【当時の情勢と合致する推定結果🔥】「The Effects of Japanese Foreign ExchangeIntervention: GARCH Estimation and Change Point Detection」：先行研究解説 No.11 2023/09/25

Although the GARCH estimation of section 4.1 provides a clear result for the impact of Japanese foreign exchange intervention, it gives only one comprehensive result for the whole observation period.
The global estimation might not account for parameter changes that are for instance frequently observed for the volatility of financial time series.

Comparing the sum of autoregressive parameters for the global GARCH estimations with and without exogenous variables provides evidence for parameter changes.

The global GARCH(1,1) estimation without exogenous variables yields a λ equal to 0.98 (λ=∑αi +∑βi ) which corresponds to a mean reversion time of about 50 days.
The estimated sum of autoregressive parameters being close to unity indicates that parameter changes might not be accounted for (Hillebrand 2003).

In contrast, the estimation of the GARCH model with exogenous variables as specified in equations (1) to (3) yields λ equal to 0.92, which in turn corresponds to a mean reversion time of about 12 days when the intervention term is introduced.

The reduction of the time of mean reversion (smallerλ) shows that the interventions at least coincided and possibly caused the parameter changes in volatility.

To cope with this problem we re-estimate our GARCH model for sub-periods as proposed by Bonser-Neal and Tanner (1996).
In a first step we subdivide our observation period into calendar years.

Although this partition is somewhat arbitrary we get a notion of changing parameters. The results of the local yearly GARCH estimations are reported in Table 4.

There seems to be a change in the effects of Japanese foreign exchange intervention starting in 1999.

In the period between 1991 and 1998 there is no significant impact from foreign exchange intervention on the mean despite considerable intervention volume and a large number of intervention events.

Intervention seems to have increased exchange rate volatility in the years 1993 to 1995.

In the second sub-period starting in 1999 the results undergo a change. In all four years up to 2002 the intervention coefficients b1 are positive and highly significant.
Japanese foreign exchange intervention seems successful during this time period.

The impact on exchange rate volatility is less clear-cut. After 1998 negative signs dominate (except for 2001), but are mostly insignificant.
Only in the year 2000 foreign exchange interventions seem to have reduced exchange rate volatility significantly at the 10% level.

The combination of successful foreign exchange intervention and reduced exchange rate volatility might not be accidental, as successful intervention reduces the scope of appreciation and thus dampens relative exchange rate changes on the respective day.

Understanding that data segmentation considerably affects our estimation results we have pursued an alternative approach.
Beine and Szafarz (2003: 6) suggest that Japanese foreign exchange intervention exhibits clear patterns of clusters.

Based on Figure 1 we build eight periods of intervention clusters, which are indicated by the numbered bars in the lower part of the Figure.

Then we set the boundaries of the segments mid-way between the last and the first day of each intervention period.
Although these intervention clusters are again chosen somewhat arbitrarily we obtain additional evidence on the effect of data segmentation.

The main findings as reported in Table 5 widely match the findings of the yearly estimations.

Between 1991 and 1998 Japanese foreign exchange intervention seems to have been ineffective and between 1994 and 1996 it seems to have increased exchange rate volatility.

Starting from late 1998 there is strong evidence for—in the short-term—effective foreign exchange intervention.
In contrast to the yearly estimations, the intervention coefficients in the volatility equation are not only negative but also highly significant.

Thus, since 1999 successful Japanese foreign exchange intervention has coincided with reduced exchange rate volatility at highly significant levels.

Based on the findings reported in Table 4 and Table 5 we identify two sub-periods in which Japanese foreign exchange intervention had different impacts on exchange rate returns and exchange rate volatility—1991 to 1998 and 1999 to 2002.

To consolidate this result we re-estimated our model as specified in equation (1) to (3) for the two sub-periods. The results are reported in Table 6 and confirm our main conclusion.

Up to 1998 Japanese foreign exchange intervention was ineffective, but coincides with increased exchange rate volatility. Since 1999 Japanese foreign exchange intervention had the intended short-term effect with evidence for less exchange rate volatility.

Changing the boundary between the two subsamples by one to six months prior to and after January 1, 1999 does change the meaningful results with respect to the mean equation (but less robust with respect to the volatility equation)