Thursday, August 12, 2010

Thoughtful Thursday: General Equilibrium Models

Journal of Economic Literature, John B. Shoven and John Whalley "Applied General Equilibrium Models of Taxation and International Trade" September 1984, Volume XXII, Number Three, Page 1007 to 1051

The demos is voting on the tax structure and its parameters. They need to know what their preferences for a tax code will do for the budget deficit and to the broader economy. We tax income 30%. Will revenue go up or down? This is the famous Laffer curve, beautifally explained by Dr. Krugman Equilibrium Models solve non-linear equations representing demand and supply, production curves, how much will people work if taxes go up? how much will they save? Drs. Shoven and Whalley give as example: simple model with a manufacturing sector and a nonmanufacturing sector, labor and capital and those who live primarily off their investments and those who live off their wages. Assume the demos decided to tax the rich, that is the income they get from dividends and interest, at fifty percent. This would reduce manufacturing output. And thus this tax would get less revenue than expected as people would consume more nonmanufactured goods.

Also, we sometimes hear that a trade or tax policy has a cost to the economy of so many billions of dollars. The general equilibrium model says how much of each good each group would receive after the change. These are converted to a single money amount in two ways. We use the new equilibrium incomes and prices and ask how money would we have to give or take away from each household to return them to their previous utility level. This is called compensating variation (CV). Or we could take the old equilibrium prices before the tax or trade change, and say how much do we have to add or take away from each household to make them "whole." The demos can look at this summary statistic for the whole economy or look at how each group in the economy does. Probably, some want to "soak" the rich and do not care about their happiness. Others may have similar feelings about those who receive welfare or who work for the government.

In the simple model with which the article started, the tax caused a welfare loss of 0.66% of the whole economy, but the tax revenue was only a few percent, so the deadweight loss from the tax by giving people less manufactured goods than they wanted was one quarter of the revenues earned. However, rich households that were living off their savings suffered all of the decline while the average Joe was better off. So whoever is making a policy decision can think in a more reasoned way.

Some tax codes say that the income from certain investments is taxed at a lower rate. A sophisticated model found that these that they reduced the interest rate for those vehicles. Thus, the tax code was more progressive than it might appear. And at the margin, one estimate says that the deadweight loss from a marginal change in the tax code is 100%. Or the cost for an extra dollar of government revenue to the economy is two dollars.

Of course, economists simulate with lots of parameters. One used thirty-three products. They used 100 different categories of households. Joseph Pechman and Benjamin Okner looked at 87000 different categories to see "Who bears the tax burden?" And as the article, but more importantly, a Cato Institute review, it matters whether we look over a lifetime or single year. A student in a high-end business school or medical school obviously has a low income but just as obviously expects a high income soon. But can we simulate or get information from the demos on this number of categories? If the tax code were a decision tree, people would try to move themselves to a more favorable branch of the tax tree.

This reminds me of the Simulation and its Discontents" of which I read a review in the Chronicle of Higher Education. One can put in one's simulation an equation making rioting likelihood an increasing function of taxes, income, or inequality. People playing with the simulation will learn "taxes causes rioting" regardless of whether there was there was empirical basis for the equation.

These modules depend upon "elasticity estimates" and input-output tables. If the prices of manufactured goods go up by ten percent, will people consume ten percent less, twelve percent less, eight percent less. And we simply don't know these numbers. An economist illustrates these issues, bacic micro economics. Similarly, there is a basic engineering question: how much steel does it take to build one windmill, how many doctors or nurses does it take to provide a certain quantity and type of health care. The latter is an input-output matrix. We only know what the prices are today and the amount consumed today. Yes economists can make estimates by looking over time, but although prices were different ten years ago and demand was different ten years ago, lots of other things were different then such as tastes and family structure, and even technology such as how long a car lasted.

Thus as each voter enters in their preference for what the taxes should be, they must also be asked to put in their estimates for all these parameters, how do each worker and invester respond to taxes. Of course, the web site would have estimates from the best economists a click away. But if the economists disagreed, the voter must choose.

And these estimates vary. Drs. Bruegger and Loertscher showed that even in a simple model, voters could be stuck in a local optimum. This was particularly if there are large external shocks to the economy so that they can't tell what changes are do to the policies for which they vote and which changes come from the external shock. For example, basic intermediate macroeconomics says that if interest rates increase, people will save more. But some economists said that the elasticity would be close to zero. My mother said that those who had a temperment for spending would rationalize not putting away for a rainy day and those who were by nature thrifty, would find a rationalization for saving. If interest rates are higher, they may save less, as a lower retirement nest egg would grow sufficiently to meet their retirement needs. These are the kind of debates of which voters must confront or resolve in their own minds, as they input their votes about the economy and taxes.

The second part of the article concerns trade effects and trade policies. This is not as relevant to the theme of this blog, but one thing that struck me is for large countries at least, the effect of free trade is very small. One estimate was that forming the European Economic Community only improved cost-benefit by the equivalent of 0.05 percent.

No comments:

Post a Comment