The other day I linked to the Assistant Village Idiot, but I forgot to post the nice note he sent me about pharmaceutical samples:

I echo the email from the public-sector psychiatrist, as I work with the same population. Samples Are All. Because different parts of the mental health budget come from different pockets, there is seldom a way to make the obvious tradeoff of $800/day in a hospital for $200/month for medications. If Big Pharma didn't make a whole lot of Consta and Zyprexa available for free, the states and Medicare would be out a whole bunch of money in increased hospitalization. (And crime. And preventable injury. And substance abuse.)

By the way, when you read the persistent poverty statistics - that 4% that is impoverished decade after decade (as opposed to the new immigrants and graduate students who are counted among the poor now) - half of that is the various mentally disabled. I see the temporary poor as sucking up a lot of resources that might have otherwise gone to my people.

Mark Kleiman questions whether they should be valued at the market price of the drugs, or the marginal cost of producing them. Well, this is an interesting accounting quandary. Pricing promotional products is tricky. Some of them will go to people who would otherwise have bought your product, and should therefore be priced at the full retail price, because that's what you lose every time they take one of your free pills. Some of them will go to people who would never have bought your product, and should therefore be priced at the marginal cost of production. The problem is, we don't have any way of sorting Group A from Group B.

We could guess, of course. But accountants do not like numbers that give people leeway to guess, because down that road lie the creative extravagences of Enron. Oh, they permit it when they really have to, for example in making allowances for bad debt. When banks make loans, they record the loans as an asset, but then record a corresponding liability of some percent of the loan value, to represent the people who will not repay them. Since we don't know who they are--if you did, the bank would hardly loan them money--or even how many of them there are, the bank's officers take a guess about how many people will default. A look at the current subprime crisis will tell you why accountants like to keep this sort of thing to a minimum.

So, accountants generally stick with the one price they do know. Assets are recorded at historical cost, even if they've appreciated--and free samples are recorded at the retail price of the product. Mark seems to think that the drug companies are, by recording the samples at retail cost, putting one over on the rest of us; they're just pretending that over half their marketing budget is free samples. But if the drug companies recorded the samples at the marginal cost of making the pills, rather than retail, their marketing expense would fall by about half, thereby depriving the pharma-bashers of one of their biggest sticks.

The intuition that companies ought to record the cold, hard cash outlay, rather than some airy fairy opportunity cost, is common. But it is not right. Giving free samples to customers who would have bought your drug costs you real cash: the cash he would have given you. More importantly, by reducing the value of the company, it costs the shareholders real cash; their stock is worth less than it otherwise would be. If Wal-Mart opened its stores and told everyone to take stuff for free, would their profits be reduced by only the value of the lost merchandise?

Obviously not. In general, accountants strive to be conservative, that is, to stick with known values rather than estimates. This usually, though not always, has the effect of making the book value of the company less than the "true" value, and both boosters and bashers can find something to quarrel with in that. But in this case, the pharmas aren't getting away with anything more than a lower EPS number to report to their shareholders.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.