General wrong way risk is defined as due to a positive correlation between the level of exposure and the default probability of the counterparty due to general market factors. (Specific wrong way risk is when they are positively correlated anyway).
Given that the different market factors tend have a stronger positive correlation when one is talking about the same country/region(mainly the base curves), the same industry (mainly the spreads), etc, should be the concentration risk (per region, industry,..) be used to model the general wrong way risk?
With 5 regions (Americas, UK, Europe(ex UK), Japan, Asia-Pacific(ex Japan) and 10 sectors (Energy, Basic Materials ,Industrails, consumer Cyclical, consumer Non-Cyclical, Health Care, Financials, Information Techniology, Telecomunication Services and Utilities), you should be able to get the GWWR from a sort of variance of the concentration from the mean_sectors(ideally 10%) and mean_region (ideally 20%).
Ex: You have 40% of your exposure in Energy, 30% in Financials 20% in Telecomunication services and 10% in whatever else, really well diversified/entering practically into the error. What I mean is, given that the rest is all the same (same maturities, types of instruments/bonds -to simplify, pricipals, etc), the GWWR should be much larger for 40-30-40-10 than for 30-30-30-10. The absolute minimum (og GWWR) should be with the 10-20% means, but, if you practically have only 3 sectors in which you can exchange your 90%exposure within, it shoud be at least with respect to the spread within the available sectors, and have calculations giving a local minimum for the 30-30-30.
Now, of course that one has still to calibrate/weight all these differences from the mean (what is more important, industry or region?), but above is the sort of the main idea.