Help in Fama McBeth regression

[Tobias Meyer]

Hi,

I'm working on fund performance persistence. I want to run a Fama McBeth regression using market exess return as dependent variable and the four factor model (excess_ret , mktrf smb hml mom) as independent variables. I took the risk factors from French's Website and tried to use the xtfmb command: xtfmb excess_ret_USD mktrf smb hml mom, lag(4).

As a result i dont get any coeffiecients, because they are being "omitted".

 input int(fund_id month_id) float(excess_ret_USD mktrf smb hml mom) 1 602  .05430361  .0631  .0143  .0211  .0366 1 603  .00999733    .02  .0497  .0281  .0322 1 604  -.1095922 -.0789  .0005 -.0238 -.0013 1 605 -.01955375 -.0556 -.0197  -.045   -.03 1 606  .09630439  .0693  .0017 -.0026  .0188 1 607 -.02092465 -.0477   -.03 -.0195 -.0019 1 608  .10529506  .0954  .0392 -.0313  .0135 1 609   .0439209  .0388  .0113  -.026  .0164 1 610 -.05590116   .006   .037  -.009  .0254 1 611  .08198454  .0682  .0069  .0382 -.0319 1 612  .04905964  .0199 -.0247  .0082 -.0032 1 613  .02209111  .0349  .0153   .011  .0203 1 614  .01389508  .0045   .026 -.0158  .0351 1 615  .04914045   .029 -.0034 -.0252  .0006 1 616  -.0368168 -.0127  -.007 -.0208 -.0058 1 617  -.0279024 -.0175 -.0016 -.0032  .0177 1 618 -.03197328 -.0236 -.0135 -.0121  .0002 1 619 -.09962872 -.0599 -.0305 -.0248 -.0033 1 620 -.11215988 -.0759 -.0353 -.0141 -.0257 1 621   .1880953  .1135  .0342 -.0018 -.0142 1 622 -.07380099 -.0028 -.0017 -.0034  .0398 1 623  -.0213312  .0074  -.007  .0177  .0195 1 624  .09463428  .0505  .0215 -.0113 -.0796 1 625  .05866757  .0442 -.0175  .0008 -.0028 1 626 -.02237838  .0311 -.0062  .0092  .0149 1 627 -.03707525 -.0085 -.0052 -.0048  .0384 1 628 -.14057542 -.0619      0 -.0059  .0659 2 526          .  .0135  .0202  .0185  .0162 2 527          .  .0429   -.03  .0241 -.0571 2 528  .01875062  .0215   .028  .0197   .026 2 529  .02962958   .014 -.0143   .005  -.011 2 530 -.00338228 -.0132  .0175  .0022  .0021 2 531  .00146869 -.0183 -.0206 -.0262 -.0539 2 532  .01983686  .0117 -.0021 -.0039   .016 2 533  .03231329  .0186  .0226  .0139   .021 2 534 -.01912357 -.0406 -.0381  .0411 -.0229 2 535  .01624903  .0008 -.0162  .0103 -.0155 2 536  .05584915   .016  .0303 -.0025  .0524 2 537  .03868681  .0143  .0031 -.0063 -.0152 2 538  .07802767  .0454   .039  .0179  .0323 2 539  .06796304  .0343  .0011 -.0008 -.0284 2 540  .00153099 -.0276 -.0151  .0196   .032 2 541  .06052558  .0189  -.005  .0165  .0317 2 542 -.03397468 -.0197 -.0137  .0161  .0055 2 543 -.01904827 -.0261 -.0398 -.0035 -.0086 2 544 -.03953657  .0365  .0288 -.0081  .0044 2 545  .00774018  .0057  .0262  .0263  .0205 2 546  .05042563  .0392  .0293 -.0052      0 2 547  .04032664 -.0122 -.0092  .0127  .0227 2 548  .01342819  .0049 -.0058  .0077  .0345 2 549 -.02079714 -.0202 -.0121  .0022 -.0129 2 550 -.01738569  .0361   .009 -.0119  .0025 2 551  .02311369 -.0025 -.0046  .0044  .0077 2 552  .08447222  .0304  .0542  .0113  .0268 2 553   .0091292  -.003 -.0037 -.0025 -.0184 2 554  .04377983  .0146  .0354   .006  .0126 2 555  .05861366  .0073 -.0134  .0261  .0061 2 556 -.01162121 -.0357 -.0305  .0254 -.0365 2 557 -.02408048 -.0035 -.0035  .0087  .0148 2 558  .01670569 -.0078 -.0407  .0294 -.0222 2 559  .02189086  .0203  .0091 -.0171 -.0346 2 560 -.01034228  .0184 -.0137  .0004 -.0095 2 561  .05575942  .0323  .0173 -.0003 -.0023 2 562  .04849438  .0171  .0086  .0008 -.0101 2 563  .02862881  .0087  -.011  .0316  .0084 2 564  .02077733   .014   .001 -.0011  .0021 2 565  .01534891 -.0196  .0132  -.001 -.0135 2 566  .04329991  .0068 -.0005 -.0022  .0249 2 567  .05803964  .0349 -.0206 -.0115 -.0016 2 568  .01747987  .0324  .0002 -.0004 -.0027 2 569  .01855638 -.0196  .0077 -.0112  .0038 2 570  .02647226 -.0373 -.0251 -.0333   .028 2 571 -.03780674  .0092 -.0013 -.0224  .0011 2 572  .08132631  .0322 -.0229 -.0186  .0459 2 573  .06148668   .018  .0022  -.026  .0496 2 574 -.05693742 -.0483 -.0262 -.0118  .0093 2 575  .04329164 -.0087   .002 -.0051  .0652 2 576 -.03780195 -.0636  -.009  .0365 -.0789 2 577  .08509144 -.0309 -.0024 -.0094  .0611 2 578  .00719363 -.0093  .0095 -.0014   .041 2 579  .08600324   .046 -.0164 -.0095  -.002 2 580  .05833438  .0186  .0322 -.0138  .0321 2 581  .01241657 -.0844  .0128 -.0242  .1253 2 582 -.08111806 -.0077  .0245  .0588 -.0514 2 583 -.06775746  .0153  .0361  .0152 -.0404 2 584 -.18227392 -.0924 -.0116  .0633  .0035 2 585 -.27378786 -.1723 -.0232 -.0289  .0779 2 586 -.09775573 -.0786 -.0296 -.0602   .071 2 587   .0877123  .0174  .0361 -.0025 -.0508 2 588 -.05113091 -.0812  .0008 -.1118 -.0184 2 589  -.0388609  -.101  .0022 -.0727  .0416 2 590  .03864679  .0895 -.0008  .0356 -.1138 2 591  .10172112  .1019  .0477  .0553 -.3439 2 592   .1678667  .0521 -.0234 -.0023 -.1244 2 593 -.04437621  .0043  .0262 -.0272  .0529 2 594   .0779318  .0772  .0207  .0528 -.0536 2 595  .02397303  .0333  -.009  .0775 -.0884 2 596  .07202368  .0408  .0246  .0093 -.0493 2 597 -.01166996 -.0259 -.0423 -.0418  .0266 2 598  .01087924  .0556  -.025 -.0017  .0029 end

I don't understand what I am doing wrong so help would be highly appreciated.

[Arthur Morris]

Tobias,

You have not have shared enough data to actually estimate a Fama-Macbeth regression. Remember that you need enough data in each time period to run the second-stage regression.
Thus, I'm unable to replicate the behavior you are reporting based on the data and code that you've shared.
It may help you to read and follow the guidelines for asking questions. Particularly the section about sharing data and code.

The following observation is based on the names of your variables, but is only a guess because I do not know how you calculated them:

You appear to be regressing excess returns on the right-hand side of the equation that produced them. Often excess returns are calculated as the residuals from regressing "raw returns" on some thing like: "mktrf smb hml mom". It is probably worth revisiting Risk, Return, and Equilibrium to clarify your thinking (unless I'm misunderstanding what you're trying to do and need to revisit the paper myself).

[Attaullah Shah]

xtfmb or asreg (from SSC) estimates the second-stage cross-sectional regressions of the FMB procedure. Issues similar to yours have been posted here several times in the past and I have replied to them in various forms, including two blog entries. The risk factors ie. smb, hml, and rm_rf vary across time, they do not vary across companies. In a cross-sectional regression, all such variables are doomed to be omitted as the regression does not see any variation in the values of these variables. Therefore, the method you are trying to apply for FMB is theoretically incorrect. Here is one of my blog entry that provides a short walk-through on how and when of Fama-MacBeth(1973) regressions.
https://fintechprofessor.com/2019/06...sreg-in-stata/

Here is another example of using asreg for fmb regression when using portfolios.
https://fintechprofessor.com/2019/05...sreg-in-stata/

Concerning the excess returns, in the majority of the asset pricing models and tests, the excess returns are defined as the return on a given asset minus risk-free rate.

[Tobias Meyer]

I understand your point. Thank you both!

[Ferhat Demirci]

Hello community,
I want to include political risk factor to CAPM (political risk do not vary across companies). When I run Fama-Macbeth two step regression, Political risk factor omitted. How can I determine the factor loadings of political risk?