How much productivity growth was there during 2007-2009?

Michael Mandel has a long and excellent blog post on this question.  He claims that the supposed productivity gains were concentrated in a small number of sectors (one of which, by the way, was financial services ha-ha) and that they are mostly illusory when cross-checked with other sources of data.  Here is his final conclusion:

However, the effect of the adjustment on the 2007-2009 period is spectacular.  Productivity growth, which had been 1.6% annually in the original data, basically disappears. The decline in real GDP is twice as large  (-1.3% per year in the original data, -2.9% in the adjusted data).  And economists are no longer presented with the confounding puzzle of why unemployment rose so much with such a modest decrease in GDP–it’s because the decrease in GDP was not so modest.  (see a piece here on Okun’s Law, which links GDP changes with unemployment changes).

His redone figures, by the way, are based on the assumption that intermediate inputs are growing and shrinking roughly at the rate of final product (Mandel believes we are mismeasuring these intermediate inputs and thus finding illusory productivity gains over that period.)  Think of his alternative numbers as illustrative rather than necessarily his best estimate.  The implications of his analysis include:

1. Productivity statistics aren’t well set up to cover outsourcing.

2. Beware of measured productivity gains, reaped over short periods of time, based on supposed drastic declines in intermediate inputs.  We’re probably mismeasuring those inputs.  Mike’s examples on these points are pretty convincing, walk through what he does for instance take a look at his numbers on mining: “Mining, for example, combines a 10% drop in real gross output with an apparent 46% drop in real intermediate inputs,  leading to a reported 23% gain in real value-added and a 26% gain in productivity.  It’s very hard to understand how intermediate inputs decline four times as fast as output!”

3. During the crisis, output fell more than we thought and thus our recovery isn’t going as well as we think.  (By the way, this is the most effective critique of the ZMP hypothesis, since the implied decline in true output now comes much closer to matching the measured decline of employment.)

4. Issues of “international competitiveness” are much more important than either economists or the Obama administration have been thinking.  Excerpt:

…the mismeasurement problem obscures the growing globalization of the  U.S. economy, which may in fact be the key trend over the past ten years. Policymakers look at strong productivity growth, and think they are seeing a positive indicator about the domestic economy.  In fact, the mismeasurement problem means that the reported strong productivity growth includes some combination of domestic productivity growth, productivity growth at foreign suppliers, and productivity growth ”in the supply chain’.  That is, if U.S. companies were able to intensify the efficiency of their offshoring during the crisis,  that would show up as a gain in domestic productivity.

5. Read #4 directly above, think about who captures those gains, and you can see that the Mandel productivity hypothesis is broadly consistent with some of the data on income inequality.

6. There really is a structural unemployment problem and it stems from ongoing low productivity growth.

7. At the risk of sounding self-congratulatory, if you combine Mike’s estimates with the new Spence paper, and the reestimation for male median wages (down 28 percent since 1969), in my view the TGS thesis is looking stronger than it did even two months ago when the book was published.


Tyler, your reliance on data massagers is laughable.

Seriously. How much did "productivity" grow? You want a number (a unitless number at that!)?

Tyler, you are a smart guy. Take a deep breath. Take a step back. Look at this discussion from the broad historical perspective. Can you not see that it is, in reality, retarded?

"His redone figures" made me laugh. This seems massaging indeed!

Note that Tyler is careful, as usual, not to endorse the ideas, but only to enumerate the implications. It seems reasonable he doesnt make his mind up quickly whether he he buys sometghing or not.

If hidden declines in output are a strong critique of ZMP, and if the drop in productivity thus indicated provides evidence for TGS, then is TGS actually compatible with ZMP?

This isn't one of the reasons I keep coming here. When confronted with the puzzle of rising productivity, neutral GDP, and rising unemployment, it simply wouldn't occur to most economists that one or two of the inputs have been mismeasured. It has occurred to Tyler, and he backs it up.

Theoretically, it could be that GDP and productivity are increasing, and it is unemployment that is being mismeasured or misreported. And it turns out that unemployment almost certainly is being misreported, but the evidence indicates that it is actually much higher than the reported figures! I think Tyler's explanation is more likely.

There is a good comment on the site that mirrors my feelings:

"it makes perfect sense to me that the proportion of intermediate inputs for computing has fallen. You simply assume that outsourcing replaces internal labor with foreign labor, when it’s probably more likely that they replaced US domestic outsourcing with cheaper and more efficient foreign outsourcing, which would perfectly explain why their intermediate inputs fell. You assert that the price drops aren’t taken into account somehow, but implicit in that seems to be an assumption that if a foreign company produces twice as many rotors at half the price, that the number for “real” intermediate input should also double. But I don’t think that’s what’s being measured at all, it’s simply the real cost of those intermediate inputs, after adjusting for overall inflation for all goods."

I followed the post up until the part about how the intermediate inputs can't be correctly measured, at which point I was lost.

I don't know how to put it in econ speak, but I worked for a company that was enamored with outsourcing prior to the recession. It required a near full staff of American expats to run the plant and a bunch, not sure how many but I'll say 3x foreign workers milling about as the Americans worked on the machine. I'm just saying that was my experience, not necessarily typical.

(My experience wasn't with China proper, but that was the typical experience in Asia, Eastern Europe, and Central/South America. And many of the locals were very impressive, but there are a hella lot of Americans on the phone every day talking to locals and expats keeping the places running.)

Your experience is atypical and an outlier. Maybe your product was something really complex or high-tech.

I have seen many automotive ancillary plants in India. I am not saying they are always as advanced or as high-quality as the American ones I have seen. But $-per-unit or $-per-quality-unit they are almost always better off for most low and middle rung items.

All of the anecdotes I've heard from other software developers agree with this. Though, it usually sounds more like educational/organisational issues than problems of people "milling about".

For example, someone told me that the firm that they use promotes all of their developers with more than 3 years experience to managers. Meaning, there are no developers with more than 3 years experience. Yikes.

This makes sense since productivity is a measure of the efficiency in the use of input relative to output, and if input prices decline, productivity increases.

So, does this mean that real wages for final assembly workers don't have to decline because we now source inputs from China? Does this mean that if the dollar declines we are headed for some real superinflation and/or dramatic decline in productivity?

It would be interesting to test the hypothesis by taking two similar businesses or industries, one with a high percentage of foreign input and outsourcing, and the other with high domestic content and little outsourcing, but with common characteristics--basically, industries which assemble or add value in the same way, using the same technology, etc., and see how productivity changed.

If the productivity of labor measured by price or by time? It's my understanding that we use time (length of time needed to produce one unit of output) so that the price of labor is irrelevant to the calculation.

I think you are right, although the Wiki entry on measurement of labor productivity states: 'Worker productivity can be measured in physical terms or in price terms.'

In looking at the BLS description of labor productivity, it seems that it is unit labor costs per output, so price and costs are in. Here is the BLS link:

I wonder why they do it that way? It obscures the crucial factor of productivity, which is how efficient labor it (not how much it costs). I can see that you might use labor cost to figure out labor time (assuming you know the pay rate per hour), but otherwise you have a number that doesn't mean much.

It's worth noting that mismeasurement in financial services productivity probably doesn't have much impact on aggregate productivity (there's a working paper, I thought from the NY Fed, that I can't find). Financial services are almost always an intermediate input, so overstating FS output simply understates productivity in other sectors.

Productivity = Output (final product) / Input (wages)

So if Input (wages) goes down either because you squeeze more out of American's for the same price because they are afraid of the recession, or you outsource to China and pay them 20% of what you pay Americans, "productivity" goes up even if you didn't improve the process at all.

You did improve the process. Outsourcing was the improvement.

Only if the people you outsource two are more productive. If you simply pay them less for the same or worse productivity your not doing anything but figuring out a better way to capture surplus for yourself.

"More productive" could be from two reasons for a worker: (1) Higher output or (2) Lower wages.

If a Chinese and American worker both assemble, say, 20 IPODs an hour but the Chinese guy asked for $2 and the American $20; this should mean that the Chinese worker was more productive? Both produced the same output but the input (wages) needed for the Chinese guy were lesser.

How does surplus enter this calculation?

Let's say it takes the American an hour to make 100 IPODs. With the same inputs is takes the Chinese man the exact same hour to make 100 IPODs. The Chinese man isn't any more productive then the American, the same number of IPODs are being made per man hour, but he's willing to accept a lower wage because his bargaining power sucks and he's starving. So the same value (IPODs x sale price) is getting produced, the pie didn't grow, but the capital owner gets to keep a higher percentage of it.

If the American used the time he had from not having to produce IPODs to produce something else, preferably something where he had a competitive advantage with the Chinese man, then maybe the pie could grow and you'd get genuine productivity growth (more output per man hour). But generally he used that opportunity to flip houses with his neighbors and bomb brown people, so the pie hasn't grown a whole lot.

Our argument is over whether productivity is (output per man hour) or (output per man dollars). I maintain that from the producing firm's point of view the latter is more important i.e. For every $ I pay a man how much does he produce?

The pie size can actually grow (though not always). Let's say the Chinese man works for fewer $; this translates to a lower Ipod price; ergo more people buy Ipods or the consumer surplus for Ipods increases.

I feel both definitions of productivity are valid viewpoints, it just depends which side you are sitting on.


Here's the relevant paper:"Implications of the Financial Crisis for Potential Growth:Past, Present, and Future"

Krugman had a post, including the following summary "What he finds is that even with fairly strong assumptions about phony financials and wasted investment, you can’t make more than a minor dent in growth estimates. On the financial side, the point is that we measure growth by output of final goods and services, and fancy finance is an intermediate good; so if you think Wall Street was wasting resources, that just says that more of the actual growth was created by manufacturers etc., and less by Goldman Sachs, than previously estimated. On the housing side, the point is that residential construction, even though it was at high levels, never got much above 6 percent of GDP. So even if you believe that a large part of the construction taking place late in the housing boom had very low usefulness, it only subtracts slightly from growth over the course of the whole period."

form the abstract: "Slowing
growth in the finance, homebuilding, and real estate sectors could hold back aggregate
growth. A detailed examination of these sectors’ direct contributions to GDP, however,
suggests that overstatements of past growth would likely not have made a large difference
in recorded GDP growth. Slower growth in these sectors would have, at most, a
moderate direct effect on aggregate economic activity. The recent experience’s longer
term effects on GDP would seem to stem largely from factors other than the retrenchment
in these sectors."

"7. At the risk of sounding self-congratulatory, if you combine Mike’s estimates with the new Spence paper, and the reestimation for male median wages (down 28 percent since 1969), in my view the TGS thesis is looking stronger than it did even two months ago when the book was published."

It's worth pointing out that the peak year for male median earnings was 1973, and they have declined 32% since then.

Focusing on "median wages" is puzzling because it neglects immigration. You are not following the same group of people. It's not as if the same families are earning lesser in new generations; we have a lot of new immigrants coming in that keep doing the lesser income jobs.

E. Barandiaran's comments are typical illustrations of the deep rot in the economics "science."

Someone illustrates the huge problems with the way data are measured -- and those huge problems are related to huge theoretical problems -- and apparently professional economists can not be bothered with it.

Economists look more and more like politicians: as long as what they say _seems_ good and some people are fooled by it, who cares about being right?

For some time I have suggested Tyler to review the research work of academic economists that during the last 50 years have focused on productivity and econ growth. These economists have developed different approaches to overcome the many methodological and data issues to measure productivity over time and across activities and countries. I hope Tyler reviews in detail their work and reports the results, those that may support his TGS hypothesis and those that do not support it.

I don't particularly care about Tyler's TGS.

I do care that the chain of models from the raw data to the high-level models is full of poorly justified --theoretically and empirically-- models and I wish economists would not be so cavalier about those defects.

Perhaps you were just using Tyler's put down against him -- that would have been funny to me, but I did not know the context.

My comment to this Tyler's post is related only to his TGS idea and its authentication. He has been cherry picking evidence and data to support his idea. This is not the sort of empirical research I expect from a scholar. Other people --including economists that are not scholars-- may have other views about what knowledge is and in particular about how ideas can and should be authenticated. I respect all other epistemological views but scholars are not expected to rely on data massaging to authenticate their ideas.
If your point is that Tyler is not a scholar (perhaps because you like his posts on Mexican, Chinese and Korean food), I expect him to say you're wrong (at least in relation to his post on economics). Anything else you have to say is irrelevant to my comment.

Blogging has made economists more popular but less rigorous. A lot of economist-bloggers seem to keep some hypotheses on their pet-agenda and then week after week scan the literature for articles that will prop up that point of view. e.g. Tyler and TGS etc. ; I think this is counterproductive and adds so much more of the economist bias into the analysis.

"There really is a structural unemployment problem and it stems from ongoing low productivity growth."

Does that sentence really go together? If you recant on ZMP, recognize low productivity growth and acknowledge GDP growth wasn't as good as previously thought --- aren't we more or less back to a traditional this-time-it's-not-different-recession?

As I keep saying, look at the stock market and housing markets. They are both so far below historical trends it seems the simple answer is there just isn't enough money around to get domestic prices moving. Everyone is blinded by global commodities to think otherwise.

What in the world are you talking about? How are either the stock market or housing markets "so far below historical trends"?? The housing market is almost back to trend (from way above). The stock market is above in PE, is it not?

The stock market looks way below trend to me. I hope we don't revert to long-term historical average PEs. As a forward looking indicator, the stock market expects lower corporate earnings going forward than it did a few years ago. Tyler has pointed out a number of times that earnings are back to where they were pre-crisis, but current earnings aren't what's relevant, it's expected future earnings. Also, since interest rates are lower than pre-crisis, a lower discount rate should apply, implying expected future earnings are even less than the % difference in price, which is about 16%.

One might argue that stock prices don't always predict accurately -- which is obviously true -- but they are a very good predictor of what corporations are expecting, which is what they base their hiring decisions on. The correlation between the change in the unemployment rate and changes in the stock market is high. The last time the stock market was below its 11 year ago highs was in the late 70's and early 80's, when the unemployment rate looked similar to what it does now.

As for real-estate, we have a non-functioning market. You are correct that house prices are above their price trends -- I wasn't thinking price trend when I wrote the above -- though that's obviously what it reads like. Home sales is the relevant metric and they are at 48 year lows. The problem is that prices are too high and too sticky, so the market isn't clearing. The quickest way to get the market to clear is to lower real prices by raising the general price level, which the Fed is obviously doing but hasn't done enough of yet (despite what goldbugs think). Once people start buying houses and furnishings and durable goods to go in them, the economy will be more or less back to normal.

i suspect one could estimate the unemployment rate fairly accurately as a function of two variables:
1) change in stock market vs. its 10 year moving average; and
2) change in number of home sales vs. a 3 year average

Let me change my 2) to 10 year average as well.

"I hope we don’t revert to long-term historical average PEs"
I don't know how long it will take, the Fed will certainly fight it, but I'm thinking this is coming. Quite frankly, except for some legitimate post 70s bounce back I consider 1980-present pretty much one giant bubble in financial assets, a bubble that has hit a wall now that interest rates can't decline further.

If homes won't clear because A) Prices are too high and B) sellers don't want to then address those problems directly.

A) If prices are too high, probably 30% too high on average for bubble purchases, it would take unfathomable inflation to raise the general price level to the point where nominal housing prices rose to bubble levels. Moreover, the last place any of that new money is likely to go is housing. Every other price level is going to rise more then housing, and inflation will ultimately drive up mortgage rates and make it even more difficult. If the money drives up commodities faster then wages it will even have a negative income affect on potential buyers.

A better way to make housing more affordable just create money via a payroll tax holiday. That increases real wages and makes the price/income ratio more affordable. When you give new money directly to workers its a lot more likely to drive up wages then giving it to banks to speculate, which is all the Fed does.

Of course, there is a way simpler way to deal with nominal price mismatches then gamble on various money creation schemes...

B) If the principal on a loan is 30% higher then the market value then mod the mortgage principle so that its in line with the current market price of the house. Its better for all parties involved that way. The only people who don't want it are bank CEOs who get huge bonuses to run "extend and pretend", that's not an interest we need to protect as a matter of public policy.

Your PE ratio chart is backward looking 10 years. Ridiculous. Stock prices are forward looking. Schiller is a flake. I'm long as shit the market right now because I believe Bernanke will eventually boost the general price level to the point where the housing market clears and that the Dow will be around 15,000 at that point, probably sometime next year.

Bernanke may indeed raise the price level to such a degree that even with a correction in P/E ratios and a collapse of real prices of all sorts of things nominal prices continue to go up. That's a legitimate investment thesis. It doesn't necessarily make it good public policy. If the DOW is 15,000 but regular folks can't eat or heat their homes, its certainly a failure of public policy.

By what metric do you consider stock and housing markets below historical trend?

The following chart is my metric for the housing market:

What does the sales rate have to do with the price level?

If you look at the official data in 2007 & 2008 non-farm productivity growth was 1.6% and 1.0% while non-financial productivity growth was only 0.4% and 0.3%.

This implies that financial productivity growth was much stronger than the 1.6% and 1.0% rates reported.

So who thinks we were getting very strong gains in financial productivity in 2007 & 2008?

What is financial output and shouldn't it include "risk" management?

". . . think about who captures those gains . . . ." OK, I'll bite: who does capture those gains? When a single company develops a more efficient production process (whether involving foreign outsourcing or not), it--its shareholders--gain in the short run. In the long run, as the whole industry adopts the more efficient process, the gain goes to . . . whom?

This picture is much more how non-economists have viewed the transformation in the economy. We will continue to trust our intuitive observations about outsourcing as a better gauge than unintuitive economic models. What are economists good for again?

Keep your economist hands off my NGDP targeting.

When the current recession started, I remember watching a number of store fronts collapse (such as Circuit City): store fronts that were being systematically replaced by Internet presences, such as Amazon. The largest music retail outlet in the United States is now Apple's iTunes Store. And with the collapse of the housing market, it struck me that what we were seeing was not just a recession triggered by the housing market. Rather we were seeing a fundamental shift in our economy, one which the housing run-up had been hiding, and one which was just now being exposed.

When I first read Dr. Mandel's blog post, it seemed to capture and lay bare the sinking feeling that has been in the pit of my stomach, and I think you perfectly captured it with excerpt 4. There is a very fundamental shift going on in the economy, and one aspect of it is a greater internet retail presence knocking off certain key retailers: an internet retail presence whose maintenance work can all be exported to Bangalore.

Here's one problem I see: The transition from brick-n-mortar stores to online ones must have had a cost saving and efficiency motive. Unfortunately the surplus seems to have been largely tapped by the firms; e.g. Apple, Amazon etc. No wonder they are showing record profits and cash. Ideally the government would be finding ways to reallocate this surplus.

I for one save hundreds of dollars every year thanks to the presence of Amazon. Perhaps they make even more profit from having me as a customer, but I doubt it.

My intuition would be that competitiveness is higher among online retailers than among brick and mortar stores. Entry is cheaper for sure but maybe there are differences in economies of scale that could explain higher monopoly power among online retailers.

I think entry-barriers get higher for online stores. No way you can open a credible online book or music store to challenge Amazon or Itunes unless you stock hundreds of thousands of titles and mount a huge advertising onslaught. A multi million dollar investment.

OTOH I could conceive starting a brick-n-mortar store for a few hundred thousand dollars.

No way you can open a credible online book or music store to challenge Amazon or Itunes unless you stock hundreds of thousands of titles and mount a huge advertising onslaught. A multi million dollar investment.

Well no kidding. Just like you couldn't open a credible bricks-n-mortar store to challenge Wal-Mart unless you stock hundreds of thousands of titles and mount a huge advertising onslaught. Also a multi million dollar investment.

But just as you could open a small bricks-n-mortar store for a few hundred thousand, you could also easily open a small online store for even less. Geez, you could hop on Amazon or eBay's back and open an online "store" for not much more upfront cost then a computer and an internet connection. For example:

Sure, but housing prices are still too high in real terms. The practice of buying and selling houses has virtually come to a standstill and this isn't explained merely by structural unemployment. It is more easily explained by the failure of the money supply to keep up with the demand for money and therefore everybody has less money to exchange things with. Here's the chart of the day that matters:

The housing market is the best proxy for the economy. When people aren't buying houses and furnishings and other durable goods for those houses, there's not much economic activity going on in the country. That is why there is so little demand for labor.

When people can't afford houses they are less likely to plan families. This has all sorts of economic consequences, short term and long term.

Maybe they could rent a house?

But they're not:

Perhaps the same reluctance to sell at a nominal loss creates a reluctance to rent at a nominal loss.

You have a post just above where you argue houses are too cheap, now you are arguing they are too expensive.

No wonder I'm not making any sense...

In the post above I said the housing market was below trend, by which I didn't mean the price level but the number of sales, which is at 48 year lows on an absolute level and despite the fact the population is 50% larger now. See the chart I attached above.

I meant this chart:

A foolish consistency is the hobgoblin of little minds.

Re: #7

Can't help thinking of the Dale Jorgenson long-ago paper that should fastest GDP and productivity growth for US economy in the deep, dark Depressions days of the 1930s. Any light sheddable on comparisons?

Mandel is not making any sense.

You produce a $100 computer using one worker and one $60 intermediate good. The price of intermediate goods falls to $10. Productivity rises from $40 to $90. The economy does not care whether the drop in the price of intermediate good reflects technological advance in the U.S., or outsourcing from a U.S. to a Chinese supplier.

Again, is the price of the inputs the relevant factor, or the time needed to produce them? I've always understand that productivity measurse the amount of time needed to produce goods (e.g., one widget per hour), in which case outsourcing doesn't affaect the result, except insofar as the outsourced labor is either more or less efficient.

The important thing in life is to have a great aim, and the determination to attain it. (Johan Wolfgang von Goethe, German Poet and dramatist)

I suspect he may be missing the impact of falling inventories on his measure of intermediate products.

Normally inventories growth is roughly in line with production so it does not impact the measure of intermediate products. But in a recession firms cut inventories and this could significantly impact the value of intermediate products that his analysis depends on. If in the recession firms cut inventories by cutting their stock of intermediate products the measure of intermediate products Mandel calculates could be significantly distorted. I would like to see how Mandel handles this issue before I would give his work too much credence.

I do not know if this happened, but I would like to see him explain how he dealt with this potential data distortion for intermediate products that his analysis depends on.

Comments for this post are closed