Synopsis: I investigate a mystery about rising conference costs and reach a different conclusion than expected.
A few days ago I was registering for the American Physical Society (APS) March Meeting, the major yearly gathering of Physicists, and noticed that the registration fee is quite a bit higher than it was last year. I made the same observation last year comparing it to the year before, and I updated the graph I had made to illustrate this:
Yes, I have a strange compulsion to make graphs.
The plotted registration fees are for early registration of APS members for the full meeting. The inflation-adjusted numbers come from the CPI calculator.
The rise is remarkable, especially since about 2016. This year’s price is 7.5% higher than last year’s, or 5% higher after adjusted for inflation.
The cost for graduate students has an even sharper increase:
Why has the APS meeting registration fee risen so much, and so incessantly?
I was curious to see if data I could dig up would shed any light on this, specifically on whether any of the following are plausible:
- APS is increasingly making a profit off of the March Meeting, using it to fund other activities.
- The cost of running a meeting — renting a convention center, etc. — have been increasing a lot in recent years.
- Greater attendance leads to to greater cost per person.
I think I can rule out all of these given the data that I’ve got, as illustrated below. Instead, I’m led to another possibility, one that I hadn’t thought of initially.
I should note that this is a pointless investigation: presumably, people at the APS know exactly why prices are what they are, and they’d probably tell me if I figured out who to ask and asked them.
In addition to the meeting registration costs, I wanted the total revenue and cost associated with each meeting, and the attendance. The Annual Reports of the APS, list total meeting-related revenue and expense. This includes meetings other than the March Meeting, such as the April Meeting — yes, there’s some creative nomenclature at work here — but since the March Meeting is much larger than all the others it seems reasonable to use the total meeting values. (The March Meeting has about 10,000 attendees; the April Meeting about 1500.) March Meeting attendance was difficult to determine. The number of attendees was given in each of the 2012-2018 Annual Reports, but the earlier ones simply stated “over 7,000 attendees,” or other such lower-bounds.
The number of people at the APS March Meeting has grown from about 7000 in 2008 to 11000 in 2018:
Not surprisingly, the total meeting revenue and the total expenses of APS meetings have also grown:
The difference between the two is positive but not large — about $270k/year on average — and it isn’t increasing at any appreciable rate:
It therefore doesn’t seem that the APS is using conference revenue as a means to fund a lot of other stuff, at least not more now than in the past. Thus we can cross out possibility #1 above.
How much does the March Meeting cost per attendee? I don’t know, but as an approximation, let’s plot the total meeting expenses (March + other meetings) divided by March Meeting Attendance. This graph surprised me:
It’s pretty flat! (At least in constant dollars.) Inflation-adjusted, the average annual increase is about 0.7 +/- 0.9 % per year, essentially zero. It certainly doesn’t have the large upswing of the March Meeting Registration Fees (1.9 +/- 0.2 %/yr, or 5.1 +/- 0.1 %/yr since 2015).
Resolutions to the Paradox
We therefore have a puzzle:
- The registration fee for APS March Meeting attendees has risen considerably in recent years;
- The per capita cost of running the meeting has remained roughly constant;
- The profit APS Makes on the meeting is small and roughly constant.
How can all of these be true? Here are five possibilities:
- My data or my approximations are flawed. My meeting expenses are summed over all APS meetings, and perhaps the March Meeting is not as dominant as I think. My attendance figures are rough and don’t distinguish between full and partial meeting fees. And so on. I’ve made several simplifications that could be far off from the reality of APS Meeting finances.
- I’m using only the Early Registration Fees to track the cost of the meeting — perhaps it used to be the case that lots of people registered late, essentially subsidizing those who register early. The number of disorganized procrastinators has been declining.
- Similarly, perhaps the fraction of student attendees has risen sharply. It’s clear from the overall expense that students are subsidized; this subsidy must come mostly from non-student attendees.
- An increasing fraction of the cost of running the APS March Meeting comes from attendee registration fees, rather than, for example, commercial exhibitors (vendors).
- I’ve missed something major in my assessment, or my understanding of how the APS and March Meeting work.
I strongly doubt #1 and #2. #5 is plausible. I think #3 and #4 are very plausible, however. I’m especially curious about #4. I’ll admit that I don’t pay much attention to the exhibitors at the March Meeting; I’d be surprised if their number has increased as much as the overall attendance has, leaving a higher fraction of expenses to be picked up by regular conference goers.
I went into this thinking that the rise of conference fees was yet another example of service-related cost increases, as in health care and college tuition, or that it relates to meeting growth, with large conferences costing more (per capita) than small ones. I now don’t think that either of these are the case. My current guess is that attendees are bearing more and more of the meeting cost.
As mentioned at the start, this is a mystery that isn’t a mystery. My hypothesis that APS meeting expenses are increasingly shouldered by academic attendees would be easy to verify or reject by anyone who has access to the actual meeting finances; it’s not some deep secret of the universe.
Moreover, the fact that attendance has grown despite rising fees implies that few people really mind the high cost of registration; demand is fairly inelastic. I wonder how high the price would be before a sizeable number of people decided to skip the March Meeting. This raises another question: Why has attendance increased so much? Who are the “new” people? I would guess that the answer to this is not only known, but discussed in APS newsletters; I’m not going to go digging through them; however.
Though it’s unimportant, I enjoyed this exercise. It reminds me that there’s a lot of data out there, and looking for patterns and meaning in the “small” things we encounter, like prices, is satisfying.
A rough version of an illustration of DNA reptation through a gel.
— Raghuveer Parthasarathy, January 14, 2020