Last Thursday, we discussed Amy Finkelstein's very interesting above-named article, which shows that adoption of EZ-Pass appears to lead to higher tolls, as well as to reduced consumer elasticity of response to the tolls, because not paying cash at the tollbooth reduces one's awareness of the charge. Oddly, EZ-Pass users appeared to over-estimate rather than under-estimate the toll they were paying, which might seem to make the empirical response backwards, except that it's consistent with the notion that people under-measure CHANGES in the toll. Also, more from our colloquy than from the paper itself, it seems that people simply were not thinking much about the toll because they didn't have to pay it in cash - one had to prod them to get an estimate of its likely level (raising a question of whether they were really thinking in terms of the cost estimates they eventually furnished).
Tentative conclusions: hard to doubt that EZ-Pass is good for us on balance as consumers, especially since its use as voluntary. Is it good for us as voters? (A key motivation of the paper was to test the Milton Friedman idea, a la withholding, that making tax payments less salient or noticeable leads to a bigger government.) Here the problem is that political choice has so many flaws (aggregation problems, externalities, collective action) that anyone trying to say whether one more defect, in the form of reduced understanding of one specific element, makes things better or worse faces severe second-best problems. Easy to be concerned about it if one has Milton Friedman-James Buchanan type priors, in which government systematically is too big and does too much. But public choice defects apply to supplying valuable public goods in addition to everything else, so unless one has the Friedman-Buchanan prior (or, rather, settled faith) it's hard to know where EZ-Pass-type reduced salience sends us relative to the optimum.