Following Nakajima (2011), we set up the TVP-VAR model with stochastic volatility as: y_t=c_t+B_1t y_(t-1 )+⋯+B_qt y_(t-q)+u_t (1) for t=q+1,⋯,T, where y_t is a (4×1) vector of remittance outflows, inflation, savings and TFP; B_1t,⋯,B_qt are (4×4) matrices of time-varying VAR parameters. The unobservable …show more content…
6 (i) displays the endogenous responses of gross domestic savings to positive remittance outflows shock. A positive remittance shock has a diminishing negative effect on savings over the whole sample. The result suggests that a rise in remittance outflows were likely to have an adverse negative effect on savings. Furthermore, the result supports the cyclical patterns of the savings-remittance nexus over time.
On the other hand, a positive remittance shock in 1980 (Fig. 6 (ii)) leads to the gradual rise in savings in short-run and savings starts picking up in long-run, while the remittance shock in 1991 (Fig. 5 (iii)) has a similar effect as the remittance shock in 1980. The positive remittance shock in 2004 (Fig. 4 (iv)) has a short-run negative effect on savings, while in long-run the impact is insignificant.
As shown in Table 2, the results of the posterior probability for the difference in the impulse response of savings to remittance outflows shock indicate that the difference in the impulse response between 1980 and 1991 is weaker, while that between 1980 and 2004, 1991 and 2004 are milder.
6. Conclusion and