Although we know that public policies have different effects across a population, examination of the distributional effects has been largely neglected. Instead, the principal criteria now used in public policy assessments are measurements of net benefits, which estimate the sum of a policy’s benefits minus its costs, and benefit–cost ratios. But using these criteria can result in a policy that is viewed as having positive overall effects while still generating many more losers than winners, obscuring its true welfare impact.
In order to better understand the true welfare effects of public policies, together with my colleagues, we explore in a new RFF discussion paper how a disaggregated general equilibrium model helps us better understand the ways that policies affect people at various locations and income levels. To do this, we use a computable general equilibrium model (CGE) developed at RFF known as the Land Use, Surface Transportation, and Regional Economics (LUSTRE) model, which simulates urban transportation and land policies in a complex and realistic setting. In our examination of data from the DC metro area, we demonstrate how this model could shed light on policy distribution outcomes by using it to compare the outcomes of four local anti-sprawl policies. These include three variants of “Live Near Your Work” (LNYW) policies, in which homeowners are offered modest subsidies for home or employment relocation to reduce commuting costs, and a vehicle miles traveled (VMT) tax, which taxes a driver’s total miles in a metropolitan area.
Our research reveals that all versions of LNYW policies, which are already used in a few states, are inefficient, and become even more so if our simplifying assumptions about program financing are relaxed. The LNYW policies only had a moderate transportation impact, inducing few residents to change their mode of transportation or reduce the length of auto trips. By comparison, the VMT tax had a much larger impact on transportation and remained efficient.
Though there is nothing new in finding an efficiency advantage of policies that penalize driving rather than encourage relocation, we also find that the VMT tax has a more progressive distributional profile than any of the LNYW policies we studied. Progressive distribution is an important factor in this case; “promoting fairness” is one of the characteristics used to justify LNYW policies, but the features of those policies that promote fairness (such as restriction of eligibility to lower-income groups) actually reduced efficiency while only modestly improving their distributional features.
Ultimately, we found that the differences in model results vary substantially according to each policy’s details, and for each policy by income level and location. In our welfare distribution assessment, a VMT tax set at about 2 cent per mile was able to produce “smart” growth with progressive benefits, characteristics which were not seen in any of our LNYW policies. The development of disaggregated behavioral models such as those we employed is and will likely continue to be complicated, but models such as LUSTRE can be useful to understanding the consequences of such regulations.
