Tuesday, June 7, 2011

There Is No Great Stagnation, Only Great Redistribution

. Tuesday, June 7, 2011

Tyler Cowen's The Great Stagnation has gotten a lot of attention for both its form and content. (I.e., there's more than a little irony in the fact that a book alleging that technological progress has markedly slowed was the first notable electronic-only book, although it has since been released in pulp-and-glue as well.) In the video above he presents his main thesis at TEDxEast. For those unaware, the argument runs basically like this: since 1973 or thereabouts, there has been a slowdown in median American income growth, and that trend has increased in the past decade. That slowdown is mostly attributable to a decline in technological innovation. We've reaped the gains of past innovations -- cars, planes, electricity, plumbing -- but haven't made many new ones. We tweak the old innovations to our advantage -- we've made cars safer and added GPS -- but those are marginal improvements, not fundamental advances. The exception is the internet and communications more generally, but while those improve quality of life they do little to improve typical incomes.

Cowen's argument has bothered me on a number of levels. First, I think he understates the real, and monetary, value of the internet and improved communications technology for standards of living. Second, I think he makes a mistake by looking almost entirely at the U.S., and almost entirely at median income. I want to focus on the second of these, placing it in the context of the first.

I'm really late to this party... Cowen's book has been covered by everyone in the blogosphere and almost everyone in the corporate press, so I'm sure someone has written more or less exactly what I'm about to write, but I've haven't seen it in quite this form before. So to see why I think Cowen's thesis is wrong, or at least incomplete, let's start with some global data.

This graph shows global real gdp per capita from 1960-2009 (blue line). I've highlighted 1973's income level -- $1,148 -- to show what the world looked like around the time that Cowen thinks the Great Stagnation started in the US. In the following 35 years, per-person income increased by nearly 800%. If the pre-1973 trend had continued (red line), that number would be more than halved. If growth post-1973 had stagnated, we'd be below the red line. But that didn't happen, as we can see from this series. First, global growth in the 1970s was faster than in the 1960s. And while that trend wasn't consistent through the 1980s and 1990s (dark green line), global GDP growth in the 2000s was the fastest during the period. In fact, by the end of the decade we'd caught back up to where we'd be if the 1970s trend had been consistent, before the financial crisis knocked us back a bit. But the story here is of pretty rapid growth on a global scale that actually accelerated in the most recent decade. No Great Stagnation, on a global level at least.

Cowen agrees that global growth has been strong as other countries adopt the innovations the U.S. has already exploited. This "catch-up" growth may be fine for developing countries, which have a lot of low-hanging fruit, but he wants to focus on those at the edge of the technology frontier, especially the US. So let's look at what's happened to US growth over the same period.

The green line represents approximately where US incomes would be if we had stayed at the pre-1973 rate of growth. Average incomes would be less than half what they are now. If the economy had stagnated, as Cowen claims, average incomes would be below the green line. Instead, the rate of US growth actually increased over that period, at a more rapid pace even than the increase in global growth depicted in the first graph. This doesn't look like stagnation at all, much less a Great Stagnation. So what is Cowen going on about?

Ah, the picture looks a bit different if you compare mean GDP/capita to median GDP/capita. Before 1973 the two tracked each other very closely. Post-1973, mean GDP/capita (the white circles) kept growing at roughly the pre-1973 trend rate, while median GDP/capita (black diamonds) stagnated. But the economy overall did not. Just median incomes. That indicates, to me, that Cowen's preferred causal mechanism -- a stagnation due to slowdown in innovation -- is missing what's actually happened. There's been enough growth, it just hasn't gone to the median earner. The result has been higher inequality.

Why has that happened? Theories abound. Some political scientists have recently made the case that rising inequality is a result of wealthy groups hijacking politics for their own economic benefit. In other words, the distribution of growth is zero-sum, and it's been redistributed towards the wealthy in the form of tax cuts, decline in union membership, erosion of the welfare state, and deregulation. I think there's something to that, but I think it's too focused on developments specific to the US. To get the whole picture, I think we need to situate the US in a global context.

It's difficult to find reliable estimates of global median income in a time series (in fact I couldn't... pointers welcome), but indications are that inequality is increasing within many countries, and across them as well. This is also not consistent with Cowen's argument, since the movement towards the technology frontier in the US was associated with rising median income, not rising inequality. If that's the process that rapidly-growing economies like China and India are in, then we should see less inequality, not more. And if the Great Stagnation is something that afflict the US specifically, we might expect the gap between the US and the rest of the world to narrow, not widen.

So I think a more nuanced theory is needed. Specifically, we need to be able to explain two things: stagnating median, but not mean, incomes; global, not just local, trends. So what do we know about the major ways in which the global economy has changed over the past 40 years? I think three things are most relevant:

1. The global economy has become more integrated. This is partially due to politics, as more countries opened their economies to trade and investment. Average tariff rates have fallen dramatically during the GATT/WTO tenure. Capital accounts have been opened by many countries. Additionally, technological improvements have lowered transaction costs. International trade and investment have increased dramatically as a result. The consequence of this movement is a larger (global) market with more middle- and high-income consumers, and increased competition in production. This leads to point #2.

2. The US's post-WWII advantage was conducive to broad-based growth. The US share of global manufacturing was nearly 50% immediately after the war. The other industrialized economies were mostly decimated by the war, and many countries had not yet industrialized. For an American worker during this period, a high marginal product (relative to a foreign worker) did not require large amounts of human capital. Relatively low-skilled workers could mix with (non-human) capital in fairly lucrative ways. In a sense, the median American worker was able to collect rents from the rest of the world from 1945-1973, because the de-industrialization in Europe and pre-industrialization in much of the rest of the world operated as barriers to competition. By the early 1970s those advantages had waned, and trade agreements made it difficult for the US to protect domestic workers. The increased competition from workers in Europe and the Asian NICs (which shifted to export-biased development in the 1960s-70s) led to the US's share of global manufacturing output to fall to 20-25% by 1973, where it has stayed more or less ever since. This hit high-wage/less-skilled workers in tradable industries the hardest, since those were the workers that would face international competition directly. It isn't surprising that incomes would stagnate as those "rents", born of circumstance, are competed away.

3. These same processes benefit high-skilled workers with lots of human capital, as did the technological improvements, particularly in information technology and communications. The rise of the rest has increased the market into which they can sell their labor (demand curve shifts right), but the high skills required to compete with them provide a continuing barrier to entry (supply curve sticks). Compensation for those high-skill workers (and innovators) goes up, but is stuck for everyone else. We get a weak version of "superstar economics", where the highly-skilled are able exploit lower transaction costs to sell into an ever-enlarging global market, while the lower-skilled face increased competition. It's a two-track economy.

Cowen dismisses globalization-rooted theories of the Great Stagnation (around minute 12 in the video above), but (to my knowledge) he hasn't dealt with the sort of mechanisms I'm discussing in any kind of detail. Somewhat bizarrely, Cowen also claims that modern innovations (the internet, satellite-based telephones) have not contributed to GDP very much. But then how to explain how GDP growth, and total worker productivity, have increased post-1973 at the same rate as pre-1973? Median incomes have stagnated because those innovations, unlike previous innovations in manufacturing, do not require the mobilization of huge numbers of workers to increase output. A few computer programmers or financiers can create generate output on their own.

There's another aspect to this that I think Cowen has missed. Increased inequality and a move to a superstarish economy should create more of an incentive for innovation, not less. And while Cowen complains that scientists are no longer heralded by society as they once were, innovators most definitely are. We make movies about them and their social networks, and then give awards to the movies. We make them the richest people in the world. And, contra Cowen, we have seen a lot of innovation in the past 35 years. Cowen focuses on innovations in two major areas that led to the pre-Stagnation growth: transportation and energy. He may be correct that innovation in transportation has declined, although the rise in high-speed rail (outside the US) might be one counterpoint), but part of that is because innovations in communication and information technology has made transportation less necessary. In terms of energy, there have been more breakthroughs in new energy sources from 1980-now than there was from 1945-1973.

There's more I could discuss, but this is long enough. So in short: I do not see a world economy that has stagnated overall. I don't even see a US economy (pre-2008) that has stagnated. I see a redistribution from a certain class of American workers to workers with similar skills in other countries, and to workers with very high skills in the US that can market those skills to a global economy. This doesn't have to be a bad thing, if the government can respond by encouraging innovation by high-skilled workers, and even encourage a lot of compensation for them, but provide for the rest with a fairly robust safety net. And, in fact, the major political cleavages of the present focus on precisely these issues. The political battles aren't about stagnation, but about distribution.

UPDATE: A few folks thought the graphs above are misleading, and they've got a point. So rather than just draw some lines in Powerpoint, I did a more reasonable comparison here. It doesn't change the substantive conclusion of this post, but it was worth doing.


Iamreddave said...

Really interesting post.

In relation to how much computer and telecommunications technology is worth it seems to be around $5000 a year

Which sounds low to me. I think I would take more than that off my wage to keep the internet and mobile.

dd said...

The red line in the first chart isn't right - it shows slowing percentage growth, not constant. You would need a log scale for a straight line to represent constant %growth over such a long period. Ditto the green line in the second chart

Kindred Winecoff said...

Red, I have no idea how to quantify something like that, but I would bet it's more than $5k per capita too.

DD, it's just meant to be illustrative, not perfectly accurate. And I don't see why you need a log scale to show constant growth with an angled line.

dd said...

well think about it - a straight line on a linear scale is representing a constant dollar amount increase per year (that's the definition of a linear scale!). So if you carry the line forward long enough, the amount will have doubled, and therefore the percentage growth every year will be half what it was when you started drawing the line. That's what log scale graph paper is for - cases when you want the same vertical increment to reflect the same proportional change.

I don't think it will make a qualitative difference on the second chart but it might on the first. It's easy to do in Excel; just calculate the CAGR for the pre-1973 period, and then drag and drop a new time series, growing US GDP forward at that percentage rate.

Your general point though seems to me to be spot on and I can scarcely believe it - in all the favourable reviews of this thing, has nobody else noticed that the entire concept is based on such a horrible error?

Kindred Winecoff said...

DD, I understand what you mean by compounding, but that shouldn't begin post-1973. The data I'm using is WDI (via Google's handy public data tool), and from that the pre-1973 trend looks pretty linear. And in fact a linear trend is consistent with Cowen's view, ie the one I'm criticizing, which is that growth has slowed since 1973 because of a stagnation. That's the whole point of Cowen's thesis, and that's the point I'm attacking.

I'm sure you understand... I don't think that's what actually *has* happened, I think that's what Cowen *believes* has happened. Those illustrations are simply trying to capture what the world should look like if the theory I'm criticizing were true. You may be right that this distinction is more important on the first graph than the second, but that's why I drew two lines on the first one.

Perhaps I should've tilted/bent my lines a touch more (I just painted 'em up in Powerpoint rather than doing any sort of real simulation), but again I was going for an illustrative point rather than a precise estimation.

From what I've heard (from feedback) Cowen has heard this criticism before in various forms. And his reaction has been that he just doesn't believe mean GDP/capita for the US. He thinks a lot of GDP is ephemeral, not real. So he thinks median GDP/capita is closer to reality, because he doesn't think any significant innovations have occurred since the 1890s or so. At least that's my impression of his view, although I can't recall seeing him put it quite that way. If that is his view I think that's pretty weak, but whatever.

Since the view that GDP is wrong is essentially non-falsifiable, I think the strongest test of his thesis is the comparative international data. And even in the US case, situating the US experience in an international perspective is necessary. That's my point #2 (building off of #1 and leading into #3), and that's really what I wanted to focus on. I did a bit more throat-clearing than I wanted to to get to that point, but I think that's what's really interesting. I'm not sure if that emphasis is what you think is spot on, or just pointing out the different b/t median and mean GDP, but either way I'm happy that you got something out of it.

Anyway I'm disappointed that Cowen doesn't really deal with the international dynamic at all. I was hoping to provoke a response from him on those points with this post, and a few folks pushed it on blogs and Twitter, but it doesn't look like any response is forthcoming. I guess I don't have that much clout. But I think the intl dynamics are really important.

Alex Tabarrok said...

DD is correct. You are forgetting about the magic of compounding.

In the second figure, for example, you are projecting something like a $5000 increase over 1960-1975 and saying if growth had increased at $5000 every 15 years since 1975 we would be much poorer today than we actually are! When what you want to do is look at the percentage change over 1960-1975 and project that. You can do that with a log scale or by finding the growth rate over 1960-1975 and then plotting a curve with that growth rate.

Note that all growth processes start out flat and then curve upwards dramatically!

Anonymous said...

What about the median GDP of other advanced economies? Those numbers should indicate if this is a systemic problem or the result of specific U. S. policies.

Kindred Winecoff said...

DD, AT, I took your suggestions in a follow-up:


Doesn't really change my substantive conclusion, but there you go.

Note that *my* argument is that growth hasn't stagnated, which is what the new graph shows (at least pre-2000s). The graphs in this post were intended to depict what things might look like if TC's argument were true (ie if growth *had* stagnated), although maybe I wasn't charitable enough with my line-drawing.

Anon, I don't know of a data source that has median GDP/capita collected in time series cross-national form. I could probably get it by going to each country's finance ministry's web site, but that seems like too much effort for a blog post. At least right now... I'm busy with some other stuff. But that's why I linked to the ILO study which argues that intra-country and inter-country inequality has been increasing.

Tom Grey said...

Great post, excellent points. I love MR, but don't believe Great Stagnation much.

"There's been enough growth, it just hasn't gone to the median earner. The result has been higher inequality."

I think this is the crucial Stagnation issue; a new, superstarish distribution.

Where hard-working low skill workers don't produce much more value than the unemployed lazy workers -- so how much should either or both of those groups benefit from (forcible) redistributing from the 5* median income earners?

dd said...

Doesn't really change my substantive conclusion, but there you go

Yes I didn't think it would (how could it? your conclusion is right!), but we graph quibblers are insatiable.

Kindred Winecoff said...

Tom, that's obviously a normative question. I'm not any kind of strict egalitarian, but I tend to support measures that promote broad-based prosperity if only because I think the political economy of extreme inequality is bad.

DD, in other contexts I am too, so thanks for the correction. I can sometimes get sloppy in my blog-haste.

There Is No Great Stagnation, Only Great Redistribution




Add to Technorati Favorites