The advent of microcomputers has radically altered both the costs of and opportunities for empirical research by economists. Even more rapid changes can be expected over the next few years. As the master of the small screen said, "You ain't seen nothing yet'. Large machines will still be necessary where substantial computing power is required, but microcomputers offer real advantages in three areas: (i) interaction between researcher and data, (ii) ease of use and (iii) portability and communications.

The traditional description of economic research draws a clear distinction between the development of a theory and its subsequent testing by rigorous statistical methods. Indeed it is often thought that these two steps are performed by different researchers, the theorist who formulates a hypothesis and the econometrician who tests whether the hypothesis is rejected by the data. In fact, empirical work rarely proceeds in this way. Ideas are suggested by empirical regularities (or irregularities, as with the sharp rise in the savings ratio in the 1970s), and there is a constant interplay between specification and testing of a model. This process can be accelerated if the researcher can work interactively with his data. Microcomputers offer much greater opportunities for such interactive work than do existing mainframe facilities.

These opportunities are also enhanced by the ease with which microcomputers can be used and the availability of specialist software for economists. The phenomenal expansion of microcomputing has encouraged the market to provide a wide range of software. Economists already benefit from the financial incentives facing software houses to produce high quality and "user-friendly' programs for business use.

Finally, access to data banks and programs produced by economists both at home and abroad will be made easier by the growth of communications facilities and the "program portability' afforded by microcomputers. For example, on a recent visit to the US, I was able to demonstrate a program an hour after my arrival! In contrast, transferring data and software from one mainframe to another can be tiresome at best and impossible at worst. Microcomputers can significantly lower the cost of research projects involving international collaboration and comparisons.

A wide range of programs suitable for economists are now available for microcomputers. These include statistical packages - my own favourite, shortly to be released in the US, is "PSA', which is specifically designed for cross-section analysis. The economist now has access to a variety of programming languages (the problems with the early versions of FORTRAN have largely been overcome), as well as graphics packages and word processors. "Spreadsheet' packages such as Lotus 1-2-3 are particularly useful for the manipulation, transformation and display of data sets. Two of my own projects can further illustrate how microcomputers can be used in applied, policy-oriented research.

In the 1984 Budget the Chancellor changed the corporate tax system, removed tax relief on life insurance policies and abolished the investment income surcharge. He said that his main aim was to reduce the enormous variation in effective tax rates on different types of saving and investment. The extent of this variation, both before and after the Budget changes, can be calculated using the microcomputer-based program "PTAX' that I constructed for just this purpose. The user of "PTAX' can alter the corporate and personal tax rates and simulate the effects of various reforms. "PTAX' was the result of a research project to examine effective tax rates in different countries, and the results of the project were cited by the Treasury in its evidence to the Select Committee following the Budget.

To evaluate the consequences of "supply-side' policies (such as a reduction in marginal tax rates to improve efficiency) it is important to measure the gains and losses they create. Michael Weisbach (MIT) and I have developed "MICROTRAP', a programme that runs on a personal computer. "MICROTRAP' calculates the distribution of gains and losses from a change in the taxation of earnings and the resulting changes in labour supply. The user of "MICROTRAP' can, without any knowledge of computer programming, change the entire structure of the tax system and, for example, simulate the recent US Treasury proposals for a much smaller number of tax brackets and lower tax rates. A very important feature of "MICROTRAP' is its ability to examine the sensitivity of the gains and losses to a wide range of assumptions about labour supply responses. The user does not have to take anyone else's results on trust. This DIY approach to the analysis of tax reform will increase the impact of empirical research on the analysis and formulation of economic policy.

Mervyn King

Mervyn King is Professor of Economics at the London School of Economics and a CEPR Research Fellow. The cross-country comparison of effective tax rates is described in "The Taxation of Income from Capital' by M A King and D Fullerton (University of Chicago Press, 1984). Information on the "PTAX' and "MICROTRAP' programs is available from Professor King at the London School of Economics.