90 / 330
以參考一籃子貨幣為名：人民幣匯率機制之驗證
90
rate of USD/RMB to crawl upward slowly since the announced regime shift, the flexibility
of the USD/RMB exchange rate is very limited and the RMB is pegged to the US dollar
more closely than it is to the non-dollar currencies.
By and large, results from the Kalman filter estimate indicate that the RMB basket is
essentially a one currency basket of the US dollar. Contrary to the expectation of the general
public, the weight in the RMB basket hasn’t been shifted from the US dollar to other
currencies.
We perform several robustness tests. Regardless of the choice of the numeraire and the
frequency of data used for investigation, we find that the RMB basket is essentially a one
currency basket of the US dollar.
In view of these findings, China’s new exchange rate regime can best be characterized
as a discretionary crawling peg to the US dollar. This regime seems to be the optimal and
logical choice for China given its circumstances. First, by allowing the RMB to appreciate
against the US dollar, it defuses pressures from the US and other countries, and prevents
trade conflicts from escalating. Second, as the appreciation comes at a very slow rate, it
gives time for the Chinese enterprises, their managers and other parties of concern to learn
and adapt to change. Third, as the appreciation takes place in an on-and-off manner and at an
erratic rate, it helps China to curb speculative activities in the financial market betting on the
movement of the RMB exchange rate. Moreover, by not committing to a predetermined
exchange rate mechanism, China can have more flexibility in managing the RMB exchange
rate.
Originality/Contribution
This paper contributes to the literature in the theoretical as well as the empirical aspect.
In the theoretical aspect, we show that to estimate the weights of the anchor currencies in the
basket, the correct specification is to use the rates of change in exchange rates and to write
exchange rates in quantity term. If the estimation equation is based on the levels of exchange
rates and exchange rates are in quantity term, the estimated coefficients are the amounts, not
the weights, of the anchor currencies. When the estimation equation is based on the levels of
exchange rates and exchange rates are in price term, the estimated coefficients are anything
but the weights or amounts of the anchor currencies. In the empirical aspect, by using the
state space modeling, we can utilize all the sample information. It is more efficient than the
rolling regression and the regression on a split-sample to examine the time evolution of the