The Fed's Wishful (And Wrong) Thinking About Unemployment

Photo: Medill DC

No one seems to have noticed that the Fed’s latest unemployment projections just don’t make sense.  While most economists are concerned about a jobless recovery, the Fed is forecasting lots of jobs, but little recovery. Yes, today’s projections suggest only tepid output growth in the next few years. And given this, it’s hard to see how we will make much of a dent in the unemployment rate.  Yet the Fed believes otherwise, cheerfully (wishfully?) forecasting declining unemployment.

The blue dots in the graph below show how economic growth has historically translated into changes in unemployment. The dashed line shows Okun’s law, which fits the data pretty well. This rule-of-thumb tells us two things:

1.       You need economic growth in excess of around 3% to lower the unemployment rate.

2.       If GDP grows a percentage point faster—which counts as a very optimistic forecast at the moment—then unemployment will fall by half a percentage point, from say, 9% to 8.5%.

Now keep your eye on the red dots, which show the Fed’s unemployment projections. These unemployment forecasts are all more optimistic than Okun’s law suggests. That is, the Fed is projecting better news on the unemployment front than is justified by their economic growth forecast. More tellingly, the red dots showing the Fed forecasts are all close to the bottom of the data cloud, which means that it will be quite unusual to see the labor market perform so well, given their projections for only moderate economic growth.

All told, the Fed is projecting GDP growth to average 3.4% over the three years to December 2013. Plugging that into Okun’s Law suggests that it is reasonable to expect unemployment to fall by the end of 2013, but only by around half a point (or 0.4%*3 years *0.5 Okun coefficient = 0.6 percentage points).  Instead, the Fed is projecting the unemployment rate to fall by nearly two full percentage points.

Obviously there are a million caveats that come with any forecast. The Fed could end up being right here, but if history is any guide, it seems unlikely. Their key risk is that they’re too optimistic about the prospects for reducing unemployment anytime soon.

Bottom line? The Fed’s output forecasts combined with Okun’s law suggests that there’s real reason to be concerned that by early 2014, the unemployment rate may still be as high as 8.5%. If so, we’ll have nothing to show for our lost decade but a sick economy.

And yes, the usual disclaimer applies.


Francisco

Your analysis are very right and I would agree, but I would first have to have more evidence on how the Federal Reserve projects or forecasts GDP growth. Maybe their projections are overstated, which would be consistent with the last 2 years or so, but I'd like to know for sure. Great article though.

billb

Wait. Those 3 points fall squarely within that cloud of points between 1% and 5%. Yes, they don't follow the trend line for a sequence of related measurements, but Okun's Law doesn't require that. It's an linear fit to a cloud of data, and those 3 points are as close to the trend line as many of the other points. I don't think you can extrapolate from a correlation to require new data to fall on the line!

Andy

I agree with you per se but he didn't say the forecast was "wrong" based on Okun's law or that accurate forecasts must lie squarely on the trend line. He only said it was optimistic and unlikely. The placement of the forecast in the cloud seem to support a "best case" which is usually not very likely.

Joseph Ryan

Spot on.
Once again, "the Emperor has no clothes"
Thanks for the tip

Pierre-Louis

2 questions?

- why use post 1980 data?
- what varaiables explain deviations from Okun's law? maybe these explain the forecast.

Quinton

You use 1980 data to present to account for the regime change at the Fed. Basically, Volker established the credibility of the Fed in terms of fighting inflation. Before that the Phillips curve, whereby you push down unemployment by increasing inflation, was all the rage, but after the experience of the 1960's and 1970's, most economists think the long run phillips curve is bogus. There is of course a healthy debate about rational expectations but I think that's the generally accepted story

Quinton

My understanding was that Okun's Law had broken down somewhat during the last recession, in a number of ways. Unemployment rose faster that the law would suggest and then fell more sharply as well. Not sure if the Fed is taking this into account but it seems possible that some of the growth we've already had hasn't been converted into jobs yet. I'm skeptical of this argument but I'd like to hear a more in depth discussion.

I didn't hear it here originally but here's am op-ed with more detail: http://blogs.wsj.com/economics/2011/03/10/okuns-confounding-law-here-we-go-again/

mherts

My question/concern to the WSJ oped is two-fold:

1) On the first point ("employers were so frightened..."), I am unsure how this would allow for the Fed's current projections (obviously the WSJ piece was not written in response to Fed data published three months later, but the two are still connected) other than that the Fed is trying to take into account a perceived "over-reaction" in their projections; and,
2) Regarding the final point ("worker productivity explained the unemployment surge...") the author lists a decline of 0.6 in the labor force participation rate. This decline represents less than a 1% (.93%) drop over the course of a year; such a small decline - in a significant recession, with the baby boomers fast approaching retirement ages and only a one year sample - does not seem like a statistic to weight an argument.

The explanation I could see to the Fed data is that there is some perceived decline in productivity either currently or going forward. I would also be interested to see wage data for hourly employees as it relates to this. If the employees retained by companies during the layoffs occurring over the past three years are of a certain wage point, it could make sense for an employer to keep these employees at a lower set of hours (eliminate overtime) and hire additional employees at lower rates (this also only makes sense if an employer expects to offset the cost of benefits/taxes of all employees in the short term and/or if the new workers on a temporary/contract basis). If Okun's law requires both unemployment to drop along with an increase in hours worked, there could be a larger decline in unemployment to offset a realized drop in hours worked (disclaimer/clarification: I write this from the paradigm of the manufacturing model where it is relatively simple to change a shift from week to week between 30 hours per week up to 80 and anywhere in between or offset the OT with new labor either temp or full-time).

Read more...

kharris

The one way that Fed projections do make sense (and they do know the math) is if the participation rate falls. Read that way, the Fed is quite grim about the health of the labor market. It is betting that opportunity for employment is so bad that it becomes generally recognized, and a bunch more of us begin behaving like the guys in pre-boom Irish bar songs.

rana

another way the Fed's prediction makes sense is to pdate Okun's "Law" for the current estimates of potential GDP growth that are in the 2 -21/2 percent range rather than the 3 percent used here. Moveover, we have just seen a drop in the unemployment rate over the first half of the year far in excess of Okun's law, which makes the 2011 dot very reasonable

Jacob Emmons

This may be seemingly unrelated to the topic of unemployment. However, my perception says most all things are related. I will not elaborate on why at this time. I would only like to ask a question and use/or coin a term. I would like to know if anyone else has taken notice or mentioned something I perceive, which I will call an opportunistic employer. I will refrain from defining the term now to alleviate biasing your thinking with my ideas. If defined as I see it know, it is strongly related to the economic future of America.