Competitive Balance and the CBA

“This is a process that in the past has been tortured, troubled, difficult. It’s a very long time since we’ve seen a collective bargaining agreement without a stoppage in play. This could not have been accomplished without a lot of effort by everyone.” – Donald Fehr, head of the MLB Players Association, after signing the last CBA, in August of 2002

“I’m very, very happy to report that we’ve reached a historic agreement. There are a lot of people who didn’t think they’d live long enough to see these two parties ever reach a deal without the loss of a single game.” – Bud Selig, same day

The 2006 season is less than two months from opening day. While that’s a great relief to those of us who don’t feel right unless there’s a baseball game being played, it also means that we’re drawing closer to the negotiations for baseball’s next collective bargaining agreement (CBA) as the current CBA expires on Dec. 19, 2006.

I have to admit that the business of baseball is not my usual bailiwick, since Maury Brown and Brian Borawski do an excellent job here on THT of covering the topic. However, this week I wanted to expand on a small part of Maury’s excellent article in The Hardball Times Baseball Annual, titled “Crystal Ball: The 2006 CBA and the Battles Within It.”

A Step Forward

As Maury noted in his article, since Marvin Miller became the head of the MLBPA in 1965, there have been eight work stoppages, totaling 366 days and 1,719 games lost.

The good news is that baseball avoided another work stoppage when negotiating the last agreement signed in August of 2002, which retroactively took effect at the start of the 2002 season. During that negotiation the owners and Bud Selig were armed with recommendations geared toward improving competitive balance from their “blue ribbon economic panel” featuring former senator George Mitchell and conservative columnist George Will. Two of the major topics on which the owners (at least some of the owners anyway) ended up scoring victories were increased revenue sharing and the creation of a luxury tax aimed at high payroll teams (that is, George Steinbrenner’s Yankees).

Under that agreement local revenue sharing increased from 20% of net local revenues (local revenue less ballpark expenses) under the 1997-2001 agreement to 34%. That money was put into a pool and redistributed equally to all 30 teams. In addition, revenue sharing from a central fund was introduced using a formula, where payers into the local revenue sharing plan were assessed a fee, which was calculated by dividing $72.2 million (based on 2001 revenue figures) by the total net local revenues of all payer clubs and then multiplying that number by a payer club’s net local revenues. That pool of money is then redistributed based on the distance each of the recipient teams are from the average.

The second victory for smaller market teams was the implementation of a stronger luxury tax beginning in 2003, whereby teams whose payrolls exceeded a maximum would face increasing penalties for each offense. The Yankees were payers into that system all three years, the Red Sox paid in 2004 and 2005, and the Angels in 2004. However, the approximately $72 million dollars collected over the last three years ($66 million of which was provided by the Yankees) is not divided amongst the teams, but rather allocated to player benefits (50%), the industry growth fund (25%), and developing players in countries lacking organized baseball (25%).

But despite this, Maury cuts to the chase and asks in his article:

“Does revenue sharing really create parity? Or more precisely, is the current revenue sharing system working to allow the franchises with lower payrolls a chance to get into the postseason?”

Although Maury notes the correlation of playoff appearances and payroll ranking, I wanted to go a step further and break down payroll ranking in a more rigorous way.

Payroll Balance Over Time

To look at this question, I calculated team payrolls using the Lahman database and then grouped the results by year and according to the last three CBAs; the 1990-1996 agreement, which, of course, actually expired at the end of 1994, leading to the strike and lockout, the 1997-2001 agreement, which included greatly increased revenue sharing and a smaller luxury tax on the five highest payroll teams, and the 2002-2005 agreement. It should be noted that since I used the Lahman database, which contains individual player salaries, not every player on the 40-man rosters was included.

First, let’s take a look at how payrolls have changed over time, starting with the first year of the 1990-1996 CBA. The following graph shows the mean payroll (in millions of dollars), the standard deviation (SD) of payroll and the coefficient of variation (CV). CV is calculated by dividing the standard deviation by the mean, and is a measure of the dispersion or statistical variability of a population, and is useful for comparing distributions with different means. CV uses the Y-axis on the right hand side of the graph, while the other two measures use the axis on the left.


What this shows is a steady linear increase in average payroll, coupled with an overall linear increase in the variation amongst teams, as measured by the coefficient of variation. It’s interesting that the increasing gap between the small and large market teams was tempered briefly between 1992-1994, and between 2000-2002, before continuing it’s upward trend.

This increasing gap between the haves and have-nots can also be seen by looking at the average payroll, standard deviation, coefficient of variation, and minimum and maximum payrolls under each of the last three CBAs. In addition, we’ll throw in a measure I call Normalized Payroll (NPayroll), calculated as a team’s payroll divided by the average for that year. The table includes the both the minimum and maximum NPayroll during the period.

Period              AVG      SD     MIN     MAX   NP-MAX  NP-MIN      CV
1990-1996          29.5    10.3      9.4    54.5    1.85    0.32   0.350
1997-2001          50.9    21.3     10.6   112.3    2.21    0.21   0.418
2002-2005          70.1    29.7     19.6   208.3    2.86    0.28   0.424

This table illustrates the same trend, as CV increases along with the gap between the minimum and maximum payrolls, as well as the gap between the normalized minimum and maximum. Under the 1990-1996 CBA, the highest payroll team spent 185% of average, while the lowest spent 32% of average. That gap widened to 221% and 21% under the 1997-2001 agreement, and then to 286% and 28% under the most recent agreement. The largest gap was of course in 2005, when the Tampa Bay Devil Rays came in with a payroll of around $30 million, while the Yankees were at $208 million—almost a seven-fold difference.

Using Recurrent Neural Networks to Predict Player Performance
Technology is rapidly advancing possibilities in decision-making.

So payrolls are rising and the gap is widening. So what? Does that necessarily mean that competitive balance is decreasing?

Well, to answer that question, I took a look at the how strong the link is between winning percentage and payroll with the results in the following table.

Period                r
1990-1996          0.28
1997-2001          0.40
2002-2005          0.46

This demonstrates that indeed the correlation coefficient r (the measure of the linear relationship between two variables, in this case winning percentage and payroll, where 1 is a perfect positive correlation and -1 a perfect negative one) has increased during each successive CBA. In other words, with each successive agreement payroll has become a stronger predictor of on-field success. That’s of the course the inverse of increasing competitive balance.

For those thinking that perhaps the correlation over the last four years was the result of two teams—the Yankees and the ownerless Expos/Nationals—I removed both from the analysis and the correlation coefficient for the most recent CBA was still .42, higher than in the 1997-2001 period.

A scatterplot of the period 2002-2005 with the Yankees and Expos/Nationals put back in will give you a feel for how strong that relationship is. The trend line shows the linear relationship and the R-squared of 0.21 is a measure of how much of the variability in winning percentage is accounted for by differences in payroll.


While an r of .46 wouldn’t be considered a strong correlation by statisticians, it does have some interesting effects when you look at payrolls in association with division ranking. The following tables show the division ranking by winning percentage, the average NPayroll for that rank, and the minimum and maximum NPayroll under each of the last three agreements.

1990-1996   Teams NPayroll  NP-MAX  NP-MIN
1               34    1.13    1.59    0.58
2               35    1.05    1.59    0.48
3               34    0.96    1.38    0.62
4               33    0.97    1.58    0.45
5               28    0.91    1.49    0.30
6               16    0.88    1.45    0.32
7               10    1.03    1.41    0.74

1               30    1.26    1.74    0.56
2               30    1.04    1.68    0.27
3               30    1.00    1.67    0.36
4               30    0.88    1.70    0.25
5               24    0.79    1.24    0.30
6                4    0.89    1.25    0.35

1               24    1.32    2.86    0.59
2               24    1.09    1.84    0.57
3               25    0.92    1.41    0.50
4               23    0.94    1.56    0.43
5               20    0.76    1.65    0.28
6                4    0.56    0.75    0.40

As you can see, teams that finished first in their division have gone from spending 13% more than average on payroll during the 1990-1996 agreement to 32% more during the last four years. You’ll also notice that there is an almost perfect correlation between where a team finishes in their division and their payroll. Furthermore, the gap between first and fifth place teams has grown with each successive period, from 22%, to 47%, to 56%. The min NPayroll column, however, does show that despite these increasing differences, teams spending around 60% of the average still occasionally win their division (Minnesota and Oakland in 2002—the two teams that seem to regularly buck the trend—along with San Diego in 2005, accounted for the only other three division titles by teams that had NPayrolls of under 1.03).

Once again, I took out the Yankees and Expos/Nationals in the most recent period, and the teams that ranked first in their division still had the highest NPayroll at 1.11, just slightly higher than second place teams.

But of course it’s all about getting to the postseason, so let’s take a quick look at teams that made the postseason fared versus those who didn’t.

Period      Post Season   Teams  NPayroll
1990-1996       No          158     0.97
               Yes           32     1.16
1997-2001       No          108     0.91
               Yes           40     1.25
2002-2005       No           88     0.89
               Yes           32     1.30

Despite the addition of the wild card in 1995, which allows 27% of the teams to make the playoffs, the gap in payroll between teams that make the postseason and those who do not has also increased during each successive agreement. The change between 1990-1996 and the later periods would be more pronounced, but just 17% of the teams made the playoffs from 1990-1996, which included five years without the wild card and one year without any playoffs.

Let’s Get Ready to Rumble

So where does all of this leave us?

Maury offers some hope that cooler heads will prevail, noting that the revenue picture for all teams has brightened, with a new ESPN deal signed in 2005, the impending (perhaps that’s too strong a word?) sale of the Nationals, and the increased valuation of Major League Baseball Advanced Media (MLBAM). After all, a rising tide lifts all boats.

However, he also points out, and I would agree based on what I’ve illustrated here, that despite the steps made during the last agreement, competitive balance doesn’t appear to be increasing; instead the widening gap in payroll has translated to widening gaps on the field. As a result, he thinks it likely that smaller market teams will use this fact as leverage in the upcoming negotiations to lobby for increased revenue sharing.

And that means there may be battles looming and we baseball fans, as we have since 1965, should prepare ourselves.

Print This Post

Comments are closed.