With the Super Bowl matchup set, serious fans will be poring over all kinds of statistics for clues as to who will claim the Lombardi Trophy on Feb. 5. But a RiseSmart analysis shows that one of the most accurate predictors for the past two decades has come from an unlikely source: U.S. Bureau of Labor Statistics unemployment data.
The team whose metropolitan area boasts the lower unemployment rate during the previous calendar year has won 17 of the past 20 Super Bowls – a remarkable 85 percent success rate. Based on this correlation, the New England Patriots should claim the NFL championship over the New York Giants. Through November, the 2011 unemployment rate for the Boston metropolitan area was 6.8 percent, compared to 8.5 percent for the New York metropolitan area.
On January 26, 1992, the Washington Redskins defeated the Buffalo Bills in Super Bowl XXVI; that year, the Washington, D.C. metro area’s unemployment rate of 4.6 percent was substantially lower than Buffalo’s 7.2 percent. So began the string in which 17 out of 20 times, the Super Bowl winning city had a lower unemployment rate than that of the losing hometown. The predictor has been correct in the past three championship games, including Super Bowl XLV, in which Green Bay (7.7 percent 2010 unemployment) defeated Pittsburgh (8.0 percent).
Other facts of note:
On the seven previous occasions that both teams’ metro areas have had unemployment greater than 5.5 percent – as is the case this year — the team from the metro area with the lower jobless rate has won in every instance.
During the five previous occasions when at least one team represented a metro area with 7+ percent unemployment – as is the case this year, with the New York Giants – the team with higher unemployment lost in every instance.
The Giants’ upset victory over New England in Super Bowl XLII, when the Patriots entered the game undefeated, represents one of the three times in the past two decades when the unemployment rate predictor failed to predict the outcome of the game.
Correlation does not imply causation, of course. But the data is interesting, to say the least.