No Panaceas

Friday, March 21, 2003
More on Rally Effects: Yesterday I showed some data on rally events and talked about the implications of the ongoing war for President Bush's approval ratings. What does it matter? Within political science there is some controversy about the impact of approval on presidential power. High approval ratings obviously do not weaken a president, but do they strengthen a president?

I think the answer is yes, but it is a qualified yes. High approval ratings can help you get more members of your party elected to Congress (e.g., 2002); obviously it can help you get re-elected; it might give you a freer hand in making foreign policy because Congress is more afraid to criticize you; it can help you hold your party-members in line and even successfully pressure members of the other party who come from competitive seats; and it can help you enlist public support in favor of legislation or foreign policy endeavors. In short, approval is a resource that helps presidents get what they want.

Much depends on context, however. For example, high approval ratings will not help you pass legislation if the other party controls Congress, especially if the other party is large and reasonably homogeneous. The current President Bush enjoys unified government. Consequently, he should be able to parlay his approval ratings into policy. He will have to choose his fights selectively, but Congress will have greater trouble resisting him on tax cuts, ANWR, and confirmation battles.

At least for a while. The good news for the Republicans is that Bush will be more powerful. But he should use it while he has it because the good news for the Democrats is the approval ratings will wear off quickly. The effects of war are likely to go away very fast. After that, barring some other major foreign policy battle, such as North Korea, Bush’s approval will depend much on the economy. For better or worse, fair or not, the economy, and not foreign policy – even if that foreign policy is largely viewed as a success – will not dictate Bush’s approval levels by this time next year. Let’s look at the first President Bush’s approval levels extending from the first Gulf War to the 1992 election.

As with yesterday the data is from various Gallup publications. Here I show Bush's approval levels for his complete administration. Just prior to the war's start he was at 54%. (The earlier rally effect was from the Iraqi invasion of Kuwait.) The Gulf War rally wore off quickly and a worsening economy eventually wrecked Bush’s approval levels. By October 1992 his approval was 34% and he was done. He won the war but lost the election. (Note the rebound after the election. Sympathy? An improving economy?)

Will we see a repeat? Obviously it depends a great deal on what happens to the economy over the next year and a half. (I know better than to try and predict the economy.) What makes much of this unfair is that in reality presidents do not affect the economy much, at least not in any kind of meaningful short-run. Yet, we blame presidents for poor economies and we credit them for good ones. (Presidents are partially to blame since they always trumpet “their” success when the economy rocks.) It was not George H.W. Bush’s fault that the economy was soft leading up to the 1992 election any more than it was Bill Clinton’s brilliance that boosted the economy in time for the 1996 elections. (On the other hand, perhaps we can credit Al Gore since he invented the internet.)

Thursday, March 20, 2003
Rally Effects Great and Small: The Gallup Organization has produced a national poll on presidential approval almost every month since the Eisenhower administration. Specifically they ask, “Do you approve or disapprove of the way ________ is handling his job as president?” As a consequence we have an extraordinary historical record of presidential approval, and over the years we have learned a lot about what drives approval ratings.

One factor is the so-called Rally Effect. This occurs when the United States is involved in a militarized dispute with another nation. Actual fighting does not have to take place – as with the October Missile Crisis. Rally events have a quite consistent impact on presidential approval. What follows are three notable rally events: the October Missile Crisis, the taking of the hostages in Iran, and the invasion of Panama. The figure shows the movement in presidential approval from just before the event (Month 1) up to the point where approval returns to the pre-event level. (The data come from a variety of Gallup publications.)

This illustrates several quite general properties of rally events. First, they yield a sharp increase in approval. This is a consistent phenomenon. Other examples include Johnson sending troops to the Dominican Republic, the invasion of Granada, the Mayaguez incident, and the 1993 missile attack against Baghdad. Second, the incident does not have to be a U.S. success to yield the effect, e.g., the Iranian Hostage Crisis. Third, the effect is ephemeral. It wears off, usually quite fast. Finally, “bigger” events yield bigger, longer lasting, rally effects. The Panama effect was small (nine points) and was probably gone in a month. The October Missile Crisis effect was larger (fifteen points) and arguably some amount of it lasted for nine months. (I say arguably because other variables are at play, e.g., the economy, that might hasten or inhibit the decay.)

What is going on here? In truth we do not have a rock-solid theoretical explanation of rally events. But what seems to be going on is that an external threat motivates internal cohesion. That is, we rally-around-the-flag against a foreign adversary. It’s us versus them and even if “us” disagree about some things we are united in a fight against “them,” at least for a while.

The two largest rally events in recorded history were September 11th and the first Gulf War. Here are numbers for the younger Bush, beginning just before September 11, 2001 and extending up to early March 2003.

Understandably, it is a huge rally effect and the decay, while persistent, has been quite slow. I think there is still quite a bit of a rally effect lingering from September 11. Generally speaking the economy has the greatest impact on a president’s approval. Chances are if September 11th had not occurred then Bush’s ratings right now would be far lower. It is too bad we don’t have systematic polling from the Pearl Harbor period. I suspect that Franklin Roosevelt’s popularity took a similar leap and it would be interesting to know how long it took for the Pearl Harbor rally to disappear. Certainly the September 11th effect is atypically slow to decay. (Perhaps it would have gone away faster had the U.S. not gone into Afghanistan.)

The point of all of this is to talk about today. What will the war do to the current President Bush’s approval ratings? Let’s take a look at the first Gulf War.

As the figure shows, the elder President Bush’s approval levels jumped up dramatically. What occurred then was a steady decay. I think we are about to see something much like the first Gulf War rally. It will be interesting if the current President Bush’s approval ratings fail to jump upward somewhere in the neighborhood of twenty or so points. But unlike the September 11 rally, this one will decay fast. Real fast. I am guessing that within six months it will be all but gone.

Of course I am making an assumption that the war will be short. That is not a controversial assumption. But just in case you are wondering, John Mueller (War, Presidents and Public Opinion) argued a while back that long wars appear to be very bad for a president’s approval.

What will Gulf War II’s boost and decay mean for the Republicans and Democrats? More later.

I've seen numerous references to this article by Fareed Zakaria of Newsweek, but just now got around to reading it. It's worth reading and remembering.

Tuesday, March 18, 2003
America's Non-Existent Debate: You would think that at this time the United States Congress would be the locus of debate over the pending war against Iraq. There was a time when the Senate, especially, served such a role. Now the role is inadequately filled by a cacophony of yapping t.v. "news" shows. Meanwhile, over the last weeks, and especially the last two days, exceptional debates have gone on in the British and Canadian parliaments. Last night in Britain was the extraordinarily dramatic moment of a Commons packed with MPs riveted to Robin Cook's resignation speech. An impassioned debate is ongoing right now in the British Parliament over a motion to support going to war. Yesterday PM Chretien fielded heated questions from the left and right about Canada's decision to sit the war out. (Hurray for C-Span.) Meanwhile the U.S. House today is considering a variety of relatively minor bills. The Senate is considering the budget resolution and the Estrada nomination. But, of course, we have Hardfire, and Capitol Ball, and Connie Chung asking Michael Jackson whether the war in Iraq will increase or decrease his chances of having more plastic surgery.

Tractor Dude Strikes: The most common observation today in DC is that one guy with a tractor can cause more disruption than protesters and threats of terrorism. Tobacco policies?

Monday, March 17, 2003
March Lameness I: Let’s all hail the mighty University of Northern Iowa. Simple transitivity, the foundation of March Madness, suggests that the 11-17 Panthers are the finest team in men’s college basketball. Who did you think was best? Kentucky? UNI beat Evansville who beat Illinois-Chicago who beat Northwestern who beat Purdue who beat Louisville who beat…you guessed it. What about Arizona? UNI beat Montana who beat Stanford who beat Arizona.

Just about every year someone does an exercise like this, usually it’s done by an alum to show how his or her supposedly poor team is really great. (For the record, Texas A&M beat Missouri who beat Oklahoma. Rice beat Colorado who beat Texas.) But does this mean anything? Let’s go a little farther.

Kentucky beat Florida who beat Florida State who beat Iowa who beat UNI. We can lay all this out as follows: UNI > Evansville > Illinois-Chicago > Northwestern > Purdue > Louisville > Kentucky > Florida > Florida State > Iowa > UNI > Evansville… I can do this with the Arizona line as well, but you get the point. In short, we have intransitivity, a violation of the transitive property if a>b and b>c, then a>c.

Consider a more abstract example. Suppose we have four teams: Red, Blue, Yellow, and Green. Let’s assume the following:
Blue > Red
Yellow > Red
Red > Green
Green > Yellow
Green > Blue
Blue > Yellow

First Tournament
Round 1
Blue v. Red, Blue wins
Green v. Yellow, Green wins

Blue v. Green, Green wins

Second Tournament
Round 1
Red v. Yellow, Yellow wins
Blue v. Green, Green wins

Green v. Yellow, Yellow wins

Third Tournament
Round 1
Blue v. Yellow, Blue wins
Green v. Red, Red wins

Blue v. Red, Blue wins

Three different tournaments with the same teams, but different matchups in the first round yield three different “champions” Green, Yellow, and Blue.

So what? Does this have any bearing to the real world? Well, yes. It turns out that unless one team will beat every single other team in a one-on-one (i.e., pairwise) matchup, then some amount of intransitivity exists, and consequently the tournament winner is a champion of contrivance. A team capable of beating everyone – in voting theory a “Condorcet winner” – almost certainly does not exist in contemporary college basketball. Short of having a massive tournament where every team plays every other team, you can’t prove a Condorcet winner exists, regardless.

But there’s another problem. It’s not just that the tournament falls prey to intransitivity; it’s also badly flawed because of the small-n problem. More later.