US kids are taught a myth in civics classes, that every vote counts. In presidential election this is certainly not true. The fact that the winner takes all electors when winning a state and some states consistently vote for one party (and if not, it is a landslide) makes that many states become completely irrelevant in the campaign. It is all about the battleground states.
There are two ways in which campaigns realize that not every vote is equal: they allocate unevenly their resources, both in time (candidate visits) and money (TV ads). Do they do this in a manner that would be predicted by a model of electoral competition? How would these allocations change if the electoral college were replaced with another system? These answers are answered by David Strömberg. The critical thing here is to measure the uncertainty that a state would vote for either camp and determine how visits (and TV ads) influence this likelihood.
It turns out the proposed model works remarkably well to explain the actual campaign decision. Or put in another way, the campaigns seem indeed to optimize following such a model. Now that we have estimated the components of the model, one can look at alternative election scheme. This is where it becomes really interesting. If one lets electors be attributed proportionally to the number of voters in each state or even switches to a direct vote, the battleground state become significantly less interesting. Large states do not benefit much, however, due to the decreasing returns of visits. It is the small states that are big beneficiaries of this system, both because of the higher return of a visit, but also because they have disproportionately many electors to begin with.
It is time to make all states relevant. There have been previous attempts to reform the presidential election. One other aspect that a reform would bring, as demonstrated by Strömberg, is that the likelihood of razor thin victories would decrease 40-fold. Not good for the 24-hour news channel, but good for the country.