Prospectus Today: The Decline

On Monday, the Associated Press wrote the story that no one had yet, reporting that home runs were down 8.8% as compared to the first five weeks of the 2004 season. Their lead was something less than subtle:

In the first year of toughened steroid testing, home runs are down in the major leagues for the first time since 2002.

That line makes the connection, and the rest of the story strings together quotes from players to further establish the idea that the new testing plan and the decline in offense are related.

There’s no arguing the core fact: offense is down, and power is way down. Home runs, slugging and isolated power are all well off from their 2004 season totals. OBP is down, although largely due to a drop in batting averages; walk rates are steady, while strikeout rates have actually dropped. Defensive Efficiency is up slightly.

A completely unscientific analysis of that information could lead to the following hypothesis: players scared by the increased penalties for positive tests have reduced or eliminated their steroid usage, and are hitting for less power because of it. Instead of doubles and homers, they’re hitting flyball outs. Pitchers have also dropped the juice, and with lowered velocity due to that, are striking out fewer batters. Fewer hard balls put in play have made things easier for defenses.

Neat, tidy…and completely ridiculous.

Let’s start with the numbers. In 2003, the first year of steroid testing–survey testing, with no punishment phase–96 players tested positive, per Will Carroll. Per the 2002 Basic Agreement, that was enough to trigger random testing, with penalties, for 2004. In ’04, the number of positive tests dropped to 12. As was well covered last winter, players testing positive in 2004 would not be punished, nor would their names be released. However, in the current environment, it seems unlikely that a notable star’s positive test would have remained a secret for long. This isn’t grand jury testimony.

The low positive rate in 2003 and the lack of news generated by the new testing program in 2004 would seem to have indicated that baseball’s steroid problem had been blown out of proportion. In the winter of ’04-’05, however, the combination of the government’s investigation into the Bay Area Laboratory Co-Operative (BALCO)–and the leaks from it–and the buzz about Jose Canseco‘s book Juiced brought about media attention, and not far behind it, the U.S. House of Representatives.

In that chaotic environment, and ignoring the considerable evidence that MLB did not have an industrywide problem, MLB and the Players Association hastily revised the testing program. In January, the “administrative track” was eliminated, with a first offense now carrying a 10-day suspension. Androstenedione and the notorious THG were added to the list of banned substances.

That was just over four months ago. It was less three months before Opening Day. And in that short a period of time, so many players are to have given back the benefits of their supposed steroid use that it would drive offense down across the league?

Let’s run at this from a different angle. In 2003, the first year of survey testing, runs and power were up over 2002. In 2004, the first year in which positive tests were met with penalties, however weak, offense was up over 2003’s higher numbers. So testing showed a very small number of players using steroids, and despite the testing–which would, by 2005’s logic, lower offense–home runs and runs scored went up.

Do you see where I’m going with this? To tie the reduced offense so far this season to reduced steroid usage in the face of greater penalties for usage and more substances on the banned list, you have to not only ignore the evidence that two years of testing coincided with higher offense, but you have to conclude that the effects of steroid use were so short-lived that three months after the new plan went into place, all the users were weaker.

Sorry, I can’t get there.

Here’s AP’s data:

HR's per game through the season's first five weeks:

Year  Games   HR   HR/G

2005   460   908   1.97
2004   459   990   2.16
2003   461   953   2.07
2002   456   878   1.93
2001   454  1047   2.31
2000   457  1183   2.59
1999   457  1016   2.22
1998   437   862   1.97
1997   398   742   1.86
1996   423   989   2.34

The 8.8% drop from ’04 to ’05 isn’t even the most significant dip we’ve seen in the past ten seasons. The dropoffs from ’96 to ’97 and from ’01 to ’02 were much larger, and well before the era of steroid testing.

Before steroids were the hot-button issue, the primary culprit for changes in offense was the baseball. Any time home runs went up, there would be stories about how the ball was different, harder, smoother, bouncier, not made of Nerf like it was in the good old days. We’re not seeing those stories this year, because there’s another easy explanation in the room. Easy, however, is not necessarily accurate.

What I can’t get away from is that drop between the 2000 and 2001 seasons. If steroids really are, and have been, the critical factor in the current high-offense era, then how do you explain such a significant decline in home runs nearly two years before anyone was asked to pee in a cup? Was everyone cycling down at once? The current decline is about half of that one; is it at all reasonable to point to a proximate cause for it when the ’01 decline goes unexplained?

How about this: offense bounces up and down for not one reason, but for an amalgamation of reasons. Tying the five-week year-over-year decline in home runs to a steroid testing program that’s not as old as some items in my fridge is facile, sensationalistic, and obstructs the effort to find truth.

If all of this hasn’t convinced you to reserve judgment, let me toss out one final data point:

AL:
        AVG  OBP  SLG  RA/G   ISO  HR/9
2004:  .270 .336 .433  5.03  .163  1.15
2005:  .259 .324 .404  4.65  .145  0.99

NL:
        AVG  OBP  SLG  RA/G   ISO  HR/9
2004:  .263 .329 .423  4.69  .160  1.11
2005:  .258 .331 .411  4.63  .153  1.02

While slugging, ISO and HR/9 are down in both leagues, they’re down much more significantly in the AL. Additionally, the NL hasn’t seen much falloff in the other statistics that measure offense. If steroid testing were the determining factor, we would expect to see the two leagues lose offense at a comparable rate.

This is going to be a story all year long. There is a significant segment of the media–and quite probably, the game’s administrators–that would love to be able to point to reduced offense as some evidence that the new new testing program is having an impact. Be wary of these claims, and examine all the data in the context not only of recent seasons, but the game’s history.