(Wow, just realized I haven’t written anything this month. I’m up for tenure this year and am working on a bunch of stuff, but I’ll try to do better than once a month.)

I was listening to some talk radio yesterday and the host was talking about the absurdity of taxing a company (health care companies in general, pharmaceutical drug companies in particular) to subsidize its customers. His argument went like this: I sell a product of my own invention, a pill, for $10.00, and that price includes a decent profit margin for my efforts. The government thinks the price is too high and people can only afford $8.50, so they tax me $1.50 and then give that $1.50 to the customer so they can afford the product. His argument was that when you tax him $1.50, he’s just going to raise the price by $1.50 so he can keep the same profit as before. Thus, the new price is $11.50 and customers getting the subsidy end up paying $10.00, leaving them in exactly the same position as before.

Sounds good, but there’s a significant problem with his argument. The only way that prices rise by the entire $1.50 is if you have a competitive market with zero economic profit — thus, prices HAVE to rise by $1.50 just for him to stay in business. But he prefaced the entire argument by saying this was his intellectual property and he included profit in his prices. This implies he has market power (the ability to raise prices above marginal cost and not lose all his customers). Thus, his assumptions contradict each other.

When firms have market power, they should produce where marginal revenue equals marginal cost (MR = MC, aka the “golden rule of profit-maximization”). If MC rises by $1.50 because of the tax, the firm should cut back output until MR rises by $1.50 also. Since the MR curve has twice the slope of the demand curve (if demand is linear), reducing output so that MR rised by $1.50 results in prices rising by $0.75. Thus, he should charge $10.75 for his product and the subsidized consumer ends up paying $9.25.

He’s right that the consumer won’t end up paying the $8.50 that the government wanted, but wrong in saying that he’ll just pass the entire tax onto his customers. Granted, he COULD do that, but he would be losing more money than he would by only passing on half the tax. He’d keep the same profit margin on each unit, but the higher price would cut back his sales volume and his total profit will be lower. Mitigating the price increase a bit reduces his profit margin, but he sells more units and profit ultimately is higher (lower than without the tax, of course, but higher than if he raises prices by the full $1.50).

This is just one example of the overuse of “supply and demand.” When people think about economics, the first thing they say is “supply and demand.” The problem is that model only works when markets are perfectly competitive (lots of firms making the exact same product, like in most agricultural markets). And if there’s any market that is not perfectly competitive, it’s pharmaceutical drugs. When you apply principles of supply and demand to non-competitive markets, your conclusions are wrong.

If there’s anything the financial crisis and its proposed remedies should have taught us, it is this: assumptions matter. For example, Christina Romer’s model predicted that unemployment would not exceed 8% if we passed the stimulus bill, and clearly it was wrong. Why? Maybe the economy was worse than they thought. And perhaps it’s also partly because her model did not account for the way consumer behavior changed. Before the recession, the personal saving rate in the US was around 2%. Given that, it may seem reasonable to think that if you give people a job or give them tax cuts they’ll go out and spend all of this new disposable income, so firms will have to produce more output and hire more people. Thus, we were told the stimulus bill would “save or create” 3 million jobs. But that’s not what happened. After the recession, the personal savings rate tripled to about 6% for the last 2 years.  People are paying down debt to increase their ability to get credit, and saving what they can because they’re worried they might lose their jobs.

So the conditions on which the model was based changed, which makes one wonder how many jobs were actually “saved or created?” Surprisingly, the White House is still claiming 3 million jobs. This seems curious, when the White House also says that the economy was so much worse than they thought it was, and consumers aren’t spending as much as they assumed the would. Baffling.

This article sheds some light on the situation. In its latest report, the CBO says about 750,000 jobs were directly paid for by stimulus money, but notes that we can’t say all of these jobs were “saved or created” because “some of the jobs included in the reports might have existed even without the stimulus package, with employers working on the same activities or other activities.” And they’re not counting a job the way most of us would think of it, as a “job-year” — so if stimulus money funds someone in a job for two weeks, it counts as a job. And what about the indirect effect on the entire economy? According to the CBO, the stimulus “increased the number of people employed by between 1.4 million and 3.3 million,” which the article notes is ”not far from the administration’s claim that the stimulus ’saved or created’ 3 million jobs.”

I have two major problems with this. First, between 1.4 and 3.3 million is a pretty wide margin of error. Only if all the ”best-case scenario” assumptions hold do we come close to the 3 million jobs that the White House claims on its web site it has ”saved or created,” and the fact that the unemployment rate went higher than expected and consumers are saving more money than they used to should be an indication that these best-case scenarios are not realistic.  Second, if you read the actual report, you’ll see that the high estimate used for the government spending multiplier is 2.5. You may remember that the multiplier used by Romer initially was only about 1.7, and that was used to say that the stimulus bill would create 3 million jobs. And that was when people were spending almost all of their income. Now, when people are saving more than before, the CBO is using a multiplier of 2.5 so that it can say there’s a chance the stimulus bill saved 3 million jobs. Hogwash.

When your model is proven wrong by the facts, perhaps you should adjust the model before making claims about counterfactuals whose validity come only from the model you use. In this case, the multipliers should be revised downward, not upward. And when talk show hosts make claims about the impact of government policies, it behooves them to realize that the real world is a bit more complicated than a simple model of supply and demand. It might not make for good talk radio, but it would certainly make them more credible.