Factor model attribution analysis should assume that factors are non-linear. Current factor model attribution analysis assumes that gains are due to market timing or a better understanding of systematic risk. Current theory holds that idiosyncratic risk and systematic risk are more applicable than non-linear factors. To make factor model attribution analysis more effective on should assume that non-linear factors are more important than idiosyncratic risk and systematic risk.
Factor model attribution analysis is a method of achieving alpha against the market. For example an equity research analyst could predict that the price of petroleum will rise from $50 to $80 over the course of the next six months.
A more comprehensive view of factor model attribution analysis should assume that factors are non-linear. Currently it is held that idiosyncratic risk and systematic risk are more important than assuming factors are non-linear. In financial services if an investment bank has regional specific knowledge, it should price in a comparative advantage for underwriting governments in that region. For example HSBC should more aggressively push for market share in Hong Kong rather than compete in New York.