What if We Used the Electoral College to Elect State Governors?

by Nathan Nicholson // Published October 15, 2014

Interested in this topic? Sign up to receive our newsletter and other updates on elections and electoral reform. 


Few causes are easier to champion than a status quo, especially a status quo of long standing—like the Electoral College rules used in nearly all states today, under which the presidential candidate who wins a state's popular vote receives all of that state's electoral votes. It would be unfair to portray resistance to states’ enacting a National Popular Vote plan for president as nothing more than a reflexive resistance to change, but the long entrenchment of the state-winner-take-all method of apportioning electoral votes does call for caution: it’s easy to shy away irrationally from changing a system that’s been in use for so long, even if its chronic shortcomings are clear and pervasive. In trying to assess such a longstanding institution, one obvious question has to be: would we design a system like this today? And another: if given the choice, would we extend its principles to other similar systems?

The current winner-take-all Electoral College* is so baked into American political thought that its features are hard to examine with fresh eyes. But we can use a simple thought experiment to explore these questions. All 50 states currently elect their governors by a statewide popular vote (for that matter, all members of Congress and state legislators are elected by popular vote too). But suppose a state instead decided to model its gubernatorial elections on a winner-take-all Electoral College-esque system, perhaps having each county vote for “gubernatorial electors” in the same way each state does for presidential electors now?

The presidential Electoral College gives each state a number of electoral votes equal to the size of its House delegation (apportioned by state population for a total of 435) plus the size of its Senate delegation (a flat two per state for a total of 100). Combined with 3 votes for the District of Columbia, this makes for a grand total of 538 electoral votes. The 538 figure is arbitrary but iconic, so for our thought experiment we’ll loosely mirror the real-world Electoral College by giving each county a flat two electoral votes and then assigning the remainder by population** until we get to 538. As in the real Electoral College under today's winner-take-all rules, whichever candidate for governor wins a majority of the popular vote within a given county receives all of that county’s electoral votes, and a candidate needs 270 electoral votes to win the governorship.

What would happen if, say, Pennsylvania were to elect its governor this way? Take a look:


Each of the state’s 67 counties is labeled with its electoral vote allotment and color-shaded according to its 2012 Cook Partisan Voting Index or PVI, which describes voters’ underlying partisan preferences (measured by their votes in the 2012 presidential election relative to the nationwide vote that year). Deep red or blue counties respectively lean heavily Republican or heavily Democratic, with grey representing an even split.

What’s quickly clear is that the zone of meaningful competition becomes very small. There’s simply no chance of a Republican candidate’s winning Philadelphia or of a Democrat’s winning any of the deep-red outlying counties, so those areas would drop off the campaign map—time spent by either side persuading or turning out voters in those counties would be wasted, as all of their electoral votes would be a lock for a single party. Instead, the important targets would be the relatively few light-shaded “swing” counties, those whose partisanship happens to hover right around 50%. Bucks, Northampton, Luzerne, and Dauphin would be the most tempting prizes. A Republican candidate might reach for Allegheny’s 42 votes in a good year, and a Democrat might conceivably put resources into the western swing trio of Mercer, Lawrence, and Beaver, but on the whole the race would be won or lost on the strength of the outer Delaware and Lehigh Valleys, with the cities of Harrisburg, Wilkes-Barre, and their suburbs in supporting roles.

If we generously define swing counties as those with less than 55% support for either party, Pennsylvania has only 10, containing barely a quarter of the state’s population. In a governor’s race using a state electoral college, these counties would be the equivalent of swing states like Ohio and Florida in a presidential campaign, receiving the great bulk of attention and campaign resources and being targeted by relentless get-out-the-vote efforts. Meanwhile, nearly another quarter of the state resides in counties with 65% or greater support for a single party—counties in which the governor’s race would be a foregone conclusion, as Texas, California, and New York are today in presidential races. If presidential elections are any indication, these landslide counties would see substantially lower turnout rates, lower campaign spending, and fewer campaign visits than swing counties.

Like the presidential Electoral College, a Pennsylvania electoral college would risk producing wrong-way winners—situations in which a candidate receives enough electoral votes to win while losing the popular vote. As little as 20.4% of the population could in theory combine for 270 electoral votes in Pennsylvania.

Also like the presidential Electoral College, a Pennsylvania electoral college would structurally favor one party over the other, in this case the GOP. In a “neutral” year in which each county voted according to its baseline partisanship, the Republican candidate would win, 323-215. (This, too, would be a wrong-way winner scenario, with the Democratic candidate narrowly winning the statewide popular majority. In fact, a Democratic candidate for governor would likely lose even if winning 57% of the two-party vote.) That’s partly because of the two-electoral-vote minimum, which gives outsized representation to the deep-red small counties, and partly because Democrats are inefficiently concentrated in Philadelphia County, in which Democratic partisanship is an astonishing 84.2%.

In the real world of presidential elections, such partisan advantages have gone back and forth over the years.  Today it’s actually Democrats who enjoy a similar advantage nationwide: despite Republicans’ reaping the same benefit of two baseline votes per state (which, recall, is designed as a reflection of each states’ two U.S. Senators), at a national level Democrats have an edge in a majority of the large competitive states. Nate Silver has calculated that Barack Obama could have won the 2012 election even had Mitt Romney won the national popular vote by two percentage points, for example – a finding supported by FairVote’s November 2012 analysis, although an advantage tenuous enough to be reversed by a state as small as Iowa or Colorado starting to tilt toward Republicans.

Pennsylvania isn’t a cherry-picked case—we can see the same patterns in almost any good-sized state. Let’s take a look at a prospective race for governor of Oregon using an Electoral College-type system:

OR EC Map no border bmp

Again, most of the map drops out of competition. The urban enclaves of Portland, Hillsboro, Eugene, Corvallis and their surroundings become invincible Democratic strongholds, while the rural southern and eastern reaches of the state are immutably Republican. Clackamas, Columbia, and Tillamook Counties become the only real battlegrounds. Just 7 out of 36 counties, containing under a quarter of the state’s residents, are within 5% of even partisanship, while almost a third of the state lives in landslide counties whose votes are unlikely ever to be in doubt. In a neutral year, a Republican gubernatorial candidate would win the electoral college vote 304-234, despite a Democratic win in the statewide popular vote by nearly four percentage points. Clackamas County’s voters would have the luxury of being kingmakers—or the misfortune, depending on how much one enjoys campaign ads and canvassers: the county is almost perfectly split between Republicans and Democrats, and its 48 electoral votes would nearly always hand its winner the keys to the governor’s mansion in an otherwise close race.

These hypotheticals make it clear that imposing a winner-take-all Electoral College-like system on statewide elections would have bizarre and undemocratic effects. It would arbitrarily give some voters more clout than others depending on the partisan makeup of their region; it would create structural advantages for one party; and it would risk allowing a candidate to win even if more people voted for a different candidate. These problems, of course, are the same problems that plague every real-world presidential election carried out under the winner-take-all Electoral College in the United States, as laid out by FairVote’s Rob Richie and Andrea Levien in their 2013 article in Presidential Studies Quarterly.

To be sure, states aren’t composed of counties in the same way that the United States is composed of states, so there’s an asymmetry here: the U.S. Electoral College has a historical and philosophical grounding that a state electoral college would lack. But if the winner-take-all Electoral College is really a good way to elect a powerful chief executive from a big and diverse population, then there’s no reason why that effectiveness shouldn’t translate to the level of a statewide race. (It’s worth noting that at least 26 states currently have more people than the entire country did at the time of the first US Census in 1790.) Instead, its failures only become more apparent.

Under state-winner-take-all rules, the Electoral College is anything but a harmonious fulfillment of the Founders’ democratic ideals. It’s long since diverged from its original intent of making all states matter in selection of the president. There’s no reason to hold it up as an example of good democratic design: no rational framer would create a state-winner-take-all rule today, and attempting to extend its principles to other elections only goes to show how unwieldy and undemocratic it is. It’s time to shake ourselves awake from the lull of the status quo and reform the Electoral College.


 *The Electoral College, precisely defined, is the mechanism by which states are granted the power to elect the president under the Constitution: each state chooses a number of presidential electors based on the size of its congressional delegation, and those electors vote for the president. As often used casually today, by contrast, the term “Electoral College” refers to the specific means by which most states currently choose their electors: by having all of their electors vote for the presidential candidate who wins the popular vote within that state. But this method isn’t specified in the Constitution—it evolved over time at the discretion of the states, and in fact not all states use it (Maine and Nebraska choose their electors semi-proportionally rather than by winner-take-all). Hence we refer here to the “winner-take-all Electoral College” to describe current policy. The terminology distinction is tedious but important, because phrases like “abolish the Electoral College” are often used inaccurately to describe proposals like the National Popular Vote Interstate Compact, which in fact preserves the constitutional structure of the Electoral College in full and changes only the way in which states wield their power to choose electors.


**The apportionment of the U.S. House is computed by a statutorily defined algorithm that’s designed to avoid a paradox in which a state risks losing representatives if the size of the House is increased. In our casual experiment we simply use the county’s population as a percentage of the state’s population to assign each county a whole-number share of the remaining electoral votes out of 538 after the baseline 2 votes per county have been assigned. The formula for each county’s electoral votes is therefore:

county population / state population * (538 – number of counties * 2) + 2

Rounded to the nearest whole number. If this results in a total other than 538 due to rounding, we make up the difference by adding or subtracting 1 vote from the county whose vote/population ratio is furthest from the statewide vote/population ratio and repeating this process until the counties’ votes total 538.

Population data source: US Census, 2010.

County PVI data source: Dave Leip’s Atlas of U.S. Presidential Elections, via David Jarman.