Each electoral system creates specific incentives to (mis)allocate government resources. Would putting the National Popular Vote (NPV) in lieu of the Electoral College improve matters, or just replace old political distortions with new ones?

In the aftermath of the Trump Presidency and the voting reforms in various states, the debate about the US electoral system is back high on the agenda. A recurring proposal is to reform the Electoral College, often in favor of adopting the National Popular Vote (NPV). Under the latter system, the winner of the presidential election would be the candidate obtaining the largest number of votes across the whole country. This proposal, in its NPV Interstate Compact form, has been adopted by 16 states and DC. But, by construction, it will go into effect only if it is adopted by states representing at least 270 electoral votes.

Proponents of the reform argue that the new system would resolve most issues present in the Electoral College system, in particular the possibility of electing a president without a majority of the votes, malapportionment, and the excessive targeting of battleground states. However, they rarely mention what would happen under the new system. We explore the consequences of various reforms in a recently published academic article.

All electoral systems give politicians incentives to target policies at specific groups of voters. For instance, it is widely known that the Electoral College gives presidential candidates and parties incentives to “abandon” some states—they only need a fraction of the 50 US states to win a majority in the Electoral College—while they devote most of their attention and energy to so-called “battleground” states like Arizona, Florida, Georgia, Pennsylvania, and Michigan. The unlucky states are typically the “partisan” states, which are least likely to switch from the Republican to the Democratic column, and conversely. The empirical literature confirms this common wisdom: battleground states disproportionately benefit from federal funding and campaign effort.

Targeting—i.e., treating groups of voters differently—is not undesirable per se. Building a park in a densely-populated area is more beneficial to society than building the same park in a sparsely populated one. Yet, targeting can become socially undesirable when it is only meant to improve a politician’s (re-)election chances. For instance, politicians could prefer to build the park in an area with fewer citizens just because those citizens are also more likely to turn out to vote, or more likely to contribute money and time to their electoral campaign. In such cases, politics “distorts” governmental interventions, which is bad for society as a whole. The targeting of battleground states simply because they are electorally relevant is a clear instance of political distortions.

Adopting the NPV would eliminate the incentive to target battleground states and abandon partisan states. Candidates would have to look for votes everywhere in the country: battleground states, blue states, and red states. Now, this is not enough to justify the reform. We also need to take into account the new political incentives that would develop under the reformed system. These too will distort the allocation of governmental resources, although in favor of other groups of voters. There is no escape from political distortions.

One way to anticipate which new distortions may arise is to rely on analytical models: we find that, should the NPV be adopted, candidates would target communities for which one extra dollar of spending wins the most votes, independently of the state they belong to. These are communities with, for instance, large populations—i.e., cities—with high turnout rates, and with many independents voters. Removing the need to win different US states could therefore unleash an intense targeting of those communities. This is particularly acute when one takes into account that the targeting of governmental resources can be done at a small geographical scale (e.g., counties), and not just at the state level as assumed by traditional analyses. Why? Simply because smaller communities such as counties vary much more in size, turnout, and number of independent voters, than states.

For instance, our data shows that in the 2016 presidential election Maine was the state with the highest turnout rate, whereas Texas was the state with the lowest turnout rate. Turnout rate in the former was 1.74 times higher than in the latter. A similar comparison at the county level increases the ratio by a factor of 3: Mineral County in Colorado had a turnout rate 5.76 times higher than Chattahoochee County in Georgia. Focusing on population size, the ratio is 67 at the state level, and 86,500 at the county level.

“Adopting the National Popular Vote would eliminate the incentive to target battleground states and abandon partisan states. Candidates would have to look for votes everywhere in the country: battleground states, blue states, and red states.”

The NPV will not simply eliminate all targeting and political distortions. Informing the debate requires shifting towards a more careful analysis of (i) whether there will be more or less targeting under the new system, and (ii) whether this targeting will be more or less desirable socially. The answers to these questions depend on the specifics of the situation at the time of the reform: Are there many or few swing states? Are there large variations in politically relevant characteristics (population, turnout rate, number of independent voters) across and within states?

We performed this exercise using numerical simulations based on US historical data from 1980 to 2016. Our model predicts that the NPV would increase targeting but decrease waste and political distortions. This is because under the NPV, differences in government intervention across counties would mainly stem from differences in population size. As hinted at, targeting based on population is generally deemed socially desirable in economic models. But, economic efficiency may not be the sole criterion: if the objective is to reduce differences in spending between counties–independently of their size–then our model predicts a worsening following the adoption of the NPV. Our model also predicts that a constitutionally feasible hybrid system, which maintains the Electoral College but allocates the electoral votes of each state in proportion to the vote shares of each candidate within that state, would lead to fewer political distortions than the NPV.

We also considered the redistributive effects of adopting the NPV and find that public resources would flow to a very different set of communities. For instance, California and Arkansas would receive more, whereas Florida and Arizona would receive less. The presence of winners and losers are an obvious obstacle to this possible reform. Perhaps of interest to some, according to our numerical simulations, a few of the states that have not yet enacted the national compact, such as Alabama, Louisiana, Oklahoma, and Utah, could end up winning some extra resources after that reform, whereas states like New Mexico and Virginia could end up losing some.