Analyzing the Implications

Achieving these goals, however, requires not only a major political push, but also an enormous analytical effort. The Bush White House and many organizations in the Washington think tank community have assembled a wide array of policy databases and analytical models to assist in achieving these goals. The databases contain information on taxpayers, families, business organizations, and so forth; and analysts use these data to test the degree to which a proposed tax policy change would accomplish its equity and simplicity goals. The analytical models are employed in assessments of how proposed changes would affect economic performance.

This last test is particularly important to the Bush economic and tax plan, since so much of the justification for cutting taxes is higher economic growth. When tax policy affects the level of economic activity and the pace of economic growth, it likely changes the pool of income and sales from which it draws its own revenues. Tax policy changes that might strike some as being too expensive for the government to undertake turn out to be highly affordable if they stimulate higher economic growth.

For example, many Washington policymakers initially opposed repeal of the national wealth tax on the grounds that it would reduce federal revenues too much and would not benefit any group in America except the wealthy. However, database work showed that many who paid the national wealth tax are farmers, ranchers, small-business people, and others who would not consider themselves wealthy at all.

Economic analysis using sophisticated models of the U.S. economy showed that repeal would generate higher economic growth and more income tax revenue to the U.S. government. For example, our own econometric work and similar studies published over the years by a wide range of economists supports this interesting relationship between estate taxes and general economic activity. The U.S. taxes estates at tax rates from 17 to 50 percent. The U.S. passed legislation in 2001 that will repeal this tax in 2010 after phasing it down over the intervening years.

And, it is a good thing we are repealing it. I used a mainstream model of the U.S. economy to estimate the estate tax's economic effects and found that:

• Total civilian employment would jump an average of 142,000 to 215,000 jobs per year over the 10-year period, 2002 through 2011.
• Inflation-adjusted GDP would increase by an average of $15.1 billion per year, reaching $24.6 billion in 2011.
• Inflation-adjusted fixed investment would grow by an average of $10.1 billion per year and increase to $18.1 billion in 2011.
• The user cost of capital would fall by 120 basis points by 2011.
• Inflation-adjusted disposable income (what households have left over after paying taxes) would grow by an average of $22 billion and reach $32.7 billion by 2011.
• Repeal would lower government revenues for about four years before economic growth leads to more revenues than before repeal.

Preparing economic estimates such as these requires a host of analytical tools, from models of the U.S. economy to databases that allow the analyst to study the wealth holdings of all Americans. Indeed, every major policy change lends itself to just this type of analysis: extensive use of data about individual groups or types of people and employing a model of the general economy to estimate the policy change's economic effects.

Tax Analysis at The Heritage Foundation

In U.S. policy circles, the state-of-the-art in policy analysis now consists of a complex integration of database analysis and specialized modeling, generally of the sort that produces forecasts of policy effects. Let me illustrate how we perform our tax analysis at The Heritage Foundation's Center for Data Analysis. (Much the same work is being done at the U.S. Treasury Department's Office of Tax Analysis.)

I will use the major tax legislation that just became law on May 27 for my illustration. President Bush set forth the basic elements of his tax proposal in November of 2002. The initiative had many parts, but it became known for two bold moves. First, President Bush proposed that all of the tax law changes that had been enacted but would not become effective until 2004 or 2006 would take effect in 2003. The reduction in the tax rates paid by upper income taxpayers was the major part of this move. Second, the President urged Congress to end the taxation of dividends received by individual taxpayers. He justified this policy change by arguing that double-taxing income is bad tax policy, and dividend income is taxed first at the corporate level and again at the individual level.

It may not surprise you that a presidential proposal to cut taxes and end the double taxation of income drew praise from The Heritage Foundation. The tax economists at the Foundation jumped into action nearly as soon as the President had finished describing his tax plan.

Permit me a small digression from my story. The Heritage Foundation maintains a state-of-the-art Individual Income Tax Model. This model contains equations representing every part of the U.S. tax code relating to individual taxpayers. Because the U.S. tax code is a swamp of complexity, the model is complex. However, it permits analysts to calculate quickly how much a proposed tax law change would affect an individual's tax bill.

Of course, there are nearly as many different tax bills as there are different taxpayers. Thus, our Income Tax Model sits atop an enormous database composed of data on hundreds of thousands of taxpayers in order to capture this variety. We have about 300 variables for each taxpayer in our model. These data come from public use datasets provided (at cost) by the Department of the Treasury and the Bureau of the Census. We take these two datasets that are produced from two very different surveys by two unrelated departments of the U.S. government and match variables in both in order to come up with the 300 or so characteristics of each taxpayer. We then create taxpaying families and households from the individual records. All of our work is tested by how close we come to matching the exact amount of taxes paid in certain historical years.

Now, back to the President's plan. The Foundation's tax economists first introduced the provisions of the President's proposal into this Individual Income Tax model in order to calculate the tax effects on the entire spectrum of taxpayers. Our model is able to estimate these effects for each of the 10 future years that constitute the budget projection period required by Congress's budget rules. When we subtract the annual amount of total tax collections under the President's plan from the annual amount without that plan and sum these differences over 10 years, we come up with the 10-year "accounting" or static cost of the proposal.

We next calculated the effects of the plan on the economy. These calculations are performed using a model of the U.S. economy. The model of the U.S. economy that we employ contains 850 equations and over a thousand variables that are organized in a sequence of six blocks, each block of which represents a major sector of the economy. We introduced our "accounting" or static estimates into this model, made a few assumptions about labor and capital supplies, and let the model run. Economists call the resulting set of estimates "dynamic" because they reflect how consumers, producers, and other economic actors change their economic behavior after a change in tax law.

Our "dynamic" analysis of the President's proposal showed:

• Higher gross domestic product. Gross domestic product (GDP) would increase by an average of $69 billion in inflation-adjusted dollars from FY 2004 through FY 2013, with roughly 73 percent of this increase derived from the plan's dividend exclusion component.
• More job opportunities. The employment level would average 844,000 additional jobs from FY 2004 through FY 2013, with projected increases of 997,000 in 2004 and 1,036,000 in 2005.
• Added disposable personal income. Disposable personal income (after adjusting for inflation) would be almost $179 billion higher in FY 2004 and would increase by an average of $121 billion from FY 2004 through FY 2013.
• Higher economic growth lowering the Treasury's static revenue effect by 57 percent.Static estimates suggest the President's Economic Growth Package would reduce federal revenue by about $638 billion from FY 2004 through FY 2013. However, the CDA's "dynamic" estimates show that the proposal would reduce federal revenue during the period by only $274 billion.

We are not done yet. After estimating the effects on the U.S. economy, we produced tables that show how the major economic indicators (employment, income, tax payments, and so forth) change in each state and in each congressional district. These tables translate national numbers into the quantities that politicians and local news writers can use in communicating the benefits or harms of policy change. Our state and district tables proved highly valuable in the effort to educate the public on the benefits of the President's plan.

These same steps--static estimates, economic analysis, and translation of national results into state and congressional district estimates--also were performed by analysts at the Treasury Department and within the White House. Indeed, it is so common now to see analysts trace these same three steps in analyzing policy change, that those who do not are viewed as performing well below professional standards.

Greater Role for Statistical Analysis

What we do to analyze tax proposals we also do in analyzing policy changes in national energy, welfare, crime, education, and trade policy. A wide array of databases and specialized models support our work in this diverse set of issue areas. In fact, the Heritage Foundation's Center for Data Analysis maintains over 70 major databases, ranging from ones that follow large groups of children over long periods of time to see how income and family structure affect their health to others that have organized vast amounts of information on entrepreneurs. Our specialized models range from one focused exclusively on reforms to our publicly provided retirement pensions to another that estimates the economic welfare effects of changing tariffs on goods and services.

Members of this audience who rightly think of the United States as a country that highly values individual liberty and privacy may be shocked to hear about how much data are collected and how freely this information is distributed to private research organizations, like my think tank. Given the enormity of data collection in the U.S., it is more surprising that so little political abuse comes from it. Citizens are protected by datasets that give researchers representative cases but not the actual data on any particular citizen. Our privacy and data collection laws also protect citizens from an overly prying government.

I suspect that post-industrial economies are more, not less, sensitive to changes in tax policy. Capital and labor move easily between sectors, products, and countries; capital markets play an enormously more important role in these economies than in less developed ones; and, the relatively older age of the population in post-industrial than in less developed economies makes the after-tax return to labor and capital more important. If I'm right, then all of these factors mean that even seemingly small tax policy changes have large effects on the overall economy.

This greater sensitivity of the economy to tax policy expands the role of sophisticated data analysis in the process of policy formation. Indeed, policymakers are relying more and more on data analysis to shape and guide their efforts. If the future of policy work in the United States is clear at all, it is that database analysis and analytical modeling will occupy a greater place in policy debates than they do today and that the results of statistical analysis will play a greater part in determining how those debates are resolved.

William W. Beach is the John M. Olin Senior Fellow in Economics and director of the Center for Data Analysis at The Heritage Foundation. He spoke in Paris, France, at the conference "Mettre l'ISF au service de l'emploi" ("Putting the Wealth Tax to Work for Labor") sponsored by the Institut Français de Recherche sur les Administrations Publiques (iFRAP or The French Research Instiute on Public Services).