The business economist at work: the economics function at the Colorado Legislative Council

Business Economics, Jan, 1992 by Nancy J. McCallin

In order to forecast revenues we use many different models. We have elaborate structural econometric regression models for all of the major tax sources. For instance, sales taxes are disaggregated by industry classification, then each component is modelled accordingly. Individual income taxes are disaggregated into the components of withholding, estimated payments, case with returns, and refunds. Individual income taxes are modelled by component and in aggregated form. An additional input into the individual income tax forecast is a simulated model of 3,000 Colorado income tax returns maintained by the Department of Revenue. We provide the Department with our income and employment forecasts, and they then apply the forecasts to their sample data.

In addition to structural econometric models, we also have ARIMA and exponential smoothing models for the tax concepts. These time series models are good only for short-term forecasting and are thus only relied upon when there are significant data consistency problems or when the tax concepts are relatively small.

In addition to these more standard models, we are experimenting with neural computing for some of our more difficult revenue streams. In a joint venture with Colorado State University, we are attempting to use neural computing to forecast corporate taxes. In the past, we have tried disaggregating corporate taxes by industry and modelling the components as a function of industry profits per Colorado worker. Given the poor data on corporate tax receipts, this and other models did not prove fruitful. Neural computing uses a computer model that mimics the functionality of the human brain, identifying patterns in data and learning from them.

After the models are run and forecasts are chosen, the data must then be adjusted for accruals. Once the accrual forecast is complete, the net General Fund revenues are analyzed with respect to appropriations, transfers, and the ending balance to ascertain whether the Fund is in surplus or deficit. If the reserve has fallen to 2 percent or less of appropriations, the General Assembly and the Governor must take action to restore the reserve to the 2 percent level.

POLICY RESEARCH AND MONITORING THE ECONOMY

In addition to revenue forecasting, other functions are performed by the Economics Department at the Legislative Council. Whenever the General Assembly is considering increasing or decreasing taxes, they request an analysis of the resulting revenue and economic impacts. When the Legislature considers tax reform or a significant alteration of the tax structure, we are called upon to analyze the revenue impacts.

A good example of the type of research we do was recently evidenced by United Airlines' request for thirty-year job tax credits in return for locating a maintenance facility employing 4,500 people in Denver. We were asked to analyze the fiscal and economic impact of the request. The tax credits were not tax credits in the purest form, however. Rather, if their value exceeded United's Colorado income tax liability, the state would have been obligated to write a check to United for the difference. In total, we valued the tax credits at $206 million in inflation-adjusted 1991 dollars.