Friday, August 03, 2012

What happened to Japan?

Sorry, dude, no one looks good in skinny jeans. Dig the curly mullet though.

Since I'm in Japan, I had better do some Japan-related blogging. Brad DeLong poses a question, originally asked by one of his commenters:
Japan is now 40 to 50 percent below what the world in 1991 would have estimated their GDP to be in 2012. 
Do we attribute this to: 
The forecasting community was just wrong--Japan was having adverse technology shocks that few foresaw, and so no matter what macro policies they followed and no matter what antibubble policies they followed their GDP today would be about what it is, for the prevailing potential estimates back in 1991 were just wrong? 
If Japan had avoided its bubble and the resulting financial crisis, it would today have far higher GDP--but once the crisis happened, it ruined into an adverse supply shock and most of what has happened since was then predestined. 
If only Japan had followed the Posen plan rapidly after their bubble burst, their world would be very different today and GDP in Japan would indeed be 30-40% higher than it is. 
What is the best way to think about this?
Well, I can't answer this properly, because I don't really know what the forecasting community was smoking saying back in 1990. But I have a few thoughts.

Basically, by 1990, Japan had caught up to the richest large nations in terms of per capita GDP. The only way for Japan to have continued at its previous high rate of growth post-1990 would be for either A) an unprecedented technological boom to power a rapid expansion among all the world's rich countries, or B) for Japan's productivity to significantly exceed that of the other rich countries. In other words, anyone who forecasted continued rapid Japanese growth in 1990 was predicting that Japan was capable of doing far better than the other countries of the world, and indeed that this was the most likely outcome.

Now, let's look at Japan right now. Wikipedia tells us that Japan's per capita GDP, in market-exchange-rate (sometimes called "nominal") terms, is $45,900. That compares with $44,500 for Germany, $43,100 for France, and $39,600 for the UK. I picked Germany, France, and the UK because these are other rich developed nations with populations between 50 and 150 million and growth rates similar to Japan's. In other words, in nominal terms, Japan is richer, per person, than any comparable country. I suspect that what difference exists is due to labor inputs, since Japan does not force its citizens to take lots of time off of work the way Germany and France do.

At purchasing power parity, the numbers are a little less favorable for Japan - $34,300, compared with $38,400 for Germany, $35,900 for the UK, and $35,000 for France. However, I am suspicious of these numbers, since PPP does not take into account quality differences between similar products across countries. Japanese consumers have a famous preference for quality. Japanese plastic kitchen wrap, for example, is so strong and easy to use that it doesn't even deserve to be mentioned in the same sentence as Saran Wrap. Japanese pens last longer. The cheapest, lowest-quality Japanese milk is what we Americans call "organic milk" and pay a price premium for. Etc. (To be fair, Japanese houses are built of shoddy materials, have terrible floor plans, and don't have central AC.)

So the "real" level of Japanese GDP, if such a thing can be said to exist, is probably somewhere right around that of Germany, France, and the UK. "Lost decade" or no, Japan in 2012 is right where the Solow Model says it should be. 

What could forecasters in 1990 have been smoking that made them see this as anything other than the inevitable outcome? I suggest two things: 1) dumb trend projection, and 2) attribution error.

Dumb trend projection is people's tendency to view trends as structural. Examples of this include: "Housing prices have never fallen; hence they will never fall." "This stock returned 17.2% over the past three decades; hence it will continue to do so." "The center of gravity of the global economy is inevitably shifting to Asia." And so forth. But in reality, past performance is no guarantee of future results. Or, to put it more pithily, "The trend is your friend til the bend at the end."

Attribution error is our tendency to attribute phenomena to the wrong causes in certain reliable ways. One of these is that we tend to attribute outcomes to fixed person-specific characteristics rather than to circumstance and situation. In Japan's case, this means that people thought that Japan was growing fast in the 70s and 80s because of Japanese culture, superior Japanese government, or - and I suspect that this was more significant than people will admit - the inborn superpowers of the Japanese race.

Whatever superpowers Japanese people may have (I have found that they tend to be extremely good at "Where's Waldo?"), it did not turn out to give them a decisive productivity advantage over the Germans, French, and British. I guess in 1990, Japan was the only East Asian country ever to have gotten rich, and so there was perhaps more cause to believe that maybe East Asian countries could do things European countries couldn't. There are few who now believe that South Korea will blast through the productivity frontier. 

So maybe we learned something from the Japan episode. We learned that East Asians do not have superpowers. They have not figured out a better economic model any more than they have found a way to look good in skinny jeans. No one looks good in skinny jeans. And no one beats the Solow Model.


Update: Ryan Avent says I'm completely wrong. Market-exchange-rate GDP numbers, he says, are distorted by currency fluctuations; the yen is strong right now, so Japan looks rich. And this is true, although the Big Mac Index, produced by Mr. Avent's employer The Economist, says that the yen is not currently overvalued. My point is that neither market-exchange-rate nor PPP numbers tell the whole story. Mr. Avent does not address the limitations of PPP calculations, which go far beyond what I wrote about in this post. I still maintain that when trying to make cross-country comparisons, you should look at both numbers.

Mr. Avent is trying to defend his thesis of a "Japanese tragedy" caused by tight monetary policy. I agree that tight monetary policy has hurt Japan to some degree, and I'd draw attention to the fact that Japan's best post-bubble performance (in 2004-2007) came on the heels of a program of Quantitative Easing. But  that does not change the fact that Japan's standard of living is comparable to that of the major Western European economies (Australia is not an appropriate comparison, being a sparsely populated economy powered by abundant natural resources). And in addition, during the worst years of the "Lost Decade", Japanese unemployment was only 3% higher than at the peak of the bubble. So yes, there have been policy mistakes here, but the end of catch-up growth was no "tragedy", it was inevitable.

Update 2: On Twitter, my advisor Miles Kimball suggests that I look at the change in the yen's real exchange rate in order to better evaluate Ryan Avent's claim about market-exchange-rate GDP measures. Here, via Wikipedia, is a long-term (but up-to-date) chart of the yen/dollar real exchange rate:


Ryan's claim is that changes in this exchange rate distort Japan's market-exchange-rate GDP (making PPP the only good measure). But from this chart, we can pretty clearly see that he's wrong. The yen is currently at or weaker than its long-term average value (when the blue line is higher the yen is weaker). There has been no trend of yen strengthening since the mid-80s (when Japanese growth was robust). Hence, neither the trend nor the level of the yen exchange rate indicate that market-exchange-rate GDP is a worse measure than PPP GDP when discussing Japan's wealth relative to the United States.

But what about the European countries? If the pound, deutschemark, and euro have gotten steadily weaker since 1990, then their market-exchange-rate GDP numbers are "too low". But this is also not the case. In other words, Ryan is not right...exchange rates are not distorting the comparison between Japan and other rich countries. Japan is right up there with the richest large countries in the world.

33 comments:

  1. For me, Japan (like many other modern things) is a story about debt, and deleveraging traps.

    Japan hit an excessive level of debt relative to GDP and industry at the peak of the bubble. As debt rises, debt servicing costs rise, leaving less income for investment, consumption, etc.

    Throughout the lost decade, and indeed the years that followed, total debt levels (measured in GDP) have remained consistently high. Simply, the central bank has not devalued by anywhere near enough to decrease the real debt load, but nor have they devalued too little to result in a large-scale liquidation episode. They have just kept the economy in stasis, with enough liquidity to keep the debt serviceable, and not enough to really allow for severe reduction. The main change has been a transfer of debt from the private sector, to the public sector.

    If Japan had deleveraged faster, who knows what their example would say about diminishing returns and Solow-Swan. We can't really say how much or how little productivity or growth they would have gained from the additional technological and entrepreneurial investment they would have gained from having lower real debt levels.

    My hunch is that there is no real reason other than excessive debt levels why Japan could not have continued to grow above and beyond the examples you raise. Economic development seems to gain a lot from momentum, something Japan had a lot of before the bubble burst, and which they have lost a lot of in the intervening years.

    ReplyDelete
  2. I took a crack at explaining Japan's decline a while back. The conclusion to that post:

    So... a decade before the end of Japan's economic miracle, the fabled bureaucracy that drove Japan Inc. started to erode - more of its members started coming from lower quality, private universities, it stopped being held in as high esteem by the public, it lost its ability to impose its will on the economy, and to boot (or perhaps I should say on a related note), it started to adopt new policies and philosophies being pushed by the Reagan administration. How long do you think it takes for something like that to have a long-term, possibly irreversible effect on an economy that used to be the envy of the world?

    ReplyDelete
    Replies
    1. "fabled bureaucracy that drove Japan Inc."

      The following paper says it is in fact fable.

      http://www.cirje.e.u-tokyo.ac.jp/research/dp/2002/2002cf177.pdf

      Delete
    2. And, as always, Krugman pointed it out a long time ago...

      http://www.pkarchive.org/crises/perspire.html

      Delete
    3. himimaginary,

      The Krugman post said only this about Japan: "I visited the MOF in 1985 and saw what looked less like the Pentagon's War Room than like the Department of Motor Vehicles"

      In my post I cite a few different sources that indicate a rapid deterioration (depending on the source, and the factor deteriorating) beginning in either the mid-70s or early 1980s. So the fact that Krugman saw a mess in 1985 would not be inconsistent with the documents I cited or the conclusion I reached.

      As to the first document you cite, have you read it? They indicate that the bureaucracy lacked the legal tools to accomplish what people have claimed they pulled off. But that isn't a surprise. The big surprise to westerners was how much influence the bureaucracy had despite the fact that none of it was codified. People were similarly surprised about how the zaibatsu/keiretsu companies worked together too.

      Delete
    4. "The Krugman post said only this about Japan:"

      You dropped the most important part:
      "Then things started to go wrong, and the MOF proved itself as hapless in action as it was in appearance. It's easy to look competent in a prosperous economy (ask Bill Clinton), but the true test is whether you can cope with adversity. So much for the legendary managers of Japan Inc."

      And the sources you cited depend largely on "Todai-myth", and ignore the possibility that it's just that the other universities' quality was catching up that of Tokyo university.

      "The big surprise to westerners was how much influence the bureaucracy had despite the fact that none of it was codified. People were similarly surprised about how the zaibatsu/keiretsu companies worked together too."

      Professor Miwa is famous for debunking those conceptions as pure myths (although maybe the paper I linked was not clear enough about his position on this matter). As for keiretusu, he wrote the following paper:
      http://www.cirje.e.u-tokyo.ac.jp/research/dp/2001/2001cf109.pdf

      Delete
    5. himimaginary,

      You are misreading what I wrote. I noted it wasn't a surprise that the bureaucracy wasn't able to react competently in 1985 given it had been so effectively gutted by that point. You might as well be arguing that Mark Spitz wasn't much of a swimmer since you didn't see him on the podium yet in London.

      I cannot comment on the paper you cited as I don't have time to read it now. Apologies.

      Delete
  3. Hey Noah, great post but a couple of things I wanted to point out:
    - a Japanese once told me that the government creates national holidays in order to force more workers to take a break; those workers then need to make up those lost hours on the weekend!
    - I've tried many varieties of cow's milk in Japan and all smelled funny when opened and tasted nothing like as good as British cow's milk. I admit to not having tried milk in the US for fair comparison with your post ;-)

    ReplyDelete
  4. Noah, I entirely share your views. This is fundamental stuff, and I strongly encourage you to develop related issues further:
    - did we reach the steady state?
    - can cultural differences explain non synchronized cycles (say for example if the American is creative, and rather good at innovating, and the Japanese is more disciplined and technical and thus rather good at producing)?
    - can we use observations of the past to make predictions on the future?

    ReplyDelete
  5. (this is an edit of a comment left on Delong's post)

    You can explore the question by comparing Japan with Switzerland, that shares some features of Japanese economic culture. Switzerland experienced a crisis similar to Japan in the 90's (10 years of ultra-flat GDP, and then got back to an average 2.5% growth rate in the 00's.

    The recovery in Switzerland was most likely due to
    - free trade and free immigration agreements with EU,
    - economic ties with booming Germany
    - overvalued currency in south eurozone
    - luck (sectors in which Switzerland specializes were first booming, and then not much hit by the crisis)
    Japan had nothing of the sort.

    So, Switzerland, like Japan, may have reached some kind of steady state, and they can only grow at the pace of technological progress and population growth. Less advanced economies grow at a faster rate, but as they catch up, the marginal productivity of capital diminishes.
    Bubbles, cycles and delayed price adjustments may generate some overshooting, but apart from that, that's just plain classical Solow growth.

    In the 90s there was a fashion for endogenous growth theories, but even at the time, empirical evidence was very weak (and the econometric tools -time series- used for proving there was no convergence were methodologically not adapted).

    It should not be very complicated to test convergence again, and I'm certain that convergence would hardly be debatable now. And this is why the question of Japan is essential: if there is a finite prospect for growth, how does the economy work when it has reached the steady state? Does it generate inequalities? Are people putting all their energy in rent-seeking?

    Those questions are more relevant now than ever, because long term rates in most advanced economies are nil in real terms. Does that suggest that markets believe we reached the steady-state?

    ReplyDelete
  6. I expect the rapidity of the trend had a lot to do with it.

    ReplyDelete
  7. "No one looks good in skinny jeans. And no one beats the Solow Model."

    This must be one of the best lines I have read on convergence!

    ReplyDelete
  8. Zorblog,

    there are test of endogenous growth theory that are quite favorable (Zachariadis, 2003; Ha and Howitt, 2008).

    Keep in mind, however, that convergence in growth rates is consistent with endogenous growth. See for example Howitt (2001), and Ertur and Koch (2011). The reason is that knowledge has no borders. If ideas flow, then the whole world will eventually benefit from the expansion of the technological frontier by countries like the U.S. or Germany, so in the long run everyone should converge at the same growth rate. Regarding the level of income, just like the Solow model, endogenous growth predicts conditional convergence. Countries that invest less in R&D or education will be further away from the technological frontier since they will adopt the new technologies with a greater lag (but that distance will be stable, everything else being the same).

    Everything put together, it seems that in the end the growth process will resemble that in Lucas's (2001) "Some Macroeconomics for the 21st century". Countries that have the "right" economic environment will be able to harness the full potential of technical progress that was unleashed around the industrial revolution and will move away from other economies. Then, as those who are left behind reform, they will join one-by-one the cluster of leaders. There may still be differences reflecting the peculiarities of each economy, but these will be small.

    ReplyDelete
    Replies
    1. Of course, there are some goods and services with positive externalities (which makes a strong for gov spending, by the way), but overall, the impact of diminishing returns is stronger.
      What really hinders growth in developing countries is corruption, crime and political violence. Once you factor that into your regressions, evidence of convergence becomes overwhelming. It was already the case in the mid 90's, and I would expect that is even more evident now, with a decade of record growth in BRIC and other emerging economies.

      Delete
  9. Anonymous7:36 PM

    You had me laughing at "inborn powers of Japanese race" because in most european countries people still have somewhat similar opinion on Japanese and Korean societies.
    Having lived and worked in Korea I find those things damn funny. True, they have impressive work ethic but they also have a big productivity problem. Rigid deference to people higher in hierarchy, endless meetings that produce nothing, pay base on seniority, workers surfing on smartphones while waiting for their boss to go home so they could go after him. I found the whole thing absurd(only the working part, I loved almost everything else).
    After my one year stint I went to Vienna, to work for the same MNC. To keep it short, 40 hour work week in Vienna equates to 70h work week in Seoul.

    About Japan, why does nobody mentions demographics?
    Japan’s population in median age terms is the oldest on the planet and it's getting worse. To offset the worsening of population dependency ratio Japan will eventually need higher growth than most developed economies just to maintain the same standard of living.
    Reader from Croatia

    ReplyDelete
  10. dilbert dogbert11:48 PM

    "No one looks good in skinny jeans."
    I kinda like the way young girls butts look with skinny jeans. That said, please remember just because you are wearing "relaxed fit" jeans your butt still looks fat (huge).

    ReplyDelete
  11. This post assumes that there isn't something puzzling about European growth rates over the past 30 years, and I am not so sure that is true.

    ReplyDelete
  12. I hope you will post on the aging and shrinking population and the effect on the economy and on social security. When you say that Japanese houses are built of shoddy materials are you talking about their non-concrete worshiping traditional houses or the modern ones? You are completely wrong about skinny jeans which look good on short people and even better on short skinny people.

    ReplyDelete
  13. Anonymous11:01 AM

    Go back to 1985, before the bubble, and Japanese GDP per capita (PPP) was three times Korea's. Now it looks like Korea will overtake Japan on the same measure by 2014... and while Japan is definitely comparable to much of Europe in terms of GDP (PPP), Japanese do work substantially more than, say Germans.

    Another comparison, in 1990, Japan had lower suicide rates than Norway -- now Japan has much higher suicide rates...

    True, Japan has not turned into Sub-Saharan Africa, but the real question is whether Japan has done better than Japan could have done with a better monetary policy. You need to argue that all those times the Bank of Japan raised interest rates with inflation below 1% just before the onset of those recessions, mild as they were, was the right policy.

    ReplyDelete
  14. Noah: " I suspect that what difference exists is due to labor inputs, since Japan does not force its citizens to take lots of time off of work the way Germany and France do."

    I'm sure that those salarymen just love those long hours.

    Seriously, why would you look at Japan and talk about what Germany and France 'force' people to do?

    ReplyDelete
    Replies
    1. Well, think of taking vacations as a coordination problem...

      Delete
  15. "No one looks good in skinny jeans." And turn the music down.

    There's a recent felix salmon post about a large proportion of Japanese voters (the elderly) favouring deflation over growth. Sadly as an old fogey like you I can't do links.

    ReplyDelete
  16. Some points about the post:

    There are three separate factors at work here:

    1) Long Run
    2) Cyclical
    3) Feedback from Cyclical to Long Run

    1) Noah's right in a sense about the long run. Japan's labor productivity (GDP per hour) growth was slowing down sharply before 1990 and has held close to 70% of the US level since 1990. Japan's nontradable sectors such as retail and wholesale trade and food service are notoriously inefficient and this is unlikely to change. Hours worked per worker has also followed a long term trend downward. This accelerated during the 1990s and may be Lost Decade related. Labor force participation started to decline in the 2000s due to demographics.

    2) Ryan Avent is right about the cyclical factors. Japan's output gap grew from 3% of GDP in 1990 to 12% of GDP by 2002 according to BOJ estimates. Noah makes much of Japan's low unemployment rate but it doesn't tell the whole story as the natural rate of unemployment is far lower than in the US and there are far greater ties between employer and employee in Japan than in the US so hours are cut far more before workers are let go (a form of informal work sharing). The output gap narrowed by 2007 but is now large again, as it is almost everywhere.

    3) There is some research that claims that constraints on corporate lending since 1990 have slowed the rate of productivity growth in Japan. In Japan corporate debt is a bigger problem than household debt. Corporations are very dependent on banks for financing. Banking is also highly inefficient in Japan and loans usually involve a lot of nepotism so the least efficient corporations have been securing the finance since 1990.

    ReplyDelete
    Replies
    1. Mark,

      but how is (3) feedback from cyclical to Long Run? These are structural issues, so by definition long run. Morover, were these inefficiencies absent during the miracle years? If not, why didn't they matter? Finally, how does one go about finding what the natural rate really is? I say, if for over twenty years your unemployment rate has been hovering around x%, then that is your natural rate!

      Delete
    2. During the miracle years firms were not financially constrained. The OECD has time series estimates of the natural rate of unemployment (NAIRU) among other sources.

      Delete
  17. Anonymous6:51 AM

    Incidentally, the commonly used PPP adjustments(cooked up by the OECD?) are way off, in my view. As Noah, says the yen is not overvalued at 75-80 - check ou not just Big Mac, but minimum wages, hotel rooms, anything else.

    The idea that the South Korean standard of living is on the brink of overtaking Japan's is likewise way off.

    ReplyDelete
    Replies
    1. Anonymous11:27 PM

      OK, then look at Hong Kong, Taiwan, and Singapore then. In 1985 (before bubble started), Japan was richer than all three. Today, all three are substantially richer than Japan.

      And South Korea is on the brink of overtaking Japan. In 1985, their GDP pc was 1/3rd japan's.

      Delete
    2. Richard5:19 PM

      HK and Singapore have caught up, but I have family in Taiwan and my wife is from Japan, so I speak from personal experience when I say that the standard of living for someone with the same type of job (say, a schoolteacher) isn't really the same between the 2 countries. Taiwan is still quite a bit more primitive.

      Delete
  18. Anonymous1:29 AM

    Er - no - South Korea is NOT on the brink of overtaking Japan. As mentioned, the PPP adustments are flawed and market rates are close to fair value. On a per head basis Japan is TWICE as wealthy as South Korea and Taiwan too, for that matter.

    ReplyDelete
  19. Anonymous12:03 PM

    Unfortunately the chart at the bottom is completely wrong, seeing as the exchange rate now is 78 yen to the dollar and in 1970 it was 300...

    ReplyDelete
  20. Anonymous12:04 AM

    ah, real exchange rate...

    ReplyDelete
  21. You remind me increasingly of Lutz.

    ReplyDelete
    Replies
    1. Coming from most people this would worry me, but somehow I feel like this is a compliment...

      Delete