Larry Bartels responds

by Larry Bartels, guest blogger

Professor Rodrik has kindly invited me to respond to some of the questions and comments generated by his recent post describing my analysis of partisan differences in income growth. I'll briefly address what I take to be the most important points. The relevant chapter of Unequal Democracy includes much more detail regarding these points and others; I hope everyone will read it before deciding to believe or disbelieve the results. Other chapters of the book focus on cultural issues and voting behavior, partisan biases in economic accountability, the Bush tax cuts, the politics of the minimum wage, the relative political influence of rich people and poor people, and more.

Are the partisan differences in income growth merely coincidental?  It is never possible to rule out coincidence in historical analyses of this sort, but the odds are very much against it. The differences in average growth rates for middle-class and poor families are "statistically significant" by conventional social-scientific standards. (A post at www.conservativescientist.com suggests otherwise, reproducing my figure with ludicrously wide "error bars"; but those "error bars" are based on a simple confusion between the standard deviations of income growth reported in one of my unpublished papers and standard errors of the average growth rates, which are quite another thing.) More importantly, the differences appear consistently in the pre-1974 (high growth) and post-1974 (low growth) eras; they persist even when any one or two administrations (or election years, or transition years) are omitted from the tabulations; and they persist even after taking account of a considerable variety of potential confounding factors.

Isn't it unrealistic to consider presidents in isolation from all the other forces affecting income growth?  My book includes a variety of statistical analyses embedding the partisan presidential effects in more general models of real income growth incorporating linear and non-linear time trends, oil shocks, changes in labor force participation, "trickle down" effects of growth at the top of the income distribution, and other factors. The partisan differences in income growth implied by those analysis are, if anything, slightly larger (and more precisely estimated) than the raw differences summarized in my figure.

Aren't all presidents different?  Yes, but the parties' distinctive ideological priorities shape their policy choices with remarkable consistency. Every Republican president in the past 60 years has presided over increasing income inequality, including Dwight Eisenhower in the midst of the "Great Compression" of the post-war decades. And every Democratic president except one (Jimmy Carter) has presided over decreasing or stable inequality. Bill Clinton managed to stem the trend of accelerating inequality, producing significant real income growth for middle-class and poor families, in marked contrast to Republicans Ronald Reagan, George Bush, and George W. Bush.

Isn't Congress more important than the president?  Congress probably has an additional impact on the income distribution, over and above the impact I attribute to presidents. However, there is too little historical variation in the partisan composition of Congress in the post-war era, and that variation is too strongly correlated with time, to produce a reliable estimate of how much Congress matters--so I relegated it to a footnote.

What about post-tax income?  Most of my analyses focus on pre-tax family income from 1947-2005 because that is the longest data series included in the Census Bureau's Historical Income Tables. Post-tax income data are available beginning in 1979, and the pattern of partisan differences in post-tax growth over that period are similar in magnitude to the pre-tax differences in Figure 2.1, with large disparities for poor and middle-class families and smaller disparities for affluent families.

Could income growth patterns be causes rather than consequences of presidential election outcomes?  Incumbents of both parties do well when incomes are growing in the election year; there is no consistent tendency for voters to choose Democrats when income growth is robust or when growth rates for different income classes are more equal. In any case, the largest partisan differences in income growth, by far, occur in the second year of each administration. It is very hard to see how those differences could cause election outcomes two years earlier (and I show that they have no bearing on election outcomes two years later). However, it seems quite plausible to see them as reflecting the policy choices of new (or newly reelected) presidents in their first "honeymoon" year, when their political influence tends to be at its peak.

How do presidents produce these substantial effects?  One of my aims in writing Unequal Democracy was to prod economists and policy analysts to devote more attention to precisely that question. Douglas Hibbs did important work along these lines in the 1980s, documenting significant partisan differences in post-war macroeconomic policies. He found that Democrats favored expansionary policies producing substantially higher employment and growth rates, while Republicans endured and sometimes prolonged recessions in order to keep inflation in check. (Not coincidentally, unemployment mostly affects income growth among relatively poor people, while inflation mostly affects income growth among relatively affluent people.) In recent decades taxes and transfers have probably been more important. Social spending. Business regulation or lack thereof. And don't forget the minimum wage. Over the past 60 years, the real value of the minimum wage has increased by 16 cents per year under Democratic presidents and declined by 6 cents per year under Republican presidents; that's a 3% difference in average income growth for minimum wage workers, with ramifications for many more workers higher up the wage scale. So, while I don't pretend to understand all the ways in which presidents' policy choices shape the income distribution, I see little reason to doubt that the effects are real and substantial.

Comments

[Nathan Smith]

"Douglas Hibbs... found that Democrats favored expansionary policies producing substantially higher employment and growth rates, while Republicans endured and sometimes prolonged recessions in order to keep inflation in check."

This explanation is congenial to the no-free-lunch instincts of economists. It just seems to be too good to be true that electing a Democrat is a free pass to higher economic growth. This could explain why the voters keep electing Republicans: there's a time to spend, and a time to tighten our belts.

The minimum wage story assumes that minimum wages affect income distribution. The case for this has, I believe, little theoretical or empirical support.

[Rich C]

I just wanted to thank Larry Bartels for his guest post, which I think is very interesting and makes me want to read his book even more eagerly. I want to note three issues: first, JK Galbraith - in Created Unequal - illustrates the close connection between wage inequality and unemployment. Even if Presidents can't directly control the unemployment rate, they have at their disposal a lot of tools (including appointing the Fed Res Chair, and managing the dollar exchange rate) which can help to balance the economy and either keep unemployment low or goose it to drive down inflation. Second, in addition to the minimum wage (which requires Congressional action), Presidents have the power to appoint members of the National Labor Relations Board (much more union-friendly under Democrats), and to manage the Federal Government's huge number of contracting relationships with private sector employers (eg Project Labor Agreements under Clinton). Finally, Larry suggests in his post that taxes and transfers have become more important recently than expansionary policies; wouldn't that affect the post-tax income distribution rather than the pre-tax growth rates he's reporting? Maybe there's some connection to the pre-tax growth rates, but its not easy to see exactly what those would be , or why they would be so powerful.

[Matilde]

Seems that a first step for understanding the mechanism by which presidential administrations impact real income growth would be to break out growth in income by source of income (i.e. wages, transfer payments, dividend income, etc.). I confess I haven't read the book, but Census data do provide these breakouts, have you examined them?

[shah8]

I wish people would talk more about regulatory and resource management aspects of executive influence over economic outcomes. For instance, women and minorities make up over 50% of the population...hardening and softening of outlook towards women and minorities should be a pretty big bump in inequality up and down. Regulatory effort also affects businesses as well. Futhermore, things like the Plaza Accord, and the shaping of the S & L Bailout are also determined by presidents. Thinking on the S & L bailout, it could just be that how a president *fixes* problems introduced by Congress can determine inequality.

[Remy Piwowarski]

I would agree that more attention should be paid to regulatory and resource management aspects.

Scrapping welfare programmes or keeping minimum wage unchanged seems to have a larger and more permanent influence on the poor than economic fluctuations.

Remy Piwowarski
Exec. Director
Economics International Open Forum
http://www.economicsinternational.blogspot.com/

[CJS]

Re: "Isn't it unrealistic to consider presidents in isolation from all the other forces affecting income growth?"

I don't quite understand how this works if the Presidential effect is somehow supposed to be isolated from the Congressional effect (yes, I did see the separate listing - no, I didn't understand how it answered the question) and the Fed effect (which I haven't seen mentioned at all but which I would wager is the most important).

[CJS]

Re: "Isn't it unrealistic to consider presidents in isolation from all the other forces affecting income growth?"

I don't quite understand how this works if the Presidential effect is somehow supposed to be isolated from the Congressional effect (yes, I did see the separate listing - no, I didn't understand how it answered the question) and the Fed effect (which I haven't seen mentioned at all but which I would wager is the most important).

[Sean]

One question on the presumption of the book, though: If these effects are just being proven now (to be fully vetted when published) how do you presume to speculate on voter myopia and psychology? The Occam's Razor answer would simply be that it doesn't factor with them.

[William Perera]

My high school economics teacher has a saying which he stole from some professor at UC Davis; “There are two parties in this country, both of them are right-wing.”
Case and point: When Bill Clinton was campaigning for the first time, he pledged to implement a North American Fair Trade Agreement, rather than Bush Sr.’s proposed North American Free Trade Agreement. Among other differences from NA(Free)TA, the fair agreement would require Mexico to boost its environmental and wage standards over time and ensure better protection for American industry. Of course, once elected Clinton decided to implement Bush’s NA(Free)TA with no changes whatsoever. So, no matter who you voted for you got the same thing.
Then again, I’m just a high school student so what do I know about history…
;)

[tt]

Interesting stuff.

What if the dependent variable is unemployment rather than the income distribution?

Some policies may be progressive in the income space but regressive in the employment space.
For example, a higher minimum wage may at the same time raise the wages at the bottom percentile and reduce employment at the bottom percentile.

[HispanicPundit]

I'm curious to hear what Bartels thinks about Cowen and Tabarrok's response on Marginal Revolution blog?

[Elliott Parker]

I wrote a brief paper a couple years ago showing some similar results for GDP growth and a host of other variables. The paper is available online:
Parker, E. (2006), "Does the party in power matter for economic performance?," UNR Economics WP-06008, url
http://www.coba.unr.edu/econ/wp/papers/UNRECONWP06008.pdf.

[Elliott Parker]

The link is http://www.coba.unr.edu/econ/wp/papers/UNRECONWP06008.pdf (without the period at the end).

[RedCharlie]

It's just like Harry Truman told us, "If you want to live like a Republican, you better vote for a Democrat!"

[Chris Ellison]

As I reported on the original post this graph has been debunked. See here:

http://corner.nationalreview.com/post/?q=OTVkNDlmZmJmOTNlYzRmZmM3N2Q5NjQyZWJkZTU5YzM=

The outcome changes as lag times are moved making the association spurious. Other variables actually show almost as strong an association yet no one would think they are factors. Read the entire debunking.

[Penis Enlargement]

Blog are goods for every one where we can get more knowledge nice job keep it up !
http://free.7host05.com/herbals/

[skeptonomist]

Bartels seems to overlook the fact that the census data on income inequality, in the plot that has been much reproduced, are much better explained as a result of a gradual change over time:

www.skeptometrics.org/IncomeGrowth.htm

For example, if the data in his income inequality plot are replotted to show the period 1947-1976 vs. 1977-2005 instead of Democratic vs. Republican president, there is a distinctly better separation. The supposed separation by presidential administration does not appear in other views of the data - what is seen is a decrease in inequality to about 1970 and an increase thereafter.

If Bartels's handling of this issue is representative of his general approach to other data I see to reason to buy his book.