“…Fan, Li, and Wang (2017) claimed that the recently proposed penalized least-squares methods in high-dimensional settings are sensitive to the symmetry of the error distributions when selecting the non zero parameters in the mean regression, and these penalized methods usually need the assumption that the distribution of model error is symmetric about zero. To measure the symmetry or asymmetry for a continuous variable or the model error, there are amount of work in the literature, see for example, Ngatchouwandji (2009); Ngatchouwandji and Harel (2013); Allison and Pretorius (2017); Fan and Gencay (1995); Ahmad and Li (1997); Neumeyer, Dette, and Nagel (2005); Davis and Quade (1978); Harald (1928); Rothman and Woodroofe (1972); Srinivasan and Godio (1974); Butler (1969); Orlov (1972); Randles et al (1980); Gupta (1967); Csörgő and Heathcote (1987); Antille, Kersting, and Zucchini (1982); Hill and Rao (1977); Bhattacharya (2009); Li and Morris (1991); Gastwirth (1971). Recently, Patil, Patil, and Bagkavos (2012) proposed a measure of asymmetry by using the correlation coefficient of the density function and distribution function, and showed this correlation coefficient-based measure works well in capturing the visual impression of asymmetry in a given density curve.…”