Tuesday, September 23, 2008

Do Democratic Presidents really produce better economies?

Just because they are Democrats? With no other particular explanation? That's the claim made by the likes of Michael Kinsley, Alan Blinder, and host of other "authorities" and bloggers, this election year (and just about every other one). And they cite plenty of data to back it up. For instance Professor Blinder writes in the NY Times....

Data for the whole period from 1948 to 2007, during which Republicans occupied the White House for 34 years and Democrats for 26, show average annual growth of real gross national product of 1.64 percent per capita under Republican presidents versus 2.78 percent under Democrats.
Wow, that looks hard to argue with! Can we really believe Democratic presidents produce better economies ... just "because"?

No. Of course not. Don't be stupid. You should decide for yourself.

I'll submit four items on point for your consideration, and then you can make up your own mind. These items are: (1) logic, (2) common sense, (3) actual data, and (4) a few basic principles of statistics.

1) Logic. The cited "presidential performance numbers " are invariably tallied by the calendar years of each president's term. But how is the mere party affiliation of a president supposed to affect the real economy from his very first day in office (in fact, from before his first day in office, when not inaugurated until January 21) ... even if the Congress is of the opposing party ... and then cease having any affect at all on the economy as of December 31 of his last year in office?

Logic seems to dictate that a president has to do something to affect the economy -- take some policy action. And when I studied economics in school they taught me that there is "lag time" between when an economic policy is adopted and its having an effect -- such as up to 18 months for monetary policy and longer for fiscal policy.

What policy is it that all Democratic presidents follow to produce such superior results? The Kinsley-Blinderites don't say. What rationale do Kinsley/Blinder & Co. give for dismissing the lag time before policy has effect? They give none. Is the party affiliation of a new president supposed to instantaneously drive the economy in a new direction through "magic totem power"? They don't say.

"Lag time", hmmm, this leads to...

2) Common sense. Jimmy Carter, D, left office leaving behind double-digit inflation created during his term, the worst in US history. As the direct result of this Paul Volcker and the Fed (not Reagan and the Republicans), to break this inflation, engineered the worst recession since the Depression during Reagan's first two years, causing double-digit unemployment -- but curing the inflation Carter had left and setting up a long-term boom economy.

The Kinsley/Blinder-type presidential performance tallies count that as: "Carter, D, growth years, Good! ... Reagan, R, bad recession, Bad!"

Doesn't common sense tell us there is something wrong with that?

As a matter of fact, Eisenhower left Kennedy a boom economy ... Lyndon "guns and butter" Johnson left a stock market bubble that busted and turned into a recession at the start of Nixon's term ... and Clinton left a bursting bubble that had just started a recession that surely would have landed on Gore if he'd been elected (even though a Democrat!). Plus Carter's double-digit inflation that lead to double-digit unemployment.

In each case assuming presidential "magic totem power" that affects the economy instantaneously and ends its influence on the last day of the president's term, instead of assuming that presidential policies affect the economy with a texbook lag, works to reward the Democrat and punish the Republican. And just these four cases stretch across fully eight of the 14 presidential terms since 1948.

Which brings us to....

3) Factual data. There've been 14 full presidential terms from 1948 to 2004, 6 Democratic and 8 Republican. Count real GDP growth from January 1 of the year each president was inaugurated to December 31 of the fourth year, and we get this striking result:

Of the 7 terms with the highest GDP growth, Democratic presidents have 6 of them ... and average GDP growth per term is 17.95% for Democrats versus and only 11.75% for Republicans.

Wow! That's pretty impressive!! But still ... it assumes the president's party affiliation drives the economy from his day one in office without his even actually having done anything yet.

Let's instead consider the old boring textbook proposition that policy takes effect with a lag, and look at GDP growth for each term lagged 18 months.

Now of the 7 terms with the highest GDP growth, 5 belong to Republicans -- and average GDP growth per term is 14.7% for Republicans versus 13.9% for Democrats.

So, do we believe the textbooks when they say that policy affects the real economy only with a lag, or do we not?

If we do, that's the end of that story.

In fact, looking at the lagged data one could even credibly argue that Democrats have repeatedly left office leaving behind overheated, broken economies that have goosed their own short-term numbers up, and that Republican presidents have paid a price to inherit.

But I wouldn't say that's the true nature of Democratic presidents either, because of ...

4) Statistical basics. First of all, small sample size. C'mon what are we talking about, all of 11 presidents?

And consider all the noise in this data. E.g: Historic oil price increases to all-time highs slowed the economy under Nixon and Bush the Younger ... but Clinton received the gift of oil falling to $10 a barrel. That's two Republicans punished and one Democrat rewarded by chance in a sample size this small! How are things like that -- all the other external, random forces that influence the economy -- adjusted for?

Second, there's the lack of independence between events. Consider this: The Democrats tried seriously to recruit Herbert Hoover to be their presidential candidate in the 1920s -- even FDR favored him(!)

Imagine that Hoover had run as a Democrat, won, and the Great Depression thus had happened on the Democrats' watch. The Republicans would have swept in for a generation and party alignments would have been reversed for 40 years, maybe until this day. That one single possible event would have had a pretty big impact on all these numbers.

Third, most bascially, there is the false assumption that "correlation is causation", as pointed out by Prof Mankiw. Even if, measuring with an 18-month lag, Democratic presidents had taken all the top 6 performance spots (instead of just their actual 2 of 6) it still wouldn't have meant anything unless there was some explanation of how they affected the economy. What was their common policy action or influence that enabled them to do this? The Kinsley/Blinderites offer nothing at all here. Zip.

But without an explanation that shows and explains causation ... heck, I can detail the movements of astronomical bodies that correlate 1-to-1 with the past performance numbers of professional athletes. Of course that means nothing, it's just data mining.

Now, these four problems with the thesis under examination are neither difficult nor obscure. Persons as informed and educated and usually critically astute as Kinsley and certainly Blinder surely would see and appreciate them in a moment in any other context. So why are these two blind to them here?

I can only think because it's an election year, and this is yet another example of "your brain on politics".