Daily Chart: Pet Theories About Health-Care Spending

By Conor Clarke

This chart by Cato's AEI's Andrew Biggs has been snaking its way around the blogosphere for the past week:

And it's gotten approving [update: and skeptical!] chirps from Megan McArdle, Tyler Cowen, Jim Manzi, Arnold Kling and Greg Mankiw and others, a good deal of whom parrot the old line about how this shows that "The reason that we spend more [on healthcare] than our grandparents did is not waste, fraud and abuse, but advances in medical technology and growth in incomes." If it were waste, fraud and abuse, wouldn't you see the difference in the animal market?

But let's not flap about this too much. The chart is hounded by some fatal problems. John Schwenkler gently badgered me into trying to make a new version of this chart that deals with some of them, and I've been monkeying around with the data for the past couple of days. But, for reasons I'll grouse about after the jump, I can't reproduce a better version of this chart. (Scott Winship and Zubin Jelveh have ferreted out some of the missing data.) What I can do is graph the growth of pet food spending over the same period, and then list some of the reasons why the original chart doesn't prove much at all. (And cut out the dumb animal puns.)

1. This data is drawn from the same source (the Consumer Expenditure Survey) as the original chart. The raw slope of the pet food spending line is actually higher than the raw slope of the veterinary spending line. The normalized slope of the veterinary care line is a bit higher, but both are higher than average economic growth over the same period. Does this mean there is something unique about the two health markets, or something unique about the two animal markets? Or neither? I have no idea.

2. As Schwenkler and Manzi and others have pointed out, the original chart does not have per capita data. But of course we only care about how much is being spent on health care per person or dog. If the population grows quickly, the overall level of spending will grow with it. (Incidentally, this is why I'm having trouble reproducing Biggs' chart exactly: I can't find the number of total pets per person in the country between 1984 and 2006. And, to be extra cautious about it, I'd also need to know something about how the population has changed -- more ponies or parakeets or whatnot.)

3. Even if the chart made the same point on a per capita basis, I'm not sure why it would be surprising. You don't really have insurance or adverse selection in the veterinary market. But you do have large information asymmetries (the vets know more), large demand uncertainties (the need for veterinary care springs up uncertainly), large supply constraints, and a whole series of new patent-protected treatments that can lead to market failures.

4. Even if none of the problems in # 3 turn out to exist, I'm not sure why the growth of veterinary spending is a point in favor of conservative theories about the growth of health-care spending. Two of the most commonly cited conservative reasons for the rise in health-care spending are (1) The tax exclusion for employer-sponsored health insurance; and (2) malpractice liability, which is supposed to lead to defensive medicine and higher costs. But neither of those things happen in the veterinary market! If the original chart is correct, then are these things not really problems?

My overwhelming suspicion is that the chart does not tell us much that is useful about the market for medical care. I spoke with Andrew Biggs yesterday, and he very kindly shared his data from the expediture survey (which is not publicly available). He also cautioned against taking any of this too seriously. 700 words and two charts later, I agree.