He says a lot. I’ll try to address piece by piece.

Next, some people have shown interest in this paper by Diamond and Saez. A key result that seemed to get these people excited is the calculation of a top optimal marginal tax rate (including all taxes) of 73%, relative to the current rate of 42.5%. There are two key assumptions that Diamond and Saez make to come up with the 73% optimal rate. First, we should not care about the welfare (at the margin) of the rich people. This argument is based solely on the notion that marginal utility of income is low for the top income-earners. Second, Diamond and Saez use a "behavioral elasticity" of tax revenue with respect to the tax rate of 0.25. To see how this matters, if you use their formula and an elasticity of one, you get an optimal top tax rate of 40%.

This is definitely right. How people respond to taxes is a big deal. This is especially true for the very wealthy.

Now, I know it is fashionable to dump on rich people, but I’m not sure we want to discount their welfare as much as Diamond and Saez want to. Preferences will matter here. For example, if we take internal habit persistence seriously, as some people like to, that could make us want to weight the rich and poor equally, by Diamond and Saez’s logic. I’m not committed to habit persistence, but there may be some features of behavior that are not consistent with log utility, for example. Further, Diamond and Saez are thinking in static terms. In reality, there is mobility within the income distribution, and how much mobility is an important issue here. Given mobility within the income distribution, we all care, for selfish reasons, about how the rich are treated, as we all could be rich some day, or our descendants could be rich.

This argument cuts both ways though and indeed supports higher tax rates. For D&S the “function” of higher tax rates is to ease the burden on lower income folks.

If people might both be lower income and higher income, or have uncertainty about their children’s income then you want more progressivity not less. That’s because progressivity functions like insurance. It helps you when times are bad but hurts you when times are good, thus smoothing out overall fluctuations.

Importantly – and I actually think this is the right way to think about it – it serves as insurance for unborn generations. If your grandchild might be rich then she might also be poor. It would be natural to want to write a social insurance contract whereby we agreed that the rich grandchildren would give to the poor, thus making us all feel more secure about the economic future of our grandchildren.

Though I do think people don’t take into account the enormous assumption you are making when you use log utility and of course my hobby-horse, the separable log. Effects that actual human beings talk about on a daily basis simply disappear automatically when you use that functional form.

Finally, I have no idea where that "behavioral elasticity" is coming from, and I don’t trust it. My best guess is that it includes none of the factors that I think are important in addressing the problem. What we need here is a dynamic general equilibrium model that can take account of the short run and long run effects of a change in the income tax schedule. My best guess is that "behavioral elasticity" means that Diamond and Saez are measuring the effects of tax evasion and the intensive margin of labor supply, and that’s all. If so, I think they miss most of what is important:

I am all about a dynamic GE model but the results don’t sound crazy to me.

1. There’s also an extensive margin. Tax people at a higher rate, and some drop out of the labor force

Its going to be really hard to make the numbers work here. Unless you are raising total taxes then higher marginal rates mean lower infra-marginal rates. You lose the entire thing if you drop out of the labor force.

You really need piece rate workers like freelance writers and actors to get big effects.

2. Taxes affect occupational choice. Some work byManuelli/Seshadri/Shin says that the effect of taxes on human capital is big time. Why do I want to undertake a costly and risky investment for a very small payoff?

Haven’t read the paper but the obvious answer is that the payoff is status. Plumbing pays pretty well and is steady work. I see a lot more 1300+ SAT kids signing up for a Gender Studies major than a plumbing apprenticeship.

3. Entepreneurial activity has to be very elastic with respect to tax rates at the top end. Why would I want to risk my own wealth or that of my close family for a very big payoff with very low probability, if that big payoff is taxed at 73%?

If I teach the blogosphere anything it will be this: NEVER PUT YOUR OWN MONEY IN THE SHOW!!

If God had intended for you to risk your own resources he never would have invented credit markets and compulsive savers. These things exist for a reason – use them.

Here is a short quiz to see what we have learned.

You have a business deal that is *guaranteed* to payoff. If you put in $50,000 and a bit of sweat equity you’ll get back $200,000 in two years. Where do you get the funding

(a) Savings

(b) Mom and Dad

(c) Maxing out every credit card you can find, paying off the balance with cash advances from other cards, pushing your credit limit as high as you can get it and then dumping all the money into your scheme

If you answered anything but (c), then out of the gene pool immediately!


You just failed Optimism Bias 101. Only you and your mom think this plan is guaranteed. In fact this plan will not work. You are not special. Most of your ideas suck. Your understanding of the relevant issues ranges from shallow to non-existent to not-even-wrong.

How do I know? Are you you a person? Well, then.

However, you can succeed using the law of large numbers, if you keep trying and remember this golden rule: NEVER PUT YOUR OWN MONEY IN THE SHOW!!

I have more comments related to entrepreneurship but it is crucial that this point sink in.

4. The United States is highly dependent on highly-skilled labor that migrates here from other countries. With a top tax rate of 73%, the Indian engineers might prefer to work in India, and the Canadian professors might prefer Canada.

Again, its going to be hard to make this work. Remember that high top marginal rates reduce the rates on lower levels. This means that the vast majority of folks are going to get a positive total income effect.


As always I think the big concern that you would want to model is how highly progressive tax rates effect the speed of economic evolution. Its going to mean that the after tax cash flow from more successful operations and capital investing techniques will be less differentiated from  less successful ones. Because internal financing is cheaper than external financing this means slower expansion.

UPDATE: You may ask, doesn’t the concern over internal financing conflict with the golden rule?

If you are dealing with a pure gamble then yes, there is no such thing as playing with the house’s money. Any money you win is yours and you need to treat it as such.

However, with business this isn’t true because there is a huge amount of serial correlation in business success. A business that was successful this year is far more likely to be successful next year than a business that was failing this year. This means reinvestment can make sense even if – importantly – you have no real understand of why this business works. You just need to be aware of the serial correlation and spread between internal and external financing.