An Intertemporal Benchmark Model for Turkey’s Current Account

statistic for (ca_t ) is 2.4443 and this variable is statistically significant at 10%. Therefore, we
conclude that the current account Granger-causes future movements in national cash flow. We
also test if national cash flow Granger-causes the current account. The F-statistic from table V is
0.5792 which implies that national cash flow is statistically not significant for future movements
in the current account. Our result supports the first implication of the IBM that the current
account Granger-causes future changes in the national cash flow.

From the VAR results reported in tables IV and V we can compute the optimal consumption
smoothing current account series. Our estimated coefficients using the average world rate of
interest in our sample are reported in table VI. To do a sensitivity analysis, we compare these
coefficients to those computed over a range of rates of interest which are reported in table VI as
well. The results suggest that the coefficients are robust because they do not change significantly
when we compute them using higher and lower rates of interest.

Computation of the optimal current account allows us to test the second implication of the
IBM of equality between the actual and the optimal consumption smoothing current account. We
map the relation between actual and optimal consumption smoothing current accounts in figure
(4). The graph shows that the actual consumption smoothing current account deficit exceeds the
optimal consumption smoothing current account deficit. Visually, it does not seem that there is
equality in the two accounts.

The formal test of equality of the two accounts for a higher order VAR are that the coefficient
of the contemporaneous current account is unity and all other coefficients (for both the current
account and the national cash flow) are zero. This places restrictions on the Γ matrix, implied in
equation (11) and explained in equation (13) for the general case. Equation (13) can be rewritten
for our VAR with five lags as follows

γ= [γ_∆z γ_ca]=[00000 10000]                                            (14)

15

<<   13   14   15   16   17   18   19   >>

More intriguing information