WhatDoEconomicAgentsKnow

The formal model in BayesianVotingGame requires the assumption that agents have knowledge of the distribution function

F(c) of players' costs.

Nowadays this kind of thing is standard in theoretical economics. Of course, it is also completely unrealistic. Or am I being picky? I swallow all kinds of extreme assumptions in economic and political theory so why baulk now?

One way of getting round this issue is to take a view something like this:

Economic models aren't 'really' saying that agents have this kind of knowledge. Rather, saying 'agents know the distribution function' is a kind of shorthand for saying – 'they act as if they know ….'.

In this vein, the (in)famous Rational Expectations hypothesis in macroeconomics is often interpreted as 'Model-consistent expectations.' I.e., agents act as if they have internalised the information specified in the model. And it's not economists' job to worry about what's actually going on in peoples' heads. If economics theories are testable at all, the test is whether the statistics – 'stylised facts' and econometrics - fit the models.

But is this enough? When I was thinking about the voting model I wanted something that makes intuitive sense. I am not sure that the BayesianVotingGame does, even if the implications are correct. I think the 'handwaving' version in StrategicVotingModel does make some kind of intuitive sense, but is it formally correct?

I thought about the voting issue in terms of 'guesstimates' before I read the Palfrey & Rosenthal paper and I came up with the same basic structure as their model, but I didn’t know how to get a rigorous definition of equilibrium in these terms (not that someone who really understands this stuff couldn’t do it.) The formal model has a formal equilibrium, but to do so it makes assumptions I can't explain intuitively (again, not that someone who really understands this stuff couldn't.)