Exchange Market Efficiency: A Semi-Strong Test Using Multiple Markets and Daily Data
采用Hansen和Hodrick（1980）的方法，利用1971-1980年加拿大-美国外汇市场五种期限合约的日度数据，检验了市场信息效率与无风险溢价的联合假设，发现数据拒绝了该假设。
Following the method of Hansen and Hodrick (1980) we test the efficiency of the Canada-U.S.A. foreign exchange market by pooling information from contracts of five lengths. The test is based on daily data from the period 1971 to 1980 inclusive. We show that these data reject the joint hypothesis of exchange market informational efficiency and no risk premlum. The degree to which foreign exchange markets are inefficient is of obvious importance. McKinnon (1976) has argued that a lack of speculative activity has led to excessive exchange rate turbulence and associated economic cost; this would be manifest in exchange inefficiency. Additionally, tests of the efficiency of the well organized foreign exchanges are of obvious value in the ongoing debate concerning the validity of the Rational Expectations Hypothesis. In this note we report the results of a test of the joint hypothesis of exchange market informational efficiency and no risk premium. The test is based on a generalized concept of the foreign exchange market. The study uses daily data on the Canada-U.S.A. market for the period 1971 to 1980 inclusive. In its most general form, Fama's condition for market efficiency is f(i,t) = E(f(i -j,t + j)14(t)) (1) where f (i, t) is the forward exchange rate prevailing at time t for a contract of i months; s1(t) is the information set available to market participints at time t; and E(-) is a conditional expectations operator. If condition (1) did not hold and transactions were Received for publication August 3, 1982. Revision accepted for publication February 10, 1984. *Nuffield College, Oxford, and Bank of Canada, respectively. The views expressed in this paper are those of the authors and no responsibility for them should be attributed to the Bank of Canada. We appreciate the comments and assistance of David Burton, Kevin Lynch, Donn Maccara, Barbara Macpherson, George Pickering, Heather Robertson, and two anonymous referees. This content downloaded from 157.55.39.163 on Wed, 21 Sep 2016 05:12:50 UTC All use subject to http://about.jstor.org/terms 670 THE REVIEW OF ECONOMICS AND STATISTICS costless, speculators could make risky profits by taking the appropriate long or short position. Thus the hypothesis of efficient markets which we test consists of two parts: the assertion that expectations are rational; and the postulation that speculators swiftly arbitrage away economic profits by exploiting valuable information. Equation (1) recognizes the fact that a speculator in the foreign exchange market has many trading strategies open to him (Caller, 1980). For instance, a speculator can buy a two month forward contract and hold it until it matures, in which case the corresponding speculative rate is the spot rate two months hence: alternatively one can buy a three month contract and match it with a one month contract which starts two months hence. If the foreign exchange market is efficient, the speculator should be indifferent between these two, and all other possible strategies, and no strategy should yield extraordinary profit. This formulation is in contrast to the more commonly used formulation of informational efficiency, f(i,t) = E(s(t + i) l4(t)), where s(t + i) is the spot exchange rate observed at time t + i. Clearly, this condition reflects merely a single instance of the general formulation (1). By examining the complete set of strategies implied by (1) we are better able to address the question of foreign exchange market efficiency. Consistent with the view that there exists a wide array of speculative strategies open to a speculator within the Canada-U.S.A. market, we model the speculator as basing decisions on prediction errors recently realized from various maturities of the same market. In previous studies of this nature, tests have often been based upon an information set involving a single and a single (Cornell and Dietrich, 1977; Levich, 1978; Blejer and Khan, 1980; and Longworth, 1981). Usually this has been done to avoid the problem of overlapping observations, which will be further discussed below. However, the choice of a relatively coarse data frequency makes these tests less powerful, if the phenomena of interest are of extremely short duration, as was proved by Hansen and Hodrick (1980). Alternatively, economists have modelled speculators as basing their decisions on the information from multiple markets, all of the same (Geweke and Feige, 1979; Hansen and Hodrick, 1980). This is the normal method of transforming a weakform test (a test which uses only lagged regressands as explanatory variables) into a semi-strong test (which in addition to lagged regressands incorporates into the test other publicly available information). Our test is composed by pooling information from varying maturities for the same currency, instead of pooling information from different currencies of identical maturities. Our procedure seems to be an interesting alternative, primarily because of the information costs associated with speculating with many currencies. Speculation may be pictured as being a currency activity (with arbitrage occurring readily between lengths of a given ratio), as well as a maturity length specific activity. I.e., speculators may concentrate their attention on the dollar-yen rate, rather than the three-month market. Thus, we would expect information to be disseminated quickly across lengths; finding that information from one is not quickly disseminated to other maturities would be strong evidence against market