Economic Cycles and the Monthly Effect in the OTC Market

Article excerpt

Introduction

Ariel (1987) and Linn and Lockwood (1988) find that stocks traded in the organized exchanges exhibit a monthly effect. The first halves of trading months produce returns that significantly exceed zero, according to these studies, while the second halves of trading months provide returns insignificantly different from zero. These studies further indicate that the returns in the first halves of months significantly exceed the returns in the second halves of months. Linn and Lockwood also examine the NASDAQ composite return index and provide evidence of a monthly effect in the OTC market as well.

Fama and French (1989), however, document that economic conditions produce a variation in expected returns on stocks and bonds, while Liano and Gup (1989) provide evidence that business cycles significantly influence the pattern of the day-of-the-week effect in exchange-listed stocks. Because stock returns can vary according to economic conditions, changes in economic conditions also may alter the pattern of the monthly effect in the OTC market. The objective of this study, therefore, is to investigate the effect of business cycles on the monthly effect in the OTC market. Business cycles are incorporated into an analysis of the monthly effect in OTC stocks to determine if the monthly effect exists during both economic expansions and economic contractions. The current study is the first to examine the impact of business cycles on the monthly effect in the OTC market. Although the impact of business cycles on the monthly effect in the OTC market has not been previously examined, the implications for portfolio managers are extremely important because the continuous absence of a monthly effect during either economic cycle would indicate that portfolio adjustments for the monthly effect should be ignored in that cycle. Differences in returns between small and large firms also are examined during expansions and contractions for the existence of a size effect in either economic cycle.

Data and Methodology

Daily value-weighted (VWID) and daily equally weighted (EWID) return indices of OTC stocks, provided by the Center for Research in Security Prices (CRSP), are used in this study to examine the persistence of the monthly effect in the OTC market from 1973 through 1989.(1) Because the value-weighted index is invested more heavily in larger firms than is the equally weighted index, Roll (1983) indicates that a comparison of the value-weighted and equally weighted indices can provide a measure of the size effect. To examine the size effect in the OTC market, an additional measure, DIFF, is constructed by taking the difference between EWID and VWID.(2) A positive (negative) DIFF value would indicate that small (large) firms outperform large (small) firms. For purposes of comparison, the NASDAQ composite return index (NCRI) investigated by Linn and Lockwood (1988) also is included in this analysis.

To examine the monthly effect, both Ariel (1987) and Linn and Lockwood (1988) define a trading month as the period from the last trading day of a calendar month to the second-to-last trading day of the following calendar month. Adopting the same definition, this study equally divides the trading month into two halves, the first and second halves, with the odd middle trading day to be included in the second half. A t-test is performed to determine if index returns in the first and second halves of months differ significantly from zero. Although the magnitudes of the NCRI, VWID, and EWID provide information on average daily returns in the OTC market, evidence of a size effect during the entire sample period is produced if returns on the DIFF measure differ significantly from zero. The four return indices, NCRI, VWID, EWID, and DIFF, are analyzed subsequently for the existence of a monthly effect over the entire sample period by comparing the returns on each index in the first halves of trading months to the returns in the second halves of months using parametric ANOVA and nonparametric Kruskal-Wallis procedures. …