AndyH:
There is a very good economic paper done by some researchers at the Federal Reserve about two years ago. They build off of the generally accepted academic theory that the stock value should be equal to the present value of future income plus the current book value (equity), but perhaps discounted for equity that will get "used up" in the production of future income.
Their study shows that the only thing that investors pay attention to in pension accounting is the expense, and do not factor in any under or over-funding that is in the footnotes (which should be added to equity first under the general theory). However, other comprehensive income from an underfunded plan will affect investors' pricing of a stock in a direct manner (because of the immediate decrease in equity), without relating it to being a pension accounting item. In other words, investors are only looking at net equity and not adjusting it.
So, your comment is correct on companies that did record an additional minimum liability in recent years. When interest rates go up and they reverse this out, it will have an effect on stock price because of the immediate effect on equity.
On the flip side, the change caused by interest rates in over or under-funding for plans that have not generated an additional minimum liability will have very little effect on stock price, except to the extent that the expense figure changes (and then it needs to be a large compenent of overall net income to have a real effect).
So, yes, buy large caps, but only those with previous AML.
The Fed paper (Did Pension Accounting Contribute to the Stock Market Bubble? by Coronado and Sharpe - it doesn't say so, but they are with the Federal Reserve) is the last one at the bottom of the linked page (when I first read it, a group of us were in the middle of organizing the financial economics & pensions seminar in Vancouver last summer; I immediately wanted the paper to be the keynote address):
http://library.soa.org/library-html/m-rs04...ofcontents.html