« US Productivity Exceptionalism | Main | Does the rest of the world subsidize research on the American economy? »


Feed You can follow this conversation by subscribing to the comment feed for this post.

It may be a combination of the housing bust AND general human capital in construction workers. If they were expecting demand to return they would be a lot more inclined to hoard labour.

This story is more of a compliment to the idea that the decline in employment relative is due to structural shift and the zero MP worker in the face of depressed AD idea.

A couple more points may be relevant to your construction hypothesis:

1. Construction is dominated by small firms, at least compared to other goods-producing industries. Small firms presumably have less financial and institutional capacity to hold onto employees during downturns even if they would like to.

2. I believe that construction has experienced relatively little productivity growth over time. If a low-productivity sector contracts more than higher-productivity sectors, the result is higher average productivity.

Mark: Yes. I think that makes sense. It's a bit off both.

Erin: Good points. Your point 2 especially helps my explanation. I can explain why Spain's productivity *didn't fall*. But your point 2 could explain why it *increased*.

I still wonder if this could be at least partly due to the way GDP figures are calculated. Suppose you have two countries. Country A suffers a collapse in both demand and prices in the housing market. Country B does not. What will that do to the GDP deflator that is being used to correct the value of output to get "real" GDP? Both countries have the same inflation rates for all other goods, and the same output of all other goods.

If house prices have a weight in the price level being calculated that reflects historical housing consumption, the overall price growth calculated will be lower in A than B, if all other goods inflate at the same rate in both. So the inflation correction applied to A's output will be less, making it look as if productivity is higher than B's.

Or are they smarter about calculating those figures than that?

Paul: I'm not that great on calculating GDP deflators (despite teaching it just yesterday). But I think the chain index method they use should handle that. Hope someone else can give a better reply.

What about the possible effect of those who have lost their jobs simply having had lower than average income (if that's the case)?

Since GDP sums to an income equivalent, wouldn't that alone tend to increase productivity numbers?

Hiring practices are quite different than they were in 1962 when Okun formulated his law; could the rise in dependent (non-standard) employment also be a factor in the abnormal countries? Would companies be less likely to 'labour hoard' for jobs held by dependent employees?

OECD stats (http://www.oecd.org/document/53/0,3746,en_2649_33927_42788213_1_1_1_1,00.html) show dependent employment in the US rose from 4.5% in 1999 to 11.6% in 2009. In Ireland, it went from 4.9% to 11.6%. Part-time employment in both countries was relatively flat (2-3% difference). In Canada, the dependent employment was flat with a 0.4% change over the same 10 year period; part-time employment also showed little change.

The anomaly is Spain where dependent employment fell from 32.7% to 11.6% while part-time employment doubled, moving from 7.8% to 16.2%. Do companies relying on part-time labour, historically, 'labour hoard'?

"[Off-topic slightly: notice how the countries that had the worst recessions tended to be the ones where productivity increased? That doesn't look good for the theory that says recessions are caused by a negative productivity shock.]"

What about positive productivity shocks and other things that cause price deflation and not handled correctly in terms of the medium of exchange causing recessions because real aggregate demand is not unlimited?

I also believe you need to define productivity growth for what it is, a job killer. IMO, producivity growth and other things do not automatically lead to higher living standards for everyone.

anon: I already discussed that in the post, when I talked about lower-skilled workers suffering a disproportionate share of the job losses.

Janegca: that sounds interesting. Can you expand on what "dependent" employment means?

TMF: you are totally off-topic.

Your construction hypothesis sounds plausible, but some counterarguments:
1) Surely there are other sectors that use general human capital, which would dilute the construction effect.
2) Productivity in Ireland began to increase at the same time it did in Germany.
3) Didn't the UK have housing boom ?

Looking at your latest graph, I can think of an ever simpler explanation:
Productivity increased the most in countries that started with the lowest productivity.
That certainly fits with the extremes Germany (Grundlichkeit), Japan (Kaizen) and Spain (manana).

Anyway this is good news for the Euro: Spain has increased it's productivity relative to Germany with 8%. That means that the needed wage deflation in Spain is also reduced with 8 percentage point.

I don't see what is off-topic.

This post talks about GDP, employment, and productivity. It is this simple. If productivity growth is greater than real GDP growth, couldn't there be job losses?

Did you ever think that Okun's Law is just plain wrong?

I agree that construction is a low-productivity industry. In terms of innovation, we now have nail guns and prefabricated kitchen cabinets, sometimes even prefabricated homes (there is one such factory near where I live) but on the whole the construction industry is based on on-site methods which haven't changed much in decades. It's one industry that just hasn't innovated as much as other have over time.

Secondly, perhaps the Okun's Law violation really encapsulates how this recession unfolded. There are two categories of countries:

Category A: Those that had a real estate boom and then collapse. This is where the trouble really started. This led to a banking and credit crisis in those countries through mortgage defaults. These are the countries which have shown a productivity increase through reduced low-productivity home construction.

(I would also note that the decline in US real estate prices as of last month is now 26% from peak. That beats the 1928-1932 decline of 25.9%. Regardless of a tenth of a percentage point, the US collapse in real estate is Depression-sized. These are the bad old days.)

Category B: Those countries which did not have a banking or real estate crisis. The main transmission mechanism for the recession was a decline in terms of trade for these countries. This is what happened to Canada.

While the case of the UK may make it seem like it was a Category A country, it reality it was a Category B country. The recession started in the UK with the Northern Rock bank run which was causes when that bank had a liquidity crisis. That crisis was precipitated by the fact that Northern Rock couldn't secure market debt funding when investors re-evaluated its risk in light of defaults on mortgage-backed securities in the US. The mechanism here was B even though it was real-estate related.

Surely it would be better to look at hours than employment.

The OECD Employment Outlook view is that labour market institutions made a big difference. In Germany hours fell a lot, but employment hardly at all, due to formal work sharing. Ditto to some degree elsewhere in Northern but not Southern Europe.

Work sharing under EI partly explains the Canada - US difference as well - saved some 40,000 jobs (ma

@Nick Rowe: Apologies, I'm not an economist or an economic student so I can only explain what led me to asking the question. Not knowing what Okun's law was I went on a google hunt and on seeing he came up with it in 1962 started wondering about what might have changed in the labour force since then (assuming GDP is essentially the same). The biggest change seems to be in employee/employer relationships.

Found the OECD data and the variation in 'dependent employment' which it defines as:

"Under the OECD definition – as well as Eurostat guidelines – temporary jobs are those forms of dependent employment which, by their nature, do not provide the prospect of a long-lasting employment
relationship. The definition includes employees under fixed-term contracts, seasonal workers, temporary
agency jobs, and those with a training contract of limited duration."

The other definition I found for it was:

"The concept of 'economically dependent workers' refers to those workers who do not correspond to the traditional definition of 'employee'- essentially because they do not have an employment contract as a dependent employee - but who are economically dependent on a single employer for their source of income." at http://www.eurofound.europa.eu/eiro/2002/05/study/tn0205101s.htm

The noticeable change in this type of employment in the US and Ireland struck me as significant; especially as it was flat in Canada's data. Don't think this type of employment was a norm in the '60's when Okun formulated his law. The data he used would have been based on employment models used during his time, and earlier and so his rule might not apply given existing labour practices.

Surely it would be better to look at hours than employment.

Yes it would. Unfortunately, the OECD base I was using only has annual data, and the last observation was 2009.

What happened to Frances' piece? It's still on my reader but I was looking for the interesting debate that would have surely followed...

Nick's guess: a big drop in construction employment...and those 3 "abnormals" just so...labor intensive in the run up...knitting their houses together?

Absolutely. Financial knitting.

In those countries, MEW or similar allowed not only extra housing in the domestic US, but in those destination areas for those foreign cottages.

It takes ~6 months to build a house and 20 years to pay it off...so, how does the product get registered in the GDP tally? Is it uniform across this OECD spectrum?
The financial component of this product: Is there any doubt that whenever it is registered, it is dominated by US origins...partially accounting for this sunny-looking US GDP, no?

Nick, great post! Have you seen Mark Thoma's latest post: http://economistsview.typepad.com/economistsview/2011/01/capacity-utilization-and-unemployment.html
and the graph, which shows that the output gap in the US increased during the last decade without empoyment decreasing, but then the relationship reversed during the Great Recession. I think that supports your theory about the construction industry. I also think one can generalize your theory along the lines of Andrew Jackson's comment: The US now uses dependent workers, called contractors in the US, in all industrial sectors, not just construction, but especially in construction. During the last decade, many US workers migrated into contract work (e.g. professional services), where they mostly provide services that are temporarily expendable. As a result, corporate earnings skyrocketed at the end of this recession. Conversely, German industry participates in government-mandated vocational education, which builds worker solidarity almost as strong as in Japan. Of course, Germany also subsidizes worker retension in recessions.

Hasn't the US exhibited this since the 90 recession? It wasn't about construction then. It may be large companies have been offshoring with no intent to rehire while small companies don't have the capital depth to sustain workers through one particularly when they are at the center of it. It may be just that these countries are at the center of the recession, just as the US was at the center of the tech wreck.

ws: "1) Surely there are other sectors that use general human capital, which would dilute the construction effect.
2) Productivity in Ireland began to increase at the same time it did in Germany.
3) Didn't the UK have housing boom ?"

1. Yep. My guess is that construction is the most important (in this recession?). I could be wrong.
2. Yes, but Ireland continued into recession, while Germany began to recover. It's normal (Okun's Law) for productivity to rise during the recovery. It's abnormal for productivity to rise as the recession deepens.
3. House price boom yes. Construction boom, not anywhere near as big, I think. (There's nowhere left to build, in England).

Jeff: I know that Okun's Law is wrong. That's what this post is about. But why is it sometimes wrong and sometimes right? What's different?

Determinant. Plus, in some countries (UK) the house price boom and crash couldn't translate into a construction boom and crash because of planning restrictions. I think a big percentage of Spanish and Irish employment was in construction.

Andrew Jackson: "Surely it would be better to look at hours than employment."

Yes and no. The production function should have a relationship between GDP and hours, not number of workers. But the empirical relationship known as Okun's Law was between GDP and number of workers. If an existing empirical relationship breaks down, we are always interested in knowing why. Plus, sometimes we are just interested in number employed in its own right.

Janegca: what you have found is interesting. Doesn't fit the facts perfectly though, but it may be part of the story. I had never heard of that concept "dependent employment" before.

calmo: what is "MEW"? I think (especially if it's OECD data) that the GDP figures of all countries would measure housing the same way. The value of newly-constructed houses, with the standard chain index to adjust for inflation. As far as I know.

Thanks Detlef! I hadn't seen Mark's post. Thanks for the link. But his is about capacity utilisation and unemployment. I have never really understood capacity utilisation figures, and what they really mean.

Lord: Brad DeLong's post (linked above) has a chart that shows something similar for the US in the 2000 recession, IIRC. But that recession wasn't big enough to see it clearly.

Guillame: I think it's coming back very soon!


I thought that was the case, that in the UK there was little construction due to planning restrictions/lack of space, but I didn't have any real evidence to put that in my post.

The Home & Garden Channel used to show a British series about first-time real estate buyers. The prices people paid for homes in the UK blew my mind, especially after I converted the sum to Canadian dollars.

I joked that now I knew why my ancestors got on the boat. Actually that's not far from the real truth.

MEW is not only the sound that a cat makes, but the hiss coming out an economy that has been ransacked by financial arms that persuaded many people in the US and Ireland and Spain and Greece and...to not just let the equity in their homes just sit there rotting away.
And, in contrast to Lord who I ordinarily defer at the slightest provocation, believe this *did* have its center in the US...but concede that it prolly was connected to the previous tech bust (powerful interests being powerful enough and not altruistic nor bored enough to channel Mother Teressa).
The US GDP numbers to my knowledge have not been rebenchmarked to reflect the decline in house prices,(how was this handled in the 30s? let alone all those fluffy jobs of walking the dog, the cat, the snake...part of the great job dislocation from those golden days watching your house work for you.

Being the center of the recession, layoffs would be indiscriminate in those areas because there is no expectation of recovery any time soon, so in 2000 it was in tech with employment still down in it and this time in construction for the foreseeable future. Where the boom is strongest, the bust is most severe (though China isn't doing too badly). If instead we had an energy bust, Canada might not be doing so well though it would have a much stronger customer.

calmo: Just a note on how real GDP is calculated: Each year (actually each quarter, I think) the price index is updated. So 2010 output is evaluated at 2009 prices, and compared to 2009 output evaluated at 2009 prices, to see if real GDP went up from 2009 to 2010. And 2009 output is evaluated at 2008 prices, and compared to 2008 output evaluated at 2008 prices, to see if real GDP went up from 2008 to 2009. And so on, in a chain.

It's not like the CPI, where they only update the basket every 8(?) years.

Someone who understands this better and can explain this more clearly than me, please restate and correct me.

But the US also experienced rising productivity during the previous recession as well and that was caused by the bursting of the dot com bubble. How many construction workers were laid off in 2000 and 2001?

Randall: I don't know. But the previous recession was much smaller. It's not as easy to see the effects of something when it's small, because those effects can be swamped by other noise. It might be, for example, that workers in the dot.com sector had higher productivity, so when their employment fell average productivity would fall.

Nick: "It might be, for example, that workers in the dot.com sector had higher productivity"

God help us! I thought sucking up investment capital to create the stuff of boundless *future* consumption (petabit fiber-optic switches, Garage Mahals in the desert) was a hallmark of bubbles.  Could it be that pets.com was skewing *current* productivity upwards?

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad