Extrapolating the Unemployment rate

This post is a bit of "extrapolation fun". What is the unemployment rate going to look like when the U.S. faces the next Presidential election in 2012?

First, we have to answer another question.

• is this (largest and deepest post-war) recession qualitatively similar to other post 1940 recessions?... or,
• does this recession (and the Great Depression) have certain special characteristics that make them different?

Ordinary recession, but huge scale: If it is just a question of scale, "dumb" extrapolation should give us a pretty good guess.

Here's a chart I posted earlier, taken from from the excellent Calculated Risk blog, with my scribbles added.

The blue rectangles to the right are the future. The red line shows the current recession. The time it took to (apparently) bottom has only been matched by the recent "dot.com"recession (the dark brown line).

It has taken about two years for unemployment to climb (downward in the chart) by 6%: i.e., from about 4% to about 10%. We're now at 9.6%. If we assume that we're slowly coming off the bottom, we would see 4% in another two years (orange line, marked "straight reflection").

Normal but slower: What if we climb more slowly? Suppose we never come back to 4%? Suppose we're going to hit 5% as the new structural best at the top of the next business cycle? What if it's going to take a little more that two years to get there? In that case, we get a slightly slower extrapolation (the yellow line). This would imply an unemployment rate of about 6% during the 2012 election.

This time it's different: Certain types of recessions have been called "banking crisis recessions" or "balance sheet recessions" or "asset bubble recessions". Reinhart & Rogoff (R&R) studied such recessions across decades and across various countries. They find that the unemployment rate rises about 7% from the low boom-time rate, and takes about 5 years to peak.

Here is the U.S. unemployment-rate chart. That average, marked by the red dot, is the R&R average. From where we are today, only a significant "double dip" would take us up there.

Three other lines are plotted:

• B. A slow recovery, that is headed for a "new normal" of 7% of so.
• C. The beginnings of a more typical recovery that peters out and the either heads to a "new normal", or turns into a "double dip"
• D. A more typical recovery, falling the way it rose, and ending up almost where it started (better than 5% unemployment).

Consider a longer-term chart. If we do see a spurt in the unemployment rate,

heading to more of a R&R style

"banking crisis recession",

what will that look like relative to a longer term history of the U.S. unemployment rate?

See the point "A". If we compare it to the level in the early 1980s, it would be worse than others, but not far fetched.

Also re-plotted on this is "C", the more normal recovery that peters out.

R&R found a very narrow range for the duration, but a very wide range for the jump in unemployment rates. We've already seen unemployment deteriorate worse than many "banking crisis recessions" in other countries.

Still, even if one takes the 4 or 5 mid point examples, R&R's examples have a wide range, shown by the blue bar to the left. If we're in a R&R style recession, it would be typical for unemployment to hover at current high levels for at least another year. Yet, given the wide range it would also be consistent with past experience for the unemployment rate to come back down liken the more normal recoveries. (In fact, this would even be consistent with a double-dip scenario.)

That's it folks: The idea of this post is to understand past scenarios and possible outcomes, in order to give meat to any predictions of the future. I'm not making any public predictions here. It is food for thought.