Monday, 15 June 2009

The Jobless Recovery Scenario

The jobless recovery scenario goes like this. (1) During a recovery (as at all other times) the basic labour market identity holds: the number of people employed depends on overall demand and productivity. (2) Productivity growth depends on innovation. (3) While the relationship between employment and demand is short term, the relationship between productivity and innovation is long term. Therefore, when there is a short-term drop in economic activity, you can expect to see a fall in employment, but innovation goes on. (4) Thus when demand rebounds, there are a lot of new technologies around, and it pays to be the first to invest into them.

There was an important ‘if’ at the core of the worse case scenario: the recession will not end if the global confidence level goes where it should - to the bottom of a deep mine. Unlike with real gravity, economics is an odd place, where one might actually lift oneself up by pulling one's own hair, Baron Münchhausen-style. We saw a similar global bubble before the crisis, where most of the justification for one’s enthusiasm kept being others' enthusiasm. Bubbles can work for a surprisingly long time.

For if people want to believe that the non-existent global policy framework works, if people want to believe that the recent return to growth by asset prices has anything to do with fundamentals, and if people are willing to believe that everybody else is willing to believe, then such a global pyramid scheme might even work. In a Startlingly Pink World, there might not even be a reason for this new would-be bubble to burst; reality might grow up to it.

If that happens, an old phenomenon might return: a jobless recovery. Except, this time, potentially on a global level.

When the troubles started, there was an inevitable comparison to the last Bad Times: the 2001 burst of the bubble. Though fun, a consensus quickly emerged that there was no similarity whatsoever… Maybe, on the way out, there will be.

The most conspicuous feature of the last time the mature economies returned to growth was the delay in job creation. When the economy was already booming, employment kept lagging.

Here is the original version:

In the above graph: the red line is US industrial production, and the blue line is unemployment (both are re-based to 1997 M7 = 100%). While production returned to previous levels, unemployment stayed at a high and surprisingly stable level for years.

There has a been a lot of discussion about it; here is my version:

The above graph takes the change of the monthly rate of US unemployment data, and regresses it on the change in the monthly rate of US industrial production, and time. All are moving averages of 20 months (that’s where the R-squared peaked). A one-month lag is employed. The time series is the longest I found: 1961 to 2009, and the relationship is very close: 82% R-squared. The graph shows the fit residuals. The interesting part comes when there is a lengthy period away from the mean: the end of the 1960’s and early 1970’s see a long period of ‘too many’ jobs being in the economy, while most of the 1980’s is ‘jobless’, and then comes the ‘jobless recovery’ after the 2001 bubble.

The whole point is that there is an underlying story about productivity. If you think that in the long run, productivity dynamics rule, then the recovery is catch-up time for firms. During recession, investment into new technologies halts, and thus when there is a return to growth, the surviving companies have the option to buy into the latest technology. As the innovation cycle is much longer than most troughs (if you want evidence, look at the OECD patent stats here), this means that for a while you can increase production by investing into new technologies, and thus delay re-hiring people.

The first time this hit me was during the 2001 crisis, when I was heading an investment banking research unit and a physicist friend of mine told me about the latest developments in the physics of computer technology. We realised that if half of what he saw was really getting into production, the economy coming out of recession would be all about this and not about giving jobs back.

Investment into technology has almost entirely stopped all around the world in the past 18 months. Yet, it is very clear that the advancement of technology has been as fast as ever. And thus, we are wondering if there will be a repeat show.

(Although, I am not at all convinced that this really is the end of the recession, see the previous post. But, if it is…) You could well argue that many of the forces that created the jobless recovery seven years ago, are here now, too. In particular, the productivity mechanism might be very similar.

So, are there any signs of this?

Well, look at an interesting confidence measure from the US:

The above graph is the differential between the demand confidence and employment tendencies, survey based, from the US. (That is above zero means that the demand confidence is above the employment confidence level. The red dots are the 2002-04 years, as well as the last month's data.) Not only it confirms the jobless nature of the post 2001 recovery, but it also provides us with one point, which is very much in the same direction.

Furthermore, if you look at the other countries from which comparable data is reported, support for this argument emerges. Here is the picture of the industrial confidence and employment sentiment data from 19 OECD countries (the red line stands for the annual averages, similar confidence differential as above, positive signalling that industrial production confidence is stronger than the propensity to employ people):

And here is the same, dissected: the red countries are the four that saw the most negative differentials during the burst recovery, and the blue is the rest. (The lines are averages for the respective groups):

Just like the US case, it is clear that at least for the countries that saw a jobless recovery last time around, there are signs that this time might not be that different.

All this depends on the optimistic assumption that the crisis is nearing its end. Which is either still (a) in the confidence measures only; or (b) directly related to fiscal spending. (Sorry, no time to illustrate the second one in this post). This makes this scenario just as likely or unlikely as the worst-case one.

If you wanted to pick, probably it will be the size and length of the boost in self belief that will decide. If there is a moral here, then it must be about there being no point in having small Münchhausen lies.


  1. very interesting piece. you suggest that this jobless recovery will be global. what are the signs for that. the mechanism would be difficult to defend in the case of emerging markets, no?

  2. great post. i love the Baron Munchausen analogy!

  3. tamas, what do you think this means regarding when job levels will be back? do we have to wait for as long as three years again?

  4. peter, thank for the great question. i am now looking through the blogosphere's comments about jobless recovery, and it seems that the most central question is indeed about the politics of it, and especially the length of the reaction by the labour market to the end of recession.

    contrary to others arguing that the speed of the recovery is what matters, i would think that it is the length and depth of the halt to the technology updates that should be the key factor -- as long as the productivity hypothesis is correct.

  5. So people who have lost their jobs should give up on trying to get a new one and just sit around trying to invent something.

  6. Re Anonymous
    I don't think the message is to sit around. The underlying tonic to any problem, inculding this recession, is to be proactive and be that on any level. There are many good points made here and thousands of other articles and some scenarios may happen, and some won't happen. What won't hurt though is proaction. In my personal case, appying for work and improving my skill set to b ready for the time that this earthy weight lightens it's load. Which it will. If misery loves company,(which on the internet it certianly does!) well than what of the oppisite of misery? What will that be like and how will that affect every thing when we (as much as humans can) move out from under the current cloud? I have seen consumer confidence numbers but what about something deeper, colectively a more positive movement, taking hold? That may be a powerful boost too

  7. there must be a strong policy implication to this. the post hints at possible policy mistakes that might enlarge the problem. fine. what are the other implications. what should the policy makers do when they DO understand this relationship? it is sure not about the delaying adoption of new technologies...