September 08, 2004
Growing Public, Peter Lindert
Hi all. A little change of pace from the coffee mug guy today. I've been reading Peter Lindert's Growing Public: Social Spending and Economic Growth Since the Eighteenth Century again. It's a troubling book that kicks economic received wisdom in the teeth, and I am puzzled at the lack of attention it's gotten in the blogosphere. (Two posts of Tyler Cowan's, here and here.)
Why is it such a troubling book? Well, here is Lindert's seventh of his nine conclusions:
The net national costs of social transfers, and of the taxes that finance them, are essentially zero. They do not bring the GDP costs that much of the Anglo-American literature has imagined. Accordingly, differences in those costs play almost no role in either the rise or the deceleration of social spending's share. No Darwinian mechanism has punished the bigger spenders.
Readers of The Economist know that one major theme in the Anglo-American literature is that welfare kills economic growth. It's widely used to explain Eurosclerosis (as opposed to, I dunno, core-periphery issues in EU policy), and it's become something of a conservative-libertarian shibboleth in the United States. "There ain't no such thing as a free lunch", or TANSTAAFL, a phrase the influential Missouri-born science fiction author Robert Heinlein countrified from the writings of the economist Alvin Hansen.
Turns out there might actually be a free lunch. Lindert's regressions are pretty good, unlike a lott of ideologues I could name. I suspect much of the silence has to do with Lindert's interpretive framework, which follows Albert Hirschman's little book, Exit, Voice, and Loyalty, and is rather different from the simplified models of public choice theory and class interest that many people, either consciously or unconsciously, use.
There's some food for thought for this US election year as well. As Lindert comments,
Nor is the puzzle strictly international. Within the United States since the 1960s, social transfers have taken a rising share of state product, and the variance in their generosity has also risen -- and has been positively, not negatively, correlated with the level and growth of state product per capita. How can the generous states, like Connecticut, New Jersey, and California get away with giving out more generous welfare and other transfers year after year? Why haven't they grown more slowly than other states? Why haven't businesses deserted them, leaving them with fewer firms and more welfare families?
Like I said, I'm surprised at the lack of commentary Lindert's book has received on the Internet. More on it later.
Here are Lindert's nine conclusions, from pages 20 and 21 of Growing Public:
1. There was so little social spending of any kind before the twentieth century primarily because political voice was so restricted.
2. The central role of political voice is shown by an exceptional case. Both Britain's relatively high poor relief in 1782-1834 and its cutbacks in 1834 and 1870 fit the changing self-interest of those with voice.
3. Similarly, just noting the interests of those with voice helps to explain Chapter 1's education puzzles: Why did Germany and laissez-faire North America lead the way in tax-based public schooling, and why did Britain lag behind in the nineteenth century? How did the United States remain a leader in educational attainment, yet end up ranked about fourteenth in students' test scores? Again, the concentration of voice was the enemy of education.
4. The great advance of social spending since 1880 is explained partly by the same political-voice motif, partly by population aging, and partly by income growth. Roman Catholicism was a negative influence on taxes and transfers before, but not after, World War II.
5. Postwar welfare states developed more fully in countries where the middle and bottom ranks traded places more and were ethnically homogenous.
6. The same forces that explain the growth of social spending until the 1990s carry implications for the future of social spending in all regions - in the affluent OECD countries, in the transition countries, and in Third World countries. In Western Europe, the political power of the elderly and the generosity of their public pensions have already matured and will fade. The social transfers that aging societies have supported will not decline as shares of GDP, but the generosity of pensions per elderly person will decline. Support for the elderly will also be under pressure in the formerly communist countries. Among prospering countries in the Third World, however, pensions will probably become more generous as income grows.
7. The net national costs of social transfers, and of the taxes that finance them, are essentially zero. They do not bring the GDP costs that much of the Anglo-American literature has imagined. Accordingly, differences in those costs play almost no role in either the rise or the deceleration of social spending's share. No Darwinian mechanism has punished the bigger spenders.
8. That large social programs have cost little in practice is consistent with the rise of European unemployment since 1970. Differences in social insurance did play a role in the OECD differences in unemployment over time and space, but only a partial role. Furthermore, the loss in output was less severe because those who remained out of work tended to be less productive workers anyway. Therefore any percentage loss of output tends to be much smaller than the percentage of jobs lost.
9. Two general principles seem to explain why the welfare state does no net damage to GDP per capita and why welfare states will not collapse. The first is that high budget democracies show more care in choosing the design of taxes and transfers so as to avoid compromising growth. The second is that broad universalism in taxes and entitlements fosters growth better than the low-budget countries' preference for strict means testing and complicated tax compromises.
Posted by coyu at September 8, 2004 12:16 AMHeh heh. I think I'm going to deconstruct his last paper on the way to a PhD. Right off the bat, I recall that the R^2 on those regressions implied that the model left _much_ to be explained. (Which, of course, appears to be a distressingly common failure.)
That said, I'll be sure to catch this extended follow-up.
Bernard "Battle!" Guerrero
Posted by: Bernard Guerrero at September 8, 2004 03:41 PMHey Bernard,
There's an extended discussion by Robert Margo on the econometric model Lindert uses here. It's not particularly damning; basically Margo wishes Lindert had used a somewhat different and perhaps smaller set of variables in his model to make his causality clearer, but that he doesn't disbelieve the results. (And, after all, Lindert's data sets are available through the Cambridge website; you can roll your own regressions should you choose.)
As for the unexplained residuals in R^2, yeah, they're sometimes high, but they're not very analyzable. This isn't psychohistory! One trend I have noticed is that the outliers tend to be countries with 'isolated' political histories, ones resistant to borrowing from foreign models: the US, Britain, Switzerland and Japan. However, they don't clump in a consistent direction. For example, premature years of life lost in Japan are much lower than the trend line, while for the US, they are much higher, despite similar low percentages of social transfer per GDP.
The Scandinavian countries also tend to clump together, not really a surprise; and so do the nouveau riche European countries. But I am not sure whether Lindert's model would be improved with a variable for belonging to the Nordosphere. (This would be the upshot of Margo's suggestion to use Barro's religion variable anyway. But what does that actually mean? How does the number of Lutheran pastors per capita translate into anything?)
C.
Posted by: Carlos at September 8, 2004 05:25 PMOkay, I'll make my little micro-blog a venue for discussion on this, if it helps.
Mind you, my impression is that this sociological/economic finding is rather analogous to the mathematical finding that "circular shapes are very often very round" ...
Posted by: Pouncer at September 8, 2004 08:03 PMLindert's findings are actually rather subtle. Turns out that the high welfare states tax labor relatively more and capital relatively less than low welfare states. For instance, the top marginal tax rate on dividends is lower in Sweden than it is in the US. This keeps investment high, which helps maintain growth.
At the same time, the high welfare states also have lower effective marginal tax rates on the poor, usually by allowing them to keep their benefits should their income somewhat increase. This has the opposite effect of the punitive means testing used in the United States before Friedman's Earned Income Tax Credit was adopted, which created a strong disincentive for the poor not to earn (or at least, report) outside income.
There are added fillips. For instance, the high welfare states' unemployment programs tend to be designed to weed out the net drags on the work force, which is probably why French hourly worker productivity is higher than that of the US. After all, why should a French business hire a lemon worker?
Finally, the high welfare states all have strong government-mandated parental and maternal leave provisions -- usually on the order of 1% of GDP. I think this probably saves 2% of GDP losses from stress alone.
There's enough here to annoy both the "eat the rich" crowd and the "starve the beast" types, I think.
C.
Posted by: Carlos at September 8, 2004 10:28 PMIt is an interesting study... but as a scientist I feel that it lacks consideration of all the varibles. Technological advances of the 19th-20th centuries increasingly require less "output" of labor while still increasing production. For example, now one farm family can feed hundreds to thousands of people with only added seasonal workers added. One factory can put out millions of bolts of cloth. Etc.
Thus the work output per person required to sustain or even grow an economy has decreased to a fraction of what it used to be. And even the workers who are productive are expending more efforts on creation of art, comforts, services etc rather than basic needs.
Thus sociey as a whole can carry more underproductive workers or even total dead weight and still grow.
This effect continues to increase at slightly behind the rate of technological expansion.
1) What's his definition of "voice" here?
2) I'm a bit confused by the references to English history. I thought the Poor Laws went right back to Elizabethan times, with only modest tweaking until the 19th cen.
3) Why does he think that pensions will decrease in Eastern Europe, but not in "prosperous countries in the Third World"? In 2020 -- or 2030, 2040, or 2050 -- Romania and Bulgaria will have (proportionately) a lot more old folks than India or Thailand.
More, but those will do for starters.
Doug M.
Lynette, technological productivity growth in economic history is usually considered as a contributing factor to national wealth, although this method does have its difficulties. So the average Briton in 1800 is computed to have an income of only ~ $1000+ current US dollars, even though said Briton may have been at the technological cutting edge of the time, and almost certainly did not consider himself or herself extremely poor.
It turns out that to keep a modern economy even at a break-even level, it still has to produce enough to match the depreciation of its assets. That's usually considered a fixed percentage of the size of the economy. (Oddly, the depreciation rate does not seem to have changed much since the Industrial Age began.) Anyway, the upshot is that instead of the amount of labor per person needed to maintain an economy decreasing as labor productivity increases, it has stayed roughly steady as an economy expands with productivity.
I should note that many pre-industrial agricultural societies had massive rural underemployment, part of what some guy once called "the idiocy of rural life". And hunter-gatherers only have to work two, three hours a day to sustain their economy, such as it is.
Doug: 1) "voice" is used as a shorthand for political representation, not necessarily democratic. 2) Lindert is starting as soon as the surviving records become good to analyze econometrically, the late 18th century. The Netherlands, by the way, was the world leader at that time, with possibly up to 1.5% of its GDP used in social transfers, combined government, church and private. 3) Exactly.
C.
Posted by: Carlos at September 9, 2004 01:15 AM? Exactly /what/? I'm not catching this.
viz., why won't the old folks of Romania -- who will be something like a quarter of the population, if not more -- just "voice" themselves more pension-y goodness?
Another q.: are high welfare states' unemployment programs really /designed/ to "weed out the net drags on the work force"? I grant that France, for instance, has high hourly worker productivity because a lot of low-quality workers are on welfare instead. But was this a deliberate policy, or a happy(?) side effect of a an entitlement system that was chasing other goals altogether?
Also, is this really such a good thing? I've recently been working with a lot of high-energy type-A workaholic consultants. These people are probably ridculously productive, both in gross and per hour. If we put everyone but these few on welfare, I don't doubt that worker productivity would soar... but I don't think it would be good for the economy as a whole.
Unless a worker is so bad s/he is actually destroying value, it seems that we'd all be better off if that person was employed (though the individual worker might strongly disagree). Am I missing something?
Doug M.
Hi Doug,
1) Lindert predicts that in transition nations (very roughly, and some of this may have already happened in countries like Poland and Hungary) social transfer spending will rise with voice, level off due to other voice concerns that this is unsustainable for the economy, but as percent share of GDP, become politically untouchable. Nevertheless, the brute force of demographics will cause pensions to decrease per person.
I understand that Romania has had some wild swings regarding its pensions. So this may be something that works better in theory than in practice.
2a) It's the workers, hm, at the margin that are weeded out. (Interesting double meaning there.) If by "destroying value" you mean "net loss for companies as employed", that's right. Consultants not on line for an early heart attack need not worry.
2b) Some of it is design, and some of it is serendipity. One of Lindert's themes, which I wish was better articulated, was that 'Darwinian' pressures on social transfers tend to play out at the fine-tuning level. The Blind Policy Maker? Sometimes.
C.
PS you're posting as Claudia.
Posted by: Carlos at September 9, 2004 02:15 PM