The Presidency and Distribution

Dani Rodrick posts this chart from Larry Bartels' forthcoming book. It shows that under Republican administrations, incomes grow strongly for rich people, but barely at all for those of more modest means. Under Democratic presidents, by contrast, lower income groups see stronger income growth but everyone's income grows faster than it does under the GOP.

That's striking stuff. Indeed, it's almost too striking. The president's control over domestic policy is pretty circumscribed and public policy has only a limited influence over the economy, so it's surprising to see such a strong effect. I'll be interested to read the book and see Bartels' account in more detail.

Comments

This Graph is really quite a poor way to represent the data set. At first look it says "Democrats make things go down". Since that's the opposite of the actual message, I wonder what graph could have been less-well-suited to convey this information?

Posted by Punning Pundit | April 1, 2008 12:54 PM

simple fix: turn the graph around

Posted by lynn | April 1, 2008 1:02 PM

John Irons has written about this, though I can't find anything online at the moment. The only explanation I remember was that it might be an expected-inflation phenomenon, in which Democratic presidents choose higher-inflation policies than Republicans. The numbers work for unemployment, too (declines under Democrats, rises under Republicans), which is unsurprising given the income numbers I suppose.

Posted by minderbender | April 1, 2008 1:03 PM

Three words with one little word in between:

Marginal Propensity to Consume

If the stylized facts that Robert Solow wrote lo these many many years ago are true, the Republican policy of using income distribution policies that make the capital accumulating rich richer should lead to a higher long run growth path.

Of course, in the long run we're all dead. And, in Summers' rejoinder, death is endogenous to the model.

Posted by jayackroyd | April 1, 2008 1:06 PM

Probably means nothing.

Posted by Korha | April 1, 2008 1:13 PM

Matt, regulatory agencies matter. There is a reason why the S&L crisis occurred.

Posted by Rob | April 1, 2008 1:14 PM

I would think there is a strong correlation with the point in the business cycle and the probability the electorate will choose a Democrat. That is striking though, and I remember that the increased Dem growth holds under a variety of circumstances.

Posted by theCoach | April 1, 2008 1:16 PM

Indeed, it's almost too striking. The president's control over domestic policy is pretty circumscribed and public policy has only a limited influence over the economy, so it's surprising to see such a strong effect.

Republicans are the party of Peace and Depression, Democrats are the party of War and Inflation.

Or so it used to be.

max
['Sounds too good to be true usually is.']

Posted by max | April 1, 2008 1:20 PM

I wonder how much of this effect is the Clinton-era dot-com bubble?

Posted by Sebastian Holsclaw | April 1, 2008 1:21 PM

The graph really is terribly designed. Why is it even a line graph? If you're looking at the income percentile's separately, shouldn't it just be a bar graph?

Posted by Sam L | April 1, 2008 1:23 PM

Democrats were in power 1948-1953, 1961-1969, 1977-1981, and 1993-2001. Except for 1977-1981, that's all pretty booming times for the economy.

Republicans get 1953-1961, 1969-1977, 1981-1993, and 2001-2005. 1953-1961 was a pretty good time (although there was a recesssion in 1958-59, I think.) The Nixon/Ford era, especially the second half of it, was a pretty bad time for the economy. The Reagan/Bush I era was mediocre at best, as has been the Bush II era. So it's not too surprising it works out like that.

Posted by John | April 1, 2008 1:24 PM

The first thing I wondered is which way the causation is going. For example, it seems to me a plausible hypothesis would be something like that Democrats are seen as more likely to increase taxes for the purpose of increasing governmental services, and more people might be willing to vote for that deal under circumstances where they are experiencing strong income growth.

Posted by DTM | April 1, 2008 1:25 PM

Is income growth before or after taxes?

Posted by Al | April 1, 2008 1:40 PM

It's not actually true that everyone does better under the Democrats. As Kevin Drum pointed out about three years ago, on Bartels's statistics the top 5% do about as well under the Republicans, and the top 1% do better under Republicans.

I'm sure this is just a coincidence, completely unrelated to the GOP's pro-rich policies. [Rodrik says that Bartels shows that it is not a result of Democrats coming to power in better times.]

Posted by Matt Weiner | April 1, 2008 1:40 PM

Is income growth before or after taxes?

Before taxes, according to the extremely similar graph posted by Drum.

Posted by Matt Weiner | April 1, 2008 1:44 PM

Whether this study is an accurate reflection of the phenomenon, I think the Republican strategy is extremely short-sighted. Giving money away to your friends can make a few people wealthier in the short run, but it doesn't build an effecient economy- not even an economy biased towards the wealthy. Not to mention that a refusal to invest in the work force (the lower 80%) is certainly going to hurt you in productivity in the long run.

That's why even though I'm in the 95 percentile, I can vote Democratic, confidently supporting my own financial interests and morality at the same time.

Posted by mpowell | April 1, 2008 1:46 PM

Agreed - this is a really poor visualization of the (very interesting) data. Somebody could do it justice by reworking it a bit.

Posted by Gabriel | April 1, 2008 1:47 PM

I'm sure Bartels points this out in the book, but there obviously is too small of a sample to make much hay out of this graph; it wouldn't surprise me if average income growth in Australia, for instance, is historically better under conservative governments.

There's no way to use 4 data points for each party in order to tell a convincing story here. Further, the most important Federal choices (such as, Should we Build an Interstate?) won't affect income until many years after, and I don't know how we can decompose the effects each president is responsible for.

Posted by cure | April 1, 2008 1:49 PM

"The president's control over domestic policy is pretty circumscribed and public policy has only a limited influence over the economy, so it's surprising to see such a strong effect."

Why do you believe that the President's control over domestic policy is so narrowly circumscribed?

Why do you think he has such a limited influence over the economy?

You've just lived thru 7 years of George W. Bush. How circumscribed was he? Was he circumscribed in driving up health care cost inflation? Was he circumscribed in rolling back banking regulation? Was he circumscribed in creating a corrupt mess out of the War on Terror?

Posted by Bruce Wilder | April 1, 2008 1:49 PM

Agreed - this is a really poor visualization of the (very interesting) data. Somebody could do it justice by reworking it a bit.

Posted by Gabriel | April 1, 2008 1:51 PM

Is income growth before or after taxes?

Net income.

Measuring income after deductions is one thing, but "after taxes" would be a bit difficult to trend between middle and high income brackets, given the variables for each....no?

Although perhaps a parallel chart on % taxes paid on income would be interesting...if you really wanted the net, net view on the trend...

Posted by AKBY | April 1, 2008 1:57 PM

I'm sure Bartels points this out in the book, but there obviously is too small of a sample to make much hay out of this graph; it wouldn't surprise me if average income growth in Australia, for instance, is historically better under conservative governments.

There's no way to use 4 data points for each party in order to tell a convincing story here. Further, the most important Federal choices (such as, Should we Build an Interstate?) won't affect income until many years after, and I don't know how we can decompose the effects each president is responsible for.

Posted by cure | April 1, 2008 1:57 PM

This is akin to the Great Reagan Illusion. The economy under every Republican president from time out of mind has grown slower than that of every economy under a Democratic president. Except one. Reagan oversaw slightly more growth than Jimmy Carter.

Posted by Jeffrey Davis | April 1, 2008 2:04 PM

if true (i.e., assuming causation of some sort), this could single-handedly turn many people who vote republican on economics (while holding their noses on social issues) into democratic voters. all of which suggests to me this isn't true (or we'd hear a lot, lot more about this). obviously, the data are what they are. the question is what it means. anyone read (a preview copy of) the book or seen articles by the same guy?

Posted by dj superflat | April 1, 2008 2:15 PM

It would be interesting to see the Clinton years plotted separately. I suspect they'd look more like the Republic curve.

Posted by observer | April 1, 2008 2:22 PM

Uhhh ... Republican, not Republic.

Posted by observer | April 1, 2008 2:23 PM

Cactus over at angrybear.blogspot.com has done a lot of similar graphs as well as tracking a bunch of other metrics in his Presidential Performance series. Check it out.

Posted by sambaoo | April 1, 2008 2:55 PM

I'm sure Bartels points this out in the book, but there obviously is too small of a sample to make much hay out of this graph

As Drum points out, there's 50 years' worth of data, and it's robust to all sorts of adjustments (for instance, if you adjust the lag, or if you omit a single administration). Bartels claims that there'd only be a 1% possibility of getting this distribution by coincidence.

it wouldn't surprise me if average income growth in Australia, for instance, is historically better under conservative governments.

Which, if true, would show that the Australian conservative party isn't as dysfunctional and destructive as the GOP. Not really relevant to the current discussion.

Spin away, folks, but if you want to convincingly refute the case that the Democrats are better for the economy, you'll have to engage with Bartels's analysis. Start (pdf). I'm no expert -- maybe there's something he's obviously doing wrong. But it doesn't look like you can blame it on sample size or the Clinton boom.

Posted by Matt Weiner | April 1, 2008 2:56 PM

"The president's control over domestic policy is pretty circumscribed and public policy has only a limited influence over the economy, so it's surprising to see such a strong effect."

Well, of course you think like that.

Trust fund scumbags like Matthew Yglesias and Marty Peretz join General Electric and Barack Obama in fighting against universal healthcare because they don't think governmental economics matters.

And for them, of course it doesn't matter. Yglesias and Peretz are going to be rich and taken care of no matter what.

Posted by Petey | April 1, 2008 3:01 PM

cure writes:

"There's no way to use 4 data points for each party in order to tell a convincing story here."
-------------------------------------------

How is it "four data points?" The "data points" in this case appear to be year-by-year changes in income for every year between 1948 and 2005. Bartels examines those changes in five income categories - so, 57 data points for each income category.

If Rodrik asserts that the results are statistically significant, I don't see much reason to question him.

Posted by N | April 1, 2008 3:06 PM

Petey, don't embrace the dark side.

Your inner hatred may draw you to the candidate of Rush Limbaugh and Richard Mellon-Scaife.

The candidate who thinks Fox News is a good news network.

The candidate who thinks John McCain is qualified to be President.

The candidate who thinks raising the payroll cap on Social Security taxes is a ridiculous tax increase.

The candidate who voted for the Iraq War.

The candidate who trails among Dems nationally by ten points -- 52 percent to 42 percent.

Don't do it, Petey. Don't be consumed by the darkness.

It won't be easy and I know you feel alone. But come on now. Together we can change things and bring you back into the light.

Repeat after me now: yes we can.

Yes we can.

Posted by Barbar | April 1, 2008 3:18 PM

Measuring income after deductions is one thing, but "after taxes" would be a bit difficult to trend between middle and high income brackets, given the variables for each....no?

I don't think so. Doesn't the Treasury or Labor Department already gather that info?

In any case, I think the after tax information is much more relevant than the before tax information. If the question is, what government policies put more money in the pockets of people in each of the 5 quintiles, then adjusting for the amount government takes in taxes seems no different than adjusting for the amount lost to inflation.

Posted by Al | April 1, 2008 3:32 PM

Man, Nicholas Nassim Taleb would beat this author to death with his giant brain for that graph and that argument.

Posted by Wilson | April 1, 2008 3:36 PM

Wilson,

Rather than appealing to the imagined authority of Taleb and his giant brain, perhaps you could precisely state what is wrong about the argument and the graph?

Posted by N | April 1, 2008 3:55 PM

There are two factors that could account for this: minimum wage levels and the EITC expansions under Clinton. Both have big, rapid effects on income.

Posted by mq | April 1, 2008 3:57 PM

This would confirm a tentative hypothesis I've been nursing: that the best economic growth policy is trickle-up. Increasing the material well-being, or at least hope and opportunity, among the lowest wage-earners and poorest produces more beneficial effects throughout the economy than trickle-down ever could, and for logical reasons: Better breaks for lower-income folk via education, tax relief, job security and income improve their productivity, drive their consumption, and in the long run, reward economic efficiency. Gains by people on the lower end trickle up as big investors and major capital owners identify and supply their increased demand for goods. And as lower-income people gain financially they become better positioned to invest in the stock market themselves, enhancing profits and investment. There's a limit to the effectiveness of socialist distribution, of course, and we can see plenty of historical examples of overreach. But there does seem to be a "sweet spot" mix of socialism and capitalism that produces the most desirable social effects from top to bottom, and it ain't laissez-faire.

Posted by beejeez | April 1, 2008 6:13 PM

The average voter has known this for some time. I had an auto-worker friend tell me several years that whenever a Democrat is in office they get lots of overtime, while when a Republican is president it's all layoffs all the time. And a teacher friend commented right after Bush took officet hat her bank account would soon be empty, sincet hat happened whenever a Republican was at the helm.
So why do people vote GOP? Because they are freaked out by terrorists or other foreign foes, or because their religious or moral scruples force them to make abortion the only issue that matters. The only time large numbesr of people voted for a Republican for reasons having to do with the economy was back in 1980.

Posted by JonF | April 1, 2008 7:12 PM

I was a student of Bartels and was quite skeptical at first myself. The story gets more interesting when you look at the differential Dem & GOP timing of income growth across the years of a president's term and the income distribution.

The book is full of robust regression techniques, sensitivity tests, and other statistical checks and re-checks. It is quite persuasive and I buy most of the story.

Of course the story is much, much more than one chart! It is available this month I believe.

Posted by Jake | April 2, 2008 9:29 AM

Some points:

1) It's on it's face useless to generalize about Rs vs Ds unless you also include congressional composition.

2) Party is only a rough proxy for policy and trends that move the political center will move the positions of the parties accordingly.

3) It's silly to assume the causation goes any particular way without evidence to back it up. It could just as easily be that better economic times make people more likely to accept more abitious social programs or stricter environmental protections and thus more likely to elect Democrats (the case that more prosperity leads people to be willing to spend more on a social safety net and environmental protection in the long run is pretty well established) or that Republicans tend to be more popular when people are worried about slow economic growth relative to other concerns.

4) I wonder what this graph would change as the lower bound is moved further back. The in-the-middle-of-the-Truman-Administration cutoff seems like cherry picking to avoid the slowdown during Truman's first term.

Posted by MattXIV | April 2, 2008 11:59 AM

Has there been any similar research on the effect of the party affiliation of governors?

Posted by Joseph Hertzlinger | April 2, 2008 3:33 PM

I wonder if Petey's repeated personal attacks on MY and company might be considered "obscene, harassing, defamatory, or otherwise objectionable." Of course, we have Chris Ford's rather blatant anti-semitism and such, so I'm thinking not. In fact, I'm really not sure what would cross the line at this point.

Anyway, though, bear in mind that one unfortunate result of recent research on questions like "would you rather live in a world in which you make $50,000 but everyone else makes $25,000, or one in which you make $100,000 and everyone else makes $200,000? (prices are the same in both worlds)" is that many people care more about relative position than absolute wealth. As such, it's possible that the overall welfare of the rich is still better under Republicans than Democrats, if they place sufficient value on the poverty of those with lower income than they have, or are sufficiently distressed by seeing their circumstances improve. In fact, declining incomes for everyone, with a particularly precipitous drop for poorer people, could be better for many richer people than quickly but evenly rising incomes for everyone.

Although perhaps luck and the law of small numbers (the presidential administration being a crude unit of selection), or just "Republican are incompetent" is more elegant.

Still, remember that "equity" does not mean "equality." It means some particular conception of justice. One way of trading off between efficiency and equity is to try to create inefficient levels of economic equality, if you regard the equitable level of inequality as being lower than that compatible with efficient functioning of the economy. Another way of trading off between efficiency and equity is to try to to create inefficient levels of economic inequality, if you regard the equitable level of inequality as being higher than that compatible with efficient functioning of the economy.

Posted by Julian Elson | April 3, 2008 1:10 AM