July 16, 2003
From the desk of Jane Galt:The Dangers of Data Mining
I want to have a little chat about an economic fallacy that one sees all over the place, from commentators both left and right: the idea that because some number was particularly good under the presidency of one's chosen political party, that this therefore means that their policies are correct. Specifically, they use this kind of data to claim that some thing of which they are in favor is good for the economy -- without reference to what that piece of data would have looked like absent their favored item.
This fallacy is extremely widespread. I will use one example apiece from left and right.
For a while, leftish bloggers were quite caught up in the notion that the correlation of good economic performance with Democratic presidencies had to be causal. Clinton was thus "good" for the economy.
But when you want to argue that a president is good for the economy, you cannot just point to the fact that the economy was good while he was president, QED.
Even the most policy wonkish can concede, surely, that there are some influences on the economy larger than that of the president. If green men from Mars had invaded during Clinton's presidency and razed the capital infrastructure to the ground, forcing us to return to subsistence farming with the attendant loss of life -- surely we would not blame Clinton for the entirety of the resulting 97% contraction of GDP, would we? We might argue that he should have prevented the invasion, or at least valiantly climbed into a fighter plane like Bill Pullman and at least try to chase them off, but we would not argue that better fiscal, monetary, or social policy would have averted the, er, recession.
Indeed, if you begin totalling up the outside influences, you quickly realize that they are much greater than the influences of fiscal and monetary policy, at least within the rather staid limits set by the American voting public. (Our staggering new budget deficit is a dainty 4% of GDP. The Argentinian government would have laughed such miserly sums off the playground during its heyday.)
If you want to claim Bill Clinton was good for the economy, you must first establish what the economy would have looked like without the benevolent policies you give him credit for. If we were due for a nice long internet bubble, Clinton's policies may have helped, hindered, or done little either way. But simply pointing out that GDP and tax revenues grew doesn't give us any clue as to which of the three is true.
Now, conservatives may be chortling. But you have your own wing-nuts to contend with, my frosty friends. Like the fellow I saw the other day trying to refute the standard macroeconomic belief that if you raise budget deficits, you decrease national savings, and thus investment, and thus future productivity gains. His argument? In the 1980's the budget deficit increased but interest rates went down.
Well, yes, they did. But that might have had a little something to do with the fact that the government abruptly halted its massive inflation of the money supply. In fact, if you see how neatly falling interest rates dovetailed with Volcker's aggressive monetary targeting, it becomes obvious that by far the largest influence was the Fed's war on inflation.
It is thus insufficient to point out that interest rates were dropping. This tells us nothing about the effect of the budget deficit on interest rates, which is generally, IMHO, relatively trivial. We can only figure out that effect by first determining how much interest rates would have fallen in a neutral environment, and then analyzing whether the budget deficits measurably added to it.
But how do I do that, I hear you cry. Why, it is difficult, my little chickadees; that is why people have to get PhD's and things. It is so difficult, in fact, that when you see a blogger who has claimed to prove some grand theory, such as the superior economic performance of their political party, or the ability of budget surpluses to generate astonishing rates of growth, using only numbers they can find on the internet in fifteen minutes or less, you should be very, very suspicious.
Posted by Jane Galt at July 16, 2003 6:07 AM | TrackBack | Technorati inbound linksAll this is true enough, but still it evades the point that liberals are actually making and targets a strawman instead.
The Lib point is not that the correlation of "Good Economic Times" with Dem presidents proves that Dem economic policies are correct. No one is really making that argument except for you Jane.
The "Liberal" point is that it disproves that usual Republican argument that their policies are better for the economy than Democratic policies.
I think it possibly, even likely, that Presidents don't really have much effect on the economy, and the effect that they do have is the result of ALL of their policies taken together rather than the subset that are directly economic.
One doesn't have a theory of gravity to know that gravity exists. And similarly, dispite your desire to deny it. There is 100% correlation between Democratic presidents and better economic outcomes for the past 50 years.
Yes, correlation does not prove causation. But since I only care about the outcome when I vote rather than about proving a theory. The smart person votes Democrat, at worst it's irrelevant to the economic health of the country, and at best it's good.
Think of it as Pascal's Wager for presidential elections.
Bones, your post is a veritable festival of innumeracy.
The correlation neither proves, nor disproves, the theory that Republicans are better for the economy; it tells us nothing unless we know what the state of the economy would have been without the Republicans.
I think it's lovely that you're going to vote Democratic, and am touched by your belief in the magical powers of the Party. But might I suggest that your other posts indicate that you would probably vote Democratic even if the correlation ran the other way?
And the smart man doesn't bet that the roller's going to turn a seven just because he saw him do it six times in a row -- not unless he's got better evidence than that that he's rolling with loaded dice. But if you're the kind of fellow who does -- why, drop by the next time you're in New York, and we'll shoot some craps.
Just as a gentle note: comparing a political or economic philosophy to Pascal's Wager is guaranteed to bring a look of horror to the eyes of anyone with more than a freshman-level knowledge of philosophy. I read such comparisons and mentally translate, "Okay, so it's superficially appealing to people who accept its unvoiced assumptions, but falls apart completely upon thoughtful examination?"
What you seem to be saying is that economics is too complicated to take into account when choosing a candidate for president.
Gotcha, that leaves foreign policy and domestic social issues and no rational reason (for me) to ever consider voting for a Republican ever again.
No, what I'm saying is that if you want to choose a president based on economics, you have to look at more than his party, and you need to examine his proposed policies more carefully than the kind of analyses I cited.
Before you smack anyone down too hard over their understanding of economics, you may want to correct the sentences "...the government abruptly halted its massive inflation of the money supply..." and "...Volcker's aggressive fiscal targeting..." ;)
Presidents don't have the tools to have a major effect on the economy in the short run (except for ill, like Nixon's price controls). But the short-term effects of their actions are grossly exaggerated -- plus and minus -- for political purposes, for political reasons.
Presidents have to spend a lot of time to get an economic policy enacted, assuming they have one. And then the policy effects work with a lag.
E.g., Carter's major economic acheivement and influence wasn't seen in his own term, which was affected by oil crises and bad monetary policy that weren't his fault. It was the *degregulation* of many major industries that started in his administration -- and which didn't have a big effect for a long time after. And arguably Reagan's biggest economic effect, more than tax rates and deficits, was seen in his continuing of the deregulation Carter started.
LBJ's big economic effect wasn't the boom that lead to the big stock market top in his own term -- it was his overheating of the economy to get "guns *and* butter" that built into the lasting inflation problems of the '70s (and Nixon's price controls).
Even FDR's big economic effect wasn't economic growth during his own time in office, that *had* to happen after a Depression, and his policies arguably slowed it. His *big* effect was the creation of goverment agencies and programs such as Social Security that remained modest in size in his own time but which grew and grew and grew afterward and are still driving the agenda in the 21st Century, for better or worse.
Each president has to work with the legacy left to him, for better or worse, and each president's own biggest effects will be seen only in the legacy he leaves behind for others (unless one *really* screws up). So you can't just compare them by what happens within their own terms with the day they leave office the cut-off date for comparing. That's missing 80% of what they do.
Yep, people on both sides of the debate exaggerate the good their side does and the bad the others create.
It's always been that way and will probably always continue to be so.
Having said that it would be wrong to say that one can't learn anything from history. If, as has been claimed in other threads, the economy has consistently been better under Dem presidencies (well, at least in the modern era) that is telling us something.
Maybe I wouldn't go as far as Bones and say it disproves the Republican's view. But at the very least it makes any GOPer's claim that they are better for the economy highly suspect.
Just FYI Jane: Doing a heteroskedastic t-test on Real GDP growth rates for Democratic versus Republican presidencies for 1949-2002 (lagging one year), gives a significance level of >95%.
Now there are all kinds of philosophical problems here - including the underlying assumption that exogeneous factors are random (otherwise we couldn't use predictive statistics here). But it shows that the differences are very unlikely to be just due to 'random' chance.
Greg Mankiw, in his intro Macroeconomics book, speculates that the differences in growth between Democratic and Republican presidents may be because of the GOP being more hawkish on inflation. Unfortunately for him, a t-test on CPI inflation for Democratic versus Republican presidents gives
Unlike most other occupations, you can do useful work in economics using publicly available data & a spreadsheet. If some liberal blogger/blog commentators (like me) want to challenge folk myths dating from the age of the Gipper, why object to it?
As Bones & Michael have commented, I think you're evading the point. [Surely you can't want Amity Schales job that much.]
I didn't state that it was not possible to disprove either the proposition that Republicans are better for the economy (although I think there would be insurmountable technical problems with the huge variance between presidents), or that raising budget deficits has negative effects on growth; I was simply pointing out that you can't disprove them using those sorts of silly numbers. Not every single thing I say is a political argument, dears. And Achilles, I found the typo before you did, and fixed it even as you typed. ;-)
Yep, I noticed that just after I posted. By the way you may want to fix the other one too since the government does not really control the money supply... not yet at least ;)
I agree that the historical evidence between presidential party and econ performance does not absolutely disprove the idea that GOP presidents are better than Dems.
But it is completely wrong to say it tells us nothing. Because it does. It's just a first step, but a step nevertheless.
Dwight's original point remains a valid, if limited one.
Jim Glass wrote:
[A lot of good points]
Jim, I'd agree with you up to a point, but I think you're undervaluing the effects of short-term fiscal policy. Now, it's arguable that such effects are less under the control of a US president than a UK prime minister; but the tone set for fiscal policy by the Whitehouse is important Frex, economists such as Bob Solow or Paul Samuelson were been pretty critical of Bush's tax cut, because of the lack of short-term stimulus in the package; it's not that surprising that we didn't see much bang for the buck subsequently.
I'd still argue with you over FDR & the New Deal, BTW; if you look under the old thread a week or so back, I posted a response to your reply to my comment.
How does one adjust for
a) the tendency of most administrations only to achieve the policy objectives that go most against the grain of their own party (e.g. Clinton:Welfare Reform but not Health Care)
b) the substantial changes in the defining elements of the party's stance on the economy (e.g. current Democratic rhetoric about the budget being closer to Eisenhower/Ford/Dole)?
Policies may affect the economy, but certainly not the label, mascot or color of the party in power.
Bones is right to point out that many of the disputed analyses have correctly addressed disproving a relationship, which is about as far as the stats will take you. However, to use his counter-strawman, there is only a "usual Republican argument" correlating these factors in a pure party-boosting setting.
Watching politicians today, I'm not sure that Republicans and Democrats can even be said to act reliably in their own longer term interests.
Incidentally, the most vivid recent example of "a)" above is Lula in Brazil. Now he's cutting pension benefits. Only an insider could get away with that.
Megan, kinda makes you want to go back to "All abortion, all the time," doesn't it? ;-0
"b) the substantial changes in the defining elements of the party's stance on the economy (e.g. current Democratic rhetoric about the budget being closer to Eisenhower/Ford/Dole)?"
Again, FYI: Democratic presidents tend to run deficits that are lower as a %age of GDP than GOP presidents. [Though I'd say this is a side-effect of the differences in growth].
When talking about Presidential achievements - I find it frustrating that the legislative branch is ignored. Sure Presidents have significant influence over legislation - but it is the Congress which writes and passes the budget, spending bills, tax bills, and every other law - and Congress has by far the predominant influence over the substance of new laws.
Presidents do a good job of claiming credit for whatever passes Congress but by and large the end product looks nothing like what the President proposed (assuming the President proposed anything). Few things are as nonsensical as "Clinton's Welfare Reform." If you are going to have a serious discussion about the influence of Presidents, there ought to be some acknowledgemnet that by and large Presidents accept the policies that Congress originates. Congress has ceded a lot of power by allowing the regulatory state to grow so large - but domestic policy in the US is still largely controlled by the Congress.
"When talking about Presidential achievements - I find it frustrating that the legislative branch is ignored. Sure Presidents have significant influence over legislation - but it is the Congress which writes and passes the budget, spending bills, tax bills, and every other law - and Congress has by far the predominant influence over the substance of new laws."
Who's ignoring them? I've got a spreadsheet breaking down the GDP growth numbers according to Dem/GOP control of Whitehouse, Senate, House. (And I sit and argue with myself as to what time length I should use for policy lag.)
It takes 2-3 hours tops to get the data and load it into excel; go knock yourself out.
I'd recommend using a lag of at least 10 years, Tom. At least, that's about the length of time the chorus of moaners over at Brad DeLong's seem to believe it will take for Bush's policies to bring doom upon us. Sure would be a pity to see whoever gets elected in 2012 to get stuck with the blame.
Maybe I wouldn't go as far as Bones and say it disproves the Republican's view. But at the very least it makes any GOPer's claim that they are better for the economy highly suspect.
Holy loads of reading comprehension problems. Her post doesn't prove anything, but, in fact, all she is stating is that there are a number of factors that must be taken into place when making an anlysis of the economy and the economic impact of one party's policies. Thus, arguments like:
Economy Good under Clinton, thus Democrats are good for the economy
or
Economy good under Reagan, thus Republicans good for economy
are equally simplistic and stupid. Sheesh.
In 1989 a financial analyst looked at the influence that the baby boomers had on the price of housing in the 80a and made a very simple prediction. In his book "The Coming Global Boom" he said that the boomers were reaching the time in life where people save. Thus, equity markets were going to boom in the 90s. He was right and for the right reasons.
I understand that Clinton has a prolific sex life, but I don't think we can give him credit for the baby boom following WWII.
Congressional dominance, state fiscal policies, judicial "policies", climate, demographics, war -- if you don't control for all of those, you don't have any control at all.
Gee, I thought it was weenie liberals who said averything was complex and hated simplistic thinking. And we got our asses whipped for two decades by simplistic thinkers.
Whoopee! Let the wild rumpus begin!
Zizka- Only when people try to provide complex answers (excuses) for simple problems do I call them weenies. And weenieness (new word?) is bipartisan.
The president in 2012 will sure have a lot more debt and interest, if not blame, to deal with. I am sure George H.W. Bush has much to say about how much political fun it is to have to raise taxes to pay off debt accumulated via past structural deficits.
At first I read it as "razed the capitol infrastucture to the ground", and I thought "hmmm, ya know, that might actually boost GDP". =)
A President's largest economic influence may derive from policies that are not viewed as economic. The EPA, a Nixon creation, has had a large if ill-defined influence on the economy, but I don't think Nixon viewed the issues prompting the creation of the agency as primarily economic. The economic consequences of such policy choices may take decades to manifest themselves -- making it very difficult to determine how a particular president "did on the economy." Of course, if it's hard to grade a president, it's even harder to grade his party.
Another reason it's so difficult to compare GOP and Dem president's economic performance is the policies of the two parties have not remained constant over time. Wasn't Regan fond of saying, "I didn't leave the Democrat Party. It left me"?
Who knows what JFK or Nixon would propose if either were President today? We do know that Nixon's economic policies (price controls) are closer to today's Democrats and that Kennedy's (tax cuts) are closer to today's GOP. Simply doing a statistical analysis based party affiliation divorces the actual policies from the results.
"We do know that Nixon's economic policies (price controls) are closer to today's Democrats and that Kennedy's (tax cuts) are closer to today's GOP."
Are yes, simple statistical analysis will definitely prevent one from reaching the conclusions that can be attained by relabeling the guys on your side who had terrible economic policies as being "closer to the other side" and relabeling the guys on the other side who had good economic policies as being "closer to our side".
Jane: I more or less agree with your basic point, but I think you're trying to push it too far. The postwar period now encompasses nearly 60 years and 11 presidents, which is close to being a large enough data set to draw some tentative conclusions.
So...if Democrats rather consistently outperform Republicans, even with time lagged comparisons, then in order to say that the data means nothing you have to argue that by some weird coincidence the exogenous variables *consistently* favor Democrats more than Republicans. This is a little hard to swallow.
Even with a very sophisticated analysis we may not be able to discern the mechanism, but when enough data piles up, it makes sense to suspect that *something* is going on, even if we're not quite sure what, and even if we think the influence of presidents on the economy is relatively small. Mendel figured out the laws of inheritance in such a way, even though he had no clue what the mechanism might be.
On a different note, I'd also like to point out that your basic point about the complexity of economic behavior, correct though it may be, also applies to most other fields of human knowledge. If having a PhD is a prerequisite to discussing anything important, the entire blogosphere would disappear in a puff of smoke. So I say: go to it, bloggers! Don't let Jane's harrumphing keep you from trying to convince your fellow man of the righteousness of your cause using every tool at your disposal! And may the best man win.
Mindles: I agree that the "only Nixon can go to China" syndrome is genuinely fascinating. I wonder, for example, if a Democrat could ever have gotten the prescription benefit bill passed?
Surely, achilles, you admit that more Democrats support price controls in general, and more Republicans support marginal income tax cuts in general. It's definitely worth saying that Nixon and Kennedy's economic policies are both not generally what is advocated by their party.
Kevin, I've long argued for the "only Nixon can go to China" argument as having some validity. It definitely does on free trade. Republican legislators tend to be more free trade than Democrats-- note how a majority of Republican voted to give Clinton fast track authority, but relatively few Democrats. (One has to adjust for particular districts, but generally the Republican candidate in a district is at least as pro-free trade as the Democratic one. It's certainly the case in all my observations in North Carolina, including the Erskine Bowles - Elizabeth Dole Senate race last year, when Bowles ran a ton of protectionist ads.)
However, members of both parties tend to support their own presidents. And presidents try to reach out to the other side by defying their party, relying on being able to keep their base anyway. When a Republican president tries to reach out to Democratic-leaning voters that they might be able to win (like union voters), he compromises on free trade. A Democratic president reaches out to Republican-leaning libertarians and the like by being pro-free trade.
Kevin, if I thought that only a PhD could discuss economics, I wouldn't have much of a blog, now would I? But myself, I let the PhD's do the heavy lifting most of the time and then interpret their results -- I don't go out and grab some random number to "prove" the economic veracity of my political beliefs.
And Kevin, am I to take this to mean that you would, say, stake your company's money on a marketing survey that used results from eleven people over sixty years to extrapolate what the next guy who buys will do?
“If having a PhD is a prerequisite to discussing anything important, the entire blogosphere would disappear in a puff of smoke. So I say: go to it, bloggers! “
“But myself, I let the PhD's do the heavy lifting most of the time and then interpret their results -- I don't go out and grab some random number to "prove" the economic veracity of my political beliefs.”
The Ph.D. behind someone’s name is sometimes nothing more than proof that they are an intellectual slut. Still, it’s prudent to normally give them the benefit of the doubt---until further evidence indicates that they are idiots. An individual possessing a Phd, after all, usually knows more about their area of specialty than the proverbial red neck anti-intellectual slob that Richard Hofstadter wrote about in his insightful “Anti-Intellectualism in America.”
"Even the most policy wonkish can concede, surely, that there are some influences on the economy larger than that of the president."
Politicians are naturally motivated to present the argument, "Look what I have done for you." So presidents are essentially obligated to overstate their "accomplishment" in improving the economy.
These arguments are grounded in the mistaken belief that the economy is sensitive to top-down control(i.e., "I can do things to deliberately and significantly help OR hurt the economy.") I suspect that any administration's effect is based more upon psychological mechanisms than upon what are regarded as technical economic mechanisms.
A process which is even *more* foundational than overstating the presidents' effects upon the economy, is the basis for making *any* claim in the political arena: say good things and only good things about your own position, and say bad things and only bad things about alternative positions.
Nixon was a Republican, he ran with a set of ideas on economic policy (and other policy) that beat the competing Democratic platform, so I am afraid his policies are the policies of a Republican president. Similarly, our current president has a track record that includes increasing Federal deficits, expanding discretionary spending and protectionism - these have to be considered (just as much as the tax cuts) as being the economic policies of a Republican president - there's no disavowing it. I was simply pointing out that it is silly to throw out the underperformers (embrace the otehr side's overachievers) and then claim your side is superior.
For a particularly humorous piece along these lines, you can check out the Larry Kudlow piece in NRO that basically relabeled presidents who presided over economic booms (Kennedy, Clinton, Coolidge) as being 'supply side presidents', which, lo and behold, proved the overwhelming power of supply side economics!
This is of course, all wrapped up in the caveat that one wants to play the game that Jane says we should not play. As for myself, I prefer to live by the axiom my old college professor taught me: presidents always get more credit and blame than they deserve for the economy. I also think that one can judge president's performances on a lot of dimensions: discretionary spending, structural deficits, allowing the Fed to do its job, leadership in trade policy, appointing competent economic policy makers, coherence of overall economic vision that are more directly under their control than unemployment or GDP.
The deeper stuff I leave to smart people, some with Ph.Ds even, who have analyzed political business cycles and examined the impact of different parties being in power on the economy. Such analysis often weakens conclusions reached from crude statistical analysis, they rarely reverse them.
achilles - I wasn't advocating relabeling prior presidents, such as Nixon and Kennedy. I was simply pointing out that, using Nixon's or Kennedy's performance (policies) to try and predict the performance of a current member of their parties is, at best, problematic. Each party has moved on and, if party has any influence on policy, it would seem that past performance would have little predictive value. In other words, you shouldn't vote for a generic Republican just because you are enamored with Nixon's economic policies.
Jane: And the smart man doesn't bet that the roller's going to turn a seven just because he saw him do it six times in a row -- not unless he's got better evidence than that that he's rolling with loaded dice.
True. Although, if forced to bet, I'd go with seven. At worst, I've still got a 1/6 chance of being right. And Bayesian logic would start to suggest that the dice aren't quite right at this point. :)
Kevin: If having a PhD is a prerequisite to discussing anything important, the entire blogosphere would disappear in a puff of smoke.
Heh. There are actually an inordinate number of Ph.D. bloggers out there.
"For a particularly humorous piece along these lines, you can check out the Larry Kudlow piece in NRO that basically relabeled presidents who presided over economic booms (Kennedy, Clinton, Coolidge) as being 'supply side presidents', which, lo and behold, proved the overwhelming power of supply side economics!"
My word. That's the same rigorous statistical analysis that I used in undergraduate physical chemistry experiments.
hi jane,
"But how do I do that, I hear you cry. Why, it is difficult, my little chickadees; that is why people have to get PhD's and things. It is so difficult, in fact, that when you see a blogger who has claimed to prove some grand theory, such as the superior economic performance of their political party, or the ability of budget surpluses to generate astonishing rates of growth, using only numbers they can find on the internet in fifteen minutes or less, you should be very, very suspicious."
and,
"But myself, I let the PhD's do the heavy lifting most of the time and then interpret their results -- I don't go out and grab some random number to "prove" the economic veracity of my political beliefs."
so, why wouldn't you need to have a phd to actually interpret the results. why would that task actually be easier than doing the initial research. after all, doesn't the ph.d actually have to interpret the data they have, in order to ascertain its meaningfulness? and isn't that ability to see meaningfulness a product of all those years of arcane learning?
in a word, i think you are wrong on that, and i suspect that you are exaggerating, perhaps deliberately, to get a discussion going (which is ok). as far as i can tell, someone who has done some econometrics or some basic statistics can do quite a decent job in following these debates and participating. the major issue is time, and academics usually have more of it than other folks to do these sorts of analysis. why are you worshipping at the altar of academic credentialism? it has its place, but please, a little decorum!
Jane's faith that Ph.D.'s in economics can explain the causal connections between economic performance and policy is not shared by the admninistraton.
In a quite remarkable statement, OMB Director Josh Bolten is reported to have said:
"I think the art and science of economics has not yet advanced to the stage where we can really properly capture all the positive effects the tax cuts do have on the economy."
"And Kevin, am I to take this to mean that you would, say, stake your company's money on a marketing survey that used results from eleven people over sixty years to extrapolate what the next guy who buys will do?"
Heh heh. You would be shocked and appalled, perhaps, at how close this is to the truth about how most companies make marketing decisions.
James: there are indeed an awful lot of PhD bloggers. However, I note (ahem) that virtually none of them confine themselves to their areas of expertise....
Tom, a heteroskedastic t-test might be useful if your dataset consisted of independent observations.
Which it doesn't.
"Tom, a heteroskedastic t-test might be useful if your dataset consisted of independent observations.
Which it doesn't."
Yes, I am well aware of that, thank you, as you would have realised had you noted that I spoke of the "philosophical problems" with using a t-test.
That pink thing you see at the edge of your vision is your nose, by the way.
And yet those "philosophical problems" didn't stop you from claiming that the test "shows that the differences are very unlikely to be just due to 'random' chance." When you seem to know enough statistics to know that the test makes assumptions that aren't valid in this case.
Thanks for clearing me up on the nose thing. Was wondering what that was.
Don wrote:
"And yet those "philosophical problems" didn't stop you from claiming that the test "shows that the differences are very unlikely to be just due to 'random' chance." When you seem to know enough statistics to know that the test makes assumptions that aren't valid in this case."
Note the quotes on 'random' there, Don. The differences aren't due to 'random' chance (in the sense of 'random' being sampling error). [There is no sampling error - we have the entire population.] I'm well aware of the limitations of a statistical analysis at this level.
I've also gazed at the data sufficiently to be convinced that GOPers bollix up the economy more than Dems' appears to work as a rule of thumb. [Either that, or believe that the Dems are exceptionally fortunate in the exogeneous factors they receive & inherit]
There *is* a difference between growth under GOP vs. Democratic presidents, particularly in years 2-3 of an administration (c.f. http://www.yale.edu/leitner/pdf/drazen.pdf for a long discussion on GDP growth rates and the electoral cycle).
Reading that paper, I see the author couldn't figure out a policy variable difference; neither does Mankiw pull one out in his textbook. I won't feel too bad that I haven't figured out the policy variables to which I can attribute the GOP/Democratic differences.
Don, I noted this in another comment (at http://www.janegalt.net/blog/archives/004258.html)
In that thread, some other bloke said:
"What this means is that the statistical results could be result of random chance. Precision is related to sample size, the smaller the sample the less percision you tend to have."
I said:
"'Err, no. In this case, we're talking about descriptive statistics, not predictive ones; we *have* the entire population of post-1948 GDP annual growth rates.
Growth rates under Dem v. GOP presidents *are* different, particularly in Years 2 & 3 of an administration; the differences are *not* a result of sampling of a portion of a population of data.
(Now of course, assigning those differences to differences in policy variables is a non-trivial matter. But, given the way the data breaks down GOPers have the uphill battle to fight).'"
But Tom, that's not true. If the sample you are trying to describe is merely the people who have all been president all ready, you're absolutely right -- your observation is valid, but trivial. There is, in fact, a 100% chance that the Democrats will have had the economy growing faster under them than under Republicans. We don't need statistical analysis to prove it; we can just look at the GNP/GDP data.
If, however, you are trying to predict future presidential performance, the population gets bigger. From the very, very crude way you're applying it, all nine Democratic contenders belong to it; since "Democratic" is the variable, we could stick any of them in office and, according to you, have a good chance of getting improved growth in years 2-3. Presumably we could also elect my Dad and have the same salutary effect, since he's a Democrat. Or my doorman. Since all that apparently matters is that he is a Democrat, regardless of policy choices, then the same magical effect will apply no matter who y'all stick in office.
Tom, I'm not talking about sample size at all. I'm talking about serial correlation, the bane of the time series econometrician. You simply don't have 54 independent data points. Whether it's a population or a sample is quite beside the point.
Here's the simple truth. There are scads of PhD economists who are loyal Democrats and would give their eyeteeth to be able to claim that Democratic presidents "manage the economy" better and create more growth than do Republicans. It would be one of Those Papers -- the ones that get you in top journals, the ones that get you in glossy news magazines that impress your parents, the ones that help advance your political views, the ones that get you tenure. But at this point, one simply can't make that claim and retain one's intellectual honesty. The PBC literature (Nordhaus, Alesina, Rogoff...) is interesting, yes. But until we have a causal link between actions taken by Democratic vs Republican presidents and GDP growth, it's simply inappropriate to draw the conclusions that you draw.
And even if the relationship were definitive, there are any number of reasons why it wouldn't be a reason to vote Democratic. Consider the following. Suppose that when a Republican is in the White House, people expect future smaller government, lower taxes and more wealth, while if a Democrat is at 1600 PA Ave, people expect the converse in the future. The income effects of these expected changes in future wealth would lead people to consume more leisure now under a Republican administration and less under a Democratic one. Ergo, more current economic activity under a Democrat, and a political business cycle consistent with the data -- but certainly not a reason to vote Democratic. Now, I'm not saying that those expectations are reasonable, and this is just a sketch of a model off the top of my head. But there you are.
So, all we have to do is elect Democrats and the economy problem is solved forever? Sounds like wishful thinking to me.
All of this economic theorizing appears to me like trying to figure out what's inside the black box without actually being able to open it up. You can infer relationships by correlating the historical behavior of variables, but you can't really model the actual mechanisms. And you certainly can't do any experiments with the box under any kind of controlled conditions to test your hypotheses.
This is a hard thing to do when you only have a few dozen state variables; I think we're probably talking orders of magnitude more complexity than that.
David Perron and Jane are setting up a false straw man here. The crude statistical analyses, and the sophisticated ones, are using data on how the economy actually performed under Democratic presidents and Republican presidents.
These studies do not have any predictive power for how the economy would do if we were to "stick" Jane's doorman or father or Carol Mosely-Braun or me into the office of POTUS. They may have some predictive power for how the economy will do under the condition that Jane's doorman or father or Carol Mosely-Braun or I WIN the Democratic primary and become the candidate that the Democratic party nominates for president.
If you add that second equation, for winning the presidential primary into the model, then the answer to the question "what is the probability that the economy will boom more under Jane's Democratic doorman than under GWB?" should be pretty clear. [Bite your tongue, achilles, bite your tongue...]
So don't misrepresent the discussion by turning into a crude caricature that goes something like "hey we are Democrats, we are always better". You may (and perhaps should) disagree with the actual claim being made here "hey, if you elect the Democratic nominee you will invariably get a better economy than if you elect the Republican nominee". Weakening that argument may be possible by appealing to econometric flaws, but reversing it is harder. In either case you will have to challenge Alesina and Drazen and others of that ilk so may the Force be with you.
Maybe those Democratic presidents do so well because they come after Republican presidents who have fixed things up, but the changes take 6-10 years to really take hold.
(I don't really believe this theory, but this is the level of analysis that is really going on here, which is to say nearly none.).
Sebastian, you are on to something here. The reason I never liked to play this game too much (outside of reading the more sophisticated political business cycle models) is because of the following two propositions:
a) Democratic presidents are good for the economy while in office, Republican presidents are bad for the economy while in office.
b) Republican presidents are forward thinking people who do things to make sure the economy grows well in the long run but don't worry about something as crass as getting re-elected.
Democrats on the other hand are only worried about getting their jollies in the short run and then leave the economy in shambles after they leave office.
This is a great pair of theories for all to use, if your favorite debating partner is a Democrat (Republican) who believes proposition A (B), all you have to do is espouse proposition B (A). You can both go home happy in the knowledge that you are read.
Of course, one does have to make sure not to question too much whether trade barriers, farm subsidies, structural deficits etc. are really forward looking programs. Also 12 year runs in office really cause problems but one can always abstract away.
Just making correlations between growth and the party of the President in office; without trying to account for their policies and other factors affecting economic growth and activity is simply fallacious reasoning.
Its a form of post hoc ergo propter hoc...Party X is in power when the economy is growing at Y average rate; Party Z is in power when economy is growing at A average rate; therefore Party X or Z is better in office...is like arguing that since the Cock crows, and the sun rises; therefore the sun rises because the cock crowed.
Unless you can show that Party X or Z's administration and its policies caused the different growth rates, or interest rates, its all post-hoc fallacious reasoning. Correlation is not causation. It may be wise to bet that since the cock crowed, the sun will rise shortly. But they are independent variables.
Economic systems are immensely complex, and have a multitude of variables that affect them.
Its also not the case that the President is entirely able to control economic activity. They make grandiose claims to that effect, but the President is not the CEO of USA Inc. Even then, there are other factors that could cause problems despite a President's best efforts. Congressional control over the budget, the Federal Reserve's not inconsiderable influence, economic growth and trade abroad, the weather, and a host of other influences affect the economy and money supply beyond the control of the President.
Presidents of both parties also often go against the normal "grain" of their parties. JFK wasn't a Republican, but his economic policies resembled the GOP "ideal" more than Nixon's did, for instance. Both Bushes have proposed or implemented economic programs not normally associated with Republicans. The policies of Presidents of both parties are hardly consistent and do not follow party lines all that well all the time.
But I am starting to repeat points made earlier in the thread.
"Tom, I'm not talking about sample size at all. I'm talking about serial correlation, the bane of the time series econometrician. You simply don't have 54 independent data points. Whether it's a population or a sample is quite beside the point."
To make an analogy: I doubt that statistics on accidents on Jeep Wranglers versus Volvos are adjusted to take into account that one type is far more likely to be driven by frat boy adrenalin junkies, and the other has an overrepresentation of drivers who are risk-adverse tenured academics with kids. Nonetheless, I know which one I'd rather be inside in an accident.
Achilles has made points I was planning to make more eloquently.
As to cocks crowing and the sun rising: does your cockerel propose appropriation bills amounting to 20% of the solar radiation output? Does it nominate members to the Federal Open Solar Radiation Markets Committee?
Tom (who drives an 18-year old Honda, BTW)
"There are scads of PhD economists who are loyal Democrats"
I thin we may be getting into the heart of the issue here... (BTW, in a New Yorker article, Karl Rove was asked to define a Democrat, and he answered 'Someone with a doctorate').
"and would give their eyeteeth to be able to claim that Democratic presidents "manage the economy" better and create more growth than do Republicans. It would be one of Those Papers -- the ones that get you in top journals, the ones that get you in glossy news magazines that impress your parents, the ones that help advance your political views, the ones that get you tenure."
For what's it's worth, I'm not an economist, nor am I looking for tenure. You seem to have high expectations out of comments on blogs; but I don't think I'm capable of performing the Clark-medal winning quality work you describe.
Using e.g. a t-test involves some heroic assumptions, you're correct; but to assume that the position re. effectiveness were reversed would involve proving that the GOP had had some exceptionally rotten luck. I'm looking forward to those arguing on the other side marking their beliefs to market by showing this.
I can qualitatively propose reasons why the economy might have performed worse under the GOP; overly contractionary fiscal policy in Eisenhowers first term; pretty lousy economic policy (fiscal and monetary) under Nixon & Ford (though, as Jim Glass noted, LBJ should share the blame), and needless to say, the malign influence of supply-siders on Reagan (and the clean-up that Bush senior had to do).
I think Alesini draws a plausible picture of the pattern (strong growth in the early-middle part of a Dem administration, slowing later because inflation fears; conversely, weak growth in the early-middle part of a GOP administration, with better growth in Year 4 due to monetary/fiscal stimulus), but I don't think the underlying inflation evidence is that strong.
Also, I'd consider the stature of economic policymakers in administrations: Rubin & Summers are much, much sharper than Snow or O'Neill, and for the CEA, Yellen & Tyson seem to have had more influence than Lindsey had or Mankiw has. (Jude Wanniski claims that he was consulted on Larry Lindsey's appointment to the FOMC & as head of the CEA [I thought Larry Lindsey was actually a fairly good economist, but don't know why cranks like Wanniski should have *any* input into who the chair of the CEA is])
"But at this point, one simply can't make that claim and retain one's intellectual honesty."
I'll draw a Denbestian engineering analogy here; if I'm designing a distillation tower for a refinery, I don't know the fugacity, activity coefficient, or specific heats of all the hydrocarbons in the mixture; but by using a 0.45 kcal/lb for the specific heat, I can get a design and a budget cost that will be within 20% of the actual.
I'd use a more sophisticated method for the actual design to be built(and I'd probably use a different engineer), but a quick-and-dirty method will be good enough for a quick cost estimate & cost comparison. I believe BCG's slogan is to do the 20% of the work that gives you 80% of the insight.
I'm not talking about a level of proof suitable for a AER paper. I'm talking about sufficient to call into question the rhetoric that GOP supply-sider have been trading on since Carter.
Achilles: Go raibh maith agat, a chara.
The argument about consistency in policies may be beside the point. I think that if a single policy applied consistently in all circumstances was the answer to economic growth, we'd have found it out long ago, and both parties would be using it.
Oh wait, such a policy has been known for 200+ years. The 20th Century conspicuously and definitely proved it, even though no country completely followed it: the more it was violated, the poorer that nation was by the end of the century, almost without reference to starting conditions. This policy, of course, is economic freedom together with the enforcement of laws protecting property.
Too bad neither the Dems nor the Repubs have the moral fiber to actually follow it consistently.
It was the worst that I have every read it was crape it was shit
It was the worst that I have every read it was crape it was shit
It was the worst that I have every read it was crape it was shit
It was the worst that I have every read it was crape it was shit
Comments are Closed.