What difference do independent directors actually make?

I see the Italians have updated their corporate governance code. The new code, most of which comes into effect on 1 January 2015, requires, amongst other things, publicly listed companies to have at least two independent directors. This sounds like a good move; one which is consistent with codes elsewhere, including New Zealand and Australia for example. The basis for requiring at least two independent directors (also called outside directors in some jurisdictions) on the boards of publicly-listed companies sounds robust: independence is said to be conducive to improved decision-making and to transparency, and two directors have more chance of exerting influence than one lone voice.

But what of the holy grail question? Do independent directors enhance business performance? 

Many practitioners think that the approach to discussions, debate and decision-making by independent directors is more deliberate and objective (than executive/insider directors), primarily because independent directors are thought to be less emotionally involved in the day-to-day business and that they have less to gain or lose. Over the last three years, I have read upwards of 50 research papers on independent, non-executive and outsider directors.  While the research is not unequivocal, the general tenor seems to bear practitioner perceptions out. 

However, the impact of independent directors on business performance far less clear cut. A variety of conclusions are apparent in the research. Cause has not been established. It's a bit like saying that female directors cause companies to perform better. Increasingly, people are realising that board performance is more likely to be contingent on what directors do in certain situations than on who they are or any specific board structure or composition. Like gender, the independence attribute is likely to be a proxy for something else. We need to discover what that might be, so it can be used to qualify the suitability of director candidates and inform board performance assessments. Only then will the writers of codes be able to move beyond the reasonably blunt instrument currently in use: proxies.