TEST FOR TRADING COSTS EFECT IN A PORTFOLIO SELECTION PROBLEM WITH

The test statistic used to test the null hypothesis in this situation is the Wald test statistic given by: T δ̃2 W = σ̃2δ where σ̃2δ is a consistent estimator of the asymptotic variance of δ̃ and T is the number of observations used in the estimation process. [...] Hence, if we denote by cα1 the critical value of the test implemented in the first step then in the second step the J-test is implemented as follows: If W > cα1 then the critical value of the J-test is the standard one χ 2 α (K − L)2 If W ≤ cα1 then the critical value of the J-test is (0.5χ2(K − L) + 0.5χ2(K − L+ 1))α3 Let c1α2 denote the (1− α2)- quantile of a χ2 (K − L) and c2α3 the (1− α3)- q. [...] The first row (case µ = 0 ) corresponds to the size of the tests whereas the four other rows correspond to the power of the tests. [...] The return on the market portfolio and the interest rate are 19 from the Fama-French database and the CPI are from the Federal Reserve Bank of St. [...] The Panel A of those two tables gives statistics when the investor accounts for trading costs in the portfolio selection problem and the panel B contains the same statistics when the investor ignores the trading costs.