There has been a fair amount of discussion on the amount of games for the National Basketball Association (NBA) season as of late. A majority of this discussion has been centered around player health as it relates to the apparent elevated risk from playing in back-to-backs on the road (article here, it’s interesting to note that their analysis is at the season level and not the game level^{1}). Some of this discussion has also revolved around the business of basketball and how having star players rest impacts the NBA’s bottom line. When that happens, it piques my interest and I must say, I’m disappointed with what some people have said regarding this issue.

Some people have suggested that decreasing the number of games will make each game more important and thus driving up interest, allowing the NBA to charge more, and increase revenue. This is pretty much true if we’re talking per-game, that’s called the Law of Demand. The problem is that the NBA is not focused on per-game revenue per se, they care about total revenue. When you reduce the number of games, you’re also reducing your revenue opportunities which becomes problematic because fewer games means lower revenue, all else equal. But all else is not equal, each game generates more revenue than before which means higher revenue. And now we’re at an impasse in determining how total revenue changes due to restricting the number of games. One aspect says total revenue will increase while the other says revenue will decrease. So which is it? And does it matter how much the decrease in number of games is?

Welcome to Econ 101, please take your seat and I’ll try to be brief with this lesson. But it’s economics, so it’ll still sound verbose. (Sorry.)

## NBA Revenue

As far as I can tell with respect to the NBA, there are two main data sources for NBA revenue. One is directly from the NBA and is called Basketball Related Income (BRI, click the link for Larry Coon’s description). This particular measure is defined in the Collective Bargaining Agreement (CBA) and is one of the major components for determining the Salary Cap. BRI attempts to quantify the NBA’s income streams that are due to their players and games but less customary operational costs that are not related to the player salaries. There are certain components of BRI which are fixed in nature such as proceeds from arena signage or team sponsorships. Some components are mostly fixed, such as national and local broadcast rights. And then there are components which vary based upon the number of games played such as regular season and playoff gate receipts, parking, concessions, etc.

Last season, the NBA’s BRI was $5.289 billion dollars and the value is currently projected to be around $6.5 billion. The next two Salary Cap projections (of $102 million and $103 million) imply that the NBA has BRI projections in the neighborhood of $7.2 billion and $7.8 billion^{2}. You can find data on BRI going back to 1998 through Larry Coon’s cbafaq.com except for the 1999 lockout season (determining some of these values might involve some indirect calculations). As a word of caution, this should not be strictly interpreted as revenues but it is close enough for our back of the envelope calculations.

The other data source comes from Forbes’ NBA valuations, which you can find going back to 1990 from Rod Fort. These are the more conventional measures of what a business would consider as their revenue sources. There are naturally some components of revenue which do and do not depend on the number of games played. The amount of fixed sources of revenue will come into play later, but last season Forbes estimated that total NBA revenue was $5.866 billion.

Since the 2000 season when BRI was $2.257 billion, the NBA has averaged 5% annual growth in revenues. Alternatively, Forbes estimated a value of $1.874 billion total NBA revenue in 1998 which implies around 6% annual growth. If we focus on the Forbes gate revenues, the 2000 value was $729 million and increased to $1,490 million in 2016 for about 4% growth. All of these growth rates have been fairly stable with the exception of 1999 and 2012, which coincide with the NBA lockouts and will come in handy later. For a bit of context, inflation was around 2% throughout this time period. In other words, investing in an NBA team in 1998 would have been a good investment.

Data on specific streams of revenue within the NBA are hard to come by. Forbes does produce an estimate for each team’s gate revenues, but only consistently since 2003. Gate revenues have trended towards making up a smaller share of total revenues over time. Forbes estimated gate revenues to take up 50% of all revenues in 1990 and this share declined to 31% in 2000 and 25% in 2016. At the same time, the remaining share of total revenues has not be filled with National TV revenue even though the TV contracts have increased over time. In 2002, national TV revenue made up approximately 26% of BRI per my estimates back when I evaluated the new TV contract (also this and this provide more information on the NBA’s financials that I’ve gone over). This share has declined most every year all the way to less than 20% last year. However, the new TV contract kicks in this year and the estimated $2.1 billion in national TV revenues should make up around 30% of BRI this season. The national TV contract will increase by about 5-6% each year for the next 9 years, so it’s share of total revenues will depend on how the other revenue sources increase over the next 9 years.

As far as describing revenue trends and compositions, we really only need to focus on revenue which is tied directly to the number of games. It’s clearly been declining but still makes up at least 20% of total revenues. And at the same time, it’s clear that new revenue streams have been emerging within the NBA since the 2002s as both gate and TV revenues have decreased their shares. While we know what some of the new revenue streams will be in the future (ie jersey advertisements and e-league), it is still unclear what all these revenue streams will be and how we might project their growth since we don’t have data on them.

So we have an OK picture of the NBA’s revenue. But that is not the main point of this post. The main point is describing how the NBA’s revenues may change due to a decrease (or increase) in number of regular season games played. We need to recognize that we are fundamentally describing the demand for the NBA. And in doing this, we ultimately need to know how sensitive the demand curve is to a change in quantity. So I’ve brought the data, now I need to describe a concept that some may not be familiar with and then finally churn some numbers to answer how a reduction in games may affect the NBA’s bottom line.

## Elasticity of Demand

I think it is reasonable to assume that the NBA, as a whole, is concerned with maximizing their total profits (revenues less costs). This is not to say that the NBA does not care about game integrity/quality or player health, they clearly do at least to the extent that these issues affect their profits. It may even be the case that the NBA has non-profit driven motives on these issues, but this is a difficult proposition to measure and analyze so I will omit this discussion^{3}. Our focus here can now be described as how total profits change due to a change in the number of NBA games in the regular season. If we ignore costs for the moment, this can be rephrased as understanding how total revenue changes due to a change in quantity.

To put this particular issue into economic jargon, we would use the term elasticity of demand. The NBA is interested in how their consumers’ willingness to pay for an NBA game respond to a change in the quantity of NBA games available. This situation is a bit different from how a typical Econ 101 course would set up the concept, usually it is phrased as how responsive the quantity demanded is to a change in price. But price and quantity are tied together, so the order is of limited importance. The business model of the NBA is to set the quantity of games first, and then have price-per-game adjust to their market’s demand.

Anyways, per our set-up we are concerned with how a quantity change will affect total revenue (which is price times quantity). So how would the NBA’s total revenue increase/decrease due to a change in the number of games? Well, this is where the elasticity of demand comes in handy.

The elasticity of demand measures the responsiveness of the demand curve -- ie how much price will change due to a quantity change and vice-versa^{4}. Total revenue is defined as price times quantity, which implies we can determine how much total revenue will change due to a quantity change if we know the elasticity of demand. We already know that if price goes up (a positive change), then the quantity demanded will go down (a negative change). It might be helpful to set down the actually definition of elasticity of demand:

E_d = %changeQ / %changeP

We already know that this particular value will be negative because when the %changeQ (percentage change in quantity) is positive (or negative) then the %changeP (percentage change in price) will be negative (or positive). But what we really want to know is how total revenue changes. Let me describe it like this: if the percentage change in quantity demanded (purchased) is larger than the percentage change in price, then the elasticity of demand will be less than -1 (so -2 or -10, etc. which is called "elastic demand"). This tells us that reducing the price will lead to a larger increase in quantity, which will result in an increase in total revenue. Or in other words, an increase in price will drive a lot more people away than the resulting increase in revenue per-person. Using the same logic, if the elasticity of demand is greater than -1 (so -0.99 or -0.5, etc. which is "inelastic demand"), then the NBA should decrease the number of games.

All we need is the elasticity of demand to figure this out, but how can we determine the elasticity? Well, because we are focused on a quantity change we need to know how a change in quantity affects either price or total revenue. To do this, we need a couple of assumptions, some elbow grease, and ingenuity.

## Making the NBA Lockouts Useful

To determine the elasticity of demand for NBA games, there needs to be some change in the number of games played^{5}. The NBA has played 82 games per each team since the 1967-68 season with the exception of the 1999 season (50 games) and the 2011-12 season (66 games). It’s basically only those two seasons that one can try to determine the elasticity of demand and we can utilize Forbes revenue data with those years to estimate the elasticity of demand. Since BRI data only goes back to 1999, we can only use this as a cross-check for the 2011 lockout.

The 1999 NBA Lockout resulted in a reduction from 82 games to 50 games at a time when there were only 29 teams. This implies a 39% reduction in regular season games from 1,189 to 725, but their playoff games were unaffected^{6}. The 1998 to 2001 playoffs averaged 70.75 games (71, 66, 75, and 71), which we need to add onto the game reduction. That’s approximately a -37% change in games due to the lockout. For the 2011 lockout, the regular season game reduction was from 82 to 66 per 30 teams (1,230 to 990) for a -19.5% change in games but, again, their playoffs were unaffected. The 2010 to 2013 playoff seasons averaged 84 games (86, 81, 84, and 85), which implies a change of approximately -18% games due to the second lockout.

Now comes a bit of an assumption. We know what the revenues for the lockout seasons were -- $1.2 billion and $3.861 billion per Forbes while the BRI in 2012 was $3.375 billion. The problem is we don’t know for certain what the revenues would have been if there were no lockouts. Since we know that total revenues have increased every year since 1990, we can say that the lockouts were expected to have higher revenues than the previous year but not as high as the following year.

As a leap of faith, and to be simplistic, I’ll just assume that the lockout years were expected to have the mid-point of total revenues for the year before and after the lockout. This implies Forbes revenues of $2.095 billion in 1999 and $4.285 in 2012 while a BRI of $4.055 billion plus gate revenues of $1.209 billion per Forbes. If you think my values are too high or too low, then feel free to substitute your own value in this next step.

Since determining the elasticity of demand is our ultimate objective, we need the change in revenue-per-game (ie price) for estimating the elasticity of demand. We know what the total number of games were expected to be in 1999 (1,259.75) and 2012 (1,314), plus what we just assumed to be the expected total revenues. That implies expected prices of $1.66 million in 1999 and $3.26 million in 2012 ($1.208 million for gate revenue and BRI measured as $3.09 million). The actual values for these prices were $1.54 million in 1999 and $3.14 million in 2012 ($0.892 million per gate revenues and $3.43 million per BRI). Now if we take the actual price per game less the expected price per game and divide that by the expected price per game, we get the change in price due to the lockout^{7}. I’ve worked out those values below as a nice point of reference in case you don’t agree with how I calculate percentage changes:

### Lockout Changes

Season | Actual Games | Actual Price | Expected Games | Expected Price |
---|---|---|---|---|

Season | Actual Games | Actual Price | Expected Games | Expected Price |

1999 (Forbes) | 791 | $1.54 million | 1,259.75 | $1.66 million |

2012 (Forbes) | 1,074 | $3.14 million | 1,314 | $3.26 million |

2012 (Forbes Gate) | 1,074 | $1.208 million | 1,314 | $0.892 million |

2012 (BRI) | 1,074 | $3.43 million | 1,314 | $3.14 million |

One thing that is interesting here is that the actual price is only greater than the expected price for the 2012 Forbes Gate and BRI estimates, which would appear to defy the law of demand. That is a bit problematic, but for our back-of-the-envelope calculations let’s look at the elasticity of demand from these three events:

### Estimated Demand

Season | %changeQ | %changeP | E_d |
---|---|---|---|

Season | %changeQ | %changeP | E_d |

1999 (Forbes) | -37% | -7.20% | 5.14 |

2012 (Forbes) | -19.50% | -3.70% | 5.27 |

2012 (Forbes Gate) | -19.50% | 35.40% | -0.55 |

2012 (BRI) | -19.50% | 9.20% | -2.12 |

Well this is certainly interesting. From our rough estimates, we have that gate revenue is inelastic, BRI as inelastic, and Forbes’ revenue is the illusive Giffen good which violates the law of demand. What this would imply is that the NBA should reduce the number of games played if they are concerned with gate revenues. But seeing that this only makes up one-fifth of the league’s revenues, they would likely be more concerned with BRI. Which the data show the NBA should increase the number of games.

This data and the estimated changes should not be taken as infallible, but at the same time the warts of the data should not lead to an immediate dismissal of the results. Instead, we can usually determine in which direction the data issues (and economic issues) that might be present fall towards. And if we know the direction the problems lean towards (ie the price/quantity change is too high/low), then we’ll get a better idea of the elasticity of demand.

## Conclusions

I’ve brought some data to the table on the issue of the optimal number of games in a season. I don’t believe any of the major media outlets covering this topic have done such. If they have, I apologize for missing it but please let me know and I’ll correct this.

But overall, here’s some data. If you disagree with the data, I don’t know what to tell you. Push your angst towards Forbes or the NBA. But for the others, you’ll realize that the idea of reducing the number of NBA games needs some more data to support the movement. Because as of right now, there’s not much evidence to support a decrease in the number of games except for gate revenues, which only makes up 20% of the revenues..

But please, if you’re only advocating a shorter season because it drives your page views then by all means go for it. Or if it supports the opposite side, go for it! I’m only here to try and bring some actual data and insight to the discussion, which seems to be lacking in all of this discussion.

^{1. In layman’s terms, you cannot actually make a claim about elevated risk at the game level. This is the ecological fallacy. The 3.5 times risk claim is not accurate and is a gross overestimation, especially since the number of back-to-backs does not vary much at the team level.↩}

^{2. Albert Nahmad has estimated that the NBA projects these values to be $7.0 billion and $7.6 billion. The discrepancy is based on our assumptions of what the National TV deal pays out each season with his values being lower than mine.↩}

^{3. One might even note that discussing this type of an unmeasurable activity is like counting the number of angels that can dance on the head of a pin. Even if we could determine the degree to which the NBA cares about player health, it won’t invalidate anything else in this article. We’d just have to determine how to combine the two effects.↩}

^{4. Some people reference this as the slope of the demand curve, which is not right. The elasticity of demand is actually the inverse of the slope multiplied by price divided by quantity. It’s still a function of the slope, I guess I get their point...↩}

^{5. The change in number of games played needs to be exogenous from price changes, ie that the change in games played was not due to a change in the demand for basketball games. This is problematic if one believes that a labor dispute affects NBA fan’s willingness to pay.↩}

^{6. In 2003, the NBA converted the first round series from best-of-five series to best-of-seven series.↩}

^{7. There is some disagreement over how one should calculate a percentage change. I’ll leave it up to an intrepid commenter to go through and calculate the other versions of percentage changes and then see how the elasticity of demand changes because of it.↩}