What Determines Homeland Security Spending? an Econometric Analysis of the Homeland Security Grant Program
Prante, Tyler, Bohara, Alok K., Policy Studies Journal
The Department of Homeland Security (DHS) annually disburses grants to each U.S. state. The distribution of these grants has been attacked widely. DHS grants are being distributed as "pork," say critics, alleging that politicians fighting for their respective state's share of DHS funds has resulted in a disproportionate amount of funding going to states with relatively small populations and relatively little risk of terrorist attack (e.g., de Rugy, 2006; O'Beirne, 2003). Given the serious nature of terrorism and the obvious consequences of misallocating grant funding, the legitimacy of this assertion is significant. However, claims of politics driving DHS spending have relied largely on anecdotal and descriptive evidence.
We attempt to provide a more rigorous examination of the issue. Using statistical revealed preference analysis, we address a series of simple but important questions: (i) Is the distribution of DHS grants determined by risk of terrorist attack? (ii) Does the party affiliation of a state's elected officials drive DHS grant funding? (iii) Does the Congressional influence of a state's elected officials determine funding? and (iv) Has recent legislation introduced to change the grant disbursement procedure had the desired effect of increasing the importance of risk in determining the funding distribution? By comparing the varying grant disbursements, the relative risk of terrorist attack, and the political backdrop in each state, our objective is to identify what factors determine DHS grant outcomes.
The Problem: Are Politics Trumping Terrorism Risk?
Made clear by the attacks of September 11, 2001, the landscape surrounding terrorism prevention and policy has undergone a sea change in recent years. As one response, the U.S. Congress passed the Patriot Act of 2001. The Patriot Act reorganized government dramatically by bringing a myriad of federal agencies and programs (e.g., the Federal Emergency Management Agency, the Office of Domestic Preparedness, and the Nuclear Incident Response Team) under the domain of the DHS while also creating several new agencies (DHS, 2006a). Though the list of responsibilities left to the DHS is lengthy, department objectives have been summarized as follows: Preventing terrorist attacks, reducing America's vulnerability to terrorism, and minimizing the damage should an attack occur (DHS, 2004a).
A primary instrument the DHS uses to address these objectives is the Homeland Security Grant Program (HSGP; DHS, 2007). Through a number of individual grant programs, the HSGP distributes money annually to U.S. states and territories. These grants are significant in scale; over 1.6 billion dollars were disbursed in 2006 (DHS, 2007). Individual grant programs under the HSGP have particular eligibility requirements and objectives; however, they share the core mission of the DHS: "to enhance the ability of State, local and tribal governments to prepare, prevent, and respond to terrorist attacks and other disasters" (DHS, 2007).
One fallout of the changed landscape is that national security has emerged as a prominent issue in the national consciousness (Roberts, 2006). A hailstorm of criticism has surrounded the distribution of Homeland Security grants (e.g., de Rugy, 2005, 2006; Earle, 2004; McLaughlin, 2002; Roberts, 2005; Wisloski & Connor, 2006). Sparsely populated states have received more per capita grant dollars than their more crowded counterparts, and this has been the precipice for much of the controversy surrounding the department (Roberts, 2005).
Critics allege DHS funds are being spent trivially in rural areas and that grants awarded to higher profile areas can potentially provide a larger benefit. A series of outlandish examples, including surveillance cameras in a small Alaskan fishing village; kennels for stray animals in Modoc County, California; gym memberships; and a 14-mile bridge in Mobile, Alabama, are cited as evidence of wasteful spending (de Rugy, 2006; Wisloski & Connor, 2006). Perhaps not surprisingly, representatives from less populous states counter that the natural resources provided in these states are also at risk of terrorist attack, and that rural areas require protection as well (Earle, 2004). (1)
Beyond the assertion of "wasteful" spending, it is alleged that politicians, fighting for monies on behalf of their, constituents has superseded the objective of preventing terrorist attacks and that DHS grants are being distributed as "pork" (de Rugy, 2005, 2006; Earle, 2004; McLaughlin, 2002; Roberts, 2005; Wisloski & Connor, 2006). The claim of political considerations driving funding has been forwarded not only in popular media but also by some members of Congress. (2)
The procedure itself by which grants are distributed has also come under scrutiny. Prior to 2006, many homeland security grants were distributed by formula. Under this initial disbursement process, each state received a base level of funding (0.75 percent of the total distribution for a given grant program), with the remaining funds being distributed proportionately to a state's population (DHS, 2004b, 2005). This process was used to allocate funds for three of the grant programs analyzed in this paper: the State Homeland Security Grant Program (SHGP), the Law Enforcement Terrorism Prevention Program (LETPP), and the Citizens Corps Program (CCP). A fourth program, the Urban Areas Security Initiative (UASI), distributes funds based on perceived risk (DHS, 2004c). This initial disbursement procedure proved to be undesirable. Using a population-based stencil and not risk analysis as the mechanism to allocate DHS grants resulted with what has been called an inefficient distribution of funds (Earle, 2004). In reference to the grant distribution process, Congressman Chris Cox (R-California) argues, "This should be all about national security and less about pork and politics. Right now, we're using funding formulas that look more like a highway bill" (Earle, 2004).
The Department of Homeland Security Appropriation Act of 2006 fundamentally changed the procedure by which two HSGP grants are distributed. While still awarding each state a base of 0.75 percent of the total distribution for a respective program, the remaining SHGP and LETPP funds are now allocated based on risk (not on population as before; DHS, 2006b). The disbursement procedure for the UASI and CCP did not change. In summary, the legislation has had the effect of allowing the DHS more discretion over how to distribute funds in 2006. Table 1 describes disbursement procedure for the SHGP, LETPP, CCP, and UASI for 2004, 2005, and 2006.
While the assertions of malfeasance are made widely, the department has been subject of little empirical research. Given the obvious importance of the issue, it is worthwhile to ask whether the claim of politics determining the distribution of grants is statistically supported. In addition, although the Department of Homeland Security Appropriation Act of 2006 nominally changed the process by which grants are distributed, it is unknown if the Act has had the desired effect of increasing the importance of terrorism risk in allocating grants. The objective of this paper is to statistically analyze the funding pattern of the DHS from 2004 to 2006 to address these questions.
Review of Related Research
Given the short existence of the department, the body of research specifically directed toward the DHS is limited. Exceptions exist: Providing background, Roberts (2005) and Clarke and Chenoweth (2006) discuss the complexities surrounding the distribution of HSGP grants. Sunstein (2003) and Fischhoff, Gonzalez, Small, and Lerner (2003) identify the trouble people have accurately assessing and processing terrorism risk. Caudle (2005) notes the difficulty of measuring the successes and failings of homeland security policy.
More germane to this paper, however, is a line of research that investigates the determinants of national government grants to states. Notably by Anderson and Tollison (1991) and Wallis (1996), researchers have often posed a variation on the question we investigate here: Is it political, economic, or other factors that determine a state's allocation under a federal grant program? Examples of such studies include Wright (1974), Levitt and Snyder (1995), Hoover and Pecorino (2003), and Ansolabehere and Snyder (2006). In our attempt to tease from the observed actions of the DHS the underlying decision-making framework used by the department, our work can more broadly be thought of as an application of revealed preference analysis, with roots to McFadden (1975).
Recently, Roberts (2005) has called for increased focus on the distribution of DHS grants. Coats, Karahan, and Tollison (2006) have started the process, analyzing the distribution of DHS grants for 2004. Coats et al. find that political variables are not a statistically significant determinant of funding while proxies for terrorism risk provide mixed results. Our work is intended to build on this study. Given that we analyze the program after Coats et al., we are able to extend the analysis in several ways: (i) the available data set has been expanded in that three years of funding decisions have now taken place; (ii) the Department of Homeland Security Appropriation Act of 2006 has been passed, allowing for study of the effect of the legislation on the funding behavior of the department; and (iii) we use different measures of terrorism risk and political influence.
Empirical Model and Data
Using data from 2004, 2005, and 2006, we analyze funding decisions of the DHS for four grant programs: SHGP, LETPP, CCP, and UASI. (3) Theory on intergovernmental grant behavior predicts that disbursements are determined by government attempting to maximize social welfare, by elected officials looking after their own interests, or by some combination of these goals (Becker, 1983; Borck & Owings, 2003; Oates, 1999). Following this convention, we offer a model of DHS grant funding decisions from which the "pork hypothesis" and "risk hypothesis" may be cleanly evaluated. We propose that the amount of funding awarded to each state i can be characterized by the following model:
[FUNDING.sub.i] = f([RISK.sub.i], [POLITICS.sub.i], [POWER.sub.i]) (1)
where [RISK.sub.i] is a vector of one or more variables measuring a state's risk of terrorist attack, [POLITICS.sub.i] is a vector of one or more variables measuring the party affiliation of a state's elected officials, and [POWER.sub.i] is a vector of one or more variables measuring the connectedness or influence of a state's elected officials within Congress. Table 2 provides variable definitions and descriptive statistics. The model is estimated with ordinary least squares (OLS) using robust standard errors to account for heteroskedasticity.
As noted earlier, the allocation for various DHS grant programs is decided with different processes. The allocation under some programs is determined by formula; others are determined by perceived risk and need. It is unclear then which disbursement should be modeled: the total amount of funding a state receives or the funding left to the discretion of the DHS. We therefore model the funding a state receives using a set of dependent variables. Our first dependent variable, TOTAL, is the sum of the funding awarded to a state from the SHGP, LETPP, CCP, and UASI programs. (4) DISCRETIONARY is defined as the DHS grant funding a state receives that is not determined by formula (for 2004 and 2005, UASI grants; for 2006, UASI grants + SHGP + LETPP - 0.75 percent of the total disbursement for SHGP and LETPP). NON-DISCRETIONARY is defined as the funding a state receives that is determined by formula. DISC-SHARE is the proportion of total funding a state receives that is left to the discretion of the DHS.
A major difficulty in modeling the funding process of the DHS is finding a suitable measure of terrorism risk. While much of the data we use are empirical, assessment of terrorism risk is inherently subjective. Further, while the DHS has attempted to measure terrorism risk by area of the country, these assessments are sensitive in nature and not publicly available. Therefore, any measure we use for terrorism risk could be subject to the critique of being imprecise in that the informational advantages of the DHS make their internal assessment of risk more accurate than any we would use. With a nod to the difficulty associated with doing so, we implement risk with a set of indicator variables classifying each state into one of three categories of terrorism risk. Ranging from the highest risk of terrorist attack to the lowest, these indicator variables are denoted as HIGH-RISK, MED-RISK, and LOW-RISK. Risk categorizations are taken from Applied Insurance Research Terrorism Loss Estimation Model. AIR is a firm specializing in modeling catastrophic risk for insurers (Boyle, Malinowski, & Gannon, 2002). Convening terrorism experts and former employees of the FBI, CIA, and Department of Defense, AIR's model is an application of the Delphi method. (5) In the course of expert deliberations, the importance of infrastructure, tourist attractions, and high profile targets are included in AIR's estimates. (Boyle et al., 2002). Each state's categorization is presented in Ripley (2004).
Political factors having undue influence on funding outcomes could result from two distinct happenings: (i) partisan politics are determining funding, or (ii) the influence of a state's elected officials, irrespective of party affiliation, is determining funding. We therefore separate the two when addressing the "pork hypothesis." The [POLITICS.sub.i] vector is composed of a constructed index variable, BLUE-RED-INDEX, measuring the party affiliation of the elected officials in a state. BLUE-RED-INDEX is the sum of four variables: BUSH-2000 and BUSH-2004 are indicator variables coded as 1 if a state was carried by George W. Bush in the 2000 and 2004 presidential elections, respectively. (6) R-SENATE is the proportion of a state's senators who are Republican and R-HOUSE is the proportion of a state's representatives who are Republican. (7) BUSH-2000, BUSH-2004, R-SENATE, and R-HOUSE are summed to create BLUE-RED-INDEX. Combining these variables into an index allows for a wide measure of the party affiliation of a state's elected officials while avoiding potential problems associated with multicollinearity. Given that each political variable ranges from 0 to 1, BLUE-RED-INDEX would take a value of 4, as an example, if a state had voted for President Bush in the 2000 and 2004 elections and had elected all Republicans to Congress.
The vector [POWER.sub.i] is also implemented using an index variable. Following the example of Wallis (1996), we use leadership positions within Congress as a proxy for political sway. The variable INFLUENCE-INDEX is the sum of a series of indicator variables, each coded as 1 if a state's elected official is: Speaker of the House, House Majority or Minority Leader, Senate Majority or Minority Leader, House Majority or Minority Whip, Senate Majority or Minority Whip, a member of the House DHS committee, or the Chair or ranking minority member of any House or Senate Committee. (8)
Control variables are also included in our model. POP is a state's population. (9) MEXICO is an indicator variable coded as 1 if a state borders Mexico, 0 otherwise. CANADA is an indicator variable coded as 1 if a state borders Canada, 0 otherwise. OCEAN is an indicator variable coded as 1 if a state is not land locked, 0 otherwise. The variable WHITE measures the percentage of a state's residents racially categorized as white in the 2000 census. INCOME is the per capita income in a state as indicated in the 2000 census (U.S. Department of Commerce, 2000).
Finally, we account for the possibility that the funding pattern of the program changed after the Department of Homeland Security Appropriation Act of 2006 by including several temporal variables. YEAR-06 is an indicator variable coded as 1 for grant disbursements taking place in 2006, 0 otherwise. To test for changing influence of our explanatory variables in 2006, we construct the interaction variables HIGH-2006, MED-2006, BLUE-RED-INDEX-2006, and INFLUENCE-INDEX-2006. In each case, the indicator variable YEAR-06 is multiplied by the respective explanatory variable.
From these data, the following hypotheses are tested.
Hypothesis 1: DHS grant allocation is determined by a state's risk of terrorist attack.
Hypothesis 2: DHS grant allocation is determined by the party affiliation of a state's elected officials.
Hypothesis 3: DHS grant allocation is determined by the power within Congress of a state's elected officials.
Hypothesis 4: The significance of terrorism risk, party affiliation, and congressional influence in determining grant allocations changed in 2006.
In the following section, these hypotheses are addressed through evaluation of the estimated sign and significance for the above-mentioned variables.
OLS regression estimates with robust standard errors are presented in Tables 3, 4, and 5. Table 3 presents a set of models where the three years of funding outcomes are pooled. Table 5 presents results where funding outcomes are parsed into two groups: funding decisions made for 2006 and funding decisions made for 2004-05. Table 4 again presents pooled models but presents specifications that include the temporal interaction variables and thus allows for testing of hypothesis 4. Model performance is acceptable as indicated by the coefficient of determination value of [R.sup.2] ranging from 0.635 to 0.916. F-statistics are significant to the 1 percent level in all specifications. Results are generally similar across each of the four dependent variables and coefficient sign and significance is generally robust across multiple specifications. (10) To start, we discuss the results from the pooled models (models 1-4).
On the whole, the results suggest that claims of the DHS not funding on the basis of terrorism risk are misguided. The estimated coefficient for the variable HIGH-RISK is positive and statistically different from zero in nearly every case. Relative to the base category LOW-RISK, states categorized as being of high risk of terrorist attack receive more funding, all else equal. The evidence with respect to states categorized as being of MED-RISK is not as clear. Indicated by estimated coefficients that are positive and significant, MED-RISK is a positive determinant of discretionary grant funding. However, with respect to the total disbursement a state receives, MED-RISK states do not statistically differ from LOW-RISK states.
Perhaps surprisingly, the results are inconsistent with the suggestion of DHS grants being distributed as "pork." Across all models and specifications, the estimated coefficient for BLUE-RED-INDEX is not statistically distinct from zero. We find a similar result when evaluating INFLUENCE-INDEX. Regression coefficients for INFLUENCE-INDEX are generally statistically significant and negative where the dependent variable is DISCRETIONARY, NON-DISCRETIONARY, or DISC-SHARE and not statistically significant where the dependent variable is TOTAL. We therefore find no support for the argument of congressional influence being a positive determinant of funding. (11)
With respect to sign and significance, the control variables show a mixed influence on funding outcomes. With some exceptions, POP, CANADA, and INCOME are generally positive determinants of funding. WHITE and OCEAN are usually negative determinants of funding. The estimated coefficients for the variable MEXICO are not statistically different from zero. Also noteworthy, the constant for both models of nondiscretionary grant funding (model 2 and model 6) is positive and statistically significant, indicating that base-category states (e.g., low-risk, nonborder states) are associated with positive funding levels. This result is consistent with the observation that some funds are awarded to each state irrespective of risk.
Focusing on Table 4, we observe several notable changes in the funding pattern of the DHS in 2006. First, the coefficient for the indicator variable YEAR-06 is positive and significant for models 5 and 8, and negative and significant for models 6 and 7. Because the total disbursement in 2006 was less than in 2005 and 2004 and the Department of Homeland Security Appropriation Act of 2006 increased the proportion of funding left to the discretion of the DHS, this result is expected.
More interesting though is the change in influence of several variables. Beginning with model 5, the estimated coefficients for both HIGH-2006 and MED-2006 are negative and significant. This implies that the influence of terrorism risk in determining discretionary grant funding was less in 2006 than in 2005 and 2004. This result is surprising given the spirit of the Department of Homeland Security Appropriation Act of 2006. This finding is further supported when the data are segmented into funding outcomes corresponding to before and after the Act. As indicated in model 9 and model 10, the estimated coefficients on HIGH-RISK and MED-RISK are positive and significant determinants of funding prior to the Act (2004-05) and not statistically significant after the Act (2006). Therefore, despite the intent to increase the emphasis terrorism risk carries in determining funding outcomes, the data suggest that the Department of Homeland Security Appropriation Act of 2006 has had the opposite affect on discretionary grant spending.
The significance of terrorism risk in determining total funding outcomes appears more stable across the two time periods. Model 7 shows the estimated coefficient for HIGH-RISK to be positive and statistically significant while the interaction variable HIGH-2006 is not statistically significant. The results thus suggest that in contrast to discretionary disbursements, the importance of terrorism risk in determining total funding outcomes is consistent.
For all categories of grant spending, the party affiliation and congressional influence of a state's elected officials does not appear to change in significance in 2006 as indicated by insignificant coefficients for BLUE-RED-2006 and INFLUENCE-INDEX-2006. These results are further supported by evaluating the Akaike Information Criterion (AIC) for each model. In comparing the restricted models of Table 3 with the corresponding extended models of Table 4, the AIC is minimized in the extended models where the interaction variables are found to be significant (models 5, 6, and 8). Where the interaction variables are not significant (model 7), the AIC is minimized in the restricted model (model 3). Therefore, while there is evidence to suggest a change in the determinants of funding left to the discretion of the DHS, the process by which the total distribution of grants are allocated appears consistent across the two time periods.
Our results can be summarized by returning to our four hypotheses presented earlier. First, our results support hypothesis 1: DHS funding outcomes are generally consistent with assessed terrorism risk, in particular with respect to high-risk states and with mixed support for medium-risk states. Despite the assertions to the contrary, we identify no positive relationship between party affiliation or congressional influence and funding outcomes. Therefore, our results do not support hypotheses 2 or 3.
The evidence is mixed with respect to hypothesis 4: The importance of terrorism risk in determining discretionary funding outcomes is diminished in 2006. However, with respect to the total grant disbursement a state receives, the impact of terrorism risk, party affiliation, and congressional influence is consistent across the two periods.
The process by which DHS grants are distributed has come under attack. The claim is that grants are not being allocated on the basis of terrorism risk but rather by political factors. In this paper, we use revealed preference analysis to statistically investigate this assertion.
Robust across a series of models, we find evidence that the risk of terrorist attack is a positive and statistically significant determinant of funding. In contrast, political factors, from the perspective of both party affiliation and congressional influence are not found to positively influence funding outcomes. Despite legislation espoused to increase the importance of terrorism risk in funding decisions, the determinants of the total grant allocation awarded to each state is consistent between 2004-05 and 2006. In fact, the results suggest that this legislation has decreased the importance of risk in determining discretionary funding outcomes.
Our analysis is admittedly limited by the quality of terrorism risk data available. However, while exact metrics of terrorism risk do not exist or are not publicly available, policymakers and researchers alike have shown interest in the distribution of DHS grants. The difficulty of doing so should not prevent analysis of the funding decisions of the department. Therefore, though our results should be taken with appropriate consideration, we find the funding pattern to be inconsistent with the claim of DHS grants being distributed by political and not terrorism risk considerations.
Tyler Prante is an assistant professor of Economics at Central Washington University. His current research interests are in natural resource and environmental economics.
Alok K. Bohara is a professor of Economics at the University of New Mexico. His research interests are in issues related to the environment, development, and political economy of poverty and violence.
1. Senator Ted Stevens (R-Alaska) argues, "I don't believe that every spot that's got a name on our map necessarily is a threat, you can have a threat where there's no population and you can have 'no threat' where there is a large population." In a similar thought, Senator Harold Rogers (R-Kentucky) argues, "It was a realization that every state has a degree of threat. You never can tell where the problem is going to originate" (Earle, 2004).
2. As an example, Senator John McCain (R-Arizona) states, "We need to adjust the funding; we really do. If that means Arizona may get less funds, then my constituents will understand. It'll be tough, because everyone wants to get their share of the pie" (Earle, 2004).
3. As noted earlier, the DHS distributes money to the states through a litany of grant programs. Therefore, the grants we consider do not represent the total disbursement awarded to each state. However, many
DHS grant programs were not in existence for each of the years considered in this study. We therefore restrict our analysis to four primary DHS grant programs available for each of the years of study. Description of the varying grant programs used by the DHS in each year is available at http://www.ojp.usdoj.gov/odp/grants_programs.htm#fy2007hsgp (accessed May 15, 2007).
4. Data for these variables are taken from annual DHS reports on the SHGP (DHS, 2004b, 2005, 2007).
5. The Delphi method describes a process of iterative, structured discourse among a group of experts. For detailed discussion of the Delphi method, see Linestone and Turoff (1975).
6. These results can be observed at http://opensecrets.org.
7. Congressional elections have changed the composition of elected officials over our time period of study. Political variables are taken from the 108th Congress for observations from 2004 and 2005 and from the 109th Congress for observations from 2006. Data for these variables are taken from http://www.opensecrets.org/politicians/candlist.asp?Sort=S&Cong=108, available May 15, 2007.
8. As before, congressional leadership positions are observed from the 108th Congress for observations taking place in 2004 and 2005 and the 109th Congress for observations from 2006.
9. Population data are taken from the 2000 Census.
10. Alternative specifications have been omitted for space but are available upon request.
11. Though concerns regarding heteroskedasticity are addressed by estimating with robust standard errors, we perform the analysis in per capita terms as well. Results from these models are consistent with the models presented herein and are available upon request.
Anderson, Gary M., and Robert D. Tollison. 1991. "Congressional Influence and Patterns of New Deal Spending, 1933-1939." Journal of Law and Economics 34 (1): 161-75.
Ansolabehere, Stephen, and James M. Snyder, Jr. 2006. "Party Control of State Government and the Distribution of Public Expenditures." Scandinavian Journal of Economics 108 (4): 547-69.
Becker, Gary S. 1983. "A Theory of Competition among Pressure Groups for Political Influence." Quarterly Journal of Economics 98: 371-400.
Borck, Rainald, and Stephanie Owings. 2003. "The Political Economy of Intergovernmental Grants." Regional Science and Urban Economics 33: 139-56.
Boyle, Susan, Sarah Malinowski, and Mike Gannon. 2002. AIR Worldwide Releases Terrorism Loss Estimation Model to Analyze Financial Impact of Terrorist Attacks on Insurers, Reinsurers, and Corporations. Press Release. http://www.iso.com/press_releases_2002/09_03_02.html. Accessed May 15, 2007.
Caudle, Sharon. 2005. "Homeland Security Approaches to Results Management." Public Performance and Management Review 28 (3): 352-75.
Clarke, Susan E., and Erica Chenoweth. 2006. "The Politics of Vulnerability: Constructing Local Performance Regimes for Homeland Security." Review of Policy Research 23 (1): 95-114.
Coats, R. Morris, Gokhan Karahan, and Robert D. Tollison. 2006. "Terrorism and Pork-Barrel Spending." Public Choice 128: 275-87.
de Rugy, Veronique. 2005. "Homeland-Security Pork." Washington Times (August 1). http://www.aei.org/include/pub_print.asp?pubID=22924. Accessed September 8, 2006.
______. 2006. "Pork Won't Protect Us." The Dallas Morning News (April 23). http://www.dallasnews.com/sharedcontent/dws/dn/opinion/points/stories/DN-derugy_23edi.ART0.State.Edition1.3e9314e.html. Accessed September 8, 2006.
Department of Homeland Security Appropriations Act of 2006, Public Law 109-90, 109th U.S. Congress (October 18, 2005).
Earle, Geoff. 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. Accessed March 1, 2008.
Fischhoff, Baruch, Roxana M. Gonzalez, Deborah A. Small, and Jennifer S. Lerner. 2003. "Judged Terror Risk and Proximity to the World Trade Center." The Journal of Risk and Uncertainty 26: 137-51.
Hoover, Gary A., and Paul Pecorino. 2003. "The Political Determinants of Federal Expenditure at the State Level." University of Alabama Economics, Finance and Legal Studies Working Paper No. 03-04-01. http://ssrn.com/abstract=395084. Accessed May 15, 2007.
Levitt, Steven D., and James M. Snyder, Jr. 1995. "Political Parties and the Distribution of Outlays." American Journal of Political Science 39: 958-80.
Linestone, Harold A., and Murray Tufoff. 1975. The Delphi Method: Techniques and Applications. Reading, MA: Addison-Wesley.
McFadden, Daniel. 1975. "The Revealed Preferences of a Government Bureaucracy: Theory." The Bell Journal of Economics 2: 401-16.
McLaughlin, Abraham. 2002. "Homeland Security-Pork or Protection?" The Christian Science Monitor (March 6). http://www.csmonitor.com/2002/0306/p03s01-usmi.html. Accessed September 8, 2006.
Oates, Wallace E. 1999. "An Essay on Fiscal Federalism." Journal of Economic Literature 37 (3): 1120-49.
O'Beirne, Kate. 2003. "Introducing Pork-Barrel Homeland Security: a Little Here, a Lot There." National Review (August 11). http://www.findarticles.com/p/articles/mi_m1282/is_15_55/ai_105891459/print. Accessed September 8, 2006.
Ripley, Amanda. 2004. "How We Got Homeland Security Wrong." Time (March 28): 163.
Roberts, Patrick S. 2005. "Shifting Priorities: Congressional Incentives and the Homeland Security Granting Process." Review of Policy Research 22 (4): 437-49.
______. 2006. "FEMA and the Prospects for Reputation-Based Autonomy." Studies American Political Development 20: 57-87.
Sunstein, Cass R. 2003. "Terrorism and Probability Neglect." The Journal of Risk and Uncertainty 26: 121-36.
Uniting and Strengthening America by Providing Appropriate Tools Required to Intercept and Obstruct Terrorism (USA Patriot Act) Act of 2001, Public Law 07-56, 107th U.S. Congress (October 26, 2001).
U.S. Department of Commerce, Economics and Statistics Administration, U.S. Census Bureau. 2000. Census 2000. http://www.census.gov/main/www/cen2000.html. Accessed March 1, 2008.
U.S. Department of Homeland Security (DHS). 2004a. Securing Our Homeland U.S. Department of Homeland Security Strategic Plan. http://www.dhs.gov/xlibrary/assets/DHS_StratPlan_FINAL_spread.pdf. Accessed May 15, 2007.
______. 2004b. Fiscal Year 2004 Homeland Security Grant Program. http://www.ojp.usdoj.gov/docs/fy04hsgp.pdf. Accessed May 15, 2007.
______. 2004c. Fiscal Year 2004 Urban Areas Security Initiative Grant Program. http://www.ojp.usdoj.gov/odp/docs/fy04uasi.pdf. Accessed May 15, 2007.
______. 2005. Fiscal Year 2005 Homeland Security Grant Program. http://www.ojp.usdoj.gov/odp/docs/fy05hsgp.pdf. Accessed May 15, 2007.
______. 2006a. History: Who Became Part of the Department. http://www.dhs.gov/xabout/history/editorial_0133.shtm. Accessed May 15, 2007.
______. 2006b. FY 2006 Homeland Security Grant Program. http://www.ojp.usdoj.gov/odp/docs/fy06hsgp.pdf. Accessed May 15, 2007.
______. 2007. Overview: FY 2007 Homeland Security Grant Program (January 5). http://www.dhs.gov/xlibrary/assets/grants-2007-program-overview-010507.pdf. Accessed May 15, 2007.
Wallis, John Joseph. 1996. "What Determines the Allocation of National Government Grants to the States." NBER Working Paper, Historical Paper 90. http://www.nber.org/papers/h0090. Accessed May 15, 2007.
Wisloski, Jess, and Tracy Connor. 2006. "Homeland's Pork." The New York Daily News (June 4). http://www.nydailynews.com/front/v-pfriendly/story/423576p-357369c.html. Accessed September 8, 2006.
Wright, Gavin. 1974. "The Political Economy of New Deal Spending: An Econometric Analysis." Review of Economics and Statistics 59: 30-8.
Table 1. Grant Disbursement Processes Program 2004 2005 2006 SHGP Each state gets 0.75 Each state gets 0.75 Each state gets percent of total percent of total 0.75 percent of SHGP disbursement SHGP disbursement total SHGP with the remaining with the remaining disbursement funds distributed funds distributed with the according to according to remaining funds population share. population share. distributed according to terrorism risk. LETPP Each state gets 0.75 Each state gets 0.75 Each state gets percent of total percent of total 0.75 percent of LETPP disbursement LETPP disbursement total LETPP with the remaining with the remaining disbursement funds distributed funds distributed with the according to according to remaining funds population share. population share. distributed according to terrorism risk. CCP Each state gets 0.75 Each state gets 0.75 Each state gets percent of total percent of total 0.75 percent of CCP disbursement CCP disbursement total CCP with the remaining with the remaining disbursement funds distributed funds distributed with the according to according to remaining funds population share. population share. distributed according to population share. UASI All funds All funds distributed All funds distributed according to risk. distributed according to risk. according to risk. SHGP, State Homeland Security Grant Program. LETPP, Law Enforcement Terrorism Prevention Program. CCP, Citizens Corps Program. UASI, Urban Areas Security Initiative. Table 2. Summary Statistics and Variable Descriptions Variable Description Mean (SD) Total Sum of a state's annual 43,468,412 (49,281,718) allocation under four DHS grant programs (SHGP, LETPP, CCP, and UASI) DISCRETIONARY Annual allocation left to the 17,473,884 (34,340,758) discretion of the DHS. (UASI in 2004-06 and SHGP + LETPP beyond 0.75 percent of the SHGP and LETPP awarded to all states and territories in 2006) NON- Total minus discretionary 25,994,528 (24,835,095) DISCRETIONARY DISC-SHARE Discretionary/total 0.284 (0.279) HIGH-RISK Indicator variable, 1 if a 0.22 (0.416) state is categorized as being of high risk of terrorist attack, 0 otherwise MED-RISK Indicator variable, 1 if a 0.36 (0.482) state is categorized as being of medium risk of terrorist attack, 0 otherwise BLUE-RED-INDEX The proportion of a state's 2.328 (1.411) senators that are Republican + the proportion of a state's congressmen that are Republican + 1 for each time a state was carried by George W. Bush in the 2000 and 2004 presidential elections INFLUENCE-INDEX Number of senators and 1.02 (1.348) congressmen that chair a committee, serve in the Homeland Security Committee, or serve in Senate or Congress leadership positions (Speaker of the House, House Majority/ Minority Leader, House Majority/Minority Whip, Senate Majority/Minority Leader, Senate Majority/ Minority Whip) POP State population (2000 census) 5,616,997 (6,143,925) MEXICO Indicator variable, 1 if a 0.08 (0.272) state borders Mexico, 0 otherwise CANADA Indicator variable, 1 if a 0.22 (0.416) state borders Canada, 0 otherwise OCEAN Indicator variable, 1 if a 0.46 (0.500) state is not land locked, 0 otherwise WHITE Percentage of state population 79.774 (12.810) categorized as white (2000 census) INCOME Per capita income (2000 30,512.02 census) (4238.925) HIGH-RISK-2006 High-risk*year-2006 0.073 (0.262) MED-RISK-2006 Med-risk*year-2006 0.12 (0.326) BLUE-RED-INDEX- Blue-red-index*year-2006 0.784 (1.381) 2006 INFLUENCE- Influence-index*year-2006 0.5 (1.236) INDEX-2006 Sources: Homeland security grant program reports (DHS, 2004b; 2005; 2007), AIR Terrorism Loss Model (Boyle et al., 2002), U.S. Census Bureau. DHS, Department of Homeland Security. SHGP, State Homeland Security Grant Program. LETPP, Law Enforcement Terrorism Prevention Program. CCP, Citizens Corps Program. UASI, Urban Areas Security Initiative. Table 3. Ordinary Least Squares Regression Results Model 2 Model 4 Model 1 ln-NON- Model 3 ln-DISC- ln-DISCRETIONARY DISCRETIONARY ln-Total SHARE HIGH-RISK 4.025 0.118 0.316 7.884 (2.05)** (0.97) (2.33)** (1.92)*** MED-RISK 4.336 -0.029 0.040 9.024 (3.47)* (-0.45) (0.56) (3.46)* BLUE-RED- 0.022 0.017 -0.007 -0.008 INDEX (-0.07) (0.68) (-0.28) (-0.01) INFLUENCE- -0.968 -0.051 0.018 -2.023 INDEX (-3.13)* (-1.80)*** (0.56) (-3.05)** YEAR-2006 7.692 -1.377 -0.593 16.183 (9.69)* (-24.47)* (-9.20)* (9.58)* ln-POP 3.587 0.337 0.585 6.625 (5.53)* (7.67)* (13.01)* (4.85)* MEXICO 0.908 0.195 0.086 1.878 (0.62) (1.39) (0.78) (0.60) CANADA 1.041 0.096 0.187 1.942 (1.18) (1.39) (2.50)** (1.02) OCEAN -1.658 -0.031 -0.154 -3.125 (-1.92)*** (-0.48) (-2.75)* (-1.69)*** WHITE -0.062 -0.002 -0.007 -0.118 (-1.46) (-0.74) (-3.85)* (-1.31) ln-INCOME 7.794 0.124 0.513 15.400 (1.68)*** (0.54) (1.80)*** (1.58) Constant -121.665 10.963 3.769 -250.538 (-2.34)** (4.39)* (1.20) (-2.31)** N 150 150 150 150 [R.sup.2] 0.6683 0.8640 0.8597 0.6347 AIC 5.947 0.587 0.645 7.447 F-stat 46.38* 71.49* 98.79* 37.52* Note: *,**,***indicate significance at the 1%, 5%, and 10% levels respectively. t-statistics in parentheses. AIC, Akaike Information Criterion. Table 4. Ordinary Least Squares Regression Results Model 5 Model 6 ln- ln-NON- Model 7 Model 8 DISCRETIONARY DISCRETIONARY ln-Total ln-DISC-SHARE HIGH-RISK 7.919 0.369 0.257 16.513 (3.55)** (2.64)* (1.86)*** (3.55)* MED-RISK 7.845 0.105 0.032 16.739 (4.91)* (1.30) (0.41) (4.99)* BLUE-RED- -0.066 0.025 0.003 -0.089 INDEX (-0.18) (1.03) (0.10) (-0.12) INFLUENCE- -1.071 0.039 0.003 -2.218 INDEX (-2.43)** (1.14) (0.08) (-2.45)** YEAR-2006 12.820 -0.850 -0.584 27.666 (8.99)* (-8.59)* (-3.57)* (9.42)* ln-POP 3.470 0.333 0.589 6.367 (4.87)* (7.16)* (12.98)* (4.23)* MEXICO 0.659 0.125 0.105 1.307 (0.58) (1.45) (0.96) (0.57) CANADA 1.003 0.097 0.188 1.861 (1.68)*** (1.59) (2.49)** (1.50) OCEAN -1.789 -0.019 -0.153 -3.406 (-2.68)* (-0.37) (-2.74)* (-2.46)** WHITE -0.064 -0.002 -0.007 -0.121 (-1.55) (-0.93) (-3.85)* (-1.40) ln-INCOME 7.720 0.116 0.514 15.231 (1.75)*** (0.56) (1.82)*** (1.67)*** HIGH-2006 -11.645 -0.883 0.203 -25.845 (-9.53)* (-7.74)* (0.90) (-10.03)* MED-2006 -10.111 -0.417 0.017 -22.836 (-8.00)* (-4.71)* (0.13) (-8.51)* BLUE-RED- 0.123 -0.426 -0.024 0.217 INDEX-2006 (0.28) (-1.35) (-0.48) (0.24) INFLUENCE- 0.573 -0.099 0.005 1.227 INDEX-2006 (1.54) (-2.68)* (0.09) (1.60) Constant -120.883 10.885 3.712 -248.833 (-2.33)** (4.47)* (1.19) (-2.31)** n 150 150 150 150 [R.sup.2] 0.7718 0.9155 0.8628 0.7585 AIC 5.627 0.165 0.676 7.086 F-stat 137.04* 114.16* 69.08* 124.12* Note: *,**,***indicate significance at the 1%, 5%, and 10% levels respectively. t-statistics in parentheses. AIC, Akaike Information Criterion. Table 5. Ordinary Least Squares Regression Results Model 10 Model 9 2004-05 Model 11 Model 12 2006 ln- 2006 2004-05 ln-DISCRETIONARY DISCRETIONARY ln-Total ln-Total HIGH-RISK 0.71 5.75 0.45 0.26 (1.37) (2.22)** (1.47) (1.76) MED-RISK 0.27 6.76 0.04 0.04 (0.76) (4.12)* (0.30) (0.50) BLUE-RED-INDEX -0.27 -0.04 -0.02 0.00 (-0.27) (-0.10) (-0.34) (0.03) INFLUENCE-INDEX -0.09 -1.21 0.02 -0.00 (-1.07) (-2.55)** (0.42) (-0.05) ln-POP 1.19 4.46 0.58 0.59 (6.54)* (5.54)* (6.13)* (11.11)* MEXICO -0.27 0.85 -0.25 0.16 (-0.63) (0.54) (-0.11) (1.35) CANADA 0.15 1.38 0.14 0.21 (0.50) (1.57) (0.97) (2.32)** OCEAN -0.43 -2.64 -0.20 0.21 (-1.89)*** (-2.73)* (-1.99)** (-2.03)** WHITE -0.02 -0.09 -0.01 -0.01 (-2.35)** (-1.46) (-2.71)** (-2.75)* ln-INCOME 1.18 10.81 0.65 0.44 (0.89) (1.72)*** (1.06) (1.43) Constant -12.60 -164.58 2.07 4.34 (-0.84) (-2.30)** (0.30) (1.27) n 50 100 50 100 [R.sup.2] 0.7919 0.7416 0.8158 0.8678 F-stat 24.79* 89.38* 23.55* 59.13* Note: *,**,***indicate significance at the 1%, 5%, and 10% levels respectively. t-statistics in parentheses.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: What Determines Homeland Security Spending? an Econometric Analysis of the Homeland Security Grant Program. Contributors: Prante, Tyler - Author, Bohara, Alok K. - Author. Journal title: Policy Studies Journal. Volume: 36. Issue: 2 Publication date: May 2008. Page number: 243+. © 1999 Policy Studies Organization. COPYRIGHT 2008 Gale Group.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.