1231

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=rter20

Critical Studies on Terrorism

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/rter20

Pork, risk, or reaction? The determinants of US counterterrorist funding

Kyle Kattelman

To cite this article: Kyle Kattelman (2021) Pork, risk, or reaction? The determinants of US counterterrorist funding, Critical Studies on Terrorism, 14:4, 514-535, DOI: 10.1080/17539153.2021.1996936

To link to this article: https://doi.org/10.1080/17539153.2021.1996936

Published online: 08 Nov 2021.

Submit your article to this journal

Article views: 173

View related articles

View Crossmark data

Pork, risk, or reaction? The determinants of US counterterrorist funding Kyle Kattelman

Department of Social Sciences and History, Fairleigh Dickinson University, Teaneck, USA

ABSTRACT A wealth of literature exists on the relationship between a country’s domestic politics and its counterterrorism policies, yet few examine this relationship in the US. Past studies examining counterterrorism spending in the US find that domestic politics have little impact and that funding is distributed to the states largely as a function of its risk from a terrorist attack. Yet these studies suffer a number of shortcomings, including limited data on counterterrorist funding and a superficial conceptualisation of the variables that measure domestic politics. Using data on US outlays from the Department of Homeland Security Grants Program from 2003 to 2018, this study seeks to illuminate the process of counterterrorist spending in the US by reconciling three separate sets of predictors in the literature. Specifically, is counterterrorist spending distributed according to legislative politics (pork), genuine, rationally calculated population vulnerability (risk), or public fear in the wake of particularly notor- ious terrorist attacks (reaction)? The results support and expand the existing literature in that risk remains an influential variable, though grant distribution is still affected by political incentives. Finally, states do tend to see an influx of funds in reaction to terrorist attacks, though the source of the group plays no role.

KEYWORDS Terrorism; counterterrorism; funding; Homeland Security grants

Introduction

The effects of a country’s domestic politics, culture, and public opinion on its counter- terrorism policies is well documented, both abroad (Rees and Aldrich 2009; Perliger 2012; Miller 2007; Kattelman 2016), as well as in the US (Crenshaw 2001; Davis and Silver 2004). Yet few have examined how this relationship applies to counterterrorism (CT) spending. The attacks on 9/11 drastically shifted security priorities for the West, as well as the developed world. The most visible aspect of this shift was the advent of the US-led Global War on Terror in which a coalition of states engaged in open military operations, strengthened law enforcement and security, and severed the finances of terrorist organisations (US Department of State 2002). Since then, states have been very public about the wide and far- reaching actions taken to combat the spread of global extremism (Country Reports on Terrorism 2020). Yet, they have been less forthcoming on how these actions are funded at anything beyond a general assessment (Immenkamp et al. 2020).

CONTACT Kyle Kattelman [email protected]

CRITICAL STUDIES ON TERRORISM 2021, VOL. 14, NO. 4, 514–535 https://doi.org/10.1080/17539153.2021.1996936

© 2021 Informa UK Limited, trading as Taylor & Francis Group

This pattern is readily apparent in the US, which has seen an average annual increase of $360 billion between 2001 and 2015 on homeland security spending (“European Parliament: Think Tank” 2016). In spite of this growth, our understandings of the specific mechanisms behind it remain obscure. One of the more visible acts in the fallout from 9/ 11 was the creation of the Department of Homeland Security in 2002. Though its tasks are myriad, its primary responsibilities consist of preventing terrorist attacks, reducing the US vulnerability to terrorism, and minimising damage from terrorist attacks (US Department of Homeland Security 2005). Crucial to achieving these objectives is the Homeland Security Grant Program (HSGP), a programme comprised of a number of individual grants that disperse funds to US states and territories, and a programme that has endured repeated criticism of trivial “pork” spending (Friedman 2005; Kaplan 2006).

Despite these criticisms, the few statistical studies that take a direct look at CT spend- ing in the US find little evidence of pork behaviour, largely concluding that domestic politics play a minimal role and that funds are primarily driven by risk assessment – that is, funds are directed to areas of the country that are the most at risk from terrorist attacks. One of the first to examine this in the post-9/11 era, Coats, Karahan, and Tollison (2006) do find that funds are directed to states with larger Electoral College votes, providing some support for political influence, but they conclude that risk of terrorist attack remains the primary driver. Prante and Bohara (2008) expand upon this study by incorporating data after the passage of the Department of Homeland Security (DHS) Appropriation Act of 2006 (2005), which allowed the DHS more discretion on grant outlays and thus more room for potential political influence. The authors again show risk to be the primary driver and find little evidence of political influence. However, their data only cover 2004–2006, a limited time frame and one where the degree of party polarisation was much lower than it is today. Taken together, these studies suggest that more work is needed in under- standing the specific processes that affect CT spending.

Despite the DHS’ insistence that grant money is distributed according to the most terrorism-prone areas, allegations still abound that many politicians are securing money for constituents (Lipton 2004), special interests (de Rugy 2005), and even academic institu- tions (US Department of Homeland Security 2005). Compounding this spending issue is a fearful public that constantly demands visible government action after every incident of terrorist violence (Jordan, Kosal, and Rubin 2016). Therefore, the specific goals of this study are to analyse the sources of counterterrorist spending in the US by reconciling three separate sets of predictors in the literature. More specifically, are HSGP outlays distributed according to legislative politics (pork), genuine, rationally calculated citizen vulnerability (risk), or public fear in the wake of particularly notorious terrorist attacks (reaction)?

Due to the lack of robust research upon which to build, this study serves as a first- generation, exploratory approach to the predictors of US CT spending. The organisation of the article proceeds as follows. In the next section, I discuss the three possible categories of motivators gleaned from the literature. For each category I derive hypoth- eses, which I then test using a series of OLS and logit regression models on several comprehensive grants from the HSGP from 2003 to 2018. Following a discussion of the results, I then draw conclusions and call on a more focused and robust approach to understanding the allocation of CT funds.

CRITICAL STUDIES ON TERRORISM 515

Motivators of counterterrorist spending in the US

A large body of literature shows that electoral incentives influence the national security politics of democratically elected governments (Hess and Orphanides 1995; Marinov, Nomikos, and Robbins 2015; Schneider 2013; Smith 1996). Looking at US defence Thorpe (2010, 2014) finds that economic dependence on defence spending in rural districts results in congressional members from these districts being more likely to support weapons expenditures, independent of whether such measures improve national security. Constituent dependence on a war economy in these areas gives rise to candidates who seek to gain and keep their elected seats by funnelling military expenditures back to their home districts. The attacks of 9/11 likely deepened this dependence, as research on other grant programmes show how exogenous shocks to the political system create “policy windows” in which political entrepreneurs can pass legislation for a variety of unrelated causes.1 Evidence of this shock can be seen in the dramatic increase in defence spending earmarks shortly after 9/11 and the complaints that followed it (Newman 2021). Dubious spending allocations in the name of national security have also plagued the HSGP specifi- cally. For example, in 2005, landlocked Tulsa, Oklahoma, received $725,000 in port security funds. In that same year, Grand Forks, North Dakota, received $1.5 million in federal funds to respond to nuclear attacks (Friedman 2005). Such examples are numerous throughout the country.

To confront the allegations of rural patronage, in 2006 the DHS announced a renewed focus on the most vulnerable areas through its Urban Area Security Initiative (UASI), which provides recourses that cater to the needs of urban areas deemed a high threat. In spite of this, critics are quick to point out the large number of urban areas deemed high-threat; furthermore, a number of the included cities seat congressional members on the House Appropriations Homeland Security Subcommittee (Kaplan 2006).2

In short, the evidence indicates that there should be some form of politically motivated spending with regard to the HSGP, yet direct, statistical examinations of the programme find that it is primarily driven by risk assessment. These findings warrant a new examination. The next section derives a series of hypotheses of pork spending by tying it to the larger body of literature on legislative behaviour. Following that, additional hypotheses are derived from competing explanations (i.e. risk and reaction) before being tested against one another.

Pork: domestic political motivators

As previously stated, there is a robust set of literature that finds electoral incentives play a role in national security politics. Regarding the proximity of elections, Bueno de Mesquita (2007) notes that politicians favour defence spending to the extent that it increases security and provides an observable contribution for which they can claim credit. These two motivations are not mutually exclusive; however, when voter and government preferences are not perfectly aligned, the government will increasingly favour allocating CT resources that are observable to voters. Thus, re-election preferences play a role in the allocation process, leading governments to increase spending as their proximity grows near (Bueno de Mesquita 2007; Lonardo 2019; Nanes 2017). Applying this logic to the US allows for the first hypothesis:

516 K. T. KATTELMAN

H1: CT funding to states should increase during a presidential election year.

Partisanship can also play a role. With regard to government behaviour, right-leaning parties trend more hawkish on policies of national defence than left-leaning parties (Budge and Hofferbert 1990; Fordham 1998; Russett 1990). Translating this theory to CT funds directed to individual US states, we can test for similar patterns of behaviour by utilising the partisanship and ideology of individual, state-elected legislators. In the US, federal statutes define domestic terrorism, yet there are no specific laws that prohibit it. By contrast, there exists a robust set of federal statutes, in addition to an officially designated list of foreign terrorist organisations which ban acts of international terrorism. As a result, elected officials in the US primarily view terrorism as a foreign threat (Rotella 2021). From this, it can be theorised that legislators with a conservative ideology would be in favour of increased CT spending, both at the national level and in the interest of securing funds for their respective state, in the name of national security from a foreign threat. This logic results in the second hypothesis:

H2: CT funding to a constituency should increase if its congressperson is conservative.

Beyond individual partisanship, majority party control can also expedite funding approvals if the policy is in line with its interests. In his analysis of Western European democracies, Askoy (2018) finds that CT spending during election years is conditioned on partisan ideology, particularly when right-oriented governments are in power. This is due to the fact that supporters of right-wing governments tend to favour strong national security, while supporters of left-wing governments may be turned off by an increase in counterterrorism funding, believing that the money can be used for alternative policies. This dovetails with previous literature arguing that defence spending is partly predicated on the ideological orientation of the elites in power (Askoy 2018; Budge and Hofferbert 1990; Fordham 1998; Russett 1990). Taken together, the findings from this subset of research can be modified for testing within the US. That is, when right-wing governments are in power they leverage their majority party status to acquire more CT funding, leading to the third hypothesis:

H3: CT funding to a constituency should increase if the legislature is Republican controlled.

Finally, a central feature of classical theories on legislative behaviour is the ability of party leaders to advance collective partisan goals, most importantly the maintenance of majority-party status (Cox and McCubbins 1993, 2005; Aldrich and Rohde 2000, 2001), an ability that includes influencing the distribution of spending projects (Lazarus 2009). A vast literature points to the effects that House and Senate Appropriations Committee membership has on the value of earmarked funds to districts (Allen 2007; Binder 2008; Lazarus 2009, 2010). Based off of the distributive model, by which legislators utilise their advantageous committee positions to procure funds for their district to secure re-election, it can be theorised that members of the House and Senate Department of Homeland Security subcommittee are in an advantageous position to engage in pork politics to direct funds to their districts, especially if one

CRITICAL STUDIES ON TERRORISM 517

is a chair (i.e. cardinal) of the subcommittee (Berry and Fowler 2016). Indeed, one key criticism of the UASI is that the cities that receive eligibility have legislators in key positions on the DHS subcommittee. These institutional characteristics lead to the final set of hypotheses in this section:

H4: CT funding to a constituency should increase if its congressperson is a member of the DHS appropriations subcommittee.

H5: CT funding to a constituency should increase if its congressperson is a chair of the DHS appropriations subcommittee.

H6: CT funding to a constituency should increase if its congressperson is a party leader.

Risk: statistical motivators

In spite of its criticisms, the DHS remains steadfast that its funding allocation is guided by risk assessment and that the most terrorist prone areas receive the most funding. Prior to 2006, many HSGP funds were distributed by a formula that allocates 0.75% of the total to each state with the remaining distributed proportionally to a state’s population (H.R. 2360 2005). While it is reasonable to assume that terrorist attacks would correlate with popula- tion, this method was still noted as inefficient (Earle 2004). Thus, the Department of Homeland Security Appropriation Act of 2006 (2005) gave the DHS more discretion in its distribution process, ostensibly allowing them greater leeway to distribute based on risk.

The few robustness checks on this formula seem to confirm that risk has an influence. Prante and Bohara (2008) cite risk as the main determinant in CT distribution, more so than their measures of political power and partisanship, which they found to be insignif- icant. Coats, Karahan, and Tollison (2006) find that terrorism risk factors more into grant distribution than political variables. However, they do find a political effect in that funds are directed to states with greater Electoral College votes. Their reasoning is that the US winner-take-all system exists as a political motivator for presidential candidates to use the DHS as an arm of their re-election campaign to win the states with the largest number of votes. Nevertheless, there are two methodological problems with this approach. The first is that during the period their data cover, grant allocations were distributed according a formula that accounts for a state’s risk, population, and population density. Hence, there is little discretion in executive ability to manipulate this distribution pattern. Moreover, the theoretical explanation outlined in this present study flows from literature that stresses congressional committee oversight of the bureaucracy (MacDonald and McGrath 2016).

The second relates to the utilisation of Electoral College votes as a political variable. Since these vote allocations are correlated entirely with a state’s population, it is impos- sible for the authors to separate the political effects of this variable from its relationship to risk. Furthermore, these analyses conflict with statements from the DHS itself. A DHS Inspector General’s audit of its Port Security Grant Program found a gross misallocation of funds, finding that “Grant award decisions are made with the intent of expending all

518 K. T. KATTELMAN

available funding and spreading funds to as many applicants as possible.” (Office of Inspector General 2005, 4). These discrepancies warrant a renewed look at risk-based allocation, which leads to the seventh hypothesis:

H7: CT funding to a constituency should increase with its level of risk from a terrorist attack.

Reaction: psychological motivators

Despite the fact that the handful of major terrorist attacks in the US that have occurred in the past 20 years are nowhere comparable to the magnitude of those on 9/11, public fears of terrorism since that day remain heightened (Haner, Sloan, and Cullen 2019). This is partially attributable to our cognitive biases. People rarely apply expected utility calcula- tions to everyday thinking and instead tend to overinflate the probability of big payoff or epic disaster scenarios. Furthermore, individuals tend to put greater weight towards potential losses rather than gains, resulting in a need to eliminate risk and protect the status quo (Sunstein and Zeckhauser 2008). The catastrophic nature of the 9/11 attacks imprinted into the minds of millions of Americans the necessity of preventing additional loss, and the public’s singular focus on that event leads to an inflation of its likelihood of repeat.

These feelings are reinvigorated each time a new attack happens, resulting in renewed calls for security measures, due to our tendency to view unrelated and scattershot acts of violence as part of an intentional, centralised agency (Jordan, Kosal, and Rubin 2016; Bloom and Veres 1999; Jervis 1976). A recent example of this includes the bipartisan call for the expansion of the HSGP in the wake of the anti-Semitic shooting in Jersey City and the US assassination of Maj. Gen. Qasem Soleimani (Roberts 2020). These fears, when coupled with a military-industrial complex that benefits from military spending, create what Friedman (2011) calls “ . . . a permissive environment for overreaction to terrorism. Security politics becomes a seller’s market where the public will overpay for counter- terrorist policies.” Hence, the problem of spending could be less a matter of government supply and more one of public demand, resulting in the next set of hypotheses:

H8: CT funding to a constituency should increase with the frequency of terrorist attacks it suffers.

H9: CT funding to a constituency should increase with the severity of terrorist attacks it suffers.

Furthermore, this fear varies with partisanship in the citizenry. The role of party identification as a heuristic for interpreting political information is widely established in American politics research (Campbell et al. 1960; Norpoth, Lewis-Beck and Jacoby 2008; Lodge and Hamill 1986). Overall, Republicans are seen as internationalists who favour authoritative foreign policy tactics and strategies, while Democrats are seen as accommodationists, a pattern that has intensified over time (Holsti and Rosenau 1988; Holsti 2004). To this end, polls consistently show that a majority of Republicans favour

CRITICAL STUDIES ON TERRORISM 519

an increase in anti-terrorism defence spending within the US, while only roughly 1 in 3 Democrats favour an increase (Pew 2017). Examining counterterrorist threats more generally, over two-thirds of Republicans think that government policies do not go far enough in preventing terrorism in contrast to just 46% of Democrats (Pew 2016a). The mitigating effects of partisanship can vary the intensity of public calls for government action on a state-by-state basis, leading a legislator to react to that demand in the interest of preserving their seat. This leads to the final hypothesis:

H10: Constituencies with a more conservative citizen ideology should receive more CT funding.

Data and methodology

Dependent variables

Beginning with the dependent variables, three sources are used. The first two are discretionary funding from the State Homeland Security Grant Program(SHSGP) and the Emergency Management Performance Grant Program (EMPG) (Department of Homeland Security 2020). These two sub grants of the Homeland Security Grant Program were chosen because they afford the greatest time span of data. Other subcategories either do not last as long or draw their funds off of the programmes used here.3 These two grants will be combined to create a total funding variable, formatted in millions of raw dollars. Prior to the DHS Appropriation Act of 2006 (2005), each state received a base of 0.75% of the total distribution for a respective programme with the remaining funds distributed based on state population. Since 2006, the base funds are still allocated, but risk and the anticipated effectiveness of proposed investments are used instead of population, which gives the DHS more discretion over distribution. As a result, 0.75% of each state’s allocation is subtracted from the total to ensure that only discretionary funds, subject to variable influence, are used. The unit of analysis is a state-year.

The second dependent variable measures the eligibility of a metropolitan area to receive funds from the Urban Areas Security Initiative (UASI), a programme that provides funding to enhance security in high-threat, high-density areas (FEMA 2021). Eligibility is determined by assessing the relative risk of a terrorist attack in the 100 most Metropolitan Statistical Areas (MSAs) in the US (Homeland Security Act of 2002). The unit of analysis, district-year, is constructed by decomposing each MSA into its respective congressional district(s). The variable is coded as 1 if a city’s district is eligible for UASI funds in a given year and 0 otherwise.

Taken as a whole, the majority of funding from the Homeland Security Grant Program stems from these programmes.4 It is important to note that these three variables do not represent the entire domain of security spending. Due to the opaque nature of the subject, a number of sources are omitted,5 though these contributions satisfy a number of conditions outlined in the literature that are being tested. Specifically, each outlay/ eligibility decision flows directly to a state/district, that allocation can ultimately be traced back to specific congressional action, and that action is visible to the public. The timeline

520 K. T. KATTELMAN

of the data extend from 2003, which represents a seismic shift in counterterrorist spend- ing the wake of the 9/11 attacks, until 2018, the latest report on outlays and jurisdictions for the 115th Congress. A timeline of spending and city eligibility is presented in Figure 1.

Two main peaks are visible – the first in 2003 captures the quick acceleration of the invasion of Iraq, and the second spike beginning roughly in 2008 reflects spending increases when the wars in Afghanistan and Iraq were at their height. A second notable feature is the prominence of UASI grants that roughly follow the contours of the outlays. UASI grants are specifically allocated out of a risk calculation based on population density, the presence of critical infrastructure, and the presence of a credible threat (Reese 2004). However, as a block grant, a number of city officials began to lobby congressional legislators who felt unfairly left out. As a result, the original pool of 7 cities grew to 30 at their peak, with the pool of money increasing from $100 million to $700 million (Ripley 2004). Efforts to revise the funding formula to a more valid measure of risk have been met with pushback in Congress. For example, HR10 included language to reduce automatic state grants from 0.75% to 0.25%. This was denounced by Senator Leahy (D-VT), a member of the Homeland Security Appropriations Subcommittee, as an effort to “short- change rural states” (Ripley 2004, 5). Such findings present a vivid account of the dynamic between pork and risk on CT spending and lend tacit support for the hypotheses outlined above.

To get a preliminary view of the distribution of spending, Table 1 presents a simple per capita outlays-by-state in descending dollar amount. A clear pattern that emerges is that spending disproportionately favours the rural states. Indeed, the states that routinely fall

Figure 1. Federal homeland security outlays and UASI eligibility: 2003–2018.

CRITICAL STUDIES ON TERRORISM 521

in the bottom 10 of population are among the highest earners of Homeland Security outlays. The reverse is true for the most populous states, as California, Florida, and Texas are the lowest receivers. The pattern does not map perfectly, as New York sits in the middle of the pack. It is important to stress that this study examines discretionary spending only, as including the baseline each state automatically receives would dispro- portionately favour the rural states. The distribution of these specific grants fit into the broader patter of US defence spending that favours rural regions (Thorpe 2010, 2014).

Independent variables

To measure the effects of pork, risk, and reaction, three subsets of independent variables are used. Beginning with the electoral and institutional features that are hypothesised to influence pork, a dummy variable is constructed for a presidential election year. The theory outlined here relies on the public visibility of presidential rather than congressional elections as a primary driver of CT outlays. To capture the partisanship of individual legislators, DW- NOMINATE scores are used (Lewis et al. 2021). Scores are averaged across legislators by state within a specific year for each chamber, resulting in separate measures for House and Senate. Individual legislator scores from the House are used at the district level. Since the dependent variables are spending decisions, only the economic ideology dimension is used, with a higher value indicating a more conservative economic ideology. Additional partisan effects include a dummy variable indicating majority party status for the Republican Party and a count variable indicating the number of party leaders within a state or district.

Table 1. Average per capita outlays by state, 2003–2018. (1) Wyoming $7.20 26. New York $1.62 (1) Vermont $6.54 27. Kentucky $1.59 (1) North Dakota $5.95 28. South Carolina $1.52 (1) Alaska $5.76 29. Alabama $1.51 (1) South Dakota $5.10 30. Maryland $1.48 (1) Delaware $4.74 31. Colorado $1.48 (1) Montana $4.33 32. Minnesota $1.43 (1) Rhode Island $4.10 33. Washington $1.39 (1) New Hampshire $3.42 34. Missouri $1.39 (1) Maine $3.40 35. Massachusetts $1.39 (1) Hawaii $3.34 36. Wisconsin $1.39 (1) Idaho $3.03 37. Arizona $1.33 (1) Nebraska $2.72 38. Indiana $1.32 (1) West Virginia $2.68 39. Tennessee $1.32 (1) New Mexico $2.47 40. New Jersey $1.31 (1) Nevada $2.10 41. Virginia $1.30 (1) Utah $2.01 42. Illinois $1.24 (1) Kansas $2.00 43. Georgia $1.21 (1) Arkansas $1.94 44. Michigan $1.20 (1) Mississippi $1.92 45. North Carolina $1.16 (1) Iowa $1.89 46. Pennsylvania $1.16 (1) Connecticut $1.80 47. Ohio $1.15 (1) Oklahoma $1.69 48. California $1.14 (1) Oregon $1.66 49. Texas $1.05 (1) Louisiana $1.63 50. Florida $1.03

522 K. T. KATTELMAN

For institutional power, since committee membership can influence the direction of funds, a count of a state’s legislator membership on the Subcommittee on the Department of Homeland Security, seated within the Appropriations Committee, is used. These measures are used as specific committee assignments and have been shown to result in particular constituent benefits (Evans 1994; Lee 2003; Balla et al. 2002; Lazarus 2009, 2010). Finally, it has been shown that the chair position rather than membership is the most influential position for pork distribution (Berry and Fowler 2016). As such, a dummy variable is coded for whether a state has a legislator who is the cardinal of the DHS Appropriations Subcommittee, from either chamber.

Measuring the degree of terrorism risk is inherently tricky, as assessments are often sensitive in nature and not publicly available. Prior to 2006, the DHS-based discretionary outlays on state population, switching to more comprehensive risk assessment mea- sures afterwards. Therefore, it is important to employ a measure alongside state popu- lation to check for evidence of this change (Prante and Bohara 2008). While the DHS’ exact risk assessment methodology remains unknown, a relatively objective approxima- tion is captured in the Applied Insurance Research Terrorism Loss Estimation Model (AIR Worldwide 2016), which utilises the Delphi method to condense input from terrorism experts and former employees of the FBI, CIA, and Department of Defence. The measure puts heavy emphasis on the presence of high-profile targets, tourist attractions, and infrastructure rather than state population or per capita income (Boyle, Malinowski, and Gannon 2002). The original measure consists of three ordinal values based on a state’s level of risk (low, medium, and high), but for purposes of modelling it will be composed into two dummy variables of medium and high, with the low risk serving as the base category. Due to data limitations, the risk data are only available for 2004. It is assumed here that there would be little temporal variation if more years were available, and as such the data are applied to the entire timeframe of the dataset. Still, while it is likely that a state such as Wyoming will remain low and New York high, variation in the middle category is a concern.

The last subset, reaction, is measured according data on terrorist attacks taken from the Global Terrorism Database (START 2021). The organisation defines terrorism as “ . . . the threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation” (GTD Codebook 2017, 9).6 The data include geo-coordinates on the specific location of the attack, allowing for attacks to be matched to the specific location in which they occurred. To measure severity, a casualty count is also included. Both variables are expected to hold positive relationships with both dependent variables.

Finally, to capture public demand for CT funding, Berry et al.’s (1998) measure of citizen ideology is included. The measure consists of an index of the ideological positions of incumbent and challenger weighted by the level of citizen support within each district, with larger values indicating a more liberal citizen ideology.

Control variables include separate dummy variables indicating whether a state or district borders Mexico, Canada or an ocean. The natural log of a state’s population is included, since the DHS used population as a metric of grant allocation prior to 2005. To capture the general dependence on federal funding, a state’s per capita GDP is included (Pew 2016b). Finally, evidence suggests a racial dynamic to perceptions of terrorism. Recent evidence finds that white anxiety over demographic shifts leads to demand for

CRITICAL STUDIES ON TERRORISM 523

harsher counterterrorism measures (Patel 2017; Piazza 2020). To control for this possibility, the percentage of the state’s population categorised as white is included. All demo- graphic data are taken from the US Census Bureau (2020). For application to UASI eligibility, data on a given metropolitan area are used. Summary statistics for both units of analyses are presented in Table 2. To account for temporal variation not specified by the included variables, year dummies are used in all models.

Results

Funding

Table 3 presents OLS models for funding outlays to the states. A Fisher test indicates no unit root presence, indicating that the underlying process is stationary. Model 1 in the first column serves as a baseline that exclusively features the control variables. The results largely hold to expectations, with population and a state’s GDP exerting positive influence on funding outlays. Both measures are often used as an instrumental variable for risk, and the DHS (2009) directly employs state population to determine the amount of funding allocated. All models include a one-year lag of the dependent variable on the right side of the equation, and the lagged DV behaves as predicted, with current funding a function of the previous year’s outlays.

Moving to the second column, model 2 includes the set of variables testing hypotheses 1–6 (pork) and finds support for institutional and political variables, though none with legislator partisanship. Interpretation of these OLS coefficients against low count and dummy variables is relatively straightforward. Each number in the model represents the change in raw dollars per each unit increase in the independent variable. For example, the coefficient on DHS committee chair membership (i.e. Cardinal) is 0.31. To translate that into a substantive interpretation, simply multiply that coefficient by the different values that the cardinal variable could be. Since there are only two values to this variable (i.e. 1 if a state contains a cardinal and 0 otherwise), there are only two predicted relationships.

Table 2. Summary statistics of state outlays (district values in parentheses). Variable Obs. Mean Std. Dev. Min. Max.

Total funds (in millions) 918 8.13 7.48 .05 54.16 UASI funding 2098 0.50 0.50 0 1 Election year 920(2098) 0.25(0.25) 0.43(0.43) 0(0) 1(1) Cardinal dummy 920 – 0.04 – 0.19 – 0 – 1 – GOP majority 920(2098) 0.62(0.63) 9.42(0.41) 0(0) 1(1) DHS member 920(2098) 0.41(0.04) 0.58(0.18) 0(0) 3(1) Party leader 920(2098) 0.16(0.07) 0.42(0.25) 0(0) 3(1) House NOMINATE 796(2098) 0.05(−0.10) 0.23(0.44) −0.49(−0.70) 0.68(0.75) Senate NOMINATE 800 – 0.06 – 0.35 – −0.64 – 0.67 – High Risk 816(2098) 0.06(0.29) 0.24(.45) 0(0) 1(1) Medium Risk 816(2098) 0.16(0.28) 0.36(0.45) 0(0) 1(1) Attack count 920(2098) 0.48(1.50) 1.28(2.60) 0(0) 14(14) Casualties 920(2098) 0.38(1.27) 2.94(5.10) 0(0) 59(59) Citizen ideology 700 – 51.91 – 15.80 – 13.48 – 97.00 – Log population 832(2098) 15.12(14.61) 1.02(1.11) 13.13(13.15) 17.49(16.77) Canada 920(2098) 0.17(0.17) 0.38(0.37) 0(0) 1(1) Mexico 920(2098) 0.07(0.23) 0.25(0.42) 0(0) 1(1) Ocean 920(2098) 0.37(0.65) 0.48(0.48) 0(0) 1(1) Percent white 896 – 66.32 – 25.09 – 0.70 – 96.5 – GDP per capita 832(1951) 49813.89(10.82) 10100.16(0.27) 19820.21(9.84) 79894(11.76)

524 K. T. KATTELMAN

Table 3. Predictors of CT outlays to US states, 2003–2018. Model1 Model2 Model3 Model4 Model5 Model6 Model7

Test of Pork Hypotheses

H1:Election year t – 1 – 0.90*** (0.16)

0.90*** (0.16)

0.80*** (0.20)

0.90*** (0.16)

0.08(0.05) 0.09(0.05)

H2aHouse ideology t – 1

– 0.41(0.26) 0.37(0.26) 0.44(0.28) 0.39(0.25) 0.41(0.26) 0.38(0.28)

H2bSen. ideology t – 1 – 0.01(0.26) −0.004 (0.27)

−0.01 (0.27)

0.01(0.26) 0.01(0.26) −0.02 (0.28)

H3GOP majority. t – 1 – −0.94*** (0.17)

−0.84*** (0.22)

−0.59* (0.33)

−0.97*** (0.17)

−0.94*** (0.17)

−0.54* (0.33)

H4:DHS mem. t – 1 – 0.07(0.10) 0.09(0.11) 0.35* (0.18)

0.07(0.10) 0.07(0.10) 0.35** (0.18)

H5:Cardinal dummyt – 1

– 0.31*(0.16) 0.39*(0.21) 0.27** (0.13)

0.07(0.17) 0.31*(0.16) 0.06(0.13)

H6:Party leader t – 1 – −0.08(0.16) 0.18(0.29) −0.06 (0.16)

−0.10(0.16) −0.08(0.16) 0.16(0.27)

Party leader*GOPmaj t – 1

– – −0.59(0.44) – – – −0.54 (0.42)

DHS*GOPmaj t – 1 – – – −0.63* (0.34)

– – −0.60* (0.31)

Cardinal*GOPmaj t –

1

– – – – 0.52*(0.31) – 0.61(0.44)

Election Year*GOPmaj

– – – – – 0.82*** (0.15)

0.72*** (0.21)

Test of Risk Hypotheses H7a:High risk dummy – 2.16*(1.17) 2.13*(1.14) 2.24*

(1.30) 2.15*(1.16) 2.16*(1.17) 2.19*(1.25)

H7b:Med. risk dummy – 0.17(0.24) 0.16(0.24) 0.13(0.26) 0.16(0.24) 0.17(0.24) 0.11(0.26)

Test of Reaction Hypotheses H8:Total attacks t – 1 – −0.004

(0.07) 0.001(0.06) 0.01(0.06) −0.004

(0.07) −0.004

(0.068) 0.01(0.06)

H9:Casualties t – 1 – 0.04*(0.02) 0.03*(0.02) 0.04** (0.02)

0.03*(0.02) 0.04*(0.02) 0.04** (0.02)

H10:Citizen ideology t – 1

– 0.001(0.01) 0.001(0.01) 0.001 (0.01)

0.001(0.007) 0.001(0.007) 0.001 (0.007)

Control Variables DV t – 1 0.93***

(0.02) 0.85***

(0.04) 0.85***

(0.04) 0.85***

(0.05) 0.85***

(0.04) 0.85***

(0.04) 0.85***

(0.05) (ln) Population t – 1 0.23**(0.10) 0.37***

(0.14) 0.38***

(0.14) 0.41**

(0.16) 0.38***

(0.14) 0.37***

(0.14) 0.42**

(0.17) Canada dummy 0.32(0.20) 0.23*(0.14) 0.23(0.14) 0.25(0.16) 0.23(0.14) 0.23*(0.14) 0.24(0.16) Mexico dummy 0.43(0.34) 0.78(0.48) 0.76(0.48) 0.81(0.50) 0.78(0.48) 0.78(0.48) 0.79(0.51) Ocean dummy 0.07(0.06) 0.22(0.13) 0.20*(0.12) 0.23*

(0.14) 0.22*(0.13) 0.22(0.13) 0.22*(0.13)

Percent white t – 1 −0.004* (0.002)

−0.0004 (0.004)

0.000 (0.004)

−0.00 (0.00)

−0.0004 (0.004)

−0.0004 (0.004)

0.0003 (0.004)

GDP per capita t – 1 0.56**(0.28) 0.41(0.29) 0.47(0.32) 0.37(0.30) 0.42(0.29) 0.41(0.29) 0.43(0.34) Constant −10.26**

(4.17) −9.29**

(4.46) −9.97**

(4.97) −9.38*

(4.83) −9.52**

(4.58) −9.29(4.46) −10.26

(5.42) N 780 696 696 696 696 696 696 R2 0.97 0.97 0.97 0.96 0.97 0.97 0.97

All models OLS regression estimations with year dummies. Robust standard errors in parentheses, clustered by US state. * p ≤ 0.1. ** p ≤ 0.05. *** p ≤ 0.01.

CRITICAL STUDIES ON TERRORISM 525

Multiplying the coefficient by 1 yields 0.31. Since the dependent variable is measured in millions of dollars, it means that, all else equal, a state will receive $310,000 more if its legislator is a member of the DHS committee chair than if it is not (since the value is 0 if a state does not feature a member). There is strong support for this relationship, as it remains significant and positive across four of the six different models.

There is also some support for hypothesis 4, as models 4 and 7 indicate that DHS committee members are able to independently secure $350,000 in CT funds for their states.7 There is no support for hypothesis 6, as the party leadership variable is consis- tently insignificant. Turning to partisanship effects, we see that member ideology has no effect. Majority party status is significant, though it runs opposite to hypothesis 3. It was anticipated that support for counterterrorist spending would increase with a GOP major- ity. However, the opposite is true, which falls in line with the more conventional con- servative stance on government spending (Jacoby 1994, 2000, 2008; Oldendick and Hendren 2018).

Much of the research cited here notes a conditional relationship between political dynamics and public spending. The next series of models test this by including interac- tions between electoral year, majority party control, and institutional effects. Previous

Table 4. Logistic regression of UASI eligibility, 1 = received funds. Model1 Model2 Model3 Model4 Model5 Model6

Test of Pork Hypotheses

H1:Election year t –

1

– −1.13(0.83) – – −1.16(0.87) 1.64*(0.89)

H2:House ideo. t – 1 – −0.93(1.02) – – −0.74(0.51) −0.91(0.63) H3:GOP majority t –

1

– 4.72***(0.83) – – 9.39***(1.67) 9.28*** (1.78)

H4:DHS mem. t – 1 – −0.93(1.02) – – −0.88(1.01) −0.58(0.78) H6:Party leader t – 1 – 1.39**(0.61) – – 1.28**(0.63) 1.90**

(0.94)

Test of Risk Hypotheses H7a:High risk

dummy – – −0.84(1.02) – −0.87(1.00) –

H7b:Med. risk dummy

– – −1.38*(0.83) – −1.07(0.88) –

Test of Reaction Hypotheses H8:Total attacks t –

1

– – – 0.02(0.11) 0.01(0.11) 0.05(0.09)

H9:Casualties t – 1 – – – 0.01(0.03) 0.01(0.03) −0.01(0.03) Control Variables DV t – 1 3.74***(0.62) 3.74***(0.60) 3.74***(0.62) 3.74***(0.61) 3.74***(0.60) 2.26***

(0.37) (ln)Population t – 1 3.98***(0.94) 4.04***(0.93) 4.27***(0.96) 3.97***(0.94) 4.26***(0.95) 8.31*(4.77) Canada dummy 0.92(0.67) 0.97(0.68) 1.08(0.89) 0.92(0.67) 1.21(0.89) – Mexico dummy −0.55(0.80) −0.48(0.81) −0.42(0.82) −0.58(0.81) −0.32(0.83) – Ocean dummy −0.01(0.52) −0.19(0.54) 0.56(0.74) −0.03(0.53) 0.29(0.78) – (ln)GDPpc t – 1 1.68(1.68) 1.49(1.76) 1.10(1.49) 1.70(1.68) 1.09(1.55) 2.64(3.28) Constant −76.27***

(21.61) −79.67***

(21.61) −73.96***

(20.63) −76.44***

(21.28) −78.52***

(21.31) –

N 1947 1947 1947 1947 1947 705 Wald X2 384.63*** 403.57*** 411.13*** 389.51*** 501.56*** 333.18*** Log likelihood −301.88 −297.88 −300.34 −301.79 −297.00 −135.87

All logit regressions include year dummies. Robust standard errors in parentheses, clustered by US state. H5 not included because no DHS subcommittee chairs present in data. * p ≤ 0.1. ** p ≤ 0.05. *** p ≤ 0.01.

526 K. T. KATTELMAN

works have noted that committee member power is contingent upon majority party status (Cox and McCubbins 1993: Kiewiet and McCubbins 1991). To test this, three interaction variables are created between Republican majority status and membership on the DHS subcommittee, chair position, and party leadership. Of these, there is robust support of a conditional relationship between a DHS seat and Republican majority control, in addition to independent DHS effects, as evidenced by models 4 and 7. Hence, there is evidence that institutional effects work in conjunction with partisan control, though it works in the opposite direction presented in H3. As model 4 shows, a Republican controlled legislature reduces a DHS member’s ability to garner funds for their state by $870,000.8 The effects of a cardinal seat appear more robust, with an individual’s ability to collect on average $30,000 across models 2–4.9 Additionally, the positive coefficient on the interaction term is in line with existing literature. The discre- pancy in direction between these two variables may be a product of the partisanship of the individual members. Of the DHS members, 40% are Democrats, a sizeable minority compared to 29% of Democratic cardinals. Thus, cardinals may be more successful at working with a Republican majority since the overwhelming majority of them are Republicans themselves.

Overall, however, the results show robust support for the independent effects of DHS and cardinal seats, irrespective of party control, supporting hypotheses 4 and 5. Finally, the effects of majority party control are also interacted with an election year, as previous studies have indicated that conservative members tend to lean hawkish on defence spending in election years (Askoy 2018). Models 6 and 7 show robust support for a Republican reversal of defence spending during election years, as indicated by the interaction terms. Hence, it appears that Republican controlled legislatures adhere to the fiscal conservative orthodoxy with regard to defence spending until election years.

The effects of risk are consistently significant across all models, supporting hypothesis 7. However, only states with a high risk receive additional funds, accepting on average an additional $2.17 million. Thus, it appears that DHS allocations are truly allocated, in part, by the degree of inherent risk. A robustness check includes two additional independent measures of risk. The first is a population measure weighted by density. The second is an aggregated Risk Management Solutions (RMS) model that calculates expected annual consequences from a diverse range of terrorist threats (Willis et al. 2005). As noted earlier, DHS outlays are a partial function of state population. However, these two measures account for location or the degree of clustering of groups of people. Unfortunately, much like the risk dummies, data are only available for 2004. Both variables show positive significance and render the institutional variables insignificant while election year, Republican majority control, and casualty count retain significance in the predicted direction. The results with these additional measures are presented in the online Appendix.

Finally, we see no support that either attacks or citizen ideology drive CT funding, though casualty counts show robust effects. Each individual casualty within a state generates an average increase of $36,000. This is no small effect and is in line with existing research on the knee-jerk reactions in the US to terrorist events (Mueller and Stewart 2012). Terrorism is a notoriously subjective label, and the theory outlined here is contingent upon whether the public perceives a violent act as terrorism. In the US, racial stereotypes influence the inherently subjective concept of terrorism (Whidden

CRITICAL STUDIES ON TERRORISM 527

2001; Aziz 2009; Elver 2012), with individuals more prone to label people perceived of as “Arab” or “Muslim” as terrorists (Corbin 2017, 458). To test for the possibility that the American public view certain acts as more salient than others, the data are disaggregated into a separate count of attacks from Islamist organisations or indivi- duals thought to be affiliated with Islamist organisations. Perpetrators groups were coded as Islamist according to Carson and Suppenbach (2018), and incidents in which there is tentative evidence of ties to Islamist groups were hand coded by checking the motive, incident summary, and cited sources within the GTD database. Overall, there were 43 separate Islamist attacks identified, yet they produce no significant effects.10

UASI eligibility

Table 4 presents logistic regression of whether a metropolitan district received UASI outlays. Model 1 presents base controls, which behave as expected. UASI outlays are dependent upon the population of the metro area and whether it received funding in the previous year. This is consistent with DHS (2009) reports that model risk off of a metropolitan area’s population. Model 2 lists pork findings, and it is here that two notable effects emerge. The first is that institutional effects perform similarly with state outlays. However, instead of cardinal and DHS seats it is a party leadership position that influences UASI eligibility. The second finding is a departure from the outlay models. Election year has no effect on eligibility, but a Republican controlled legislature has a negative effect. This finding lends support to hypothesis 3 while at the same time pointing to separate underlying processes that drive funding and the decision to fund. Whereas the effects of a Republican majority were conditional on election year, here they exhibit consistent positive effects.

Further analysis of partisanship effects finds that more conservative district members lead to a lower probability of receiving grants, which runs counter to Republican perfor- mance as a whole. Indeed, one of the defining features of this model is that institutional and partisan features play the lone role in the variation of metro area qualification, as model three shows that risk is not taken into consideration, nor are the reactionary measures of attacks or casualties. In fact, risk is significant in the opposite predicted direction. There is no theoretical reason to suspect that officials target lower risk states, but the findings could be an artefact of the pork process – the initial cities were the most at risk, but as more cities are added to the list of UASI recipients they are inevitably going to add cities that are less at risk. A cursory glance at the data supports this. A number of high-risk metropolitan areas (e.g. New York City, San Francisco, Chicago) received funds across all years of observation, while less risk states slowly accumulated, peaking in 2010, before dropping again.

The full model illustrates the robustness of partisan and institutional effects, though individual membership ideology fails to reach significance. Taken together, once popula- tion is controlled for, eligibility to receive UASI funds meant for counterterrorism defence functions largely as an exercise in pork and partisan politics. These results also translate over to within-observation effects, as the fixed-effects model 6 demonstrates. Interpretation of logit coefficients is not as straightforward as OLS, as they are representa- tions of log-odds. However, by running the coefficients through the logit model, pre- dicted probabilities for specific variables of interest can be obtained, in this case party leadership and Republican control of the legislature. Table 5 lists the predicted

528 K. T. KATTELMAN

probabilities from model 5, and the results are indeed substantive, with party leadership contributing to an 8% increase in the probability of UASI eligibility and Republican majority control contributing to a 7% bump.

As with the OLS models, several robustness checks are run. Islamist attacks do not have any significant effects on the results, and the RMS and density-weighted population measures also are negative and significant, which is consistent with the categorical risk measure in model 3. Taken as a whole, the findings on UASI eligibility show robust partisan influence at play, calling into question the purpose of counterterrorist funding to metropolitan areas. Furthermore, the effects of majority party differ between the two grant processes, suggesting a disconnect between the decision to fund and the amount of funding that warrants a closer look. A condensed version of the hypotheses and the support they receive is provided in Table 6.

Discussion

This exercise is considered a first-cut attempt at delineating the specific factors that influence public CT spending since 9/11. The data cover the entire period of the phase commonly known as the “Global War on Terror,” and include two measures meant to

Table 5. Predicted probabilities of receiving UASI funding conditional on GOP control and party leadership.

H6: Party Leader

No Yes

H3: GOP Majority No 59% 68% Yes 65% 74%

All predictions significant at *** p ≤ 0.01. All other variables held at their mean values

Table 6. Summary of results.

Predicted finding Support for

Increased funding? Support for UASI

eligibility?

Pork

H1: CT funding to states should increase during a presidential election year. Yes No H2: CT funding to a constituency should increase if its congressperson is

conservative. No No

H3: CT funding to a constituency should increase if the legislature is Republican controlled.

No Yes

H4: CT funding to a constituency should increase if its congressperson is a member of the DHS appropriations subcommittee.

Yes No

H5: CT funding to a constituency should increase if its congressperson is a chair of the DHS appropriations subcommittee.

Yes –

H6: CT funding to a constituency should increase if its congressperson is a party leader.

Yes Yes

Risk H7: CT funding to a constituency should increase with level of risk. Yes No Reaction H8: CT funding to a constituency should increase with the frequency of

terrorist attacks it suffers. No No

H9: CT funding to a constituency should increase with the severity of terrorist attacks it suffers.

Yes No

H10: Constituencies with a more conservative citizen ideology should receive more CT funding.

No –

CRITICAL STUDIES ON TERRORISM 529

capture both the degree of CT spending as well as its scope. The collections of indicators are meant to capture the three most commonly cited reasons behind CT variation and expansion.

The results of this study have several implications for future research on the dynamics of CT spending in the US. The first and most obvious is that, contrary to the few existing studies, spending on the prevention of terrorism in the US is as much a product of politics as any other spending decision. Committee chairs, election cycles, and party control all work to alter the dynamics of funding flows to states as well as district eligibility. Perhaps that should not been seen as a surprise, yet it justifies the anecdotal evidence of over- spending and pushes back on the narrative that risk is the primary driver of national security spending. Most notably in the case of UASI eligibility, risk plays a negative role in the decision to grant funds to metropolitan areas.

Second, the results here suggest that risk plays a large role in grant allocation, with a small handful of the most at-risk states receiving the largest shares of funding. The time variant controls of population and GDP per capita are traditionally thought of as rough approxima- tions of risk, and there were robust effects of population increases on both funds and eligibility. This is to be expected, since the DHS explicitly relied on population as a metric of distribution. However, the other common approximation of risk, GDP per capita, exerted sporadic influence among state outlays and no influence in USAI eligibility. A benefit of including the AIR risk measure in regression models is that it allows for direct comparison to the common proxy measures of population and GDP, which provides us with more valid evidence that the DHS does employ some degree of risk assessment in funding allocation.

Third, it is important to mention the effect that the human toll has on CT spending. The analysis here makes clear that individual attacks do not matter as much as the actual degree of destruction that they render. The US has a low tolerance of death at the hands of terrorists, and a close interpretation of the data bears this out. This finding comple- ments previous works that detail the vast degree of overspending and suggest that casualty-induced spending measures intend to sooth a psychological gap rather than correct a lapse in an already over secure nation.

As previously stated, the research presented here is a first cut at a very small subset of federal CT spending. The theoretical logic rested primarily on a top-down approach. That is, spending decisions were a function of actions at the federal level. State and local demand were not captured in these models beyond simple demographic and partisan measures. One promising method of measuring local dependence and demand on defence spending involves examining subcontracts. As of now, however, data at this level remain limited in scope, yet patterns of local economic dependence on defence spending using this method exist (Thorpe 2010, 2014). Further research into the HSGP should employ data at the subcontract level to expand on the findings here, as it remains unclear how any of this grant money is actually used.

Lastly, it should be stressed that statistical models cannot provide definitive answers to the specific processes that drive CT spending to the states. At best, they infer that key political positions and party control play a role in HSGP distribution. While their advantage is that they show these effects as part of a broader pattern, detail is lost on how these processes actually play out. Future research should also focus on qualitative, deep-dives into the DHS sub- committee, for example, to uncover specifically how this relationship drives spending.

530 K. T. KATTELMAN

Notes

1. Gimpel, Lee, and Thorpe (2009) find evidence of this with the American Recovery and Reinvestment Act of 2009, which was passed in response to the exogenous shock of the financial crisis of 2008.

2. In particular, Representative Harold “Hal” Rogers (R-KY) has brought over $200 million in DHS grants back to his state in the three years since the panel was created, which is approximate to the amount New York City received in 2005 (Kaplan 2006).

3. For example, the Law Enforcement Terrorism Prevention Programme (LETPP) was a separate line item until 2009, when it was determined that SHSGP and UASI will dedicate 25% of their funds to its continuance (US Department of Homeland Security FY 2009 Overview). Other programmes, such as the Metropolitan Medical Response System (MMRS) and the Citizen Corps Program (CCP), discontinued in 2012.

4. Specifically, the SHSGP and UASI (US Government Accountability Office 2008). 5. For example, state and local investments in counterterrorism, dual-use programs and spend-

ing, private donations, classified spending, and economic losses and secondary effects stemming from the long-term cost of homeland security.

6. There are also additional filtering criteria that allow for customisation of the definition – 1. Act must be aimed at attaining a political economic, religious, or social goal. 2. There must be evidence of intention to coerce or convey a message to a larger audience. 3. The action must be outside of the context of legitimate warfare activities. All three criteria were selected to obtain the data for this study.

7. This value is simply the result of multiplying the DHS committee membership coefficient of 0.35 on the value of a state having a member of the DHS committee, which is 1. Since the dependent variable is in millions of dollars, the coefficient is multiplied by 1,000,000 to generate the predicted dollar amount.

8. Interpreting interaction effects must take into account the separate coefficients as well as their interactions. Hence, we add the coefficient for GOP majority (−0.59) with DHS member- ship (0.35) with the interaction term between the two (−0.63).

9. This is compiled by averaging the three Cardinal coefficients (i.e. 0.31, 0.39, and 0.27). 10. Results presented in the online Appendix.

Data availability statement

Data available at https://kylekattelman.wixsite.com/kylekattelman

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes on contributor

Kyle Kattelman is associate professor of political science at Fairleigh Dickinson University.

References

2016. “AIR Worldwide Expands Its Terrorism Model Globally.” AIR Worldwide. Press Release. https:// www.air-worldwide.com/news-and-events/press-releases/AIR-Worldwide-Expands-Its-Terrorism- Model-Globally/

Aldrich, J. H., and D. W. Rohde. 2000. “The Republican Revolution and the House Appropriations Committee.” The Journal of Politics 62 (1): 1–33. doi:10.1111/0022-3816.00001.

CRITICAL STUDIES ON TERRORISM 531

Aldrich, J. H., and D. W. Rohde. 2001. “The Logic of Conditional Party Government: Revisiting the Electoral Connection.” In Congress Reconsidered, edited by L. Dodd and B. Oppenheimer, 269–292. 7th ed. Washington, DC: Congressional Quarterly Press.

Allen, J. 2007. “The Earmark Game: Manifest Disparity.” CQ Weekly, October 1, 2836–2850. Askoy, D. 2018. “Electoral and Partisan Cycles in Counterterrorism.” Journal of Politics 80 (4):

1239–1253. doi:10.1086/698847. Aziz, S. F. 2009. “Sticks and Stones, the Words that Hurt: Entrenched Stereotypes Eight Years after 9/

11.” New York City Law Review 13: 33–72. Balla, S. J., E. D. Lawrence, F. Maltzman, and L. Sigelman. 2002. “Partisanship, Blame Avoidance, and

the Distribution of Legislative Pork.” American Journal of Political Science 46 (3): 515–525. doi:10.2307/3088396.

Berry, C. R., and A. Fowler. 2016. “Cardinals or Clerics? Congressional Committees and the Distribution of Pork.” American Journal of Political Science 60 (3): 692–708. doi:10.1111/ ajps.12192.

Berry, W. D., E. J. Ringquist, R. C. Fording, and R. L. Hanson. 1998. “Measuring Citizen and Government Ideology in the American States, 1960–93.” American Journal of Political Science 42 (1): 327–348. doi:10.2307/2991759.

Binder, S. 2008. “Lend Me Your Earmarks.” The Monkey Cage, February 20. https://themonkeycage. org/2008/02/lend_me_your_earmarks_1/

Bloom, P., and C. Veres. 1999. “The Perceived Intentionality of Groups.” Cognition 71 (1): 1–9. doi:10.1016/S0010-0277(99)00014-1.

Boyle, S., S. Malinowski, and M. Gannon. 2002. “AIR Worldwide Releases Terrorism Loss Estimation Model to Analyze Financial Impact of Terrorist Attacks on Insurers, Reinsurers, and Corporations.” AIR Worldwide. Press Release. Accessed 13 April 2021. https://www.verisk.com/archived/2002/air- worldwide-releases-terrorism-loss-estimation-model-to-analyze-financial-impact-of-terrorist-atta/

Budge, I., and R. Hofferbert. 1990. “Mandates and Policy Outputs: US Party Platforms and Federal Expenditures.” American Political Science Review 84 (1): 111–131. doi:10.2307/1963632.

Bueno de Mesquita, E. 2007. “Politics and the Suboptimal Provision of Counterterror.” International Organization 61 (1): 9–36. doi:10.1017/S0020818307070087.

Campbell, A., P. E. Converse, W. E. Miller, and D. E. Stokes. 1960. The American Voter. Chicago, IL: University of Chicago Press.

Carson, J. V., and M. Suppenbach. 2018. “The Global Jihadist Movement: The Most Lethal Ideology?” Homicide Studies 22 (1): 8–44. doi:10.1177/1088767917733783.

Coats, R. M., G. Karahan, and R. D. Tollison. 2006. “Terrorism and Pork-Barrel Spending.” Public Choice 128 (1/2): 275–287. doi:10.1007/s11127-006-9054-8.

Corbin, C. M. 2017. “Terrorists are Always Muslim but Never White: At the Intersection of Critical Race Theory and Propaganda.” Fordham Law Review 86 (2): 455–485.

Cox, G. W., and M. D. McCubbins. 1993. Legislative Leviathan: Party Government in the House. Los Angeles, CA: University of California Press.

Cox, G. W., and M. D. McCubbins. 2005. Setting the Agenda: Responsible Party Government in the US House of Representatives. New York: Cambridge University Press.

Crenshaw, M. 2001. “Counterterrorism Policy and the Political Process.” Studies in Conflict & Terrorism 24 (5): 329–337. doi:10.1080/105761001750434204.

Davis, D. W., and B. D. Silver. 2004. “Civil Liberties Vs. Security: Public Opinion in the Context of the Terrorist Attacks on America.” American Journal of Political Science 48 (1): 28–46. doi:10.1111/ j.0092-5853.2004.00054.x.

de Rugy, V. 2005. “What Does Homeland Security Spending Buy?” American Enterprise Institute. Working paper #107.

Department of Homeland Security. 2020. “DHS Announces Grant Allocations for Fiscal Year 2020 Preparedness Grants.” https://www.dhs.gov/news/2020/06/30/dhs-announces-grant-allocations- fiscal-year-2020-preparedness-grants

Department of Homeland Security Appropriations Act, 2006. 2005. H.R. 2360, 109th Cong. 109–90. Earle, G. 2004 “Homeland Security Funding Part I — Money Not Flowing to the Places in Danger.„

The Hill, April 15. http://archive.newsmax.com/archives/articles/2004/4/14/163610.shtml

532 K. T. KATTELMAN

r, Hilal.2012 “Racializing Islam Before and After 9/11: From Melting Pot to Islamophobia.” Transnational and Contemporary Problems 21: 119-172.

European Parliament Think Tank. 2016. “Counter-Terrorism Funding in the EU Budget.” https:// www.europarl.europa.eu/thinktank/en/document.html?reference=EPRS_BRI(2016)580904

Evans, D. 1994. “Policy and Pork: The Use of Pork Barrel Projects to Build Policy Coalitions in the House of Representatives.” American Journal of Political Science 38 (4): 894–917. doi:10.2307/ 2111726.

Federal Emergency Management Agency. 2021. “Homeland Security Grants Programme.” https:// www.fema.gov/grants/preparedness/homeland-security

Fordham, B. 1998. “Partisanship, Macroeconomic Policy and the US Use of Force, 1949–1994.” Journal of Conflict Resolution 42 (4): 418–439. doi:10.1177/0022002798042004002.

Friedman, B. 2005. “Think Again: Homeland Security.” Foreign Policy July-August: 22–25. Friedman, B. 2011. “Managing Fear: The Politics of Homeland Security.” Political Science Quarterly

126 (1): 77–106. doi:10.1002/j.1538-165X.2011.tb00695.x. Gimpel, J. G., F. E. Lee, and R. U. Thorpe. 2009. “Geographic Distribution of the Federal Stimulus of

2009.” Political Science Quarterly 127 (4): 567–595. doi:10.1002/j.1538-165X.2012.tb01128.x. Global Terrorism Database. 2017. Codebook: Inclusion Criteria and Variables. College Park, MD:

University of Maryland. Global Terrorism Database. 2021. (National Consortium for the Study of Terrorism and Responses to

Terrorism [START]); Accessed 13 April 2021. https://www.start.umd.edu/gtd Haner, M., M. M. Sloan, and F. T. Cullen. 2019. “Public Concern about Terrorism: Fear, Worry, and

Support for Anti-Muslim Policies.” Socius: Sociological Research for a Dynamic World 5. doi:10.1177/2378023119856825.

Hess, G., and A. Orphanides. 1995. “War Politics: An Economic, Rational Voter Framework.” American Economic Review 85 (4): 828–847.

Holsti, O. 2004. Public Opinion and American Foreign Policy. Revised Edition. Ann Arbor, MI: University of Michigan Press.

Holsti, O., and J. Rosenau. 1988. “The Domestic and Foreign Policy Beliefs of American Leaders.” Journal of Conflict Resolution 32 (2): 248–294. doi:10.1177/0022002788032002002.

Homeland Security Act of 2002. 2002. H.R. 5005, 107th Cong. 107–296. Immenkamp, B., G. Sgueo, S. Voronova, and A. Dobreva. 2020. “The Fight Against Terrorism.”

European Parliament. https://www.europarl.europa.eu/RegData/etudes/BRIE/2019/635561/ EPRS_BRI(2019)635561_EN.pdf

Jacoby, W. G. 1994. “Public Attitudes toward Government Spending.” American Journal of Political Science 38 (2): 336–361. doi:10.2307/2111407.

Jacoby, W. G. 2000. “Issue Framing and Public Opinion on Government Spending.” American Journal of Political Science 44 (4): 750–767. doi:10.2307/2669279.

Jacoby, W. G. 2008. “Comment: The Dimensionality of Public Attitudes toward Government Spending.” American Journal of Political Science 61 (1): 158–161.

Jervis, R. 1976. Perception and Misperception in International Politics. Princeton, NJ: Princeton University Press.

Jordan, J., M. E. Kosal, and L. Rubin. 2016. “The Strategic Illogic of Counterterrorism Policy.” The Washington Quarterly 39 (4): 181–192. doi:10.1080/0163660X.2016.1261564.

Kaplan, F. 2006. “Hog Heaven.” Slate, January 4. https://slate.com/news-and-politics/2006/01/the- homeland-security-budget-is-still-full-of-pork.html

Kattelman, K. T. 2016. “Party Structure, Information, and Coalition Durability: The Relevance of NATO in the Global War on Terror.” Studies in Conflict and Terrorism 39 (9): 803–818. doi:10.1080/ 1057610X.2015.1131030.

Kiewiet, R., and M. Mathew. 1991. The Logic of Delegation: Congressional Parties and Appropriations Process. Chicago, IL: University of Chicago Press.

Lazarus, J. 2009. “Party, Electoral Vulnerability, and Earmarks in the US House of Representatives.” Journal of Politics 71 (3): 1050–1061. doi:10.1017/S0022381609090872.

CRITICAL STUDIES ON TERRORISM 533

Lazarus, J. 2010. “Giving the People What They Want? The Distribution of Earmarks in the US House of Representatives.” American Journal of Political Science 54 (2): 338–353. doi:10.1111/j.1540- 5907.2010.00434.x.

Lee, F. E. 2003. “Geographic Politics in the US House of Representatives: Coalition Building and Distribution of Benefits.” American Journal of Political Science 47 (4): 714–728. doi:10.1111/1540- 5907.00050.

Lewis, J. B., K. Poole, H. Rosenthal, A. Boche, A. Rudkin, and L. Sonnet. 2021. Voteview: Congressional Roll-Call Votes Database. Accessed 13 April 2021. https://voteview.com/

Lipton, E. 2004. “Big Cities Will Get More in Antiterrorism Grants.” The New York Times, December 22. https://www.nytimes.com/2004/12/22/politics/big-cities-will-get-more-in-antiterrorism-grants. html

Lodge, M., and R. Hamill. 1986. “A Partisan Schema for Political Information Processing.” American Political Science Review 80 (2): 505–520. doi:10.2307/1958271.

Lonardo, L. D. 2019. “The Partisan Politics of Counterterrorism: Reputations, Policy Transparency, and Electoral Outcomes.” Political Science Research and Methods 7 (2): 253–269. doi:10.1017/ psrm.2017.19.

MacDonald, J. A., and R. J. McGrath. 2016. “Retrospective Congressional Oversight and the Dynamics of Legislative Influence over the Bureaucracy.” Legislative Studies Quarterly 41 (4): 899–934. doi:10.1111/lsq.12138.

Marinov, N., W. G. Nomikos, and J. Robbins. 2015. “Does Electoral Proximity Affect Security Policy?” Journal of Politics 77 (3): 762–773. doi:10.1086/681240.

Miller, G. D. 2007. “Confronting Terrorisms: Group Motivation and Successful State Policies.” Terrorism and Political Violence 19 (3): 331–350. doi:10.1080/09546550701424059.

Mueller, J., and M. G. Stewart. 2012 “The Terrorism Delusion: America's Overwrought Response to September 11.” International Security 38 (1): 81-110.

Nanes, M. 2017. “Political Violence Cycles: Electoral Incentives and the Provision of Counterterrorism.” Comparative Political Studies 50 (2): 171–199. doi:10.1177/ 0010414016635198.

Newman, R. 2021. “Why Congress Is Bringing Back ‘Pork.’” Yahoo!Finance, March 16. https://finance. yahoo.com/news/why-congress-is-bringing-back-pork-184519151.html

Office of the Inspector General. 2005. “Review of the Port Security Grant Programme.” Department of Homeland Security. https://www.oig.dhs.gov/sites/default/files/assets/Mgmt/OIG_05-10_ Jan05.pdf

Oldendick, R. W., and L. B. Hendren. 2018. “Party Differences in Support for Government Spending, 1973–2014.” American Politics Research 46 (3): 499–533. doi:10.1177/1532673X17719718.

Patel, T. G. 2017. “It’s Not about Security, It’s about Racism: Counter-Terror Strategies, Civilizing Processes and the Post-Race Fiction.” Palgrave Communications 3 (1). doi:10.1057/ palcomms.2017.31.

Perliger, A. 2012. “How Democracies Respond to Terrorism: Regime Characteristics, Symbolic Power and Counterterrorism.” Security Studies 21 (3): 490–528. doi:10.1080/09636412.2012.706505.

Pew. 2016a. “15 Years after 9/11, a Sharp Partisan Divide on Ability of Terrorists to Strike U.S.” https:// www.pewresearch.org/politics/2016/09/07/15-years-after-911-a-sharp-partisan-divide-on-ability -of-terrorists-to-strike-u-s/

Pew 2016b. “How Closely are State Economies Tied to Spending?” https://www.pewtrusts.org/en/ research-and-analysis/articles/2016/03/22/how-closely-are-state-economies-tied-to-federal- spending

Pew. 2017. “With Budget Debate Looming, Growing Share of Public Prefers Bigger Government.” https://www.pewresearch.org/politics/2017/04/24/with-budget-debate-looming-growing-share- of-public-prefers-bigger-government/#where-partisans-differ-and-mostly-agree-on-federal- spending

Piazza, J. A. 2020. “White Demographic Anxiety and Support for Torture of Terrorism Suspects.” Studies in Conflict & Terrorism 1–29. doi:10.1080/1057610x.2020.1792733.

Prante, T., and A. K. Bohara. 2008. “What Determines Homeland Security Spending? An Econometric Analysis of the Homeland Security Grant Programme.” The Policy Studies Journal 36 (2): 243–256.

534 K. T. KATTELMAN

Rees, W., and R. J. Aldrich. 2009. “Contending Cultures of Counterterrorism: Transatlantic Divergence or Convergence?” International Affairs 81 (5): 905–923. doi:10.1111/j.1468-2346.2005.00494.x.

Reese, S. 2004. “Fiscal Year 2005 Homeland Security Grant Programme: State Allocations and Issues for Congressional Oversight.” CRS Report for Congress, RL32696.

Ripley, A. 2004. “How We Got Homeland Security Wrong.” Time Magazine, March 22. http://content. time.com/time/subscriber/article/0,33009,993684,00.html

Roberts, Ed. 2020. “House Panel Seeks Restoration of Homeland Security Grants.” Homeland Preparedness News. https://homelandprepnews.com/countermeasures/42329-house-panel- seeks-restoration-of-homeland-security-grants/?amp

Rotella, S. 2021. “Domestic Terrorism: A More Urgent Threat, But Weaker Laws.” Propublica, January 10. https://ctmirror.org/2021/01/10/domestic-terrorism-a-more-urgent-threat-but-weaker-laws/

Russett, B. 1990. “Doves, Hawks and US Public Opinion.” Political Science Quarterly 105 (4): 515–539. doi:10.2307/2150933.

Schneider, C. J. 2013. “Globalizing Electoral Politics: Political Competence and Distributional Bargaining in the European Union.” World Politics 65 (3): 452–490. doi:10.1017/ S0043887113000129.

Smith, A. 1996. “Diversionary Foreign Policy in Democratic Systems.” International Studies Quarterly 40 (1): 133–153. doi:10.2307/2600934.

Sunstein, C. R., and R. J. Zeckhauser. 2008. “Overreaction to Fearsome Risks.” Harvard Law School Programme on Risk Regulation Research Paper No. 08–17.

Thorpe, R. U. 2010. “The Role of Economic Reliance in Defense Procurement Contracting.” American Politics Research 38 (4): 636–675. doi:10.1177/1532673X09337107.

Thorpe, R. U. 2014. The American Warfare State: The Domestic Politics of Military Spending. Chicago, IL: University of Chicago Press.

US Census Bureau. 2020. “Decennial Census Datasets.” https://www.census.gov/programs-surveys /decennial-census/data/datasets.html

US Department of Homeland Security. 2005. “National and Federal Laboratories and Research Centers.” https://www.dhs.gov/science-and-technology/national-federal-laboratories-research- centers

US Department of Homeland Security. 2009. Fiscal Year 2004 Homeland Security Grant Programme. https://www.dhs.gov/xlibrary/assets/grant-program-overview-fy2009.pdf

US Department of State. 2020. Country Reports on Terrorism. Washington, DC: US Department of State.

US Department of State. 2002. Patterns of Global Terrorism 2001. Washington, DC: U.S. Department of State.

US Government Accountability Office. 2008. “DHS Risk-Based Grant Methodology Is Reasonable, but Current Version’s Measure of Vulnerability Is Limited.” https://www.gao.gov/assets/gao-08-852. pdf

Whidden, M. J. 2001. Unequal Justice: Arabs in American and the United States Antiterrorism Legislation, 69 Fordham L. Review 2825, 2849.

Willis, H. H., A. R. Morral, T. K. Kelly, and J. J. Medby. 2005. Estimating Terrorism Risk. RAND: Center for Terrorism Risk Management Policy.

CRITICAL STUDIES ON TERRORISM 535