Examining the effects of announcement days (a-days) of significant macroeconomic news, the authors find support for the accuracy of traditional asset pricing models, such as the capital asset pricing model (CAPM). Namely, such days validate market beta as an accurate measure of systematic risk with beta strongly related to average returns across various asset classes, including individual stocks, test portfolios, bonds, and currencies. Furthermore, there is a pronounced “on/off switch” phenomenon because expected variance is positively related to future aggregated quarterly a-day returns but has a minimal relationship with aggregated returns for all other days (i.e., those with no announcement). Investors anticipating the release of important economic news demand higher returns to hold riskier assets.

Market timing has been extensively discredited as a “loser’s game” for long-term investors. The authors indicate that most of the excess returns from holding risky assets occur within a narrow slice of the trading calendar. But in contrast to returns following a characteristically random pattern, which would argue for passive strategies, they demonstrate that over an extended period (1964–2011), nearly all the annual excess returns generated by diverse US assets (small stocks, growth stocks, high-beta stocks, the entire stock market, and even long-term bonds) are earned on a-days. Those days represent roughly 10% of total trading days and can be anticipated via scheduled news releases on inflation or unemployment and/or the meeting of the Federal Open Market Committee. Furthermore, for the overwhelming balance of time (i.e., non-announcement days, or n-days), available gains, on average, are minimal, and there is even a statistically significant negative relationship between average returns and beta. These findings would seem to offer potential for suitably designed active strategies.

The authors note this a-day versus n-day dynamic may be because the a-days reduce investor disagreements about expected growth in aggregate economic variables, which limits potential overvaluation of higher beta assets. Their evidence shows betas to be stable across a-days and n-days. So, identifying assets that could be the greatest beneficiaries of increased clarity of sentiments could provide guidance for tactical allocation strategies.

The authors rely on the CRSP database for US stock and bond return data and on Kenneth French’s website for portfolio return data organized by size, book-to-market ratio, and industry. Both a single unconditional full-sample beta and time-varying betas based on rolling one-year windows are derived for each asset type. Inflation and unemployment a-days for 1964–2011 are identified from the US Bureau of Labor Statistics website, whereas scheduled interest rate a-days are drawn from the Fed’s website. The authors run a series of regressions to calculate annual returns for the various assets using the classic two-step CAPM testing procedure, with the Fama–MacBeth procedure used for second-stage regressions to compute a-day and n-day coefficients. They also test beta coefficients via a single direct regression to determine their stability across a-days and n-days. Results for the various equity portfolios are sorted by beta, book-to-market ratio, size, idiosyncratic risk, and downside beta. They offer parallel findings for bonds across a wide range of maturities, currency carry trade portfolios, and individual stocks.

The authors try to challenge their conclusions and find no supporting evidence for their alternate hypotheses. Those rejected alternatives include the following: (1) a-day returns are intrinsically more volatile; (2) the findings are correlated with periods of high-equity risk premiums; (3) a-day and n-day betas are different; and (4) the findings can be explained by simple modifications of standard models of asset returns (i.e., CAPM or various multifactor representations).

The analysis of a-day returns seems to offer potential for noteworthiness comparable with other calendar-related market anomalies, such as effects identified for January, turn of the month, and Mondays. The findings have a fundamental basis (substantive macroeconomic news) rather than merely being determined by the passage of time. The authors recognize that further research is needed to provide a thorough explanation. Perhaps other research could cover such questions as, Do the findings hold in non-US markets? What effect do subsequent announced revisions of aggregate economic data have? Is a-day profit potential being exploited by such techniques as high-frequency trading?