I would require independent cost-benefit analysis done for proposed legislation and for regulatory agency decisions that affect major sections of the economy or human life (canonical examples being the FDA and NRC in the US that are pretty much only required to consider costs today). Ditto permitting for construction projects of all kinds, I want to reduce the number of veto points and consider all the costs and benefits at once instead of serially.
I would replace complex, detailed regulations with market-based ones wherever possible. E.g. we'd have had no need for complicated climate policies/subsidies/mandates/treaties if we'd set up a carbon tax or fee-and-dividend system with built-in annual percentage increases back when we first realized this was a problem in the ~1990s.
Ditto for public services and programs: cut the bureaucracy and means testing and general nanny-ing as much as possible, and accept that this will lead to some misuse will still being a benefit on average.
In general, tax negative externalities far more heavily. Pollution (including noise and light), waste, health impacts of products.
Redirect subsidies from the elderly to younger families. Lack of population growth is a much bigger threat to the well-being of retirees, long term, than most of what we actually focus on.
Increase funding for basic research across all the sciences.
Increase support for scaling up promising innovations, making the valleys of death smaller. Mainly an issue for hardtech, which is a big part of why we focused so much on software innovation and fintech etc. for so long. Quit reporting too much on the very-much-expected cases where this fails anyway and "wastes" public money.
Generally encourage minding-your-own-business on the margin relative to today.\
Generally accept that there is a limit to how quickly society can assimilate new ideas and norms, which changes over time but which is essentially independent of how good an idea something is. Discourage demonizing those who are just slower overall to adopt new ideas, it never helps in the long run.