Election law in the United States depresses voter turnout and particularly for lower income and minority groups who are more likely to elect candidates supportive of economic redistribution. Monetary expenditures on campaigns and media also impact voting behavior to detriment of economic redistribution. In this chapter, I will discuss some of the evidence on the impact of election law and then discuss some proposals to increase voter turnout and reduce anti-redistributive bias in election law.
Turnout
Early Voting: Voter turnout as a percentage of the voting-age population in the United States was 56% in the 2016 election (http://www.pewresearch.org/fact-tank/2018/05/21/u-s-voter-turnout-trails-most-developed-countries/). Though turnout was high compared to other recent U.S. elections, it was the sixth lowest among the 32 OECD (Organization for Economic Cooperation and Development) countries for which voting age population is available. Only Luxembourg, Slovenia, Poland, Chile, Latvia, and Switzerland had lower turnout rates. Moreover, turnout in non-Presidential federal elections is often around 40%. The countries with the highest voter turnout are Belgium (87%), Sweden (83%) and Denmark (80%).
In contrast to most developed countries, federal elections in the United States do not fall on weekends nor are they national holidays. This makes it difficult for voters to make it to the polls in order to cast a ballot. Voting on a working day is a likely contributor both to the overall low turnout rates in the United States as well as to the substantially higher turnout of higher income Americans. Those earning more than $150,000 per year vote at a 50% higher rate in presidential elections and at a 100% higher rate in midterm elections than those making less than $5,000 per year. More generally, turnout is roughly monotonically increasing in income. This is unsurprising since low income workers have less flexible work schedules (Enchautegui, 2013).
One policy which many states have used to increase turnout is early voting. Currently, 32 states have mandated or allowed polling stations to open before election day for voters to cast ballots in person. Minnesota has the most generous early voting system with 46 days of early voting including openings on weekends and weekday nights. It also had the highest turnout rate of any state in the country in 2016.
Kaplan and Yuan (forthcoming) estimate the impact of expanding access to voting using an early voting homogenization law in Ohio from 2012. They find that expanding early voting increases turnout by 0.22 percentage points per day of early voting. This implies that a state like Minnesota yields an additional 10 percentage points of turnout from its 46 days of early voting. Moreover, Kaplan and Yuan (forthcoming) show that the effect is 20% higher for women than for men. The effects are also higher for those of child-bearing and working age. Additionally, they find higher impacts on political Independents and Democrats. Other research on the impact of early voting in Florida has shown higher impacts on African-American voters than on whites (Herron and Smith, 2012). Simulations of the creation of a national early voting law at the level of Minnesota suggest that such a policy would have increased turnout, made the electorate more representative of the population, and altered the outcome of the 2016 Presidential election and the majorities in the Senate in both 2012 and 2016.
Alternate (or additional) possible legal changes that would likely have a similar impact would be to declare election day a national holiday or to place it on a weekend. This also would likely increase turnout and make the electorate more representative of the voting population.
Photo Identification: A number of laws have been passed in recent years which restrict access to the ballot box. Two examples of these are (1.) photo identification laws which require voters to present certain forms of official forms of identification and (2.) felon disenfranchisement laws which disallow felons and former felons from voting. Ethnic minorities and young voters are most likely to show up to polling stations without a voter id. Both of these groups are under-represented in the national electorate. Citrin, Green and Levy (2012) show that informing voters about new voter id laws has a 1 percentage point higher impact than just encouraging voters to turn out. Thus, photo id laws do both reduce turnout and make the electorate less representative of the population. Laws preventing incarcerated felons or former felons no longer in jail from voting also have similar impacts.
The reasons given for photo identification laws are that they protect elections from voter fraud such as double voting. However, photo identification laws, as discussed above, have substantial negative impacts upon voter turnout and tilt the electorate towards over-represented voters. There is recent evidence on voter fraud and find it to be minimal at best. Goel, Meredith, Morse, Rothschild and Shirani-Mehr (2017) find that, in contrast, to the sizable negative impact of photo id laws on voting, double voting accounts for at most 0.02% of all votes and almost all of these are consistent with small amounts of measurement error in the voter files. Getting rid of photo id laws would improve the representativeness and fairness of the electoral system as would elimination of felon disenfranchisement laws.
Voter Registration: Though only a slim majority of voters participate even in Presidential elections, voters are required to register to vote in order to vote. Only North Dakota does not require voter registration. Moreover, most states require registration well in advance of the election. In the 2016 Presidential elections, 86.8% of registered voters cast ballots. Thus, a substantial portion of the barrier to full participation in elections is a barrier from registration requirements. Moreover, this problem is exacerbated by the practice by election commissions who manage the voter rolls of purging the rolls of inactive voters. Voters with absences in participation across multiple federal elections are often dropped from voter registration databases. In most states, when on election day, these voters then go to vote, they are unable to do so because election day registration is not allowed.
Ansolabehere and Konisky (2006, Political Analysis) show that the introduction of statewide voter registration (some counties already had pre-existing registration systems in place) in New York and Ohio in 1965 and 1977 respectively depressed turnout by 10 percentage points. In some states, voters may register to vote and vote on election day. This is called election day registration. Burden (2011) shows that election day registration lowers the registration turnout penalty by 2 percentage points. Braconnier, Dormagen and Pons (2017) show similar reductions in voter turnout from registration laws in France. A recent paper by Cantoni and Pons (2018) shows that election day registration has the single largest impact upon turnout of any election policy which varies across states in the United States. In Ohio, voters must be registered 28 days or more in advance of the election in order to vote on election day. There is some overlap between the period of early voting and the mandatory registration period so that voters can register to vote and vote early in the same location at the same time. Kaplan and Yuan (forthcoming) show that the effect upon turnout of same day registration days is much larger than for normal early voting days.
In recent years, some states have created automatic voter registration systems where voters are automatically registered by the state government at age 18. Oregon became the first automatic voter registration state in 2016. Rhode Island and Vermont have now followed Oregon and the state of Washington has passed legislation to enact automatic voter registration as of 2019. Since these laws are new, not much analysis has been done on their impact. However, they represent one possible way to increase voter turnout. Overall, many laws in the United States discourage full participation and differentially make voting difficult for lower income families and minorities who tend to have less flexible jobs.
Gerrymandering
In the United States, representatives to the House of Representatives are elected in winner-take-all districts. Every ten years, following the census, these districts are redrawn for the purposes of equating representation across districts. This is done at the state level. However, particularly when there is unified government, the party in power shapes the districts in order to maximize their party’s seat shares in the state and federal legislatures. For example, in Wisconsin in 2012, the Republicans gained 60 out of 99 seats (almost 2/3) with 48.6% of the vote share. Nationally, in 2014, the Democrats in the House of Representatives won 48.8% of the national vote against 47.6% for the Republicans; however, the Republicans won 234 seats and the Democrats won 201. The Brennan Center for Justice estimates that Republicans had 16 to 17 additional seats in the House of Representatives due to partisan gerrymandering during the 115th Congress (2017-2018). In other words, in the absence of partisan gerrymandering, the Democratic party would have had majority control in the House of Representatives in the 115th Congress.
Though gerrymandering is not necessarily biased towards over-representation of higher income and white voters, in recent years, it has been more heavily used by the Republican party. This is due to two factors: (1.) more progressives states often pass laws which make gerrymandering more difficult and (2.) the Republican party won a record number of state legislatures and governorships in 2010 which then allowed them to gerrymander in a large number of states. A recent paper by Jeong and Shenoy (2017) estimates that obtaining majority control over the legislature and thus the gerrymandering process increase the probability of majority victory in a given district in the first election after redistricting of 11 percentage points.
In 14 states, including the large state of California, the legislature has shifted redistricting decisions from elected governors and legislatures to independent commissions. In 6 cases, these are non-partisan commissions and in 8 cases, commissions are bi-partisan. In addition, measures have been developed to assess the imbalance in partisan representation across districts. The “efficiency gap“ measure computes the number of wasted votes (votes above majority) as a fraction of total votes cast in a district. With modern data, we have both policies to implement non-partisan or bi-partisan gerrymandering as well as the ability to assess its success. Possible remedies for gerrymandering include national legislation or state-level legislation establishing non-partisan or bi-partisan redistricting commissions. These are currently 21 state with bi-partisan or non-partisan commissions. Alternatively, states could pass legislation electing representatives in proportion to state-level vote shares for Congressional seats.
Campaign Contributions and Political Influence
Election laws in the United States allow for monetary contributions from private individuals and from corporations. Courts have interpreted campaign contributions as protected under freedom of speech provisions of the first amendment to the U.S. Constitution. Contributions have historically been capped due to the strong state interest in reducing the possibility of “influence” or the “perception of influence”, either of which the courts have claimed could be destabilizing for democracy. However, what do we know about the actual impacts of campaign contributions?
First of all, the amount of monetary expenditures is large and rapidly increasing. In the year 2004, political campaigns for presidential candidates spent $850 million, campaigns for House elections spent $660 million, and Senate campaigns spent $550 million. By 2016, those numbers had risen to $1.5 billion of spending by Presidential campaigns, $970 million by House campaigns, and $670 million by Senate campaigns. However, in addition to campaign expenditures, independent expenditures by individuals and corporations have become substantially more important in the past decade. Two important decisions by federal courts (Citizens United v. FEC, 2010 and Free Speech v. FEC, 2013) allowed for unlimited independent expenditures by individuals or corporations in favor of political candidates. In the wake of these decisions, independent expenditures rose from $330 million in 2008 to $1.0 billion in 2012 and $1.4 billion in 2016.
But do campaign expenditures have an impact? Most political campaigns spend the vast majority of their funds on political advertisement and “Get Out The Vote” (GOTV) operations[1]. Gerber and Green (2015) have done field experiments on the impacts of GOTV operations. In particular, they have randomized contact with voters in various formats: canvassing (door knocking), phone calls, mailings and even text messages. They then use official voter registration data to check turnout for those whom they contact and those whom they do not. They find positive impacts upon turnout for all forms of contact. More recently Vincent Pons analyzed the effects of a massive randomized voter persuasion effort (5 million doors were knocked) by the Socialist Party in France in the 2012 French Presidential elections. The socialist party randomized its canvassing efforts. It differs from the Gerber and Green experiments in that the object was not just voter turnout but also voter persuasion. Pons (2018) did not find a statistically significant impact on voter turnout; however, he did find a 3 percentage point increase in the vote share for the Socialist Party in the precincts which were randomly canvassed. In addition, he found that some of the persuasive effects of the initial canvassing experiment persisted to future elections. So campaign expenditures on GOTV operations is highly effective.
There has also been recent very convincing research done on the impact of political advertisement. Spenkuch and Toniatti (2018) estimate the impact of political advertisements by looking at turnout and vote share differentials across media market boundaries within states where there was a differential partisan composition of advertisements. Similar to Pons (2018), Spenkuch and Toniatti (2018) fail to find a statistically significant impact upon voter turnout though the estimated effect is positive. However, they do find that an additional ten ads in favor of a candidate raises that candidate’s vote share by 0.22 percentage points. Surprisingly, these results are estimated for Presidential campaigns where one would expect communication overload to have driven the impact of marginal impact of advertisement to zero.
Most campaign expenditures, therefore, have sizable impacts upon election outcomes. Thus contributors to campaigns, either through direct contributions, through Political Action Committees, or through independent expenditures, influence politics and policy outcomes. There are two main ways in which campaign expenditures can influence political outcomes. They can influence who is elected (“election motive”) and they can influence how politicians vote once elected (“influence motive”)[2]. Either way contributors exert undue influence over the political system.
Who are the contributors? Particularly with the rise in independent expenditures, contributors are increasingly the wealthiest members of society. Since the Citizens United ruling, over 20% of contributions have been from the top 0.01% of income. Moreover, the fraction of contributions coming from the top 0.01% of the income earners has been rising over time even since the passage of Citizens United (Bonica et al., 2013). These contributions likely both support more conservative candidates (even within the Democratic party) and also make candidates more conservative particularly on economic issues. A new paper by Kaplan, Spenkuch and Yuan (2018) show that members of Congress are much more likely to support an interest group’s position on a bill when they have received a contribution from the interest group. Moreover, they find that when votes take place on high news pressure days (i.e. when there is likely to be little coverage of the bill in the press), the responsiveness of members of Congress to campaign expenditures increases by up to a factor of 10. This shows that interest groups not only influence who gets elected but also how politicians vote once elected. The skewness in the earnings distribution of contributors thus likely pushes policy, and particularly economic policy, strongly towards low levels of taxation and government expenditures. It is not surprising that as campaign contributions have increased at very rapid rates over the past 5 decades, tax rates on income and particularly on capital income have declined so dramatically.
Finally, the fact that individuals, corporations and interest groups can spend so much money on political campaigns means that they can threaten to spend money if candidates do not vote in accordance with their policy desires. With partisan gerrymandering, it is increasingly difficult for wealthy individuals or interest groups to threaten candidates in general elections. However, in highly partisan districts, there is often room for a primary challenge to an elected politician. Chamon and Kaplan (2011) write a theoretical model where interest groups or individuals can contribute to a campaign or threaten to contribute to an opposing campaign. Direct evidence on threats is difficult to find since such threats are technically illegal. Chamon and Kaplan (2011) provide empirical evidence in support of their model of threats. However, there is also anecdotal evidence. For example, Grover Norquist’s Americans for Tax Reform has been successful in getting almost all Republicans to sign a pledge that they will not raise taxes. These tax pledges received press coverage during the Obama-era government shutdowns when legislators were threatened with primary challenges for agreeing to expenditure increases. Also, the Senate Conservatives Fund, headed by Ken Cuccinelli, threatened to recruit and fund primary challenges to any Republicans not voting for repeal of the Affordable Care Act. The ability for individuals and corporations to spend unlimited amounts of money in support of a candidate in the post-Citizens United era increases the influence of money on politics and policy while simultaneously increasing the difficulty in detecting the impact of money.
Many countries limit campaign expenditures. In many countries, government provides funding for campaigns in lieu of private funding. Some countries such as Canada not only put limits on individual contributions but also put limits on aggregate expenditures by individual campaigns. Well-crafted empirical work on campaign contribution limitations are rare. This is because campaign contribution limitations are usually decided at the country level. This makes causal estimation of the impact of such laws difficult. Recent work by Avis et al. (2017) show that campaign contribution limitations in Brazil led to greater political competition and entry by candidates with lower average income and wealth. The United States could tighten campaign contribution limits to individual campaigns, to political parties and could pass a constitutional amendment limiting or banning independent expenditures.
Ideological Influence
Political advertising influences voters but so do the media and so do think tanks and membership organizations. In the 19th century, newspapers were explicitly partisan. This died out towards the end of the 19th century (Gentzkow and Shapiro). Starting in the late 1980s, conservative talk radio reintroduced partisan news on a large scale. This was followed by the introduction of the conservative Fox News Channel (FNC) in 1996. FNC soon became the most popular news channel on cable television. It was clearly much further to the right than other news organizations (Martin and Yurukoglu). Studies have used the staggered introduction of Fox News as well as randomness in the channel order of Fox News across towns to estimate the impact of Fox News on voting patterns. DellaVigna and Kaplan (2007) and Martin and Yurukoglu (2016). They both find a positive impact on the Republican vote share from the expansion of FNC of around 0.5 percentage points in the 2000 Presidential elections. They find similarly sized effects in other races (Senate, House of Representatives). Martin and Yurukoglu (2016) also cover later years and find that as FNC expanded, its influence grew to 6.3 percentage points in the 2008 elections. These are extremely large effects. Martin and McCrain (2018) also find a substantive conservative shift and increase in viewership in local news due to consolidation by the Sinclair Media Group.
What policies exist to potentially mitigate the impact of conservative news? Fox news slants their news strongly in favor of the Republican party (as MSNBC does in favor of the Democratic party). Other countries such as the United Kingdom have laws mandating minimum levels of impartiality. Moreover, the United Kingdom has purely publicly financed television (the BBC). Impartiality laws regulating particularly TV media as well as funding for a public television news station could help restore the imbalance created by FNC. As an example, in the U.K., a regulatory agency (Ofcom) reviews all forms of media for content and mandates corrections for misleading or untrue information. Their power comes from the ability to levy fines. In extreme cases, Ofcom (Office of Communications) can deny operating licenses or deny mergers and acquisitions.
In addition to the media, membership organizations potentially wield political influence and political power. As income and wealth inequality have risen, organizations have proliferated supporting the interests of business owners: the Chamber of Commerce has expanded and become a partisan organization and wealthy donors have formed other organizations (i.e. the American Enterprise Institute, the Cato Institute, and the Heritage Foundation). Think tanks also exist which support the Democrats or are on the political left (i.e. the Center for American Progress, the Economic Policy Institute). However, the organizations supporting business owner interests are unsurprisingly substantially better funded. The main set of organizations which have historically represented the interests of working people have been labor unions. However, unions have been in strong decline over the past 45 years. With a peak of 35% representation in the private sector, unionization rates in the private sector are now below 7% (Naidu, 2019). During this time period of declining unionization, we have also seen large reductions in top marginal tax rates from 94% in the late 1940s to 70% in the 1960s to the current 35%. Interestingly, the conservative move in economic policy has not been mirrored in social policy suggesting a potential role for unions.
Moreover, the conservative supreme court has through multiple decisions (Harris v. Quinn, 2014; Janus v. AFSCME, 2018) eroded the ability of public sector unions to collect fees from their members. This will further reduce the size and power of public sector unions going forward.
Though we do not have a credible estimate of the impact of unions directly on policies, recent work by Hertel-Fernandez, Feigenbaum and Williamson (2018) estimate the impact of unionization upon the Democratic vote share. They compare changes in county-level presidential vote shares in states when a right to work law is passed to neighboring counties in other states that never passed a right-to-work law. Right to work laws ban contracts which allow unions to charge dues to all workers. Right-to-work laws have been shown to have very large impacts on state private sector unionization rates. Hertel-Fernandez et al. (2018) find no difference in right-to-work versus non-right-to-work law state presidential vote shares before right-to-work laws are passed. However, immediately following passage, they find a decline in the Democratic vote share of 3.5% which expands over the following decade to 15 percentage points. Thus strengthening labor law by banning Right-to-work laws, allowing for public sector union due collection, and streamlining the union election process (i.e. allowing for a card check process) would also help lower income individuals express themselves politically at state and national levels. Naidu (2019) discusses in greater policies and ideas for increasing the strength of the union movement in the United States.
Summary
In sum, political institutions in the United States favor higher income individuals over lower income individuals and ethnic majorities over ethnic minorities. This is accomplished through a myriad of policies which impact who votes, allow for differential influence and access by the wealthy, structure voting districts to dilute the impacts of under represented voters, and allow for oversized influence of pro-business owner ideas through media and membership organizations.
There are many policies, some implemented in other countries, which could help restore greater balance to political competition and ideological competition. Some policies which we have discussed in this piece which would help equalize the political process include:
Turnout
- (1.) Expand early voting, particularly weekend voting and same day registration.
- (2.) Make election day a national holiday.
- (3.) Get rid of photo identification laws.
- (4.) Eliminate voter registration or at least introduce automatic voter registration.
- (5.) Eliminate felon disenfranchisement laws.
Monetary Influence
- (1.) Limit campaign contributions to very small amounts such as $50 per individual per candidate per election or eliminate private funding and go to a public funding system.
- (2.) Ban or strongly limit independent expenditures.
- (3.) Ban expenditures for corporations.
Gerrymandering
- (1.) Mandate Independent Redistricting Commissions
Ideological Influence
- (1.) Disallow politically biased television news.
- (2.) Make union representation easier.
Endnotes
[1] Information on campaign and independent expenditures come from opensecrets.org, which synthesizes data from the Federal Election Commission.
[2] Grossman and Helpman (2002).