Thursday, March 31, 2011

John Taylor draws a Phillips Curve

On Tuesday, Greg Mankiw linked to a two-month-old blog post by John Taylor of Stanford University. Taylor posts the following graph, which shows a striking inverse correlation between fixed investment (as a percentage of GDP) and unemployment: 

The obvious conclusion is that investment is very important for the business cycle. Should we be surprised? No! It's common knowledge that although investment and consumption are both correlated with GDP, investment is far more volatile - in other words, investment makes up a very outsized portion of business cycle swings. And in fact, this is exactly what theory would predict - since capital goods are durable, it's relatively easy to shift purchases of capital goods (i.e. investment) into the future or into the present. 

(Paul Krugman points out that most of the recent collapse in fixed investment came from housing...sure, that's true, but nonresidential investment is also highly procyclical and highly volatile.)

So Taylor is pointing out something that everyone already knows. But here's the conclusion he draws:
Encouraging the creation and expansion of businesses should be the focus on government efforts to reduce unemployment. The recent compromise agreement to prevent the increase in tax rates on small businesses and the move to lighten up on the anti-business sentiment coming out of Washington are two steps in the right direction. 
Sorry, but this just doesn't follow. Taylor seems to have in mind an RBC-type model, in which the business cycle is driven by supply shocks. But his graph in no way implies that we are living in a supply-driven world, because demand-based models also imply that investment and unemployment are inversely correlated! 

This is a clear result of the standard New Keynesian macro models, for example (see Woodford 2005). In these models, investment is low when aggregate demand is low - if customers are buying even a little fewer of the goods that firms sell, it makes sense for firms to delay much of their purchases of the capital equipment that they use to produce their goods. The way for policymakers to boost investment, therefore, is to boost aggregate demand, by printing money and buying stuff - or, in some models, by having the government borrow and spend money. Thus, Taylor's graph is perfectly consistent with a world in which expansionary government policy is exactly what we need.  Paul Krugman also basically makes this point.

Anyway, thinking about investment in the context of New Keynesian models made me realize something. John Taylor is just drawing a Phillips curve! A Phillips curve is a curve that shows an inverse correlation between inflation and unemployment. And in New Keynesian models, inflation and investment go hand in hand - when the economy is booming, prices rise. So a New Keynesian would expect to see something just like what Taylor drew.

And then I remembered something else about Phillips curves - they shift over time. Policy shifts and structural shifts in the economy will, over time, change the tradeoff between inflation and unemployment. If you look at data over more than a decade, you see a series of different curves for different time periods:

I then noticed that John Taylor's curve covers only 1990-2010. If Taylor's curve is in fact just a Phillips curve, then if we extended the time period we'd expect to see a series of several nested curves, with the 70s and 80s being farther to the upper right.

Which is, in fact, exactly what we see! Justin Wolfers does the leg work and comes up with this:

Looks familiar, right? Somehow I'm guessing Taylor probably didn't set out to redraw the Phillips curve...but that's what he's done.

To reiterate, Taylor's graph is perfectly consistent with a world in which demand shocks drive the business cycle and demand-side policy is the key to stabilizing investment. His supply-side conclusions are purely a function of his assumption that we live in an RBC-type world. But there are good reasons to think he's wrong. For an argument as to why New Keynesian models are better than RBC models, see this 1989 paper by...Greg Mankiw.

Update: The data support the demand story. Not exactly a huge surprise...

Update 2: For a great primer on Phillips curves, see Brad DeLong.

Wednesday, March 30, 2011

Tyler Cowen buys straddles on humanity's future

Just a quick post here...I've been working hard on dissertation (let's hope Miles Kimball and Uday Rajan are reading this sentence), and haven't had much time for bloggage. But a little gentle needling of Tyler Cowen doesn't take too long. :)

Cowen links to a piece by Mike Mandel, which claims to show that most of the reported "productivity growth" of this recession was actually just a mistake. The idea is that productivity has been mismeasured, and has actually stagnated or fallen in the last few years; that would make this recession more in line with previous episodes, and less of a weird "jobless recovery." For a critique of Mandel's piece, see Karl Smith. (For the record, I agree with Smith on this one - Mandel appears not to consider the fact that GDP can be measured by income, not just by value added, and hence mismeasurement of intermediate goods would not, as Mandel claims, result in mismeasured productivity.)

But my point is of considerably less import, and regards Tyler Cowen's reaction to Mandel's piece. Cowen writes:
By the way, this is the most effective critique of [my] ZMP [Zero Marginal Product] hypothesis, since the implied decline in true output now comes much closer to matching the measured decline of employment...

At the risk of sounding my view [my] TGS [The Great Stagnation] thesis is looking stronger than it did even two months ago when [my] book was published.
The Zero Marginal Product hypothesis is that technological change has rendered many workers completely useless, while the Great Stagnation theory holds that technological change has slowed down dramatically. 

As I have noted before, these two theses are mutually exclusive. Technology can only make humans obsolete if it produces a lot of stuff very cheaply. But if this is happening, then productive technology has not stagnated. You can believe in one of these, but you can't believe in both; Cowen admits as much in his post.

Now, Tyler Cowen is a "big-think" guy, and it's perfectly legitimate for big-think guys to make contradictory hypotheses. Science fiction authors do it all the time. What's not quite so legit is to do this and then pat yourself on the back for being so perspicacious when events confirm one or the other of your contradictory predictions. Doing so is basically the definition of confirmation bias.

What Cowen has basically done is to buy straddles on humanity's future. If productivity goes up, he gets kudos for the ZMP; if it stagnates, it's TGS. Basically, the only way Cowen goes 0-for-2 is if the economy rapidly returns to full employment and historically average productivity growth. At that point, Cowen's career as a futurist would be in extreme danger...he might even have to make a third hypothesis...

Update: Tyler Cowen responds in the comments:
Note that this post doesn't describe the ZMP hypothesis correctly. The workers are ZMP for demand-side reasons, for "recalculation" reasons, or because they're simply not productive or cooperative, not because some new technology replaced them. It could be that new technologies help us better measure worker productivity, however. So the postulated contradiction doesn't hold.
But most of these alternative reasons just don't seem to make sense.

Demand changes don't affect a worker's marginal product. When there is insufficient demand, workers still have the ability to produce stuff, it's just that people won't buy the stuff they produce, so they become unemployed. "Recalculation" should operate through a similar channel - firms can use workers to produce stuff, they just don't know which type of stuff they'll be able to sell. In both of these cases, marginal product is not zero. The firm's production function is not changed.

As for workers being "simply not productive or cooperative", it's theoretically possible (although one has to wonder what the workers are expecting to be hired to do!). But it's a bit unclear why a bunch of workers who had been productive/cooperative before would suddenly become totally unproductive and/or uncooperative. And it's even less clear why these suddenly-unproductive or suddenly-uncooperative workers would still be actively seeking jobs and expecting to find jobs.

Finally, there's the idea that new technology has allowed firms to see that some of their workers are utterly useless. But this means that these workers must have been replaced by technology in the past, and we are only now realizing it due to a second, different kind of technological improvement.

Basically, if all humans have some productive power- and I think they probably do - the only way you get ZMP where you had no ZMP before is if we invent something (or some combination of things) that can do anything a ZMP human can do, but at lower cost, and which is in elastic supply (i.e. we can make more of it). Suppose we invent a humanoid android which can be produced and maintained more cheaply than a human, and which can and will do anything a human can do. Then any time we consider hiring a human, we just build an android and hire it instead. Humans go the way of the horse(-and-buggy). Even in this setup, it's still hard to get ZMP for humans, because the humans can just go off and form their own economy if there's enough space for them to do so (kind of like wild horses).

Basically, for even some humans to be truly ZMP, they must either refuse to work or be totally disabled or be squeezed off the planet by competitors.

Friday, March 25, 2011

Japan - truth, fiction, and stereotypes

Peter Tasker is regarded as "the foremost authority on the Japanese equity market". He lives in Japan, and makes a living studying and explaining the country. This is probably why his column on Japan's recovery from its unprecedented natural disaster gets a few important things right. But, frustratingly, these points of light are embedded in an article that is otherwise the usual mix of outdated stereotypes and bad economics that so often characterizes Western analysis of Japan.

I'm going to go out of order here. First, here are some things Tasker laudably gets right.
When the Japanese economy stagnated, the praise and warnings turned to lectures and self-congratulation, as the West patted itself on the back for having bested the Japanese threat. But...[i]n my three decades of residence here, Japan's underlying reality has changed a lot less than volatile foreign perceptions. 
This is basically true. Japan is often held to have suffered two "lost decades," but in fact suffered only one; per capita GDP growth rebounded strongly in the 2000s, catching back up with a stagnating America.
[Western "experts"] misread the causes of Japan's postwar success. The supposedly farsighted technocrats praised by Johnson in his 1982 book, MITI and the Japanese Miracle, were the same people who tried to stop Honda from getting into the auto market, poured public money into sunset industries, and built nuclear power plants on a tsunami-prone coast at sea level. 
Also true. Japan's massive bureaucracy has frequently been more of a hindrance than a help to the company's businesses.
After the collapse of the bubble economy in 1990, Japan did indeed descend into stagnation and banking crisis. At the time it seemed as if Japan's policymakers and bankers were uniquely incompetent in their fumbling attempts to tackle the problems. With the hindsight offered by the global financial crisis, it is clear that there are no easy fixes to the damage caused by the implosion of a large-scale bubble. And the United States is not one to judge: Washington has refused to make Wall Street take the harsh medicine it urged on Japan a decade earlier. 
All very true, and worth repeating.

But although he does a good job of skewering certain Western misconceptions about Japan, Tasker seems more than willing to indulge in some of his own...
The Japanese economic miracle had nothing to do with had everything to do with the resilience of ordinary Japanese people and the country's deep reservoir of social capital.
This is, to put it plainly, rubbish. Do all "resilient" cultures do well economically? Was China's poor economic performance until 1979 due to its lack of "resilience", and has its miracle growth since 1979 been due to a massive sudden increase in "social capital"? Did Korea and Taiwan only become "resilient" in the 70s? 

Hardly. Japan's economic miracle was due to A) rapid post-WW2 catch-up growth, B) a high savings rate, C) stable macroeconomic management, D) openness to foreign technology, E) stable export markets, and especially F) a well-educated hard-working labor force. 
Japan was urged to make radical economic reforms by many foreign observers, who were then disappointed by Tokyo's glacial progress in making them. But economic efficiency was never the end goal, whether Japan's economy was rising or falling. It was social stability...don't expect to see a plethora of Japanese billionaires emerging...or the adoption of hostile takeovers, Reagan-Thatcher-style supply-side reforms, and the rest of the neoliberal agenda.
Oh yeah? Where were you in the early 2000s, when Junichiro Koizumi and Heizo Takenaka instituted a whole bunch of Reagan-style neoliberal reforms (and the economy, possibly coincidentally, boomed)? Did you miss the fact that Koizumi won a landslide victory in 2005 purely on the strength of a promise to privatize the postal system? And as for "billionaires" do know about all those super-rich Japanese industrialists who emerged from the post-WW2 high-growth period, right? 

Sure, many Japanese people prize stability and security over growth and efficiency. More than Americans, on average...but much less than, say, the French! Growth is a big deal for Japan, and the country has shown a distinct willingness to do what it needs to do to keep the wheels of commerce turning.
The generation of Japanese brought up amid the postwar devastation was driven by a hunger to reconstruct everything -- their lives, their society, their country's standing in the world. Once Japan was strong enough to be left alone, the target had been achieved.  
If this is true, why do so many Japanese people I meet - from businesspeople to scientists to government officials to housewives - bemoan Japan's growth slowdown? Why the unprecedented ouster of the long-ruling Liberal Democratic Party? Why the outcry over growing poverty? And besides being patently false, the claim that "Japanese people only care about social stability and national prestige, not economic well-being" strikes me as pretty insulting as well.

But here's the part that really got me steamed:
[E]conomists usually discuss GDP without reference to currency markets, but this can obscure what's really going on. Japan's tight monetary policy has caused the yen to strengthen significantly against the dollar and dollar-linked currencies -- which raises the global purchasing power of Japanese households and corporations. In comparison, U.S. growth looks impressive when denominated in dollars, but not so much when taking into account the weak dollar policy followed by Messrs. Greenspan and Bernanke.
OK, this is just bad, bad economics. And false. First of all, "tight monetary policy"? This is a country where interest rates have been zero for a decade, and "quantitative easing" first became a buzzword. And a strong yen policy?! Japan actively intervened in currency markets to weaken the yen for decades, including a massive dollar-buying spree in early 2004. The yen was, until very recently, undervalued against the dollar by most measures. And to say Alan Greenspan pursued a "weak dollar policy" is, quite frankly, the kind of thing I expect to read on a second-rate investment advice website or on a conservative blog rant. 

A little more attention to facts, please.

Anyway, after getting a few things right and a few face-palmingly wrong, Tasker fills the remainder of his column with cliched stereotypes of Japanese culture. Here's a condensed version:
[T]he country's extraordinary social cohesion and stoicism haven't budged an inch...

If modern Japan has a common ethic, it's based on mutual respect, not victory in competition...

Japan does gaman (endurance) superbly. It copes with the challenges of success less well...

In the Japanese hierarchy of needs, social cohesion ranks higher than top-line growth...

The Japanese also fear a dilution of shimaguni konjo, the "island nation spirit" [that would result from increased immigration]...

The quiet strength of today's Japan is that the janitor in your apartment building is not a representative of "the other." He is you.

As they set about the task of recovery, [the Japanese] will become more like themselves. 
Heh. Been rereading The Chrysanthemum and the Sword on those long plane flights, have we? I swear, I never cease to be amazed by the willingness of Western analysts to ascribe everything about Japan to the mysteries of the country's ancient, immutable, alien culture (and/or race).

And in conclusion...I guess being "the foremost authority on the Japanese equity market" entitles one to say pretty much anything about Japan and get away with it. But it does not entitle you to say something like this:
In a sense, Japan has been waiting for a crisis just such as this to show its inherent strengths. 
No, Mr. Tasker. Japan was not waiting for a crisis just such as this, in any sense of the word.

P.S. - I realize, as I write this, that I never posted anything on this blog about giving to Japanese tsunami/earthquake relief. That was a huge mistake on my part. It was, I suppose, a holdover from back when I assumed that so few people read my blog that I couldn't make a difference. But in any case, if you haven't already given money to disaster relief, you should. Sorry I failed to say that when it would have counted a bit more.

Wednesday, March 23, 2011

Predicting China's inevitable slowdown

"If something cannot go on forever, it will stop."
- Herbert Stein

China's rapid growth has been going on for 30 years, but it can't go on forever. This is one of the most basic facts in macroeconomics, confirmed by every single observation and supported by every single theory at our command. When a developing country's per-capita income approaches that of the richest countries, it cannot grow much faster than the richest countries.

The question is, when? How rich will China get before it stops posting blistering growth rates? Answers vary widely.

Economist Barry Eichengreen, for example, claims that the big slowdown is coming in just three years:
In recent work, Kwanho Shin of Korea University and I studied 39 episodes in which fast-growing economies with per capita incomes of at least $10,000 experienced sharp and persistent economic slowdowns. We found that fast-growing economies slow when their per capita incomes reach $16,500, measured in 2005 US prices. Were China to continue growing by 10% per year, it would breach this threshold just three years from now, in 2014.

There is no iron law of slowdowns, of course...slowdowns come sooner in countries with a high ratio of elderly people to active labor-force participants, which is increasingly the case in China, owing to increased life expectancy and the one-child policy implemented in the 1970’s.

Slowdowns are also more likely in countries where the manufacturing sector’s share of employment exceeds 20%, since it then becomes necessary to shift workers into services, where productivity growth is slower. This, too, is now China’s situation, reflecting past success in expanding its manufacturing base.

Most strikingly, slowdowns come earlier in economies with undervalued currencies. One reason is that countries relying on undervalued exchange rates are more vulnerable to external shocks. Moreover, while currency undervaluation may work well as a mechanism for boosting growth in the early stages of development, when a country relies on shifting its labor force from agriculture to assembly-based manufacturing, it may work less well later, when growth becomes more innovation-intensive.

Finally, maintenance of an undervalued currency may cause imbalances and excesses in export-oriented manufacturing to build up, as happened in Korea in the 1990’s, and through that channel make a growth deceleration more likely.

For all these reasons, a significant slowdown in Chinese growth is imminent. 
I have a slight problem with Eichengreen's math here. He claims that slowdowns happen around a per capita GDP of $16,500 in 2005 dollars. China's per capita GDP in 2010 was about $7,500 in purchasing power parity terms (at market exchange rates, it was less than 2/3 of that). In 2005 dollars, it was about $6,500 (subtract 14% for inflation). Assuming a continued 10% rate of growth, as Eichengreen does, I calculate that China will pass the $16,500 mark in 2020, not 2014 as Eichengreen states.

So it seems that Eichengreen was either using some very different numbers, or made a math error. Assuming that the other factors he cites knock another 3 years off the time until the slowdown, we get 2017.

Of course, that probably should be regarded as a lower bound. The World Bank predicts that China will not slow substantially until 2030:
The World Bank today said China has the potential to achieve an annual growth rate of 8 per cent for another 20 years, with latent possibility to outgrow the US economy.

Estimates showed that China's current relative status to the US was similar to Japan's in 1951, and South Korea's in 1977, who were in their high-speed development period, World Bank Chief Economist and Senior VP Justin Yifu Lin said. 
Wow, that's 16 years later than Eichengreen's prediction - which, at high growth rates, is a factor of 4 difference in total GDP! Even if Eichengreen got his numbers wrong and his model predicts a slowdown in 2020, that's a 10-year difference, or a factor of 2.5. Why the big discrepancy?

Well, three reasons: 

1. The World Bank assumes that some of China's slowdown will come immediately. They predict 8% growth rather than 10% (and keep in mind, Eichengreen's predicted slowdown was a drop to 7%). Mildly slower growth will delay the day of much slower growth.

2. The World Bank assumes that China's growth will fall when it reaches half of U.S. GDP; for Eichengreen's model, the cutoff is 41%.

3. The World Bank assumes that the trigger for slower growth happens at a percent of future U.S. GDP, while Eichengreen assumes that it happens at a percent of 2005 U.S. GDP. Since the U.S. is still growing, this subtracts from the effective growth rate that the World Bank is using for its calculations. With a very optimistic assumption of 3% U.S. growth, that means that a China growing at the World Bank's 8% figure would only be approaching the World Bank's slowdown cutoff at a rate of about 5%.

So who's right? It depends on two things: A) China's growth rate in the near future, and B) the location of the mysterious "slowdown cutoff" at which rapid catch-up growth hits diminishing returns. 

On the former issue, I tend to agree with Eichengreen. China's leaders regularly declare their intent to lower growth to a slower, more sustainable level, but so far it has never happened. Not only do local leaders often ignore the central government's diktats, but the central government itself seems unwilling to appreciate its exchange rate or to halt the flood of cheap capital that it forces its banks to provide. The government is too afraid of a sharp slowdown to allow a gentle one. So I also predict that 10% growth will continue to be the norm for this decade. 

Making that modification, the World Bank's prediction drops to 2023. The remaining 3-year discrepancy between the (math-corrected) Eichengreen base case and the World Bank's prediction is down to the different assumption about the cutoff. If you believe Eichengreen's list of exacerbating factors, the discrepancy is 6 years (2017 vs. 2023), or still nearly a factor of 2 for the Chinese economy.

So, basically, expect China to slow sometime late this decade or early the next decade. For my own on-the-record prediction, I'll go with an average of the professionals, and make it 2020. At that point, China's total GDP (PPP) will be about $25 trillion, or about 1.5 times the size of the U.S. economy.

Friday, March 18, 2011

Economists' mistakes on climate change

Some climate scientists have been getting mad at economists lately (see also here). As Karl Smith notes, most economic analyses of climate change tend to downplay the threat caused by climate change:
I wrote before that part of the problem for climate hawks is that even expected damages from climate change are not that large. Not zero by any means but not as high as they would like...Getting the result that the US would fall off a cliff due to climate change as projected over the next century is hard to produce.
Climate scientists, in contrast, are increasingly alarmed by the data. Who is right?

Well, I would not call myself an expert on the subject by any stretch, but I did take a class that introduced me to some well-known economic analyses of climate change. I have to say, I was not hugely impressed. The papers I read seemed to have two big limitations:

1. Backward-looking, local analysis

2. Limited models with few or no interaction terms

The first of these is basically what the climate scientists are complaining about. Empirical economists use the past to predict the future, but climate change of the type we are facing is unprecedented. All economic data comes from a time when climate change wasn't very severe. 

The problem with this is that very extreme bad events are not always simple extrapolations of mild bad events. Suppose we want to make a model of the cost of five-kilometer-long asteroid impacting the Earth. So we look in the past, and we find the average costs caused by small meteorites, and we multiply these by the ratio of the volume of these small meteorites to the really big one. If a 0.5-meter-long meteorite causes an average of $100,000 in damages, then a 5-kilometer-long meteorite will cause $12.5 trillion in damages. Not fun, but not too bad.

But this would obviously be silly talk! A small meteorite doesn't cause a very small amount of things like global crop failures, tsunamis, mega-earthquakes, supervolcano eruptions, and ice ages. A large meteorite causes large amounts of these things. In other words, there is a high degree of nonlinearity at work that just cannot be estimated from past data. Economists don't look at the tail risks of climate change - the low-probability doomsday scenarios - because we just don't have the tools to do so.

But we rarely admit this.

The second big problem with economic analyses of climate change is that the models used are very limited in scope. For example, consider this NBER working paper by respected climate-change economists Deschenes and Greenstone. It analyzes the cost of increased heatstroke deaths from higher temperatures. Or consider this paper by the same authors, published in the American Economic Review (the most prestigious economics journal) in 2007. It estimates the losses in agricultural output due to climate change. 

These are careful, serious analyses, and the authors are up-front and honest about what they are and are not trying to model. But they are nevertheless limited by the fact that each model studies only one isolated effect of climate change. Interactions between effects are just ignored. For example, suppose agricultural losses make us less able to afford the air conditioning that would prevent us dying from heatstroke. Or suppose the danger of death from heatstroke makes it too dangerous to go out and farm! 

Even worse, there are a whole plethora of possible effects of climate change - shifting coastlines, species destruction, increased migration, higher rates of disease, etc. There are possible interactions between any pair of these. There are even possible interactions between any three of these!  A realistic model of climate change costs - one you could hold up in front of a Congressional committee and say "Hey, here's what we're looking at!" - would have to take most of these into effect. But economists aren't yet into making that sort of model. (Note: if you know of exceptions, please let me know!)

As another example of this, take Karl Smith's defense of climate change economics, in which he analyzes the impact of shifting coastlines on the United States. He concludes that the costs of resettling Americans will be small. But what about lost farmland? What about disrupted supply chains and networks of trade? What about the effects of the inundation of all the coastal cities worldwide that buy the U.S.'s export products? What about the uncertainty created by not knowing where to rebuild, because we don't know how much more the seas will rise? And so on.

How OK is all of this? Ryan Avent defends climate change economics, saying that bad analysis is worse than no analysis at all:
I think it's possible—indeed, likely—that current models are failing to adequately capture the impact of climate change on the global economy. But this is how science works. You approximate reality poorly, then you learn from that and do a better job, and then you do a better job still.
But what exactly are we learning from these limited, flawed analyses? Does making a backward-looking, linearized model of one "isolated" effect of climate change help us make a better model in the future? Well, yes, if by "in the future" you mean "after climate change has already run its course, and we can see how wrong those limited models were." Because that is how science works - you test your models against the data. Lowball estimates of climate change costs won't get a rigorous test until it's too late.

Ryan continues:
Environmentalists and economists are fundamentally on the same side, supporting the use of data and the scientific method to reach reasonable, peer-reviewed conclusions and appropriate policy recommendations.
That may be true (at least for those of us who are not political shills). But it seems to me that economists could improve their backward-looking models by incorporating forward-looking climate science. We could loudly reiterate how little they (or any of us) know about the doomsday "tail risks" of climate change. And we could try to make more comprehensive models.

Thursday, March 17, 2011

Is econ a science? On its good days, yes.

Karl Smith Adam Ozimek has a blog post up about whether economics is a science. This is a questions I've thought about a lot. It's a difficult one, for (at least) two big reasons:

1. There are a whole bunch of branches and sub-fields of economics.

2. There is a big difference between how economics is practiced and how it could be practiced.

For an example of the first problem, consider Prospect Theory. This Nobel Prize-winning theory is a kind of cognitive psychology - Daniel Kahneman put people in a lab and tried to figure out how they calculate probabilities and take risks. There are a lot of economists doing stuff like this - another example is Vernon Smith, who also won a Nobel.

But then again, consider macroeconomics. You can't put a national economy in a lab, so falsification has to wait for history to come prove your theories wrong. In the meantime, people make lots of assertions that just can't be backed up. This is how we got Real Business Cycle Theory (a favorite whipping boy of mine).

So some economists are doing science, and some aren't. Actually, this is generally true of most of the fields we call "sciences". Biology, for example, contains plenty of science, but it is also home to "Evolutionary Psychology," which consists mainly of non-falsifiable just-so stories. Even physics, which is usually held up as a paragon of scientific rigor, generates its share of seemingly non-falsifiable theories.

For an example of the second problem, take the recent progress that has been made in game theory. Game theory started out as just math - you assume players follow certain rules and face certain choices, and then you derive what they would (or might) do. For a long time that's all it was. You had (and still have) people trying to think up games whose results seemed to mirror real-world phenomena, but this was not science, since there were many games and many equilibrium concepts that could yield the same result (and nobody really tried to test which was really in effect). But more recently, experimental economists have started putting
people in labs and finding out how people really play games. This is science.

So a non-scientific branch of economics may become scientific in the future.

Now to me, here's the more important question:  Should economics be a science?

To answer this question, observe that there are two reasons why some economists do science and some don't. Sometimes economists don't do science because they don't have the tools to do it. Others don't do science because they don't think they should have to. In general, I think the first is a good reason not to do science, and the second is a bad reason. Let me explain why.

Many times, we find ourselves able to observe the world, but not in a controlled way. For example, take history; we can see what happened, but we don't know why, since we can't put history in a lab. In the "social sciences," you have lots of people doing empirical work by observing statistical correlations - but, unless they can find natural experiments, they can't infer causation. Are all of these useless endeavors? NO! Observation without falsification isn't science, but it is still useful. Naturalistic observation narrows down the set of possible explanations for phenomena that are too big and complex to test - and there will always be some phenomena that are too big and complex to test.

So I think that much non-scientific empirical work has value.

However, there are times when economists have the tools to do science, but simply fail to make use of those tools. This is generally due to the academic culture of the field. Behavioral economics experiments, for example, encountered (and still sometimes encounter) huge resistance from theorists who thought that their speculations about human behavior were more powerful than actual empirical tests. As another example, take Ed Prescott's assertion of "theory ahead of measurement" - the idea that observations should be ignored when they contradict intuition.

In cases like these, it seems to me that economics suffers from cultural holdover from the time when economics was basically philosophy. For centuries, even as chemists and physicists and biologists were constructing labs, "economists" were still basically writing literary tomes full of speculation and hand-waving. In the mid-20th century, this hand-waving approach was (mostly) replaced with a combination of rigorously defined mathematical explication and empirical statistical analysis. But neither math nor statistics (nor both together) is sufficient to make a discipline scientific; for that, you need experiments.

Although, as I said, I believe that science is not the only valuable way to explain the world, it seems to me that, when it can be done, it is the best way. Falsifiable theories, when they've withstood our best attempts at falsification, are just better at predicting the future than any other kinds of theories*. Thus, when economists can do science, we should do science.

And when we can't do science, I think we could stand to be a bit more humble and circumspect when advancing our theories and making our policy prescriptions. Take macro, for example. The debate between Keynesians, monetarists, and neoclassicals won't be resolved any time soon, because we can't do experiments to find a falsifiable theory of business cycles (and because history has weighed in against every theory we've ever constructed). The general public has caught on to this fact, and has come to expect less of its macroeconomists. But macroeconomists seem not to expect less of themselves, and still tend to radiate an aura of intellectual confidence that seems a bit unjustified.

And finally,  to critics of the economics field in general, just remember: Economics is a very young discipline. As a formal field of study, it's really less than a century old; as a science, less than half of that. Remember that chemistry, which is now probably the hardest of the hard sciences, was for long centuries a mix of alchemy, superstition, and real science. Ptolemaic astronomy persisted for centuries before we had the telescopes to prove that Copernicus had a better theory. As humans develop the tools to more accurately observe big, complex systems and groups, economics will become progressively more scientific. Maybe with a lag; after all, it takes time to bury accumulated centuries of philosophy and religious canon under a mountain of experimental data. But it will happen.


1. Mike the Mad Biologist has a rant about econ vs. biology. It is a pretty good rant, though a bit tangential to what I've been discussing here. The key takeaway line is this: 
[T]he key difference [between biology and economics] is that biology has accepted modes of confronting theories and, importantly, discarding them. 
Exactly. Econ doesn't always have the means to discard bullshit theories. But sometimes we do have the means and we choose not to use them. That is bad. 

Mike also links to Barry Eichengreen saying:
The late twentieth century was the heyday of deductive economics. Talented and facile theorists [could] build models with virtually any implication, which meant that policy makers could pick and choose at their convenience. Theory turned out to be too malleable, in other words, to provide reliable guidance for policy. 
In contrast, the twenty-first century will be the age of inductive economics, when empiricists hold sway and advice is grounded in concrete observation of markets and their inhabitants. Work in economics, including the abstract model building in which theorists engage, will be guided more powerfully by this real-world observation. It is about time.
What Barry said.

2. Peter Dorman has some musings on the meaning and value of science that echo my own, and he concludes with an excellent prescription for the field of econ:
So what would a scientific economics look like? I have mostly answered this already: it would look like other sciences whose objects of study are complex, heterogeneous and context-dependent [like geology and hydrology]. It would study mechanisms primarily and end states only for heuristic purposes. It would be predominantly empirical, where this encompasses both statistical work and direct observations on economic behavior (which may also entail statistical analysis)...Experimentation, in the lab and in the field, would become more common, but even more important, primary data collection of all sorts would be accorded a very high value, as is the case in all true sciences. Its macro models would come to look like macro models in hydrology or biogeochemistry: simultaneous differential equations representing mechanisms rather than static end states embodying (a single) equilibrium. Economists would increasingly find it useful to collaborate with researchers from other fields, as their methodological eccentricities are abandoned. Finally, there would be a much clearer distinction between the criteria governing scientific and policy work, insulating the former from some of the influence exerted by powerful economic interests and freeing the latter to adopt an ecumenical and risk-taking approach to tackling the world’s problems.
Sounds about right to me!

Wednesday, March 16, 2011

Libertarianism is a "low-end" strategy of state formation

Reading Ian Morris' Why the West Rules -- for Now, I was struck by the following passage:
In a nutshell, the Eastern and Western cores avoided collapse in the first millennium BCE by restructuring themselves, inventing new institutions that kept them one step ahead of the disruptions that their continuing expansion itself generated.

There are basically two ways to run a state, what we might call high-end and low-end strategies. The high end, as its name suggests, is expensive. It involves leaders who centralize power, hiring and firing underlings who serve them in return for salaries in a bureaucracy or army. Paying salaries requires a big income, but the bureaucrats’ main job is to generate that income through taxes, and the army’s job is to enforce its collection. The goal is a balance: a lot of revenue goes out but even more comes in, and the rulers and their employees live off the difference.

The low-end model is cheap. Leaders do not need huge tax revenues because they do not spend much. They get other people to do the work. Instead of paying an army, rulers rely on local elites—who may well be their kinsmen—to raise troops from their own estates. The rulers reward these lords by sharing plunder with them. Rulers who keep winning wars establish a low-end balance: not much revenue comes in but even less goes out, and the leaders and their kin live off the difference.
(emphasis mine)
There is not much economic analysis in this discussion of "high-end" and "low-end" strategies. Morris is a historian. He doesn't really try to answer the question of why the taxes and bureaucracy of "high-end" strategies might enable a state to grow and prosper, instead of dragging it down with inefficiency and deadweight loss. Thus, it might seem strange to economists that Morris would so naturally assume that "high-end" strategies can be a very good thing.

But we should not dismiss Morris' observations simply because they don't square with our conventional wisdom. After all, Econ 101 provides a perfectly plausible theoretical reason why "high-end" strategies might succeed better than "low-end" ones. That reason is public goods. The tax money that governments take in doesn't all get wasted or mailed off as Medicare or welfare checks. Some of it goes to pay for roads, courts, police, schools, research, and a bunch of other very important things that private companies cannot or will not provide in sufficient quantity. If high-end states outperform low-end ones, it is because they provide the public goods that low-tax, low-spending governments can't create.

Now let's think about libertarianism. The modern American libertarian movement seeks to minimize taxes, government spending, and the size and scope of the bureaucracy. In the view of this movement, public goods either don't really exist, or aren't worth the price, or can be efficiently provided by private actors (charities, etc.). The alternative libertarians favor is exactly what Ian Morris calls a "low-end" strategy. Low taxes will be matched by low spending; private security contractors will replace many of the functions of the army, police, and prison system; schools will be privatized; basic research will simply be neglected; tort reform will limit the role of courts; and as for roads, who knows. To the extent that they voluntarily give away their wealth, rich individuals will assume the "noblesse oblige" role of the feudal lords of old. The Bush Administration even discussed returning to that lowest of low-end techniques: tax farming.

Why are we seeing so many smart people calling for a return to statehood-on-the-cheap? Most observers blame either the vanishing nationalism of a globalized elite, or the ethnic fragmentation of the American voting public. I give credence to both of these. But it may also be the case that libertarianism is simply a response to what Ian Morris calls the "paradox of development":
[P]eople’s success in reproducing themselves and capturing energy inevitably puts pressure on the resources (intellectual and social as well as material) available to them. Rising social development generates the very forces that undermine further social development. I call this the paradox of development.
High-end states that succeed in times of economic expansion can fail dramatically in times of stagnation or contraction. As growth hits its limits, the big bureaucracies that helped generate and stabilize the growth are no longer needed...but institutional momentum keeps them around past their sell-by date, dragging the economy down with inefficiency until a breaking point is reached, and a high-end state collapses into a new low-end equilibrium. 

That is Morris' view, and I agree with it.

Are the U.S., Europe, and Japan running up against this paradox? Many argue that our best days are behind us - that resource limits, or the natural slowing of scientific discovery, are putting an end to the dizzying growth of the Industrial Revolution. Even if you don't believe those theories, it's hard to ignore the stagnation of wages in rich countries over the past few decades. If stagnation is here to stay, maybe a high-end state just doesn't make sense anymore.

This is, basically, the libertarian thesis. I'd like to point out that it is a profoundly pessimistic one - instead of "morning in America," it's twilight as far as the eye can see. The best we can hope for is the little one-time bump we get from eating the institutional capital we built up during the Industrial Revolution - and even that is pretty much used up at this point. The Reagan/Thatcher/Koizumi revolutions cannot be repeated.

What's more, I can't easily dismiss this thesis. Is there definitive proof that a big, active government is still the best way to go? Not that I know of! Looking at history gives us only anecdotes, not data. Sure, it's true that the richest, fastest-growing states in history were all high-end (ever see a country get rich without government-build roads?), but the causality could run the other way; high-end states might simply be the best response to an era of plenty.

Libertarianism's detractors, of course, can cite recent empirical studies showing the net economic benefits of government investment in highways, rail, digital infrastructure, and R&D. We can point to the negative results of privatization of the prison system, and the high costs of replacing the military with mercenary contractors. And we can point out that our productive power is not nearly as stagnant as our earnings (which hints that our "stagnation" does not mean we're hitting a technological ceiling).

We also have some anecdotal evidence that a strong state still beats a weak one. Of all rich economies, Germany has done the best lately, boosting employment even in the middle of a global recession. Many credit Germany's strategy of export promotion and vocational education. This seems like activist government at its best. (Update: I may have spoken too soon with this example!)

But in the end, this evidence cannot crush the libertarian case. The specter of decline looms a bit too frighteningly. The neoliberal reforms of the 80s look a bit too much like a worldwide phenomenon. It is hard to forget Bill Clinton saying that "the era of big government is over."

Anyway, the debate over what kind of state we should have will continue to rage. But at least we can keep it honest. Libertarians like to paint liberals as a bunch of socialists obsessed with redistribution at the cost of efficiency. But many (most?) of us are just people who think that the high-end model of state formation - an active, growth-oriented government providing public goods - is still better than the low-end alternative. Maybe not as clearly better as it was in the 50s and 60s. But still better. Maybe.

Tuesday, March 08, 2011

In a pinch, upgrade the humans or redistribute the robots

In my last post, I took a somewhat dim view of economic theories that rely on extremely specific but unobservable changes in technology to make them tick. Let's not be futurists until we absolutely have to.

BUT, that said, it's really fun to be a futurist! I read a lot of science fiction, and as an economist I love speculating about how our economy and society would respond to various technological advancements. So I'm going to take off my scientist's hat for a minute, and put on my fun-but-ultimately-useless futurist's hat, and with that caveat I'll try to answer the question: what do we do when robots replace workers?

Will Wilkinson explains how automation can increase inequality:
[S]uppose [as a factory owner] I replace all my workers with machines. Questions of distributive fairness disappear! I own the machines; I don't owe the steely suckers anything!...Now, I don't know what all those workers I laid off are doing...Certainly my demand for robots...has created vigorous growth in the robot manufacturing sector. But...job growth there can last only as long as it takes to come up with robots that build robots that build robots, and so on. So workers are slowly squeezed out of manufacturing by automation...This squeeze has many implications, one of them being that here is an important sector of the economy in which more or less all the gains accrue to the owners of capital and more or less none to the working class, simply because the working class doesn't work here anymore...the robot-owning class moves up a bit relative to the no-longer-manufacturing working class...
In the past, technological improvements have never made human labor obsolete; no matter how big the change, there's always been something new for us to do. But that may not always be the case, because humans are fundamentally limited beings. Yes, some human or another will always be needed as labor, but that might end up just being a few really brilliant people working on robot improvement.

So if that day comes, what do we do to prevent the massive inequality that will result? As I see it, there are two basic approaches:

1. Improve the human being, or

2. Redistribute capital ownership rights among human beings.

Following Bruce Sterling, I will call these the "Shaper" and "Mechanist" approaches* (which just goes to show that anything economists think of in 2011 was thought of by sci-fi authors three decades earlier).

The Shaper approach is simple: use technology to make humans smarter, or otherwise more capable, so that we maintain our usefulness even in an era of better and better robots. Alex Tabarrok makes this case:
Will robots and artificial intelligences take human jobs? Perhaps but the nature of humanity is not carved in stone. Genetic enhancement (GE) is within a hairsbreadth of reality...In the not so long run it's not about computers substituting for labor or even complementing labor, it's about designing labor to complement computers (and vice-versa). 
This option is attractive because it allows us to preserve the same sort of labor market we have now. Workers are valuable, so capitalists hire them; workers get both money and a sense of personal worth and dignity

One problem with the idea, of course, is that humanity has all kinds of moral hangups about this kind of human enhancement. Though Tabarrok believes that the market, and competition among nations, will eventually overcome our qualms, I'm not so sure. There is a vast web of institutions that we have created to slow the progress in human sciences - human subjects review boards, the FDA, the entire medical system for that matter. In any rich country, research into human improvement is severely curtailed, and technologies are strictly regulated. It may be that markets will eventually win out and give us superbabies who will be able to hold their own against robots, but we might have to wait for quite a long time. 

A second problem, of course, is that human improvement might create its own form of inequality.  If left to the market, superbaby technology would be available only to the rich, locking in inequality across generations. That is something we will want to avoid. But if governments were to appropriate and distribute the best baby-enhancing technology, it would go a long way toward creating equality of opportunity.

The Mechanist response to an automated society, on the other hand, is to redistribute ownership of the robots (Tyler Cowen favors this approach). As Will Wilkinson points out, robots are slaves (even if self-aware, self-motivated artificial intelligence is possible, there's not necessarily any need for us to use it). Given a world where every product and service is produced by robot slaves, we'll have massive inequality between those who own the robots and those who don't. But if we simply redistribute ownership of the robots amongst society, equality will be back. This would be a real, working version of George Bush's failed "ownership society" concept.

There are, of course, problems with this approach as well. How do we accomplish this redistribution fairly? Won't people who get fewer robot-shares try to seize others by force? Will this redistribution severely reduce the efficiency of the robots' use? What happens if some people sell all their robot-shares, only to spend all the proceeds and see their children become poor again? It's clear that widely-distributed capital ownership isn't stable over the long run. Labor tends to create an egalitarian society because everyone gets a brand-new endowment of labor when they're born. It would be difficult to replicate this happy coincidence with widely distributed capital ownership.

And finally, both of these remedies assume that humanity will still be in control of policy and will still own the means of production. True replacement of humanity would mean artificial intelligence - machines that can not only do everything we do, but can make decisions about what to do better than we ourselves can. If we manage to replace ourselves in every last respect, then we're out of the Bruce Sterling universe and into the Isaac Asimov universe. Or maybe Terminator or The Matrix. When two intelligent species share a planet, the result seems unlikely to be pretty...seen any Neanderthals lately?

In any case, if it ever does look like our species is becoming obsolete, I suggest we employ a hybrid of the "Shaper" and "Mechanist" approaches. We should get used to the idea that upgraded humans will command much higher wages, which they will use to support their unmodified parents in our old age. Government redistribution of human-upgrading technology could make equality of opportunity even more of a reality than it is now. And in the meantime, we can try more sensible approaches to creating a true "ownership society," one that is more than just a housing bubble...maybe by turning Social Security into a forced-contribution pension scheme that only buys shares in an index-fund-of-everything (and forcing companies of a certain size to go public)? And maybe by giving out welfare checks not as cash, but as shares in this omni-fund?

Anyway, fun goofy futurism aside, I think that the problem of robots replacing us is still MUCH MUCH farther away than many economists seem to think. The current global economic turmoil might be a tiny bit about the end of labor, but my hunch is that the vast majority is about rapid globalization and a good ol' once-in-four-generations financial crisis, with a few resource constraints and government policy mistakes tossed in. But if and when the robots do really replace us, rest assured that we have options. Options that are fair and equitable and safe, if more than a bit weird to our modern sensibilities.

(Unless Skynet takes over. In that case, all we can do is reach for the plug.)

* Actually, the Mechanists also boosted themselves with cyborg enhancement. My concept of human enhancement is that we'd use both genetic manipulation and cyborg technology. So I'm not exactly copying Sterling. Widespread distribution of the ownership of robotic capital is really more like a Vernor Vinge setting.

Monday, March 07, 2011

Technology in the gaps

It must have been strange to be an ancient Greek. Storms happened because Zeus woke up in a bad mood. Seas were calm because Poseidon was off playing XBox...or something. Basically, everything happening around you was due to the whims of mercurial, ineffable superbeings.

I feel like economists often treat technology the same way - as a capricious god who lives in all the gaps of our theories, pulling the levers and making the clockwork run.

The most famous example is the "Real Business Cycle" (RBC) model, for which Ed Prescott won the Nobel 2004. Briefly...In this model (as in many others), you split economic production into two "factors": capital and labor. Whatever's left over - that is, the fraction you can't measure as capital or labor - is called "total factor productivity," or TFP. The RBC model says that TFP is, basically, technology. When technology gets better at a rapid rate, the theory says, we have an economic boom, and when it gets better only slowly (or gets worse), we have a recession. Ta-da! Business cycle explained!

Of course, there are many, many problems with this theory, and that's a blog post for another day. I want to focus here on one point: When you look at the wider world, you can't actually see the changes in technology that would be needed to cause the economic fluctuations we observe. RBC theory basically says that technology causes the movements of the economy, but that the only way you can see changes in aggregate technology is by...watching the movements of the economy. The storm is the proof that Zeus is angry.

Recently, many economists are using technology to explain a different set of phenomena - income stagnation, unemployment, and inequality. One prominent example of this is Tyler Cowen's The Great Stagnation, which claims that a slowdown in the rate of innovation and scientific discovery is causing the long-term flatlining of U.S. median income. This idea has been somewhat endorsed by Paul Krugman.

But there's a second strain of technologist theories bouncing around out there in the econosphere. These claim that technology is finally replacing humanity, making many of us irrelevant.

One of these is the theory of skill-biased technological change - basically, the idea that information technology is so hard to use that it's creating inequality between those who are smart/educated enough to use it and those who aren't. In the 90s, this theory was put forth as a reason why education could help fight rising inequality. But these days, some economists are saying that it's too late for that - that technology is replacing educated people too. Here's Paul Krugman:
[My magazine piece postulated] that information technology would end up reducing, not increasing, the demand for highly educated workers, because a lot of what highly educated workers do could actually be replaced by sophisticated information processing — indeed, replaced more easily than a lot of manual labor. Here’s the piece: I still think it’s a fun read. 

So here’s the question: is it starting to happen?...
Computers, it turns out, can quickly analyze millions of documents, cheaply performing a task that used to require armies of lawyers and paralegals. In this case, then, technological progress is actually reducing the demand for highly educated workers..

[S]oftware has also been replacing engineers in such tasks as chip design...

The fact is that since 1990 or so the U.S. job market has been characterized not by a general rise in the demand for skill, but by “hollowing out”: both high-wage and low-wage employment have grown rapidly, but medium-wage jobs — the kinds of jobs we count on to support a strong middle class — have lagged behind...

Why is this happening?...Some years ago, however, the economists David Autor, Frank Levy and Richard Murnane argued that...[c]omputers...excel at routine tasks, “cognitive and manual tasks that can be accomplished by following explicit rules.” Therefore, any routine task — a category that includes many white-collar, nonmanual jobs — is in the firing line.
(For a skeptical response, see Brad DeLong, and Brad DeLong again. Also Ryan Avent.)

And the most extreme example of this class of theory has got to be Tyler Cowen's "Zero Marginal Product Workers" hypothesis, which holds that many unemployed workers have been completely, utterly replaced - that they have no skills that are worth even minimum-wage compensation.

Now, I can't say with any conviction that any of these "technologist" theories are wrong (except for RBC, but that's for other reasons), so I'm not going to violate DeLong's Rules of Krugman. Maybe these things really are happening! And the bald fact is, technology is hugely important to long-term economic growth, to the composition of the labor market, and to our modern wealthy existence. This is not in dispute. So for me to say "Bah, technology is always just Zeus by another name" would be very silly.

But I will say that technologist theories deeply trouble me. They imply that economics, as a science (yes, I used the S-word!), is ultimately of limited use in explaining the economy. If all of the interesting stuff is happening in research labs and tinkerers' garages, then economists are basically reduced to being futurists, speculating on the rate and type and social impact of future technological advances. Our patron saint would be not Paul Samuelson, but Alvin Toffler.

Furthermore, frequent use of technologist theories forces economists to do some pretty tricky mental gymnastics. Cowen and Krugman, for example, claim that overall technology is advancing too slowly to raise our median incomes, but that certain kinds of technology are advancing fast enough to replace a huge number of workers. Of course, it's possible that's true; we've seen a lot of innovation in computers in the last 20 years, and less innovation in kitchen appliances. But it illustrates the fact that, the more phenomena you attribute to "technology," the more hyper-specific claims you are forced to make about a process that you can't accurately observe. 

How far are we willing to go with this? Taken to their absurd extremity, technologist theories would have us explaining every single fluctuation in any economic variable in terms of invisible changes in technology...oh wait, never mind, that's called "RBC."

Therefore, I recommend that economists be very sparing in our use of technologist theories. When we see something we can't easily explain - stagnating incomes, rising inequality, shifts in job opportunities - we should try very hard to explain the phenomenon in terms of things we can understand, observe, and predict. For example, Krugman offers globalization, and the outsourcing of white-collar desk jobs, as an alternative explanation for stagnant employment in those job categories. That might turn out to be less important than technology, but it's something we should look at first, because we can understand it and we can observe it and we can even predict it to some degree.

So, in conclusion: Using technology to explain short-term or medium-term economic phenomena may be perfectly spot-on correct. It is also a kind of giving up.

Update: I do want to point out that I'm not trying to trash Krugman's NYT column. That column's main thrust is that education will not be a panacea for inequality, stagnating wages, or unemployment. I think that that is an excellent point, for many reasons. And Krugman's hypothesis about technology replacing skilled workers is really meant as an alternative to the standard theory of skill-biased technological change, not as an assertion that technology drives everything about the labor market.

Update 2: Via Thoma, a paper investigates "skill-biased technological change" and finds that changes in skills were a much bigger factor than changes in technology in the inequality increase of the 1990s. Just another reminder of the danger of putting "technology in the gaps"...