Inequality and American Democracy

by Henry on March 16, 2006

The current issue of _PS: Political Science and Politics_ has a “symposium”: on inequality and American democracy, collecting together various responses to the “report”: and book issued on the topic by the American Political Science Association’s taskforce. There’s a lot of valuable commentary and empirical data in there; also well worth reading are the accompanying critical papers on “inequality and American governance”: and “inequality and public policy”: A lot of meat in there, including the below graph drawn from Larry Bartels’ “paper”: . It shows income growth by income level under Democratic and Republican administrations from 1948 to 2001. The solid line shows how families at the 20th percentile (lowest), the 40th percentile etc have done under Democratic adminstrations, the dotted line how they’ve done under Republicans. The difference is startlingly obvious. Under Democratic administrations, growth has been fairly egalitarian, ranging from 2.6% average growth for the poor at the 20th percentile to 2.1% for the rich at the 95th percentile. Under Republican administrations, the rich have done about as well as under Democratic administrations, but the poor at the 20th percentile have only seen .6% income growth. As Bartels says:

bq. Are partisan differences in the economic fortunes of American families really this stark? The arithmetic calculations from the Census Bureau data are straightforward. Their political significance can only be gainsaid by supposing that the apparent pattern is the result of a massive historical coincidence. Elsewhere, I have provided extensive checks on the robustness of the partisan disparity evident in Figure 2, including comparisons based on alternative economic units, time periods, and income definitions, statistical controls for historical trends, nonparametric tests, and the like ( “Bartels 2004”: ). It seems hard to escape the conclusion that, over the past half-century, Republican presidents have been consistently bad for the economic health of middle-class and poor people.




Kieran Healy 03.16.06 at 4:30 pm

Truman said it first: “If you want to live like a Republican, vote Democrat.”


otto 03.16.06 at 4:48 pm

I’m surprised that this seems to be true even when Eisenhower’s presidency is included. Must take a look at the paper.


BigMacAttack 03.16.06 at 5:24 pm

Interpreting from about 6 business cycles and a heterogeneous set of Democratic and Republican economic policies will probably only convince either the ignorant, gullible, or the already convinced.

There are plenty of reasons not to vote Republican, that wasn’t one of them.


M. Gordon 03.16.06 at 6:17 pm

I’m curious how much of this trend is due to Clinton’s presidency alone, and how it would look without it in the data.

bigmacattack: Are you suggesting that there isn’t enough data here? That seems odd. Speaking not as an economist (which I’m not) but as a scientist, if there wasn’t enough data, you would expect to see little or no trend. The fact that a very obvious trend exists suggests that there is a correlation. You could run a t-test and show the likelihood that this was accidental, and it would be very low. Thta being said, it’s not clear to me that this is not due to one or two large outliers, which might make you question the _reasons_ for the trend. But the trend is very clear and impossible to deny, unless you’re squinting hard to avoid seeing it.


wcw 03.16.06 at 6:37 pm

I think BMA is confusing this paper for a canonical partisan trope which measures aggregate economic performance vs a dummy variable for presidential party.

If this paper had anything to with that, his standard response would be correct. As this paper is about the performance of the top and bottom of the income scales, not of the aggregate, he’s just confused. This is a very different matter than aggregate economic performance and much more amenable to executive interference via tax and distribution policies.

That said, I have not read the paper or looked at this particular data set, but I do have a back-of-the-envelope acquaintanceship with income strata performance over time, and this conclusion rings true for me.

Even if it’s driven heavily by the Clinton years (benefiting the bottom) and Reagan, Bush I and especially (tax cuts!) Bush II years (benefiting the top), that doesn’t exactly make the data go away, does it?


Sebastian Holsclaw 03.16.06 at 7:20 pm

What policies are supposed to account for the disparity? What effect does immigration policy/reality have on this?


Sebastian Holsclaw 03.16.06 at 7:27 pm

“also well worth reading are the accompanying critical papers on inequality and American governance and inequality and public policy.”

Is that two papers? You have two links to one paper–a very interesting paper but only one. :)


mpowell 03.16.06 at 7:54 pm

That’s a pretty impressive discrepancy. There’s been a lot of discussion around here about inequality recently and although I’m not ready to get on board with the whole program that I think people might be hinting at, I’d certainly admit that assuming this result isn’t extremely sensitive to the removal of one or two presidencies, this is a strong argument to vote democrat. And I am neither ignorant, gullible or already convinced. (Well, to be fair, I was already convinced for the next election cycle. But that wouldn’t always be true.)


Henry 03.16.06 at 8:21 pm

Sebastian – link fixed. Bartels includes some along-the-way speculations about the intervening policies – but obviously a statistical exercise of this sort is unsuited to determining which policies had these effects.


Z 03.16.06 at 8:25 pm

Even if it’s driven heavily by the Clinton years (benefiting the bottom) and Reagan, Bush I and especially (tax cuts!) Bush II years (benefiting the top)
Actually Bartels checks this, maybe not in the paper linked but in another one called Partisan politics and the U.S income distribution (readily accesible from his page). Excluding any one or two administrations changes nothing substential. Otto was right to express surprise about the inclusion of the Eisenhower administration: removing it makes the trend much more impressive. Removing Clinton’s does not weaken the trend, it reinforces it.

What policies are supposed to account for the disparity?
Bartels show that the general superiority of Democratic administration seems to be strongly correlated to their superior macroeconomic output. Low-income households are disproportionately sensitive to macroeconomic indicators, so it is not surprising that they do so much better under Democratic administration. As he writes “[this] seems to suggest that post-tax income growth is less subject than pre-tax income growth to partisan effects, which is certainly counterintuitive.” He also remarks that datas are too scarce yet to assess this correlation yet. Maybe looking at international comparisons would be a good idea. Why do Democratic administration produce better macroeconomic outputs? I don’t know, but I’m sure it is not too hard to find reasonnable ideas.


yabonn 03.16.06 at 8:27 pm

Speaking not as an economist (which I’m not) but as a scientist, if there wasn’t enough data, you would expect to see little or no trend.

I don’t agree with BigMacAttack, but this point surprises me.

If there is not enough data, you can fit (unconclusively) many elaborate explanations to it, no? Or, seen from the other side, you’ll always have a (meaningless) trend?


'As you know' Bob 03.16.06 at 8:49 pm

The vast majority of Americans are glaringly better off during Democratic administrations.

And this is news, how, exactly?


dm 03.16.06 at 9:27 pm

Might this effect be the result of raising the minimum wage? The resulting increase in demand would trickle up through the economy (raising all boats, as it were), but would, of course, have the biggest impact at the minimum-wage-earning end of the economy.

Minimum wage increases seem to mostly happen only under Democratic presidents.


Gareth Wilson 03.16.06 at 11:21 pm

Given that in peacetime the president is just a civil servant with a flashy office, I’d be much more interested in a graph showing how party control of Congress relates to the statistics.


ab 03.17.06 at 1:22 am

Without being an expert on the subject matter, but…

(1) What about the role of Congress?

(2) What about local and state policies?

In a highly federal and complex political system such as in the US, I would be always sceptical of a simple graph like this one above.

Also, don’t you need to account for different worldwide economic cycles? The 1950s and 1960s saw high growth, while the 1980s only little.

It also might be that the causal arrow points the other way. In high-growth times Democrats get elected to distribute more fairly the benefits of growth, while in low growth times Republicans get elected to help growth?

I haven’t read the paper, though; perhaps it answers some of these points.


Pablo Stafforini 03.17.06 at 1:48 am

Many liberals find in these results a vindication of their political preferences. As a man of the left, I too feel vindicated, but for quite different reasons. The study, in my interpretation, doesn’t so much show the political superiority of the Democrats over the Republicans, but the moral bankruptcy of both. For an inference to the best explanation accounts for the historical pattern by assuming

a. a principled bipartisan coincidence regarding the aims of economic policy, both Democrats and Republicans subscribing to a sort of inverse Difference Principle: “economic inequalities are to be arranged so that they are both to the greatest benefit of the most advantaged”; and

b. a tactical discrepancy between the two parties regarding the means to achieve the aims outlined in (a), Democrats believing that only some degree of inequality promotes the interests of the most advantaged, Republicans believing that no degree of equality would.

The study, in other words, supports the view that the history of recent American politics is nothing but the story of an “overlapping consensus” between two “comprehensive doctrines,” the key features of which were admirably summarized by one of its key protagonists and an eminent admirer of John Rawls—namely, Bill Clinton: “The truth is,” he said, “that there is an ideological struggle between those who believe that the best way to grow the economy is to give more money to the wealthy [sc., Republicans], and the Democrats who believe that the wealthy will make more money if average people do better.”


Z 03.17.06 at 1:57 am

To ab (#15),
I believe it answers all your questions, especially if you supplement it by reading Partisan politics and the U.S income. What he shows (in details) there is that there is a strong correlation between presidency and economic well-being. This correlation cannot be explained solely by economic cycles (unless you can explain why favorable economic cycles are strongly correlated to Democratic administration). As for the causal arrow pointing the other way, this is shown to be extremely improbable because economic well-being changes most radically in the second year of a democratic presidency (i.e the first year where new policies have effects).

However, Bartels points to a very peculiar fact: Republican administrations are very good at achieving a better economic situation in election years while Democratic administrations are bad at it. To quote “As a result, families at every income level have experienced much more election-year income growth under Republican presidents than under Democratic presidents.” As voters apparently tend to vote in function of the yearly trends, this explains while economical success has not necessarily meant electoral success for Democrats.

Of course the other papers are worth a read too. Of particular interest to me is the notion (developed in Inequalities and public policies) that income inequalities may foster an environment where elites can access power by looking “upward” rather than by following the more traditionally democratic way of trying to speak for large numbers of people. Since democracy matters (to me), this is reason enough to worry about extreme inequalities.


stuart 03.17.06 at 4:46 am

I think the thing that would be more telling is if the graph broke down what happens within the top percentile, as while Republicans are shown as equivalent for the top 20th percentile, the trend would tend to give the impression that the top 5 or 10% would likely be better off under their administration than under Democrats (of course, to do this it would mean that the households in the 80-90th were likely to be worse off, but I think this follows equally).

Of course this raises the question of how you can consistently get 50% or so of people to vote for a party that is economically beneficial for only about 5% of the population. Either 45% of the people are expected to vote for them because they aim to be in the top 5% within a couple of years, or otherwise at least gain enough ground at the expense of those around them to make up for the general lack of growth in their income/wealth band. Or maybe they are just under the false impression they are likely to be wealthier, or vote for non financial reasons.


abb1 03.17.06 at 5:29 am

Or maybe they are just under the false impression they are likely to be wealthier, or vote for non financial reasons.

Imagine that you live in a small town. You own a diner or you’re a car salesman or a bank clerk. You make $40K, you own a house that’s worth $70K, your mortgage is $400/month. Life is good.

Then you look into your paycheck and you see $500 taken out every month; if you’re single it could be as much as $800. You don’t feel that you use any government services, especially federal services. Defense – sure, but what else? You have never seen a federal employee in your life, no federal law or regulation ever gave you anything other than a headache.

If you are this guy, you’re likely to vote republican – and for reasons that are mostly financial.


Brett Bellmore 03.17.06 at 7:11 am

I wonder if the data could be deconvoluted to get an estimate of the step response of the economy to the election of one party or the other?

If Democrats and Republicans alternate, and Democrats follow “quick fix” policies, and Republicans aim at long term growth at the price of short term sacrifice, it would tend to produce a pattern of economies rising towards the end of Republican adminstrations, and peaking in the middle of Democratic ones. But it wouldn’t mean that Democrats were actually good for the economy in any long term sense, just that they were coming along in time to harvest what Republicans had planted.

Note, I am NOT asserting this is true, just saying the data appear to be consistent with it, and that a more sophisticated analysis would be necessary to prove/disprove the hypothesis. However, I will say that the economy strikes me as more an oil tanker than a powerboat, and I don’t see it responding rapidly to policies that are good for the long term.

“Of course this raises the question of how you can consistently get 50% or so of people to vote for a party that is economically beneficial for only about 5% of the population.”

By having a large segment of the population that thinks there are other things that are important besides economics? Like whether babies are killed in the womb, or their guns are taken away?


Brett Bellmore 03.17.06 at 7:15 am

In other words, if the economy tends to be upward bound at the beginning of Democratic administrations, and downward bound at the end of them, that really does NOT suggest that Democrats are good for the economy. LOL


mc 03.17.06 at 7:21 am

This looks v interesting – must take a look. But in the meantime did any british readers see the guardian story on tuesday reporting a study by the IFS on inequality in Britain? The story was accompanied by a graph (unfortunately not available on the guardian website as far as I can see – you may be able to find it on the IFS or IPPR websites, this graph has been kicking around for a while) comparing income growth by percentile (1) 1979-1990 and (2) 1997-2003. obviously given it is a single comparison, it is much more vulnerable to objections about not comparing like with like, not comparing across cycle etc., but still the graph is incredibly striking (and rather undermines the headline to the story – ‘inequality unchanged since thatcher’ – the graph indicates pretty clearly that the reason aggregate measures of inequality are unchanged since 1997 is because of what’s happening to the bottom 5% and especially 1%, and the top 1%. Clean those out and the story is very different. Which is not to say that what happens to the bottom and top 1% is irrelevant, it certainly isn’t – just that it shouldn’t be allowed to obscure completely what’s happening in between.)


Brett Bellmore 03.17.06 at 7:39 am

In the interest of honesty, though, it’s pretty hard to paint the Republican party as the party of fiscal rectitude with Bush in office.


Steve LaBonne 03.17.06 at 8:19 am

#11 might be one of the best capsule explanations I’ve ever seen of the different worldviews of natural and social scientists. ;)


Barry 03.17.06 at 8:44 am

Brett, or with Reagan.

Of the five post-WWII Republicans presidents (Eisenhower, Nixon, Reagan, Bush I, Bush II) how many have been fiscally responsible?

Please note that I’m not talking about shrinking government, because (to the best of my knowledge) none of them has done so.


Brett Bellmore 03.17.06 at 9:01 am

With respect to #11, speaking as somebody who didn’t flunk statistics 101, if there’s not enough data, I’d have no expectations whatsoever about whether trends would appear, because any that did appear might be meaningless noise.


Kevin Donoghue 03.17.06 at 9:14 am

The way I read Yabonn’s comment (#11) is that with a small dataset you can always expect to find spurious relationships – a sound point. But I really don’t think that’s what’s involved here. The comments about fiscal profligacy ring true. If Republicans prefer lax fiscal policy, offset by tight monetary policy, ye olde IS/LM model suggests that Republican administrations will enrich the rich.


Hektor Bim 03.17.06 at 9:18 am

Brett Bellmore,

Your argument is strange. The dataset covers more than 50 years. Can you say with a straight face that we should inflict pain for 50 years in the pursuit of some mythical long-term future, like say in 300 years, when we are all very much dead. I don’t think many Republicans would say that their economic plans have a longer than 50 year window for developing, because if they did, no one would vote for them.


Brett Bellmore 03.17.06 at 9:26 am

Hektor, your argument is strange: The dataset covers more than 50 years, during which the parties repeatedly swapped places. If Republican policies bore fruit as little as 4 years after they’re instituted, it would still, on average, be Democrats plucking it.

Having the economy peak during your watch is nothing to brag about, you know; It only peaks when it stops improving…


Henry 03.17.06 at 9:50 am

Brett – before jumping in with your theories, you should read the papers – especially the 2004 paper (which I link to in the middle of the Bartels quote). Bartels tests for simple alternation type explanations and finds that they don’t explain what’s going on. Indeed, he performs quite a number of statistical tests, checks for robustness etc. To put it mildly, Bartels didn’t flunk Statistics 101 either (he’s one of the most highly rated statistical methodologists in political science). There’s room for argument about the intermediating variables here. But claims that this is sloppy statistics are wishful thinking – especially when they don’t show any evidence that you’ve read the paper where he answers your questions.


Kevin Donoghue 03.17.06 at 10:05 am

If Republican policies bore fruit as little as 4 years after they’re instituted, it would still, on average, be Democrats plucking it.

It might be wise to revise Recent US History 101 and do a quick count on who held power when.


Steve LaBonne 03.17.06 at 10:09 am

“Revising” history- in the US rather than British sense of “revise”- is exactly what conservatives do for a living…


BigMacAttack 03.17.06 at 10:10 am

M. Gordon,

Yes, I am suggesting that there isn’t enough data.
Once you make the assumption that each business cycle played some independent role in determining well being and you realize that there where only 6 business cycles, you realize that some part of this analysis is subject to what I think what they call a low level of confidence.

If you think the business cycle is irrelavent and/or controlled year to year by the president, well then you have plenty of data.


Nope, my assumption was correct, see Z’s comments.


I doubt 8 or so darts thrown against say a 4 region board divided into real good, good, bad, and real bad would yield a correlation with any great level of confidence.

Congress, lag, and the precise nature of the marcoeconomic policies that are supposed to result in growth are also important issues.

On the surface it seems to me that Clinton’s macroeconomic policies differed from LBJ’s and Reagan’s differed from Nixon’s.

Look, if you think nationalizing health care is a good idea, vote Democrat, but don’t put too much stock Bartel’s graph.


BigMacAttack 03.17.06 at 10:38 am


I am sure the man is a stat genuis.

But he assumes that the president controls the business cycle and yet he cannot state what policies result in more growth (or at least he doesn’t bother to in the link.) and yet he knows it is related to Democratic policies, whatever those are.

He is searching for a pattern he knows is there and by making a few key assumptions he has found it. Suprise.


yabonn 03.17.06 at 10:43 am

with a small dataset you can always expect to find spurious relationships – a sound point. But I really don’t think that’s what’s involved here.

That’s what i meant, only better put. Thank you kevin.

I don’t think neither it’s what’s involved here. Actually, i know realize gordon meant “trend ” as “statistically meaningful trend”, so my bad.

if there’s not enough data, I’d have no expectations whatsoever about whether trends would appear

“Dear students of statistics 101, never forget that if there’s not enough data, you should have no -repeat no– expectation about wether trends would appear”.

Yeah, i can picture that.


Barry 03.17.06 at 10:45 am

“Given that in peacetime the president is just a civil servant with a flashy office, I’d be much more interested in a graph showing how party control of Congress relates to the statistics.”

Posted by Gareth Wilson

Gareth, perhaps should should both read a (USA) Civics 101 book, and then read newspapers for a few years. Some recent history books are called for, as well.


abb1 03.17.06 at 10:47 am

I think BigMac has a point. The graphs seem to be showing correlation, but you need more to demonstrate causation.


Kevin Donoghue 03.17.06 at 10:50 am

Z (#10): Why do Democratic administration produce better macroeconomic outputs?

Only “better” if what you want is more economic growth and less inequality. Those who want to spend more on weapons systems and less on welfare, while cutting taxes and keeping inflation in check with the help of tight monetary policy, are likely to have different priorities.

BMA(#33): On the surface it seems to me that Clinton’s macroeconomic policies differed from LBJ’s and Reagan’s differed from Nixon’s.

Z has already remarked that removing Eisenhower makes the trend much more impressive and removing Clinton reinforces it.

Look folks, this isn’t some flighty theory sketched out on a napkin over a liquid lunch. Policies matter. If you want more data look at other periods and other countries. There is a reason why farmers flocked to William Jennings Bryan and stockbrokers drank the health of Margaret Thatcher.


MQ 03.17.06 at 11:38 am

I’m with comment #13, this could really be a straight up effect of minimum wage increases under Democratic administrations. Your 20th percentile of income person is probably earning around the minimum wage. Did Bartels control for the minimum wage level? Until you check that all other more elaborate explanations are premature.


Henry 03.17.06 at 12:10 pm

bigmacattack – he isn’t making the assumptions that you say he is – insofar as he can (and this is always the problem with stats – you do what you can but it may not be enough), he’s controlling for them. I do suggest that you read the paper if you haven’t (the 2004 one linked from the Bartels quote) and see what he does, and how he justifies it. He runs this under a wide variety of different assumptions and finds that the underlying pattern is pretty robust. He also reports t-statistics which give us some idea of whether there is a real relationship there or not. We can certainly argue about the intermediating variables between the occupier of the presidency and the results for different income segments – i.e. what is driving this relationship. But unless he’s being quite dishonest (and he has a valuable academic reputation to defend), the underlying statistical relationship exists.


Kevin Donoghue 03.17.06 at 12:23 pm

mq (#39): I’m with comment #13, this could really be a straight up effect of minimum wage increases under Democratic administrations.

Nice one, mq. But doesn’t the case against a higher minimum wage turn on the claim that it discourages the employment of low-skilled workers? So now we need to square this nice theory with the awkward fact that unemployment has been lower under Democratic presidents. Maybe the adverse effects of raising the minimum wage kick in after an eight-year lag, so that Republicans are made to look bad?


matt butler 03.17.06 at 12:29 pm

The criticisms about the small number of data points in this study, and the resulting lack of statistical confidence, are correct. Unfortunately, this is a limitation of politics generally. In the huge, slow, and uncontrolled domain of a political system, there are never enough data points available to support generalizations with anything like the level of robustness we demand from controlled experiments. Even if there were, parties and conditions can change rapidly enough that past performance would not guarantee future results.

And yet (in democracies) we are called upon to make political judgements. In so doing, it seems prudent to make use of the best statistical analysis available, and to take seriously the correlations it reveals. We can’t conclusively rule out the null hypothesis that they are spurious (or that are real but non-causal, or that they are causal but non-predictive), but in the absence of a better analysis, they are all we have to go on.


Brett Bellmore 03.17.06 at 12:57 pm

” especially the 2004 paper (which I link to in the middle of the Bartels quote).”

You mean the one where all the documentation of those checks he made is to be found in a couple of unpublished papers he mentions in a footnote?

I’ll do more reading this evening, when I have the time.


Barbar 03.17.06 at 1:36 pm

It’s a fascinating pattern.

On one hand, Republicans think inequality is not important.
On the other, Republicans think that their politicians are very unlikely to increase inequality.

On one hand, Republicans want to cut the size of the federal government.
On the other, Republicans think that tax cuts increase total tax revenues collected by the government.



abb1 03.17.06 at 1:44 pm

As dm noted above minimum wage could lead to increase in economic activity (and have ripple effect on increasing other wages near the bottom, for that matter).

What’s more difficult to explain imo is seemingly non-existing result of massive keynesian intervention during Reagan, Bush1 and Bush2.


spencer 03.17.06 at 1:50 pm

Not that it makes much difference, but since WW II there have been 10 business cycles, not 6.

But the conclusion of this paper are supported by a wealth of other data that show the economy does tend to do much better under democratic administrations.

It is even true about the stock market as average annual increase of the S&P 500 has been some 12% under democrats vs about 8% under republicans — and this comparion does not include Bush II that would bring the Rebublican average down..

The suprise to me is that for the top quintile it did not make any difference. I wonder what that says about political influence– if anything?


Barry 03.17.06 at 3:57 pm

“What’s more difficult to explain imo is seemingly non-existing result of massive keynesian intervention during Reagan, Bush1 and Bush2.”

Posted by abb1

Not all massive deficit spending is equal. Not all tax cuts are equal.


abb1 03.17.06 at 4:18 pm

Look, you borrow a few trillion dollars and order some big airplanes to build – you are bound to generate some serious economic activity.

Even if you borrow a few trillion dollars and pay a few million people to carry rocks from one pile to another and back – you’re still going to generate some serious economic activity.


abb1 03.17.06 at 4:24 pm

For a while, that is.


Kevin Donoghue 03.17.06 at 4:30 pm

Well abb1, suppose you borrow a few trillion to cover a tax-cut, the Federal Reserve raises interest rates to forestall a rise in inflation, so that massive capital inflows push up the dollar and consumers take advantage by buying foreign goods. You generate economic activity all right – in Asia.

That in brief was the story of Reagan’s first term as I remember it.


abb1 03.17.06 at 4:44 pm

Well, fair enough, I guess; although I seem to remember them imposing some serious tariffs (like 100% tariffs?) on Japanese stuff. But I am not an expert; you’re probably right.


MQ 03.18.06 at 2:10 am

Kevin Donoghue (#41): not a single serious economist, even those who believe the minimum wage causes employment loss, has found that the employment loss due to U.S. minimum wages is even close to large enough to cancel out the income gains from the wage increase. At current and historic levels in the U.S., minimum wage increases substantially raise total income shares going to the poor, period. Not controversial.


Kevin Donoghue 03.18.06 at 5:07 am

mq, I thought you were being facetious when you wrote (#39) that the effects being discussed by Bartels “could really be a straight up effect of minimum wage increases under Democratic administrations”. It seemed like a parody of the efforts, which some commenters are making, to trivialise what Bartels is saying. It now appears you really meant it. Apologies for misinterpreting you.

I am sceptical of your theory. As Bartels points out, unemployment has been 30 percent lower under Democratic presidents, on average; GDP growth has been 30 percent higher under Democratic presidents, on average. I’m pretty sure you can’t name “a single serious economist” who believes that raising the minimum wage could have had beneficial effects of that magnitude. At any rate I would find it hard to take such an economist seriously.

To me it seems that what we are looking at is the effect of differing macroeconomic policies. Republicans are happiest when cutting taxes. If austerity is called for they prefer to tighten monetary policy rather than cut spending. In these and other respects their policies are less favourable to the poor than those of Democrats. I think that’s roughly how most mainstream economists would interpret Bartels’ findings. It’s not an elaborate or implausible explanation as your comment seems to imply. By all means suggest alternatives, but persuading me that minimum wage increases have much to do with it would be a tough assignment.


Hektor Bim 03.18.06 at 10:12 am

The results here are broader than that, though. The Republicans’ policies are not just less favorable to the poor, they are less favorable to 80% of the population. Unless minimum-wage laws are more powerful than I understand them to be, there is much more going on that that.


BigMacAttack 03.18.06 at 1:17 pm


He most certainly does and did. The numbers might back up those assumptions but he certainly did and does make them.

There is a long standing correlation between the conference of the Super Bowl winner and the stock market. Appearently there is an even greater correlation between Jan’s barometer and the stock market.

Now, causation between party and economic growth is certainly on the surface more plausible than causation between the conference of the Super Bowl winner and stock market.

But one key thing here is that Bartels cannot identify what Democratic policies lead to economic growth. Minus that you might as well place stock market bets based on the Super Bowl.

Some of his baseless speculations are almost comical. It’s Democrats easy monetary policy. Only as he notes economists haven’t been able to make that link. And who controls monetary policy? Oh yeah the Fed. So Brett would be right, Clinton’s growth belongs to Reagan who appointed Greenspan. And the Reagan years belong to Carter who appointed Volcker. As does the disaster of 1979-81 under Miller. The rest of the 70s we can blame on the evil bastard Nixon and Burns. Looks like LBJ made his own monetary policy. So we can give the good 60s to Democrats. The 50s still belong to Ike and Martin.

Without a real causal link it is all just interesting bunk.


Henry 03.18.06 at 1:44 pm

No bigmacattack, not so. You’re confusing two different problems here. One is where two variables appear to be correlated, but there isn’t any reasonable sounding causal path connecting them, as with the Superbowl analogy. Here it’s pretty reasonable to dismiss the relationship as almost certainly being spurious. The other is where there a number of plausible causal paths relating the one variable to the other, but it’s impossible on the basis of the data to decide which. Here, it’s highly likely (although never certain – that’s what measures of statistical confidence are all about) that there _is_ a real causal relationship, but it’s difficult to decide which individual causal path or combination thereof are causing it. Two quite different problems.


Kevin Donoghue 03.18.06 at 3:47 pm

BMA: Some of his baseless speculations are almost comical. It’s Democrats easy monetary policy. Only as he notes economists haven’t been able to make that link.

If there is a comedian in this show I think it’s you, BMA. You certainly found a funny way to read this:

Hibbs (1987, 218) argued that, given the class composition of their respective supporting coalitions, “Democratic administrations are more likely than Republican ones to run the risk of higher inflation rates in order to pursue expansive policies designed to yield lower unemployment and extra growth.” His empirical analyses (based on data from 1953 through 1983) supported these claims….

There is also this:

Previous work by Hibbs (1977; 1987), Keech (1980), Beck (1982), Alesina and Sachs (1988), and others has documented consistent partisan differences in economic policy, with Democrats striving to reduce unemployment and Republicans focusing primarily on controlling inflation.

Couldn’t make the link, eh? But perhaps you are referring to this remark:

Notwithstanding these investigations of the “political business cycle,” we know less about partisan differences in economic outcomes than we do about partisan differences in economic policies.

If so then you are simply misreading the paper. Bartels is just pointing out that what he has to say is new, or at any rate a new angle on a familiar topic. It certainly is not news that tight monetary policy hurts the poor and that inflation frequently sends the real after-tax return on investments into negative territory. Really, what is new here is just the fact that a scholar is ignoring the demarcation line between politics and economics. Of course J.K. Galbraith did it years ago but in best-sellers, not in scholarly journals. Those who have read Money: whence it came, where it went may wonder what the fuss is about. Of course the rich pursue policies which enrich them further. Was it ever otherwise?

And who controls monetary policy? Oh yeah the Fed.

Even if we take that exceedingly simplistic remark at face value, the Fed doesn’t operate in a vacuum. If lax fiscal policy threatens to generate inflation, the Fed will raise interest rates. Even without people like Robert Rubin and Larry Summers to advise him, Bill Clinton could figure that out. So, wanting low interest rates, he needed to buy off Greenspan with a tight fiscal policy. The fact that the Fed controls interest rates doesn’t mean that a president can’t seek to maximise economic growth; it’s just one of many constraints that need to be taken into account.

Of course the degree of independence enjoyed by the Fed varied quite a bit over the period studied by Bartels. That, in part, is why Clinton had to follow a more conservative policy than JFK or LBJ. They had old-style Keynesians encouraging them to play the Phillips Curve (though not as recklessly as they actually did) and a tame Fed. So by all means distinguish between the different generations of Democrats and Republicans. The brands have evolved. But if your contention is that there isn’t a persistent difference between the parties, then you have an awful lot of evidence to brush aside.


MQ 03.19.06 at 7:41 pm

Oh, come on. Do you really believe that macro policies pursued by Democratic presidents raise GDP growth by 30%, that the relation is causal? And cut unemployment by 30%? That we have sufficient knowledge of the economy for policy to have these kind of impacts in the short run? Especially given that the President is at best the fourth most powerful influence on macro economic policy (after the Fed, the House, and the Senate)? If what’s going on here is that growth is that much higher under Dems, the relationship is either conincidental, or is driven by selection on when Democratic presidents are elected (e.g. coming off of recessions that discredit Republican leadership, when growth is likely). Certainly minimum wage increases could not produce these macro growth effects. But they do have the virtue of being clearly related to Presidential action and directly raising incomes at the 20th percentile. In fact, just about the only way this could be a true causal effect driven by Presidential policies is if it is the result of minimum wage increases. Anything else is most likely a spurious correlation.

Comments on this entry are closed.