Markets are not complete. Two major ways in whuch they are not complete is that we have borrowing constraints and that we cannot exchange with future generations. The latter can be a big deal when we think about the valuation of future amenities (like the environment) or long term risks. In particular, future generations could make us behave in certain ways if they could influence some of today's markets. This is precisely why the overlapping generation literature emerged, and a principal conclusion of it is that the government needs to intervene, in particular by providing an security that lives beyond generations: the government bond. While there is obviously a welfare cost to the lack of future generations on current markets, how large is it? The literature tells us the welfare benefit of the government bond is large.
Roel Mehlkopf just defended a dissertation on this topic, focusing on risk. In a nutshell, the cost is not that large, and it all has to do with distortions on the labor market. For one, those you ex-post need to transfer to another generation face a commitment problem in the sense that they want to reduce their labor supply, for example by retiring early. Once you take this into account, there is little to redistribute, and it can even be welfare-decreasing to transfer. This rationalizes why pension funds needs to be solvent at all times, even if they are solvent in the long run. One important implication is that when cuts are necessary in pensions, they should be larger for the young workers, as this reduces the labor market distortions.
Also, the dissertation points out that comparing to a situation with fictitious markets between non-overlapping generations can be misleading. Indeed, this implies that they all have the same weight in a social welfare sense. But there can be good reason for a social planner to deviate from this, and the analysis above, fr example, implies that future generations benefit more from risk sharing than current ones (who are at least partially locked in by past decisions). This should entice the social planner to give more weight to current generations, even beyond normal discounting of the future. And as only current generations for for the current government, we are not far from that optimum.