Here’s two things you thought were true but aren’t:
The unemployment rate for 2006 was 4.6%
E.R.A. is nearly a complete measure of a pitcher’s effectiveness
Let’s break down the unemployment rate first.
The U.S. unemployment rate is constantly cited as an economic indicator and a consistently low rate is seen as proof of the efficiency of the market based economy. There’s only one problem: it does not measure the prevalence of joblessness in a community. There are two reasons the U.S. unemployment rate severely under-represents precisely what is claims to represent.
First and foremost, the way in which the stat is calculated is inconsistent with its title. It does not include those who are jobless but have not applied for or exhausted their benefits. The unemployment rate is a measure of what percentage of the population is actively seeking work. Our European counterparts do not suffer from higher unemployment rates because they are too foolish to adopt the hyper-efficient neo-classical economic model that we do. They post higher rates because they define jobless and unemployed to mean the same thing.
The other reason why our unemployment rate is disingenuously low is that we use the legal system as a cruel step-sister surrogate to our social welfare system. We incarcerate our citizens at 5 to 8 times the rate of most industrialized nations, according to the Sentencing Project, a rate that increased dramatically during the 1990s. Our prison population dwarfs that of any other Westernized nation.
Good research on the subject estimates that more than one in three inmates were unemployed at the time of incarceration. Hundreds of thousands of our unemployed are uncounted, hidden in our penal system, many for petty offenses and low-level drug crimes. The impact that the penal system has on joblessness is manifold. Upon release, the stigma of incarceration becomes a powerful deterrent to potential employers, which leads individuals back toward illegal behavior. Joblessness leads to incarceration which leads back to joblessness. Not collecting unemployment while in jail, lacking the immediate work history to qualify for benefits upon release, and often landing back in prison, a prisoner may go uncounted in the unemployed population for years.
A study by Bruce Western and Katherine Beckett published in the January 1999 edition of the American Journal of Sociology estimated that when the U.S. unemployment rate was readjusted for our imprisoned population, we fare no better than industrialized European nations in terms of providing work for our citizens. In fact, the study showed that the prison boom negated the job creation of the economic upturn in the 1990s, and then some.
“U.S. employment performance looks weaker once the size of the prison population is taken into account… The modified estimate suggests that unemployment in the economically buoyant period in the mid 1990s is about 8% – higher than any conventional U.S. unemployment rate since the recession of the early 1980s.”
We also don’t count the underemployed, involuntary part-time workers, involuntary early retirees, those with disability who would like to work but are not working, or those who chose to return to school because of the job market.
While the unemployment rate doesn’t measure enough to accurately portray what it claims to, a pitcher’s earned run average (ERA) measures many confounding variables beyond a pitcher’s control. It is also a precarious statistic that has garnered far more trust from its general community than it has earned.
A pitcher’s ERA is highly dependent on the defense that plays behind him. Voros McCracken’s research has shown that the percentages of balls in play that become hits are surprisingly consistent from pitcher to pitcher. Batting Average on Balls in Play (BABIP) is about .300 for virtually all pitchers over the course of their careers.
To understand that statistic, let’s first note that in any given at-bat, there are three possible outcomes that do not involve the defense: the homerun; the strikeout; and the walk. All other outs must be recorded by players other than the pitcher (except for the small percentage of balls hit to the pitcher himself). You will often hear baseball men refer to a particular pitcher’s ability to “induce weak contact,” that is, his ability to get hitters to hit weak ground balls or pop ups that wind up as outs. Such a pattern has rarely been documented as an attributable skill. BABIP measures the batting average in at-bats that do not result in strikeouts or homeruns. Some of the absolute best pitchers of all time have shown ability (at least in their primes) to consistently post below-league average BABIPs, like Pedro Martinez and Mariano Rivera, but they are rare exceptions (and future first-ballot Hall of Famers).
Take a look Roy Oswalt and Ramon Ortiz. Oswalt is consistently one of the best pitchers in the game while Ramon Ortiz has been a below average pitcher for most of his career. Yet their BABIPs are very similar. When you look at the defensive independent numbers though, a huge difference is apparent. Approximately speaking, Oswalt, on a per nine inning basis, strikes out 50% more, walks 33% fewer, and gives up homeruns half as often. This is the difference between an annual Cy Young candidate and a retread journeyman.
The differences in ERA between pitchers who do these three things at similar rates are attributable to two things, the defenses behind them and luck. Sometimes stellar defense and/or random fluctuation allows a pitcher to defy BABIP when balls are hit within the range of fielders at an unusually high rate or an exceptional set of defenders are able to field more balls than normal, especially when these events happen to occur in prime scoring situations. But only a handful of the all time greats can do so regularly, and with differing casts accompanying them.
I like to say that most things that happen on a baseball field that fans traditionally chalk up to luck are really just the random variation expected to occur in outcome over a large sample size. Whether those random variations happen at particularly high leverage points in the game is the luck. That is to say that a ground ball that scoots through, two inches beyond the shortstop’s glove is not luck at all, though the batter had no control over whether that ball was two inches to the left or right. What is luck is if that event happened to occur with the bases loaded or with two outs and nobody on. Of course, the pitcher would have not done his job any better had that ball been hit two inches to the left, or if there was nobody on and the runner was left stranded. His ERA would reflect failure though, even if the reason the shortstop didn’t get to the ball was because he got a bad jump, misread the trajectory or has atrocious range up the middle (cough, Derek Jeter, cough).
ERA alone is not a reliable predictor of year to year success; it is too subject to fluke strings of events and the performance of one’s defense. ERA measures a variety of things, but none of them particularly precisely. The reading it gives is partially driven by skill, partially driven by luck and partially driven by the skill of those whom it is not measuring. Sometimes a confluence of circumstances will create a perfect storm that plays to a pitcher’s tendencies. Brandon Webb’s Cy Young performance last year coincided with Arizona improving its infield defense drastically from the previous year, including the signing of Orlando Hudson (arguably the best defensive 2B in the game). Webb’s balls in play are ground balls more frequently than any other Major League starter.
This doesn’t even begin to address the differences in home ballpark, quality of opposition, not having to face your own team’s offense, the impact of unearned runs, relievers inheriting runners already on base (who are charged to the starter if they score) and the myriad other issues that impact ERA.
To get an accurate picture, one has to look at the core components of performance, components controlled solely by the person whose performance the stat indicates. So, while some people marvel at the phenomenal start of 2007 for sub par pitchers like Jason Marquis and point to his tiny ERA, his BABIP is 70 points lower than he has posted at any point in his career. Mr. Marquis is simply benefiting from an anomalous string of events taking place over a comparatively small sample size. It will not last.
In the world of baseball, there is a burgeoning movement to analyze the game from a more scientific perspective and really test the validity of long-standing assumptions. I don’t see that as in mainstream political and economic rhetoric. The dramatic rise in incarceration rates has coincided with declining crime rates. This should be a big story. Either there is some egregious error in the way crime is being reported or our practices of incarceration have to do with things more profound than simple law breaking.
Michael Lewis’s Moneyball has become one of the most influential sports books ever written. In it, Oakland A’s general manager, Billy Beane, talks about how much money is invested in the game of baseball and how big the stakes of winning are. One of the themes of the book is how preposterous it would be to entrust the making of such huge financial and strategic decisions simply to the hunches and gut feelings of coaches, managers and GMs, yet teams do just this – all the time. Beane adopted a scientific approach to the game that was expounded by a marginalized group of niche baseball students. This epistemological approach to baseball is called sabermetrics and it is not new, but, since the book’s appearance, perhaps only porn has grown more as a result of the internet. Meanwhile, the Oakland A’s compete with the big boys every year with a fraction of the budget and seem to have an endless supply of talented young players, especially pitchers.
The Oakland A’s rarely sacrifice bunt, they rarely steal bases. Run expectancy matrices calculated over thousands of games illustrate mathematical truths that contradict accepted baseball wisdom. Trading an out for a base advanced worsens your chances of scoring runs. Getting caught stealing is anywhere from two to three times as detrimental as a successful steal is beneficial (depending on the overall offensive context of the game). Still, small ball is praised by the old guard who lament the lost art of the bunt, ad nauseam. Of course there are many variables and under different situations and with different players certain strategies become more viable than others. But isn’t it at least beneficial for the managers and commentators to know the base from which deviations occur?
One of FireJoeMorgan’s favorite quotes is from Ron Gardenhire, manager of the Minnesota Twins, who claimed that their second baseman, Luis Castillo was worth 15 extra wins to their team. Well there are stats that try to determine these things. According to WARP (Wins Over Replacement Player), Barry Bonds was worth about 15 extra wins in 2004, when he set the single season record for on-base percentage and had the fourth highest single season slugging percentage of all time! This is like an economic analyst saying that Google is worth a “shitload of cash” or “a bazillion dollars.” Of course, these mathematically driven total player metrics have their problems as well and there is no shortage of debate about how to make them more accurate.
We’ve seen only half-hearted pop-culturish attempts to look beyond generally accepted assumptions in an attempt to analyze things at a much more elemental level when it comes to paradigm shifts, social trends our own decision making processes—books like Malcolm Gladwell’s Blink and Levitt and Dubner’s Freakonomics. Of course, I’m speaking outside the realm of academia.
This type of insight is threatening to mainstream pundits. Many of ESPN’s baseball analysts outright deny the veracity of anything sabermetric, often using generic platitudes like, “games aren’t played by computers.” They are understandably defensive; the iconoclast ideas presented threaten to expose those batting average worshippers and NASDAQ watchers as naked emperors.
Joe Morgan, ironically, takes every opportunity to bash sabermetrics and Lewis’s book, though he’ll freely admit that he never read it. He says that he’s played the game and there is nothing about it that a computer could teach him. I say it is ironic because it is the sabermetric interpretation of statistical analysis that transforms Morgan from the lower-tier Hall of Famer the public sees him as, into, arguably, the greatest 2B of all time.
You’ll hear about the prison system as an ironfisted surrogate for the social welfare system on CNN as often as you’ll see BABIP listed in the New York Post. Just further proof why you shouldn’t be getting your information about things you care about from either.
Back in March, Meta had a post, “The housing market: everything you know is wrong” (by which he meant, everything he thought he knew was wrong) that looked at a surprising graph of housing prices. The post remains stubbornly popular, having been viewed on all but 11 of the past 79 days, perhaps a tribute the willingness of visitors to this site to have their everyday assumptions challenged.
We live in a data-rich world and economists and social scientists are just getting used to using the vast resources of computing power and cheap data storage available to us. Using the data effectively often takes great cleverness and imagination. The greater difficulty, however, may be to get people used to the idea that the things everyone knows can now be verified—and all too often falsified.