Is There a “Self-Selection” Effect in FQHC Executive Compensation Surveys?

Many of our clients and colleagues have expressed a concern that surveys of FQHC executive compensation, including those conducted by the National Association of Community Health Centers, State and Regional Primary Care Associations (“industry surveys”), and even those conducted by ElementOne, are subject to a “self-selection” effect — that is, only certain types of people or organizations participate, meaning you only get part of the picture.  The concern, or the hypothesis presented by some folks, is that it’s the higher-paying organizations that aren’t participating, making the survey results artificially low.  I’m not sure I buy the logic, but it’s worth a look.

We decided to take a look at the hypothesis to see whether our recent study of FQHC Form 990 submissions (which you really can’t dodge, at least not for long), validates the concern.  We began by comparing the industry data to our 2012 FQHC CEO Compensation survey, and then to the predicted base salaries from the 990 study.  For the entire lurid detail of the methodology, ask for a copy, which we’ll provide as soon as it’s available.  We won’t be providing you with the industry data, however… that’s something you need to acquire legitimately from the publishers on your own.

When it comes to health centers with less than $10 million in revenue, there was essentially no difference between the industry surveys, our FQHC CEO survey, and the 990 study.  The middle 50% of the samples overlap reasonably well, and the “average” rates were generally within five percent of each other.  The story changes, however, as health centers get larger — at $10 to $20 million, the predicted salaries from the 990 study were about five percent higher than other surveys at the 25th percentile, but about ten percent higher at the median, and 12.5% higher at the 75th percentile.   At $20 million and over there is a 10% difference at the 25th percentile, and a 25% difference at the 75th percentile.  Clearly there are higher rates showing up in the 990s, but is it because there are fewer larger organizations in the industry surveys, or is it because higher paying organizations don’t respond?

What the findings suggest is that for health centers of less than $10 million in revenue, it does not appear that higher paid individuals (or organizations) are any less likely to report their salaries than lower paid individuals.  On the other hand, there is something going on in the larger health centers — the salaries reported in the surveys are lower than those reported in the Form 990s.  Why?

It may be that larger FQHCs just don’t see their much smaller industry colleagues as peers, and just look elsewhere for data on organizations they consider to be more like peers.  There’s a lot of common sense to this.   At a certain point, size and complexity may become more important than similar business models.  One industry survey has about half the number of possible participants in the over $20 million category, which is an outstanding sample — but we don’t know which of the “over $20 million” health centers they are.  If they’re the $20 to $50 million folks, and the $50 million plus folks aren’t in it, we wouldn’t expect the industry survey numbers (or our numbers, which parallel them) to be as high as the Form 990 figures.

It may be that better-paying organizations choose not to participate because they don’t want their pay levels out there.  I’m not convinced that there’s a reason to buy into the logic for doing so, but that doesn’t mean it isn’t happening.  Since there aren’t very many surveys that actually present individual lines of data, it’s not as if your information will get out there and be seen by anyone.  What’s the rationale?  I suppose some might say “well, we participated in the survey, and now we’re obligated to use it,” but that seems like a stretch.  How would using other sources instead be any different from saying “I don’t like this diagnosis, so I’m going to get a second opinion?”  Then again, people aren’t always logical.

The short answer is… if you’re at a smaller health center, it doesn’t appear that hidden data is keeping the reported market rates down.  If you’re in a larger health center, look at the survey data you have to try and determine if it reasonably reflects organizations similar in size to your own.  This is particularly important if your health center is more than $20 million in size.  Twenty million to 200 million is a huge range — you need to look for data for truly comparable organizations.  In our 990 study, for example there were 69 FQHCs with revenue between $20 and $30 million, and the average salary was $210 thousand; there were 56 with revenue between $30 and $50 million, and a $246 thousand average salary.  Both groups are clearly “over $20 million,” but which ones you pick result in a 17% difference in pay.

It may be that the lack of participation among larger health centers is why the numbers seem too low and why many folks we talk to say “but you can’t just look at FQHCs.”  Just goes to show that you have to be very careful when looking at data…