Network Ties and Congressional Delegation to U.S. Federal Agencies

Abstract

Many public administration studies have argued that agency network activity brings about important policy benefits such as informational advantages and stakeholder support. Given the proposition, agency network activity would positively affect congressional delegation because policymaking authority tends to be delegated to the agencies that can achieve appropriate policy outcomes. This hypothesis is examined by several regression tests using data from U.S. federal agencies. Statistical results indicate that the agencies with significant network ties are likely to yield more appropriate public policies and have greater statutory discretion.

Keywords

network ties policy appropriateness discretion congressional delegation

Introduction

Public administration studies have long examined the factors and conditions that determine agency policymaking authorities (Wirgau, 2013). Regarding the issue, the studies have generally agreed that agencies can attain high discretion only if they are capable of selecting proper policy alternatives with knowledge about policy environments and the consequences of different administrative actions (Epstein & O’Halloran, 1999; Gailmard & Patty, 2007; Wilson, 1989) or if they have sufficient political support in their policy communities (Baekkeskov, 2017; Carpenter, 2001, 2010; Rourke, 1984). In other words, bureaucratic policymaking authority would be dependent on the agency’s capability to achieve appropriate policy outcomes without administrative problems (MacDonald & Franko, 2007; Rourke, 1984; Wirgau, 2013).

Regarding this issue, traditional public administration studies have emphasized internal factors of agencies, in particular, the professionalism of agency staff. The studies have generally agreed that agency professionalism that facilitates the improvement of agency performance also increases policymaking authority (Eisner & Meier, 1990; Rourke, 1984). Recently, however, external interactions with state and non-state stakeholders have been highlighted as an important factor affecting administrative performance (Herian, Hamm, Tomkins, & Zillig, 2012; Neshkova & Guo, 2012). A plethora of studies have emphasized that agency network ties not only allow sharing large amounts of information (Gargiulo & Benassi, 2000; Hansen, 1999) but also promote social support from network members (Wellman & Wortley, 1990; Wirgau, 2013). In particular, the actors in central network positions tend to have significant informational benefits and stakeholder support (Putnam, 2000). Therefore, if an agency has more diverse and stronger network ties with stakeholders, the agency could obtain more administrative information and stakeholder support, thereby producing appropriate policy outcomes. Consequently, agencies with significant network ties would acquire greater policymaking authority.

It is not novel to argue that external factors can affect agency policymaking authority. For example, Daniel Carpenter (2001, 2010) claimed that the agencies with significant reputations within their policy communities can operate autonomously by drawing stakeholder support and reducing congressional intervention in agency actions. Similarly, Gregory A. Huber (2007) argued that agencies that behave in a neutral manner can limit the complaints of interest groups and maintain their discretion. These studies have generally maintained that favorable relationships between agencies and stakeholders enhance bureaucratic discretion. Nevertheless, the relationship between agency network ties and bureaucratic policymaking authority has not yet been fully studied, especially in a statistical manner. Although Carpenter (2001, 2010) analyzed agency reputation in policy networks regarding bureaucratic autonomy with several case studies, he was more focused on agency reputation than network ties. Similarly, G. A. Huber (2007) emphasized consistent law enforcement rather than network ties with the case study of the Occupational Safety and Health Administration.

To address the lacuna, this article examined the effect of agency network ties on policymaking authority quantitatively with panel data from U.S. federal agencies. The main hypothesis is as follows:

Hypothesis 1: Agency network ties enhance policy appropriateness and consequently, have positive effects on congressional delegation.

It implies that an agency’s network ties promote its informational and political capabilities in administration, thereby positively affecting statutory bureaucratic policymaking authority. In this article, policy appropriateness denotes legally and technically suitable selection of policy alternatives and politically appropriate policymaking that promotes stakeholders’ support. Several regression tests indicate that Congress would provide more statutory discretion to agencies if degree centrality (i.e., the number of network ties that an actor has) and tie strength (i.e., the potency of the bond between network members) of the agencies are sufficiently high in policy networks.

Networking and Policymaking Authority in Public Administration

There has been growing interest in interaction among government agencies and non-state stakeholders in public administration studies in recent decades. Issue network and advocacy coalition framework (ACF) studies have claimed that policy networks in the United States have been characterized by multiple policy stakeholders and the presence of conflict (Heclo, 1978; McFarland, 1987). Moreover, the policy stakeholders tend to make advocacy coalitions with those who share policy beliefs, and try to translate their beliefs into public policy (Sabatier & Jenkins-Smith, 1993). Given the nature of policy subsystems, advocacy coalitions frequently fail to reach satisfactory compromises by themselves. Rather, policy brokers who suggest reasonable solutions and promote compromises play a crucial role in mediating conflicts between coalitions (Ingold & Varone, 2012). In this vein, administrative agencies, which have significant monetary and human resources to interact with policy stakeholders directly (Rourke, 1984; Wilson, 1989) have taken meaningful positions in mediating conflicting coalitions (Jenkins-Smith, Clair, & Woods, 1991). If agencies have significant network ties in subsystems, they could be more likely to resolve coalitional conflicts and make more appropriate policies in a timely manner.

Moreover, diverse studies have emphasized that when agencies enhance their network ties with stakeholders, they can accumulate “social capital” that reduces potential policy problems. In other words, interactions with stakeholders strengthen the appropriateness of agency decisions (Desario & Langton, 1987; Herian et al., 2012; Neshkova & Guo, 2012; Wang & Van Wart, 2007). Agencies can identify legally and technically proper policy alternatives by increasing learning opportunities for bureaucrats to produce more informed policy decisions (Kennedy & Carpenter, 1988; Nicholson-Crotty & O’Toole, 2004). For example, Neshkova and Guo (2012) argued that the U.S. state transportation agencies’ network ties provide administrators with site-specific information and contributes to improved administrative performance. Nicholson-Crotty and O’Toole (2004) showed that U.S. municipal police departments’ network ties enable the agencies to better understand policy environments and enhance their administrative performances (e.g., arrest rates).

In addition, discourses in administration increase stakeholder support in policy networks (Desario & Langton, 1987; Herian et al., 2012; Wang & Van Wart, 2007). Broad and frequent communications promote intergroup trust and reduce transaction costs in policy communities (Doney & Cannon, 1997; Nicholson, Compeau, & Sethi, 2001) while eliminating misunderstandings between unconnected groups (Burt, 2004). Consequently, when agencies arrange frequent contacts in central network positions, the agencies can produce appropriate policies that promote stakeholder support. For example, survey results from Langbein and Kerwin (2000) and Lubell (2000) suggested that network ties of the Environmental Protection Agency (EPA) such as diverse collaborative programs (e.g., National Estuary Program) and negotiated rulemaking have produced greater attitudinal support from environmental stakeholders relative to conventional rulemakings and non-participatory programs. Innes, Connick, Kaplan, and Booher (2006) also showed that California’s collaborative water management program called CALFED effectively improved policymaking gridlock and litigation by drawing favorable responses from stakeholders. In sum, network ties with stakeholders increase policy appropriateness, allowing agencies to choose appropriate policy alternatives and promote stakeholder satisfaction by making public policies properly (e.g., Susskind & Cruikshank, 1987).

Given the proposition that agency network ties enhance policy appropriateness, those activities might also positively affect agency statutory discretion. Because legislators are unlikely to have sufficient professional knowledge about policy issues, they prefer to delegate information acquisition to expert agencies (Bawn, 1995; Gailmard & Patty, 2007). Thus, the information flowing through network ties is an important factor for policy authority delegation. Admittedly, agencies have their own expertise derived from human capital such as professional staff, yet it is hard for them to fully understand policy problems and environments without networking with stakeholders (Gray, 1989; O’Toole, 1997; Van Waarden, 1992). Therefore, without adequate information sharing with stakeholders, agencies cannot maintain sufficient expertise to select appropriate policy alternatives and consequently, may not attain significant policy discretion.

Stakeholder support developed by networking with stakeholders allows agencies to make clear organizational goals (Stazyk & Goerdel, 2011) and to implement public policies in an efficient manner (Ansell, 2011; Putnam, 2000), thereby promoting policy appropriateness and congressional delegation. Furthermore, as Rourke (1984) noted, “strength in constituency is no less an asset for an administrator than it is for a politician” (p. 48). Elected officials cannot ignore public responses to policymaking and implementation. Thus, if agencies receive high political support from stakeholders or advocacy coalitions, it would be costly for legislators to deny the policy authorities of those agencies. In other words, stakeholder support allows agencies to avoid hostile responses from legislators (Carpenter, 2001, 2010). These administrative benefits from networking would contribute to increased congressional delegation.

From the theoretical arguments above, it can be hypothesized that broader and stronger agency network ties increase policy appropriateness in terms of selecting proper policy alternatives and receiving political support from stakeholders, thereby encouraging legislators to apply more discretion. This hypothesis will be examined in the next section using several regression tests. To make samples uniform, data on the largest domestic and regulatory U.S. federal agencies over a 10-year period were used (see the appendix).¹

Statistical Tests

Measuring Main Variables

Agency network ties

In measuring network ties, centrality is a core concept. As a result of many network studies, it has been generally accepted that if an actor is in a central position (and has broader network ties), the actor can enjoy significant information flow and trust (Burt, 2004; Huang & Provan, 2007). Given the definition of centrality as “a function of the centrality of those to whom one is connected through direct and indirect links” (Ibarra, 1992, 432), many network studies have used the number of network ties that an actor holds (i.e., degree centrality; Akkerman, Torenvlied, & Schalk, 2011; Provan & Huang, 2012). However, not only degree centrality, but also tie strength significantly affects information flow and political support in policy networks (Isett & Provan, 2005; Lee & Kim, 2011). In other words, both the number of network branches and the quality of the branches are important. If agencies have limited contact with stakeholders, they would not have access to sufficient administrative information, even with many network branches (Lee & Kim, 2011; Wellman & Wortley, 1990). Moreover, limited contact can increase transaction costs among network actors (Doney & Cannon, 1997; Nicholson et al., 2001). Because of this reason, several studies have used degree centrality (i.e., the number [or proportion] of participants) weighted by tie strength as a measure of network ties (e.g., Koch, Galaskiewicz, & Pierson, 2015; Srivastava, 2015; Sykes, Venkatesh, & Gosain, 2009). Following this tradition, this study also used the multiplication of degree centrality and tie strength for network ties. In addition, this measure was divided by the number of agency agendas to limit possible biases from agency size:

network ties = \frac{degree centrality \times tie strength}{number of agency agendas} .

To measure degree centrality, the numbers of advisory committee participants were used. Advisory committees contribute to promoting network connections through transparent meetings (Bingham, 2010). In advisory committees, agencies learn policy issues, adjust stakeholder opinions, develop policy alternatives, and reach a consensus solution through deliberative processes (Applegate, 1998; Balla, 2004). As a result, agencies can acquire significant stakeholders support through advisory committee meetings (Moffitt, 2010). The data from advisory committees are somewhat superior to other alternative measures such as negotiated rulemakings in several aspects.²

Because advisory committees are one of the most common networking methods for U.S. federal agencies (U.S. federal agencies operated 1,077 advisory committees in 2012) and the Federal Advisory Committee Act (FACA, P.L. 92-463) provides broad guidelines for committee management, the numbers of advisory committee participants are relatively consistent across agencies, and the qualities of individual advisory committees are unlikely to be biased. In addition, because advisory committees are not comprised of ordinary citizens, but of major stakeholders, minor network nodes can be excluded. To further reduce possible biases from advisory committee characteristics, only advisory committees that were active and held at least one meeting in a year’s time were considered, because inactive advisory committees—committees waiting for legislation of committee abolishment or funding—cannot contribute to linking network participants. In contrast, regardless of establishment authority and committee functions, advisory committees can generally contribute to promoting communications among stakeholders.³ For example, the food and drug administration (FDA) tends to establish diverse scientific program advisory boards rather than regulatory negotiation committees to promote stakeholder meetings, because the FDA’s main issues are technically complex, and its stakeholders tend to have significant information about drugs and chemical substances. Thus, excluding certain committees due to establishment authority and committee functions can cause serious biases in measuring network ties.⁴ The advisory committee participant data were from the data set of the federal advisory committees database (FACD; http://www.fido.gov). In addition, the “Federal Register” was consulted for the data of some agencies whose advisory committees were not clearly indicated in the FACD data set.

Regarding tie strength, many administrative studies have measured the concept based on the frequency of interactions (Huang, 2014; Lee & Kim, 2011). Following this tradition, this article also measured tie strength by counting the number of agency meetings with policy stakeholders. However, major stakeholders tend to participate not only in advisory committees but also in diverse kinds of policy-related meetings. Thus, counting only the number of advisory committee meetings may underestimate tie strength. To correct this problem, tie strength was measured by the number of meetings with stakeholders held by individual agencies that are publicly announced in the “Federal Register.” In addition, the denominator—number of agendas—was designated as the number of agency regulatory proposals published in the “Federal Register.” To read statistical data conveniently, the network tie measure was divided by 1,000.⁵

Policy appropriateness

As argued above, policy appropriateness in this article was narrowly defined regarding proper policy alternatives and political support from stakeholders. If policy alternatives are inappropriate and stakeholders are unsatisfied with agency decisions, they would express dissatisfaction through official lawsuits or unofficial complaints (e.g., protests, oppositions, and disapproval remarks; Kelly, 2004). In this vein, this study used two proxies to measure policy appropriateness: the number of court challenges against agencies (in courts of appeals and the Supreme Court) and the ratio of the number of New York Times (NYT) articles about stakeholders’ complaints against specific agency decisions to the number of NYT articles that cited the agencies.⁶

The former indicates that agencies possibly fail to choose legally and technically proper policy alternatives in their administrative adjudications or rulemakings. If agencies can understand policy environments and choose relevant policy alternatives, they would be less likely to be judicially challenged. The latter implies the probability that agencies may be unable to make politically appropriate decisions—in other words, the probability that agencies could not respond to stakeholders appropriately— thereby losing political support from stakeholders. If agencies lack sufficient political support from stakeholders, they would face significant stakeholder complaints against agency decisions. Both measures connote that agencies may misunderstand or mismanage political demands and policy environments and would, therefore, be unable to produce appropriate policy outcomes.

Congressional delegation

Regarding congressional delegation, there is no perfectly agreed upon method of measuring the concept. However, several studies such as Epstein and O’Halloran (1999) and J. D. Huber and Shipan (2002) have suggested solid methods to measure congressional delegation. This study used both methods to measure congressional delegation to enhance statistical robustness. The measure suggested by Epstein and O’Halloran (1999; henceforth, “EO Measure”) is the most elaborate and is widely used by many other delegation studies (e.g., Ainsworth & Harward, 2009; Thomson, Torenvlied, & Arregui, 2007). The researchers set up detailed standards for executive delegation and categories of constrains in a law.⁷ Then, they determined the discretion level of a public law as follows:

EO Measure = \frac{provisions with executive delgation}{total provisions} \times (1 - \frac{categories of constraints in a law}{total categories of constraints}) .

Because the current article is about network ties of agencies, and the unit of analysis is an agency (not a public law), the number of discretionary laws based on the EO Measure is counted to measure congressional delegation. To ascertain which public laws affect specific agencies’ jurisdictions, this article used a chronological list of public laws provided by the Statutes at Large and the parallel table of authorities and rules provided by Code of Federal Regulations (CFR). The chronological list of public laws provides information about which U.S. Code sections are affected by specific public law provisions. Using the parallel table between U.S. Code and CFR, it can be inferred which agencies are related with specific public laws. Because a high value of EO Measure implies more delegation, only public laws whose EO Measure scores are higher than average are assumed to be discretionary laws. To count the number of discretionary laws, this study used Congressional Research Service (CRS) bill summaries. In addition, to filter out minor laws, only public laws with more than 10 provisions were included in the data set.⁸

Another frequently used measure of discretion is the length (i.e., number of words) of legislation, as suggested by J. D. Huber and Shipan (2002). They argued that longer statutes leave little room for agencies to behave discretionally. Their measure (henceforth, “HS Measure”) may be sufficiently valid while comparing discretionary authorities of similar agencies; however, it is necessary to consider agency characteristics while using the HS Measure for cross-sectional or panel studies, because the length of legislation is likely to be affected by agency characteristics. In this respect, Meier’s (1980) suggestion to measure agency autonomy is considerable: He recommended “the ratio of the number of pages of rules the bureau issues to the number of pages of substantive legislation that applies to the agency” (henceforth, “R/L Ratio”) as an indicator of agency legislative autonomy (p. 364). Following his suggestion, the ratio of the length of agency regulation to the length of legislation is used as a measure of agency statutory discretion.

To prevent biases in measuring the length of legislation, only provisions that clearly require agencies to perform administrative behaviors were counted to measure the length of legislation. Moreover, the length of regulation was measured by marginal increase in CFR pages. In calculating R/L Ratio, however, there are some mathematical problems: The denominators (i.e., length of legislation) are frequently zero, and the numerators (i.e., length of regulation) are sometimes negative. To eliminate the problems, the R/L Ratio was calculated as follows:

R/L Ratio = \frac{the length of agency regulation}{the length of legislation} = \frac{\max (marginal increase in CFR pages, 0)}{sum of restrictive provisions + 1} .

Results

Agency network ties and policy appropriateness

In examining the relationship between network ties and policy appropriateness, it is necessary to control for internal factors that affect policy appropriateness. In contrast to social capital gained from external relationships with stakeholders, human capital and financial capital are the main “internal” factors that affect policy appropriateness (Donahue, Selden, & Ingraham, 2000). Human capital is the most important resource for agencies with regard to selecting better policy alternatives and improving administrative performance (Brewer & Selden, 2000). Therefore, if agencies have sufficient number of professionals and experienced employees, they would be more likely to make and implement appropriate public policies. To control for internal professionalism and employee experience, the number of professional employees and the average year of service were included as control variables.⁹ Financial capital is also important for policy appropriateness. Because budget constraint determines the availability of policy alternatives, sufficient financial capital is necessary for agencies to select appropriate policy alternatives (Wilson, 1989). Thus, agency budget authority—the financial capital provided by law to a federal agency to obligate revenues—was included in regression models. The size of the budget authority is revised to be constant in 2008 dollars (in billions of dollars) to adjust for inflation.

In this article, policy appropriateness was measured by the frequency of stakeholders’ complaints and judicial challenges, which can be affected by the contentiousness of policy environments. For example, if agencies enact significant rules that include remarkable changes in the distribution of policy benefits, policy environments could become more contentious and, as a result, judicial challenges and stakeholder complaints would increase. To control for this factor, the number of significant rules was incorporated; this variable was counted using the Unified Agenda that classified significant rules based on Executive Order 12866.¹⁰ Moreover, if an agency implements substantial government programs, stakeholders tend to have more incentives to intervene in the agency’s decisions to acquire greater policy benefits (Campbell, 2003). Thus, agency budget authority and the number of total employees could be meaningful regarding the contentiousness of policy environments.¹¹ Furthermore, a conservative policy atmosphere would promote stakeholders’ complaints and judicial challenges against regulatory agencies. To control for this factor, this article used Stimson’s (1998) public liberalism index that denotes national mood in making public policies.

Because the numbers of court challenges are skewed, with a lot of the data in the vicinity of the logical lower bound (i.e., 0), zero-inflated negative binomial regressions were adopted in the count data to address the over-dispersion problem for Models A-1 and B.¹² To limit heteroskedasticity and autocorrelation problems, cluster robust standard errors were reported for the models. Because zero observations for courts of appeals cases are not significantly high, the relationship between network ties and courts of appeals cases were reexamined by panel negative binomial regressions with fixed effects to check the robustness of the regression results (Model A-2). In contrast, partly because there are a large number of zero observations for Supreme Court cases, panel negative binomial regressions are not appropriate for the dependent variable.¹³ For the ratio of stakeholders’ complaint articles (Model C), linear panel regressions with fixed effects were used to limit unobservable agency-specific factors. In addition, cluster robust standard errors were reported for the model to relieve heteroskedasticity and autocorrelation problems.

The regression models show that agencies with more network ties are less likely to face stakeholders’ complaints and judicial challenges. In particular, the effect of network ties on policy appropriateness is generally more significant relative to human and financial capitals; only the budget authority of Model A-2 and the average year of service of Model B were statistically significant. They generally imply that when agencies have broad and strong network ties, they are more likely to achieve appropriate policy outcomes in terms of policy alternative selection and policy response to stakeholders. These results were statistically robust, even with inclusion of several confounding variables such as Republican Congress (1 = Republican majorities in both houses and 0 = otherwise), honeymoon period (1 = the first year of a new presidency, 0 = otherwise), independent agency (1 = independent agency and 0 = otherwise), and year dummies.¹⁴ In addition, the network ties of specific agencies such as the Bureau of Land Management, Forest Service, and National Park Service are significantly high (see the appendix). Therefore, the regression tests of Table 1 (and also, those of Table 2) were reexamined with these agencies excluded. The statistical results remained consistent at least at the 10% significance level; in particular, the statistical significance of network ties was invariable.

Table 1.

The Effect of Agency Network Ties on Policy Outcome Quality.

Independent variables	Courts of appeals cases		Supreme court cases	Ratio of stakeholders’ complaint articles
Independent variables	Model A-1	Model A-2	Model B	Model C
Network ties	−.046**	−.011*	−.069**	−.001***
	(.017)	(.005)	(.026)	(.000)
Professionals	.176	.175**	.170	.005
	(.109)	(.067)	(.108)	(.028)
Average year of service	.010	.002	−.061^†	.004
Average year of service	(.022)	(.014)	(.033)	(.003)
Significant rules	−.002	−.008	−.007	−.000
	(.016)	(.005)	(.018)	(.001)
Public liberalism	−.030	.009	−.102*	.003
	(.021)	(.011)	(.045)	(.007)
Budget authority	.001	−.002^†	−.001	−.000
	(.001)	(.001)	(.001)	(.000)
Total employees	.038*	.016	.020^†	.004
	(.016)	(.011)	(.001)	(.005)
Constant	3.262*	1.532*	6.595*	−.204
	(1.524)	(.749)	(2.835)	(.427)
Inflate constant	−24.444***		−31.071***
	(2.237)		(.600)
Prob. > χ²	.0000	.0042	.0000
Prob. > F				.0068
Log pseudo likelihood	−933.89		−349.126
Log likelihood		−582.550
Nonzero n	227	227	99
n	270	270	270	270

Note. Zero-inflated negative binomial regressions and cluster robust standard errors in parentheses for Models A-1 and B (inflation model: logit), panel negative binomial regressions with fixed effects for Models A-2, and linear panel regressions with fixed effects and cluster robust standard errors for Model C.

†

p < .10. *p < .05. **p < .01. ***p < .001 level (two-tailed).

Table 2.

The Effect of Network Ties on Congressional Delegation.

Independent variables	Number of discretionary laws	R/L ratio
Independent variables	Model D	Model E
Network ties	.045*	2.744**
	(.019)	(.848)
Professionals	.122	5.276
	(.101)	(42.432)
Total employees	−.004	−45.885*
	(.014)	(16.634)
Budget	−.005*	−.407***
	(.002)	(.087)
NYT articles	.000	.595***
	(.001)	(.086)
GAO reports	.003	4.588***
	(.010)	(.759)
Unified	.050	−6.172
government (t + 1)	(.342)	(27.218)
Republican	.898	−101.892^†
congress (t + 1)	(1.184)	(57.151)
EGI width (t + 1)	−2.257	180.963
	(3.594)	(132.623)
Constant	−1.649	379.952**
	(1.573)	(119.691)
Inflate constant	−19.900***
	(.733)
Log pseudo likelihood	−100.381
Prob. > χ²	.0000
Prob. > F		.0000
Nonzero n	35
n	270	270

Note. Zero-inflated negative binomial regressions and cluster robust standard errors in parentheses for Model D (inflation model: logit), and linear panel regressions with fixed effects for Model E. NYT = New York Times; GAO = government accountability office; EGI = equilibrium gridlock region.

†

p < .10. *p < .05. **p < .01. ***p < .001 level (two-tailed).

Agency network ties and congressional delegation

This subsection examines the relationship between network ties and congressional delegation. Traditional bureaucratic politics literature has emphasized several independent variables that affect congressional delegation such as professionalism, divided government, and congressional oversight. In particular, a plethora of bureaucratic politics studies have argued that internal professionalism is the main factor that determines agency discretion and that legislators tend to delegate more discretion to highly professionalized agencies to acquire informational benefits (Bawn, 1995; Eisner & Meier, 1990; Rourke, 1984). Thus, the number of professional employees in each agency was included in regression models to control for the variable. Moreover, political conflict between legislative and executive branches has also been emphasized regarding congressional delegation (Oosterwaal, Payne, & Torenvlied, 2012). Legislators would like to provide more discretion to agencies under a unified government in which ideological differences between legislative and executive branches are insignificant (Epstein & O’Halloran, 1999; J. D. Huber & Shipan, 2002). The dummy variable of unified government was also included to control for the factor. Regarding this variable, this study assumes that the government of the 107th Congress was a unified government, because the Republican Party formed a unified government at the beginning of the 107th Congress, despite Senator James Jeffords (I-VT) switching from Republican to Independent during the congressional session.

In addition, if legislators are able to observe and correct agency misbehaviors through ex post oversight, they do not need to enact restrictive statues ex ante (Bawn, 1997; J. D. Huber & Shipan, 2002). In this sense, Bawn (1997) stated, “increasing ex post monitoring decreases the total and marginal benefits of statutory control” (p. 110). Therefore, more active oversight would decrease congressional delegation. The number of the Government Accountability Office (GAO) reports on specific agencies is included to control for congressional oversight. Furthermore, agency size was also controlled. Unquestionably, inappropriate policy implementation by agencies with higher policy resources or greater regulations may be more detrimental to legislators; a larger amount of policy resources can be wasted depending on agency size. Therefore, legislators may have more incentives to control sizable agencies to limit unfavorable policy consequences. Thus, budget authority and total employees were also included in regressions.

In addition, EO Measure and R/L Ratio can be affected by legislative productivity. Many legislative politics studies have argued that divided government (e.g., Binder, 2003) or pivotal legislators’ ideological positions (e.g., Krehbiel, 1998) are the main factors that cause legislative gridlock. Thus, both unified government and a variable of the equilibrium gridlock region (EGI) based on Krehbiel’s pivotal politics model were included in regressions. EGI represents the ideological distance of legislative gridlock where new lawmaking is impossible. In other words, broad EGI implies serious legislative gridlock. Given the pivotal politics model, the EGI of the model is always between the ideal points of veto pivot and filibuster pivot (Krehbiel, 1998). Because each congressional chamber has its own veto pivot, it is necessary to consider the pivotal actors of both houses simultaneously when calculating EGI widths. From this perspective, following Chiou and Rothenberg (2003), EGI was assumed as the union of both chambers’ legislative gridlock intervals. The ideological positions of pivotal legislators were based on DW-NOMINATE scores.

Furthermore, because sample agencies are highly regulatory, conservative legislators are more likely to control the regulatory agencies. In this respect, a dummy variable of Republican Congress is also included. Because annual update for statutory delegation is nearly impossible, some time lapse in enacting statutes may be inevitable. Therefore, political environment variables for legislative productivity (i.e., unified government and EGI width) and statutory delegation variables were measured based on the data for year t + 1, given network tie data of year t. In addition, the number of NYT articles citing specific agencies was also included to control for issue salience. This variable would have simultaneously positive and negative effects on EO Measure and R/L Ratio. Although legislators tend to introduce more bills under high salience (Birkland, 1997), they also have high incentive to control agency misbehaviors regarding salient issues (Bawn, 1997; Ringquist, Worsham, & Eisner, 2003).

EO Measure has lots of zeros; therefore, zero-inflated negative binomial regressions were adopted to limit the problem of over-dispersion. Because statistical tests were based on panel data, heteroskedasticity and autocorrelation problems were highly possible. To limit these problems, cluster robust standard errors were reported (Model D). In contrast, linear panel regression with fixed effects was used for the R/L Ratio model (Model E).

In Table 2, all the models show that network ties positively affect congressional delegation. When agencies have broader and stronger network ties, legislators are more likely to enact highly discretionary statutes (Model D), with agencies writing their regulations in detail by themselves, given the length of congressional statutes (Model E). These statistical results indicate that agencies are provided with more leeway in selecting policy methods if they have significant network ties with stakeholders. Moreover, for all models in Table 2, the statistical significance of the variable is greater than that of professionals, which represents the agency’s internal professionalism. This result does not mean that agency expertise has no effect on congressional delegation. Rather, it supports the proposition that agencies have insufficient expertise to make appropriate public policies with only internal professionals (Gray, 1989; O’Toole, 1997; Van Waarden, 1992). Networking with stakeholders is necessary for agencies to acquire sufficient professionalism for congressional delegation.

These results were consistent, even when inserting or changing some covariates—for example, substituting the number of GAO reports with the number of congressional hearings in which each federal agency participated as a witness or additionally including several independent variables such as average year of service and public liberalism scores of year t + 1.¹⁵ The regression results were also reexamined with dummy variables of individual agencies (for Model D), as well as year dummies and a dummy variable of independent agencies included (for both Models D and E).¹⁶ For all of the cases above, the statistical results remained robust at least at the 10% significance level.

Possible Statistical Shortcomings

The regression tests in the previous sections might have several statistical weaknesses regarding measurement validity and confounder problem. This study used several proxies for network ties, policy appropriateness, and congressional delegation. Because these concepts are highly abstract, questions could be raised about the measures used in this study. Nevertheless, the measures of the main variables have already been used by diverse public administration and political science studies. In addition, several measures were used for the same variable to enhance statistical robustness. Therefore, the validities of the measures might not be too low to make the regression results meaningless.

In addition, confounder problems might be significant, because there are a number of unobservable factors that affect policy appropriateness and congressional delegation. Therefore, the regression results may not completely eliminate the possibility of spurious relationships among network ties, policy appropriateness, and congressional delegation.

Nevertheless, several important variables such as public liberalism, unified government, and EGI width, which were assumed to be critical factors for policy appropriateness and congressional delegation by previous studies, were included in the regressions as controls. Furthermore, the suggested regression models of Tables 1 and 2 were reexamined by inserting cross-sectional and yearly dummies to limit confounder problems and/or excluding several outlier agencies. Therefore, in spite of their incompleteness, the regression results of Tables 1 and 2 may show meaningful causal relationships among network ties, policy appropriateness, and congressional delegation.

Conclusion

Networking has become an integral part of public administration. According to Meier and O’Toole (2006), “public managers inhabit a networked world, and many analyses have documented that these managers devote part of their energy and talents to dealing with their complicated, interdependent environment” (p. 697). Therefore, many public administration studies have emphasized that networking in administration has positive effects on policy implementation, program management, and innovation (Meier & O’Toole, 2006; O’Toole, 1997) and that networking improves political support in policy networks (Desario & Langton, 1987; Herian et al., 2012; Wang & Van Wart, 2007). They have also argued that agency network ties tend to enhance administrative performance (e.g., Herian et al., 2012; Neshkova & Guo, 2012; Nicholson-Crotty & O’Toole, 2004). However, the effect of policy network ties in terms of congressional delegation has not yet been fully examined. Thus, this article statistically tested the relationship among agency network ties, policy appropriateness, and congressional delegation.

Because elected officials are likely to delegate more policy authority to agencies capable of producing appropriate policy outcomes (J. D. Huber & McCarty, 2004; MacDonald & Franko, 2007; Rourke, 1984), agency network ties that can limit possible policy problems can positively affect the delegation decisions of elected officials. Supporting this hypothesis, several regression results indicate that agencies with more network ties are likely to be granted statutory discretion. In particular, in recent decades when policy stakeholders in all policy areas have developed significantly, both quantitatively and qualitatively, the implication of agency network ties may be greater (Holyoke, 2011). Given the growth and diversification of policy communities, agencies without sufficient network ties would lack sufficient information to select appropriate policy alternatives and respond to stakeholders in an appropriate manner thus, failing to maintain policymaking discretion.

These results are meaningful in clarifying the relationship between the political capabilities of agencies and the distribution of political influences in policy networks. Even though administration has frequently been assumed to be neutral, agency policymaking and implementation tend to be political (Rourke, 1984). Thus, in the policy networks that consist of diverse political actors such as agencies, policy stakeholders, and elected officials, the political capabilities of agencies could determine their political influence and policymaking authorities (Carpenter, 2001; J. D. Huber & McCarty, 2004). In this sense, both policy appropriateness and congressional delegation are meaningful. Policy appropriateness in this study is related to the political support from policy stakeholders. Similarly, congressional delegation generally implies agencies’ policymaking authority in policy networks. Therefore, these variables are important in examining the causal relationships between the political capabilities of agencies and the distribution of political influences in policy networks.

Admittedly, this study result does not fully clarify detailed causal paths from network tie to congressional delegation. Because agency network ties simultaneously contribute to informational advantages and stakeholder support, it is hard to determine which factor is the primary driver of congressional delegation using only statistical results. Nevertheless, this study clearly indicates at least that agency network ties positively affect congressional delegation by enhancing agency (administrative and political) capabilities in administration, contributing to investigations on the effect of agency network ties on effective administration and bureaucratic policymaking authority.

Footnotes

Appendix

Descriptive Statistics.

Variables	M	SD	Minimum	Maximum
Policy outcome quality
Courts of appeals cases	20.79	50.72	0	422
Supreme court cases	1.08	2.23	0	15
Ratio of stakeholders’ complaint articles	0.083	0.200	0	3
Discretion and discretionary budget
Number of discretionary laws	0.14	0.38	0	2
R/L Ratio	0.05	0.19	0	1.73
Independent variables
Network ties	2.76	9.22	0	75.44
Professionals	2.24	2.45	0.09	9.82
Average year of service	16.95	2.10	10.2	20.4
Total employees	9.97	14.29	0.92	64.00
Budget (in 2008 billion dollars)	56.53	169.40	−1.06	881.59
Significant rules	7.50	10.87	0	63
Public liberalism	58.75	2.53	54.94	62.64
NYT articles	124.92	239.07	0	1,804
GAO reports	5.96	10.68	0	93
EGI width	0.73	0.07	0.60	0.79
Federal agencies	Average network ties		Standard deviation of network ties
Agricultural marketing service	0.002		0.001
Animal and plant health inspection service	0.010		0.006
Bureau of alcohol, tobacco, firearms, and explosives	0.000		0.000
Bureau of Indian affairs	0.000		0.000
Bureau of land management	10.246		6.021
Centers for Medicare and Medicaid services	0.148		0.061
Environmental protection agency	0.367		0.069
Employment and training administration	0.004		0.009
Farm service agency	0.000		0.000
Federal aviation administration	0.413		0.204
Federal communications commission	0.065		0.043
Food and drug administration	2.095		0.562
Federal emergency management agency	0.005		0.008
Federal highway administration	0.081		0.177
Food and nutrition service	0.003		0.003
Food safety and inspection service	0.000		0.000
Forest service	19.968		15.506
Fish and wildlife service	0.053		0.011
Minerals management service	0.052		0.028
Mine safety and health administration	0.000		0.000
National park service	5.740		4.138
Nuclear regulatory commission	0.109		0.061
Occupational safety and health administration	0.042		0.022
Patent and trademark office	0.000		0.001
Small business administration	1.226		0.657
Securities and exchange commission	0.004		0.005
Social security administration	0.014		0.026

Note. The number of discretionary laws, R/L Ratio, and EGI width are measures of year t + 1. The data of individual agencies’ network ties are averages and standard deviations between 1999 and 2008. NYT = New York Times; GAO = government accountability office; EGI = equilibrium gridlock region.

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

Author Biography

Jongkon Lee is an assistant professor in the Department of Political Science and International Relations at the Ewha Womans University. He is interested in bureaucratic politics and principal-agent theory.

References

Ainsworth

S. H.

Harward

B. M.

(2009). Delegation and discretion in anticipation of coalitional drift. American Politics Research, 37, 983-1002.

Akkerman

Torenvlied

Schalk

(2011). Two-level effects of interorganizational network collaboration on graduate satisfaction: A comparison of five intercollege networks in Dutch higher education. The American Review of Public Administration, 42, 654-677.

Ansell

C. K.

(2011). Pragmatist democracy: Evolutionary learning as public philosophy. New York, NY: Oxford University Press.

Applegate

J. S.

(1998). Beyond the usual suspects: The use of citizens advisory boards in environmental decisionmaking. Indiana Law Journal, 73, 903-957.

Baekkeskov

(2017). Reputation-seeking by a government agency in Europe: Direct evidence from responses to the 2009 H1N1 “Swine” influenza pandemic. Administration & Society, 49(2), 163-189.

Balla

S. J.

(2004). Between commenting and negotiation: The contours of public participation in agency rulemaking. I/S: A Journal of Law and Policy for the Information Society, 1, 59-94.

Bawn

(1995). Political control versus expertise: Congressional choices about administrative procedures. The American Political Science Review, 89, 62-73.

Bawn

(1997). Choosing strategies to control the bureaucracy: Statutory constraints, oversight, and the committee system. Journal of Law, Economics, & Organization, 13, 101-126.

Binder

S. A.

(2003). Stalemate: Causes and consequences of legislative gridlock. Washington, DC: Brookings Institution Press.

10.

Bingham

L. B.

(2010). The next generation of administrative law: Building the legal infrastructure for collaborative governance. Wisconsin Law Review, 2010, 297-356.

11.

Birkland

T. A.

(1997). After disaster: Agenda setting, public policy, and focusing events. Washington, DC: Georgetown University Press.

12.

Brewer

G. A.

Selden

S. C.

(2000). Why elephants gallop: Assessing and predicting organizational performance in federal agencies. Journal of Public Administration Research and Theory, 10, 685-712.

13.

Burt

R. S.

(2004). Structural holes and good ideas. American Journal of Sociology, 110, 349-399.

14.

Campbell

A. C.

(2003). How policies create citizens. Princeton, NJ: Princeton University Press.

15.

Carpenter

D. P.

(2001). The forging of bureaucratic autonomy: Reputations, networks, and policy innovation in executive agencies, 1862-1928. Princeton, NJ: Princeton University Press.

16.

Carpenter

D. P.

(2010). Reputation and power: Organizational image and pharmaceutical regulation at the FDA. Princeton, NJ: Princeton University Press.

17.

Chiou

F.-Y.

Rothenberg

L. S.

(2003). When pivotal politics meets partisan politics. American Journal of Political Science, 47, 503-522.

18.

Desario

Langton

(1987). Citizen participation in public decision making. New York, NY: Greenwood Press.

19.

Donahue

A. K.

Selden

S. C.

Ingraham

P. W.

(2000). Measuring government management capacity: A comparative analysis of city human resources management systems. Journal of Public Administration Research and Theory, 10, 381-412.

20.

Doney

P. M.

Cannon

J. P.

(1997). An examination of the nature of trust in buyer-seller relationships. Journal of Marketing, 61, 35-51.

21.

Eisner

M. A.

Meier

K. J.

(1990). Presidential control versus bureaucratic power: Explaining the Reagan revolution in antitrust. American Journal of Political Science, 34, 269-287.

22.

Epstein

O’Halloran

(1999). Delegating powers: A transaction cost politics approach to policy making under separate powers. New York, NY: Cambridge University Press.

23.

Gailmard

Patty

J. W.

(2007). Slackers and zealots: Civil service, policy discretion, and bureaucratic expertise. American Journal of Political Science, 51, 873-889.

24.

Gargiulo

Benassi

(2000). Trapped in your own net? Network cohesion, structural holes, and the adaptation of social capital. Organization Science, 11, 183-196.

25.

Gray

(1989). Collaborating: Finding common ground for multi-party problems. San Francisco, CA: Jossey-Bass.

26.

Hansen

M. T.

(1999). The search-transfer problem: The role of weak ties in sharing knowledge across organization subunits. Administrative Science Quarterly, 44, 82-111.

27.

Heclo

(1978). Issue networks and the executive establishment. In King

(Ed.), The new American political system (pp. 87-124). Washington, DC: American Enterprise Institute.

28.

Herian

M. N.

Hamm

J. A.

Tomkins

A. J.

Zillig

L. M. P.

(2012). Public participation, procedural fairness, and evaluations of local governance: The moderating role of uncertainty. Journal of Public Administration Research and Theory, 22, 815-840.

29.

Holyoke

T. T.

(2011). Competitive interests: Competition and compromise in American interest group politics. Washington, DC: Georgetown University Press.

30.

Huang

(2014). Knowledge sharing in a third-party-governed health and human services network. Public Administration Review, 74, 587-598.

31.

Huang

Provan

K. G.

(2007). Structural embeddedness and organizational social outcomes in a centrally governed mental health services network. Public Management Review, 9, 169-189.

32.

Huber

G. A.

(2007). The craft of bureaucratic neutrality: Interests and influence in governmental regulation of occupational safety. New York, NY: Cambridge University Press.

33.

Huber

J. D.

McCarty

(2004). Bureaucratic capacity, delegation, and political reform. The American Political Science Review, 98, 481-494.

34.

Huber

J. D.

Shipan

C. R.

(2002). Deliberate discretion? The institutional foundations of bureaucratic autonomy. New York, NY: Cambridge University Press.

35.

Ibarra

(1992). Homophily and differential returns: Sex differences in network structure and access in an advertising firm. Administrative Science Quarterly, 37, 422-447.

36.

Ingold

Varone

(2012). Treating policy brokers seriously: Evidence from the climate policy. Journal of Public Administration Research and Theory, 22, 319-346.

37.

Innes

J. E.

Connick

Kaplan

Booher

D. E.

(January, 2006). Collaborative governance in the CALFED program: Adaptive policy making for California water (Working Paper Series Paper WP-2006-01). Berkeley, CA: Institute of Urban and Regional Development.

38.

Isett

K. R.

Provan

K. G.

(2005). The evolution of dyadic interorganizational relationships in a network of publicly funded nonprofit agencies. Journal of Public Administration Research and Theory, 15, 149-165.

39.

Jenkins-Smith

H. C.

Clair

G. K. S.

Woods

(1991). Explaining change in policy subsystems: Analysis of coalition stability and defection over time. American Journal of Political Science, 35, 851-880.

40.

Kelly

(2004). Unlocking the iron cage: Public administration in the deliberative democratic theory of Jürgen Habermas. Administration & Society, 36, 38-61.

41.

Kennedy

S. L.

Carpenter

W. J. D.

(1988). Managing public disputes: A practical guide to handling conflict and reaching agreements. San Francisco, CA: Jossey-Bass.

42.

Koch

B. J.

Galaskiewicz

Pierson

(2015). The effect of networks on organizational missions. Nonprofit and Voluntary Sector Quarterly, 43, 510-538.

43.

Krehbiel

(1998). Pivotal politics: A theory of U.S. lawmaking. Chicago, IL: University of Chicago Press.

44.

Langbein

L. I.

Kerwin

C. M.

(2000). Regulatory negotiation versus conventional rule making: Claims, counterclaims, and empirical evidence. Journal of Public Administration Research and Theory, 10, 599-632.

45.

Lee

Kim

(2011). Exploring the role of social networks in affective organizational commitment: Network centrality, strength of ties, and structural holes. The American Review of Public Administration, 41, 205-223.

46.

Lubbers

J. S.

(2008). Achieving policymaking consensus: The (unfortunate) waning of negotiated rulemaking. South Texas Law Review, 49, 987-1017.

47.

Lubell

(2000, August 31-September 3). Attitudinal support for environmental governance: Do institutions matter? Annual Meeting of American Political Science Association, Washington, DC.

48.

MacDonald

J. A.

Franko

W. W. J. R.

(2007). Bureaucratic capacity and bureaucratic discretion: Does congress tie policy authority to performance? American Politics Research, 35, 790-807.

49.

Mayhew

D. R.

(2005). Divided we govern: Party control, lawmaking and investigations, 1946-2002 (2nd ed.). New Haven, CT: Yale University Press.

50.

McFarland

A. S.

(1987). Interest groups and theories of power in America. British Journal of Political Science, 17, 129-147.

51.

Meier

K. J.

(1980). Measuring organizational power: Resources and autonomy of government agencies. Administration Society, 12, 357-375.

52.

Meier

K. J.

O’Toole

L. J.

(2006). Political control versus bureaucratic values: Reframing the debate. Public Administration Review, 66, 177-192.

53.

Moffitt

S. L.

(2010). Promoting agency reputation through public advice: Advisory committee use in the FDA. The Journal of Politics, 72, 880-893.

54.

Neshkova

M. I.

Guo

(2012). Public participation and organizational performance: Evidence from state agencies. Journal of Public Administration Research and Theory, 22, 267-288.

55.

Nicholson

Compeau

Sethi

(2001). The role of interpersonal liking in building trust in long-term channel relationships. Journal of the Academy of Marketing Science, 29, 3-15.

56.

Nicholson-Crotty

O’Toole

L. J.

(2004). Public management and organizational performance: The case of law enforcement agencies. Journal of Public Administration Research and Theory, 14, 1-18.

57.

Oosterwaal

Payne

Torenvlied

(2012). The effect of political disagreement on discretion: Mechanisms and conditions. Administration & Society, 44, 800-824.

58.

O’Toole

L. J.

(1997). Implementing public innovations in network settings. Administration & Society, 29, 115-138.

59.

Plummer

Fitzgibbon

(2004). Co-management of natural resources: A proposed framework. Environmental Management, 33, 876-885.

60.

Provan

K. G.

Huang

(2012). Resource tangibility and the evolution of a publicly funded health and human services network. Public Administration Review, 72, 366-375.

61.

Putnam

R. D.

(2000). Bowling alone. New York, NY: Simon and Schuster.

62.

Ringquist

E. J.

Worsham

Eisner

M. A.

(2003). Salience, complexity, and the legislative direction of regulatory bureaucracies. Journal of Public Administration Research and Theory, 13, 141-164.

63.

Rourke

F. E.

(1984). Bureaucracy, politics, and public policy (3rd ed.). Boston, MA: Little, Brown.

64.

Sabatier

P. A.

Jenkins-Smith

H. C.

(1993). Policy change and learning: An advocacy coalition approach. Boulder, CO: Westview.

65.

Schneider

Scholz

Lubell

Mindruta

Edwardsen

(2003). Building consensual institutions: Networks and the national estuary program. American Journal of Political Science, 47, 143-158.

66.

Srivastava

S. B.

(2015). Network intervention: Assessing the effects of formal mentoring on workplace networks. Social Forces, 94(1), 427-452.

67.

Stazyk

E. C.

Goerdel

H. T.

(2011). The benefits of bureaucracy: Public managers’ perceptions of political support, goal ambiguity, and organizational effectiveness. Journal of Public Administration Research and Theory, 21, 645-672.

68.

Stimson

J. A.

(1998). Public opinion in America: Moods, cycles, and swings (2nd ed.). Boulder, CO: Westview.

69.

Susskind

Cruikshank

(1987). Breaking the impasse: Consensual approaches to resolving public disputes. New York, NY: Basic Books.

70.

Sykes

T. A.

Venkatesh

Gosain

(2009). Model of acceptance with peer support: A social network perspective to understand employees’ system use. MIS Quarterly, 33, 371-393.

71.

Thomson

Torenvlied

Arregui

(2007). The paradox of compliance: Infringements and delays in transposing European union directives. British Journal of Political Science, 37, 685-709.

72.

Van Waarden

(1992). Dimensions and types of policy networks. European Journal of Political Research, 21, 29-52.

73.

Wang

Van Wart

(2007). When public participation in administration leads to trust: An empirical assessment of managers’ perceptions. Public Administration Review, 67, 265-278.

74.

Wellman

Wortley

(1990). Different strokes from different folks: Community ties and social support. American Journal of Sociology, 96, 558-588.

75.

Wilson

J. Q.

(1989). Bureaucracy: What government agencies do and why they do it. New York, NY: Basic Books.

76.

Wirgau

J. S.

(2013). Quality and access: Bureaucratic autonomy and the role of storylines at the national endowment for the arts. Administration & Society, 45, 296-325.