This is getting ridiculous (Litterman Chapter 10)

[Will Durant]

page 110

Our numerical results show that there is a dichotomy between optimal asset allocations for over- and under-funded plans. The latter must take a large amount of equity risk in order to improve their funding status.

page 118

A fund with a deficit is better off investing in bonds because they offer a better hedge against changes in the value of liabilities leading to a lower surplus volatility.

I assume that in the first section they are looking at it from an expected return point of view while in the second section they are looking at it from an expected volatility point of view. However, neither section actually provides this background. Each of the two (contradictory) suggestions are stated in absolute terms.

[Bell]

I stumbled on those two too:

(i) I think the first one is to to with the RACS. Overfunded pensions plan will have an increasing RACS up to a certain level, at which it will start to decline, as a function of equity allocation. Therefore, it is not optimal for an overfunded pension plan, which seeks to increase its RACS, to have an all equity investment strategy.
However, for an underfunded pension plan, the RACS increases with the equity allocation..... and the story basically is underfunded funds, seeking to increase their RACS shall increase their equity allocation.

(ii) The second statement is more about matching assets to liabilities. Like you said, about the volatility of the surplus. To reduce that volatility, the underfunded is better off investing in Bonds.

I could be wrong.

Good luck.

[Will Durant]

I agree with what you wrote, and it makes perfect sense.

But that's not what the textbook wrote ... in one place it says an underfunded plan MUST invest in equities, and in another place it says an underfunded plan is better off investing in bonds. In neither place did they qualify the context under which these statements are true.

I am continually appalled at how poor the syllabus readings are and how much a candidate needs to infer. No wonder the pass rate is < 50%.

Specifically I'm thinking of things like this
http://www.actuarialoutpost.com/actuarial_discussion_forum/showthread.php?t=131020
where what's in the text is just flat-out wrong nonsense.

This reading is not quite as meaningless, but it's close.