Robust determination for the number of common factors in the approximate factor models

This paper proposes a robust determination method for the number of common factors in the approximate factor models. The new method is based on the ratio values of some transformation function of adjacent eigenvalues arranged in descending order. Under some mild conditions, the resulted estimator can be proved to be consistent. It can be further shown that, comparing with the competitors in the existing literature, the new method has desired performance on truly selecting the value of the number of latent common factors whether there are dominant factors or not. Monte Carlo simulation is carried out for illustration.