K. Victor Chow

A simple multiple variance ratio test

Abstract Empirical applications of the variance ratio (VR) test frequently employ multiple VR estimates to examine the random walk hypothesis against stationary alternatives. Failing to control the joint test size for these estimates results in very large Type I errors. This manuscript extends the Lo and MacKinlay (1988) methodology and provides a simple modification for testing multiple variance ratios. Monte Carlo results indicate that the size of our test is close to its nominal size and that it is as reliable as the Dickey-Fuller (D-F) and the Phillips-Perron (P-P) unit root tests. For a stationary AR(1) alternative, our test is comparable to both the D-F and the P-P tests and seems to be more powerful than these tests against two unit root alternatives, an ARIMA(1,1,1) and an ARIMA(1,1,0).

Testing for Marginal Changes in Income Distributions with Lorenz and Concentration Curves

Asymptotically distribution free statistical tests for comparing absolute and relative Lorenz and concentration curves are provided. The procedures do not require independent samples and can be used to test for marginal changes in income distributions. The tests are illustrated using a large sample of tax returns that have been randomly selected for audit by the Internal Revenue Service. The tests reveal the marginal effects of systematic underreporting of income and tax liabilities on the U.S. income distribution. Copyright 1994 by Economics Department of the University of Pennsylvania and the Osaka University Institute of Social and Economic Research Association.

Long-term and short-term price memory in the stock market

Abstract In this study we examine the issue of memory in common stock returns through an analysis of short- and long-term dependencies in various equity return series. We conclude that there is no compelling evidence that would support a finding of systematic dependencies in the behavior of equity returns.

On the long-term or short-term dependence in stock prices: Evidence from international stock markets

This study examines the short- and long-term dependence in the United States and 21 international equity market indexes. Two heteroscedastic-robust testing methods, the modified rescaled range analysis and the rescaled variance ratio test, are employed to test for the existence of dependence. The evidence consistently reveals the absence of long-term dependence in these 22 stock returns indexes. The random walk hypothesis for most, but not all, stock returns indexes is not rejected. When the random walk hypothesis is rejected, the evidence supporting the rejection is weak and the stochastic dependence occurs mainly in short-horizon, rather then long-horizon holding period returns.

Did Tax Reform Reduce Actual US Progressivity? Evidence from the Taxpayer Compliance Measurement Program

Micro data from audited tax returns are used to evaluatechanges in the actual progressitivity of U.S. federal incometaxes in 1979 and 1988, which is distinct from apparent progressivity.Statistical inference methods are applied to global measuresof both actual and apparent residual and liability progression.The analysis reveals an absence of change in the overall levelof actual residual progressivity, but statistically significantdeclines in actual liability progression. The results indicatethat in assessing overall tax progression it is important totake sampling errors into account and that measures of actualtax progressivity can deviate significantly from apparent progressivity.

Value, Size, and Portfolio Efficiency

In this article the authors examine the effects of book–to–market value (BE/ME) on portfolio selection and efficiency. The authors classify U.S. stocks into six value/size categories and generate a large number of random portfolios for each value/size category. Using a range of portfolio ranking criteria, they show that portfolios composed of high BE/ME stocks. Portfolios composed of high BE/ME stocks perform better than randomly selected portfolios. These results obtained for all ranking criteria and for all portfolio sizes examined.

LORENZ DECOMPOSITION AND INTERDISTRIBUTIONAL LORENZ COMPARISONS

We use recently developed methods to perform decompositions of the Lorenz curve in the United States by race, region, and marital status. The decomposed Lorenz ordinates are used to construct interdistributional Lorenz curves (ILCs), which allow us to identify an economic advantage by one subgroup over another or changes in economic advantage over time. We propose asymptotically distribution-free estimators for the ILCs and apply these estimators to data from the Current Population Survey for 1977 and 1997. As one might expect, there are economic advantages by race, region, and marital status, even in 1997. Economic advantage is greatest for marital status and smallest for region in both years. We find significant convergence (i.e. a smaller economic advantage) over time by race and region, though not by marital status.