Contracts,Agency Vulnerability,and the Allocation of Federal Funds

Abstract

Legislators have the potential for credit claiming from agency spending in a congressional district only when the awarded contracts or grants are sufficiently large to be noticed or appreciated. Therefore, unlike previous research on distributive politics, we examine the allocation of contracts considering the size of the contracts, not just the overall spending or the numbers of contracts per congressional district. We use a difference-in-difference analysis to evaluate how agencies altered their allocations of contracts in response to the 2006 congressional elections, in which partisan control of the Congress changed to the Democrats. We find that federal agencies responded to the election by allocating larger contracts to Democratic districts. In addition, we find that agencies whose programs received lower scores on the Bush administration’s Program Assessment Rating Tool (PART) allocated larger contracts to Democratic than Republican districts than did agencies whose programs received higher PART ratings.

Keywords

procurement distributional politics pork

Introduction

These days it is not often that we have good news to report about an American manufacturer. So as a strong supporter of American manufacturing and local manufacturers, I am very proud to be able to announce today a new $670,000 U.S. Department of Defense contract for Hoist Liftruck . . . This contract is supporting almost 40 jobs here at Hoist and is contributing to the demand that is spurring Hoist to hire new workers., & it’s not just Hoist and its workers that reap the rewards. Hoist’s supply chain is based heavily in the Midwest, which means that there are a lot of other regional manufacturers that are going to see more business as a result of this contract. Rep. Dan Lipinski (D-IL), October 2011 press release.

Having founded Hoist Liftruck and built it into a U.S. manufacturing stalwart over the last 17 years, I can tell you that this contract is supporting jobs not just at Hoist but at the many other Midwest companies in our supply chain . . . On behalf of Hoist Liftruck and its employees, I want to thank Congressman Lipinski for standing up for Hoist and American manufacturing. Hoist Liftruck President Marty Flaska, quoted in Lipinski’s press release.

For the most part, the press release seems unremarkable. A Representative is claiming credit for a sizable contract being procured, through his assistance, for his district. Representatives want to show how they are getting the most for their constituents, which is how they get reelected (Fenno, 1973; Mayhew, 1974).

As for the federal agencies awarding contracts, they must choose which vendors to award contracts. They are bound by the Federal Acquisition Regulation (FAR), but typically agencies have significant discretion to administer contracts according to agency priorities (Bickers, Evans, Stein, & Wrinkle, 2007; Rundquist & Carsey, 2002; Wilson, 1973). In this article, we examine what motivates the federal agencies in awarding contracts. That federal agencies may have their own self-interested motives is not itself new to the literature (Banks & Weingast, 1992; Krause, 1999; Moe, 1984; Niskanen, 1994). Multiple authors have proposed that strategic considerations play a role in awarding grants or contracts, with additional funds directed to senators or representatives who are supporters of the president, located in swing districts (Gordon, 2011; Hudak, 2014), are ideological allies of the Secretary or President (Berry, Burden, & Howell, 2010; Bertelli & Grose, 2009; McCarty, 2000), or are in a position to help the agency itself (Arnold, 1979; Berry & Gersen, 2010; Stein & Bickers, 1995).

This article departs from the existing literature in several ways. First, it expands on the public administration literature on contracting. Second, unlike most of the prior works on distributional politics, we examine contracts rather than grants. Third, our work is also distinguished by its orientation on how decisions on strategic contract allocations may differ by contract size. Small contracts or grants do not offer the same opportunities for legislators to get recognition and acclaim for their supportive efforts as larger federal awards. Therefore, we distinguish among the sizes of the contracts to determine whether size matters. The extant literature has only looked at total spending or the total numbers of grants or contracts in a district or a state (Carsey & Rundquist, 1999; Lowry & Potoski, 2004; Stein & Bickers, 1995).

Like Berry et al. (2010), we use a difference-in-difference approach to make causal claims about the allocation of resources by federal agencies. However, where Berry et al. looked at how grant awards followed the President’s electoral priorities, we consider how the federal bureaus may appeal to Congress to address their own interests. We look at the 2006 congressional elections as a natural experiment to evaluate whether federal agencies altered their allocations of contracts to Republican and Democratic districts as a consequence of the election, which marked a shift in control of Congress from one party to the other. In addition, we supplement this analysis with propensity score matching to control for district characteristics and further understand how agencies shifted their contract awards in response to the congressional change in partisan control.

Background on Contracting

Federal procurement contracting, which refers to the process by which the government purchases goods and services from private parties, has expanded enormously in recent years (Frederickson & Frederickson, 2007). Contracting potentially offers the benefits of private competitive markets, with high-powered incentives, and the market discipline missing in government bureaucracies (DeHoog, 1990; Savas, 2002).

A primary line of research has considered the basis for the decision of government agencies to contract out. Within the private sector, the determination of whether a private firm relies on internal production or market procurement of goods and services (“make it or buy it”) is based on transaction costs (Williamson, 1981). Likewise for the government, the decision to outsource rather than internal production follows the transaction cost explanation (Williamson, 1985, 1999) based on production cost efficiency and transaction cost minimization (Brown & Potoski, 2003a, 2003b; Globerman & Vining, 1996; Savas, 2000).

Contracting out becomes less attractive as either bargaining costs (the costs of negotiating and administering the contract) or the risk of behavior opportunism by the buyers or sellers in the market rises. In that case, in-house (bureaucratic) supply is more likely to make sense (Williamson, 1985, 1999). For example, Levin and Tadelis (2010) found that as the contracts become harder to write and administer, services become less likely to be contracted out—a one standard deviation change in contracting difficulty is associated with a 40% reduction in the likelihood of privatization.

The nature of the good or service being produced matters in the “make-it or buy-it” decision. Opportunism costs refers to the price of guarding against the possibility that the contractor may have the potential for contract recipients to shirk from their responsibilities, or take advantage of unforeseen circumstances, and thereby shift costs to the government. Its potential increases based on “asset specificity,” the degree to which the good or service produced is committed to a specific task, and thus cannot be redeployed to alternative uses (Brown & Potoski, 2003a, 2003b; Globerman & Vining, 1996). Greater asset specificity increases the risk of opportunism and decreases expected outsourcing.

Politics may also matter. For example, the choice to outsource services may depend on the existing private benefits to politicians of keeping service provision inside the government (Boycko, Shleifer, & Vishny, 1996). Also, political factors, such as the presence of a city manager and the influence of public employee unions, explain some of the variations in local government human service contracting (Levin & Tadelis, 2010; Lu, 2013).

One interesting stream of public administration research has been in the consequence of difficulty in measuring performance and establishing accountability in third-party contracts with private organizations or nonprofits for the production of goods and services (Brown, Potoski, & Van Slyke, 2006; Frederickson & Frederickson, 2007; Kettl, 1993; Radin, 2006). Government agencies face principal-agent problems because it is difficult to monitor the contracts of private vendor.

Assuming a competitive market, the decision to use contracts as a tool for program execution partially explains which firms might receive contracts. In particular, when agency contracting staff is adequate, they may use the tools, such as competitive bidding, to find the most efficient vendors. Thus, firms offering the lowest cost bids, in competitive markets with high transparency associated with product production and thereby offering low risks of opportunism should receive contracts. Agencies, especially those decentralized, reliant on a local field office structure, have the capacity to steer contracts to particular locations and preferred vendors based on how the agency structures the bid proposals and selection procedures (Bickers et al., 2007; Rundquist & Carsey, 2002). What has not been explored sufficiently is the extent to which other influences, such as politics, may affect the awarding contracts.

Strategic Allocation of Contracts

Several authors have highlighted how grants and contracts may be distributed in a way to support the president’s political goals. Berry et al. (2010) looked at grant spending from 1984 to 2004 and considered the impacts of changes in presidential and congressional partisan control. To control for district-level political factors and redistricting that influence the receipt of federal outlays, the authors used a difference-in-difference approach based on district and county fixed effects. They found that partisan alignment with the president resulted in increased overall grant distributions, both in terms of their value and their numbers of grants. In addition, districts represented by a member of the president’s party received more federal funds, with mixed results for members of the majority parties in the House and Senate (significant when controlling for counties, as opposed to districts), and no significantly higher distribution levels for congressional committee chairs.

Gordon (2011) found even more direct evidence, based on contract awards by the General Services Administration (GSA). He used a White House list of 56 priority congressional districts for the 2004 elections that labeled some districts as requiring “defense” and others as “targets,” where the Democrats were vulnerable. The analysis showed that vendors in “defense” districts received larger contracts from GSA, in terms of monetary value, than did vendors in other unidentified districts. He also found that districts were not funneled more contracts; rather, the contract sizes were larger than expected. Similarly, Hudak (2014) found that for the period of 1996 to 2008, agencies distributed project grants so that a greater proportion of grant spending went to key swing districts, strategically relevant in presidential electoral politics. Similarly, McCarty (2000) described how agencies allocate greater funds to presidentially electorally important geographic constituencies or to legislative allies of the President.

Bertelli and Grose (2009) took a slightly different course, arguing that agencies distribute funds according to the ideological preferences of cabinet secretaries and the goals of presidential electoral politics. They examined Department of Defense (DOD) contracts and Department of Labor (DOL) grants to states from 1991 to 2002 and considered the allocation as a function of the ideological difference between the relevant cabinet secretaries and senators. Ideological congruence between senators and the DOL or DOD was associated with significantly larger amounts of grants or contracts, respectively.

Agencies may also use grant or contract fund distributions to serve their own ends, separate from the interests of the President, as a way to influence Congress. Arnold (1979) proposed that federal agencies will “allocate benefits strategically [among congressional districts] in an effort both to maintain and to expand their supporting coalitions,” (p. 207) primarily to protect or increase their programs’ budgets. He found that the bureaus dispersing water and sewer grants to model cities prioritized districts whose representatives served on the appropriations committees or had oversight responsibilities for the agency.

In particular, agencies are able to appeal to legislators by offering them something they want—pork. Mayhew (1974) described Congress members as “single-minded seekers of reelection” who engage in three reelection-related activities: advertising, that is, getting favorable name recognition; credit claiming, most likely on “particularized benefits,” but which must be plausibly connected to the member; and “position taking” on policy issues (p. 49). When agencies make decisions on project grants or contracts, the members of Congress may get an opportunity to claim credit for their district’s good fortune, either in a public announcement or a private communication (Bickers et al., 2007). Thus, particularized benefits or “pork” offers a route to reelection (Carey & Shugart, 1994; Katz, 1986; Shepsle & Weingast, 1984).

To establish an electoral connection between the distributive benefits of a grant or contract and a Congress member, several conditions must be met. First, efforts must be “traceable” (Arnold, 1990), meaning the member’s actions are seen as steering benefits to a particular constituency, clientele, or location. Second, credit claiming requires that the grants or contracts to the district be sufficiently large. Thus, aggregate totals of grant or contract spending (or their numbers) in a district may not offer adequate information for credit claiming opportunities.

The findings from the extant literature lead us to our first hypothesis:

Hypothesis 1: Agencies allocate larger contracts to the politicians who help them the most. When political control of Congress changes from one party to another, agencies will allocate larger contracts to those in the party now in control of Congress.

For example, Bickers et al. (2007) established both of these conditions by directly measuring the impact of Congress members’ claims of credits for distributive benefits on voters’ electoral support. From congressional websites, they collected data on legislators’ credit claiming for earmarks and federal grants to see their impact on voters’ electoral support for incumbent House members in the 2006 election. One curious finding was that the effect of credit claiming was neither unconditional nor uniform; its impacts on voter support for the incumbent were frequently negative, affected by both the voter’s and the candidate’s party (Bickers et al., 2007).

Perhaps, it is not surprising that there are potential political costs associated with not allocating grants or contracts based on merit, but opting for patronage instead. The electoral or political impacts of steering public resources may vary based on voter perceptions (Gordon, 2011). For example, Sen. Ted Stevens’ “bridge to nowhere” and other such projects became popular targets in the 2006 election, symbolizing the corruption of incumbency (Gordon, 2011). Moreover, some laws, notably the Hatch Act (5 U.S.C. xx7321-7326), prohibit federal employees such as those involved in making grant or contract award decisions from engaging in politically motivated activities. As a result, members must carefully frame their public credit claiming to avoid backlash. Very large and identifiable “pork” can become an electoral target. Private communications in which the legislator attributes his or her actions to a vendor getting a contract runs a different set of risks, particularly if the vendor is a campaign contributor. Very large grants or contract awards may be subject to extra scrutiny if there are claims of any interventions or deviations in standard procedures.

Distinguishing Among Federal Agencies

There are other factors that influence how agencies distribute grants or contracts. One factor may be if an agency were more vulnerable to losing appropriations and so more likely to aggressively curry favor with Congress. Such a situation emerged during the Bush administration as a result of the Bush administration’s Program Assessment Rating Tool (PART), a comprehensive program evaluation tool. The Office of Management and Budget (OMB) used the PART to publicly classify federal agencies’ programs as Effective, Moderately Effective, Adequate, Ineffective, or Results Not Demonstrated (see Gilmour & Lewis, 2006b; Heinrich, 2012; Joyce, 2011). Did agencies with programs that had been described as “ineffective” behave differently in allocating resources than those whose programs were designated “effective”?

The PART used a standardized questionnaire that covered a program’s design, use of strategic planning and performance measures, financial management, and results. The PART asked roughly 25 questions about each program’s performance and management, including the following: Is the program’s purpose clear? Is it well designed to achieve its objectives? Is the program well managed? Does the program achieve its goals? The question responses were coded to yield a numerical score, so program ineffectiveness could be quantified and progress (or lack thereof) highlighted.

The PART offered OMB an assessment device that might have motivated agency performance through threats of budget cuts. Some evidence suggests that PART scores influenced administration budget choices, such that higher rated programs received more funding in the President’s budget request to Congress (Gilmour & Lewis, 2006a; Mullen, 2006). It also affected agency behavior, particularly agencies that were of the same ideological orientation as the President (Lavertu, Lewis, & Moynihan, 2013).

Agencies whose programs received low PART scores might have felt at risk of budget cuts in the congressional appropriations. No administration had ever so explicitly and systematically categorized programs with quantifiable scores and labeled some programs as ineffective or as being unable to demonstrate their results (Gilmour, 2007; Joyce, 2011). Administrations previously might critique programs as part of a justification for a budget cut in the annual budget. However, the across-the-board assessment was new. Unlike the 1993 Government Performance and Results Act, which allowed agencies to explain away poor performance (agencies designed the program goals, targets, and did not control for environmental factors in their predicted annual accomplishments), the PART evaluations were controlled by the OMB and included as part of the President’s Management Agenda (Kasdin, 2010). As a result, there would be considerable uncertainty on the part of the agencies as to how Congress would respond to the evaluations, such as when OMB proposed budget reductions or program terminations for the low-rated programs. Moreover, for most of the PART’s history (2003-2008), the President, the House and the Senate were both controlled by the Republicans. Therefore, agencies might have hypothesized that Congress was unlikely to be concerned that the PART program reflected the partisan preferences of the Bush administration (Dull, 2006).¹

In response, agencies with low-rated programs may have attempted to curry favor with Congress. We look for evidence that agencies compensated for the increased appropriation vulnerability. Our second hypothesis investigates whether bureaus with programs with low PART scores allocated larger contracts to Republican legislators before the 2006 elections and to Democratic representatives afterward, when control of the Congress had shifted.

Hypothesis 2: Agency contract decision responses were not uniform in response to the change in congressional partisan control. More vulnerable agencies were more likely to allocate their resources in a strategic way and therefore act more responsive of the congressional electoral change.

Method

To determine whether agencies’ allocation of various contract sizes differed before and after the 2006 election, we first need to categorize contracts according to their value. Furthermore, the value of the 104,936 contracts ranges from US$0.01 to US$81 million, with a mean of US$109,000 and a median of US$9,000 (see Table 1). The vast difference between the mean and the median also indicates that the distribution in the value of contract is right-skewed; the majority of the contracts are very small (87% of the contracts have a value below the mean), whereas a few contracts are exceedingly large, neither of which can be simply ignored or excluded as outliers. We cannot simply categorize contracts using mean, median, and standard deviation with equal numbers of contracts in each category because they would be dominated by small contracts.

Table 1.

Categorization of Contracts.

Type of contracts	Value range (in US$)	Number of contracts	Percentage of all contracts (%)
Small contracts	(0, 100,000)	90,263	86.0
Moderate contracts	(100,000, 1,000,000)	12,198	11.6
Large contracts	(1,000,000, 10,000,000)	1,783	1.7
Very large contracts	(10,000,000, + ∞)	88	0.1

Accordingly, because the contract value is not normally distributed² and also right-skewed, we took a nonlinear ordinal logit approach based on the categorization of contract value, instead of a normal ordinary least squares (OLS) model on contract value. First, we categorize contracts by order of magnitude, a relatively more subjective method. Small contracts are those valued from US$0 to US$100,000, moderate contracts are from U$100,000 to US$1 million, large contracts are from US$1 million to US$10 million, and very large contracts have a value greater than US$10 million. As a robustness check, we also categorize contracts based on quartiles, that is, categorize contract by 25th, 50th, and 75th percentile of contract values.

We then use the following difference-in-difference analytical frame in an ordinal logit model to estimate how agencies allocate different sizes of contracts in response to the election:

\begin{array}{l} C o n S i z e_{i, t} = α + β_{1} D e m_{i, t} + β_{2} E l e c t i o n_{t} \\ + β_{D D} D e m * E l e c t i o n_{i, t} + C + D + A + e_{i, t} \end{array}

(1)

ConSize_i,t is the categorization (1 = small, 2 = moderate, 3 = large, 4 = very large) of contract i in year t. As a robustness check, we also evaluate a dependent variable using a categorization based on quartiles, using four quartiles, ranked from low to high, with equal number of contracts in each group. Dem_i,t is the district partisanship indicator, and equals one if the contract is given to a Democratic district. Election_t is the postelection indicator, and equals one if the contract was distributed after the election. β_DD, the coefficient on the interaction term between district partisanship and the postelection indicator, is the difference-in-difference estimator and indicates whether there is any statistically significant change in how agencies allocate different sizes of contracts after the election.

We also include a number of control variables in the model. C is a vector of contract-level control variables, including whether the contract is a sole-sourced contract (it does not receive multiple bids), and whether the recipient of the contract is an educational institution, is a woman-owned, veteran-owned, minority-owned, or Native American-owned entity, is a nonprofit organization, or is an emerging small business.

D is a vector of preelection district-level control variables, including the population density, education level, racial component, employment, income, and total area of the district. Regional fixed effects (Northeast, Midwest, South, and West) are also included as controls for additional site-level variations.

We also include A, a bureau-level partisanship identity index of the agency who allocates the contract. The index is based on the natural log ratio of average spending (using real dollars) for those years in which Congress is controlled entirely by the Democratic Party to the average spending in years in which the Congress is controlled entirely by the Republican Party from 1976 to 2008:

A = \ln (\frac{\begin{array}{l} A v e r a g e a n n u a l d i s c r e t i o n a r y \\ s p e n d i n g w i t h a D e m o c r a t i c C o n g r e s s \end{array}}{\begin{array}{l} A v e r a g e a n n u a l d i s c r e t i o n a r y \\ s p e n d i n g w i t h a R e p u b l i c a n C o n g r e s s \end{array}})

A large positive value of the index indicates the agency’s (bureau’s) tendency toward favor by the Democratic Party, whereas a large negative number suggests the opposite (Kasdin & Lin, 2015).³ The partisanship identity index is a revealed preference of the Congress, not a stated preference, as offered by Gilmour and Lewis (2006a) who evaluate agency partisanship identity based on the statements of political candidates. Unlike the Clinton and Lewis (2008) approach, the measure is an objective assessment, rather than a more subjective assessment of ideological orientation, based on expert opinions. It reflects bureau-level data, rather than departmental assessments in contrast to the other measures.

In addition to how agencies overall responded to the election in their allocation of different sizes of contracts across Democratic and Republican districts, we are also interested in whether more “vulnerable” agencies responded in a stronger way. More vulnerable agencies might have been more eager to protect themselves by currying favor from those in power, and thus might have been more willing to allocate larger contracts to Democratic districts when the Democratic Party was in power.

We use PART ratings, specifically its Section 4 Outcome Rating, as an indication of agency vulnerability. Agencies with lower program ratings may have received smaller budgets in the future and were therefore more vulnerable; thus, they were more likely to actively respond to the election to protect their interests. The program ratings were then aggregated to bureau level. We add this PART score variable to Model 1, making the following triple difference model:

\begin{array}{l} C o n S i z e_{i, t} = α + β_{1} D e m_{i, t} + β_{2} E l e c t i o n_{t} + β_{3} P A R T_{i} \\ + β_{4} D e m * E l e c t i o n_{i, t} + β_{5} E l e c t i o n * P A R T_{i, t} \\ + β_{6} D e m * P a r t_{i, t} + β_{D D D} D e m * E l e c t i o n * P A R T_{i, t} \\ + C + D + A + e_{i, t} \end{array}

(2)

where PART_i are the average of PART program ratings (the rating made in or closest before the period of 2006-2008) for the agency (bureau) that allocates contract i. Its interaction terms with the postelection indicator and the district partisanship indicator are added as controls and the coefficient on its interaction term with both the postelection indicator and the district partisanship indicator is the triple difference estimator. The triple difference estimator indicates the magnitude and significance of how much agencies with lower PART ratings differ from other agencies in their allocation of different sizes of contracts.

Moreover, to better control for district-level factors, we also apply a propensity score matching technique as an additional check to Model 1. We match congressional districts based on socioeconomic and demographic variables, and then compare the change in the number of each type of contract received by districts before and after the election. The matched districts should be very similar to each other socioeconomically and demographically, and the only difference should be the partisanship of the district. Based on the matched districts, we calculate the change in the number of each type of contract received after the election to see whether there is any statistical difference in the change in the number of different sized contracts between the Democratic and Republican districts.

Table 2 shows descriptive statistics of the variables.

Table 2.

Descriptive Statistics.

Variable name	M	SD	Minimum	Maximum
Contract level
Obligated amount	0.11	0.77	0.01	81,000,000,000
Contract size	1.14	0.42	0	4
Sole-sourced contract	0.55	0.50	0	1
Is educational institution	0.02	0.12	0	1
Is woman owned	0.11	0.31	0	1
Is veteran owned	0.05	0.22	0	1
Is minority owned	0.20	0.30	0	1
Is native American owned	0.02	0.12	0	1
Is nonprofit organization	0.08	0.27	0	1
Is emerging small business	0.03	0.17	0	1
PART rating (Section 4, Outcome)	74.3	8.04	33	86
Agency partisanship identity	0.29	0.62	−0.59	1.33
Election	0.43	0.50	0	1
Congressional district level
House Democrat	0.59	0.49	0	1
Percentage White (%)	62	22	0	97
Land area (1,000 km²)	26.2	157	0.00	1,480
Population density (population per km²)	0.001	0.001	0.00	0.03
School attainment	6,581	2,375	0	12,769
Employment percentage	61.2	13.7	0	73.9
Median income	54,785	18,679	0	105,627

Note. PART = Program Assessment Rating Tool.

Data

The bulk of our data come from USAspending.gov, which has a data set containing every government contract on record. We track the amount of every prime award federal contract in each federal department for fiscal years (FY) 2006 and 2008. FY 2006 started on October 1, 2005, and ended on September 30, 2006, concluding before the change in Congress. FY 2008 reflected the work of a Congress controlled by a new party. As FY 2008 started on October 1, 2007, and ended on September 30, 2008 (i.e., before the Presidency changed parties), the allocation of contracts was not affected by changes in the Presidency.⁴

Due to the large volume of records, we limit our main sample to the Departments of Commerce and Treasury, plus the Environment Protection Agency, the Nuclear Regulatory Commission, the U.S. Agency for International Development, and the Small Business Administration.⁵

We have a total of 241,227 records for 2006 and 2008 for the six departments and independent agencies comprising our main sample. We exclude three types of records from our analysis: (a) incomplete records, those without information on place of performance, are excluded because we need to link the contract to the specific congressional district; (b) records of contracts for work performed in Washington, DC, are excluded because the district is not represented by any Congressperson or Senator and therefore is not the focus of our study; and (c) records with a zero or negative transaction amount are excluded, because this designation indicates that the contract is still ongoing. After omitting these records, our data set contains 104,936 records, which form our basic sample of analysis.

We also use census data for congressional district control variables from the American Community Survey as well as White House OMB archival data on PART ratings, and agency’s annual budget authority.

Results and Analysis

From 2006 to 2008, we see an overall increase in the total funding to districts (see Table 3). Although there is a decrease in the number of contracts districts receive, the average value of contracts increased by 41.05%. This figure translates into a 35.81% increase in the size of contracts for Democratic districts and a 25.61% increase in the size of contracts for Republican districts. Although the number of contracts received in both Democratic and Republican districts decreased, the latter saw a dramatic change, falling by almost half (46.21%). This results in a decrease in the total spending to Republican districts of 32.31%, compared with an increase in the total spending to Democratic districts of 8.19%.

Table 3.

Funding to Democratic and Republican Districts Before and After the 2006 Congressional Election.

		Total	Democratic districts	Republican districts
Number of contracts	2006	59,384	31,590	27,794
Number of contracts	2008	45,552	30,575	14,977
Average value of contracts	2006	92,907	110,958	72,391
Average value of contracts	2008	131,043	150,693	90,930
Total value (million)	2006	5,517	3,505	2,012
Total value (million)	2008	5,969	4,607	1,361

The regression results for Models 1 and 2 are presented in Table 4. Coefficients are standardized to allow for comparison across coefficients. Columns 1 (a1) and 2 (a2) reflect the categorization of contracts by order of magnitude, whereas columns 3 (b1) and 4 (b2) rely on categorizing contract values by quartiles. Model 1 is tested based on (a1) and (b1), whereas Model 2 is tested in (a2) and (b2). The two forms of contract categorizations show consistent results, both in terms of the signs and the significance.⁶ Because of the similarity in the results, we will focus on the (a1) and (a2) for our discussion of the regression results.

Table 4.

Regression Results (Ordinal Logit, Standardized Coefficients).

Regression models	Categorize contracts by order of magnitude		Categorize contracts based on quartiles
Coefficients	(a1)	(a2)	(b1)	(b2)
House Democrat	−0.42***	0.71*	−0.26***	0.16
House Democrat	(0.03)	(0.29)	(0.02)	(0.19)
Election	0.16***	−2.40***	0.05*	−2.63***
Election	(0.03)	(0.34)	(0.02)	(0.22)
Dem-election	0.21***	1.69***	0.25***	1.86***
Dem-election	(0.04)	(0.39)	(0.02)	(0.26)
Agency partisanship identity	−0.16***	−0.14***	−0.18***	−0.12***
Agency partisanship identity	(0.02)	(0.02)	(0.01)	(0.01)
Sole-sourced contract	0.04*	0.05*	−0.03**	−0.08***
Sole-sourced contract	(0.02)	(0.02)	(0.01)	(0.01)
PART rating	—	−0.01***	—	0.00
PART rating	—	(0.00)	—	(0.00)
Part-Election	—	0.04***	—	0.04***
Part-Election	—	(0.00)	—	(0.00)
Part-Democrat	—	−0.02***	—	−0.01**
Part-Democrat	—	(0.00)	—	(0.00)
Part-Election-Democrat	—	−0.02***	—	−0.02***
Part-Election-Democrat	—	(0.01)	—	(0.00)
Is minority/woman/native American owned	0.27***	0.23***	0.36***	0.34***
Is minority/woman/native American owned	(0.02)	(0.02)	(0.01)	(0.01)
Is nonprofit organization	0.26***	0.44***	0.18***	0.19***
Is nonprofit organization	(0.03)	(0.04)	(0.02)	(0.03)
Is emerging small business	−0.08	0.11	0.01	0.15***
Is emerging small business	(0.06)	(0.07)	(0.03)	(0.04)
Percentage White	−0.49***	−0.53***	−0.30***	−0.36***
Percentage White	(0.01)	(0.02)	(0.01)	(0.01)
Land area	−0.09***	−0.15***	−0.06***	−0.12***
Land area	(0.01)	(0.03)	(0.01)	(0.01)
Population density	−0.08***	−0.10***	−0.06***	−0.06***
Population density	(0.01)	(0.01)	(0.01)	(0.01)
School attainment	−0.25***	−0.28***	−0.23***	−0.26***
School attainment	(0.02)	(0.02)	(0.01)	(0.01)
Employment percentage	0.23***	0.29***	0.21***	0.23***
Employment percentage	(0.02)	(0.02)	(0.01)	(0.01)
Median income	0.22***	0.20***	0.16***	0.20***
Median income	(0.01)	(0.01)	(0.01)	(0.01)
Pseudo R²	.05	.06	.03	.03
N	104,936	83,901	104,936	83,901

Note. Regression Results 1 and 2 listed in this table are based on Model 1 and Model 2, respectively. Regression is conducted with regional fixed effects (Northeast, South, Midwest, and West). Inconsistency in the number of observations between regression results from Model 1 and 2 is caused by the lack of PART rating of some of the agencies administering the contract. Standard errors are in parentheses. PART = Program Assessment Rating Tool.

p < .05. **p < .01. ***p < .001.

In testing Model 1, our key coefficients—the district partisanship indicator, the postelection indicator, and the difference-in-difference estimator—are all statistically significant. We find that before the election, when controlling for other contract and district variables, the log odds of Democratic districts receiving larger contracts were 0.42 smaller than that for Republican districts. After the election, both Republican and Democratic districts were more likely to receive larger contracts. For Republican districts, the log odds increased by 0.21, but Democratic districts saw an even larger rise, a 0.37 (0.21 + 0.16) increase in the log odds of receiving larger contracts. These results show that before the election, when the Republican Party was in control of Congress, larger contracts were distributed to the Republican districts, whereas after the election when the Democratic Party took over, larger contracts were more likely to be allocated to the Democratic districts. This result is consistent with the total funding levels to Democratic and Republican districts; after the election, contracts became larger and total spending also increased, but these changes favored Democratic districts more than the Republican districts.

We also see that the use of sole-sourced contracts is positively associated with the size of the contracts—for larger contracts, the likelihood of agencies using sole-source contracts rather than competitive ones increased. This result might indicate an oligopsony for the larger contracts; agencies are more likely to directly offer contracts to the same highly specialized contractors who have built up expertise over time. However, it may be that agencies use sole-sourced contracts as a tool for funneling larger contracts to Democratic districts. We test this possibility by interacting the sole-source variable with the district partisanship indicator, as well as interacting the postelection indicator and the district partisanship indicator (not shown). In each case, the coefficients are not statistically significant, suggesting that the allocation of sole-source contract does not differ between Democratic and Republican districts either before or after the election. In other words, agencies are not manipulating the sole-sourced contracts to funnel larger contracts to specific districts or entities.

Thus, the results of Model 1 are consistent with Hypothesis 1: Agencies responded to the election by generally allocating larger contracts to the Democratic districts after the Democratic Party took power.

Model 2 further examines how agencies with differing levels of potential vulnerability reacted to the election, and the results are presented in the columns 2 and 4 of Table 4. Because the PART measure invariably captures political dynamics, as it may be correlated with the agency ideology or partisan identity (Dull, 2006; Lavertu et al., 2013), we control for the differences across agencies based on partisanship. The partisanship index is negative and significant, signifying that as an agency is more favored by the Democratic Party—receiving higher funding levels when Democrats control Congress, rather than Republicans—the contract size decreases.⁷ The partisanship of the agencies does not determine the agency response to the vulnerability created by low PART scores.

Consistent with Hypothesis 2, the triple difference ( $β_{D D D} D e m * E l e c t i o n * P A R T_{i, t})$ is negative and statistically significant, indicating that agencies that were more “at-risk” were more responsive to the election. The regression result based on Model 2 suggests that before the election, among republican districts, a higher PART rating of agency was associated with 0.01 log odds of smaller contracts. They were less likely to allocate larger contracts to Democratic districts than Republican districts (with a decrease in log odds of 0.02). After the election, agencies with higher PART rating were more likely to allocate larger contracts (the interaction term between PART rating and the postelection indicator is estimated to be 0.04), but are comparably less likely to allocate larger contracts to Democratic districts (a statistically significant triple difference estimator of −0.02). The results for agencies with lower PART ratings are reversed, that they are more likely to allocate larger contract to Democratic districts after the election. These results show that more vulnerable agencies (with lower PART ratings) allocated larger contracts to Democratic districts than to Republican districts in response to the electoral change. The large difference between Model 2 and Model 1 in the magnitude of the postelection indicator and district partisanship indicator variables and their interaction terms is because the estimates have different meanings in the two models. In Model 1, these coefficients give estimates on variables for all agencies in the sample, whereas in Model 2, they give estimates on variables for agencies with a PART rating of 0. The large differences in these estimates for the two models further demonstrate the heterogeneity of responses of agencies depending on level of vulnerability.

The estimates of Models 1 and 2 show that after the Democratic Party came into power, agencies began diverting larger contracts to the Democratic districts and that more vulnerable agencies were more active in doing so. However, such results based on ordinal logit regression may underestimate the possible bias caused by the confounding factors of different characteristics of congressional districts. Last but not least, the ordinal logit model uses a categorical contract size variable as the dependent variable; the interpretation of the result is less intuitive and cannot give information on exactly how many contracts of each size were allocated before and after the election. To address these issues, we use a propensity score matching technique in the difference-in-difference analytical framework as an additional support to Model 1.

To employ this technique, we first use a logistic model to calculate the propensity of a district being “Democratic” using the district socioeconomic and demographic control variables in Model 1. We then match each Democratic district to a most similar Republican district according to their propensity scores. The matched districts are very similar socioeconomically and demographically, but differ in their partisanship. We then calculate the changes in the number of each size of contract received before and after the election and compare these changes in Democratic districts to the matched Republican districts.

Table 5 presents selected socioeconomic characteristics of unmatched and matched Democratic and Republican districts. Before matching, these characteristics vary significantly between Democratic and Republican districts—five out of six important estimators show statistically significant differences. After matching, none of these characteristics differ between Democratic and Republican districts.

Table 5.

Characteristics for Unmatched and Matched Districts.

	Unmatched			Matched
	Democratic district	Republican district	p value for t test	Democratic district	Matched control group	p value for t test
Percentage White	65.85	74.82		67.44	70.19	.41
Land area	1.21 E+10	1.83 E+10	.02	1.50 E+10	1.60 E+10	.69
population Density	1,289	437	.001	560	540	.98
School attainment	6,454	5,782	.002	6,051	5,623	.23
Employment rate	58.54	58.84	.786	56.88	56.41	.84
Median income	53,737	50,064	.025	52,256	52,889	.82

Note. Differences in the values of Democratic districts between unmatched and matched groups are due to the different sample of Democratic districts included, as Democratic districts with inadequate common support are excluded for the matched group.

Figure 1 shows the common support of propensity score matching. Looking at the number of observations of Democratic and matched Republican districts at each level of propensity score, it is clear that the number of observations is severely asymmetrical for the two groups when propensity scores drop below 0.3 or rise above 0.7, where not enough common support is found. Therefore, we only include in our analysis observations with propensity scores within the 0.3 to 0.7 range.

Figure 1.

Common support of propensity score.

Table 6 shows the results of the comparison based on propensity score matching by nearest neighbor, nearest five neighbors, 0.1 mile radius, and 0.05 mile radius. Despite the different methods used, the results are very robust. We find that after the election, Democratic districts received approximately 21.5 fewer small contracts and 1.2 more large contracts. Both of these estimates are statistically significant. The number of moderate and very large contracts, however, did not differ before and after the election, with most of the various estimates being insignificant.

Table 6.

Differences in the Changed Number of Each Type of Contract by Matched Districts.

DID estimator: Difference between Democratic and republican districts in the change in the number of contracts received before and after the election.	Matched by nearest neighbor	Matched by nearest five neighbors	Matched by radius = 0.1 mile	Matched by radius = 0.05 mile
Small contracts	−21.11*	−21.40*	−22.78*	−21.11*
Small contracts	(9.37)	(9.01)	(9.58)	(9.38)
Moderate contracts	1.82	2.90	1.91	1.82
Moderate contracts	(2.76)	(2.40)	(2.83)	(2.76)
Large contracts	1.21**	1.36*	1.17*	1.21***
Large contracts	(0.54)	(0.57)	(0.55)	(0.54)
Very large contracts	0.14	0.18*	0.13	0.14
Very large contracts	(0.09)	(0.08)	(0.09)	(0.09)

Note. DID = difference-in-difference.

p < .05. **p < .01. ***p < .001.

These findings are consistent with the regression result of Model 1, providing direct evidence that more large contracts and fewer small contracts were channeled to Democratic districts than to Republican districts. It also shows an interesting result on very large contracts: except for matching by nearest five neighbors, the results show that the number of very large contracts allocated to the Democratic districts did not change significantly after the election. This result might stem from diminishing returns as contract size increases. As the size of the contract grows, so too does the attention it receives. The appearance of unfairness in allocating contracts might be met with unfavorable attention and lawsuits. The value of very large contracts might be great enough to encourage losing vendors to react, perhaps with litigation or bad publicity, thereby potentially undermining an agency’s prospects for cultivating favor. Moreover, for very large contracts, the pool of potential vendors shrinks; for very large contracts, the average number of bids is 2.18 compared with 4.48 for large contracts. This statistically significant difference indicates that there is less opportunity for agencies to reward contracts to different firms. The allocations of very large contracts and large contracts are therefore different, and the incentive for agencies using contracts to curry favor will accordingly decrease when contracts grow from large to very large.

Following this line of logic, the relationship between contract size and the likelihood of allocating the contract to a Democratic district after the election should resemble an inverse parabolic curve as the contract size grows from moderate to large and to very large. At first, the likelihood of contracts being diverted to Democratic districts increases as contract size grows. But with increased contract size also comes heightened public attention, including potential criticism, which introduces a negative effect on the likelihood of allocating the contract to Democratic districts. At some point, the negative effect will surpass the positive impacts and the relationship will tip; an increase in the size of contracts will no longer motivate agency officials to include politically strategic calculations in their allocation of contacts.

Table 7 provides support for this relationship, showing results from a polynomial regression of the obligated contract amounts on whether or not the contract is channeled to Democratic districts. We find the interaction terms between the postelection indicator and both the size of the contract and the square of the amount of the contract to be statistically significant, where the former is positive and the latter is negative. This gives us an inverse parabolic curve with a tipping point of approximately US$14.5 million. In other words, agencies have the incentive to divert larger contracts to Democratic districts to curry congressional favor up until the contracts reach US$14.5 million, after which they will be more cautious in allocating the contracts such that further increases in the size of the contracts will not further motivate the members to favor Democratic districts.

Table 7.

Regression Result With Polynomials.

Coefficient	Estimate
Obligated amount	3.87 E–9
Obligated amount	(3.85 E–8)
Obligated amount²	9.27 E–16
Obligated amount²	(1.74 E–15)
Election	0.36***
Election	(0.06)
Amount × Election	1.06 E–7*
Amount × Election	(5.31 E–8)
Amount² × Election	−3.97 E–15*
Amount² × Election	(2.10 E–15)
N	14,166
Pseudo R²	.36

Note. Regression is conducted with contract- and district-level control variables as well as regional fixed effects (Northeast, South, Midwest, and West). Standard errors are in parentheses. Dependent variable is whether or not the contract was awarded to a Democratic district. Only moderate, large, and very large contracts are included in this regression.

p < .05. **p < .01. ***p < .001.

As a robustness check and to expand the external validity of our results, we evaluate the impacts of the 2010 congressional election using the same analytical design. In the 2010 election, control of the House switched to the Republican Party from the Democratic Party. This analysis is shown in Supplemental Appendix 3.

Conclusion

To the extent that congressional representatives can enhance their prospects for reelection by getting credit for bringing home the bacon, they will seek particularized benefits (“pork”) that they can point to when speaking with voters and/or donors about their efforts. Federal agencies can curry favor with representatives by providing them with these opportunities. Prior studies have found evidence that agencies use their control over program operations to ingratiate themselves with congressional decision makers through the distribution of federal funds in a number of ways: through grants and contracts (Arnold, 1979; Berry & Gersen, 2010; Krause, 1999; Mayer, 1995; Stein & Bickers, 1995), through loans and loan guarantees (Bickers & Stein, 1996), and through the determinants of natural disaster declarations and “emergency” spending in response to them (Garrett & Sobel, 2003). Other research shows that the probability of IRS audits is lower for residents whose congressional members oversee the IRS (Young, Reksulak, & Shughart, 2001). In this article, we add to this literature by focusing on the strategic distribution of contracts allocated to congressional districts.

We break from previous research by looking not at overall spending per congressional district but instead at the size of contracts awarded in a district. We reason that only contracts large enough to significantly affect a firm’s finances or make a difference in a district offer the potential for credit claiming by legislators. As such, only with the strategic allocation of sufficiently large contracts can federal agencies curry favor with legislators. We examine the 2006 congressional election to see whether agencies allocated larger contracts to Democratic districts after the election, given the change in partisan control that followed the election.

Our results show that federal agencies responded to the election by allocating larger contracts to Democratic districts after the Democrats took control of Congress. In addition, we find that the change in the allocation of contracts from small to very large is not linear: There are diminishing returns to very large contracts likely due to unwanted public attention, a higher chance of litigation from losing bidders, and a smaller pool of vendors.

Finally, we examine whether agencies with more “vulnerable” programs were more politically strategic in their allocation of contracts. In the years prior to the 2006 election, the Bush administration introduced the PART, which evaluated program design and management. We found that the agencies with more “vulnerable” programs were more responsive to the election. Before the election, those with lower PART ratings were more likely to use larger contracts, and after the election, they allocated smaller contracts to Republican districts but even larger contracts to Democratic districts.

Supplemental Material

Supplementary_Material – Supplemental material for Contracts, Agency Vulnerability, and the Allocation of Federal Funds

Supplemental material, Supplementary_Material for Contracts, Agency Vulnerability, and the Allocation of Federal Funds by Stuart Kasdin and Luona Lin in The American Review of Public Administration

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

ORCID iD

Stuart Kasdin

Supplemental Material

Supplemental material for this article is available online.

Author Biographies

Stuart Kasdin is Mayor Pro Tempore for the city of Goleta, CA and an adjunct professor of political science at Santa Barbara City College. Previously, he was assistant professor at the Trachtenberg School of Public Policy and Public Administration at George Washington University.

Luona Lin is a research associate at American Psychological Association. She has her Masters in Public Policy with concentration in program evaluation from the Trachtenberg School of Public Policy and Public Administration from George Washington University.

References

Arnold

(1979). Congress and the bureaucracy: A theory of influence. New Haven, CT: Yale University Press.

Arnold

(1990). The logic of congressional action. New Haven, CT: Yale University Press.

Banks

Weingast

(1992). The political control of bureaucracies under asymmetric information. American Journal of Political Science, 36, 509-524.

Berry

C. R.

Burden

B. C.

Howell

W. G.

(2010). The president and the distribution of federal spending. American Political Science Review, 104, 783-799.

Berry

C. R.

Gersen

J. E.

(2010, May 18). Agency spending and political control of the bureaucracy. Paper presented at the Political Economy Seminar, Stanford University, CA.

Bertelli

A. M.

Grose

C. R.

(2009). Secretaries of pork? A new theory of distributive public policy. The Journal of Politics, 71, 926-945.

Bickers

K. N.

Evans

Stein

R. M.

Wrinkle

R. D.

(2007, August 30-September 2). The electoral effect of credit claiming for pork barrel projects in Congress. Paper presented at the Annual Meeting of the American Political Science Association, Chicago, IL.

Bickers

K. N.

Stein

R. M.

(1996). The electoral dynamics of the federal pork barrel. American Journal of Political Science, 40, 1300-1326.

Boycko

Shleifer

Vishny

(1996). A theory of privatization. The Economic Journal, 106, 309-319.

10.

Brown

T. L.

Potoski

(2003a). Managing contract performance: A transaction costs approach. Journal of Policy Analysis and Management, 22, 275-297.

11.

Brown

T. L.

Potoski

(2003b). Transaction costs and institutional explanations for government service production decisions. Journal of Public Administration Research and Theory, 13, 441-468.

12.

Brown

T. L.

Potoski

Van Slyke

D. M.

(2006). Managing public service contracts: Aligning values, institutions, and markets. Public Administration Review, 66, 323-331.

13.

Carey

J. M.

Shugart

M. S.

(1994). Incentives to cultivate a personal vote: A rank ordering of electoral formulas. Electoral Studies, 14, 417-439.

14.

Carsey

T. M.

Rundquist

(1999). Party and committee in distributive politics: Evidence from defense spending source. The Journal of Politics, 61, 1156-1169.

15.

Clinton

J. D.

Lewis

D. E

. (2008). Expert opinion, agency characteristics, and agency preferences. Political Analysis, 16, 3-20.

16.

DeHoog

R. H.

(1990). Competition, negotiation, or cooperation: Three models for service contracting. Administration & Society, 22, 317-340.

17.

Dull

(2006). Why PART? The institutional politics of presidential budget reform. Journal of Public Administration Research and Theory, 16, 187-215.

18.

Fenno

(1973). Congressmen in committees. Boston, MA: Little, Brown.

19.

Frederickson

D. G.

Frederickson

H. G.

(2007). Measuring the performance of the hollow state. Washington, DC: Georgetown University Press.

20.

Garrett

T. A.

Sobel

R. S.

(2003). The political economy of FEMA disaster payments. Economic Inquiry, 41, 496-509.

21.

Gilmour

J. B.

(2007). Implementing OMB’s Program Assessment Rating Tool (PART): Meeting the challenges of integrating budget and performance. OECD Journal on Budgeting, 7, 1-40.

22.

Gilmour

J. B.

Lewis

D. E.

(2006a). Assessing performance budgeting at OMB: The influence of politics, performance, and program size. Journal of Public Administration Research and Theory, 16, 169-186.

23.

Gilmour

J. B.

Lewis

D. E.

(2006b). Does performance budgeting work? An examination of the Office of Management and Budget’s PART scores. Public Administration Review, 66, 742-752.

24.

Globerman

Vining

A. R.

(1996). A framework for evaluating the government contracting-out decision with an application to information technology. Public Administration Review, 56, 577-584.

25.

Gordon

S. C.

(2011). Politicizing agency spending authority: Lessons from a Bush-era scandal. American Political Science Review, 105, 717-734.

26.

Heinrich

C. J

. (2012). How credible is the evidence, and does it matter? An analysis of the program assessment rating tool. Public Administration Review, 72, 123-134.

27.

Hudak

(2014). Presidential pork: White House influence over the distribution of federal grants. Washington, DC: Brookings Institution Press.

28.

Joyce

P. G.

(2011). The Obama administration and PBB: Building on the legacy of federal performance-informed budgeting? Public Administration Review, 71, 356-367.

29.

Kasdin

S. R.

(2010). Reinventing reforms: How to improve program management using performance measures really. Public Budgeting & Finance, 30(3), 51-78.

30.

Kasdin

S. R.

Lin

(2015). Strategic behavior by federal agencies in the allocation of public resources. Public Choice, 164, 309-329.

31.

Katz

R. S.

(1986). Intraparty preference voting. In Grofman

Lijphart

(Eds.), Electoral laws and their political consequences (pp. 85-103). New York, NY: Agathon Press.

32.

Kettl

(1993). Sharing power: Public governance and private markets. Washington, DC: Brookings Institution Press.

33.

Krause

G. A.

(1999). A two-way street: The institutional dynamics of the modern administrative state. Pittsburgh: University of Pittsburgh Press.

34.

Lavertu

Lewis

D. E.

Moynihan

(2013). Government reform, political ideology, and administrative burden: The case of performance management in the Bush administration. Public Administration Review, 73, 845-857.

35.

Levin

Tadelis

(2010). Contracting for government services: Theory and evidence from U.S. cities. The Journal of Industrial Economics, LVIII, 507-541.

36.

Lipinski

(2011, October 18). Lipinski announces “buy American” victory for local manufacturer and workers. Congressman Dan Lipinski. Retrieved from https://lipinski.house.gov/manufacturing/lipinski-announces-buy-american-victory-for-local-manufacturer-and-workers-october-18-20111/

37.

Lowry

R. C.

Potoski

(2004). Organized interests and the politics of federal discretionary grants. The Journal of Politics, 66, 513-533.

38.

(2013). How political are government contracting decisions? An examination of human service contracting determinants. Public Administration Quarterly, 37, 183-209.

39.

Mayer

K. R.

(1995). Electoral cycles in federal government Prime contract awards: State-level evidence from the 1988 and 1992 presidential elections. American Journal of Political Science, 39, 162-185.

40.

Mayhew

D. R.

(1974). Congress: The electoral connection. New Haven, CT: Yale University Press.

41.

McCarty

N. M.

(2000). Presidential pork: Executive veto power and distributive politics. American Political Science Review, 94, 117-128.

42.

Moe

T. M.

(1984). The new economics of organization. American Journal of Political Science, 28, 739-777.

43.

Mullen

P. R.

(2006). Performance-based budgeting: The contribution of the program assessment rating tool. Public Budgeting & Finance, 26(4), 79-88.

44.

Niskanen

W. A.

Jr. (1994). Bureaucracy and public economics. Brookfield, VT: Edward Elgar.

45.

OMB (2010, July 7). Cutting waste and saving money through contracting reform. Office of Management and Budget. Retrieved from https://www.whitehouse.gov/sites/whitehouse.gov/files/omb/assets/blog/Update_on_Contracting_Reforms.pdf

46.

Radin

B. A.

(2006). Challenging the performance movement: Accountability, complexity, and democratic values. Washington, DC: Georgetown University Press.

47.

Rundquist

B. S.

Carsey

T. M.

(2002). Congress and defense spending: The distributive politics of military procurement. Norman: University of Oklahoma Press.

48.

Savas

E. S.

(2000). Privatization and public-private partnerships. New York, NY: Seven Bridges Press.

49.

Savas

E. S.

(2002). Competition and choice in New York City social services. Public Administration Review, 62, 82-91.

50.

Shepsle

K. A.

Weingast

B. R.

(1984). Legislative politics and budget outcomes. In Mills

G. B.

Palmer

J. L.

(Eds.), Federal budget policy in the 1980s (pp. 343-357). Washington, DC: Urban Institute.

51.

Stein

R. M.

Bickers

K. N.

(1995). Perpetuating the pork barrel: Policy subsystems and American democracy. New York, NY: Cambridge University Press.

52.

Williamson

O. E.

(1981). The economics of organization: The transaction cost approach. The American Journal of Sociology, 87, 548-577.

53.

Williamson

O. E.

(1985). The economic institutions of capitalism: Firms, markets, relational contracting. New York, NY: Free Press.

54.

Williamson

O. E.

(1999). Public and private bureaucracies: A transaction cost economics perspective. The Journal of Law, Economics, & Organization, 15, 306-342.

55.

Wilson

J. Q.

(1973). Political organizations. New York, NY: Basic Books.

56.

Young

Reksulak

Shughart

W. F.

II . (2001). The political economy of the IRS. Economics & Politics, 13, 201-220.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.52 MB