Thread by @lymanstoneky

The One, True, Correct way to assess public support for a policy is to ask, "Would you support X if it meant you had to pay Y higher in taxes?" where Y is simply the budgetary cost divided by the number of households in the country.

I will also accept actually doing a full willingness-to-pay framework as valid, however it's a massive pain to administer.

No, because peoples' income changes over their lifetime. Within-lifetime inequality is very large, and most peoples' average lifetime income will be closer to the national average than their year-specific income!

https://twitter.com/mathIsMight/status/1382390094229860354

spoiler alert:

you can predict a person's income better if you know their age than if you know the unemployment rate that year or their parent's income!

So when we ask people about *permanent* policy changes we should *not* quote them a price at their *current* income. In principle we should model their future income distribution, calculate taxes, then quote the average they'd pay per year over remaining life expectancy.

But that's hugely difficult to do in a survey. So quoting people the average tax cost is perfectly reasonable.

Calculating point-in-time economic incidence of taxation is hard enough. I don't know that anybody has ever even bothered to *try* to calculate *lifetime* economic incidence. We barely even have reliable estimates of lifetime income in terms of policy-usable metrics.

So in general we should make a simplifying and egalitarian assumption.

Alternatively, I'm okay with saying, "assume 25% of the cost will be paid by people over $1 mil in income and that they don't take surveys." So you only have to divide 75% of the cost out.

But like, we know that corporate taxes get passed on to workers and shareholders (i.e. retirees and pension funds). We know that individual income is life-cycle dependent so taxes you avoid now hit you later when your income changes.

Furthermore, we should not test polls by asking people to consider ***the most favorable possible hypothetical***, but rather ***a plausible hypothetical***. And a plausible hypothetical is that taxes will not be applied on some pure ability-to-pay heuristic.

An even more plausible hypothetical is that no taxes will go up at all, the program will be deficit financed, and in 20 years crowding-out effects will reduce growth rates leading to disproportionately negative consequences for poor people and the next generation.

In which case technically we should ask, like, "Would you support X if it increased your child's odds of being in poverty by Y?"

But that gets very wonky and weird. Much simpler to just ask about a budget constraint.

Of course you can also just ask, "HOW MUCH would you be willing to pay in extra taxes to get policy X?"

If the amount people are WTP equals the policy cost, you've got a winner!

But this would certainly fail, because rich people would report low WTP. Obviously, our system relies on transfering from those with low WTP to those with high WTP.

So instead you could have a question like, "Which would you rather have, X policy, or Y dollars in a cash check?" and you vary Y randomly across respondents. That gives you a subjective valuation of the policy.

The reason you need to survey budget constraints by the way is because respondents already *imprecisely* consider them.

Why do conservatives tend to oppose policy X? Sometimes, they just hate policy X. But often, it's because they believe it'll lead to tax hikes!

So you mis-measure actual beliefs about policies when you don't provide a budget constraint, because conservatives will tend to invent a budget constraint in their head, while progressives will tend to assume the burden will be borne by amorphous moneyed payers.

This actually causes partisan differences to be overstated and leads to misunderstandings about partisan disagreements.

If the proposal were, "Imagine the US government just found a massive trove of gold on Federal land and is selling it: how should the money be used?"

And then you provide a list of options, I guarantee you even the rank-order of policy preferences is going to change! People treat free money differently!

If I find $100 lying on the street, folks, I have some DLC I want.

But if I worked for that $100, y'all, baby needs DIAPERS.

Gonna stop replying to this thread because this is a classic case where I'm waaaay too correct for twitter to handle.