Info

453,516

P = implicit price deflator (1981 = 100) R = 90-day prime corporate interest rate, % GDP = C\$, millions (1981 prices) Source: Rao, op. cit., pp. 210-213.

Gujarati: Basic I III. Topics in Econometrics I 17. Dynamic Econometric I I © The McGraw-Hill

Econometrics, Fourth Models: Autoregressive Companies, 2004 Edition and Distributed-Lag

Models

684 PART THREE: TOPICS IN ECONOMETRICS

The estimated short-run demand function shows that the short-run interest elasticity has the correct sign and that it is statistically quite significant, as its p value is almost zero. The short-run income elasticity is surprisingly negative, although statistically it is not different from zero. The coefficient of adjustment is 8 = (1 — 0.9607) = 0.0393, implying that only about 4 percent of the discrepancy between the desired and actual real cash balances is eliminated in a quarter, a rather slow adjustment.

To get back to the long-run demand function (17.11.2), all that needs to be done is to divide the short-run demand function through by 8 (why?) and drop the ln Mt_i term. The results are:

STM* = 21.7888 — 1.6132ln Rt — 0.6030lnGDP (17.11.7)44

As can be seen, the long-run interest elasticity of demand for money is substantially greater (in absolute terms) than the corresponding short-run elasticity, which is also true of the income elasticity, although in the present instance its economic and statistical significance is dubious.

Note that the estimated Durbin-Watson d is 2.4582, which is close to 2. This substantiates our previous remark that in the autoregressive models the computed d is generally close to 2. Therefore, we should not trust the computed d to find out whether there was serial correlation in our data. The sample size in our case is 40 observations, which may be reasonably large to apply the h test. In the present case, the reader can verify that the estimated h value is —1.5008, which is not significant at the 5 percent level, perhaps suggesting that there is no first-order autocorrelation in the error term.