GDP per working population.) The unemployment axis is per-(working)population, so it would be cleaner (and directly show the trend in per-worker GDP increase) if the GDP axis was too.

Finally, I note the fitted line does not have a slope of one: 20% in the change-in-GDP axis corresponds to 7% on the change-in-unemployment axis. This suggests that either the most efficient (highest GDP production) workers are laid off first, or (more likely) that when unemployment is high, even those who still have jobs have decreased GDP output. (Or at least decreased relative to the secular trend towards increased GDP output over time.)

I am not an economist, so please excuse any error or misuse of terms in the above.


ray bans on my face

Damn, now I gotta go tell my friends and family I was wrong when I said "Okun's law is broken."


Dr J and MW both make excellent points. The fluctuations in the savings rate is a point worthy of consideration. Also, to point to what MW was saying, there is definitely a lag in unemployment that might indicate the numbers actually make more sense than they do when presented in the article. Of course, perhaps the "lag" is non-existant. Personally, I side with the originally posted (by MW) notion. Okun's Law has historically been a very consistent measure, and the latest recession is not as much of a statistical anomaly as, say, the stagflation in the 70's was (caused by bottle-necking of supply in the oil market, which has been run by ((basically)) a cartel since the flood!) you could easily argue that given the cybernation in conjunction with outsourcing that has occurred since the 70's, that indeed unemployment tends to lag a bit against output. If this was weighed into the model described in the article- as MW stated- we might find a more uniform distribution which adheres to Okun's law better.