Monday, July 7, 2014

When The Rate Of Return And The Rate Of Growth Do Not Matter Much For Piketty

So, I have finally gotten around to reading Piketty's monumental book.  Something struck me right away that I think many ignore, at least I have seen no one comment on it.  His emphasis on the importance of r > g, (rate of return on capital as he defines it greater than GDP growth rate) seems overblown.  Now I must admit that he repeatedly warns that the path of the distribution of income and wealth is hard to predict and that many factors are involved.  But it is also the case that he makes little of how often the direction of change does not correspond with the direction of change in this simple equation, that its effect is overwhelmed by other factors.

This can be seen by considering the first two figures in the entire book, which appear in his Introduction and that he declares are the most important in the book, what he is trying to explain ultimately.  The first shows the share of the top decile  of the US income distribution from 1910-2010 and the second shows the ratio of capital to income for Britain, France, and Germany from 1870-2010.  What is striking is how often changes in these have little to do with changes in r-g.  As it is, he argues in the book for the near constancy of r in all these nations over a long period of time, which means effectively that it is changes in g that largely drive this gap.

So, what do we see?  For the US the top decile starts out well off in 1910, reaching a peak just before the Great Crash, then it falls to the 1970s after which it rises again, reaching nearly its 1920s level by 2010.  However, what is not discussed much is that nearly all of the decline in inequality occurs in one tiny period: World War II.  He spends almost no time in this large book focusing on that peculiarity.  It is true that g was high in this period, but not that much higher than during the 1960s or for that matter the 1920s.  In the 1960s there was little change in this top decile share, which essentially got as low as it was going to go by 1945 and basically bounced around after that not moving much until it took off after 1980 or so.  And  in the 1920s rather than falling, the top decile share actually rose rather then fell.  It fell initially at the beginning of the Great Depression when g fell hard.

He does provide an explanation of the 1920s and early 30s, but it goes against his basic story about g.  What led to the peak at the end of the 20s?  A sharp rise in capital gains income.  And the following decline was the reversal of this with the stock market crash.  OK, that is almost certainly correct, but it goes against the story about g.  And how is it that essentially all of the decline in income inequality from 1931 to the 1970s occurred in the early 1940s?  He really does not focus on this, but he does indicate what is up, and it is important for his policy discussion at the end of the book.  That is when the US federal tax code went into its most progressive state, with the top marginal income tax rate moving up to the 90+% level in 1940, just in time for the war, where it would stay until the mid-1960s tax cut.  Of course there was a surge of g, and we know that increased employment of women and minorities helped lower inequality, but it is striking how much change in inequality was in these five years alone.

The story in Europe is slightly different, and, of course, it focuses on wealth rather than income.  There, with some minor differences between the three countries, wealth compared to output starts out very high in 1870 and stays there until  World War I, then only to massively collapse in all three countries, although not quite as much in the UK as in Germany and France.  Capital makes a slight recovery during the 1920s, when g was up compared to the previous decade (ooops, going the wrong way), only to decline after 1930 steadily, reaching a minimum in 1950 for all three.  For France and Germany it starts rising again, steadily in the case of Germany and continuing until now.  With Britain it sits at the same level 1950-1970, and then starts rising. 

But here is the catch.  During the 50s and 60s, economic growth was much higher in both France and Germany than in the previous two decades, and also it was higher than in Great Britain, which put in a sclerotic performance during those decades.  But if a higher g is supposed to be associated with less wealth inequality, that is the exact opposite of what we see here.  Wealth inequality declines during the troubled 30s and 40s, only to reappear in the high growth 50s and 60s in France and Germany, while failing to do so in the more stagnant UK. 

Now he does provide quite detailed stories of what was going on in these nations, although his focus is most detailed for France.  War, inflation, and bankruptcy destroy fortunes, although the direct damage of war is not really all that important.  The end of all this allows for capital to reaccumulate, at least in France and Germany, if not Britain.  But this simply emphasizes how these other factors can easily overwhelm what is going on with g.

Indeed, the vast majority of the decline is not in the Depression or the much more physically destructive World War II, but in the decade of World War I.  Here is where capital simply plunges, falling by more than half in both France and Germany, and by about a third in Britain.  What is curious is that the largest part of this decline is in foreign holdings, which simply collapse.  His discussion of the increase in foreign assets in the nineteenth century involved the colonial empires of these nations, although this was not so important for Germany, which he really does not discuss.  But, while Germany lost its foreign empire, if anything those of Britain and France increased due to  WW I, reaching their peaks in the 1920s as they picked up former colonies of Germany, as well as ones in parts of the former Ottoman Empire.  Piketty simply never acknowledges this.

What else happened?  Oh, the Bolshevik Revolution, which led to the repudiation of tsarist Russian debts, many of whose bonds were owned in these countries, and also the expropriation of foreign company holdings in Russia.  Needless to say this has nothing to do with g.

So, a number of critics have argued that policy really drives all this more than his grand dynamics of r and g.  Offhand looking at his own numbers, this would appear to be the case.  In many decades what is happening to distribution is going in quite the opposite direction from what should be the case if one just looks at r and g (mostly just g).  If it is not policy that is causing this, it is some exogenous shock.

Of course for his current policy discussion, he argues that everything is now working to increase inequality.   The growth rate has fallen and policy has moved to favor the wealthy and those who inherit (particularly in the US on the latter).  So, policy must be changed if we are not to end up a patrimonial capitalism.  On that he probably is correct, but he may well have overstated the role of his grand dynamics and understated the role of everything else in his discussion, even as he provides many caveats on this point.

Barkley Rosser

7 comments:

Peter Dorman said...

Technically you're quite right, Barkley. The way I read Piketty, his r>g mantra is that this is what you might call the default or baseline condition for the accumulation of capital relative to income. He thinks it's always positive in a capitalist society and more positive as g falls for structural reasons (since r remains about the same). Then, on top of that, you have policy. He regards the upsurge of inflation and high tax rates of the first half of the 20th c. as exceptions attributable mostly to war. (Kuttner properly takes him to task for downplaying the purely political dimension in the US, although Piketty does go on about the US as global tax leader.)

For me, this is the main thing that Piketty has in common with Marx. Both want to argue that there is a fundamental tendency to capitalist development, and that politically or institutionally driven alterations are in some sense secondary or epiphenomenal.

Dan Kervick said...

Barkley, I discussed some of the issues you are raising here in "Why Is r > g So Significant for Piketty?" on my blog, and in several previous posts.

Dan Kervick said...

Barkley, you say that "it is true that g was high in this period [WWII], but not that much higher than during the 1960s or for that matter the 1920s."

But the US economy doubled in size during WWII, and growrh rates were indeed much higher. For the years 1939 to 1944, the real growth rates in US GDP were:

7.98%
8.81%
17.70%
18.90%
17.04%
7.99%

So it seems to me that is consistent with his story about the importance of g.

You also say with regard to the 20's and 30's:

"He does provide an explanation of the 1920s and early 30s, but it goes against his basic story about g. What led to the peak at the end of the 20s? A sharp rise in capital gains income. And the following decline was the reversal of this with the stock market crash. OK, that is almost certainly correct, but it goes against the story about g."

But for Piketty, g is the growth rate of national income, which includes all income from capital, and that includes capital gains. (See pg. 18) So changes in distribution due to changes in capital gains income are consistent with a story about changes in g.

Fianlly, you say:

"But, while Germany lost its foreign empire, if anything those of Britain and France increased due to WW I, reaching their peaks in the 1920s as they picked up former colonies of Germany, as well as ones in parts of the former Ottoman Empire. Piketty simply never acknowledges this."

"What else happened? Oh, the Bolshevik Revolution, which led to the repudiation of tsarist Russian debts, many of whose bonds were owned in these countries, and also the expropriation of foreign company holdings in Russia. Needless to say this has nothing to do with g."

But Piketty does emphasize the importance for France and Germany of expropriations due to revolution and decolonization, and the extent to which the reduction in capital was due to a variety of policies aimed at reducing the importance of capital. (See pp. 148-9)

Also, national income, for Piketty, is defined as GDP + net income from abroad - depreciation. So the loss of foreign assets, and their annual yields, does have something to do with g.

Nevertheless, you are right to stress the importance of other factors behind the changes in the distribution of both wealth and income. Many of these factors are discussed in chapters 7 through 12. They include:

- Differences in the rates of saving maintained by different classes of income recipients.

- Differences in the rates of return to capital earned by different classes of wealth owners.

- Changes in the mortality rate.

- Changes in the ratio of average wealth at death to average wealth overall.

- Cultural norms and power relations that seem to be at work behind the tremendous rise in labor income inequality in the English speaking countries.

Peter said...

"That is when the US federal tax code went into its most progressive state, with the top marginal income tax rate moving up to the 90+% level in 1940, just in time for the war, where it would stay until the mid-1960s tax cut."

And the late 40s had inflation which helped reduce wartime debt. In Europe:

"War, inflation, and bankruptcy destroy fortunes, although the direct damage of war is not really all that important."

Again inflation. But in an interview Piketty says inflation is too disorderly. It hurts low income savers. Progressive taxation is more orderly and desirable. But in the absence of taxation inflation would help.

rosserjb@jmu.edu said...

Peter,

Thanks for your comments. I think inflation was more important in the 20s than the late 40s, by a whole lot.

Dan,

Sorry I failed to cite you. Had not seen your blog on this. Lots of people have been commenting, of course.

I did note that there was a surge of g in early 40s, so I did not say that was contrary to the story, just that the change is so dramatic, and I think it is both factors, the rapid growth/expansion of labor force, and the high rate of marginal taxation at the top.

I do not think your reply on the late 20s really cuts it. Yes, of course capital income counts as income, but the story Piketty is arguing is that as g rises the share of income going to capital falls. But that is not what happens at the end of the 20s, nor does that fit for France or Germany in the 50s and 60s. Yes, he says recovery is helping capital, but higher g is supposed to lower capital share. That is my point. The story he is telling does not fit his deeper story about r and g.

Regarding decolonialization, this happened for Germany in the 20s, but just the opposite happened for both France and Britain. Their colonial empires expanded after WW I, reaching their all time peaks. So, something strange is going on with this collapse of foreign assets. Part of this is indeed expropriation and repudiation in Russia, but those holdings were not nearly sufficient to explain this collapse, particularly given that the colonial empires were expanding. There are loose ends here.

rosserjb@jmu.edu said...

Peter,

Thanks for your comments. I think inflation was more important in the 20s than the late 40s, by a whole lot.

Dan,

Sorry I failed to cite you. Had not seen your blog on this. Lots of people have been commenting, of course.

I did note that there was a surge of g in early 40s, so I did not say that was contrary to the story, just that the change is so dramatic, and I think it is both factors, the rapid growth/expansion of labor force, and the high rate of marginal taxation at the top.

I do not think your reply on the late 20s really cuts it. Yes, of course capital income counts as income, but the story Piketty is arguing is that as g rises the share of income going to capital falls. But that is not what happens at the end of the 20s, nor does that fit for France or Germany in the 50s and 60s. Yes, he says recovery is helping capital, but higher g is supposed to lower capital share. That is my point. The story he is telling does not fit his deeper story about r and g.

Regarding decolonialization, this happened for Germany in the 20s, but just the opposite happened for both France and Britain. Their colonial empires expanded after WW I, reaching their all time peaks. So, something strange is going on with this collapse of foreign assets. Part of this is indeed expropriation and repudiation in Russia, but those holdings were not nearly sufficient to explain this collapse, particularly given that the colonial empires were expanding. There are loose ends here.

Dan Kervick said...

Barkley, Piketty doesn't say that an increase in g will lower the capital share. The capital share will continue to increase so long as β is less than s/g. And assuming a constant r, the rate at which the capital share increases from one year to the next is

(s/β - g)/(1 + g)

So increasing values for g can in some cases only slow down the rate of increase in the capital share, but fail to make the capital share move in the opposite direction if β is sufficiently low and s sufficiently high. In the post WWI period, France and Germany had extremely low capital-to-income ratios, so you could get a rising capital share even with rising g.