Well, the statistical evidence we have on that is from a couple of studies that were done a few years ago. One was in an international context and just looked at replacement rates right after widowhood. Generally, they showed that right after the spouse dies, it really wasn't an issue. But we're always looking at averages, medians, means, so there are always cases at the ends of the distribution where things are a little bit different.
If we look at a Canadian study that was a little bit longer term, it looked at the income tax data, and again they found there was not a lot of evidence, at least in talking about the means or medians, in terms of immediate large declines in income replacement rates once you take account of the fact that the number of people in the household had been reduced. However, over the following five years, the income declines far more for widows than it does for widowers. So over time you get a greater gap between the incomes of the two groups. That corresponds to what we see in terms of the GIS rates, where the GIS rates for women definitely seem to be going up as they age.