PCA factor extraction

Usage

pca_est(target = NULL, X, nfac, gamma = -1)

Arguments

target: Ignored; accepted for API uniformity with other estimators.
X: Numeric matrix or data frame (T x L) of factor proxies.
nfac: Positive integer; number of factors to extract.
gamma: Numeric scalar controlling mean adjustment in the second-moment matrix. `gamma = -1` (default) gives the sample covariance (traditional PCA). `gamma = 10` and `gamma = 1` give the Lettau-Ludvigson variants from He et al. (2023).

Value

An object of class "sdim_fit".

References

He, J., Huang, J., Li, F., and Zhou, G. (2023). Shrinking Factor Dimension: A Reduced-Rank Approach. Management Science, 69(9). doi:10.1287/mnsc.2022.4563

Examples

set.seed(1)
X <- matrix(rnorm(100 * 8), 100, 8)
fit <- pca_est(X = X, nfac = 3)
print(fit)
#> <sdim_fit [pca]>
#>  Observations : 100 
#>  Predictors   : 8 
#>  Factors      : 3