Academic journal article
By Nero, Peter
American Economist , Vol. 45, No. 2
Peter Nero (*)
Analysis of sports data provides a useful application of econometric techniques because the data tend to be free of measurement error. In particular, player productivity measures are well defined. The economics literature contains many applications with a focus on production and costs (e.g. Ruggiero et al., 1996), and on wage equations in labor economics (e.g Slottje et al., 1994 and Scully, 1974). Most empirical sports analyses have used Major League Baseball (MLB) statistical data to analyze these economic relations.
Professional golf, specifically the PGA Tour, also provides useful data for empirical analyses but has been largely ignored in the economics literature. One notable exception is the paper by Moy and Liaw (1998), which derived a wage equation using performance variables including driving distance, driving accuracy, the number of greens in regulation, sand save percentage and putting average. Other variables considered by fans include descriptive statistics, like the most popular golf glove on The Tour. At each tournament during the season, a golfer's equipment is scrutinized as closely as his performance on the course. Fans and manufacturers alike wish to know what ball Greg Norman is playing, what shirt Fred Couples is wearing, and what driver gives Tiger Woods his distance off the tee. However, an educated golfer knows that talent, not equipment or clothing, is what separates a tour player from an average player. To fans, it appears as though professional golfers are all excellent golfers.
However, the fact that players do not earn the same amount during the course of the season undercuts the belief that all players have equal talent. In this paper, golfer earnings are examined using regression analysis to determine which of the performance variables have the greatest impact. Also, the improvements in a player's game that would translate into greatest increases in earnings are determined. Finally, player efficiency is measured using the wage equation and the affect of the inefficiency on wages is identified. In the next section, the data and summary statistics are presented and regression analysis is used to estimate the earnings equation.
II. Empirical Analysis
Data for this study were obtained from the Golf World Special Annual Issue (December 13, 1996 pages A8-A9). Data were selected for the top 130 golfers based on earnings for 1996. Four of the players had to be eliminated from the analysis because of a lack of data: Colin Montgomerie, Jay Don Blake, Payne Stewart, and Tiger Woods. The following statistics were used in the analysis:
EARN: Earnings--The amount of prize money won.
DRIVE: Driving Distance--The average number of yards of a drive.
FAIR: Fairways Hit--The percentage of drives landing in the fairway.
PUTT: Putting--The putting average per green.
SAND: Sand Saves--The percentage of times a player uses at most two shots to score from a greenside sand trap.
STARTS: Starts--The number of tournaments played in 1996.
Descriptive Statistics are presented in Table 1. Interestingly, there tends to be relative consistency among the performance variables: the standard deviations on DRIVE, FAIR, PUTT, SAND and STARTS are all relatively low. For example, the average number of putts was 1.79 and ranged from 1.71 to 1.83. There was wide variation in the earnings however; the mean tour earnings was $432,835 with a range from $123,100 to $1,780,159. This implies that small improvements in performance should lead to large gains in earnings.
To determine the impact of the performance variables on earnings, the following regression was estimated:
Ln(EARN) = [alpha] + [[beta].sub.1]DRIVE + [[beta].sub.2]FAIR + [[beta].sub.3]PUTT + [[beta].sub.4]SAND + [[beta].sub.5]STARTS + [epsilon] (1)
Following Slottje et al. (1994), a semi-log equation was selected. …