American federalism dates back to the creation of the U.S. Constitution in 1787, which broadly divided the roles and authorities of federal and state governments. Since then, scholars have routinely debated the optimal level of power-sharing in the multi-layer governing structure. As a result, the federal and state dynamics of the public policy process have been the focal point in federalism studies (Bednar et al., Reference Bednar, Eskridge and Ferejohn2001; Rodden, Reference Rodden2002; Conlan, Reference Conlan2006).
The power and authority of U.S. cities are not defined in the Constitution, however. Since the Supreme Court ruling in Hunter v. City of Pittsburgh (1907), the legal status of cities has been regarded as the corporation of states (Clark, Reference Clark1985). This has shaped the development of the scholarly literature, as previous studies have examined either a city's relationship with state government or (more recently) local political dynamics. Thus, the relationship between the federal government and local governments in the public policy process remains a puzzle.
With the increasing nationalization of politics and strong party influence in state and local policy processes, federal politicians have interacted more and more with subnational political actors, and the executive relationship, in particular, has become more critical. McCann (Reference McCann2016), for example, finds that the federal government delegates a substantial portion of its policy implementation to state and local governments, which results in governors and mayors exerting significant influence in federal policy implementation. Local politicians also increasingly interact with the public and stakeholders as front-line politicians in various policy areas (Einstein and Glick, Reference Einstein and Glick2018). The president thus needs strong support from mayors to achieve his political and policy goals in an increasingly polarized environment.
Local politicians also need help from the federal government for their success in political careers and policy implementation. As federal issues have become more important to local elections, party cues and endorsements from federal politicians have become critical to local politicians (Hopkins, Reference Hopkins2018). Many local policies require the support of the federal government, and local governments have relied heavily on fiscal transfers from federal or state governments (Nunn et al., Reference Nunn, Parsons and Shambaugh2019). Importantly, many federal-local policy processes also do not involve state government; consequently, the political distance between federal and local governments has narrowed in our polarized society.
While previous research on distributive politics and federalism has focused almost exclusively on the federal-state relationship, this study analyzes the effects of the federal-local relationship in the allocation of federal grants.Footnote 1 As a significant amount of federal grants is distributed directly to cities, bypassing the state government's redistribution process, we examine the effects of the president–mayor party alignment on the allocation of federal block and project grants to 568 medium and large cities from 2005 to 2020.
The paper is organized as follows. Section 1 presents an overview of direct federal grant allocation to cities between 2005 and 2020. Section 2 introduces the various datasets, and Section 3 describes the analytic models and presents the results of our analysis, showing the effect of party alignment on presidential particularism in federal grant allocation. Section 4 concludes by discussing our findings and offering recommendations for future research.
1. Federal grant allocation to cities
Although most distributive politics research focuses on the dynamics between the president and Congress (Berry et al., Reference Berry, Burden and Howell2010; Dynes and Huber, Reference Dynes and Huber2015; Kriner and Reeves, Reference Kriner and Reeves2015), some studies have examined the interplay between federal and state governments. Larcinese et al. (Reference Larcinese, Rizzo and Testa2006) and Nicholson-Crotty (Reference Nicholson-Crotty2015), for example, find that governors of the president's party receive more federal funds. In addition, Ansolabehere and Snyder Jr (Reference Ansolabehere and Snyder2006) emphasize the influence of state government party control in the allocation of state funds to local governments, specifically counties. While Goldstein and You (Reference Goldstein and You2017) examine the federal-local relationship within the context of the local government's lobbying strategy towards the federal government, no research has explored the direct relationship between federal and local governments in the public policy process.
In 2020, for example, $6 billion of block or project grants was distributed directly to cities.Footnote 2 Figure 1(a) shows the allocation of direct federal grants to cities over a longer period, from 2005 to 2020. Although the yearly grant amount did not significantly increase during this period, the spending by the federal government more than doubled during two economic crises (the 2009 Great Recession and the COVID-19 pandemic).
Block and Project Grant to Cities (2005–2020). (a) Year-Month (2012 constant $) and (b) By Agency (2012 constant $).
The Office of Management and Budget (OMB) reviews each agency's grant allocation proposals and makes an apportionment plan to decide the funding level and timing. In this sense, the grant allocation process is controlled by the president's ex-post influence (Berry et al., Reference Berry, Burden and Howell2010). The federal government distributes grants directly to cities each year in the areas of transportation, housing, education, and community development. As shown in Figure 1(b), the Department of Transportation (DOT) allocated over $18 billion directly to cities between 2005 and 2020, while the Department of Housing and Urban Development (HUD) and Department of Homeland Security (DHS) distributed $9 billion and $8.5 billion, respectively, directly to cities during the same period. More detailed information is provided in Table B.1 in the Appendix.
The allocation of federal grants to cities varies at both the total and per capita levels, as shown in Figure 2. The largest cities, such as Chicago, New York, and Los Angeles, received a considerable amount over time (2005–2020). However, as Figure 3 indicates, Atlanta and Orlando received a high amount in 2019 and Springfield (IL), Atlanta, and Salt Lake City received relatively larger amounts in 2020 in per-capita terms.Footnote 3
Block and Project Grant to Cities between 2005 and 2020, (2012 constant $).
Block and Project Grant to Cities (2019 and 2020), by City. (a) 2019: Per Capita (2012 constant $) and (b) 2020: Per Capita (2012 constant $).
2. Data and variables
Formula grants are disbursed strictly based on eligibility criteria, whereas block and project grants allow discretion to the awarding agencies in deciding the recipient jurisdictions and allocation amount (Dilger and Cecire, Reference Dilger and Cecire2015).Footnote 4 Therefore, we use 702 federal block and project grants that are allocated directly to city governments (i.e., those that do not pass through state governments).Footnote 5
To investigate the effects of the president–mayor relationship on federal grant allocation, we incorporate two different datasets. First, for federal grant data, we use the Federal Assistance Awards Data System (FAADS), which has been the typical standard (Berry et al., Reference Berry, Burden and Howell2010; Kriner and Reeves, Reference Kriner and Reeves2015). The FAADS data, collected and managed by the Department of Treasury and Census Bureau, provide extensive information on federal funds allocation at the award-level, such as types and purpose of federal funds, allocation timing and amount, awarding agency, and recipient's name and address.Footnote 6 Second, we use the local political dataset assembled by Warshaw et al. (Reference Warshaw, de Benedictis-Kessner and Velez2022), which provides extensive information on local politics and elections for municipalities with populations over 50,000. It includes key variables like the partisanship of the mayor and other candidates, vote counts of candidates, and election month and year. While other local datasets include cities with population thresholds over 75,000, 100,000, and 125,000, the Warshaw et al. (Reference Warshaw, de Benedictis-Kessner and Velez2022) dataset provides broader information on the 568 medium and large cities with populations greater than 50,000 in the 2020 Census over a longer time period.Footnote 7 Detailed information on the Warshaw et al. (Reference Warshaw, de Benedictis-Kessner and Velez2022) local dataset is provided in Table B.2 and Table B.3 in the Appendix.
Of the 568 medium and large cities in our analysis, 63 (11%) do not have the mayor's party information for the entire 16 years (2005 to 2020), and 19 do not have the mayor's party information over 10 years. Therefore, 484 observations (5.3%) are missing from the full panel of 9,104 mayor-year observations. Democratic mayors are 51.8% of the sample and Republican mayors are 37%, while 11% of mayors’ partisanship is unknown. We consider candidates with unknown partisanship as independent (or non-partisan), as there is no evidence of partisan-related activities in their records.Footnote 8 With respect to mayoral party alignment, 44.5% of mayors are from the president's party, while 17.3% of mayor-year observations are classified as secure (i.e., mayor and election runner-up are from the same party). Also, 43.7% of governors are Democrats and 56.3% are Republicans during this time. In city-year observations, 23.4% exhibit uniform party alignment among the president, governors, and mayors, while 9.6% display the same party alignment among the president, governors, mayors, and runner-ups. With respect to cities by state in the dataset, California has the most (90), followed by Texas (48) and Florida (41).
For our dependent variable – Total Grant Per Capita – we use per-capita level grant allocation, based on 2012 real dollar value, to the cities. Since there are many cities that do not receive any federal grants each year, taking the logged amount of grant allocation raises concerns about zero-inflated bias in the analysis.
We include four key independent variables to analyze the effects of three types of president–mayor party alignmentsFootnote 9
Pres_Mayor_Party: Takes a value of 1 if the president and mayor are of the same party, and 0 otherwise.
Pres_Secure_Mayor_Party: Takes a value of 1 if the president, mayor, and mayoral runner-up are of the same party, and 0 otherwise.
Pres_Gover_Mayor_Party: Takes a value of 1 if the president, governor, and mayor are of the same party, and 0 otherwise.
Pres_Gover_Secure_Mayor_Party: Takes a value of 1 if the president, governor, mayor, and mayoral runner-up are of the same party, and 0 otherwise.
Why would presidents prioritize the allocation of federal funds when the mayor's position is secure? Previous research in distributive politics has demonstrated that the president allocates a greater amount of federal funds to co-partisan governors or legislators as pork barrel benefits (Berry et al., Reference Berry, Burden and Howell2010; Dynes and Huber, Reference Dynes and Huber2015; Kriner and Reeves, Reference Kriner and Reeves2015). Stated differently, these findings show that the president exhibits a higher degree of particularism towards core jurisdictions. Consequently, we hypothesize that the president prioritizes co-partisan mayors regardless of their political circumstances, as a core city. We posit that cities with secure co-partisan affiliations, denoted as Pres_Secure_Mayor_Party or Pres_Gover_Secure_Mayor_Party, exhibit similarities to core jurisdictions when analyzed at the state, county, or district levels.
Finally, we also include three city-specific socioeconomic variables as controls, following Kriner and Reeves (Reference Kriner and Reeves2015): the logged value of total population, the logged value of income per capita, and the poverty rate.Footnote 10
3. Methods and results
We use two-way (city and year) fixed effects with clustered standard errors by city to determine the effect of the four types of president–mayor party alignments on the federal grant process for the period between 2005 and 2020. Two-way fixed-effects controls for city-specific time-invariant features.Footnote 11 We thus identify initially based on change in the partisanship of the president and/or change in the partisanship of a city's mayor.
We employ four analytic models. First, we examine the effect of the president–mayor party alignment on the president's particularistic grant allocation to cities (Pres_Mayor_Party). Second, we analyze the party alignment effect in one-party election cities (Pres_Secure_Mayor_Party), where the mayor and mayoral runner-up are from the same party (which we define as secure party cities). Third, we examine the effect of the president–mayor party alignment when (a) the president, governor, and mayor belong to the same political party (Pres_Gov_Mayor_Party). Fourth, and finally, we analyze the effect of the president–mayor party alignment when the president, governor, mayor, and mayoral runner-up all belong to the same party (Pres_Gov_Secure_Mayor_Party).
Our results appear in Tables 1 and 2. For detailed information, refer to the complete tables in Appendix, specifically Table B.4, Table B.5, and Table B.6.
President–Mayor Party Alignment
Notes: Standard errors (clustered by city) in parentheses
*** p < 0.01, ** p < 0.05, * p < 0.1.
President–Mayor Party Alignment, by the President's party
Notes: Standard errors (clustered by city) in parentheses
*** p < 0.01, ** p < 0.05, * p < 0.1.
Table 1 shows the effect of president–mayor party alignment on grant distributions to 568 cities between 2005 and 2020. Although the simple effect of party alignment washes out in Model (1), we find a positive and statistically significant effect in Models (2), (3), and (4). The president allocates $3.21 per capita more to co-partisan mayors in secure party cities, and $3.545 per capita more to co-partisan mayors in states with co-partisan governors, and $5.223 more per capita to co-partisan mayors in secure party cities in states with co-partisan governors. How big are these effects? As Table B.3 in the Appendix indicates, the mean distribution was $15.98 per capita with a standard deviation of $51.7 per capita. So these significant effects would represent a 6.2%, 6.9%, and 10.1% of a standard deviation increase, respectively.
Table 2 goes a step further by breaking out the president by party, which allows us to compare effects across Republican and Democratic presidents during our time period.Footnote 12 Identification here comes from the changing partisanship of a city's mayor (models 1 and 2), the changing partisanship of the mayoral runner-up (models 3 and 4), the changing partisanship of a city's mayor or the governor (models 5 and 6), and the changing partisanship of the mayoral runner-up or governor (models 7 and 8).Footnote 13 We see that Democratic presidents are more particularistic in terms of grant allocation, and these effects are statistically significant in terms of co-partisan mayors in secure party cities (Model 3), co-partisan mayors in states with co-partisan governors (Model 5), and secure party cities in states with co-partisan governors (Model 7). Conversely, Republican presidents show nominal negative effects in cases where there is simple or secure party alignment with the president, and they exhibit a lower level of positive impact when the governors share the same party affiliation (model 6).
Democratic presidents allocate $4.439 per capita more to secure co-partisan cities, $5.689 per capita more to cities with co-partisan mayors and co-partisan governors, and $6.312 per capita more to secure cities with the co-partisan governors. These amounts represent a 8.6%, 11%, 12.2% of a standard deviation increase, respectively. Republican presidents’ particularism is only statistically significant in the case of co-partisan mayors in states with co-partisan governors, resulting in an additional allocation of $2.761 per capita, equivalent to a 5.3% of a standard deviation increase. The effect size, however, is less than half of that observed for Democratic presidents.
These results provide a definitive answer to the question motivating this paper: presidents do favor co-partisan mayors in the allocation of federal grants.Footnote 14 Particularism takes the form of significantly more per-capita grant money in the cases of co-partisan mayors in secure party cities, co-partisan mayors in states with co-partisan governors, and secure party cities in states with co-partisan governors. These significant results are driven almost exclusively by Democratic presidents. And as our timeframe spans 2005–2020, this is in fact an “Obama effect.”
What might explain this partisan difference – Democratic versus Republican presidents – in the allocation of federal grants? This is beyond the scope of the present analysis, and we believe it is question that is ripe for future study. That said, one plausible explanation, in our minds, is that presidential election (and reelection) strategies in recent decades have differed considerably by party. Democrats have focused on turning out voters in urban areas, while Republicans have relied upon mobilization efforts in more rural areas (Mettler and Brown, Reference Mettler and Brown2022). Thus, keeping cities in the party column is increasingly more important for Democratic presidents than Republican presidents. This might lead to – among other things – more grant money flowing into co-partisan cities during Democratic presidencies.
4. Conclusion
Recent research in American politics finds that the president strategically uses federal funds as one of several powerful governing resources, along with appointments and the veto, to pursue his policy and political goals. As the leader of his party, the president provides selective spending benefits to his co-partisans in Congress. We find that the partisan alignment in grant allocation goes beyond the federal level. Specifically, we find evidence of presidential particularism in the distribution of federal grant funds to co-partisan mayors from (a) secure party cities, (b) cities in states where the governor is also a co-partisan, and (c) secure party cities in states where the governor is also a co-partisan. And these results are driven almost exclusively by Democratic president. Future work should extend our timeframe to explore the robustness of this result, as well as examine the partisan motivations of presidents in distributing grants (along with funding alternatives like loans and contracts).
Our findings also suggest that recent efforts to construct advanced local datasets are critical for advancing our collective knowledge of local political economy effects. And avenues for additional data collection exist. For example, Brollo and Nannicini (Reference Brollo and Nannicini2012) analyze federal grant allocation with respect to the timing of local elections in Brazil, but there has been been no similar research in the U.S. context. Collecting the dates of local elections would solve this problem. Other research might analyze grant distribution to counties, special districts, NGOs, and private institutions by highlighting the relationship between the federal government and other subnational actors; this is possible now, as the FAADS dataset provides broad information on the receiving jurisdictions.
Finally, studies of federalism might better highlight the close relationship between federal and local governments. Mayors are deeply connected to the federal government in broad policy areas, and they have attempted to exert a stronger influence in the policy process by organizing cooperative institutions, such as Councils of Governments (COGs) and Metropolitan Planning Organizations (MPOs). These relatively new local institutions offer many opportunities for further study.
Acknowledgements
We thank Justin de Benedictis-Kessner and Christopher Warshaw for providing a great local political dataset. All errors are the authors’ alone.
Jeffery A. Jenkins thanks the Lusk Center for Real Estate at the University of Southern California for a research award to help support the research conducted for this article.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/psrm.2024.4. To obtain replication material for this article, https://doi.org/10.7910/DVN/2DM41P
Open access
