Either the Republicans are getting better or my mathematics is getting worse.
Every year about this time, I celebrate the release of the president's budget and his economic report by pouring a lot of the numbers from these documents into a spreadsheet. My goal is to reach an objective, scientific conclusion about which party governs better. (Look, everybody needs a hobby.)
The theory is that over 30 or 50 or 80 years - however many the budget documents choose to record, and it varies - any special circumstances, such as war, will average out. Or if they don't, any general claims about the superiority of one party to another are meaningless, which is also a possibility.
I look forward to this exercise because it has always told me what I want to hear: that the Democrats are better - even by the standards of Republicans. In categories such as government spending, inflation and job creation, the record of Democratic presidents has been consistently better.
In fact, the Republicans generally won in only one category, which is - not surprisingly - taxes. According to past calculations, Democrats were better at holding down the size of government (measured in dollars or in civilian employees) and holding down the deficit, but Republicans had mastered the simple trick of cutting taxes without bothering to make up the cost.
Republican presidents as a team won in three categories (lower taxes, lower spending, lower deficit). Democrats won in four (lower inflation, more new jobs, fewer government employees and a lower national debt, but the last is a fluke).
How can one party run up a smaller annual deficit and the other produce a smaller national debt? Only because the reports cover the national debt for a different set of years. Oddly, if you consider only nondefense spending, the Democrats take that category, too - which appears to be a true result, not a fluke. In the most important category - economic growth - the program coughs up an R, but the underlying numbers indicate a tie. I'd call the whole picture a tie as well.
This year's edition drags in Congress as well, comparing years when one party or the other controlled both houses. That may help to explain all the distressing R's that popped up in the boxes that tally the "winner" in various categories.
Of the nine somewhat-overlapping measures, Republican presidents have done better in six. Comparing years of one-party control to years when Congress was divided - with different parties controlling the House and the Senate - the results are equivocal. But comparing years when the same party controlled the White House and Congress with years in which each party had one of them produced a clear preference for divided government.
Of course, it all depends on how you define what you're measuring. How do you compare one party's performance in 1947 with the other's in 2003? They all inherited different situations at different stages in our nation's generally upward economic history. To account for that, this spreadsheet generally measures a variable such as government spending or jobs as a fraction of the gross domestic product at that time rather than in absolute terms. And what counts is how that figure changes.
"Average annual change as a fraction of gross national product," measuring years scattered over as much as eight decades, is a number with no real-world meaning. But it seems fair as a basis for comparison.
Some say it's unfair to judge a president's or a Congress' economic performance based on the years of actual power, since policies need time to take effect.
So what happens if we credit all economic results to the president or Congress one year earlier? It seems to have no effect in evaluating the parties' presidents, but it gives Republicans virtually a clean sweep in terms of who runs a better Congress.
There are also some who believe fervently that the world began in 1981 and that the performance of Republican politicians before Ronald Reagan is about as relevant as the performance of the old Roman Senate in judging Republicans today.
So as a bonus, a separate spreadsheet considers just the years 1981 through 2006 (and yet another considers those years with the one-year policy lag thrown in). The results are gratifying: Republican presidents swing from victory to defeat in government spending and move from a tie to a definitive loss in the all-important economic growth.
The only category they recapture is civilian government employees. On the congressional level, they swing from victory to defeat on taxes - meaning that Democrat-controlled Congresses since 1981 have a better record than Republican ones of holding down federal taxes.
Michael Kinsley is a social commentator who lives in Seattle. His e-mail is email@example.com.