Wednesday, 17 June 2009

Technology In Jobless Recovery

The technology-update dynamics (see previous post) is central to the argument that the current crisis will finish with a jobless recovery, similar to the years after the 2001 bubble burst.

The raw form of the hypothesis suggests that the firms that had shed labour during the recession subsequently invest into technology, new technology in particular, rather than expand their labour force, when demand for their products is renewed. Here are some interesting graphs (source: Fed):
Or, in a different version (using the same data as that behind the second graph above):

(The above graph: US manufacturing growth rate MoM, minus the same without the high-tech sectors. The red line is the 12 month moving average. The second half of the 1990's is clearly over the average of the rest of the period, and the return to normal is rapid after the 2001 drop. Although this is no proof, it seems at least consistent with the hypothesis.)

Both of the above graphs suggest that there was a substantial amount of over-investment into technology before the 2001 crisis, and thus the 2002-04 years saw a productivity rise as a result in part of using the previously acquired technology, and only in part because of purchasing new technologies.

Interestingly, this is not necessarily the case for the current crisis: the level of investments into technology seems to have been on par with the rest: there was no over-investment into technology that would jump out of the data, at least not the data I am looking at now. However, the hypothesis does not necessarily require a previous over-investment, only the potential to increase production without new labour. Thus a lengthy halt to the technology-intensive fixed capital could do the same. If this was to happen so, there would be a fairly straightforward equity price consequence...

I have been trying to find data for comparable variables for the rest of the world. (The usual trouble: lack of comprehensive global statistics). The best I could come up with is the share of computer and communication services in services imports by the WBDI. Each regression had a linear time variable, and a dummy for the 1996-2004 years. Here is the result:

Eastern Europe, Latin America, Middle East, South Asia: either not significant, or low R-squared. East Asia, and the Euro area behaved - at least by this measure - in line with the US and the hypothesis. And so did Sub-Saharan Africa (but why, if the rest of the emerging and developing world did not?)

In sum, these back-of-envelope results do not contradict the hypothesis that jobless recovery stems from an underlying technology dynamics. In the case of the 2001 bubble, it seems that the productivity rise came, at least partially, from the over-investment into technology during the preceding years. This seems to be true for the mature economies, not only the US, but less so for the emerging world.

It also seems to be the case, that the mid-2000's did not accumulate excess technology intensive capital. Thus the current crisis would be different.

At the same time, the length of this recession means that investment is halted for a long time. Perhaps that means a strong demand for the new technologies after the crisis, as well as another jobless recovery.


  1. This post is going against the previous post, which in turn was going against the one before. The snake bites its tail, and the World goes to hell.

  2. very interesting point. thanks. there is though an obvious further question: at what point will an emerging market start reacting jobless recovery style? there must be a reason beyond individual country stories for why this "natural phenomenon" is not there for developing countries.

  3. Some obesevations:

    1. indeed there seems to be some ambiguity here: would the level pre-crisis investment in technology then be relevant for recovery being jobless or not? (arguments in both directions are to be found in this and last post).

    2. the question arises that if the mechanism were to be indeed that simple, than wouldn't ALL recoveries have to be jobless, simply because..errr... time elapses? Going one step further, the theory would suggest that the longer the recession, the more jobless the recovery would have to be. Not sure whether this would be defendable.

    3. My instinct tells me that the decribed machanism might work only if there is a genuine technological shift that is truly sweeping.

    4. I would also expect sectoral issues to play a role in the extent to which new tech decreases demand for labour (mfg vs services etc).

    5. Could it not be that in recession companies are generally disinclined to cut workforce proportionally as production decreases, and so they'd have a workforce "buffer" once orders pick up again?

    Anyway, thanks for the interesting post.


  4. Oh, and I simply don't get this one:

    "However, the hypothesis does not necessarily require a previous over-investment, only the potential to increase production without new labour. Thus a lengthy halt to the technology-intensive fixed capital could do the same. If this was to happen so, there would be a fairly straightforward equity price consequence..."

  5. Thanks, this is fascinating: full two months ahead of time! But, please come back and clear up the sky by answering the questions.