John M. Griffin 

Empirical Methods in Finance

University of Texas at Austin

Portfolios, Assignment #4

Portfolios are essential to finance and accounting research. Form 10 equally-weighted momentum portfolios. Carefully follow details laid out in Jegadeesh and Titman (2001, Journal of Finance). Your Table should look similar to their table 1. Yes, follow their same details. You will need to pull the monthly return and market value data for all stocks from CRSP using WRDS, you may need other pieces of information like Price (see JT). It is strongly encouraged that you use SAS, STATA or as package that makes handling this data easy as you will need to use similar data for your next test.

Note: you can potentially form portfolios other than momentum, you should get the permission of the Professor as to what portfolios you propose to form.

Hint: until you make sure your program works don't run it on the whole CRSP database, just use a subset, so your not waiting on your program to run... Nor clogging up the servers.

 

  1. Show your Table 1 that is equivalent to Table 1 in Jegadeesh and Titman (01). This Table should have two Panels, Panel A with the same period as their period and a Panel B which has the results up until December?  Hint: If you can't replicate their Table you've done something wrong and you won't be able to know your right going forward.
    a) What has happened to momentum profits over the last 5 year?
    b) Explain what is happening in January?
    c) Does this change the inference obtained in Jegadeesh and Titman (01)?
     
  2. Now rank (proc Rank) your firms according to 4 groups of market capitalization and present momentum profits within each group. The results should look just as in Panel B (of Table 1) except now there are four sets of portfolios. Call this Table II
     
  3. Form a value-weighted portfolio similar to this and call it Table III. When forming value-weighted portfolios you can create value-weighted portfolios in SAS with the weight being the market capitalizations.
     
  4. Run a time-series regression with the FF 3-factor and Carhart 4-factor model and judge whether your anomalies exist. Is this a good metric?