Alan Blinder, in his classic book, "Hard Heads, Soft Hearts," coined the phrase "Murphy's Law of Economic Policy." The gist of this law is that politicians embrace tenuous and controversial economic hypotheses while ignoring well-established propositions about which there is consensus within the economics profession.

For example, consider Glenn Hubbard, who reportedly has been designated to be the Chairman of President Bush's Council of Economic Advisers. Apparently, in order to reach this exalted position, he had to establish his credentials as a supply-sider, which is a cult on the fringe of economics. Hubbard demonstrated his supply-sidism by writing a paper in which he claimed that the tax increase enacted in 1993 had a major adverse effect on entrepreneurship.

To see Hubbard's point, suppose that everyone faces two alternatives. You can work for a salary at an established business. Or you can quit your job and become an Amway distributor.

Suppose that your salary as a wage slave is $100,000 a year. However, as an Amway distributor, your income is more volatile. You average $100,000 a year, but every other year you earn $200,000 and in the alternate years you earn nothing.

Next, suppose that the tax schedule consists of a 20 percent rate on the first $50,000 in income and a 40 percent rate on all subsequent income. Over a two-year period, as a salary worker you will pay $60,000 in taxes. On the other hand, as an Amway distributor, you would pay $70,000 in taxes. Because of progressive taxation, you are discouraged from quitting your job and living the American dream.

In order to obtain the $10,000 difference in tax payments, I had to exaggerate some factors. The actual tax system contains milder increases in rates than a jump from 20 to 40 percent. And the volatility of the income of Amway distributors is probably lower than my example. (Perhaps much lower. Maybe a range of -$50 to +$50 would be more realistic.)

Looking at the numbers, one would think that the tax code ought to have a relatively small effect on entrepreneurship. However, it is the hallmark of supply-side economics to infer extravagant actual effects from small theoretical possibilities.

Indeed, Hubbard performed a sophisticated analysis, called statistical Probit (woo-woo) of the data from the Panel Study of Income Dynamics (or PSID, to the initiated). Based on this study, he finds that the convexity of the tax system has sharp dampening effects on entrepreneurship.

Specifically, Hubbard claims that the tax increase that President Clinton signed in 1993 held down the number of new Amway distributors by 20 percent! It had that much impact in discouraging people from leaving their jobs to become entrepreneurs.

To understand the tenuous nature of this conclusion, it must be stressed that Hubbard is not analyzing an increase in tax rates in general. Rather, he is talking about an increase in the "convexity" (progressivity) of the tax system.

In the language of calculus, Hubbard is focused on the second derivative of taxes with respect to income. As income increases, taxes increase. That is measured by the first derivative. As income increases, taxes also increase at an increasing rate. That is measured by the second derivative.

As a general rule, the effects of second derivatives are almost impossible to measure using the data and statistical techniques available to economists. It can be difficult enough to measure the effect of the first derivative, and second derivatives are even harder to study.

I have another data set that I believe does more to illuminate the way that taxes affect entrepreneurs. Admittedly, it consists of just one observation.

As I have described in a previous essay, in 1994 I quit my job to become an entrepreneur. Here are some findings based on this data set.

1. I became an entrepreneur not long after the Clinton tax increase. Evidently, I was not one of those discouraged.

2. I actually did think about taxes. However, I saw the tax system as my friend. Because I started my business in May, I assumed that the losses that I expected to suffer in my first six months would offset the income I had earned before quitting my job. Thus, the after-tax loss would be smaller than the pre-tax loss.

3. As it turned out, the tax system worked in my favor because the really big bucks came when I sold my business. Selling the business produced income in the form of a capital gain, which is taxed at a flat 20 percent rate, rather than the 39 percent marginal rate for ordinary income.

If we reduce marginal income tax rates, then this will reduce the value of shifting ordinary income to capital gains. It would have reduced the net benefit to me of becoming an entrepreneur. Thus, a reduction in marginal income tax rates would have had the opposite incentive effect that Hubbard predicts.

The correct economic statement may be that higher marginal income tax rates help to promote entrepreneurship. Taking into account the capital gains tax, one can argue that, if anything, entrepreneurship should have been stimulated by the Clinton tax increase.

In the grand scheme of things, changes in the progressivity of the tax system probably have only a negligible effect on entrepreneurship. Certainly, this is the case for changes in progressivity within the range that we have observed in the recent past or are likely to observe in the future.

I would hate to think that the effects of the Blinder-Murphy law are so strong that in order to be appointed to an economic policy position one has to assert something as ridiculous as suggesting that the Clinton tax increase discouraged entrepreneurship by a significant amount. I think I'd sooner become an Amway distributor.