Arttu Ahonen

Abstract: Incomplete information on the voter-level impacts of efficient reforms can undermine their political viability by preventing redistribution that would compensate anyone facing more costs than they are willing to pay. However, a policy maker that observes proxies for voters' costs and willingnesses to pay can approximate the ideal transfers, making more voters benefit in expectation. In this study, I develop a general framework for analyzing how a policy maker maximizing the expected number of voters that benefit from a reform would allocate transfers, how much they gain politically from using the information they have on the voters, and how much the voters' private information costs them. I consider the empirical application of carbon pricing, providing the first description of transfer allocations that would maximize expected support for it among self-interested voters. I estimate voter costs and preferences from novel data that links individual-level administrative records with a government-backed poll eliciting respondents' willingness to bear randomly varied personal costs to achieve a real world abatement goal. While I find that support for the climate target clearly responds to the cost it imposes on voters, voter heterogeneity is ultimately more driven by variation in willingness to pay for the policy, only 10% of which is captured by the rich set of voter characteristics I allow the policy maker to observe. This means that the cost of private information for the political viability of carbon pricing is large: the expected number of voters that benefit from optimally targeted transfers is always much closer to the benchmark of equal transfers than to the upper bound that could be achieved if it was also possible to target on unobserved voter characteristics. Maximizing expected beneficiaries also comes with a potentially large opportunity cost in terms of welfare, and leads to discriminatory transfer allocations. This follows from a fundamental tension between the political objective and welfare, as well as fairness: the voters likeliest to be on the margin between winning and losing are not necessarily most in need of the money, or equally distributed along protected classes such as gender.

Abstract: How do our past residential neighborhoods shape our current habits? We study this in the context of driving in Helsinki. We exploit granular data on the universe of cars and individual residential locations in Finland to show that variation in car ownership and vehicle kilometers driven are explained to a much greater extent by differences between individuals than between neighborhoods. Furthermore, the neighborhoods where individuals used to live when they were underage explain a substantial amount of the variation in kilometers driven between individuals, accounting for many times more of the total variation than their current neighborhoods. We also show that growing up in a neighborhood with high car ownership makes drivers likely to drive more today, even conditional on car ownership in their current residential location and their household's characteristics. These results are suggestive of the importance of neighborhood exposure during early formative years in helping form habits that persist across residential locations and across decades.

Abstract: Using detailed administrative data on all cars and residents in Finland, we analyze the distribution of fuel tax burdens both across (the ``vertical'' dimension) and within income deciles (the ``horizontal'' dimension). We estimate household-level burdens using odometer readings recorded at mandatory vehicle inspections, combined with information on fuel economy and car ownership from the state's vehicle registry. Contrary to common belief, we find that fuel taxes are not regressive in the full population. Instead, regressivity emerges only among car-owning households. Rebating the tax revenues equally, or a small fraction of them progressively, produces a progressive distribution even among car owners. However, tax burdens are highly heterogeneous horizontally, with income deciles explaining only 1.5% of the total variation. This means that even an on-average progressive distribution includes many households with high costs, which may hinder the political viability of further carbon pricing and leave some low-income households vulnerable to transport poverty. Vulnerable households contribute little to total tax revenues and emissions, meaning that they could be fully reimbursed with relatively small costs. Indeed, the highest contributions come from high-income, high-driving suburban households, with technologies playing only a marginal role: 98.6% of the variation in car-level carbon emissions is attributable to kilometers driven rather than fuel economy. However, predicting who these high drivers are is difficult: even with our uniquely detailed data, observable household characteristics capture less than 50% of household-level variation in tax burdens. This severely limits the potential for leveling horizontal inequality via targeted compensation.