mathbabe

The original goal of my blog, or at least one of them, was to expose the inner workings of modeling, so that more people could use these powerful techniques for stuff other than trying to skim money off of pension funds.

Sometimes models are really complicated and seem almost like magic, so part of my blog is devoted to demystifying modeling, and explaining the underlying methods and reasoning. Even simple sounding models, like seasonal adjustments (see my posts here and here), can involve modeling choices that are tricky and can lead you to be mightily confused.

On the other hand, sometimes there are “models” which are actually fraudulent, in that they are not based on data or mathematics or statistics at all- they are pure politics. Supply-side economics is a good example of this.

First, the alleged model. Then, why I think it’s actually a poser model. Then, why I think it’s still alive. Finally, conclusions.

Supply-side economics

At its most basic level, supply-side economics is the theory that raising taxes will stifle growth so much that the tax hike will be counterproductive. To be fair, the underlying theory just says that, once tax rates are sufficiently high, the previous sentence is valid. But the people who actually refer to supply-side economics always assume we are already well withing this range.

To phrase it another way, the argument is that tax cuts will “pay for themselves” by freeing up money to go towards growth rather than the government. That extra growth will then result in more taxes taken in, albeit at the lower rate.

Now, as we’ve states this above, it does sound like a model. In other words, if we could model our tax system and economy well enough, and then change the tax rate by epsilon, we could see whether growth grows sufficiently that our tax revenue, i.e. the amount of money that the government takes in with the lower tax rate, is actually bigger. The problem is, both our tax system and economy are way too complicated to directly model.

Let’s talk abstractly, if it’s the best we can do. If tax rates (which are assumed flat, so not progressive) are at either 0% or at 100%, the government isn’t collecting any money: none at 0% because in that case the government isn’t even trying to collect money, and none at 100% because at that level nobody would bother to work (which is an assumption in itself).

On the other hand, at 35% we clearly do collect some money. Therefore, assuming continuity, there’s some point between 0% and 100% which maximizes revenue (note the reference to the Extreme Value Theorem from calculus). Let’s call this the critical point. This is illustrated using something called the Laffer Curve. Now assume we’re above that critical point. Then raising taxes actually decreases revenue, or conversely lowering taxes pays for itself.

Supply-side economics is not a model

Let me introduce some problems with this theory:

1. We don’t have flat taxes. In fact our taxes are progressive. This is really important and the theory simply doesn’t address it.
2. The idea of a 100% tax rate is mathematically flawed, because it may well be a singular point. We should instead consider how people would behave as we approach 100% taxation from below. For example, I can imagine that at 90% taxation, people would be perfectly happy to work hard, especially if their healthcare, education, housing, and food were taken care of for them. Same for 99% taxation. I do think people want some power over their money, so it makes more sense to think about taxation approaching 100% than it does to imagine it at 100%. Another way of saying this is that the critical point may be at 97%, and the just plummets after that or does something crazy.
3. It of course does depend on what the government is doing with all that money. If it’s just a series of Congressional bickering sessions, then nobody wants to pay for that.
4. The real problem is that we just don’t know where the critical point is, and it is essentially impossible to figure out given our progressive tax system and the enormous number of tax loopholes that exist and all the idiosyncratic economic noise going on everywhere all the time.
5. The best we can do is try to figure out whether a given tax increase or decrease had a positive revenue effect or not on different subpopulations that for some reason are or are not left out, so what’s called a natural experiment. This New York Time article written by Christina Romer explains one such study and the conclusion is that raising taxes also raises revenue. From the article:
   

   Where does this leave us? I can’t say marginal rates don’t matter at all. They have some impact on reported income, and it’s possible they have other effects through subtle channels not captured in the studies I’ve described. But the strong conclusion from available evidence is that their effects are small. This means policy makers should spend a lot less time worrying about the incentive effects of marginal rates and a lot more worrying about other tax issues.

6. There are plenty of ways that natural experiments are biased (namely the subpopulations that are left out of tax hikes are always chosen very carefully by politicians), so I wouldn’t necessarily take these studies at face value either.

Supply-side economics is a political model, not a statistical model

In this recent Economix blog in the New York Times, Bruce Bartlett explains the history of supply-side economics and the real reason this flawed model is so popular. He explains an old essay of Jude Wanniski’s entitled “Taxes and a Two-Santa Theory,” which if you read it is an political, idiosyncratic argument for supply-side economics. Bartlett describes Wanniski’s essay thus:

Instead of worrying about the deficit, he (Wanniski) said, Republicans should just cut taxes and push for faster growth, which would make the debt more bearable.

Mr. Kristol, who was very well connected to Republican leaders, quickly saw the political virtue in Mr. Wanniski’s theory. In the introduction to his 1995 book, “Neoconservatism: The Autobiography of an Idea,” Mr. Kristol explained how it affected his thinking:

I was not certain of its economic merits but quickly saw its political possibilities. To refocus Republican conservative thought on the economics of growth rather than simply on the economics of stability seemed to me very promising. Republican economics was then in truth a dismal science, explaining to the populace, parent-like, why the good things in life that they wanted were all too expensive.

The Kristol quoted above is Irving Kristol, the “godfather of neoconservatism”. So he went on record saying that whatever the statistical merits of the supply-side theory were, it was awesome politics.

Conclusions

First, my conclusion is that Christina Romer should be ahead of Larry Summers on the short list to be the head of the World Bank. I mean, at least she’s trying to use actual data to figure this stuff out.

Second, I think there’s some lessons here to be learned about how people think and how they want to be convinced things work. When confronted with something they don’t like, like taxes, they are happy to believe a secondary effect, namely stifled growth, actually dominates a primary effect, namely tax revenue. It’s wishful thinking but it’s human nature.

My first question is, can Democrats come up with something along those lines too, which uses wishful thinking and fuzzy math to get what they want done? How about they come up with an economic model for how getting rid of big banker bonuses and terrible corporate governance will improve the economy, with a reference to a calculus theorem thrown in for authentification purposes?

My second question is, can we get to the point where people can figure out they are being manipulated by wishful thinking and fuzzy math with unnecessary references to calculus theorems? I know, wishful thinking.