US Forecast: April 28, 2000 |
Forecast Period 2000:2--2004:4 (20 quarters) Data The forecast is based on the national income and product accounts (NIPA) data that were released on April 27, 2000. The Latest Version of the US Model For purposes of this forecast the US model has been reestimated through 2000:1. These estimates are presented in the "Chapter 5 tables" at the end of The US Model Workbook. The rest of the specification of the model is in Appendix A at the end of this workbook. A complete discussion of the November 3, 1998, version of the US model is in Macroeconometric Modeling, which is the main reference for this site. Since the November 3, 1998, version the NIPA data have been completely revised, and the revised data have been used. A few specification changes were made in the process of updating the model using the revised data. The following is a list of the changes that have been made from the November 3, 1998, version. (No specification changes were made for the current version, April 28, 2000, from the previous version, January 29, 2000.)
Some of the above changes have lessened the reliance of the model on peak to peak interpolations. Variables Z, JJS, and the output gap variable have been replaced by UR. This means that the variables JJ, JJP, JJS, YS, and Z are no longer needed in the model. UR and JHMIN are now the only two capacity-like variables in the model. JJP had been created by peak to peak interpolations of JJ. This change also eliminates equations 95-98. In addition, the excess capital variable is no longer used in equation 12, and this means that variables EXKK and MUH are no longer needed in the model. MUH had been created by peak to peak interpolations of Y/KK. This change eliminates equation 93. All these variables and equations have been retained in the software, but they play no role in the solution of the model. See Model Versions and References for more discussion of the model versions. Assumptions Behind the Forecast To put in perspective the current assumptions that have been made about fiscal policy, it may be helpful to review an analysis of three economic plans that I did on September 4, 1996: An Analysis of the Clinton and Dole Economic Plans. The three plans were the latest (at the time) Congressional plan, the Clinton plan, and the Dole plan. The budget agreement that was eventually worked out by the Democrats and Republicans was closest to the original Clinton plan, although the spending cuts were not as large as the originally planned cuts. The following table gives for the key government policy variables in the model 1) the growth rates used for the current forecast, 2) the growth rates I used in analyzing the original Clinton plan, and 3) the actual growth rates between 1993:3 and 2000:1. Growth Rates (annual rates) Current Forecast Clinton Actual Assumptions Plan 2000:1-1993.3 TRGH 5.0/8.0 4.3 4.3 COG 2.0 -4.8 0.0 JG 0.0 -4.8 -1.0 TRGS 5.8/8.0 5.8 5.5 TRSH 6.0 6.0 5.5 COS 3.0 1.0 6.6 JS 1.0 1.0 1.6 Notes: 5.0/8.0 means 5.0% through 2000.4 and 8.0% thereafter. 5.8/8.0 means 5.8% through 2000.4 and 8.0% thereafter. Comparing the Clinton plan to the actual outcome so far, the plan was exactly right transfer payments (TRGH), but way off for purchases of goods (COG) and federal civilian jobs (JG). The current forecast assumptions about the growth of COG and JG in the future are closer to the historical values than they are to the original Clinton plan values. It is assumed for the current forecast that all tax rates remain unchanged. No assumption about monetary policy is needed for the forecast because monetary policy is endogenous. Monetary policy is determined by equation 30, an estimated interest rate reaction function or rule. The Results The current forecasts for real growth (at an annual rate) for the remaining three quarters of 2000 are 5.4, 3.8, and 2.4. The growth rate is thus forecast to remain high in the second quarter of 2000 and to be fairly large in the third quarter. The unemployment rate falls to 3.3 percent by the fourth quarter of 2000. Inflation as measured by the growth of the GDP deflator (GDPD) rises substantially. It is predicted to be 3.5 percent in the second quarter, 3.6 percent in the third, and 3.7 percent in the fourth. The Fed is predicted (through the interest rate rule) to increase the bill rate (RS) to 6.6 percent by the fourth quarter of 2000, which is in response to the lower unemployment rate and higher inflation rate. The household saving rate (variable SRZ in the model) is close to zero throughout the forecast period. The federal government budget surplus (SGP) is forecast to be between \$203.6 and \$268.3 billion throughout the forecast period. The U.S. current account deficit (variable -SR in the model) is forecast to be large throughout the period (over \$400 billion for most of the period). After 2000 the growth rate slows to about 2 percent, the unemployment rate gradually rises, and the inflation rate gradually falls. In general, this is a forecast of a soft landing. Inflation increases in 2000, but the Fed cuts it off without too much harm. There are two main uncertainties in this story. The first concerns inflation. The unemployment rate enters linearly in the price equation (equation 10), but at some point one would expect to see large (i.e., nonlinear) price increases in response to a falling unemployment rate. Unfortunately, there are not enough observations at low unemployment rates for the data to estimate this nonlinearity, and so there are no nonlinear effects in the model. It could thus be that inflation will be even worse in the future than the model is predicting if the economy is close to the point where the price response is nonlinear. This is especially likely if the unemployment rate does fall to 3.3 percent, as it is predicted to do. The second main uncertainty concerns the stock market. There is a stock price equation in the model (equation 25, explaining CG), but it cannot pick up booms and crashes. The equation is predicting that stock prices will grow modestly in the future. If, on the other hand, inflation increases, the Fed tightens, and the stock market crashes, the effect on the economy through the wealth effect could be substantial. If you want to see this, you can run an experiment in which you crash the stock market (see Chapter 7 of The US Model Workbook for how to do this). You will see that a large sustained crash leads to a recession even if you assume that the Fed substantially lowers the interest rate in response to the crash. In other words, the model has the property that the Fed does not have the power to prevent a recession if there is a large crash. You can examine the tables of this forecast memo for the details, or you can print out the forecast values from the base data set for the model. Although the model is used to forecast through 2004:4, you should not put much confidence on the results beyond about 2001. Forecast error bands are fairly large for predictions this far ahead. Possible Experiments to Run The present forecast is a good base from which to make alternative fiscal-policy assumptions, depending on what you think Congress might do in light of the rosy government budget picture. For example, there is currently considerable uncertainty about possible future tax cuts, and you can experiment with alternative tax-cut plans. Remember that the current forecast assumes that there are no tax cuts. As a historical footnote, the model has consistently been more optimistic about the size of future federal government deficits than have most others, especially regarding future tax revenues, and the recent data suggest that the model has been right. The CBO and others have now moved in the optimistic direction. You may want to compare the current CBO forecasts with those from the model. My sense is that the model has conveyed useful information in the past about the deficit that was not in the CBO forecasts at the time. You may also want to drop the interest rate reaction function (equation 30) and put in your own assumptions about Fed behavior. For example, do you think the Fed will raise interest rates more than the model predicts it will in response to what is happening in the economy? Given the above discussion about nonlinearities and inflation, you may want to experiment with the price equation (equation 10). Will inflation be higher than predicted because of the very low unemployment rates? Related to the inflation question, the current forecast is based on the assumption that the price of imports (PIM) grows at a rate of 3 percent per year throughout the forecast period. A strong dollar helps keep PIM down, but oil price increases push it up. It may be that the assumption of 3 percent is too optimistic, and you may want to experiment with higher values. This will, of course, make inflation even worse in the future. Finally, as discussed above, you may want to crash the stock market. |
US Forecast Tables: April 28, 2000 |
Table F1: Forecasts of Selected Variables--Real GDP
and Components
Table F1 (continued)--Prices and Wages Table F1 (continued)--Money and Interest Rates Table F1 (continued)--Employment and Labor Force Table F1 (continued)--Other Endogenous Table F1 (continued)--Selected Exogenous Table F2: Forecasts of the Federal Government Budget Table F3: Forecasts of the State and Local Government Budget Table F4: Forecasts of Savings Flows NIPA Table 1.1 NIPA Table 1.2 NIPA Table 3.2 NIPA Table 3.3 NIPA Table 7.1 |
Table F1: Forecasts of Selected Variables |
|
Table F1 (continued) |
|
Table F1 (continued) |
|
Table F1 (continued) |
|
Table F1 (continued) |
|
Table F1 (continued) |
|
Table F2: Forecasts of the Federal Government Budget |
|
Table F3: Forecasts of the State and Local Government Budget |
|
Table F4 Forecasts of Savings Flows |
|
NIPA Table 1.1 |
|
NIPA Table 1.2 |
|
NIPA Table 3.2 |
|
NIPA Table 3.3 |
|
NIPA Table 7.1 |
|