| Abstract: | Previous research finds a negative relation between daily index return first-order serial correlations and conditional variances. However, this finding should not necessarily be surprising since an inherent, negative mathematical relation exists between the two measures. I use progressively longer holding-period intervals for conditional variance estimation to diminish the mathematical correspondence between variances and daily serial correlation. In the process, approximately two-thirds to three-quarters of the negative relation vanishes. Thus, most of the negative relation documented previously is likely due to mathematical relations. What is left may have minor economic importance. |
