By Dylan Walsh
It happens every four years in Iowa: presidential candidates storming up and down the state, meeting after meeting and stump speech after stump speech, all for the support of a population that skews disproportionately white, retired, and rural. This year’s chaos revived the perennial question of how well Iowa’s caucus serves the institution of American democracy. Less often do people ask how it might benefit the field of microeconomics.
For Michael Sinkinson, an assistant professor of economics at Yale SOM, the return of the recurring madness in Iowa and other states provided a chance to investigate the effectiveness of pharmaceutical advertising on television.
“Of course, I’d love to call Pfizer and ask to run experiments with their million-dollar advertising budgets, to try to understand what effect these advertisements have,” says Sinkinson. “But since Pfizer is unlikely to let me randomize the placement and timing of their ads, I looked for a natural experiment instead.”
Political ads are given media priority over other advertisements, and, because primaries are chronologically and geographically staggered, statewide media markets are flooded with political advertisements on a predictable cycle. For example, 5,120 political ads aired in Iowa in January of 2008 compared to an average of 19 political ads in other areas. This deluge in Iowa, in turn, displaced other kinds of advertising.
With Amanda Starc, a colleague at Northwestern University, Sinkinson used this schedule of displacement to study the efficacy of direct-to-consumer pharmaceutical advertising. Looking specifically at a class of anti-cholesterol drugs known as statins, they collected data on pharmaceutical advertising, political advertising, and statin prescriptions across 210 media markets in several primary states.
“The logic was that, all else being equal, when the caucuses happened in certain swing states the number of people getting statin prescriptions shouldn’t change — the schedule of the primaries should really have nothing to do with the drug market,” Sinkinson says. “But if we saw a dip in the number of people taking statins in the weeks after the Iowa and New Hampshire primaries, and then we saw the same thing the following month in Nevada and South Carolina, then it would suggest that something about the political event was affecting the market.” The stranglehold of political advertisements seemed a likely cause.
This outcome—displaced ads followed by reduced statin prescriptions—is precisely what they found. Political advertising during primaries pushed out, on average, 16% of ads for Lipitor and 12% of ads for Crestor, two well-known statins. In some markets, 30% or more of Lipitor ads were displaced. In the wake of this “political shock,” as Sinkinson and Starc describe it, statin prescriptions declined. Ultimately, the researchers linked a 10% increase in advertising with a 1.4% increase in a company’s revenue, showing that Lipitor advertisements effectively persuaded people in the target demographic to “talk to your doctor,” as the advertisements invariably put it, about the drug.
This finding provides rare empirical insight into the value of pharmaceutical advertising, but Sinkinson and Starc also considered deeper implications of this result. “There are huge policy debates around whether we should allow drug ads on TV,” Sinkinson says. Many people claim that the money companies plow into marketing is partially to blame for the costliness of U.S. healthcare. (New Zealand is the only other developed country to allow direct-to-consumer pharmaceutical advertising.) “Most people have the gut reaction that this is wasteful spending, that the money should go into R&D instead. But could the spending in fact be good for society?”
Statins, it turns out, do their job wonderfully well: they prevent heart attacks with very few side effects. Doctors strongly recommend them for people with high cholesterol. Sinkinson and Starc looked at how well statins reduced the risk of heart attack in clinical trials, then took a rough estimate of the cost of a heart attack, and finally used these figures to perform a “back-of-the-envelope calculation” of the savings provided by the effectiveness of statin advertisements.
The benefits were staggering. Not only did the monetary value of avoided heart attacks outweigh the amount that pharmaceutical companies spent on statin advertisements over the period of the study, but the gains proved substantial enough to justify all direct-to-consumer advertising across the entire pharmaceutical industry over that time. Getting people to talk to their doctors about statins produced a powerful ripple of value for society at large.
Critically, these results hold only when scrutinizing statins. The value of avoided heart attacks is relatively clear and well documented. The potential side effects of statins are negligible, and so “we assumed them away,” Sinkinson says. If the researchers had chosen another class of drugs with less definitive benefits or complicated side effects, like antidepressants, the math would be much more difficult and the conclusion uncertain.
These findings present a powerful new consideration for both drug companies and policy makers when it comes to pharmaceutical advertising. Companies spend billions of dollars on media spots every year; “if there is one drug like these statins in the mix, then all that advertising could be worthwhile,” Sinkinson says.