Gross Private Domestic Investment:
A Year 2000 Forecast using Past Interest Rates
College of Arts and Sciences
Western Carolina University
The nation’s Gross Private Domestic Investment is expected to increase approximately 1% by the year 2000. However, there is a slight chance the economy will experience a recession in the early part of 2000 due to small decreases in GPDI. The Federal Reserve must watch this closely and monitor interest rates to combat the problem. The Federal Reserve should try to keep interest rates stable as long as possible and then consider a slight decrease in rates in order to boost GPDI in the latter part of 2000. This forecast is based upon the Keynesian aggregate investment function. The explanatory variables used in the forecast are the 90–day Treasury bill (secondary market) and 30–year Treasury bond interest rates, and real GDP.
Part 1. Introduction
This paper forecasts real gross private domestic investment for the year 2000. The explanatory variables are the 90–day Treasury bill interest rates (secondary market), 30–year Treasury bond rates, and real gross domestic product. The approach is based on the Keynesian investment function.

Real GDP is used because it is the most common means to measure the economy. Gross private domestic investment was selected as the forecast target because it reflects the investing habits of the typical consumer and adds to the stock of productive capital available to produce GDP. A forecast decline in GDP would indicate a recession for the year 2000. The year 2000 was chosen as the forecast horizon because of the short time span from the present, therefore producing a more accurate prediction.

The remainder of the paper is organized as follows: Part 2. contains the data concerning the interest rates and GDP from the past six years, from 1992 until 1998; Part 3. explains the basis for the forecast; Part 4. includes the regressions and describes how both the interest rates and GDP relate to GPDI, and the actual forecast for GPDI for the year 2000; Part 5. demonstrates the importance of the two different forecasts for the immediate future of the economy; Part 6. determines what the effect on the economy the forecast will have and what monetary policy the Federal Reserve must adopt.

Part 2. Data

All variables are taken from the Federal Reserve Bank of St. Louis Federal Reserve Economic Data (FRED). The measures of Gross Private Domestic Investment and GDP are FRED variables GDPIC92 and GDPC92. Both are real variables and are given in billions of chained 1992 dollars at seasonally adjusted annual rates. The primary source is the U.S. Department of Commerce Bureau of Economic Analysis.

The interest rate data given are the 3-month T–bill (secondary market) rates and the 30-year T–bond rates (FRED variables TB3Ms and GS30), both of which are given as a percent discount. The primary source for this data is the Board of Governors, U.S. Federal Reserve System. These interest rates are both monthly variables. The value given for the third month in each quarter (March, June, September, and December) was taken as the value for the quarter. The sample period runs for the first quarter of 1992 to the end of the last quarter of 1998.

From the sample data two regressions were run. The first was done using both sets of interest rates as the X variables; the second was done using real GDP as the X variable. Both regressions were estimated with the right-hand-side variables lagged two years because the forecast horizon was two years into the future.

Part 3. Economic Theory:
The Keynesian Investment Function as a Forecasting Tool
Interest Rates can change the amount of investment spending each year.  Quarterly data for interest rates are used to forecast the volume of investment spending over the next eight quarters.  The prediction of investment spending will be determined by the Keynesian Investment Function which is:
I = f (i,Y)
This is a simple model that will be used to forecast investment spending. I is investment, dollars firms commit to investment. The function shows that either variable i (interest rates) or Y (national income, i.e., real GDP) influences investment spending.

The Keynesian Investment Function is assumed to be a simple linear function used to forecast investment spending.  When using the two interest rate variables (omitting income) the formula becomes:

I = a + b*i(SR) + c*i(LR)

 When using one variable in the function, GDP, (omitting interest rates,) the formula becomes:

I = a + b*GDP

Two different regressions were run, one using interest rates and the other using GDP.  This was done to separate the effects due to interest rates from the effects due to income or GDP.  Each variable can affect investments independently of the other variable.  If SR interest decreases, firms' opportunity cost of investment decreases.  Investment can also be an independent function of LR interest.  As LR also decreases, the amount of investment firms can afford also increases.  LR interest has decreased due to the low inflationary pressures on the economy and peoples' willingness to invest in larger market securities, such as stocks.

There are a few shortcomings of this paper.  First, SR and LR interest may be altered any time the Federal Reserve Board decides interest rates need to be raised to combat various inflationary conditions.  Second, investment decisions of firms may be hard to track since there are many people with differing financial positions and various degrees of risk tolerance, which may prevent them from investing.

Part 4.  Empirical Results

After running the first regression with the interest rates (X variables) to the GPDI (Y variable) it was found that the R–square of the estimate was 0.7439.  The F–statistic was 24.69.  The regression estimate is (t-statistics in parentheses):

I = 1242.05(5.92)+111.03(6.57) i(SR) -85.17(-2.96)i(LR)

where I is investment spending, the short-term interest rate is on 3-month T-bills, and the long-term interest rate is on 30-year T-bonds.  All t-statistics are significant at conventional levels.

After running the second regression with the GDP (X variable) to the GPDI (Y variable) it was found that the R–square of the estimate was 0.937965.  The F–statistic was 272.15.  The regression estimate is (with t-statistics in parentheses):

I  = -2095.1(-10.75)+0.4869(16.49)GDP

where I is investment spending and GDP is real GDP.
After running both regressions it is obvious that consumer investment is explained well by both lagged interest rates and GDP. Annual forecasts (calculated with four-quarter averages) for 1999 are1244.55 (in billions of chained 1992 dollars) using the two interest rates and 1444.55 using GDP. For 2000 the forecasts are 1303.20 using the two interest rates and 1580.96 using GDP.

Quarterly forecasts are reported in table 1.

Table 1
Regression Estimate of RGPDI for 1999 and 2000
billions of chained 1992 dollars
Projected RGPDI(I) 
(using Interest Rates)
Projected % Change
(using Interest Rates)
Projected RGPDI(I) 
(using GDP)
Projected % Change
(using GDP)
 Part 5. Forecast Implications

A GDPI forecast suggests the amount of individual wealth and income will be used to maintain and increase the capital stock. Comparing GPDI to interest rates and GDP proved that there was a correlation between the three over the data.

The forecast predicts little or no chance of a recession in the near future. This is apparent when looking at the interest rates and GDP data. 1999 will see a strong economy with an increase of 3.18% (using interest rates) and 19.75% (using GDP) in consumer investing. In the year 2000 there will be an increase (decrease) of –2.16% (using interest rates) and 18.64% (using GDP) in consumer investing. Although this is slightly lower than the change in 1999 it is not serious and could probably be attributed to precautionary measures taken by consumers due to the changing of the century.

Part 6. Policy Conclusions
The forecast assumes little or no change in interest rates by the year 2000. However, if GPDI begins to rapidly decrease it would probably be in the best interest of the Federal Reserve to decrease interest rates to promote more investment. Because the forecast only predicts approximately a 1% increase in domestic investment, it would probably be in the best interest of the Federal Reserve and the economy to maintain current rates, since they are already very low.

It is predicted that the economy will move towards an upward trend with a slight decrease in interest rates to offset the decreasing domestic investing in the early part of 2000.