These developments mean that unless Trump can become more popular with upper-income, educated, more moderate Romney-Clinton voters or find a way to win more working-class non-white Democrats, he can only win re-election by again losing the popular vote and winning the Electoral College. Should this happen, it would be first time in our history that two consecutive presidential elections went against the popular vote winner.

Demographic changes also make it more likely that Trump’s—or any Republican building a coalition like Trump’s—vote deficit is likely to grow. That’s because America’s younger citizens are much likelier to be non-white, and much less likely to vote Republican, than the older whites who will pass away in the coming decades. A state-level analysis of political trends and demographic change by the States of Change project demonstrates this clearly. If President Trump receives the same shares of the vote among different demographic groups in 2020 as he did in 2016, even the small changes that will have occurred would cost him the crucial Midwestern states of Wisconsin, Michigan, and Pennsylvania and hence the election. If he increases his share of the vote among non-college educated whites, however, he and his successors could win the Electoral College through 2036 even as the GOP nominee loses the popular vote by ever increasing margins.

Should this arise, calls to change the Electoral College—or moves by Democrats to ratify the National Vote Compact—will surely increase.