The Electoral College Is Broken. We Should Make It Bigger.

If you’re familiar with American presidential politics, you’re familiar with some variation on the following map. It’s been the baseline output of national elections for a while now. Big cities vote for Democrats; rural areas vote for Republicans.


County-Level Presidential Winner, 2016. Clinton in blue, Trump in red.

This pattern poses a unique problem for the way Americans elect our presidents. The argument for this is relatively straightforward, so I’ll lay it out before diving into the evidence. First, though, we need a little background on the Electoral College. The number of votes that each state gets is equal to its number of senators plus its number of representatives. Each state gets two senators and at least one representative, which means that every state gets at least three electoral votes. Bigger states get more, because the number of representatives is proportional to a state’s population, but the fact that every state, regardless of size, has two senators gives small states a potent electoral advantage.

With that in mind, note the following trends. First, Americans are increasingly living in big cities. While this pattern is true across the country, it’s especially true in the biggest states. And people living in big cities have particular political preferences that are distinct from people living outside big cities. The electoral map above attests to this.

Second, the Electoral College doesn’t represent highly urbanized states very well. Actually, it doesn’t represent large states very well. But large states on their own don’t have different representational needs than small states do: hypothetically, a really big state that somehow looked a lot like Wyoming might do fine in a system where it doesn’t get as many votes as it should, because Wyoming is there, voting for the same candidates this hypothetical big state would vote for, picking up the slack with its extra votes.1 The problem comes when voting blocs develop in a way that overlaps with the unrepresentativeness in the system. And that’s what’s happening with urbanization.

The conclusion is pretty pessimistic. Increasing urbanization, especially in big states, along with an Electoral College system that gives extra votes to smaller states increases the probability of splits in the popular and electoral vote like we saw this year and in 2000. And a system where winning a majority of votes doesn’t translate into winning the government isn’t a very stable one. However, there are a few ways that we could fix this problem, including at least one that doesn’t involve doing away with the Electoral College.


The first point to make is that way more people live in cities than they used to. The Census Bureau provides data on urbanization rates — the percentage of people in each state that live in cities — going back to 1790.


Urbanization Rates Over Time

Moreover, urbanization is especially — although not exclusively — prevalent in the most populated states. For instance, in 2010 95% of California’s 37 million people lived in cities. The second biggest state, Texas, had an 85% urbanization rate, and the third biggest, New York, sat at 88%. All of these states are more urbanized than the average state; in fact, the nine largest states are all more urbanized than average. (In the graph, the red line represents the national average.)


The Relationship Between Urbanization and Population (2010)

The first point, then, is that states within the U.S. have become far more urbanized than they used to be. And, those increases in urbanization are especially concentrated in the most populous states. If urbanization is linked to patterns of unrepresentativeness in electoral politics, we’ll have a problem on our hands.


The second point to make is that urbanized states are at a particular disadvantage in the Electoral College. Again, mathematically this is in part because these states tend to be larger. But this disadvantage poses unique representational consequences because of the fact that cities tend to vote differently from rural areas, and because — obviously — the largeness of these states is due primarily to large cities.

One statistic that is useful for illustrating this relationship is the Population Per Electoral Vote, or PPEV. The PPEV is simply equal to each state’s population divided by the number of electoral votes it receives in national elections. It’s a decent approximation of how well-represented each state is in the Electoral College. States with higher PPEV, like California (PPEV in 2010 of 670,000), are less well-represented. States with lower PPEV, like Wyoming (PPEV in 2010 of 190,000), are more well-represented.

One way to approach the question of whether or not highly urbanized states are at an Electoral College disadvantage is to simply plot urbanization rates across Population Per Electoral Vote. Here I’ll plot the level of urbanization across the PPEV for every state in the years 1800, 1860, 1920, 1980, and 2010.2


And overall, it sure seems like there is a pretty strong, positive relationship between urbanization and Population Per Electoral Vote. If this relationship is true, it would be evidence that states with higher urbanization rates tend to have worse representation in the Electoral College (remember that lower values of Population Per Electoral Vote are “better”). But the aggregate data might be masking variation over time so I broke it down by year and drew fitted lines for each year:

When we examine the correlation within each year, a much more nuanced pattern emerges. We can interpret the “steepness” of the lines as the degree of correlation between urbanization and Population Per Electoral Vote: if there is no correlation, the line is flat. If there is a very high positive correlation — that is, if urbanized states tend to be much less represented in the Electoral College — then the line is sloped up and to the right.

In the early days of the United States, there appears to have actually been a negative relationship between urbanization and Electoral College representativeness. The lines for 1800 and 1860 are slightly downward sloping. But starting in 1920, the fitted line begins to slope upward, and becomes more sharply sloped as time goes on.3 In fact the line for 2010 is the most sharply sloped of them all (if only slightly).4

We can tell a few things from this graph. First, both urbanization and electoral college unrepresentativeness have grown over time. As we saw earlier, urbanization has been on the rise basically since the United States was founded, and this graph provides another illustration of this fact — more of the dots are in the upper part of the graph in 2010 than they were in 1800. Even the least urbanized states in the twenty-first century are more urbanized than the most urbanized states in the beginning of the nineteenth century. But the Electoral College has also become less representative of the states over time, as evidenced by the shift along the horizontal axis: the number of electoral votes hasn’t kept pace with the increase in population.5

Second, the relationship between urbanization and electoral college representativeness has also grown over time. This point is illustrated by how sloped the lines are. In every year since 1920, highly populated states have become less well-represented relative to their less populated neighbors. Together, these two facts lead to a disturbing conclusion: the Electoral College doesn’t represent states as well as it used to, and the bulk of this unrepresentativeness falls on the largest and most populous states. And it’s getting worse.


I said at the beginning that increased urbanization means that splits in the popular and electoral vote — like in 2000 and 2016 — are becoming more likely. Why is this the case?

Imagine there’s a country that follows the United States’ same electoral laws but only has ten states holding 100 electoral votes between them. In the first year this country exists, the population is evenly divided between the states, so that each state has exactly ten electoral votes. In this setting, the Electoral College is completely fair: each state has ten percent of the population, and ten percent of the votes. But over time the country’s population shifts, so that fifty-five percent of the population lives in big cities located in two of the states, and the remaining forty-five percent lives in the other eight states.

What do the electoral votes look like now? Remember that there are only 80 votes to divide proportionally: each state gets at least two, which means that twenty votes are already off the table. Forty-five percent of the 80 remaining electoral votes are divided among the eight less populated states, which get 4.5 each and now each have 6.5 votes (2 senators + 4.5 representatives).6 All together, the eight less populous states have 52 votes. The two populous states, in turn, have fifty-five percent of the population, which gives them fifty-five percent of the 80 proportional votes — or twenty-two each — meaning that they each have 24 votes (2 senators + 22 representatives), or forty-eight between them.

If we put everything together, we can see that after the population shift towards the urbanized states, the bigger states are now outvoted 52-48. Again, this isn’t a problem for representation unless the sorts of things that voters want and need depend on the sort of place they live. But as the map at the very top suggests, this is very much the case.

In our hypothetical example here, if we suppose that everyone votes according to which state they live in, the populous states’ candidate will win popular vote margins of 60-40 but lose Electoral College votes 48-52.7 If you’ve made it this far, you’ll probably see that this hypothetical country looks a little like the increasingly urbanized United States. This is a problem.


Where do we go from here? Below I’ll propose a few potential solutions that would help to ameliorate this problem. Next week I’ll be back with another post looking at what could happen if we don’t.

One potential solution is to do away with the Electoral College and move to a popular vote system. There are pros and cons to this. It’s clearly more representative on an individual level. It would give more weight to votes that are currently “wasted” — for example, Republican voters in California, and Democratic voters in Tennessee. But it’s also true that national policy would swing more decisively to urban interests as more and more people live there, which is undesirable, and maybe unfair, for people who live outside of cities.

Another — in my opinion better — idea is to simply increase the number of electoral votes. The current number, five-hundred and thirty-eight, has been constant since the Reapportionment Act of 1929. But it’s mathematically true that the representational effect of the constant senatorial votes decreases as the the pool of proportional votes becomes larger. Imagine you put some drops of food coloring in a bucket of water, but then you realized that those drops were systematically disempowering millions of voters: if you can’t take the drops out, you could definitely dilute their effect by pouring more water in the bucket.

Let’s go back to our hypothetical example above, and increase the number of electoral votes to 200 from 100. Each state keeps their two constant votes, and the remaining 180 votes are apportioned according to population. Now the eight small states get 81 proportional votes between them in addition to their sixteen guaranteed votes, and have 97 total. The two big states get 99 proportional votes, plus their four constant votes, for 103 total votes. These shares are a lot closer to a fair shake: the forty-five percent of people living in the small states now have 48.5% of the votes (before they had 52%), while the 55% of people living in the big states have 51.5% (before they had 48%). It’s not completely even, but it’s better than it was before.

Back in the United States, there are a few ways to implement an increase in the electoral votes. The first, simplest method would be to increase the number of Representatives in the House of Representatives. This could be done through a law passed by the House and Senate and signed by the President. There are potentially reasons why we might not want to increase the number of Representatives, though — maybe a legislature with 800 (or however many) legislators is too unwieldy.

The second method would require a constitutional amendment to sever the one-to-one link between the number of electoral votes and the number of representatives. Article II, Section 1, Clause 2 of the Constitution says:

Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, equal to the whole Number of Senators and Representatives to which the State may be entitled in the Congress….

The “fix” proposed here would replace the italicized portion with something along the lines of “a number of electors equal to the whole number of Senators and [some amount] times the number of Representatives.” This way, the number of representatives meeting in Washington would remain the same, but they would count for more in determining the number of electoral votes apportioned to each state. Having four-hundred and thirty five Representatives is basically arbitrary, and so is each Representative receiving only one vote.

What would this alteration have done to the electoral results this year? If the state results remained like they are, nothing: Trump simply won too many states for this change to matter. In fact, flipping a single state like Pennsylvania or Florida from red to blue wouldn’t have been enough to change the results, even in a system where proportional votes were doubled. But suppose that Clinton had won both Pennsylvania and Michigan — states were she was favored and ended up losing by less than 1%. Under the current rules she would have still fallen short, 268-270. But double the number of proportional votes, and she comes out ahead, 490-484. The proposed alteration is relatively conservative, but it is a move in the right direction. And there’s no real reason to only double the number of proportional votes; a higher multiple might well be preferable.

There are probably other potential fixes, as well. But I am convinced that our current electoral system is entering a period where popular vote victories are increasingly likely to result in losses. A system like this is not sustainable. The second post in this series will examine historical cases of institutional breakdown along urban/rural lines.

  1. In the real world, Texas has approximated this hypothetical state for a long time, although it is becoming increasingly urbanized. This urbanization, especially in fast-growing cities like Houston and Austin, is a big reason why Democrats believe Texas will eventually “turn blue.”
  2. In other words, every sixty years starting in 1800, and then the last year we have data for.
  3. Statistics aside: this is an example of what’s known as Simpson’s Paradox, which says that a correlation appearing in aggregated data can disappear or reverse when these data are broken into subgroups. It’s good to have a theory of what’s going on!
  4. History aside: the simple explanation for the reversed relationship in the early days of the United States is slavery. States with the lowest levels of urbanization were also the states that held the most slaves, who were only counted as 3/5 of a person for purposes of Electoral College representation. Slavery, incidentally, was the primary reason for the construction of the Electoral College: even if slave states had relatively low levels of representation in it, they had more representation than they would have had in a system where slaves didn’t count for representational purposes. The existence of slavery is also one of the main reasons the U.S. doesn’t elect its presidents through a popular vote. Slave states certainly weren’t going to let their slaves vote, but they wanted them to “count” in some way; otherwise, these states would be hopelessly outvoted by the more populated northern states.
  5. More on this below.
  6. If it bothers you that these states have non-integer electoral votes, imagine that the population isn’t perfectly evenly divided between them, and that four have 6 votes and four have 7.
  7. Of course people don’t vote strictly based on where they live. Some people in the big states will vote for small-state candidates, as will some people in the small states vote for big-state candidates.

Comments are closed.