Is the Senate a democratic institution?

By NATHAN MORSE

January 2019

The United States Senate has a unique and powerful role in the US government. Unlike the House of Representatives, the Senate has the authority to approve presidential appointments to high-ranking positions and to ratify treaties with other countries. It is also the only body that can formally remove a President from office.1 “Learn About the Senate,” United States Senate, www.senate.gov. Originally, senators were chosen by state legislatures, not elected by the people of their state. This changed in 1913 with an amendment that made senators directly elected. Chief among the goals of this amendment was expanding democracy, a response to rampant corruption in state legislatures and the federal government.2 “The Interactive Constitution: The 17th Amendment,” National Constitution Center, www.constitutecenter.org.

To analyze the effectiveness of this goal, we need to take a closer look at the word democracy. Direct democracy—where voters make all the decisions with little or no deference to representatives—is widely considered to be a poor, inefficient form of government on a large scale. Thus democracy usually refers to representative democracy, whereby the people elect leaders to govern on their behalf. The presence of elections is not enough to qualify as a democracy. Dictatorships and autocracies usually hold rigged elections and call themselves “democratic” to disguise their corruption. Taking this into account, this paper will consider a general working definition of democracy to be that a majority of citizens consent to the leaders who make decisions on their behalf.

Democracy in the seat vs. democracy in the Senate

There’s no question that each individual senator is democratically elected and has the majority consent of their state. Of course, senators do not govern states; they govern the nation. While some laws and policies may affect certain areas differently, all apply to the whole country. A New York senator has the same say in the federal tax rates for Georgians as a Georgia senator does, just as an Alabama senator has the same say in federal drug laws for Californians as a California senator does. Therefore, the democracy of the Senate should be judged not by the democracy of the parts but by the democracy of the whole.

The U.S. has 50 states ranging in population from around 600,000 in Wyoming to 40 million in California.3 “2018 National and State Population Estimates”, United States Census Bureau, www.census.gov. Regardless, each state is represented by the same number of senators. A quick visualization of the differences in state sizes:

Figure 1. Populations of all 50 states. The states to the left of the Senate Majority mark hold 50 seats. The states to the left of the Population Majority mark comprise 50% of the U.S. population.

Figure 1. Populations of all 50 states. The states to the left of the Senate Majority mark hold 50 seats. The states to the left of the Population Majority mark comprise 50% of the U.S. population.

Figure 2. Visual comparisons of states by population and by number of senators. Figure 2. Visual comparisons of states by population and by number of senators.

The largest 9 states make up 51.2% of the population, a slight majority. In other words, more than half of the nation is represented by less than one-fifth of the Senate (18 senators). Summing the populations of the 26 smallest states reveals that a majority of the Senate is elected by just 17.6% of the population. That 17.6% has 3 times more power per capita than the citizens from the 9 largest states. This fundamentally contradicts the key maxim of democacy: one person, one vote. Structurally, this is not a majoritarian democracy at all.

To what extent does this structural issue hold in practice?

We can aggregate vote shares from the past three elections to come to a hypothetical “popular vote” for the Senate. Senators are elected to six-year terms, with elections staggered every two years. Summing votes from the last three elections is the fairest metric we have of what percentage of the population voted for the senators who control the chamber.

For the 2014 and 2016 elections, data was obtained from the MIT Election Data and Science Lab, which includes data through 2016.4 “U.S. Senate 1976–2016,” MIT Election Data and Science Lab (November 2017), electionlab.mit.edu. For 2018 data, the New York Times still hosts its election night page where it totaled all votes across the country.5 “U.S. Senate Election Results 2018,” New York Times (January 2018), www.nytimes.com. We begin by loading this data in R:

# Load in data
elections <- read.csv("1976-2016-senate.csv")
elections2018 <- data.frame(rep=34981658, dem=53078387, total=89508182)

# Select variables
elections <- elections %>%
  select(year, state, party, candidatevotes, totalvotes, special) %>%
  filter(party=="democrat" | party=="republican", special==FALSE)

# Filter by year
elections2014 <- filter(elections, year==2014)
elections2016 <- filter(elections, year==2016)

Next, the total votes per party in each election were summed:

Note that this only includes regular elections, not special elections for vacancies, to avoid double-counting representation.

summary <- data.frame(election = c(2014, 2016, 2018),
                      rep = c(sum(elections2014$candidatevotes[which(elections2014$party=="republican")]),
                              sum(elections2016$candidatevotes[which(elections2016$party=="republican")]),
                              elections2018$rep[1]),
                      dem = c(sum(elections2014$candidatevotes[which(elections2014$party=="democrat")]),
                              sum(elections2016$candidatevotes[which(elections2016$party=="democrat")]),
                              elections2018$dem[1]),
                      total = c(sum(unique(elections2014$totalvotes)),
                                sum(unique(elections2016$totalvotes)),
                                elections2018$total[1]))

We now have “popular votes” for each election. Individually these mean very little because some elections happen to have mostly Democrat-leaning states up for grabs or mostly Republican-leaning states.

Election Republican votes Democrat votes Total votes
2014 51.33% 44.16% 45,974,587
2016 42.16% 52.98% 96,866,509
2018 39.08% 59.3% 89,508,182

Finally, the votes from 2014, 2016, and 2018 were summed to come to the collective popular vote for the current Senate.

summary[4,] <- c("Total", sum(summary$rep), sum(summary$dem), 
                 sum(summary$total))
Election Republican votes Democrat votes Total votes
Total 42.79% 53.67% 232,349,278

Democrats have a solid majority and an 11% lead over Republicans. Even so, Republicans hold a majority of the seats, with 10% more seats than the collective popular vote for Republican candidates. The 53 Republican senators were elected by 42.79% of the population.

Figure 3. Senate seats and collective popular vote shares. Blue represents Democrats, red represents Republicans, light blue seats represent independents that caucus with Democrats, and gray represents votes for independent candidates (regardless of ideological slant).