How to explain these two discrepancies between BLS OES wage data and salaries of pro athletes?

I refer to the Occupational Employment Statistics published by the US Bureau of Labor Statistics, May 2015 edition.

Under occupation title "Athletes and Sports Competitors" (27-2021), we have the following figures:

Total employment: 11,710.

Mean annual wage: $\$80,490$. Annual 75th percentile wage:$\$89,780$.

The first two of the above statistics imply that total wages paid out to these 11,710 workers was 11,710 × $\$80,490$=$\$942,537,900$ or a bit under $1b. But there is a discrepancy between the above figures and the salary figures across the four major US professional sports leagues. The following are the latest annual salary figures from various sources (these shouldn't be too inconsistent from salaries a year or two earlier). MLB: 862 players, of which all 862 make at least$\$507,500$. Total salaries = $\$3,781,920,218$. NHL: 1719 players, of which all 1719 make at least$\$117,300$. Total salaries = $\$3,524,086,505$. NBA: 545 players, of which 505 make at least$\$100,000$. Total salaries = $\$2,942,055,947$. NHL: 767 players, of which 766 make at least$\$216,666$. Total salaries = $\$2,142,999,943$. So there are at least two discrepancies: 1. The BLS says that the total annual wages paid to 11,170 "Athletes and Sports Competitors" was less than$\$1$b.

But in the four major professional sports leagues alone, over $\$12$b is paid out to 3,893 players. 2. The BLS says that the athlete at the 25th percentile (out of 11,170 workers) makes only$\$89,780$. This means that workers #1 — #2927 make $\$89,780$or more, while workers #2928 — #11,170 make$\$89,780$.

But in the four major professional sports leagues alone, 3,852 players make at least $\$100,000$. So far I've thought about (and ruled out) two possible explanations for the above discrepancies: • Top-coding of salaries. This might explain the low estimate for mean annual wage. However, this would fail to explain why the 75%-ile wage is listed at only$\$89,780$.

• "Athletes and Sports Competitors" does not include professional athletes.

But this is not consistent with this BLS webpage description of this category of work. There is a photo of a professional-looking baseball player and also this sentence: "Competition for most professional athlete jobs will remain very strong."

Another possible occupation title under which professional sports athletes might fall is "Entertainers and Performers, Sports and Related Workers, All Other" (27-2090, 27-2099), but here again we run into the same trouble. There are only 14,570 workers in this category and the 90th percentile hourly wage is only $\$36.35$. I'm unable to think of any other explanation. I'd appreciate help from anyone who is more familiar with how the BLS OES are collected and calculated. All data and sources used above are collected here. 2 Answers This may not be the complete answer but I believe this is a major issue: It is extremely difficult to get famous and rich people to participate in surveys. They have handlers that prevent surveyor access. They have multiple phones numbers and homes that makes it difficult to find them. They also have complex affairs that may make it difficult for them to easily summarize. For example, having a complete picture of sponsorship income can be challenging and almost no one knows what a fair value is for their accumulated pension. In summary, I assume that most of professional athletes are not covered by the BLS survey. Instead, it is people of more modest income, like minor league athletes and golf pros that are actually getting picked up. The second discrepancy is most likely a coverage/sampling issue. Actually, as I think about it, this may explain the first discrepancy as well. The OES is a survey of establishments (i.e. businesses). After a little googling, it seems that the NAICS code for professional sports teams is 711211. There are 1000 of these in the US. The 4 major sports are, what, like 120? OES samples 1 in 6 establishments over 3 years but weights probability of being sampled according to number of employees, so it's plausible that 0 or 1 major league sports teams make it into the sample. Again to the coverage issue, I'm curious that there are only 12K professional sportsmen in the US. That seems too low. Also, 767 players does not seem like enough for the NHL. The active roster for each game is 20 players per team. In addition, there are normally several "healthy scratches" whom I think get paid. There are 30 teams. So, just considering a single game, there are over 650 guys getting paid. There is a lot of movement between NHL and minors in a year, so I doubt there are only 767 guys getting paid by the NHL in a single year. It has to be more. Maybe your source considers guys who play most of their games in the minors to be non-NHL? Maybe it ignores all the guys making league minimum? Yet another coverage issue you might worry about is whether and which sportsmen are considered employees and which are considered business-for-self. I assume professional tennis players, for example, are self-employed. OES does not cover the self-employed, and I don't really know which professional sportsmen are technically self-employed. Ignoring the sampling/coverage issue, the first discrepancy is very likely due to top-coding. In the OES survey instrument, wage data are collected in ranges. There is a top range. In the (slightly outdated) form I linked above, the top range is \$208K and above annual salary. BLS fills in an average salary for that top range, but the average salary is calculated using a different survey (the NCS), and the average-over-the-topcode is for the entire employed population. This is going to give you a much too low average-over-the-topcode for professional sportsmen.

You mention that you don't think this is the explanation because the 75th %ile is only \$89K. For baseball and hockey (at least), there are a lot of minor league players. There are a bunch in soccer, too. There are goofy professional sports, too: lacrosse, roller hockey, semi-pro football, etc. These are included, I assume, in the BLS numbers. Especially in the lower rungs of competition, salaries are very low. So, 1) there are a lot more professional sportsmen than you might initially think, and 2) most of them make small salaries. So, 89K is not surprisingly low for the 75th %ile IMHO.

Another problem relative to MLB and NHL is how players who move back and forth from minor to major leagues (and among different tiers of minor leagues) are treated. In the NHL, these players are typically paid a pro-rata share of their minor league salary or the NHL minimum salary depending on how many games they play in each level. MLB I don't know.

Anyway, the OES is definitely not designed to capture boutique occupations like NFL player, and it is not surprising that it does a poor job of it.

Finally, don't be afraid to call up the BLS, ask to speak with someone who works on the OES, and ask them your questions. You might get lucky.