Ryan Avent raises this interesting comparison:
Japan and France are both highly developed countries, with a high level of technological and institutional congruence. Their levels of per capita output are nearly identical. But France raises much, much more revenue as a share of output than does Japan. Why should revenue share have, apparently, very little effect on that output relationship but a comparatively enormous effect on the relationship between outputs in western Europe and America?
I believe that a comparison of Japan and France could be used to make exactly the opposite point. Here is some hours worked data, and Mankiw’s GDP per capita data:
US: GDP/person: $46,443 Taxes: .282 Hours worked: 1777
Japan GDP/person: $32,817 Taxes: .274 Hours worked: 1828
France: GDP/person: $33,744 Taxes: .461 Hours worked: 1346
Let’s start with GDP/person. The US is much richer than France and Japan. Japan is just slightly poorer than France. I can certainly see why Avent lumped Japan and France together. And they have other similarities–distinctive national styles, good cuisine, fast trains, lots of nuclear power, etc. But notice something else. The Japanese work roughly as many hours as Americans. More specifically, Japanese tax rates are a bit lower than US tax rates, and they work a bit longer hours. The French tax rates are much higher than in American, and they work far fewer hours. (And of course in the past when French tax rates weren’t higher than ours, they worked just as much as we did.)
I know I’ve already done far too much armchair theorizing, but I can’t resist one more stab at it. Maybe Avent is wrong. Maybe Japanese GDP/person would be nothing like French GDP/person if they had similar tax rates. Maybe we should be grouping the US and France together, and Japan and Italy. Indeed, maybe Japan suffers from the same problems as the Congo and Afghanistan, just to a much lesser extent.
I’m sure that by now you’ve concluded that I have gone completely nuts. We all know that Japan has one of the most advanced and sophisticated economies in the world. The worker productivity is legendary. Their educational system produces students that are much better at math than our students. Visitors to Tokyo marvel at how everything is kept clean and tidy. How can these stereotypes be wrong? Tokyo doesn’t seem like Rome, Osaka doesn’t seem like Naples. But if the data is correct, then Italians are actually significantly more productive than Japanese workers. The Japanese work 20% more hours, but produce only 12% more output than Italy. Japan is basically a sleepy Mediterranean country that only gets to be included with the France, Germanys and Britains of the world because they work really, really hard. They’re a tiny bit more productive than Spaniards, and less productive than Italians.
Italy: GDP/person: $29,290 Taxes: .426 Hours: 1523
Spain: GDP/person: $29,527 Taxes: .373 Hours: 1745
So yes, I agree with Ryan Avent that there are many things beyond taxes that are important. In my previous post I argued that Japan had a very inefficient economic system, which explained why it raised so little revenue, despite low taxes. But unlike Avent, I think the differences between France and Japan perfectly illustrate my point. France has much higher taxes, and similar productivity to the US. The higher taxes reduce hours worked, resulting in much lower GDP/person. Japan has similar tax rates, but much lower productivity than the US. So it has lower GDP/person for completely different reasons. They work even harder than we do, but are far less productive.
I had recently read about Spain overtaking Italy in per capita GDP, which I found rather surprising. Despite my crude stereotyping of southern Europeans, it is well known that northern Italy has a very sophisticated manufacturing sector. So how did Spain do it? Very simple, they worked harder. And why did they work harder? Could it be lower tax rates?
I won’t contest Avent’s point that Singapore and HK are unrepresentative. I felt I had to pick them as they are the only two developed countries with much lower taxes than the US.
I still don’t agree with him on Texas. Avent argues:
I’d first say that this is a bizarre view of urban economics, and one that entirely ignores the well documented relationship between urban density and productivity. Most of the millions upon millions of people who live in high cost metropolitan areas—note that there are some 60 million people living in the relatively high cost metroplex between Washington and Boston—aren’t doing so because they merely want to interact with smart people and play sophisticate. They do so because it pays, because productivity and wages are higher. They have to be! If workers in such areas weren’t more productive, then they couldn’t afford to pay higher wages, and if they couldn’t afford to pay higher nominal wages, then real wages would be well below levels elsewhere, which isn’t sustainable.
I’m certainly no expert in economic geography, but doesn’t this miss the Balassa-Samuelson argument. Isn’t the argument that only the workers producing traded goods need be more productive? Are barbers, plumbers, tax preparers, etc, more productive in New York than in the Sunbelt, or are they paid more because the cost of living is higher, and the people who are more productive are willing to incur a higher cost of living to reside in NYC and SF?
I also wouldn’t lump the MetroPlex all together. The NYC area is gaining people because of its amenities, and perhaps because of productivity gains flowing from agglomeration. DC gains people for obvious reasons that have nothing to do with its economic model. The other 60% 0f the Metroplex (including Massachusetts but excluding low tax New Hampshire) is gradually losing people to places like Texas. We will lose another Congressional seat this census. Right around Boston and Cambridge things are still vibrant–due to Harvard, MIT, Mass General, etc. But the rest of the state is not that appealing from an economic perspective.
Ryan Avent also argued:
Meanwhile, who is moving to uninteresting manufacturing cities in Texas? The fastest growing metropolitan area in Texas is Austin. Austin’s industry mix is most similar to that in places like Raleigh, North Carolina (which is growing more rapidly) or Boston or San Francisco, where people continue to move.
As far as who is moving there, I don’t know their names. But I seem to recall reading that last year the Dallas and Houston areas each gained over 140,000 nobodies, which is far more than any other city in America. Austin gained 50,000, but because it is much smaller that represented a slightly higher percentage rate of increase than for Houston and Dallas. But I’d say 280,000 people is pretty impressive one year in-migration for the big two. And I didn’t mean to suggest that other places aren’t growing fast. Most have at least slightly more pleasant climates or geography (Arizona, Denver, Florida, etc.) I presume that North Carolina and Georgia also have good economic models. I know that in Georgia there are sizable suburbs that have virtually no government at all, everything is contracted out. That sounds like a libertarian paradise. But I still say Texas has done dramatically better than the rest of the south central US, and I claim it is due to their superior economic model. It’s now hard to believe, but New Orleans was once a rival to Houston for attracting oil companies.
Ryan Avent concludes:
The point, again, is that growth is complicated. Things like tax rate, and tax structure, and regulatory mix, and overall regulatory burden—they all have their place in the story. The kind of analysis that Mr Sumner is using here is just counterproductive to efforts to understand what is really driving disparities in economic performance.
Let me end by agreeing with Avent that it is all much more complicated than I suggested. My original post was confusing because I shifted from one argument to a completely different argument. The first argument was that Mankiw was actually making the supply-side side case that much higher tax rates wouldn’t raise much revenue. I tried to make the point that it was a defensible argument (defending the indefensible.) That apart from obvious special cases like Norway, countries that have dramatically higher tax rates than the US do not raise dramatically higher amounts of tax revenue, rather they raise equal or modestly higher levels of revenue. I still think that is true, but will freely concede that my blog post is far too weak to convince any skeptic. I did not mean to suggest that taxes were the only factor driving economic development. (BTW, a commenter mentioned that Mankiw used the wrong tax data, and that correct data would have actually made his case appear even stronger.)
My second argument was for using the concept of revealed preference when comparing economic systems. Progressives often portray Texas as a nightmarish place with savage inequalities, underfunded schools, a weak safety net, the death penalty, a high murder rate despite the death penalty, a gun in every pick-up truck, etc. These stereotypes are probably partly true and partly false. But the fact remains that Texas is a very appealing place for migrants. Not just for selfish rich people escaping high tax states. It is also appealing to working class whites, Hispanics, even for 100,000 poor black refugees from Katrina (who were welcomed by Houston with open arms—something the liberals in my hometown would never do.) I still think this revealed preference argument is a very powerful one, but would be the first to admit that my post was far from a definitive study of the issue. Indeed can any study be definitive?
Update: The commenter I referred to is Mark Sadowski. He says Mankiw used flawed Heritage Institute data. The OECD shows French tax rates as .436, and US rates as .283. It also occurred to me that there may be a selection bias in French productivity estimates. Suppose hours worked falls most sharply among lower-skilled immigrants from North Africa, and also young people. Then this bias would make French productivity look a bit higher (compared to the US and Japan) than it would be if we compared people of equally SES.