In my opinion there is not enough evidence to make any real statement about this decomposition. The authors of the paper have analyzed it only for a single market, while momentum have worked for almost all asset classes and in different currencies and markets. To be real effect, and not a simple overfit to a single data set over a single period (which is what the authors have done), they should have tested the effect in different markets (Japan, Eurozone, Emerging Markets, etc) and across different asset classes. Then, after there is evidence in all these cases, perhaps they could have found something other than chance.
In my opinion there is not enough evidence to make any real statement about this decomposition. The authors of the paper have analyzed it only for a single market, while momentum have worked for almost all asset classes and in different currencies and markets. To be real effect, and not a simple overfit to a single data set over a single period (which is what the authors have done), they should have tested the effect in different markets (Japan, Eurozone, Emerging Markets, etc) and across different asset classes. Then, after there is evidence in all these cases, perhaps they could have found something other than chance.