There are three estimators for linear factor asset pricing models:

  • TradedFactorModel implements an estimator which is appropriate when all factors are traded assets. When this is the case, the model can be estimated using regressions as a SUR.

  • LinearFactorModel implements a general purpose estimator based on the 2-step strategy where the first step estimated the factor loadings and the second step estimates the risk premia using the estimates from the first step. This model is appropriate for both traded and non-traded factors.

  • LinearFactorModelGMM implements a version of the 2-step model using GMM. The GMM estimator is relatively efficient and so should be preferred. The continuously updating version can is available using an options when fitting the model (use_cue=True).

All estimators implement both standard heteroskedasticity robust inference (the default) as well as kernel-based HAC estimators using eights based on the Bartlett kernel (Newey-West), the Parzen kernel or the Quadratic-Spectral kernel.

The basic usage is the same for all three estimators. Two inputs are required:

  • portfolios - The test portfolios. A T by P array of portfolio returns.

  • factors - The priced factors. A T by K array of factor returns or shocks.

This example makes use of some data from Ken French’s data library. The factors are the market, the size factor and the value factor. Sis test portfolios are used with both small and large firm size (S1 and S5) and low and high value (V1 and V5). The portfolios are transformed into excess returns prior to estimation.

from linearmodels.datasets import french
data = french.load()
factors = data[['MktRF', 'SMB', 'HML']]
portfolios = data[['S1V1','S1V3','S1V5','S5V1','S5V3','S5V5']].copy()
portfolios.loc[:,:] = portfolios.values - data[['RF']].values
from linearmodels.asset_pricing import LinearFactorModel
mod = LinearFactorModel(portfolios, factors)
res ='kernel')
                      LinearFactorModel Estimation Summary
No. Test Portfolios:                  6   R-squared:                      0.8879
No. Factors:                          3   J-statistic:                    39.109
No. Observations:                   819   P-value                         0.0000
Date:                  Sun, May 21 2017   Distribution:                  chi2(3)
Time:                          21:18:56
Cov. Estimator:                  kernel

                            Risk Premia Estimates
            Parameter  Std. Err.     T-stat    P-value    Lower CI    Upper CI
MktRF          0.0060     0.0016     3.7381     0.0002      0.0029      0.0092
SMB            0.0001     0.0011     0.1281     0.8980     -0.0021      0.0023
HML            0.0045     0.0012     3.7904     0.0002      0.0022      0.0068

Covariance estimator:
KernelCovariance, Kernel: bartlett, Bandwidth: 4
See full_summary for complete results