American political economics in one picture

Look at the figure below, and then look at it again, and again, and again.  It is the most telling picture about the U.S. political economy I have ever seen.

 

It comes from Princeton political scientist Larry Bartels' new book, soon to be released. What it shows is the difference that the President's party affiliation makes to the distribution of income during the four years of the president's term. (The distributional outcomes are shown with one year's lag.)  When a Republican president is in power, people at the top of the income distribution experience much larger real income gains than those at the bottom--a difference of 1.5 percent per year going from the bottom to the top quintile in the income distribution. The situation is reversed when a Democrat is in power: those who benefit the most are the lower income groups. If you are in the bottom quintile, the difference between having a Democratic or a Republican president in office is an income gain (or loss) of more than 2 percent per year! Strikingly, compared to Republicans, Democratic presidents generate higher income gains for all income groups (although the difference is statistically significant only for lower income groups).

Bartels shows in his book that this difference is not a statistical artifact or a fluke.  It is not the result of Democrats coming to power during better economic times, or of Republicans reining in the unsustainable excesses of Democratic administrations they replace. (It turns out that the same pattern prevails even when a Republican president is succeeded by another Republican.) These numbers are real and they are the outcome of partisan differences in policy. So if you are one of those who have bought the story that income distribution is the result of pure market forces and technological changes, with politics playing no role--think again.

Bartels' findings raise an important puzzle: if Democrats produce better income results for everyone, and particularly for the more numerous lower-income groups, why do they not always win?  Bartels offers a rather complicated, but well-supported answer to this question having to do with voter myopia and psychology.  (You will have to wait to read his book to get the full story). But Bartels does demolish two of the standard arguments regarding Republican advantages at the polls: the idea that poor Americans vote Republican for cultural reasons, or that Americans do not care about inequality.

Comments

[luci]

I've seen this claim many times before, and I still think there won't be enough data points to generate a significant result. How many presidents have we had in the modern era? Sure, we could use monthly data or something to pad out the # of observations, but how many times has the presidency changed party?

Business cycles are relatively long-lived creatures too. How many tech-booms did we experience?

I know Mr. Rodrik said, "Bartels shows in his book that this difference is not a statistical artifact [...] and it is not the result of Democrats coming to power during better economic times", but I just don't see how.

Not enough time, not enough presidents, not enough business cycles. Not to mention the President/Congress assumption, or the slim marginal difference between Democratic and Republican economic policies.

But maybe it's explained fully in the book. I actually believe such an effect could exist, but it would be small, the causal mechanism subtle, and take a lot of data to tease it out.

[Andrew Leigh]

I've done some analysis using top incomes data, and just posted the key table on my blog - http://andrewleigh.com/?p=1857

[Steve Dutch]

So who's got the worse myopia? Conservatives who vote cultural issues despite their economic interests? Or liberals who champion cultural issues even though they keep the Religious Right energized? If the cultural issues are all that trivial, why not let the Right have their prayers in school and overturn Roe vs. Wade? And if they're important enough to fight for, then it's rational for people to value them over their economic welfare. If economic corruption is such a deep problem, shouldn't we be praising people who value principle over their pocketbooks?
If you want to pull the fangs of the Religious Right, give in on the cultural issues. Abortion and school prayer are the Iraq and Vietnam of the left: an interminable quagmire against an enemy that will not give in or admit defeat.

[John H]

So what I would like to know is who thinks the Internet would have blossomed like it did in the late 90's (with the US taking point) if GWB was leading the way instead of Clinton/Gore?

Maybe not - we can surely point to the explosion in alternative energy, green technologies, and newly developed sources of independent energy which have occurred over the past eight years . . . . .

I like to distill it down to a simple point - Most republicans tend towards the status quo. They like how the rules work and abide by those rules to move their way up the ladder. Most democrats I know are far more open to change and more likely to question the rules - find a way around them.

[Kyle McCullough]

Chris and jsalvati have a reasonable question, asking just how much power does the President have, to have such a large influence--especially given Congress's control of the purse? The simple answer is MUCH. The President's veto and bully pulpit gives his party effective control, even over those matters where congress has nominal control. Probably even more important--at least for the short term--is his control of the Bureaucracy. A good example here is the recent bailout of Bear Sterns. No special legislation was passed--just the President's men deciding to spend a few billion of our dollars.

Then there's his ability to wage war, if not declare it.

[Sebastian]

First about the "culture" question: Bartels has written a much cited paper, ingeniously cited "what's the matter with what's the matter with Kansas" on this topic, a relatively easy read:
http://www.princeton.edu/%7ebartels/kansas.pdf
(note Tommy Frank's response though, "Class is dismissed" -
http://www.tcfrank.com/dismissd.pdf
Personally I think Bartels comes out the winner, but I don't think clear enough for "demolished".
As for the graph above, the fact that everyone does better under Democrats makes me suspicious(although I'm ideologically very favorable to the argument) - it seems much more plausible that Democrats just came to power under different economic circumstances.

As for what data - it's Census Bureau Data on family income and I'd assume it's post tax, but Bartels doesn't tell us.

I don't think this exists for others countries (although it would be much more convincing for European parliamentary democracy, where we don't have divided gov't) - Kenworthy and Pontusson make some interesting points suggesting that higher voter mobilization leads to more redistribution:
http://www.u.arizona.edu/~lkenwor/pop2005.pdf

[ClayMcReynolds]

Unfortunately we aren't told in this post whether or not this is real income or income adjusted for inflation. If not adjusted for inflation it is meaningless so I would rather hope that it is. I guess I'll have to find a copy of the book to find out.

Likewise on the question of taxes. The numbers must also be after taxes or they don't really mean much.

OTOH, every study of this nature that I've seen yields the same results. Dem presidents better for income. Dem congress better for income. No studies that say otherwise making it amusing the way the naysayers find ways to ignore these results with no data to back up their contention that it isn't true that dems are better for you income than republicans.

BTW, the most interesting result I've seen is the study (I don't have time to look for it now) that purported to show that while dem congress was better than gop congress and dem president was better than gop president, the best results were obtained when the white house was held by one party while congress was held by the other. (I don't recall which was best of gop WH/dem congress or dem WH and gop congress.)

Either way, I'm with DrewDuncan. You can tell from my social security statements who controlled the WH. (That would make me guess that dem pres and gop congress is the most profitable but I am but a single data point...)

[Ed Al-Hussainy]

Can't argue with the pretty figure, but the jury is very much out on whether this correlation is actually causal; I lean towards research that views economic performance as driving elections, rather than the other way around.

Here's a nice summary of what other economists think about this:
http://online.wsj.com/public/article/SB116240116581410311-JyXJXeDVPrYETGgoEliOYcxxnD8_20061109.html

[ClayMcReynolds]

"I lean towards research that views economic performance as driving elections, rather than the other way around."

As in the gop screws up the economy causing democrats to get elected to fix it after which people can start worrying about social issues again and they elect republicans who screw up the economy again and the cycle begins again?

[McLovin]

Forced to wager, I'd guess that the answer to the why incomes fare better under Democrats is that the sample is far too small to have statistical significance. It's coincidental, in other words.

But for the purpose of discussion, here are my favorite explanations for why a Democrat president is better for your bank account:

1. Oil prices. Republican incompetence in foreign affairs, and their political dependency on military adventurism inevitably creates global tension that pushes prices of oil higher. Same for gold, steel, wheat and other commodities. Such increases act as a tax on consumer spending and retard economic growth.

2. Democratic presidents use their media power to focus the nation's attention on solving domestic problems and on being optimistic about the positive role government has on the economy. Republican presidents, again with the warmongering, use their media power to inflame paranoia about communists, or Muslims. Republicans foster optimism about America's military prowess, rather than its ability to solve important domestic economic and social problems.

3. Fiscal responsiblity. Democrats end up being more responsible guardians of the treasury. The policies they advocate, broadly, tend to call for more spending than Republicans do. But because of their political position, Democrats tend to fail to push those spending programs through. Republican presidents, on the other hand, advocate parsimony, but spend far more in the end, because they are better at creating political distractions, like hating Muslims or fearing communists--and, in they end, they need to please their constituents with pork just as much as Democrats do.

[Chris]

The economic answer is easy, income and wages are not the be all and end all of economic indicators.

Under Reagan, for example, even if wages stagnated in the lower income brackets, the fed tamed inflation which brought gas prices back down and slowed the erosions of purchasing power that had occurred in the prior administrations, all the way back to Nixon.

Again, ignoring all non-economic factors like personality, party organization, etc., another reason that Republicans do well is that their policies tend to expand wealth, not wages. Anyone who owns a house, or a business, typically sees gains under a Republican administration. When housing values go up, people have more leverage, etc. Yes, this has caused the recent bubble, but it makes people very happy.

Finally, you have to understand the relationship if income distribution to voting. Younger people tend to make less, representing a large portion of low income voters, but they also tend to vote less. In fact, young people are the least likely to vote. So the people who would benefit most from a wage increase, people under 30, are the least likely to go the polls.

[Marty]

Correlation is not causation. Assuming that the data is real and actually says what it is claimed to say (a very big IF with this kind of thing), what causes this?

First, I would want to understand the data, in all its limitations. Then, if the relationship is still apparent, what caused it.

Otherwise I'm just associating hemlines with sunspots. When I was growing up I was told that Republicans ALWAYS bring depressions and Democrats ALWAYS get us into wars... same level of reasoning.

[Use your heads]

This is only slightly less absurd than buying into the super bowl stock market correlation.
http://www.forbes.com/2004/01/28/cz_jd_0128inlwatch.html

Like has been said before - how are Ike and GWB similar? how are clinton and JFK similar? congress and the Fed have more power over economic growth than presidents do. correlation does not equal causation. historical currents have as much to do with economic growth as government policies. etc etc

this party affiliation of the president is a parlor trick. nothing more. (that is not to say that once you actually do the proper comparison that the results won't be the same)

[VentrueCapital]

I haven't read the book yet, or even the original symposium from two years ago, so I have no idea if these questions are answered there:

1. Is "real income" (on the vertical axis) defined as pre-tax or post-tax, and does it include transfer payments such as welfare, farm subsidies, etc.?

2. What other factors have been tested to see if they have an even stronger correlation with income growth and/or distribution?

3. Is "income percentile" (on the horizontal axis) defined as income per household, or per person, or what? And does it include subsidies? (I assume, of course, that it is pre-tax income.)

[Sarge6]

1) Whoa, let's wait for the book so the methodology can be fully explored.

2) This is essentially W-2 gross income, right? I'd be far more interested in deposit income reflecting "Republican versus Democratic" tax rates.

3) Right there that implies you have to look at the question of divided government, i.e., a Congress in opposition to the Administration.

[McLovin]

Forced to wager, I'd guess that the answer to the why incomes fare better under Democrats is that the sample is far too small to have statistical significance. It's coincidental, in other words.

But for the purpose of discussion, here are my favorite explanations for why a Democrat president is better for your bank account:

1. Oil prices. Republican incompetence in foreign affairs, and their political dependency on military adventurism inevitably creates global tension that pushes prices of oil higher. Same for gold, steel, wheat and other commodities. Such increases act as a tax on consumer spending and retard economic growth.

2. Democratic presidents use their media power to focus the nation's attention on solving domestic problems and on being optimistic about the positive role government has on the economy. Republican presidents, again with the warmongering, use their media power to inflame paranoia about communists, or Muslims. Republicans foster optimism about America's military prowess, rather than its ability to solve important domestic economic and social problems.

3. Fiscal responsiblity. Democrats end up being more responsible guardians of the treasury. The policies they advocate, broadly, tend to call for more spending than Republicans do. But because of their political position, Democrats tend to fail to push those spending programs through. Republican presidents, on the other hand, advocate parsimony, but spend far more in the end, because they are better at creating political distractions, like hating Muslims or fearing communists--and, in they end, they need to please their constituents with pork just as much as Democrats do.

[Matt]

I think if you disagregate the periods 1948-1980 and 1980-2005 its pretty easy to explain the results of that data. Whatever dataset is being used to show growth in income inequality almost certainly shows much greater growth in income inequality from 1980-2005 than 1948-1980 because emerging economies weren't providing nearly as much competition for low income labor between 1948-1980 vs. 1980-2005. Then you've got the fact that Democrats controlled the White House 50% of the time during the 32 years from 1948-1980 and only 32% of the time in the 25 years from 1980-2005 meaning that they're a far smaller part of the 1980-2005 sample when income inequality was increasing.

That makes this an argument that the income inequality growth from 1980 is because we've had more Republican Presidents since then, but income inequality almost certainly continued apace during President Clinton's period in office. However, without seeing the dataset its impossible to be sure.

[ConservativeScientist]

If you plot the same data, but include Bartels' published standard deviations, taken from his website (http://www.princeton.edu/~bartels/income.pdf), you get an outright laughable graph. Here is a quick plot of it.
http://www.conservativescientist.com/bartels.jpg

From data that looks like that, you can't draw any conclusions. But if you leave off the error bars you can easily deceive a lot of people.

[Eric H]

I also agree with both Chris and Barkley.

In fact, if Robert Feinman actually read the study he keeps posting without his ideological blinders on, he might have noticed that Altemeyer caveats mixing "conservative" and "authoritarian", that libertarians score very low on the authoritarian scale (Robert has a bad habit of conflating them with conservatives), but most germane to this point is that there is a high correspondence between Altemeyer's authoritarians and religious belief, i.e. culture. Those authoritarians are just as likely to vote for Obama or Hillary if their promises to put a halt to cheap imports are credible enough.

[FT Woods]

Perhaps this is too over simplified an explanation, but the President presents an Annual Budget based on the administration's priorities. The budget in turn effects incomes across the nation.

So it would seem that Democratic presidents prefer policies that benefit everyone while Republican presidents prefer policies that benefit the few.

[addicted]

Even if this is pre-tax income, it is a telling graph. (Obviously, I am not convinced of the implications of the graph before I read the book and the methodology, but I am assuming it is what it is presented as in this blog). If taxes during Democratic Presidencies are lower, or equal to that during republican presidencies, then people earned more money in total. If that is not the case, then it puts paid the argument that higher taxes are the sole reason for lowered productivity. As is quite visible in this graph (assuming that D Presidents have presided over higher tax periods), that despite these higher taxes, people are earning more, and so the economy is doing better.

[Sean Kelly]

Democrats get into power. Increase gov expenditure, aggregate demand increases. Poor people who were previously unemployed now finds work, they’re made better off. Rich people already had work, so little direct effect on them.

Republicans get into power. Decrease gov expenditure to tackle inflation. Aggregate demand decreases, poor people who were previously employed now finds themselves unemployed, made worse off. Rich people have safe jobs, so little direct effect on them.

[Theo O'Brien]

You're probably aware of this, but I thought you may like to know that your posting of Bartels' chart has attracted lot of attention. Notably from Paul Krugman:http://krugman.blogs.nytimes.com/2008/04/02/bartels-alfred-wegener/

He seems to doubt the credibility of the chart but I think it was just an inevitable finding that supports what a lot of people already know.

[Nathan Smith]

Given the scarcity of data points-- there have been, what, fifteen or so presidents in the 20th century-- I'd be amazed if this chart stands up to serious econometric scrutiny. This is not to deny Bartels' good faith. But.

[skeptonomist]

Here are graphs of income growth over time for each quintile:

http://www.skeptometrics.org/IncomeGrowth.htm

See also the graph of Gini coefficient (measure of inequality) by the Census:

http://www.census.gov/prod/2000pubs/p60-204.pdf

Of course it will be interesting to see Bartel's discussion, but these graphs tend to show that the difference is mainly a temporal trend, not so much due to presidents. The real question may be whether the temporal trend is due to conservative philosophy and policies.