The energy study showed that insulation treatment caused a statistically significant, but small (0.7%-1.0%) fall in metered energy consumption. The small drop in energy use is consistent with an economic model in which energy efficiencies were obtained from the insulation so that the effective price of heating fell, in turn resulting in increased consumption of heat (i.e. a warmer house). Greatest energy savings were experienced in cool areas. Measured energy use was shown to increase slightly with the installation of clean heat installation (no data were available on nonmetered energy use).So free insulation will not help reduce energy demand: people respond to the reduced cost of heating by consuming more of it. This is worth knowing as some parties think that improved insulation is a substitute for greater generation capacity.
Health outcomes improved consequent to better heating. The MOTU reports found a 3.9:1 benefit-to-cost ratio, but that 71% of total benefits were from reduced mortality. As a public health intervention to reduce mortality, this could be fine. But it is hard to make a market failure case for the subsidy scheme.
Imagine two possible policies. Policy A gives cash to households and lets them choose whether to insulate their house with it; they're also given a pamphlet listing all of the benefits of insulation including increased life expectancy. Maybe it also has a nice narrative about how nice it is being in a warm house. There are lots of suggestions about how the money should be used for insulation, but it leaves the choice up to the household. Policy B gives a voucher for home insulation that can only be used for home insulation. If in the Policy A world households choose things other than insulation, can we really say that the the insulation subsidy programme passes cost-benefit analysis relative to the "give money to poorer households" programme? I don't think so.
Sure sure, there are other benefits where you could make a case: reduced hospitalisation and the like. But if 71% of the benefits were reduced mortality (ie reduced losses in VSL), then the benefit to cost ratio without the VSL gains drops to 1.131:1. And if people would demonstrate that they value other things by more than the VSL gains, it's hard for me to see the case for Policy B over Policy A.
Note: I've read only Grimes's summary and not the underlying work.