1
$\begingroup$

I have a fairly short and straightforward question. I am running a dynamic optimization strategy and therefore need to construct the FF5 characteristics. I am using COMPUSTAT quarterly accounting data.

Now my question is why are all accounting variables such as Book-to-market ratio, size etc. constructed using at least 6 months following the FF 1992 convention. I understand that this ensures the availability of data to investors. What I do not understand is why this is not the case with a 3/4 month lag as all firms issue quarterly earning reports and therefore the information would be available correct?

I know that Asness/Frazzini (2013), the Devil in the HML details have written a paper highlighting the fact that updating the price (in book-to-market ratio) is superior to lagging it half a year. Why not also update the accounting variables every month and only include a three month lag (at most)?

I Hope someone provide some perspective.

$\endgroup$
3
$\begingroup$

This is directly addressed by Fama/French themselves in their seminal 1992 paper The Cross-Section of Expected Returns, p. 429.


The six month lag is the most common, but minimum gap between fiscal year-end and the subsequent return test beginning in end June the following year. A firm having fiscal year-end on January would have a maximum lag of even 18 month by entering the sample!

Firms are required to file their 10-K reports with the SEC within 90 days of their fiscal year-ends. However, on average 19.8% do not comply. Furthermore, more than 40% of the December fiscal year-end firms that do comply with the 90-day rule file on March 31, an their reports are not made public until April (see Alford et al. (1992)).

Well, the six month lag appears to be really conservative. However, the overall research question in their paper is the cross-sectional effect of book-to-market ratios and firm size on stock returns (i.e. value- and size-anomaly). If the main results would heavily depend on the portfolio sorts starting on end June rather than e.g. end April, their findings would not be very robust. In summary, you may choose a different time lag for accounting data and see this as a robustness-check for your overall research question.

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.