Legislative redistricting and the partisan distribution of transportation expenditure

Abstract

I show that a state representative’s political party determines transportation expenditure in the area she represents. Previous studies of this topic consider party changes through election outcomes, which may be correlated with unobservable determinants of expenditure. To overcome this issue, I identify my estimates using Ohio’s 2012 state legislative redistricting, which moved many geographic areas into districts with opposite party incumbents. The Republican party controlled the state legislature and governorship over the period I study. I find that areas moving from governing party Republican to minority party Democratic districts received $3.4M (0.18 standard deviations) less annual highway construction funding than areas remaining in Republican districts. Areas moving from a Democratic to a Republican district, on the other hand, experienced no increase in expenditure—the negative effect of moving to a different representative’s district appears to outweigh the positive effect of a majority party representative. Descriptive evidence suggests that changing representative’s party through redistricting had a different effect on construction funding than changing through an election, underlining the importance of my identification strategy.

1 Introduction

The condition of public roads provides a particularly salient benchmark for state government performance. State governments bear primary responsibility for road construction,¹ devoting 6.2% of their direct expenditure to highways. Almost all residents use roads on a day to day basis (Fisher 2016, pp. 150–151); a functional highway system, moreover, fosters local economic development. In a 2017 Area Development magazine survey, corporate executives regarded highway access as the most important of twenty-seven location draws, having rated highway access among the two most important factors five years in a row (Gambale 2016, 2017; Fisher 2016).

I ask whether Ohio’s state government distributes transportation funding not only according to equity and efficiency concerns, but according to alignment with the majority party. Since the Republican party has controlled the Ohio state Assembly and governorship for most of the past two decades, I test whether geographic areas with Republican representatives receive more road construction funding. Because election outcomes depend on unobservable district characteristics, I exploit plausibly exogenous variation in partisan representation from the 2012 state legislative redistricting. Redistricting divided Ohio into 250 pairs of old and new districts with area in common, which I call “areas of intersection.” I apply a difference-in-differences method to estimate how construction funding changed for areas of intersection whose representative changed party. My estimates show that areas moving from a Republican to a Democratic representative received $3.4 million per year less after redistricting than areas that remained in the Republican party, a decrease of 47%.

Previous papers studying partisan alignment’s effect on resource allocation (Case 2001; Solé-Ollé and Sorribas-Navarro 2008; Berry et al. 2010; Levitt et al. 1995) have focused on district party change through elections. But election outcomes depend on voters’ beliefs about each party’s ability to provide public goods, and the selected sample of districts that change party through elections may be experiencing differential demographic and economic shifts correlated with construction spending. Besides concerns about bias, estimates identified using party changes through elections heavily weight swing districts where party change is most likely to occur. It is unlikely that the majority party would target expenditure equally between safe and swing districts (Dixit and Londregan 1998; Case 2001). Because the variation that I use derives from boundary changes rather than voting, the effect that I identify more closely represents the average over the entire distribution of voter preferences, instead of the average over swing districts only.

I argue that party changes through elections may depend on voter preferences and district characteristics, but a similar concern may be raised about redistricting. In states (like Ohio) where the party in power chooses district boundaries, redistricting patterns may depend on the political alignment of areas that move between districts. Based on evidence from court filings, I argue that Ohio mapmakers considered past voting patterns as a static factor when choosing district boundaries. From an econometric perspective, time-constant differences in partisanship between areas will be absorbed by fixed effects. While areas that change party through redistricting are not evenly distributed across the state, they are dispersed across a range of growing and shrinking areas of Ohio. They also show similar trends in employment and road condition to areas that did not change party through redistricting. Thus, I argue that selection issues associated with the redistricting boundaries are minimized in the setting I consider.

I highlight three additional results. First, I find an expenditure decrease for areas redistricted out of the majority party, but no robust expenditure increase for areas redistricted in. I offer an explanation—when an area is redistricted into the majority party, its representative changes. The new representative is likely less able to direct spending towards the new area of her district due to lack of information, as this process requires time and information. She also faces less incentive to direct more spending towards existing projects, where she may share credit with the old incumbent. Thus, areas that lose their incumbent representative will suffer a decline in funding.

Testing this hypothesis, I find that districts with the same representative pre- and post-redistricting receive more funding than those who switched to a new representative. Areas that moved from a Democratic to a Republican district lost their incumbent representative, and I find that the associated decline in funding largely offset the gain from moving to a Republican representative. Existing literature has emphasized the role of networks and experience in determining discretionary expenditure distribution (Balla et al. 2002; Bernhardt et al. 2004; Boyle and Matheson 2009; Wu and Williams 2015). But the question of whether change in representative through redistricting affects expenditure distribution has receive little attention until now.

Finally, I demonstrate the relevance of my redistricting identification strategy. I estimate different effects using variation in candidate’s party through election outcomes, compared to the effects identified using my redistricting strategy. The effects I estimate using election outcomes also show a much larger expenditure change when an incumbent representative is voted out of office rather than choosing not to run again. These comparisons illustrate the potential selection issues associated with districts that change party through election outcomes.

1.1 Related literature

Numerous empirical papers have estimated the impact of partisan alignment on expenditure distribution. Studying poverty assistance spending in Albania, Case (2001) finds that the governing party targets spending toward its own base, rather than swing districts. Castells and Solé-Ollé (2005) find that transportation expenditure distribution in Spain maximizes the governing party’s chance of reelection instead of equity or efficiency, while Solé-Ollé and Sorribas-Navarro (2008) find that Spanish municipalities aligned with the federal governing party receive 40% more grants than unaligned municipalities. Tavits (2009) shows that parties target spending towards their strongholds in the Nordic countries, a cultural setting where pork-barrel politics is less likely to occur. Others have used variation in governing party (Johansson 2003) or a regression discontinuity design in legislator vote share (Bracco et al. 2015) to identify variation in partisan alignment. Several studies show that Canadian ridings represented by the government party receive more transportation funding (Jacques and Ferland 2021) or development grants (Milligan and Smart 2005). Joanis (2011) and Mehiriz (2015) discover similar results for the Quebec parliament. Partisan alignment within parliament plays an important role even when funding distribution is determined by lower-level officials (Jacques and Ferland 2021) or by a nominally autonomous development authority (Mehiriz 2015).

The papers above study parliamentary systems. Since the party who wins the popular vote typically chooses the executive, partisanship proves particularly important in a parliamentary system (Morgenstern and Swindle 2005; Tavits 2009). On the other hand, presidential systems provide a stronger incentive to target expenditure towards local constituencies, particularly when legislators are elected from single-member districts (Shugart 1999; Morgenstern and Swindle 2005; Denemark 2000). In the US, with both single-member legislative districts and an independent executive, legislators face a strong incentive to direct funding towards their own districts. Several studies have demonstrated partisanship’s role in public expenditure distribution within the US. Levitt et al. (1995) find that areas with more Democratic voters received more programmatic spending in the 1980s, especially from programs enacted during periods when Democrats controlled Congress. Balla et al. (2002) find a majority party edge in discretionary spending, even within spending bills that receive bipartisan support. Using a regression discontinuity design, Albouy (2013) confirms that US congresspersons belonging to the majority party attract more funds to their own states, while others showed that house districts (Berry et al. 2010) or states (Larcinese et al. 2006) aligned with the president receive more federal expenditure.

This literature identifies party variation through election outcomes. But unobservable determinants of district expenditure may influence who runs in an election and who wins. A regression discontinuity design (Albouy 2013; Johansson 2003) isolates exogenous variation in legislator party, but only identifies the effect of legislator party for narrowly contested districts. Redistricting, on the other hand, generates changes in partisan alignment that do not depend on election outcomes. Voters from Republican districts are regrouped with voters in Democratic districts, and vice-versa. In many areas, the process changes the party of the legislator representing an area due to shifting district boundaries, rather than election outcomes. By exploiting this redistricting variation, I seek both to avoid selection issues associated with election variation and to identify partisanship’s effects on spending across the distribution of safe and swing districts.

I estimate that areas remaining in Republican districts before and after redistricting see a 60% increase in construction spending, relative to areas that move to Democratic districts. This figure exceeds Berry et al. (2010)’s estimate that House districts belonging to the president’s party receive 4.5% more federal spending, Albouy (2013)’s estimate that states with an entirely Republican congressional delegation gain 31% more transportation funding when the Republican party controls all branches of government² and Solé-Ollé and Sorribas-Navarro (2008) and Bracco et al. (2015)’s estimates that lower level governments aligned with the nationally governing party receive a 40% increase in grants. Estimates in my paper likely differ from previous estimates due to differences in identifying variation. I support this conjecture by estimating the change in transportation expenditure associated with a party change through elections. I estimate that areas moving from the Republican to the Democratic party in years with a Republican governor only lose $1.2M in expenditure per year, smaller than my estimate identified using redistricting.

A handful of papers have used redistricting to answer other questions. Ansolabehere et al. (2000) exploited a 1960s court ruling equalizing congressional districts’ population to estimate the component of “personal voting” in an incumbent politician’s electoral advantage. Chen (2010) uses a state Senate expansion in New York to test whether areas with more overlap in the upper and lower house receive more earmarks. He finds that areas with a senator and representative in the same party receive more expenditure, though he finds that areas moving from the minority to the majority party experience an increase in spending, the converse of my result. Similarly, Wokker (2019) asks whether incomes are higher in marginal electoral districts, using a restructuring of the Australian House of Representatives to identify changes in marginality—but he finds no significant effect. Finally, Raina and Xu (2021) use gerrymandering to identify how partisan alignment with voters affects federal representatives’ incentives—representatives grouped moving into safer districts vote with their own party less and direct less spending to their districts.

2 Redistricting

The Ohio legislature (“General Assembly”) has an upper house (“Senate”) with 33 members and a lower house (“House of Representatives”) with 99 members. Each Assembly member serves a single-member district, with each Senate district comprised of three House districts. Voters choose their representatives for two year terms every even-numbered year, while senators are elected to staggered four year terms in even-year elections. Representatives and senators then take office in January of the odd-numbered year following their election. In recent years the Republican party has dominated Ohio’s state politics: from 2007–2016 Republicans controlled the state Senate every year, the state House of Representatives all but two years, and the governorship all but four years (see Fig. 1).

Fig. 1

Partisan control of Ohio government

State legislative districts are redrawn every ten years to rebalance population. In Ohio, a commission composed of the governor, state auditor, secretary of state, and one member each selected by the House Majority and Minority Leaders drew the 2012 districts (Ohio Constitution XI § 1). Redistricting occurs the year after the Census, most recently 2011, meaning state representatives were elected from the new districts starting in 2012 to begin governing in 2013.

Redistricting created 301 common areas where an old and new House district overlapped. I focus on the House because, in contrast to the Senate, many of these areas changed party. My observations are the 250 such “areas of intersection” with population over 100.³ Depending on the continuity between an old district and its successor, an area of intersection could represent the majority of a district that continued roughly unchanged—this was uncommon, as the boundaries of all but six House districts changed with redistricting. Other areas represented part of an district that had been “sliced off” and allocated to a new district. As of the 2010 Census, for example, old House District 2 had grown to a population of 170,573, 53% larger than $\frac{1}{99}^{th}$ of state population. 116,760 people remained in new District 67, a geographic subset of old District 2, while 54,813 people were moved to new District 68.⁴ Other areas of intersection comprised the remnants of a district that had been split up more thoroughly—in thirteen House districts no intersection of the old and new district contained more than half of the old district’s population.

I’ve focused on the “area of intersection” because it is the largest type of geographic area where I can identify pairs of pre- and post-redistricting representatives. Choosing a finer level of geography—municipality, township, electoral ward, or even Census block—would entail three drawbacks. First, municipalities, townships, and sometimes even wards or blocks were divided in the redistricting process. In these cases I would have to compare heterogeneous subsets of municipalities or townships instead of comparing heterogeneous subsets of old and new districts. Second, since many projects take place over a stretch of road or multiple locations, using a finer level of geography would require me to divide more projects among multiple locations. This would induce measurement error and reduce the precision of my estimates. Finally, the benefits of construction spending may spill across boundaries—the finer geography I use, the more the perceived benefits of a road construction project accrue to those in neighboring geographies.

I estimate the effect of party change on spending using party transitions during the redistricting year: “R to R,” “R to D,” “D to R,” and “D to D.” My unit of analysis is the common area, or “area of intersection,” where an old and new district overlap. Fig. 2 plots areas of intersection, by their transition between parties, onto the map of Ohio. I also distinguish areas redistricted to a new representative in the same party. The map shows that rural areas lean Republican. Urban areas like Toledo, Cleveland, and Cincinnati lean Democrat, as do Appalachia and the “Rust Belt” region around Lake Erie. Areas that changed parties after redistricting concentrated around the borders between Democratic and Republican districts near Columbus, Cincinnati, Toledo, and in the north and east of the state.

Fig. 2

2013 redistricting, Ohio house of representatives. Data: (Ohio Secretary of State 2017; United States Census Bureau 2013)

2.1 Variation in representative’s party

Table 1 documents how areas of intersection transitioned between representatives in House of Representatives elections from 2008 to 2014, separating the first post-redistricting election, 2012, from other years. Table 1’s first panel shows how areas transitioned between parties. During the redistricting year, many more areas changed from a Republican to a Democratic representative, or vice versa, than in other years. The second panel shows the probability that an area reelects its old representative after an election. This is much more common in non-redistricting years, occurring with a 0.38 + 0.31 = 0.69 probability, compared to only a 0.32 probability in the redistricting year.

Table 1 Representative transitions and relationship with other gov. branches

The third panel tabulates the possible reasons why an area did not keep its incumbent. Voters rarely removed an incumbent candidate from office—this occurred in only 5% of non-redistricting area-by-year observations, and in only 1% of redistricting area-by-year observations (Rows 9–10). In most areas that changed representative during non-redistricting years, the area’s incumbent did not run for reelection (Rows 11–12). In contrast, the incumbent chose not to run in only 17% of areas during the redistricting year, while 68% of areas changed representative. Rows 13–14 show the share of areas whose last state representative won reelection, but where the area had moved into a different district than that representative—this category includes half of all areas in the redistricting year, including the vast majority of areas that changed representative ($\frac{26\% + 24\%}{100\% - 32\%} = 74\%$).

Table 1 shows that far more areas changed representative and changed party during the 2012 election than in other election years. More importantly, almost all of these changes derived from the redistricting process. Even among the 17% of areas that changed representative because the incumbent retired or lost, most had moved to a different district than the old incumbent and would have changed representative anyway. But even this fact does not reveal fully how redistricting drove the change in representatives. In total, there were twenty districts where at least one incumbent did not win reelection in 2012.⁵ This number includes eight districts with two incumbents—16 incumbents ended up in the same district as another sitting representative, and in seven of these double incumbent districts, one of the incumbents did not run. Redistricting reshuffled the relationship between voters, parties, and representatives with almost no change in voters’ electoral decisions.

2.2 District boundaries, gerrymandering, and demand for roads

Because a political commission composed primarily of Republicans drew the new House districts, partisan gerrymandering likely affected their boundaries (Daley 2016; Smothers 2013). A party that gerrymanders attempts to maximize either the number of seats that it wins or the probability that it will win a majority (Owen and Grofman 1988). The optimal solution to this problem involves conceding a minority of districts with a vast majority of voters supporting the opposing party, while crafting another set of districts with a “safe” majority for the gerrymandering party (Owen and Grofman 1988; Sherstyuk 1998; Friedman and Holden 2008; Gilligan and Matsusaka 1999; Gul and Pesendorfer 2010).

Two factors limit partisan manipulation of district boundaries. First, a majority of the legislature must favor the new boundaries. In practice, this constraint reduces changes in district partisanship (Sabouni and Shelton 2022). Second, state law requires that (i) each Ohio state House district contain $\frac{1}{99^{th}}$ of the state’s population (plus or minus 5%), (ii) districts must be geographically contiguous (contained within a single non-intersecting line), (iii) new districts must respect old district boundaries when possible, and (iv) when possible, districts should not split counties, townships, municipalities, or wards (in decreasing order of importance) (Wilson v. Kasich 2012; Smothers 2013). All states (except Nebraska, which does not require contiguous districts) are subject to the first two requirements—(iv) and particularly (iii) have been shown to limit partisanship in the redistricting process (Sabouni and Shelton 2021).⁶

In most states where a partisan committee draws district boundaries, the decision-making process is kept closely secret (Draper 2012; Newkirk 2017). A 2012 legal challenge to the new Ohio Assembly districts, Wilson v. Kasich (2012), revealed more information about that episode (Smothers 2013). Two secretaries, with the aid of Maptitude GIS software and consulting attorneys, drew the new boundaries (Dirossi 2012, pp.2–3). According to emails subpoenaed in the case, the mapmakers looked backward to past vote share in the 2008 presidential elections or to an average of Republican voteshare in five previous statewide races (Pierre-Louis 2012, pp.19–20, 28, 141). One of the secretaries described how “(p)reviously to retain a 50+ seat majority under 2008 Presidential year conditions, we had to win all seats above a 49.14% (Republican voteshare)—now we only have to hold 50 or more seats that are 50.94% or better” (Smothers 2013, p. 986). Besides past voteshare, the mapmakers communicated with sitting politicians through email and meetings (Pierre-Louis 2012, pp. 310, 326, 335–339, 344, 353–356) in addition to communicating with the official apportionment committee (Pierre-Louis 2012, pp. 319–321, 349–350).

Fig. 3

Changes in economic and road conditions. Data: (United States Census Bureau 2017; Federal Highway Administration 2017).

The mapmakers, then, assigned areas to districts based on past levels of partisan preferences. Areas shifting to a representative of the opposing party may have had systematically different levels of partisan support than other areas. With the difference-in-differences style estimation I employ, changes in party through redistricting may be correlated with the level of partisan support. My estimates remain consistent as long as changes in partisan preferences over time remain mean independent of changes in representative’s party through redistricting, conditional on the various control variables.

Although partisan redistricters may not have targeted growing areas, one may be concerned that areas changing parties came from growing districts, whose boundaries are more likely to change. Reference to Fig. 2 suggests this is untrue. Ten areas changed party in Franklin County, home to Columbus, which has grown 24% in population over the past two decades. On the other hand, eleven areas changed party in economically depressed Appalachian Ohio. Five areas changed party in the growing Cincinnati metropolitan area, while five changed party in the shrinking Canton-Massillon area south of Cleveland. To investigate the relationship between demand for roads and redistricting party change more systematically, Fig. 3 plots trends in employed population, employment, and the FHWA’s bridge sufficiency rating within areas that changed party, and those that did not. R to R areas had substantially larger employed population and employment than R to D areas—similarly D to D areas had a larger employed population and employment than D to R. A test for difference in the linear pre-period trend, however, never showed a p-value less than 0.32. R to D and D to R showed similar trends in economic growth and the condition of their roads, compared to areas that did not change party.

3 Road projects in Ohio

3.1 Funding process

I estimate how a representative’s party affiliation determines highway construction funding, using all Ohio highway projects with direct state funding from 2007 to 2016. My data, from the Ohio Department of Transportation’s “Transportation Information Management System,” includes 9066 projects with total estimated expenditure of $19.5 billion (Ohio Department of Transportation 2017). Most projects contain multiple work locations. Since the data include geographic coordinates for each work location to six decimal places, I can place each work location precisely within a state House district.⁷

The Ohio Department of Transportation (ODOT) is responsible for planning and executing road construction projects over 50,000 lane miles of road and 15,000 bridges including the state’s interstate highways, numbered federal highways, and state routes. The ODOT is a branch of the governor’s office whose director is a gubernatorial appointee. Large new construction projects in excess of $12 M⁸ require approval by the Transportation Review Advisory Council (TRAC), a committee whose members include the Director of Transportation, six gubernatorial appointees, and two legislative appointees.

Ohio’s transportation budget passes every odd-numbered year, with the legislature allocating funding within line-item categories and dictating general parameters for how these funds will be spent. The legislature also determines the fuel tax rate, which is distributed between the ODOT and local governments. Outside of the ODOT’s biennial budget, the state legislature and the governor exercise some control over the distribution of expenditure through specific programs that distribute state and federal funds. For example, the Bridge Partnership Program, initiated in 2013 by executive action, distributed $120 M of federal money for fiscal years 2015–2017—the legislature extended this program in 2017. Similarly, in 2013 Republican Governor Kasich signed a bill authorizing a $3 billion “Ohio Jobs and Transportation Plan” to improve state highways (Provance 2019, 2013; Fields 2011). Funding included $1.5 billion borrowed against future Ohio Turnpike revenue. Much debate over this bill hinged on whether these funds from the Turnpike, in predominantly Democratic Northern Ohio, would be diverted towards other parts of the state.⁹ Bargaining over distributed funds from such programs helps drive the effects I estimate later on.

In sum, a representative’s party could influence expenditure toward her district in several ways. If the representative belongs to the governor’s party, then she may be able to lobby the ODOT for funds more effectively. A majority party representative could more easily coordinate with her own party to affect expenditure through legislation, or threaten to introduce a bill altering the ODOT’s funding. Partisanship could also influence decisions by the TRAC, almost all of whose members were appointed by a Republican governor or Republican legislators over this period.

3.2 Project data and covariates

Table 2 Summary statistics, by party-area of intersection

Table 2 presents summary statistics on road expenditure at the area of intersection-by-year level.¹⁰ Recall that “area of intersection” refers to one of the 250 geographic areas formed by the intersection of old (pre-redistricting) and new (post-redistricting) House districts. These areas form the unit of observation in the empirical analysis that follows. The first and second column divide area of intersection by political party, while the third column totals all areas.

The first two rows describe the number of projects and the DOT’s estimated total expenditure on all projects. On average, each area of intersection receives $7.2M per year in construction spending on 35 projects. The third and fourth rows show number and estimated total cost of major projects, which belong to the six categories of road work with the largest average project cost.¹¹ These account for one-tenth of all projects, but more than half of total expenditure. The fifth row displays estimated total cost for projects where the primary sponsor is either the ODOT or some other state agency - roughly one-fifth of state spending goes toward local government projects.

Three variables in Table 2 measure contribution to road provision by local governments: public works expenditure by counties and townships, transportation expenditure by municipalities, and average property tax rate devoted to roads by all local governments. Ohio funds a large share of local road expenditure through state gas tax revenue, which it splits between the ODOT and distributions to local governments. Distribution of gas tax funds to local governments depends on number of registered vehicles, so I use gas tax distributions as a proxy for local residents’ road use. Employment, and employed population living in each area, come from the LEHD program of the Census Bureau.¹² Population is available only for 2010, from the Census Bureau.¹³ The last two variables—average bridge sufficiency rating, and vehicle miles traveled—provide two more measures of need for transportation expenditure. I discuss the construction of these variables in Online Appendix A.

4 Empirical specifications

I estimate how each area’s construction expenditure changes when it changes party through redistricting, testing: (i) whether areas that left the Republican party during the redistricting year (the “R to D” transitions)¹⁴ suffered a drop in funding, relative to areas that remained Republican (the “R to R” transitions) and (ii) whether “D to R” areas received a greater relative funding increase after redistricting than “D to D” areas. To motivate this approach, Fig. 4 plots average annual construction expenditure in the six years before (2007–2012) and after (2013–2016) redistricting for each category: R to R, R to D, D to R, and D to D. The bar chart indicates that expenditure fell in areas that switched from the Republican to the Democratic party, while generally increasing in all other areas—areas that remained in Republican districts experienced a nearly $3 million increase.

Fig. 4

Average road construction funding, 2007–2012 versus 2013–2016. Data: Ohio Department of Transportation (2017); Ohio Secretary of State (2017)

4.1 Regression specification

To uncover whether expenditure changes in Fig. 4 represent the effect of changing representation, I estimate the following specification, using all 250 areas of intersection for the years 2007–2016:

$$\begin{aligned}{} & {} expenditure_{it}\ =\ (\beta _1R\_to\_D_{i}\ +\ \beta _2D\_to\_R_{i}\ +\ \beta _3D\_to\_D_{i})\times (year_t\ >\ 2012)\ + \nonumber \\{} & {} \quad (\beta _4R\_to\_D_{i}\ +\ \beta _5D\_to\_R_{i}\ +\ \beta _6D\_to\_D_{i})\times (year_t\ <\ 2011)\ \nonumber \\{} & {} \quad +\ \varvec{\alpha X_{i,t-1}} +\ \delta _t\ +\ \gamma _i\ +\ \epsilon _{i} \end{aligned}$$

(1)

where $expenditure_{it}$ is area of intersection i’s road expenditure in year t. The party change indicators—$R\_to\_D_i, D\_to\_R_i,$ and $D\_to\_D_i$—capture how each area’s representative transitioned between parties during the redistricting year (2013). $R\_to\_D_i$, for example, equals one for areas that moved from a Republican district in 2012 to a Democratic district in 2013. The $\beta _1$ coefficient represents the difference in expenditure for these areas after redistricting for areas that changed from Republican to Democratic, relative to areas that remained Republican ($R\_to\_R_i$ is the omitted category). $\gamma _i$ denotes area of intersection fixed effects, while $\delta _t$ denotes year fixed effects. $\varvec{X_{i,t-1}}$ includes lagged covariates to better model demand for roads.¹⁵

Post-redistricting elections in November 2012 closely followed the change from Democratic to Republican governor and from Democratic to Republican House in November 2010. I want to identify the effect of changing from a Republican to a Democratic representative, or vice versa, conditional on the Republican party’s control of the legislature and governor’s office. Thus, I include an indicator for $(year < 2011)$, interacted with the redistricting party change indicators. The coefficient on the each of the post redistricting party-change variables ($(R\ to\ D)\times (year> 2012), (D\ to\ R)\times (year> 2012), (D\ to\ D)\times (year > 2012)$) indicates the increase in funding between the two years immediately preceding redistricting (2011–12) and the four years after.

Fig. 5

Trends in residualized expenditure. Data: (Ohio Department of Transportation 2017; Ohio Secretary of State 2017)

These regressions involve a difference-in-differences type approach, comparing R to D treated areas with R to R control areas, and comparing the relative change between D to R and D to D areas, again relative to the R to R baseline. To ensure that pre-trends do not bias estimates of $\beta _1-\beta _3$, Fig. 5 plots changes in residualized construction expenditure for each party transition category, with R to R areas compared with R to D areas and D to R compared with D to D. The y-axis shows the average residual from regressing construction expenditure on the controls in the second row of Eq. (1). As seen in Fig. 5, R to D areas trended up throughout the period, while R to R trended slightly downward. Following redistricting, R to D areas fell to a much lower expenditure level than R to R areas, without major changes in trends. From 2007–2012 D to R areas experienced an overall decrease in expenditure before flipping sign, while D to D districts generally trended upward. Motivated by this figure, I include linear time trends for each party change category in one of my specifications.

4.2 Regression results

The estimates from Eq. (1) are provided in the first column of Table 3. The coefficient for areas that changed from a Republican to a Democratic legislator—$(R\ to\ D)\times (year\ >\ 2012)$—shows that areas leaving the Republican party received roughly $4.2M less annual construction funding than those that remained in the specification without covariates, or by $3.4M when controlling for covariates. From Table 2, the mean annual expenditure on projects in each area is $7.2M, with a standard deviation of $18.8M. Thus, areas that switched from a Republican to a Democratic representative saw their funding drop by at least 18% of a standard deviation, a 47% increase relative to the mean. The coefficient estimate is statistically significant at the 95% confidence level.

Most coefficients on the covariates are not statistically different from zero. The negative and statistically insignificant coefficients on employment and employed population suggest that road spending is not only targeted toward growing areas. The positive coefficient on public works spending suggests that county and township investments complement state roads. Daily vehicle miles traveled is positively related to expenditure as expected, but never statistically significant. On the other hand, bridge sufficiency rating is significantly and negatively related to spending, reflecting higher state spending on deteriorating infrastructure.

There is no significant difference between the D to R and D to D post-period coefficients. While R to D areas suffer a decrease in funding relative to areas that remained Republican, D to R areas do not diverge from D to D towards R to R funding levels. This asymmetry may reflect an immediate decrease in funding to R to D areas from canceled or postponed projects, with no immediate increase in new projects directed toward D to R areas—planned projects may be canceled or delayed quickly, while it takes some time to develop and begin work on new construction. Alternatively, it may take some time for parties to build relationships with their base (Joanis 2011; Ansolabehere and Snyder 2006; Tavits 2009). Or this difference may reflect an advantage for areas that retain their incumbent representative after redistricting, a possibility I explore in Sect. 5.

Figure 5 suggests differential time trends in expenditure between areas that changed party, and areas that did not. Column 3 of Table 3 adds a time trend for each party change category to the lagged covariates. The difference between areas that left the Republican party and areas that stayed in the Republican party increases in magnitude to $7.1M and remains statistically significant. Interestingly, the coefficient for areas that switched from a Democratic to a Republican representative becomes less negative than the coefficient for areas that retained a Democratic representative (by about $3 M per year), although the difference remains statistically insignificant. When the Republican party controls both levels of the Assembly and the governorship, areas with Republican representatives benefit.

Table 3 Effect of redistricting party change on road construction spending

4.3 Percentage changes and number of projects

Given substantial variation in the dependent variable (see the first row of Table 2), it is important to estimate the proportional change in construction projects caused by party change. Ideally, the natural logarithm of construction projects would replace the level as dependent variable, but the log dependent variable is not defined for much of the sample—10% of area-by-year observations received zero construction projects. Instead, Poisson quasi-maximum likelihood estimation (QMLE) provides consistent estimates of the semi-elasticities, as long as the exponential conditional mean is correctly specified (Wooldridge 2010, Chapter 19). I estimate a model with both area of intersection and year fixed effects. Columns 4 and 5 of Table 3 present the results of this specification without and with controls. The coefficients show that areas belonging to a Republican district before and after redistricting receive 60% to 70% more highway funding than areas moving from a Republican to a Democratic district, consistent with estimates from the linear model.

Table B3 in the Online Appendix shows how redistricting party change affects both number of projects and number of large projects. Columns 1 and 2 of Table B3 correspond to Column 4 of Table 3, and Columns 3 and 4 to Column 5 of Table 3. Columns 1 and 3 of Table B3 show no change in the number of projects overall when areas move from a Republican to a Democratic district. Columns 2 and 4, however, show that large projects decrease by at least 43% in these areas, implying that a decline in large construction projects drives the substantial drop in expenditure within areas that moved to Democratic districts. Although partisan concerns do not enter into mundane construction decisions like maintenance and landscaping, these results imply that legislators in the majority party are able to reward their constituents by directing large projects toward their own districts.

4.4 Accounting for the senate

In Ohio, each House of Representatives district lies entirely inside of a state Senate district. Because no area changed to a senator of the opposite party during redistricting, every area whose representative changed parties either moved from having a senator and a representative of the same party to different parties, or vice versa. In bicameral legislatures, constituents may assign credit for funds received partially to their representative and partially to their senator (Chen 2010). It may be easier for a representative to direct funds toward her district in coordination with a senator from her own party. The governor or other members of the legislature may be less willing to direct funds to an area if they know that a member of the opposing party will receive credit.

To ensure that these senator alignment effects do not drive the previous results, I reestimate Eq. (1) including indicators for areas where the senator and representative were both Republican or both Democratic, each interacted with a post-2012 indicator. Table 3, Column 6 shows these estimates from the specification in Column 1 with the addition of these senator party alignment covariates, while Column 7 includes the same covariates as Column 2. The point estimates suggest that areas with senator and representative from the same party receive increased highway funding. But an F-test for joint significance of the party alignment coefficients fails to reject the null that both are zero (p-value = 0.37 with or without controls). Moving from a Republican district to a Democratic district still causes a decline in construction spending, but the decline depends on whether the senator and representative are Democratic as well. The estimates in Column 7 suggest that areas switching from a Republican to a Democratic representative receive $\$5.9M\ (-\$3.31M\ -\$2.58M)$ less in construction spending when their senator is a Republican (p-value = 0.03). But areas switching from a Republican to a Democratic representative experience almost no change when their senator is a Democrat.

5 Incumbent representatives and redistricting party change

According to the results in Sect. 4, shifting from a Republican district to a Democratic district creates a large drop in predicted construction funding. But shifting from a Democratic to a Republican district generates no corresponding increase in funding. As discussed at the end of Sect. 4.2, the asymmetry may reflect a loss of political connection—areas that change from a Democratic to a Republican district gain a party affiliation boost, but lose their incumbent representative. In this section, I use redistricting variation to test whether the asymmetry in party change effect partly reflects the advantage of retaining the same representative and belonging to the same district post-redistricting. Each area that changed party from 2012–2013 necessarily switched to a new representative. In all, 170 areas switched to a new representative from 2012–2013. Of these, only 50 changed party.

To test whether switching representatives drives a loss in funding, I reestimate Eq. (1), adding indicators for whether a Republican or Democratic area kept the same representative pre- and post- redistricting. As with the party-change variables, I interact each of these indicators with a $(year < 2011)$ indicator as well as the $(year > 2012)$ indicator. Table 4 shows the results of this estimation, adding trends for R to R and D to D districts that retained the same representative in Column 3, and using Poisson QMLE to estimate percentage changes in the last two columns. The results indicate that areas retaining their incumbent representatives attracted more funding that areas switching to new representatives. According to the specification in Column 1, areas that had belonged to the Republican party and retained the same representative attracted more funding than areas that moved from one Republican representative to another (the omitted category). R to D areas experienced a much smaller increase in funding when compared to R to R areas that changed representative than when compared to all R to R areas. On the other hand, D to R areas experience a post-redistricting increase in funding relative to D to D areas that changed representative ranging from a statistically insignificant $\mid- \$0.28M-\$1.04M\mid = \$1.3M$ in the first column to a marginally significant $\mid-\$6.13-\$0.69 \mid= \$6.8M$ in the third column with controls and time trends (p value = 0.11). Thus, after controlling for whether areas moved to a new incumbent representative, the effect of moving from a Democratic to a Republican district is estimated to be large and positive, relative to the specification that compared D to R areas to all D to D areas.

Table 4 Effect of redistricting party changes on construction spending, controlling for areas that changed representative 2007–2016

According to the regressions in Sect. 4, an increase in spending for areas that moved to Republican districts did not accompany the estimated drop in spending for areas that left Republican districts. The results in Table 4 suggest that areas moving to a new representative experience a decrease in funding, regardless of party. After controlling for this factor, D to R areas experience an increase in funding that equals or even exceeds the decrease in funding for areas that move from the Republican to the Democratic party.

All specifications in Table 4 estimate a large, statistically significant increase in funding for areas that retain their incumbent representative post-redistricting. Two explanations seem plausible—first, an incumbent representative whose district boundaries have changed may reward the old area of her district relative to the new area. She is less able to direct spending towards the new area of her district due to lack of information, as this process requires time and information. She also faces less incentive to direct more spending towards existing projects, where she may share credit with the old incumbent. Alternatively, incumbent representatives fight to retain key portions of their district during the redistricting process. Legislators and parties are least likely to part with loyal voters (Sabouni and Shelton 2022), particularly in areas where they may be in the process of making long-term distributive investments (Dixit and Londregan 1998; Joanis 2011). In either case, this finding merits more exploration in future work.

6 Variation from voting and redistricting

To conclude, I show that party changes through election outcomes are associated with different changes in expenditure than party changes through redistricting. I first estimate the effect of having a Republican representative across all years, with separate effects for years with a Democratic governor and years with a Republican governor. I account here explicitly for the alignment with the governor’s party because this changed between 2010 and 2011, and thus the effect of belonging to a Republican district likely changes as well—this is the same reason I interact party-change indicators with $(year < 2011)$ in Eq. (1). Each observation falls within one of four categories: Republican representative and Republican governor, Republican representative and Democratic governor, Democratic representative and Republican governor, or Democratic representative and Democratic governor. I estimate differences in expenditure levels within these categories, letting areas in the latter category serve as the baseline:

$$\begin{aligned}{} & {} expenditure_{it}\ =\ +\ \beta _1R\_rep\_R\_gov_{it}\ +\ \beta _2D\_rep\_R\_gov_{it}\ +\ \beta _3R\_rep\_D\_gov_{it}\ +\ \nonumber \\{} & {} \quad \gamma _i\ +\ \varvec{\alpha X_{i,t-1}} +\ \delta _t\ +\ error_{it} \end{aligned}$$

(2)

I use the same sample in this specification as in Eq. (1). The observation is again the area of intersection, observed between 2007–2016. Now, however, coefficients are identified using variation in representative’s party through both redistricting and election outcomes.

The estimated coefficients, in Table 5, demonstrate an advantage for areas represented by Republicans, regardless of the governor’s party: according to Column 2 with covariates, these areas gain $1.4M ($4.99M–$3.64M) more in funding during years with a Republican governor, and $1.3M more in years with a Democratic governor, although neither of these changes is statistically significant.¹⁶

Table 5 Effect of party and alignment with gov. on construction spending, voting changes versus redistricting changes 2007–2016

The next two regressions separate the effect of party changes due to election variation from the effect of redistricting party changes, by combining the post-redistricting party change indicators from Eq. (1) with the party and governor alignment indicators from Eq. (2).¹⁷ Table 5 Column 3 suggests that areas changing from a Republican to a Democratic representative during a regular election year lost $1.8M ($5.37M–$3.53M) annual construction expenditure during years with a Republican governor, and $1.9M in years with a Democratic governor. But areas changing from a Republican to a Democratic representative through redistricting lost $3.7M ($1.90M + $5.37M–$3.53M) in annual construction expenditure (p-value = 0.03). Adding covariates, the specification in Column 4 estimates similar coefficients to Column 3.

To test whether party changes due to redistricting have different effects on funding, I jointly test the coefficient on (R to D)$\times$(year > 2012) and the difference between the coefficients on (D to R)$\times$(year > 2012) and (D to D)$\times$(year > 2012) The p-value is 0.06 without covariates, rising to 0.16 when covariates are added. This, and the large coefficients on the redistricting change indicators strongly suggest that estimates identified through redistricting differ from those identified through election outcomes.

Outside of the redistricting year, areas changed parties in two ways. Either the incumbent representative lost the election, or the incumbent representative chose not to run and her party’s candidate failed to win the general election.¹⁸ The last set of regressions, in Columns 5 and 6, include an indicator for areas where the representative was in her first term and the previous incumbent lost. I interact this variable with an indicator for the representative belonging to the same party as the governor. This means that the first three variables are identified from years when a representative retired and a member of the opposing party replaced her. In years after an incumbent loses, her district suffers a large decrease in funding, particularly if the incumbent had belonged to the governor’s party. Under a Republican governor, areas where an incumbent Republican representative lost suffered $10.5M ($8.16M + $2.20M + $5.27M - $5.16M) less road construction expenditure (p-value = 0.05). If, on the other hand, an incumbent from the Republican governor’s party retired and a non-incumbent Democrat replaced her, her district would lose only $2.3M ($2.20M + $5.27M–$5.16M) in expenditure. These estimates may show that new representatives who have defeated a politically connected incumbent may face particular difficulties in bargaining for funding. Alternatively, these estimates may reflect unobserved selection issues regarding the type of districts whose voters remove their incumbents. Regardless, the effect of changing party through an election varies with the specifics of how that party change takes place, a finding that researchers of this topic should carefully consider.

7 Conclusion

I present evidence that policymakers allocate road construction expenditure across regions based not only on concerns for efficiency or redistribution, but also according to partisan preferences. Moving to a state representative from the majority Republican party leads to a large increase in road construction spending. I estimate an effect of representative’s party on spending comparable to or larger than estimates of party alignment on nationwide spending distribution in other studies, such as in Spain (Solé-Ollé and Sorribas-Navarro 2008) and Italy (Bracco et al. 2015). My estimates, moreover, exceed Berry et al. (2010) and Albouy (2013)’s estimates of the effect of congressional representation on federal spending in the United States. I find that areas moving from a Republican to a Democratic district receive 60% less expenditure in years where the assembly and governorship were Republican, while these studies found from 5% increase for districts aligned with the president to 31% increase for states whose entire congressional delegation belonged to the majority party. I also provide evidence that areas with a representative and senator in the same party receive substantially more transportation expenditure—this may indicate coordination of efforts, or a willingness of other Republican policymakers to aid districts where their own party will receive all of the credit. Areas represented by the majority party reap substantial benefits, particularly if the same party controls the House and overlapping Senate district. Areas belonging to the minority party suffer.

The large effects estimated in this paper probably reflect the institutional setting considered. At the federal level, each state has two senators who may belong to different parties. Although each federal representative is accountable to her own sub-state district, the prevalence of state-level block grants provides an incentive for state representatives to work across party lines in bringing funds back to the state. At the state level, however, one party represents each district in the House of Representatives. Because state house districts are geographically small and the entire population benefits from transportation expenditure, each representative faces strong incentives to direct funds toward her own district.

But I also provide evidence that my identification strategy drives my larger results. Areas where an incumbent representative from the governor’s party is voted out of office suffer a much larger expenditure decline than areas where party changes when the incumbent does not run. This is consistent with a non-random selection of districts changing parties through election results. More importantly, estimates of the change in expenditure associated with a party change through voting differ from estimates based on redistricting. This difference may derive from endogeneity of representative party changes due to election results. It may also reflect that districts changing party through voting form a selected subsample of swing districts. The comparison confirms the importance of isolating an exogenous source of variation in political party to estimate these effects. In this paper I advance the literature by applying such a credible identification strategy to this question.