Public Choice (2016) 168:239–263 DOI 10.1007/s11127-016-0358-z
NIH biomedical funding: evidence of executive dominance in swing-voter states during presidential elections Alberto Batinti1
Received: 15 February 2016 / Accepted: 13 August 2016 / Published online: 23 August 2016 Ó Springer Science+Business Media New York 2016
Abstract This paper explores the role of presidential politics in influencing the distribution of the National Institutes of Health (NIH) funding. In particular, it investigates how the distribution of NIH funding is stirred towards institutions located in swing-voter US states during presidential elections. In doing so, it fills a gap left in the literature on the political economy of the NIH, which previously focused on the role of membership in the Committees on Appropriations in both chambers of the US Congress. First, it is found that NIH funded performers in states where the Presidential Electoral Importance (PEI) increases by 1 %, receive, on average 0.7–0.8 % more funding. Second, this effect is robust to three additional checks. Third, I run heterogeneity tests, where the direction and change of the elasticity coefficient fit plausible assumptions on the mechanism of presidential influence on NIH funding in swing-voter states. I finally estimate, that the average lower bound of the overall impact of PEI on the NIH budget is between 2 and 3 %. It reaches a maximum of 10 % for specific states, fiscal years, and presidential mandates. Keywords NIH Political economy Congressional dominance Presidential politics JEL Classification D70 E6 H5 O38 I1
1 Introduction The National Institutes of Health (NIH) is the largest single source of federal civilian R&D spending and the first source of publicly funded biomedical research in the United States. The NIH currently funds more than $30 billion in annual grants to research institutions around the world. As shown in Tables 1 and 2, far more than a fifth of the total estimated
& Alberto Batinti
[email protected];
[email protected] 1
School of Public Economics and Administration, Shanghai University of Finance and Economics (SUFE), 777 Guoding Road, 200433 Shanghai, China
123
240
Public Choice (2016) 168:239–263
Table 1 Biotech R&D funding in the US, NIH compared to other sources of biomedical R&D funding in the US Biomedical R&D in the United States 2004–2011: million$ INDUSTRY
FED GOV
NIH
NIH (% total)
OTHER
TOTAL
2004
59,140
37,852
28,036
25.73
11,985
108,977
2005
61,140
37,316
28,644
25.72
12,917
111,373
2006
64,660
37,706
28,516
24.57
13,713
116,079
2007
68,260
38,066
29,128
23.79
16,094
122,420
2008
74,800
38,612
29,328
22.47
17,113
130,525
2009
74,347
46,798
35,508
25.56
17,794
138,939
2010
76,522
45,875
34,829
24.79
18,108
140,505
2011
77,580
39,552
29,831
21.90
19,113
136,245
Source Research!America various available annual reports available at Research!America official website
Table 2 Biotech R&D funding in the US, United States compared to other OECD countries (2004–2014)
OECD GBAORD dataset
% of total R&D spending Year
USA
Other OECD
2004
23.24
6.60
2005
22.76
7.56
2006
21.84
8.17
2007
21.90
7.91
2008
21.51
7.80
2009
26.74
8.42
2010
22.96
8.35
2011
23.23
8.88
2012
23.60
8.53
2013
24.50
8.94
2014
25.06
8.63
US investment in biomedical research comes from the NIH. On average, for the 2004–2014 period, the US government’s budget dedicated to biomedical R&D as percentage of total government spending on R&D is three to four times that allocated in other selected OECD countries. About 80 % of the NIH budget is awarded through competitive grants and contracts to more than 300,000 researchers working at 2500 organizations, including universities, medical schools, and other private and public research institutions.1 To understand the outcomes of the selection processes adopted by R&D agencies, economists have been studying how specific rules for the attribution of agencies’ funding influence researchers’ efforts to pursue careers in some fields rather than in others, and about how scientific outcomes are affected by those rules (Lazear 1997). Empirical research from Lichtenberg 1
The details reported in this section can be found at the following website: http://www.nih.gov/about-nih/ what-we-do/budget (Accessed on February 2016).
123
Public Choice (2016) 168:239–263
241
(2001) shows that the NIH’s decision process is animated by two main motives: one is rewarding scientific opportunity and productivity through basic research funding, the other is counteracting and alleviating the burden of disease through applied research. Being a governmental agency, the attribution of R&D funding to the NIH is likely to depend on political factors and targeted towards specific constituencies as well (Coate and Morris 1995; Tullock 1983). The Congress and the President have the power to define the organizational structure of the NIH by proposing and approving the creation, termination, and definitions of the scientific area of competencies of the individual institutes, and might well be in conflict in doing so.2 For example, Godefroy (2011, p. 8) reports that the Congress refused the proposal by Clinton to create an President institute specializing in HIV research.3 Authors like Weingast and Moran (1983), Weingast (1984) and Weingast and Marshall (1988) see the structure of the US Congress as the result of an institutional design conceived to give to a particular set of politicians the power to influence agencies’ spending towards specific areas of interest or constituencies. The President of the United States is involved as well in designing the NIH’s budget proposals and in defining its priorities before a proposed budget becomes a formal appropriations bill.4 The executive branch can also exercise its veto power, and has done so in the past. In June 2007, President Bush vetoed a bill that would have eased restraints on federally funded embryonic stem-cell research.5 Later in November, the president vetoed again a spending bill that would have increased the NIH’s budget by $1 billion.6 Budget requests are assembled by the individual institutes and centers that report to the NIH director and its staff. The director of the NIH then negotiates with the Department of Health and Human Services and with the Office of Management and Budget within the Executive Office of the President. The result is a budget request from the NIH that has to be consistent with the White House’s priorities. Of course, US political institutions do not formally determine the distribution of NIH grants, which is based on a peer-review process. However, the president and both legislative chambers contribute actively to the approval of the NIH budget. Politicians occupying key positions within different branches of government are players in a repeated game in which the final decisions of the NIH’s management’s likely will pay close attention to the proposals by politicians, especially the long-tenured ones in important committees, who will likely be involved in future budget approvals. Previous research finds that performing institutions represented by congressmen on the House and Senate 2
In particular, some institutes may be assigned a very broad competence on a specific disease, but can be constrained on other dimensions, for example, the research has to be dedicated only to basic or applied research, but not to both.
3
Moreover, the NIH, together with providing systematic public research for the discovery of new treatments for health improvement, supports about 432,000 jobs across the United States. Some sources report also that $1 in NIH funding is estimated to generate $2.21 in local economic growth. See: http://www.nih. gov/about/impact/index.htm (Accessed online February 2016).
4
The bill is then submitted to the Congress for adjustments, and indicates the total amount of funds for the next fiscal year, presenting details on the total budget assigned to every NIH institute or center.
5
See http://www.npr.org/templates/story/story.php?storyId=11207876 (Accessed online February 2016).
6
See http://www.the-scientist.com/?articles.view/articleNo/25677/title/Bush-vetoes-NIH-budget-increase/ (Access February 2016). Part of the Bush administration’s opposition was based on seeing unnecessary political pork and earmarking in the bill. The political struggle took went so far that, according to some analysts, the president’s veto marked the end of the stable bipartisan agreement about NIH funding that had doubled the resources dedicated to the public medical research funding for the agency during the 1998–2003 period, see: http://www.nationalhealthcouncil.org/NHC_Files/news/12-18-07-news_04-07-09.pdf (Accessed online February 2016).
123
242
Public Choice (2016) 168:239–263
Committees on Appropriations receive more funding from the NIH (Sampat 2012; Godefroy 2011; Hegde 2009; Hegde and Mowery 2008).7 However, this strand of research is limited to the investigation of the legislative oversight power of the congressional committees. Given the documented involvement of the President, this paper attempts at filling this gap by exploring how the President influences the distributions of NIH funding across US states that are of particular importance.8 Previous research on other agencies found empirical evidence of the influence of presidential politics. Grier (1987) explores the Federal Reserve System’s policy and finds systematic deceleration in money growth in the year following presidential elections, and acceleration in the next 3 years. Empirical support for presidential influence also is found for central bank policy (Chappell et al. 1995, 1993), for the Internal Revenue Service’s auditing procedures (Young et al. 2001), and for the Department of Homeland Security’s granting activity (Coats et al. 2006). More in line with the empirical strategy adopted in this paper, previous research links public spending across US states to the importance of states’ electoral votes in presidential elections, importance measured by the contestability of a specific state during such elections. Previous research finds that swing voter states during presidential elections systematically received larger grants from the New Deal’s programs (Arrington 1969, 1970; Reading 1973; Wright 1974; Anderson and Tollison 1991; Couch and Shughart 1998; Wallis 1987). Grier et al. (1995) show that presidential veto decisions are systematically used so to favor voters in closely contested states. Similar results are found also in Garrett and Sobel (2003) regarding the Federal Emergency Management Agency’s disaster payments policy. More recently, a similar methodology has been applied to the American Recovery and Reinvestment Act (ARRA), which represents the largest US fiscal stimulus package since the Great Depression. Gimpel et al. (2010); Young and Sobel (2013) show that the political variables measuring executive and legislative oversight power explain better than Keynesian variables, based on multipliers’ effects, the distribution of ARRA funds.9 Most of all, the conclusions from these papers suggest that presidential politics matters both in theory and practice, and it is worth noting that some authors claim that in some cases presidential politics plays a predominant role in influencing agencies’ behavior (Moe and Wilson 1994).10 In line with the research above, this paper finds that this is the case when looking at NIH’s geographical spending patterns; on average, a 1 % increase of 7
This strand of literature follows, among others, previous empirical tests run for other agencies, as in Faith et al. (1982) and Weingast and Moran (1983), who test Posner’s (1969) conjecture that Federal Trade Commission rulings tend to be more favorable in districts with congressional oversight committee membership. Hunter and Nelson (1995) study the relationship between Congress and the Internal Revenue Service. They find that the IRS allocates more resources for tax enforcement towards states not represented by oversight committee members. Kosnik (2006) reports that congressional representation affects the Federal Energy Regulatory Commission’s relicensing of hydroelectric dams. More recently, Ryan (2014) explores how congressional dominance affects the allocation of H1N1 (swine flu) doses. The paper finds that states with Democratic House members seated on key oversight committees were very responsive in sending about 60,000 more doses for each legislator in the first phase of the flu epidemics.
8
I will make clearer how this relevance is measured in the data section.
9
Reifler and Lazarus (2010) focus on party affiliation at the level of congressional districts.
10
On the other hand, an answer to this question is not trivial. For example, Larcinese et al. (2006) find a role for presidential politics in the distribution of the federal budget across US states, but the drivers are based on support for the incumbent president and same party affiliation, while no evidence is found supporting the swing voter effect. This paper finds instead that being a swing voter state is an important driver of more NIH funding. Another advantage of my approach is that my unit of observation is the single performer, so any concern of simultaneity bias should be cast away, in that the likelihood of a single institution manipulating presidential elections so to increase the contestability in their state of residence is likely to be zero.
123
Public Choice (2016) 168:239–263
243
the electoral contestability in one state, increases funding for NIH’s performers in that state by 0.7 %.
2 Data, data sources, and measures One challenge in pursuing this research question is that presidential elections are held every 4 years. To test my hypothesis, I build a novel and unique dataset containing NIH data on applications over a wide timeframe (1972–2015) that allows sufficient variability for gauging presidential election results. For the construction of the dataset I relied on three data sources. NIH data were first collected from the online EXPORTER database covering the 2000–2015 period. This database includes very detailed applications-level funding with about 40 variables defining, among other characteristics, the type of application, the activity code of reference, the fiscal year, and the state of residence of the funded institution/performer. Additional information and data were taken from the CRISP legacy database (1985–1999), which contains several relevant variables, but not information on funding. Finally, additional data were provided directly by the NIH’s Office of Extramural Research for the 1972–1999 period.11 By merging the EXPORTER, CRISP and the additional NIH information directly, I built the final dataset comprising funding data for the fiscal years 1972–2015. The dataset contains only applications related to extramural research for which the institution’s state of residence was identifiable and above $500. It excludes the insular areas of Guam, American Samoa, the Northern Mariana Islands, Puerto Rico, the US Virgin Islands and the District of Columbia.12 Overall, the dataset contains about 1,780,000 applications and 16,789 institutions, and for each of them the aggregated NIH funding per fiscal year was calculated. Figure 1 presents 12 maps showing the state-level distribution of NIH funds across US presidential administrations from 1968 to 2012. Data are built by calculating 4-year averages of the state-level geographical evolution of NIH spending per presidential administration. Second, I collected data on US state-level presidential election returns for the 1968–2012 period by hand. These data are necessary to construct the main independent variable. I use these data to compute a measure of contestability.13 In particular, I call the measure of states’ contestability Presidential Electoral Importance (PEI). This is based on a straightforward transformation of the percentage of votes cast in favor of the winning candidate at the state level. This is the appropriate geographical level because electoral votes and not popular votes determine a presidential election’s winner. In fact, by considering how US presidential elections are structured—winner takes all—a small statewise margin of victory implies that all available electoral votes will be assigned to the winning candidate. This variable is built by following the methodology adopted in Garrett and Sobel (2003). PEI is equal to 100 9 [1 – 4 (X - 0.5)2], where X is the percentage of total 11 Data were obtained by signing a Freedom of Information Act (FOIA). I am very grateful to the NIH Office of Extramural Research for sharing the NIH CRISP data under FOIA#40443. 12 In the case of the District of Columbia, data on presidential importance are available, but not on congressmen or possible appropriators. This implied the exclusion of about 2165 NIH performers out of more than 83,000 from the dataset. When including DC the size of presidential electoral importance is still relevant in size and significant at the 1 % level. However, a full comparison is not possible. 13
Presidential election data are available from the website http://www.presidency.ucsb.edu, which provides information on all US presidential contests from 1789 to 2012.
123
244
Public Choice (2016) 168:239–263
Fig. 1 Total NIH funding by State and Presidential Mandate (1968–2012). Notes In order to control for States’ size, for each state the per-performer average of the log of total funding is calculated. One map for each Presidential mandate
votes obtained by the winning candidate. Thus, when X is close to 1 (low contestability), PEI approaches 0, whereas when X is close to 0.5 (high contestability), PEI approaches 100. According to previous tests, it is expected that the higher is the PEI in a specific state, the larger will be the sum of grants allocated to the NIH performers in that state. Figure 2
123
Public Choice (2016) 168:239–263
245
shows a set of state-level maps presenting the geographical distribution of the PEI variable for the presidential elections between 1968 and 2012. Data on congressional membership in the Committees on Appropriations are from Charles Stewart’s congressional data page (http://web.mit.edu/17.251/www/data_page.html). To
Fig. 2 Presidential importance by State and Presidential Mandate (1968–2012). Notes PEI values by State and President for the whole available sample. PEI is equal to 100 9 [1 - 4 (X - 0.5)2] where X is the percentage vote obtained by the winning candidate. If X is very close to 1, (low contestability), PEI approaches the value of 0, while when X is close to 0.5 (high contestability), PEI approaches 100
123
246
Public Choice (2016) 168:239–263
cover the period of interest I combined two datasets, both available at this web page: One including information from the 93rd to 102nd Congress, and collected by Garrison Nelson (Committees in the U.S. Congress, 1947–1992); the other, covering the 103rd–112th Congress, collected by Charles Stewart III and Jonathan Woon (Congressional Committee Assignments, 103rd–112th). Missing committee assignment data for the 113th and 114th Congress were hand collected from the website of the Center for Responsive Politics (www. opensecrets.org), which gives details on the recent composition of both appropriations committees. Figure 3 presents a set of maps showing the state-level geographical distribution of the (Senate plus House) members of the Committee on Appropriations in the two chambers. More precisely, each map shows the House and Senate averages within each presidential administration from 1972 to 2012. The datasets collected from the three main sources were combined together by following the process described in Table 3. For lack of sufficient fiscal years under Nixon’s first administration (only 1972 is available), the 1968 electoral data were dropped from the final dataset along with appropriations data from the 92nd Congress and NIH’s data from fiscal year 1972. The resulting dataset covers 11 US presidential elections, from Nixon II to Obama II, NIH fiscal years for the 1973–2014 period, and data from the 93rd to the 114th Congress. Summary statistics for the dataset used for the baseline regressions are reported in Table 4, and the final results consists on an unbalanced panel of 87,395 observations at the institution/fiscal year level. Figures 4 and 5 present boxplots and scatters of the main variables of interest.
3 Results 3.1 Model Baseline regressions are obtained by estimating the linear model in Eq. (1), wherein the logarithmic transformation of total funding, aggregated by summing all of the NIH applications at the performer/institution level for a given fiscal year, is the dependent variable of interest. This variable is labeled by Log($FUNDijt) and it measures the amount of NIH funding to institution/performer i in state j during fiscal year t. The independent variable of interest is Log(PEIjs), indicating the logarithmic transformation of the measure of electoral importance of state j during presidential election s, so that b1 can be interpreted as an elasticity. Terms li and ct represent, respectively, performer/institution and fiscal year fixed effects. Consistent with the previous literature, I also control for the number of members of the Appropriations Committees in both chambers, referring to state j and Congress h (variable H/SAC). In some specifications I also allow for additional controls both at the institution and state level (term U0ðiÞjt X). Log $TOTFUNDijt ¼ b0 þ b1 Log PEIjs þ b2 H=SACjh þ U0ðiÞjt X þ li þ ct þ eijt ð1Þ The main hypothesis is that b^1 [ 0, showing that the larger is PEIjs, the greater is electoral contestability, thus generating more NIH funding. More precisely, b^1 is the elasticity, i.e., the percentage increase in $TOTFUNDijt associated with a 1 % increase in the PEIjs scale.
123
Public Choice (2016) 168:239–263
247
Fig. 3 Average number of appropriators by State and Presidential Mandate (1972–2012). Notes Because for each presidential mandate two values are available, maps report the average for each President
123
248
Public Choice (2016) 168:239–263
Table 3 Dataset structure and combination
Congress
Election
Winner
NIH-FY
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114
1968 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
NixonI NixonII
72 (dropped) 73–74 75–76 77–78 79–80 81–82 83–84 85–86 87–88 89–90 91–92 93–94 95–96 97–98 99–00 01–02 03–04 05–06 07–08 09–10 11–12 13–14 15
Carter ReaganI ReaganII BushI ClintonI ClintonII BushJrI BushJrII ObamaI ObamaII
Notes
Only state level data from FOIA and CRISP Legacy
From 85-99: District Level Data from FOIA and Exporter
From 00-15: District Level Data from NIH Exporter
Nixon’s electoral results were used also when President Ford was in charge. When using the veto power measure, President Ford’s one was used in that specific data were available for this type of measure
3.2 Baseline results Table 5 presents the baseline results, obtained by progressively adding control variables and fixed effects. All standard errors are clustered at the state level in order to control for collinearity in errors across institutions within the same state.14 The results in column (1) show that the effect of PEI on funding is positive as predicted, but not significant. In column (2), I add institutional fixed effects and fiscal years dummies. Note that NIH fiscal year dummies also control for congressional and presidential fixed effects in that fiscal year clusters are nested. Column (3) adds the total number of seats on the Committee on Appropriations by the state-specific congressional delegations (sum of Representatives and Senators). The result for this variable is not significant throughout the main tests in the paper.15 More specifically, significance for this variable is lost when robust standard errors, clustered at the US states’ level, are estimated. However, it is significant at 5 % level when standard errors are heteroscedastic robust, and the sign is consistent with the theoretical prediction that a larger number of appropriations committee members (be they representatives or senators) is associated with more transfers to the institution(s) in the represented geographical area.16 In column (4), I add the number of 14 Here I follow the rule given in Cameron and Miller (2015) and apply clustering on the base of the largest geographical unit, and US states are the largest geographical unit in this dataset. 15 I also tried regressions using the total appropriators for House and Senate separately. The coefficients are still insignificant for both variables. Thus for simplicity, I report the total number of appropriators. Results are available on request. 16
Results are available on request as well.
123
Public Choice (2016) 168:239–263
249
Table 4 Summary statistics Variable
N
Mean
FISCAL YEAR (NIH)
85,365
1998
CONGRESS
85,365
105.1
6.030
H/SAC
85,365
2.990
2.130
PERC_W
85,365
51.40
SEATS_W
85,365
14.84
PEI
85,365
97.29
PEI_LOSS
85,365
0.510
PEI_LOSS_5
85,365
TOTFUND (LOG)
85,365
SD 12.06
Max
Sources
1973
2015
NIH
93
114
NSW
0
9
NSW
24.70
78.20
P
0
55
P
3.570
68.19
100
P
0.500
0
1
P
0.0800
0.270
0
1
P
13.01
2.010
6.210
20.47
NIH
TOTSEATS
85,365
21.02
N_ORGSTATE
85,365
120.7
STATE-FY FUNDING (LOG)
85,365
19.66
8.110
Min
16.13
15.52 112.8 1.630
3
55
P
1
475
NIH
12.15
22.32
NIH
The unit of observation is the performer per NIH fiscal year. The panel is unbalanced. For more details and definitions, please refer to the Data section of the paper NIH National Institutes of Health, NSW Nelson, Stewart, and Woon, P Data from http://www.presidency. ucsb.edu
institutions in the state, as this variable can measure network effects that might favor additional NIH funding. This variable has both a negligible effect on funding and is not statistically significant. This does not mean that network effects are absent, but perhaps that performers belong to interstate research networks not captured by this state-level variable. At the same time, it is difficult to establish a convincing connection between interstate networks of researchers and the state-level aggregate PEI that could cause endogeneity. Columns (5) and (6) add two variables that are significant at the 1 % level; these and are the lagged value of the dependent variable as well as an aggregate trend of NIH funding at the state level. The omission of these variables creates biased estimates; the coefficient of PEI in column (4) is progressively reduced from 1.232 to 0.701, but it remains positive and significant at the 1 % level.17 Column (7), where state fixed effects are added, shows that a 1 % increase in PEI implies about a 0.75 % increase in funding to the performing institution. In column (8), congressional district fixed effects are entered instead, with the consequence of a small reduction in the estimated elasticity measure, which diminishes slightly from 0.746 to 0.701.18 Overall the result in (6) is robust to the additional checks in columns (7) and (8). Moreover, the decline in the number of observations in (8) is explained by the lack of district- level information before 1984. However, the estimate of interest is robust despite the change in the reference period and the addition of district-level fixed effects.
17 I recall here that PEI coefficients are to be interpreted as elasticities. Given that PEI is a constructed scale, this is the appropriate measure to gauge the entirety of the effect of a change in PEI on institution-level funding. 18 State fixed effects are justified in that a small number of institutions (212 out of 16,517) in fact changed their state of residence during the 43-year period under scrutiny. However, adding this control does not change the main results obtained without state fixed effects. Moreover, having such a small number of performers moving across states should remove doubt that the effect obtained is the result of the selfselection process of institutions looking to operate in highly contestable states to obtain additional funding.
123
250
Public Choice (2016) 168:239–263
(1)
(2)
Boxplot: Log of Total Funding at the State Level
Boxplot: Log of Total Funding at the State Level Source: NIH data, FY 1972-2015
10
WY1979 WY1977 WY1976
10
12
11
14
12
16
13
18
14
20
Source: NIH data, FY 1972-2015
(3)
(4)
Boxplot: Presidential Electoral Importance
Boxplot: Log of Presidential Electoral Importance: State Level Source: NIH data, FY 1972-2015
4.5
100
Source: NIH data, FY 1972-2015
60 40
3.5
DC1980 DC1978 DC1977 DC1979 DC1972 DC1996 DC1995 DC1994 DC1993 DC1998 DC2000 DC1999 DC1997 DC1991 DC1992 DC1989 DC1990 AL1972 DC1985 DC1988 DC1987 DC1986 MS1972 DC1984 DC1981 DC1983 DC1982
DC1995 DC1993 DC1994 DC1996 DC1998 DC1999 DC1997 DC2000 DC1990 DC1992 DC1989 DC1991 AL1972 DC1987 DC1988 DC1986 DC1985 MS1972 DC1984 DC1982 DC1981 DC1983
DC2008 DC2005 DC2006 DC2007 DC2015 DC2014 DC2013 DC2002 DC2001 DC2004 DC2003 DC2012 DC2010 DC2011 DC2009
DC2005 DC2006 DC2007 DC2008 DC2013 DC2015 DC2014 DC2004 DC2002 DC2001 DC2003 DC2012 DC2009 DC2010 DC2011
3
20
VT2011 MA2003 VT2012 VT2010 MA2001 MA2004 MA2002 VT2009 WY2010 UT1973 UT1976 UT1975 UT1974 WY2009 WY2011 WY2012 TN1975 KS1974 TN1976 KS1976 KS1975 TN1974 KS1973 TN1973 WY2004 WY2002 WY2003 ND1994 VA1976 VA1973 ND1993 VA1974 ND1995 VA1975 ND1996 WY2001 RI2001 RI2004 RI2003 RI2002 RI1972 ID2005 ID2008 ID2007 ID2006 OK1988 NH1988 NH1986 NH1985 OK1986 OK1985 OK1987 NH1987 AR1975 WY2006 AR1976 AR1974 WY2005 WY2008 WY2007 AR1973 WY1974 WY1973 WY1975 WY1976 AR1972 NC1976 NC1973 NC1975 NC1974 GA1972 AK1996 AK1994 AK1995 AK1993 HI2014 WY1986 HI2013 NE1973 HI2015 NE1975 WY1987 NE1976 WY1985 WY1988 NE1974 NE1994 NE1995 NE1986 NE1987 NE1988 NE1993 NE1996 NE1985 SC1974 SC1973 SC1976 SC1975 UT2006 UT2007 UT2005 UT2008 ID1993 ID1995 ID1994 ID1996 HI2009 HI2011 HI2012 HI2010 FL1975 FL1973 FL1974 FL1976 WY2014 WY2013 WY2015 ID1986 AL1973 ID1987 ID1988 AL1974 AL1975 AL1976 ID1985 UT1982 UT1981 UT1984 UT1983 OK1976 OK1974 OK1973 OK1975 UT1988 UT1986 UT1985 UT1987 GA1975 GA1974 GA1976 GA1973 UT1996 UT2014 UT1995 UT1994 UT1993 UT2015 UT2013 LA1972 MS1973 MS1976 MS1975 MS1974 DC1973 DC1976 DC1974 DC1975 DC1980 DC1977 DC1978 DC1979 DC1972
4
80
ND1994 WY2001 ND1995 WY2004 WY2003 ND1996 VA1976 ND1993 VA1974 WY2002 VA1973 VA1975 RI2002 RI2001 RI2004 RI2003 RI1972 ID2007 ID2005 ID2006 ID2008 OK1985 NH1987 OK1987 NH1988 OK1986 OK1988 NH1986 NH1985 WY2005 WY2008 WY2007 WY2006 AR1976 AR1973 AR1975 AR1974 WY1975 WY1974 WY1973 WY1976 AR1972 NC1973 NC1976 NC1975 NC1974 GA1972 AK1994 AK1995 AK1993 AK1996 HI2014 NE1975 HI2013 NE1973 WY1988 NE1976 HI2015 NE1974 WY1985 WY1986 WY1987 NE1988 NE1985 NE1987 NE1994 NE1996 NE1993 NE1986 NE1995 SC1975 SC1976 SC1973 SC1974 UT2005 UT2008 UT2007 UT2006 ID1993 ID1994 ID1996 ID1995 HI2010 HI2011 HI2009 HI2012 FL1976 FL1975 FL1973 FL1974 WY2015 WY2014 WY2013 ID1985 AL1973 ID1988 AL1975 AL1974 ID1986 ID1987 AL1976 UT1982 UT1983 UT1981 UT1984 OK1976 OK1973 OK1974 OK1975 UT1988 UT1987 UT1986 UT1985 GA1976 GA1975 GA1974 GA1973 UT2013 UT1996 UT1993 UT1994 UT2015 UT1995 UT2014 LA1972 MS1975 MS1973 MS1974 MS1976 DC1974 DC1976 DC1973 DC1975
(5)
(6)
Boxplot: Average Appropriators per State
Boxplot: Average Appropriators per State Source: NIH data, FY 1972-2015
1
10
Source: NIH data, FY 1972-2015
ND1977 ND1980 ND1978 ND1979
CA2012 CA2014 CA1999 CA1998 CA2011 CA1997 CA2010 CA2009 CA2013 CA2000
.8
8
CA2008 CA2007
CA1995 NY1999 TX1994 TX2007 NY2000 NY1986 CA2004 TX1993 CA2003 CA2005 NY1985 TX2008 TX2010 TX1996 CA1991 NY1983 CA1996 CA1992 TX1995 TX2009 NY1984 CA2006
ND1981 ND1973 ND1974 ND1991 ND1982 ND1976 AK2014 AK2013 NV1974 ND1985 MT2012 ND1986 NV1973 MT2006 VT1995 MT2015 ND1983 MT2011 MT2010 ND1992 MT2005 ND1975 VT1996 MT2009 ND1984
.6
6
CA2001 CA2015 TX2005 CA1994 CA1993 TX2006 CA2002
.2 0
.4
OR1974 HI1982 ME2014 NV1993 RI1974 NH1974 ID2003 HI1984 NV1996 HI1988 HI1989 MT1975 AR2014 ID2007 NV1994 HI1985 MT1977 MT1976 AR1978 AR1977 RI2008 NV1995 HI1990 NV1992 RI2001 RI2007 HI1981 RI1973 NH2001 HI1975 ID2008 UT1980 ID2005 HI1987 ME2015 NH2000 RI2009 NH1973 HI1986 ME2013 MT1978 RI2010 HI1973 NV1991 HI1983 ID2004 RI2002 ID2006 HI1974 NH1999 UT1979 NH2002 AR2013 OR1973 HI1976
2
NC1997 FL1974 PA1994 NY1989 IL2004 FL1984 AL2001 AL1998 OH1995 NY2009 OH2003 OH2011 FL2009 OH2006 CA1978 TX2015 FL1973 FL1999 OH2008 AL2002 OH2012 FL1988 OH2005 OH2001 CA1983 OH1992 IL1978 OH1997 FL1987 WA2000 FL2014 AL2008 PA1993 WA1996 TX2012 IL2008 FL2005 FL2006 NY1993 MA1976 OH2002 WA1997 NY1976 FL1986 IL1980 NY1997 TX1997 IL1977 TX2013 OH1990 OH1996 PA1996 NY1973 FL2002 FL1981 FL2001 OH2009 AL1999 FL1975 IL2003 PA2007 FL1980 FL1983 NY2010 NC1998 NY1994 FL1977 MA1975 IL2005 PA2008 WA1995 IL2007 FL1976 CA1979 FL2004 NY1992 OH2013 FL2000 TX2011 TX1998 OH2010 FL2013 NY1979 PA1974 NY1974 NY1998 MA1978 NY1975 FL1979 WA1999 OH1989 FL1978 PA1998 CA1977 AL1997 MA1977 AL2007 IL2006 FL1993 FL1982 OH1993 NY1980 PA1997 PA1973 FL2010 FL1994 NY1991 CA1980 OH2007 IL1979 FL1985 NY1990 FL2003 OH1998 PA1995 OH1994 OH1991 AL2000 OH2004 TX2014 NY1996 CA1984 WA1998 OH2014 NY1995
0
4
PA1987 TX1992 PA2009 NY2007 PA2006 PA1981 NY1987 NY2006 TX2000 CA1985 TX1991 TX1999 PA1986 PA1983 CA1989 IL1985 NY2004 TX1990 TX2003 TX2001 PA1989 NY1982 NY2001 PA1991 CA1974 PA1975 CA1987 FL2015 IL1986 PA1990 PA1984 PA2002 PA1988 CA1990 PA1992 NY1988 PA2005 PA1977 PA1985 CA1981 TX2002 FL2008 NY1977 CA1975 PA2004 TX1989 PA1979 CA1976 NY1981 NY1978 PA2010 CA1982 NY2008 CA1973 NY2002 CA1986 PA2003 PA1976 FL2007 CA1988 PA1982 NY2005 TX2004 PA1980 NY2003 PA2001 PA1978
Fig. 4 Boxplots of main variables. Notes Boxplots of main variables of interest. Data aggregated at the State/Presidential Mandate level for the period 1972–2015
3.3 Robustness The results in the baselines regressions are robust to three checks. Following previous literature on appropriations committees membership, I test if a reduction in PEI is associated with a loss in funding. To do so, I generated a dummy variable, PEI_LOSS, taking value equal to 1 if the percentage change in PEI is negative, and 0 otherwise.19 In Table 6, columns 4 to 7, the coefficient of PEI_LOSS is negative as expected, and significant at 1 % level. Institutions 19 Note that, in doing this, I do not lose observations (still 87,395 in the sample) because, while I do not have NIH data for fiscal years before 1972, I do have data on PEI from the 1968 presidential election and this allows building that variable for the first period in which NIH data are available.
123
Pub c Cho ce 2016 168 239–263
251
(1)
(2)
85
90
95
16
TN2012 TN2010 TN2009 TN2007 IA2011 PA2015 TN2013 TN2005 TN2015 TN2011 TN2004 IA1994 TN2003 TN2014 IA2015 TN2008 TN2002 MO2015 MO2004 MO2008 MO2012 MO2013 MA2010 MO2005 PA2013 MO2009 PA2014 NH2005 WA2010 MO2014 IA2012 IA2009 CT2012 TN2006 PA2010 MA2009 CT2010 IA2010 MA2015 PA2009 MO2011 OH2015 MO2002 CT2011VT2002 CT2009 CT2015 WA2014 MN2015 MI2015 MN2010 MO2010 AL2010 AL2009 MO2007 MO2006 NC2009 MA2012 MA2013 WA2009 WA2015 MA2011 AL2015 MN2009 AL2014 CT2014 WA2013 PA2011 MN2013 MA2014 WA2011 MO2000 NC2010 PA2012 CT2007 MN2007 IA2005 MO2001 NC2015 AL2012 TX2009 NC2008 WV2009 PA2008 IA2008 MO2003 NY2015 CT2013 TX2006 MN2012 MO1997 NC2007 MA2008 AL2011 AL2013 WA2012 CT2006 NC2006 NC2014 NY2010 IA1996 NC2002 MN2008 MA2007 MA2006 CT2008 PA2007 IA2000 PA2005 NC2001 MI2014 TX2003 TX2010 IL2015 MI2009 WV2012 AL2003 MN2014 IA2013 IA2004 NY2009 NC2003 NC2011 RI2015 GA2006 MO1998 NC2013 MD2013 WA2008 MN2011 IA2014 MD2010 MO1999 OH2010 NH2008 OH2013 NC2012 VT2014 UT2010 TX2005 CA2009 MI2010 WV2010 NC2004 IA2002 VT2006 NY2007 PA2006 TX2007 TX2015 NY2011 PA2004 PA2003 OH2014 MD2012 CT2003 CT2005 IA2006 MA2005 WI2009 NC2005 SC2015 WA2007 MI2012 IL2007 WA2006 NY2013 AL2002 GA2015 MI2007 NH2007 NH1999 MD2011 TX2008 NY2004 NY2014 TX2002 GA2007 OH2012 CA2010 TX2001 NY2012 CT2004 NY2008 NY2006 TX2004 GA2008 MI2011 PA2002 OH2011 IA1998 IA2007 UT2011 AL2008 CA2015 NE2009 TX2011 MI2008 OH2009 AL2005 CT2001 GA2005 TX2012 WA2003 TN1999 TN2001 CT2002 AL2004 MA2004 MA2003 IL2009 NY2005 IL2013 IL2006 WA2005 IA2003 NH2006 NY2003 IL2012 IA1993 MI2013 TN2000 UT2012 WI2010 PA2001 NE2008 IL2010 NE2010 MD2008 NC2000 IA1988 GA2010 UT2003 TX2013 OH2008 OH2007 KS2014 MD2015 MD2009 CA2008 IL2005 NH2004 KS2015 IN2007 IN2014 IA1990 NH2009 MI2003 MI2006 KS2008 MI2001 KY2006 KY2003 MO1996 IA1992 WA2001 GA2003 MI2002 WI2004 OH2006 MA2002 NE2014 IL2011 CA2011 TX2000 GA2014 GA2009 NY2002 RI2012 SC2010 MI2005 TN1997 IL2008 WA2004 NH2010 WI2015 MN2006 MN2005 FL2015 IA1999 RI2010 CA2012 IA1995 MO1995 CA2007 MN2002 WA2000 WI2011 VT2010 TX2014 WA2002 GA2013 MI2000 OH2005 NY2001 IN2006 IL2014 IL2003 IA1991 GA2004 GA2012 IN2012 MA2001 VT2005 MS2015 MN2003 MI2004 WI2014 NH2001 IA2001 WI2005 AL2006 CA2013 KY2005 AL2000 IN2015 NC1999 CA2006 IA1989 SC2009 MN2001 OR2015 OR2009 VT2013 UT2009 MD2014 NE2015 IA1985 CT2000 FL2010 WI2007 RI2011 CA2014 AL2001 CA2005 IA1987 FL2009 VT2011 LA2014 GA2011 MI1999 CO2015 OH2002 KY2015 KS2007 LA2005 MN2004 CO2012 WI2000 AL2007 IN2005 NY2000 LA2015 FL2012 TN1995 OH2004 WI2013 CO2013 KY2008 IN2013 WI2003 IN2008 PA1999 MD2004 KY2009 NC1995 VT1998 NH2003 AR2002 OK2015 LA2012 CA2004 KS2010 ME2009 TN1996 CT1997 IN2009 CO2010 CO2014 OR2010 MO1994 WY2007 AL1996 WA1996 MS2014 HI2005 WI2006 AR2011 LA2011 LA2006 VT1999 CT1999 FL2011 PA2000 IA1997 RI2014 CA2003 MD2007 NM2012 IN2002 WI2012 TN1998 CT1984 OH2003 TX1999 OR2014 IL2002 MS2013 RI2008 LA2010 SC2008 SC2007 AL1997 SC2013 TN1994 WI2002 FL2008 PA1998 NH2000 FL2014 IN2010 WA1997 VT2009 NY1999 WY2006 KY2010 NC1991 IL2004 WI1992 WA1992 WA1998 VT2015 RI2009 RI2013 MD2002 WV2011 MD2005 MS2012 IN2004 IN2003 SC2011 NH2011 CO2011 MN2000 WA1999 MA2000 WI1995 KS2011 ME2010 AL1999 OR2004 KS2013 LA2007 GA2002 OK2010 WV2013 OK2009 AL1998 LA2003 MN1999 NE2007 LA2013 VT2004 WI1993 KS2012 ME2013 HI2008 CA2002 KS2009 NC1996 KY2001 TN1993 CT1998 CT1982 WI2008 NE2013 MD2006 NM2009 FL2006 NH2015 PA1997 FL2013 OH2001 TX1996 SC2014 WY2012 AL1993 WA1994 IA1984 MA1999 OR2003 OH2000 NE2006 RI2006 IN1998 NC1990 IN2001 WA1995 CO2009 MI1997 IN2011 VT2001 WI1996 MN1996 ME2014 IA1986 LA2004 NH2014 WI1999 KS2005 KY2007 MN1994 OR2008 NH2013 WV2014 NH1991 LA2009 MD2003 VT2003 WI1994 CA2001 OR2013 FL2007 RI2003 AR2012 AR2010 CT1985 KS2002 SC2005 MS2010 SC2003 ME2015 KY2014 RI2007 DE2014 NE2011 WA1993 TX1998 NC1998 FL2005 MO1990 WA1989 RI2004 NY1998 IN2000 NC1993 SC2012 PA1993 VT2000 WI2001 WI1998 NE2012 NC1994 PA1994 CT1983 MO1991 VT2007 NC1992 SC2006 CT1994 SC2004 LA2008 MI1996 ME2004 NM2013 OH1999 NY1997 OR2011 MN1995 OR2005 MN1998 NC1988 CT1995 ME2011 IL2001 MI1994 CT1989 WV2015 CT1988 MN1993 NC1997 PA1995 NM2015 NM2014 CT1981 WI1997 MN1997 CO2004 PA1996 OH1998 KY2004 AZ2008 IA1982 CA2000 MN1982 TX1997 MN1991 CT1991 CO2002 TX1995 TN1992 TN1991 NM2010 OR2002 NE2001 AL1995 AL1994 WV2006 AZ2012 IL2000 NH2002 WA1990 MO1993 VA2009 AR2009 MD2001 CT1993 AZ2007 GA2001 NY1996 RI2005 ME2012 MN1984 NH2012 GA1999 MI1991 MI1998 OK2007 OR2006 ME2007 AZ2010 AZ2009 KS2003 CT1996 CO2005 NE2003 MA1998 NC1989 AZ2013 HI2007 MI1992 OR2007 MA1995 NY1995 OK2014 HI2002 VT2008 IN1999 NJ2009 IL1999 WI1991 CT1992 CA1999 NE2005 KS1998 ND2014 SC2002 NV2012 AR2007 KY1997 OR2001 MA1996 IA1983 HI2004 KS2006 FL2004 FL2003 CO2003 OK2011 AL1988 KS2004 MN1990 OH1997 VA2010 LA2002 NY1992 WI1989 CO2008 CO2007 CO2006 PA1992 NY1994 KS2001 MS2009 AZ2005 MN1992 IL1996 OR2012 VA2007 VA2005 IN1994 KY2011 AZ2015 NV2010 AZ2011 NY1993 UT2002 HI2001 NE2002 MA1997 TN1990 NM2011 MI1995 KY2000 MN1989 CO2001 UT2004 MS2011 WV2005 VT1982 AR2004 IN1993 AR2014 CT1987 NJ2014 KY2002 ME2005 ME2008 VA2008 MI1989 NV2013 MO1992 MN1987 WY2011 WA1991 VT2012 OH1996 AL1992 GA2000 CA1996 NY1991 FL2002 IN1997 HI2014 RI2002 NE1999 KS2000 AR2013 TX1994 KY2012 NJ2015 NV2009 MI1990 MA1994 WV2003 CA1998 NC1984 PA1991 NM2008 VA2006 VT1991 KS1994 IN1995 NC1987 CA1995 WI1990 CT1990 WV2002 CA1992 UT2000 IN1996 TX1993 AZ2006 WV2004 AR2001 KY2013 NJ2010 SC2001 ME2002 MA1992 IL1997 MN1985 DE2010 WV2008 ME2003 GA1997 WI1987 NY1989 NM2003 IA1979 AZ2004 OK2013 OH1995 AL1991 MI1993 CA1994 NY1990 MO1989 VA2015 AZ2003 OK2012 DE2012 NC1986 IN1991 OH1994 NJ2011 AZ2014 NV2014 VA2014 MN1986 WI1986 AL1987 DE2015 ME2001 MA1991 KY1999 MA1993 IL1995 CA1997 HI2003 NE2004 KY1995 ME2006 FL1999 WV2001 CA1990 WY2009 HI2006 CA1993 NJ2007 NY1988 WI1984 WY2010 NE2000 MO1988 OH1993 AL1989 TN1989 TX1992 WA1988 MA1990 AR2008 IL1998 NY1987 VA2011 LA2001 DE2009 OK2004 CO1994 IN1989 VT1996 MN1988 OK2005 OK2008 OR1999 IN1990 MN1981 MD2000 WI1983 CA1989 UT2001 TX1991 CA1991 PA1990 OR2000 NH1981 KY1998 WI1988 WI1985 WI1982 HI2015 OK2006 TX1988 AL1990 NJ2012 WA1987 MA1989 AR2006 FL2001 CT1986 OR1995 AR2003 IL1994 AZ1994 MO1987 MI1988 OH1992 VA2012 NH1997 RI2001 ND2013 OK2003 TX1990 MA1988 AZ2001 NM2007 WI1978 ID2010 MA1982 GA1996 WV2007 SD2009 NJ2006 PA1987 VA2004 VA2002 WI1981 TX1987 IN1992 NV2011 AZ2002 CA1988 MN1983 CO1992 NC1982 NV2008 WI1980 NH1992 IN1981 CO2000 NY1986 CT1980 SD2015 NJ2013 TX1989 LA1999 VA2013 MO1985 RI2000 DE2013 GA1994 GA1998 CT1979 VA2003 TN1983 MA1985 MA1987 WA1982 WA1981 AL1985 MO1986 HI2000 AR2000 NV2015 PA1989 TN1988 TN1987 KS1997 NH1994 GA1992 WA1980 WI1979 NM2005 AZ1993 IN1988 CO1999 NJ2008 AZ1997 NY1984 NM2002 TN1986 WY2008 KS1996 ID2009 NJ2004 GA1995 WA1986 NY1985 AZ2000 MI1982 GA1993 SC1998 PA1986 FL2000 IA1978 RI1999 RI1998 CA1987 VT1984 DE2008 NJ2005 VT1997 NY1983 UT1999 UT1998 MA1984 SC2000 SC1999 OR1998 FL1998 LA1998 IL1992 CO1995 ND2009 PA1988 NC1983 TN1985 OH1991 PA1985 NV2007 DE2011 NC1985 KY1987 WA1985 MI1983 HI2013 UT1997 NJ2003 MI1987 DE2002 VT1988 OR1996 SC1997 CO1998 LA1996 DE2007 VT1994 NM2004 LA2000 AL1983 IN1987 OK2002 AZ1992 VT1993 NY1982 MT2009 IL1993 NH1998 NH1987 ID2015 NH1990 AK2010 SD2014 MA1981 VT1987 MA1986 DE2003 OR1993 VT1983 MN1980 MD1999 ID2004 MI1984 VA1982 AZ1990 MT2010 WA1984 NH1986 NE1998 ID2011 ND2005 CO1996 OK2000 MA1983 VT1977 MD1998 RI1995 VA2001 ME2000 MO1982 NM2006 TN1979 DE2005 WA1979 AZ1999 MT2011 VT1992 WI1977 TX1986 TX1985 CO1993 GA1991 RI1997 KS1992 TX1984 MD1997 AZ1998 CO1991 RI1996 OR1997 LA1997 OR1992 TN1984 MO1984 PA1984 CA1986 CA1985 CO1997 KY1994 NC1980 CA1983 N C1981 AZ1996 SC1995 MI1986 MI1985 WV1994 IA1980 NM2000 AZ1995 OR1994 ND2011 NM2001 MO1981 DE2004 KY1996 OH1989 OR1990 MD1996 SD2012 NM1994 MO1980 ND2015 MS2006 IN1982 MI1980 CA1984 UT1980 MT2014 MA1980 AR1998 AL1984 IL1991 WA1983 OH1988 TN1980 TX1983 OH1990 AR2005 MD1980 NH1995 FL1994 AL1986 AZ1991 NM1988 OR1989 NY1981 CA1982 MO1983 CT1978 IA1981 GA1989 ND2012 OR1991 MD1981 MS2005 ND2010 MI1981 AR2015 IN1985 SC1994 KS1991 TX1982 TX1981 CA1981 PA1982 NH1985 NE1997 KS1999 FL1993 SC1996 MS2008 AK2015 IL1987 OR1988 MD1982 LA1994 TN1982 PA1983 NH1996 FL1992 OK1999 NH1982 NJ1996 NE1994 FL1991 AZ1989 VT1995 RI1994 RI1993 CA1980 NH1988WY2004 AZ1987 UT1992 CO1990 KS1995 ID2012 NE1992 MT2003 MA1979 SD2011 WV1993 MD1994 ND2007 MT2013 HI1995 NV2006 FL1988 FL1996 NJ2002 NJ2001 NY1980 UT1989FL1987 AZ1982 AK2014 CO1989 MT2012 NV2005 OK2001 VT1986 NM1999 NE1996 ID2014 AZ1988 FL1997 UT1991 FL1989 SC1993 NH1980 IL1988 MN1979 DE2006 TX1980 IL1990 GA1990 OH1987 IN1984 TN1978 MN1978 CA1979 WV2000 TN1981 MD1995 PA1981 NE1993 FL1995 MT2005 MT2007 KY1992 IL1989 ID2003UT1979 AK2009 MA1978 RI1991 IN1980 NJ2000 MD1993 NE1995 SC1991 OK1998 MS2007 KS1982 ID2013 CO1988 AK2012 SD2013 RI1992 KY1993 LA1992 TX1979 SD2008 AL1981 MD1992 MO1979 AK2011 FL1990 GA1987 KY1985 OK1997 WA1978 MO1977 IN1986 DE2001 MD1984 CT1977 VT1990 NV2002 MS2004 HI1999 NM1998 AK2013 LA1995 MI1979 ME1999 OR1987 NC1979 WI1976 MD1990 NM1997 ND2006 SC1992 NH1993 IL1985 ME1991 AR1999 CA1978 NH1989 NH1984 NY1979 MD1991 RI1989 MD1988 PA1980 AZ1986 LA1991 HI1994 WV1999 IL1984 CO1987 ND2004 GA1988 KS1993 UT1990FL1986 SC1986 MT2008 MA1977 KS1990 CO1984 VA1983 OH1986 OR1985 NM1991 MO1978 ID2002AZ1985 CO1980 NC1978 LA1989 WV1992 AL1982 VA1987 LA1987 NE1991 HI1998 IL1986 VA2000 LA1993 LA1990 AR1996 AR1995 NH1983 AL1978 KY1991 IA1977 MO1976 AZ1984 VT1979 MD1983 NJ1997 VA1981 NM1992 MD1989 VA1989 IN1983 AL1979 IN1978 IN1977 NJ1999 NV2003 CT1976 RI1990 SD2010 MD1978 NM1990 ND2008 GA1984 OH1985 MT2004 HI1997 AL1980 IL1982 NM1995 MD1979 OH1984 GA1986 KY1988 MT2002 MS2002 CO1982 IL1983 CO1986 ME1996 NM1993 HI1996 WV1995 ND2001 KY1986 AK2004 CT1975 CO1983 VT1985 NV2000 NJ1998 VA1980 IL1981 TX1978 CO1985 MA1976 IN1979 CA1977 OH1982 NC1976 WY2005 NM1987 VA1999 DE2000 NV2004 PA1978 NE1988 GA1985 OR1986 WV1991 NM1989 UT1978 SC1990 NE1990 WA1976WA1977 VA1984 FL1985 ME1994 NJ1991 KY1989 VT1989 NM1986 MT2006 NJ1995 NJ1989 FL1983 MD1987 NY1978 TN1976 NJ1994 PA1979 HI1992 MI1978 LA1988 GA1983 IA1976 NJ1990 OH1981 KY1983 NV2001 NJ1992 NM1996 H I1993 NV1994 VA1988 ME1993 OK1984 VA1990 OK1991 TN1977 KS1980 OR1981 AK2005 WV1998 MS2003 MS1999 VT1981 AR1994 MD1986 IL1980 DE1998 RI1988 OH1983 WY2003 OK1996 SC1987 FL1982 CO1981 VA1997 PA1976 CO1978 WV1997 ID2005 WY2002 WY2001 OK1994 ND2003 SD2006 GA1982 KS1983 CA1976 AR1997 SC1985 MT2000 CT1974 MT2015 AR1992AR1993 MS2000 FL1984 NY1977 RI1987 NE1987 KS1989 SC1982 VA1991 NY1976 NJ1993 KS1988 OK1993 AK2007 NH1978 MD1985 OK1995 LA1985 OK1992 MD1977 ME1998 NC1977 KS1985 VA1985 VA1993 OR1982 VA1998 OH1979 KS1984 MN1977 MO1975 SC1989 VA1979 RI1986 KY1984 NE1986 TX1976 ND2002 VA1996 KS1981 NH1979 HI1989 ME1997 OR1983 PA1977 KS1986 ME1990 ID2008 SC1988 AZ1983 OR1980 WV1996 OR1978 KS1987 AK2008 NH1977 ID2006 VA1986 AZ1981 SD2005 VA1995 KY1990 MT2001 HI1990 VA1978 OR1984 NJ1988 VA1992 VA1994 OR1979 OH1980 AK2002 HI1991 MWV1990 I1977 IL1979 IL1978 NC1975 OK1985 NV1996 NJ1987 CO1979 TX1977 AK2006 MS1995 NM1984 NM1979 OK1988 ID2007 UT1976 GA1979 SD2004 MI1976 GA1980 IN1976 NE1989 CO1977 KY1982 SC1983 NC1974 LA1982 MD1976 LA1986 VT1978 KS1979 NJ1981 RI1985 OK1990 MS2001 GA1981 VT1980 NV1999 MS1997 NE1985 NY1975 AR1991 MS1998 NC1973 AK2003 MA1975 LA1984 CO1976 WY1992 MN1976 LA1983 OK1987 WI1975 NM1985 NV1998 VA1977 OH1978 ND1999 WI1974 FL1980 IL1977 WY1991 IA1975 CA1975 MA1974 AZ1978 HI1987 OK1986 NH1976RI1984 ME1995 OR1977 PA1975 MO1974 AZ1980 AL1977 DE1999 NJ1980 IL1976 FL1981 UT1977 NM1978 NV1995 SD2007 NE1984 MS1991 MS1996 WI1973 MS1994 NM1981 CA1974 NJ1982 WY1993 MI1975 TN1974 OK1989 ME1992 FL1979 NY1974 SC1984 MD1975 SD2003 IA1974 NV1991 AK2001 FL1976 AK2000 GA1978 SD2001 WA1975 FL1978 DE1997 WY2000 AZ1979 ND2000 DE1994 HI1988 LA1981 PA1974 NV1992 CT1973 DE1992 WV1987 MS1989 KS1978 ID2001 ME1985 HI1986 FL1975 TN1975 TX1975 RI1982 MO1973 OK1983 NJ1985 NJ1979 NJ1984 RI1983 ME1988 MT1999 NV1997 LA1980 MD1973 TN1973 NM1982 RI1981 MD1974 SD2002 DE1996 NM1983 WV1984 NV1990 NV1989 OH1977 AR1990 MS1993 NY1973 MA1973 NJ1983 VA1976 AZ1977 HI1985 CA1973 TX1974 MT1994 MS1992 WV1983 KY1981 NV1993 ND1996 SD2000 WA1974 ME1989 SC1981 OK1982 MS1990 OR1976 UT1974 GA1977 IL1975 KY1980 FL1977 VT1975 NM1980 MT1998 IN1975 NJ1986 SC1980 MT1995 MS1988 KY1979 OH1976 DE1979 LA1979 UT1973 MT1996 HI1976 AR1988 AR1987 NE1982 MI1974 VT1976 KS1974 NE1980 IA1973 WA1973 IN1974 AR1989 NE1979 HI1981 DE1995 WV1989 NM1977 DE1990 RI1980 RI1979 WV1986 ME1984 NE1981 ME1983 LA1978 IN1973 NV1987 NE1983 KS1976 TX1973 ME1987 MT1997 SC1979 WV1985 OK1980 IL1974 NV1988 MS1987 OK1981 WV1988 WV1982 KS1975 UT1975 PA1973 NJ1978 MN1975 MS1982 ND1995 WY1995 MT1993 DE1988 ND1997 DE1986 HI1980 CO1974 MN1974 HI1983 KY1978 MN1973 MS1981 MS1984 AK1995 DE1987 DE1993 DE1983 HI1979 AZ1976 MS1986 HI1982 SC1978 DE1991 VA1975 FL1974 NJ1976 MI1973 OK1979 NH1975 CO1975 IL1973 HI1984 MT1991 DE1980 AR1986 ND1998 KS1977 AR1985 MS1980 ME1986 HI1978 OH1975 KY1976 DE1978 LA1977 NV1986 WY1994 MT1990 DE1982 MS1983 MS1985 OR1975 OH1974 NV1984 DE1984 OR1974 OK1978 VA1973 NV1982 ID1999 RI1978 NJ1977 NM1976 HI1977 OK1977 NH1974 VA1974 MT1987 MT1992 FL1973 NH1973 MS1979 ID2000 WY1989 MT1988 KY1977 ID1997 LA1976 WY1997 ME1982 ME1978 AK1994 AK1999 KS1973 AK1998 SD1982 CO1973 NE1977 WY1990 RI1977 AR1983 SC1977 DE1981 ME1981 HI1974 RI1976 AR1982 WY1987 ND1988 OH1973WV1980 SD1997 ME1980 ND1993 MT1989 LA1974 WV1976 WV1981 WY1988 AK1997 AR1984 MS1978 ND1982 WY1996 HI1973 WY1998 ID1998 ME1976 DE1985 OR1973 LA1973 SD1994 NE1976 AK1996 NJ1975 DE1989 NE1978 WV1974 HI1975 WY1986 VT1974 AR1980 KY1975 AR1981 ID1996 SC1976 AK1993WY1976 NJ1974 NV1980 ME1979 MS1977 VT1973 NV1983 DE1976 SD1999 DE1977 MT1986 SD1998 SD1992 LA1975 WY1999 WV1979 ND1991 AZ1975NV1975 SD1993 WY1984 AZ1974 WV1978 ND1987 ME1977 AK1992WV1977 ND1992 ID1994 NM1975 NV1979 ID1993 NV1976 SD1985 ID1995 AR1976 ND1994 AK1988NV1985 WY1985 WV1975 ND1985 ID1990 ND1989 RI1975 AK1991 NV1978 SD1983 SC1975 KY1974 ID1989 NJ1973 NM1974 MT1983 RI1973 NV1977 NE1975 RI1974 ND1984 ND1986 NV1981 WY1979 AR1977 SC1974 ID1991 MT1984 SD1991 ID1992 MT1985 AR1979 ND1983 SD1996 AZ1973 AK1989 SD1988 KY1973 MT1982 MT1981 SD1989 NE1974 ND1981 WV1973 ME1975WY1978 AR1975 ME1974 SD1995 SC1973 ND1980 SD1987 AR1974 AR1973 AR1978 DE1974 WY1977 WY1983 WY1981 MT1980 MT1979 AK1990AK1976ND1990 SD1981 SD1984 AK1978 SD1986 SD1977 NE1973 AK1987 WY1982 ME1973 AK1981 DE1975 NM1973 SD1990 ND1977 NV1974 ID1977 WY1980 ND1978 SD1978 ID1982 AK1977 DE1973AK1975 SD1979 AK1984 ND1979 AK1983 MT1977 AK1986 ID1974 ID1984 SD1980 ID1976 ND1974 AK1982 ID1983 WY1975 SD1976 ND1976 WY1974 AK1980 NV1973 AK1985 ID1978 AK1979ID1979 MT1975 ID1981 ID1975ID1980 MT1976 MT1978 SD1975 WY1973 SD1974 MT1974 ND1975 MT1973 ND1973 SD1973
UT2005 UT2007 UT2008 UT2006 HI2009 WY2013 HI2011 HI2012 HI2010 WY2014 WY2015
14
l_mean_cost
NY2004 NY2003 NY2002 NY2001
12
20 18 16 14 12
80
18
CA2009 CA2010 CA2015 CA2012 CA2011 CA2014 CA2013 CA2008 NY2010 MA2010 CA2007 CA2006 MA2009 NY2009 CA2004 CA2005 MA2012 NY2015 MA2011 NY2014 MA2015 CA2003 MA2014 NY2012 NY2011 MA2013 CA2002 NY2013 MA2008 MA2007 MD2010 MA2006 MA2005 CA2000 NY2007 MD2012 NY2005 MD2011 MD2009 NY2008CA2001 NY2006 PA2015 PA2010 PA2009 MD2013 CA1999 PA2014 PA2012 CA1998 PA2011 MD2015 TX2010 TX2009 MA2000 NY2000 MD2008 PA2013 CA1997 NC2009 PA2008 CA1996 PA2007 PA2005 NC2010 PA2006 CA1995 MA1999 PA2004 MD2014 TX2012 PA2003 TX2011 CA1994 NY1999 TX2015 WA2010 PA2002 NC2012 NC2011 CA1993 NC2013 MA1998 TX2003 TX2014 TX2013 NC2015 CA1992 WA2009 NY1998 TX2008 TX2005 TX2004 NC2014 IL2009 TX2007 TX2006 PA2001 IL2010 MA1997 NC2008 MD2004 IL2015 CA1991 MD2005 NY1997 WA2012 WA2011 WA2014 MA1996 CA1990 MD2006 TX2002 OH2010 MD2007 NY1996 IL2012 CA1989 OH2009 PA2000 WA2013 WA2015 NC2007 NC2006 NY1995 NY1994 MD2003 NY1992 NC2005 MA1995 IL2011 MI2009 NY1993 MI2010 MA1994 NY1991 TX2001IL2013 IL2014 CA1988 NC2004 MD2000 PA1999 OH2015 MD2002 MA1993 NY1990 WA2008 WA2006 NC2003 MA1992 OH2012 CA1987 MI2015 NY1989 WA2007 OH2011 IL2008 TX2000 WA2005 MA1991 FL2010 OH2013 MD2001 MI2011 WA2004 NY1988 MI2012 PA1998 OH2014 FL2015 IL2007 GA2010 WA2003 IL2005 MA1990 MD1999 NY1987 FL2009 IL2006 GA2009 CA1986 PA1997 NC2002 GA2015 MD1998 MI2014 OH2008 IL2003 MA1989 MI2013 CA1985 IL2004 FL2012 MO2009 TN2009 WA2002 MN2010 MO2010 TX1999 OH2005 CT2010 TN2010 MN2009 OH2007 OH2006 FL2014 PA1996 GA2014 FL2011 GA2012 CT2009 VA2009 MD1997 NC2001 NY1986 GA2011 MN2015 MI2008 MA1988 PA1995 NY1985 OH2004 WI2009 NC2000 GA2013 FL2013 MN2012 OH2003 WA2001 MN2014 MA1987 MO2015 MN2011 MN2013 IL2002 MI2007 TN2015 CT2015 TN2012 WI2010 PA1994 MI2006 MI2005 MD1994 TN2011 TX1998 MO2012 MD1996 CA1984 MO2011 OH2002 CT2012 CT2011 PA1993 MO2014 PA1992 TN2014 VA2010 WA2000 MI2004 TN2013 CT2014 MI2003 CO2010 NY1984 TX1997 IL2001 MO2008 CT2013 NC1999 MO2005 WI2015 MA1986 MD1993 MD1995 MO2003 TX1996 CT2004 CT2008 MO2004 MD1992 CT2007 MO2007 MA1985 PA1991 IL2000 CO2009 MO2006 CA1983 MO2013 NY1983 MN2008 WA1999 WI2011 CT2006 MI2002 OH2001 TX1994 TN2008 CT2005 MD1991 TX1995 MN2007 MO2002 NJ2009 WI2014 TN2007 WI2012WI2013 OH2000 VA2015 VA2013 NJ2010 TN2005 OR2010 NC1998 IL1999 MI2001 VA2012 VA2011 CO2012 TN2006 OR2009 CO2015 PA1990 TX1993 CT2003 AL2009 GA2008 NC1997 NY1982 CA1982 MN2006 MD1990 NY1981 CA1981 AL2015 MN2005 TN2004 WI2008 NC1996 MN2004 MO2001 CT2002 AL2010 CO2011 VA2014 PA1989 WA1998 OR2015 NY1980 TN2003 MI2000 MN2003 OH1999 CA1980 CO2013 TX1992 WA1997 MA1984 NC1995 WI2004 FL2005 CO2014 WI2007 WI2006 GA2007 GA2004 MO2000 WI2005 WI2003 CT2001 GA2006 GA2005 FL2008 OR2012 MD1989 NC1994 OR2014 PA1988 NJ2012 OR2011 MN2002 CA1979 MI1999 FL2004 TX1991 IL1998 NY1979 OR2013 FL2007 GA2003 FL2006 WA1996 WA1995 CO2008 NC1993 NJ2011 OH1998 TN2002 VA2008 VA2005 NJ2013 CO2005 IN2009 PA1987 WI2002 MA1983 IL1997 MD1988 FL2003 IN2010 AL2011 CT2000 CO2007 CO2003 NJ2014 MI1998 CO2004 VA2004 NC1992 GA2002 MN2001 MO1999 AL2014 AL2012 OH1996 TX1990 OH1997 CO2006 NY1976 NJ2015 CA1978 IA2015 MI1997 IL1996 WA1994 IN2015 TX1989 IA2009 VA2007 VA2006 CT1999 VA2003 NY1978 WA1992 AZ2010 OR2008 NC1991 IA2010 FL2002 CO2002 MA1982 OR2006 IL1995 AL2013 AL2003 WI2001 MA1981 MI1996 MN2000 TX1988AL2005 OR2007 OR2005 TN2001 CA1976 MO1998 PA1986 AZ2009 MD1987 NY1977 WA1993 MI1995 PA1985 VA2002 GA2001 OR2004 IL1994 AL2008 OH1995 NJ2004 NC1990 MI1994 CO2001 TX1987 NY1975 NJ2007 NJ2005 CA1977 IN2014 SC2009 CT1998 WA1991 WI2000 MA1980 NJ2008 UT2010 UT2009 IN2012 NJ2003 KY2009 AL2004 MI1993 SC2015 OR2003 MO1997 IL1993 MN1999 IN2011 OH1994 NC1989 NJ2006 CT1997 FL2001 IL1992 LA2012 AL2002 IA2011 AL2006 CO2000 AZ2013 IA2012 AL2007 KY2015 LA2010 LA2009 KY2010 CT1996 SC2010 AZ2011 GA2000 MI1992 OR2002 MD1986 VA2001 TN2000 WA1990 NY1974 IN2013 MO1996 CA1975 AZ2012 PA1984 OH1993 MA1979 IL1991 OH1992 CT1995 VA2000 MD1985 FL2000 TX1986 CT1994 CA1974 WI1999 NC1988 MI1991 UT2012 UT2011 LA2011 IN2008 AZ2014 IA2013 MO1995 MN1998 RI2009 RI2015 IN2006 NJ2002 AL2001 IL1990 TX1985 IN2007 CT1993 OH1991 OR2001 CT1992 IA2014 IL1989 IA2008 RI2010 SC2014 CO1999 WA1989 MO1994 VA1999 MN1997 KY2012 MN1996 MA1978 TN1999 PA1983 SC2012 SC2011 LA2013 AL2000 IN2004 MN1995 IA2007 NC1987 IN2005 AZ2015 KY2014 LA2014 KY2011 NM2009 IL1988 CO1998 MA1976 GA1999 NJ2000 MI1990 CT1991 IA2004 NJ2001 IA2003 WI1998 LA2015 IL1987 OR2000 IA2006 IA2005 MN1994 IN2003 WA1988 OH1990 FL1999 KY2013 MO1993 NM2010 AL1999 CO1997 CT1990 SC2013 MI1989 MO1992 WI1997 RI2011 PA1982 MO1991 MA1977 NY1973 OH1989 AZ2005 CT1989 TN1998 RI2012 AZ2007 LA2005 MN1993 TX1984 IA2002 PA1981 RI2013 WI1996 KS2009 IN2002 VA1998 OR1999 WI1995 AZ2006 GA1998 MO1990 TN1997 CO1996 KY2005 AZ2008 PA1980 LA2006 CA1973 MN1991 UT2008 RI2014 CT1988 CO1994 MI1988 IL1986 FL1998 NC1986 VA1993 OH1988 WI1994 IL1985 IA2001 OK2009 NE2010 KS2010 AL1998 TN1996 AZ2004 LA2003 CO1995 WA1987 LA2004 VA1997 KS2015 CT1987 KY2008 MD1984 MN1992 NJ1999 AZ2003 MI1987 OK2015 VA1991 VA1996 WI1993 AL1997 MN1990 VA1995 VA1994 AL1996 OH1987 GA1997 TN1995 FL1997 NC1985 NM2012 NM2011 MO1989 PA1979 RI2007 TX1983 TN1994 CO1993 NE2009 NH2009 OK2010 RI2008 VA1992 KY2006 IN2000IN2001 WI1992 VA1990 KY2007 KY2004 MA1974 TN1993 NE2015 KS2011 MN1989 KS2014 LA2007 MA1975 NH2010 IA2000 UT2005 UT2003 FL1996 FL1994 KS2012 SC2008 AZ2002 UT2007 UT2006 MO1988 AL1995 TN1992 LA2008 WA1986 NJ1998 OR1998 FL1995 TX1982 IL1984 CT1986 CO1992 NM2014 WI1991 RI2004 UT2002 RI2006 RI2005 VA1989 GA1996 TX1981 MN1988 IA1999 PA1978 UT2004 PA1976 SC2007 WA1985 AL1994 NM2013 RI2003 OK2014 AR2009 TN1991 NH2015 MN1987 AR2010 FL1993 CT1985 IN1999 SC2005 MD1983 NM2015 OK2013 NE2014 OH1986 GA1995 GA1994 KY2003 WI1990 IL1983 MO1987 NM2008 OK2012 NE2012 SC2004 NH2014 SC2006 SC2003 TX1980 ME2009 NJ1997 OR1997 UT2001 FL1992 OH1985 GA1993 LA2002 WI1989 AZ2001 KS2013 CO1991 NH2013 NE2013 MI1986 TN1990 MD1982 NH2012 NC1984 VA1987 PA1977 FL1991 MI1985 AL1993 NE2011 NJ1996 NH2011 MN1986 UT2000 OK2011 IN1998 ME2010 AZ2000 RI2002 CT1984 ME2012 MA1973 CO1990 IA1998 IL1982 TX1979 HI2009 AL1992 KY2002 IL1981 VA1988 GA1992 MN1985 AL1991 IN1997 TN1989 OR1996 NJ1995 WI1988 WI1987 NM2007 SC2002 FL1990 MD1981 FL1989 MO1986 ME2011 OR1995 IL1980 HI2010 VT2009 MD1980 NH2004 NH2005 IN1996 KS2008 PA1975 PA1974 NJ1994 AZ1999 NH2003 TN1988 OR1994 WA1984 GA1991 NJ1993 ME2014 CT1983 AR2011 NC1983 NM2006 AR2012 MO1985 A AL1990 L1989 MS2010 ME2015 ME2013 CO1989 OH1984 IA1997 FL1988 IN1995 KY2001 NJ1991 NH2006 RI2001 VT2010UT1999 IA1996 NJ1992 IL1979 AL1988 NH2007 TX1978 NM2004 NM2003 MI1984 NC1982 NM2005 TN1987 MS2009 NJ1990 IN1994 OR1993 WI1986 MD1979 WA1983 CO1988 KS2007 GA1990 MS2013 AZ1998 CT1982 NH2008 OR1992 IN1993 MT2009 LA2001 HI2011 OK2004 MN1984 IA1995 NM2002 NC1981 FL1987 IA1993 GA1989 IA1994 MS2015 KY2000 WI1985 CT1981 OH1983 NE2008 IN1992 SC2001 HI2005 VA1985 KS2006 NJ1989 NH2002 TX1976 RI2000 LA2000 IL1978 HI2012 OK2007 AR2013 AL1987 IA1992 NC1980 MO1984 UT1998 NE2006 IL1976 KS2003 VA1986 MS2012 AZ1997 MI1983 TX1977 CO1987 AR2014 KS2005 TN1986 OR1991 CT1980 MT2010 LA1999 VT2014 NE2005 OK2005 OK2008 OK2003 IN1991 KS2002 IA1991 IL1977 WA1982 WA1981 WV2010 WA1980 HI2014 NE2007 AZ1995 KS2004 MS2014 NM2001 NE2004 MN1983 VT2013 OK2006 TN1985 WV2009 ME2004 MI1982 MI1981 VT2011 AZ1996 AZ1994 MS2011 ME2003 IA1990 NC1979 CT1979 MD1978 OH1981 MO1983 VT2012UT1997 MD1976 OR1989 VT2003 HI2004 NJ1988 MN1982 OR1990 OH1982 AZ1993 GA1988 ME2006 AZ1992 PA1973 WI1984 MO1981 VT2004 WV2012 NV2010 HI2008 ME2002 ME2007 VT2007 VT2005 IN1990 MN1981 MN1980 MI1980 LA1998 MO1980 OH1980 VT2008 CO1986 NE2003 ME2008 ME2005 HI2007 NH2001 HI2013 OK2002 IA1989 MD1977 MO1982 NM2000 HI2015 NC1976 TX1975 FL1986 GA1987 AZ1991 NV2009 IA1988 VT2006 IL1974 KS2001 WA1979 MT2011 VT2015 AL1985 NJ1987 SC2000 LA1997 OH1979 NE2002 DE2009 IN1989 VT2002 IL1975 DE2014 AR2005 WI1983 CO1985 DE2010 AL1986 NC1978 UT1992 FL1985 AR2006 RI1999 AR2015 ME2001 LA1994 AR2007 LA1992 KY1999 AZ1990 VA1982 TN1984 MO1979 WI1981 AR2008 CT1978 TX1974 IN1988 VT2001 LA1996 LA1991 MI1979 UT1991 NH2000 OR1988 WA1978 WI1982 DE2012 IA1987 AR2004 LA1993 NV2014 TN1983 GA1986 MN1979 LA1995 HI2006 VA1983 IN1987 MT2014 NV2012 KS2000 RI1997 NC1977 MT2012 AR2002 MD1975 OK2001 RI1998 WI1980 UT1990 NE2001 WA1976WA1977 SC1999 CO1984 MO1978 VA1984 MD1974 CT1976 DE2015 DE2013 MT2013 AR2003 AK2009 CT1977 NV2013 TN1982 NH1999 WI1979 UT1989 SD2009 HI2003 AK2010 AZ1989 NM1999 NJ1986 OR1987 VT2000 OH1978 TN1981 WV2011 MN1978 LA1990 KY1997 IA1985 NC1975 KY1998 WV2015 MO1976 DE2011 GA1985 MI1978 WI1976 LA1989 AL1984 NC1974 MD1973 TN1980 FL1984 WI1978 IN1986 MO1977 KS1999 HI2002 NJ1985 MI1976 SC1998 CO1983 NV2011 TX1973 NE2000 IA1986 WV2014 WV2013 OK2000 IL1973 CT1974 ME2000 WA1974 OR1986 MN1977 MI1977 MN1976 AR2001 KS1998 LA1988 RI1996 OH1976 IN1985 WA1975 OR1985 FL1983 FL1981 WI1974 NM1997 NH1997 ID2009 LA1987 OH1977 AZ1988 HI2001 NH1996 OK1999 CT1975 KY1996 RI1995 MO1974 TN1979 MT2005 FL1982 SD2011 CO1981 NM1998 WI1977 NC1973 MO1975 CO1980 AL1983 RI1992 SC1997 AK2012 AK2014 AR2000 NV2015 IA1984 NH1998 SD2014 AK2011 SD2012 FL1980 MT2008 GA1984 MI1974 RI1994 RI1993 AZ1987 MT2015 RI1991 MI1975 NH1995 OH1974 CO1982 MN1975 SD2013 KY1995 KY1993 ND2009 VT1999 AL1981 NH1992 OH1975 KY1994 VA1981 SC1996 MS2008 SD2010 OR1984 MN1974 HI2000 IN1984 TN1978 KY1992 NM1996 WI1975 KS1997 NH1991 NH1994 AK2013 MT2007 NM1994 FL1979 AL1982 KS1996 NE1999 ID2010 NH1993 OK1998 MT2003 VA1980 VT1991 LA1985 GA1983 CO1979 RI1990 NJ1984 SC1994 MT2004 AL1980 ND2010 NM1995 MS2004 ND2014 SC1995 ME1999 IA1983 TN1976 VT1993 LA1986 MT2006 GA1982 ND2011 KS1995 MS2006 IN1983 AZ1986 WA1973 RI1989 TN1977 OK1997 FL1976 ID2014 OK1996 MO1973 ND2012 OR1983 ME1998 VT1990 VT1998 VT1992 GA1981 DE2006 AK2015 KS1991 DE2008 FL1978 ND2013 AL1979 AR1999 VA1979 IA1982 KS1994 SC1991 MS2003 IN1980 VT1994 VT1989 AZ1985 MS2007 DE2004 IN1982 DE2007 WI1973 FL1975 KS1993 CO1978 MI1973 NM1993 NJ1983 NH1990 CT1973 DE2003 IN1981 OR1982 FL1974 NE1996 SD2015 KY1991 IA1981 LA1984 WY2009 SC1993 MS2005 NM1992 MN1973 TN1974 SC1992 OR1980 OR1981 RI1988 VT1996 OK1995 NE1998 NE1997 ID2012 DE2005 NE1995 TN1975 ID2013GA1980 OH1973 MT2002 MS2002 KS1992 FL1977 ID2015 CO1976 VT1995 VT1997 NJ1981 RI1987 IA1980 ID2011 CO1974 OR1979 NM1991 ME1997 VT1988 ND2015 DE2002 CO1977 AL1978 IN1979 AR1998 GA1979 WV2007 SC1990 VA1978 NE1993 OK1994 NJ1982 HI1994 IA1979 WY2015 ME1996 WV2008 NJ1980 NE1994 KS1990 WV2006 LA1983 CO1975 NE1992 KY1990 WY2010 HI1999 IN1978 RI1986 NV2006 SC1989 MS2000 KY1989 NM1990 NV2005 UT1980 ME1991 ME1990 IN1976 OK1993 NH1989 OR1978 NV2007 GA1978 OK1992 NJ1979 MT2001 VT1987 VA1977 WY2014 SC1987 AR1996 OR1976 NV2004 UT1979 NE1991 HI1998 AR1995 NM1989 ND2005 AZ1984 WV2005 AR1997 ME1995 NH1988VA1976 WY2011 KS1989 IN1977 NV2008 NV2003 KS1988 KS1987 SC1988 HI1997 LA1982 LA1981 WY2013 TN1973 KY1988 SD2005 HI1995 RI1985 WY2012 FL1973 HI1996 OK1991 VT1986 MS1999 AR1994 WV2004 OR1977 LA1980 SC1986 SD2006 AL1977 IA1978 GA1977 NE1990 MT2000 OR1974 NM1988 OR1975 IN1974 ME1994 AZ1981 KY1987 SD2008 MS2001 NH1987 NJ1976 DE2001 NV2002 NV2001 HI1993 SD2004 ND2007 NE1989 VT1985 IA1974 NH1986 KS1986 AZ1983 SC1985 CO1973 ND2004 AZ1982 NM1987 VA1975 UT1978 IN1975 RI1984 OK1990 NJ1978 UT1976 AZ1980 OK1989 IA1977 NM1986 LA1979 KS1985 LA1974 IA1976 HI1992 OK1988VA1974 OK1987 ND2008 ME1993 AK2005 NM1985 KS1980 AR1993 LA1976 SD2007 IA1975 KS1979 NJ1975 VT1984 ND2003 KY1986 MS1997 LA1978 NE1987 NH1985 AR1992 NV2000 MS1998 WV2003 ND2006 RI1983 NJ1974 SD2003 HI1990 OR1973 WV2002 DE2000 NE1988 HI1991 MS1995 NM1984 OK1985 MS1994 KS1984 NJ1977 WV2001 OK1986 ND2002 KS1981 ME1992 LA1977 SC1984 IN1973LA1973 SD2002 KS1983 KS1982 ME1989 KS1976 AZ1979 UT1974 LA1975 KY1985 MS1996 KS1978 MS1993 NM1981 NM1979 RI1982 MT1999 SC1983 KS1977 VA1973 NH1984 RI1981 IA1973 AR1991 HI1989 NM1983 UT1975 ID2002UT1977 AK2008 OK1984 NM1980 ID2006 RI1980 ID2005 ME1988 ME1987 AZ1978 SC1981 MS1990 SC1982 KS1975 AK2007 WV2000 ND2001 AK2004 ID2003 SC1980 VT1983 NE1985 MS1989 NV1999 NM1978 ID2004 AZ1974 NJ1973 MS1991 AK2003 HI1985 NE1986 OK1983 VT1982 ID2008 KS1974 NV1998 NV1994 ME1986 AK2002 WV1998 AR1990 VT1981 HI1987 KY1979 WV1999 HI1988 ID2007 UT1973 AZ1977 SC1979 DE1999 OK1980 KY1982 ME1985 MS1992 HI1986 KY1981 AZ1976 AZ1975 MS1987 NH1983 NE1984 VT1980 VT1979 AR1989 NV1997 RI1979 NH1982 KY1980 NM1982 MS1988 NH1981 KY1984 KS1973 WV1993 HI1976 MS1986 OK1981 AK2001 SD2001 OK1979 WY2004 NH1980 NM1977 NV1996 NM1976 SD2000 ME1984 AK2006 OK1982 KY1983 MS1985 MS1981 MT1998 NM1975 MT1997 VT1978 DE1998 WY2008 WY2001 WY2003 DE1996 WV1997 WY2002 MT1996 WY2007 OK1977 HI1984 MS1982 VT1976 NM1974 HI1981 DE1997 NV1993 HI1983 MS1980 WY2006 NE1981 NV1992 ME1983 MS1984 NE1983 RI1978 WV1992 KY1976 VT1975 VT1977 NH1979 SC1978 WV1996 MS1983 WV1995 WV1994 NV1995 AR1988 KY1977 RI1976 OK1978 KY1974 NE1980 ME1982 HI1980 WY2005 NE1979 KY1978 DE1990 WV1991 WV1989 HI1979 HI1982 WV1990 RI1977 DE1994 SC1977 WV1987 ME1981 KY1975 VT1974 ID2001 MT1995 AR1987 ND1999 HI1978 NE1982 AZ1973 NH1976 DE1995 NH1978 SC1976 DE1991 MS1979 HI1974 RI1975 ME1980 SC1975 WV1985 MT1994 WV1988 NH1977 HI1977 RI1974 NV1991 DE1993 WY2000 ND2000 DE1992 ME1976 ME1979 ME1977 WV1986 NH1975 RI1973 MS1978 SC1974 AR1986 NH1974 AR1985 NV1990 NV1989 MT1992 ID2000 DE1989 HI1975 MT1993 AK1974 ME1978 MT1990 NM1973 MS1977 VT1973 SD1999 NE1978 MT1991 ME1975 MT1989 KY1973 NV1988 NE1977 ND1995 AK2000NV1987 WV1984 NE1976 ND1997 DE1988 AK1975 DE1987 NE1974 WV1983 NE1975 MT1987 NH1973 HI1973ME1974 AR1984 MT1988 WV1979 AR1983 NV1986 AR1982 AK1996 AR1974 ND1996 AK1997 AK1998 SD1997 SC1973 DE1984 ID1999 WV1982 WV1981 ME1973 ND1998 WV1980 SD1998 DE1986 WV1978 NE1973 AK1993 DE1985 ND1989 DE1983 AK1994 NV1985 AR1980 AR1981 AR1977 DE1980 AR1979 WV1974 MT1986 ND1991DE1982 WY1998 DE1979 MT1983 SD1994 AK1992WV1977 AR1975 ND1993 AK1988 AR1978 ND1992 ID1990 MT1985 MT1974 DE1981 SD1992 WY1999 MT1984 ND1988 SD1996 SD1995 AK1995 AR1976 ND1994 AK1999 SD1993 WV1976 AK1973 WY1992 MT1975 ND1974 ID1998 ND1981 WV1975 WY1991 DE1978 SD1975 SD1974 ID1989 ID1996 AK1991 NV1984 ND1990 SD1991 AR1973 WY1993 ND1987 WV1973 DE1976MT1973 NV1980 MT1982 MT1981 DE1977 ID1997 WY1997 WY1984 WY1990 ID1994 ID1991 WY1989 ND1980 NV1979 SD1978 ID1993 ID1995 NV1975 WY1987 WY1985 ID1992 DE1974AK1976 SD1988 WY1983 MT1980 WY1988 SD1989 WY1996 ND1985 ND1982 ID1974 NV1977 SD1977 WY1986 MT1976 MT1979 ND1984 SD1987 SD1973 DE1975 ND1979 SD1990 NV1981 SD1976 NV1983 WY1995 ND1986 ND1975 AK1990 ND1983 SD1981 SD1980 AK1978 AK1977 AK1989 SD1979 NV1976WY1982 ND1976 WY1974 WY1994 ID1976ID1977 SD1983 NV1978 ND1973 SD1982 NV1982 WY1975 NV1974 ND1978 SD1986 WY1981 DE1973 AK1980 ID1980 MT1978 WY1979 MT1977 AK1987 ID1981 ID1984 AK1979 AK1981 WY1978 WY1973 ND1977 WY1980 NV1973 ID1982 ID1975 WY1976 WY1977 ID1979 SD1984 ID1983 SD1985 ID1978 AK1984 AK1983 AK1986 AK1982 AK1985 ID1973 MA2004 MA2003 MA2002 MA2001
Scatter plot log of average total cost and presidential electoral importance State Level Aggregates
10
22
Scatter plot log of total cost and presidential electoral importance Pres Imp Over 80
100
85
90
(3)
ID1973
0
MD2010 MD2009 PA2015 MD2013 PA2014 PA2012 PA2011 PA2013 MD2008 MD2014 WA2010 WA2009 MD2004 MD2005 WA2012 WA2011 WA2014 MD2006 MD2007 WA2013 MD2003 NC2006 NC2005 MD2000 NC2004 MD2002 WA2008 NC2003 WA2006 WA2007 WA2005 MD2001 GA2010 MD1999 GA2009 GA2015 NC2002 MD1998 MO2009 MO2010 MD1997 VA2009 MI2008 NC2001 WI2009 NC2000 MI2007 TN2015 CT2015 MI2005 MD1996 MI2006 WI2010 MO2012 MO2011 MI2004 TN2014 VA2010 MI2003 TN2013 MO2008 MO2005 NC1999 MD1995 MO2003 MD1992 MO2004 MO2007 MO2006 MI2002 WI2011 MD1991 TN2008 MO2002 TN2007 VA2013 WI2012 MI2001 VA2012 VA2011 GA2008 MD1990 WI2008 MO2001 NC1996 MI2000 VA2014 NC1995 WI2007 WI2006 WI2004 GA2007 GA2004 GA2006 GA2005 WI2005 WI2003 MD1989 MI1999 IL1998 GA2003 NJ2013 IL1997 MD1988 WI2002 NJ2014 NC1992 IA2009 IA2010 NC1991 WI2001 MD1987 NJ2004 NJ2005 IN2014 IN2012 WI2000 IN2011 NJ2003 NJ2006 IA2012 IA2011 LA2012 KY2015 AZ2011 LA2010 LA2009 AZ2012 IN2013 WA1990 WI1999 LA2011 IA2013 RI2009 IA2014 NJ2002 IA2008 WA1989 RI2010 KY2012 IA2007 CO1998 IA2004 KY2014 KY2011 NJ2000 IA2003 NJ2001 IA2006 IA2005 KY2013 CO1997 AZ2005 IA2002 VA1998 KS2009 AZ2006 CO1996 MN1991 IA2001 VA1993 WI1994 AZ2004 KS2010 LA2003 CO1995 LA2004 KS2015 MN1992 NJ1999 VA1997 MD1984 AZ2003 OK2015 WI1993 MN1990 VA1994 RI2007 RI2008 WI1992 MN1989 KS2011 IA2000 KS2014 LA2007 AZ2002 KS2012 LA2008 NJ1998 WI1991 IA1999 MN1988 AL1994 AR2009 MN1987 AR2010 MD1983 NE2014 NM2008 WI1990 NJ1997 SC2004 TX1980 SC2003 AZ2001 KS2013 LA2002 WI1989 NE2013 AL1993 NJ1996 CT1984 RI2002 IA1998 TX1979 GA1992 NJ1995 NM2007 OR1996 FL1990 OR1995 SC2002 FL1989 MD1980 NJ1994 CT1983 IN1996 KS2008 GA1991 ME2014 NJ1993 TN1988 AR2011 IA1997 AR2012 ME2015 ME2013 IN1995 IA1996 RI2001 TX1978 TN1987 IN1994 MD1979 GA1990 KS2007 AZ1998 CT1982 IN1993 LA2001 MS2013 MT2009 NM2002 OR1992 IA1995 CT1981 GA1989 IN1992 MS2015 SC2001 KS2006 NH2002 KS2003 AZ1997 MS2012 TX1977 CT1980 KS2005 MT2010 OR1991 TN1986 IN1991 KS2004 WV2010 MS2014 NM2001 WA1980 MI1982 TN1985 WV2009 MI1981 CT1979 MS2011 OR1989 MD1976 OR1990 NJ1988 AZ1992 MI1980 MO1980 NH2001 NM2000 AZ1991 MT2011 WA1979 SC2000 NJ1987 LA1992 AR2015 LA1994 MO1979 AZ1990 CT1978 TN1984 LA1996 LA1991 MI1979 NH2000 TX1974 OR1988 WA1978 LA1993 TN1983 LA1995 MT2012 MD1975 WI1980 SC1999 MD1974 MO1978 CT1977 NH1999 WI1979 AZ1989 NJ1986 NM1999 WA1977 OR1987 IA1985 MO1976 WV2015 AL1984 WI1976 MD1973 MO1977 WI1978 MI1976 NJ1985 IA1986 TX1973 OR1986 WA1974 NM1997 OR1985 AZ1988 KY1996 MO1974 MT2005 NM1998 WI1977 AL1983 MO1975 AK2014 AZ1987 MI1975 MI1974 MT2015 KY1995 AL1981 MS2008 OR1984 NM1996 WI1975 AL1982 KY1992 AK2013 NM1994 AL1980 NM1995 MS2004 GA1982 MT2006 MS2006 AZ1986 WA1973 MO1973 OR1983 GA1981 AL1979 AZ1985 MS2007 MS2003 IN1982 CO1978 MI1973 NM1993 IN1981 OR1982 KY1991 MS2005 NM1992 OR1981 OR1980 VT1996 MS2002 VT1995 CO1976 OR1979 NM1991 AL1978 AR1998 CO1977 WV2007 OK1994 WV2008 WV2006 KY1990 CO1975 KY1989 MS2000 NM1990 UT1980 OK1993 OK1992 OR1978 AR1996 AR1995 OR1976 UT1979 NM1989 AR1997 AZ1984 WV2005 KY1988 OK1991 WV2004 LA1980 MS1999 OR1977 AL1977 AL1976 NE1990 NM1988 MS2001 OR1975 AZ1981 KY1987 NJ1976 IA1974 NE1989 AL1975 AZ1983 AZ1982 NM1987 NJ1978 AL1974 AZ1980 LA1979 NM1986 NM1985 LA1976 NJ1975 KY1986 LA1978 MS1998 MS1997 NJ1974 AR1992 WV2003 WV2002 HI1990 MS1994 OK1985 NJ1977 OK1986 WV2001 LA1977 AZ1979 KY1985 LA1975 MS1993 AL1973 AR1991 HI1989 IA1973 OK1984 ID2006 ID2005 SC1982 SC1981 MS1990 WV2000 SC1980 ID2003 ID2004 MS1989 NJ1973 HI1985 MS1991 OK1983 ID2008 NV1994 WV1998 AR1990 HI1987 KY1979 WV1999 HI1988 ID2007 KY1982 SC1979 KY1981 MS1992 OK1980 HI1986 AR1989 KY1980 OK1975 WV1993 HI1976 OK1981 OK1979 NV1996 OK1974 OK1982 WV1997 HI1984 OK1977 HI1981 HI1983 MS1980 NV1993 NV1992 WV1992 KY1976 WV1996 OK1976 AR1988 NV1995 WV1995 WV1994 KY1977 OK1973 OK1978 KY1978 WV1989 WV1991 HI1982 WV1990 WV1987 KY1975 AR1987 HI1974 MS1979 WV1985 MS1974 WV1988 RI1974 NV1991 MS1976 MS1978 WV1986 RI1973 AR1986 AR1985 NH1974 HI1975 MS1975 MS1977 HI1973 MS1973 AR1984 NH1973 WV1979 AR1983 AR1982 WV1980 WV1978 AR1980 AR1981 AR1979 WV1977 ND1991 AR1975 ND1992 WV1974 AR1976 WV1976 MT1975 ND1974 WV1975 ND1981 WV1973 ND1985 ND1982 MT1976 ND1984 ND1986 ND1975 ND1983 ND1976 ND1973 NV1974 MT1978 MT1977 NV1973
NY2015 NY2014 NY2012 NY2011 NY2013 MD2015 IL2009 IL2010 IL2015 IL2012 PA2000 WA2015 IL2011 IL2013 IL2014 PA1999 OH2015 MA1992 MA1991 WA2004 WA2003 MA1990 MA1989 FL2012 TN2009 WA2002 GA2014 TN2010 FL2011 GA2012 GA2011 MA1988 GA2013 WA2001 MA1987 IL2002 MD1994 IL2001 MA1986 MD1993 MA1985 IL2000 NJ2009 OH2000 IL1999 NJ2010 AL2009 AL2015 AL2010 OH1999 MA1984 MO2000 NC1994 NJ2012 NC1993 NJ2011 VA2005 VA2008 MA1983 AL2011 MO1999 VA2004 AL2012 AL2014 IL1996 VA2007 WA1992 WA1994 VA2006 VA2003 MA1982 IL1995 AL2003 MA1981 AL2013 TX1988 WA1993 VA2002 IL1994 TX1987 MA1980 NJ2008 NJ2007 WA1991 AL2005 AL2004 IL1993 KY2009 IL1992 AL2006 MI1992 KY2010 VA2001 IL1991 MA1979 VA2000 TX1986 MI1991 IL1990 TX1985 IL1989 VA1999 IL1988 MI1990 WI1998 IL1987 MI1989 WI1997 TX1984 WI1996 WI1995 KY2005 FL1998 MI1988 OH1988 KY2008 MI1987 AL1996 FL1997 OH1987 TX1983 KY2006 KY2004 MA1974 KY2007 FL1996 AL1995 FL1995 IL1984 TX1982 TX1981 IL1983 KY2003 OH1986 FL1992 OH1985 MI1986 FL1991 MI1985 AZ2000 IL1982 MA1973 IL1981 KY2002 WI1988 WI1987 AZ1999 KY2001 OH1984 NJ1991 NJ1992 MI1984 NJ1990 WI1986 IA1993 IA1994 KY2000 OH1983 WI1985 NJ1989 TX1976 AR2013 IA1992 IL1976 AR2014 MI1983 IA1991 AZ1994 MD1978 IA1990 OH1981 OH1982 AZ1993 WI1984 OH1980 IA1989 IA1988 MD1977 TX1975 IL1974 OH1979 IL1975 WI1983 KY1999 WI1981 WI1982 IA1987 OH1978 KY1997 LA1990 KY1998 MI1978 LA1989 IL1973 MI1977 LA1988 OH1976 WI1974 OH1977 LA1987 OH1974 KY1994 KY1993 OH1975 LA1985 LA1986 IN1980 WI1973 LA1984 OH1973 IN1979 LA1983 IN1978 IN1976 IN1977 LA1982 LA1981 OR1974 IN1974 IN1975 OK1990 OK1989 OK1987 OK1988 OR1973 MS1995 IN1973 MS1996 MS1987 KY1984 MS1988 MS1986 KY1983 MS1985 MS1981 MS1982 MS1984 MS1983
NY2010 NY2009 PA2008 PA2007 TX2012 TX2011 TX2015 TX2014 TX2013 NY1998 NY1997 NY1996 OH2010 NY1995 NY1994 OH2009 NY1992 NY1993 NY1991 NY1990 OH2012 NY1989 OH2011 IL2008 OH2013 FL2010 PA1998 OH2014 IL2007 FL2009 IL2006 IL2005 PA1997 OH2008 IL2003 IL2004 OH2005 OH2006 FL2014 OH2007 PA1996 OH2004 PA1995 FL2013 OH2003 PA1994 TX1998 CA1984 OH2002 PA1993 WA2000 TX1997 CA1983 WA1999 OH2001 NC1998 NC1997 WA1998 NY1980 CA1980 WA1997 FL2005 CA1979 FL2004 NY1979 FL2006 OH1998 WA1996 WA1995 FL2003 OH1997 CA1978 NY1976 FL2002 OH1996 AL2008 OH1995 CA1977 NY1975 FL2001 OH1994 AL2002 AL2007 NY1974 OH1993 OH1992 FL2000 AL2001 OH1991 AL2000 MA1978 MA1976 FL1999 OH1990 AL1999 MA1977 NY1973 OH1989 AL1998 AL1997 MA1975 FL1994 FL1993 IL1980 PA1974 FL1988 IL1979 FL1987 IL1978 IL1977 PA1973 FL1986 FL1985 FL1984 FL1983 FL1981 FL1982 FL1980 FL1979 FL1976 FL1978 FL1975 FL1974 FL1977 FL1973
CA2006 CA2005 CA2004 CA2003 NY2008 NY2007 NY2005 NY2004 NY2006 PA2010 PA2009 NY2003 NY2002 NY2001 PA2005 PA2006 PA2004 PA2003 PA2002 TX2003 TX2004 PA2001 CA1990 TX2002 CA1989 CA1988 TX2001 CA1987 TX2000 NY1988 FL2015 NY1987 CA1986 CA1985 TX1999 PA1992 PA1991 PA1990 CA1982 NY1982 CA1981 NY1981 PA1989 TX1992 FL2008 PA1988 FL2007 TX1991 PA1987 TX1990 NY1978 TX1989 CA1976 PA1986 NY1977 PA1985 CA1975 PA1984 CA1974 PA1983 PA1982 PA1981 CA1973 PA1980 IL1986 IL1985 PA1979 PA1978 PA1976 PA1977 PA1975
TX2009 TX2010 NY2000 CA1996 CA1995 NY1999 CA1992 TX2008 TX2007 CA1991
CA2015 CA2002 CA2001 CA1994 CA1993 TX2005 TX2006
CA2009 CA2010 CA2012 CA2011 CA2014 CA2013
A
NY1986 NY1985 NY1984 TX1996 NY1983 TX1994 TX1995 TX1993
ND1980 ND1979 ND1978 ND1977
MA2015 MI2015 MA2013 MA2014 MI2014 MI2012 MI2011 MI2013 SC2010 SC2009 OR2009 CO2015 CO2012 CO2013 TN1995 AR2002 CO2014 OR2010 TN1996 NM2012 SC2008 SC2007 CO2011 NH2011 OR2004 NM2009 NE2006 OR2003 OR2008 MO1990 ME2004 OR2011 CT1989 CT1995 OR2005 CT1991 NE2001 NM2010 OR2002 NH2012 CT1996 ME2007 OK2007 OR2006 OR2007 MA1995 NE2003 OK2014 CT1992 MA1996 NE2005 NV2012 OR2001 AZ2015 NV2010 OR2012 NE2002 NM2011 ME2005 ME2008 DC2008 HI2014 NV2009 CT1990 NE1999 KS1994 AR2001 ME2002 ME2003 DE2010 MO1989 OK2013 DE2012 ME2001 ME2006 NE2004 DC2009 MO1988 NE2000 DE2009 OR1999 OK2008 DC2007 OR2000 DC2002 MO1987 DC2010 NV2011 NV2008 LA1999 RI2000 SD2015 MO1985 MO1986 DC2005 DC2001 NH1994 DC2003 KS1996 DE2008 RI1999 RI1998 DC2006 OR1998 DE2011 NV2007 DE2002 DE2007 HI2013 LA2000 DC2004 DC2011 DE2003 RI1995 ME2000 NE1998 DE2005 RI1997 KS1992 OR1997 RI1996 DE2004 DC2012 KS1991 NE1997 RI1993 CO1990 RI1994 KS1995 CO1989 DE2006 MN1978 RI1991 CO1988 KS1982 RI1992 DC2000 DC1999 DE2001 ME1999 ME1991 NH1993 RI1989 CO1987 KS1993 CO1984 KS1990 NC1978 CO1980 DC1997 DC1996 RI1990 CO1986 CO1982 DC1993 CO1983 DC1998 ME1996 CO1985 DC1995 DC1994 NC1976 DC1992 DE2000 ME1994 KS1980 ME1993 DE1998 RI1988 CO1981 KS1983 RI1987 KS1989 KS1988 NH1978 ME1998 KS1985 NC1977 KS1984 DC1990 DC1989 MN1977 RI1986 KS1981 ME1997 KS1986 ME1990 DC1991 KS1987 NH1977 CO1979 NC1975 KS1979 RI1985 DC1987 MN1976 DC1988 RI1984 ME1995 NH1976 DE1999 NM1978 ME1992 DE1997 DE1994 DE1992 KS1978 ME1985 RI1982 ME1988 RI1983 DE1996 RI1981 ME1989 ND1996 SD2000 DC1985 DC1986 DE1979 NM1977 DC1983 DE1995 DE1990 ME1984 RI1980 RI1979 ME1983 ME1987 DC1984 MN1975 ND1995 DE1988 DE1986 MN1974 MN1973 DC1981 DE1983 DE1987 DE1993 AZ1976 DE1991 DC1977 DC1982 DE1980 MT1991 NH1975 KS1977 ME1986 DE1978 DC1980 DE1982 MT1990 DE1984 RI1978 MT1987 DC1978 MT1992 MT1988 ME1978 ME1982 DC1979 RI1977 DE1981 ME1981 DC1974 DC1973 SD1997 ME1980 ND1993 MT1989 DC1976 DE1985 ME1976 SD1994 DE1989 ID1996 ME1979 DC1975 DE1977 DE1976 SD1999 MT1986 SD1998 SD1992 AZ1975 SD1993 ME1977 ID1994 ID1993 NV1976 ID1995 NV1975 ND1994 NV1978 MT1983 NV1977 MT1985 ID1991 MT1984 ID1992 SD1996 SD1991 SD1988 MT1982 MT1981 SD1989 ME1975 ME1974 SD1995 SD1987 DE1974 MT1980 MT1979 SD1977 ME1973 DE1975 SD1990 ID1977 SD1978 SD1979 DE1973 ID1974 SD1980 ID1976 SD1976 ID1978 SD1975 ID1975 SD1974 SD1973
ID1973
4
6
8
IA2015 MO2015 MO2013 MA2010 MO2014 NH2005 CT2012 CT2010 MA2009 CT2011 CT2009 MN2015 MN2010 NC2009 MA2012 MA2011 CT2014 MN2013 MN2009 NC2010 CT2007 MN2007 NC2015 NC2008 CT2013 MN2012 MO1997 NC2007 MA2008 CT2006 NC2014 MN2008 CT2008 MA2007 MA2006 MI2009 MN2014 NC2011 NC2013 RI2015 MO1998 MN2011 NC2012 NH2008 MI2010 CT2003 CT2005 MD2012 MA2005 SC2015 NH2007 MD2011 CT2004 NE2009 CT2001 CT2002 MA2004 MA2003 NH2006 NE2010 NE2008 NH2004 IN2007 NH2009 MO1996 MA2002 NH2010 RI2012 TN19 MN2006 MN2005 MN2002 MO1995 IN2006 MA2001 IN2015 MN2003 MN2001 OR2015 CT2000 NE2015 LA2014 RI2011 IN2005 LA2005 LA2015 MN2004 IN2008 ME2009 NH2003 CO2010 CT1997 IN2009 MO1994 CT1999 HI2005 LA2006 IN2002 RI2014 OR2014 TN1994 IN2010 SC2013 IN2004 IN2003 SC2011 MA2000 MN2000 RI2013 GA2002 ME2010 MN1999 OK2010 LA2013 NE2007 OK2009 HI2008 TN1993 CT1998 IA1984 NH2015 SC2014 MA1999 MI1997 CO2009 IN2001 IN1998 NC1990 RI2006 MN1994 MN1996 NH2014 NH2013 NH1991 OR2013 MS2010 CT1985 HI2009 KS2002 RI2003 SC2005 DE2014 IN2000 NE2011 RI2004 SC2012 MO1991 NE2012 CT1994 MI1996 MN1998 NM2013 SC2006 ME2011 MI1994 MN1995 NC1988 CT1988 CO2004 MN1997 NM2015 NM2014 IA1982 AZ2008 MN1993 MN1982 CO2002 TN1992 CT1993 GA2001 AZ2007 TN1991 GA1999 ME2012 MN1984 MO1993 AZ2010 AZ2009 MI1998 RI2005 CO2005 HI2007 MA1998 NC1989 AZ2013 HI2002 IN1999 AR2007 IA1983 KS1998 ND2014 HI2004 AL1988 CO2003 KS2000 CO2008 CO2007 CO2006 KS2001 OK2011 MS2009 HI2001 MA1997 CO2001 MI1995 TN1990 AR2004 CT1987 MO1992 NV2013 AL1992 GA2000 IN1997 MA1994 NJ2015 NC1984 NC1987 IA1979 MN1985 AL1991 GA1997 MI1993 NM2003 HI2011 NC1986 OK2012 DE2015 HI2012 AL1987 AZ2014 MN1986 NV2014 MA1993 HI2003 HI2006 HI2010 TN1989 AL1989 AR2008 IN1990 IN1989 CO1994 OK2004 MN1981 OK2005 NH1981 HI2015 AL1990 AR2006 AR2003 OK2006 CT1986 NH1997 ND2013 OK2003 ID2010 GA1996 SD2009 CO1992 MN1983 NC1982 CO2000 NH1992 DE2013 HI2000 GA1998 GA1994 AL1985 AR2000 KS1997 NM2005 NV2015 IN1988 CO1999 GA1995 ID2009 IA1978 GA1993 SC1998 LA1998 ND2009 NC1985 NC1983 NM2004 OK2002 SC1997 IN1987 AK2010 NH1987 NH1998 MN1980 NH1990 SD2014 OR1993 MO1982 ND2005 NM2006 NH1986 OK2000 TN1979 CO1993 CO1991 LA1997 MO1984 NC1981 NC1980 IA1980 ID2015 AZ1996 AZ1995 KS1999 SC1995 ID2011 MO1981 ND2011 OR1994 ND2015 SD2012 MT2014 AR2005 TN1980 MO1983 NH1995 IA1981 AL1986 ND2012 MD1981 ND2010 IN1985 SC1994 AK2015 SC1996 MD1982 NH1985 OK1999 TN1982 NE1994 NH1982 NH1996 NH1988 MT2003 ID2012 ND2007 NE1992 SD2011 HI1995 MT2013 NV2006 ID2014 NV2005 MN1979 NE1996 OK2001 NH1980 SC1993 IN1984 TN1981 TN1978 MT2007 NE1993 AK2009 ID2013 NE1995 OK1998 SC1991 AK2012 SD2008 SD2013 AK2011 GA1987 OK1997 IN1986 NV2002 HI1999 AR1999 NC1979 SC1992 ND2006 NH1984 HI1994 NH1989 GA1988 ND2004 MT2008 SC1986 ID2002 HI1998 NE1991 IA1977 NH1983 NV2003 CT1976 IN1983 SD2010 HI1997 GA1984 ND2008 GA1986 MT2002 MT2004 HI1996 CT1975 AK2004 ND2001 NV2000 GA1985 NV2004 NE1988 SC1990 TN1976 HI1992 IA1976 GA1983 NV2001 HI1993 AK2005 AR1993 TN1977 MD1986 AR1994 OK1996 SC1987 ND2003 SD2006 MT2000 SC1985 CT1974 NE1987 MD1985 AK2007 OK1995 SC1989 ND2002 NE1986 NH1979 SC1988 AK2008 SD2005 MT2001 AK2002 HI1991 NM1984 AK2006 NM1979 SD2004 GA1980 GA1979 NC1974 NJ1981 SC1983 NE1985 NV1999 AK2003 NC1973 ND1999 NV1998 IA1975 AZ1978 NJ1980 NE1984 SD2007 NJ1982 NM1981 TN1974 SC1984 SD2003 AK2001 AK2000 GA1978 SD2001 ND2000 CT1973 ID2001 TN1975 NJ1979 MT1999 NJ1984 NV1997 NM1983 NM1982 SD2002 TN1973 NV1990 NV1989 NJ1983 AZ1977 GA1976 MT1994 GA1977 MT1998 NM1980 MT1995 MT1996 NE1982 KS1974 ID1988 NE1980 NE1979 NE1981 NV1987 KS1976 NE1983 MT1997 NV1988 KS1975 MT1993 HI1980 ND1997 CO1974 HI1979 AK1995 GA1974 SC1978 GA1975 ND1998 HI1978 NV1986 NV1984 NV1982 ID1999 HI1977 NM1976 ID2000 ID1997 AK1994 AK1999 KS1973 SD1982 CO1973 NE1977 SC1977 RI1976 AK1998 ND1988 LA1974 AK1997 ID1998 LA1973 NE1976 NE1978 GA1973 AK1996 AK1993 NV1980 NV1983 SC1976 AZ1974 NV1985 ND1987 AK1988 AK1992 NM1975 NV1979 SD1985 ID1990 ID1989 ND1989 RI1975 AK1991 SD1983 KY1974 SC1975 NM1974 NE1975 NV1981 SC1974 AZ1973 AK1989 KY1973 NE1974 SC1973 ND1990 AR1974 AR1973 SD1986 ID1987 AK1976 AK1990 SD1981 SD1984 AK1978 NE1973 AK1987 AK1981 NM1973 AK1975 ID1982 AK1977 AK1984 AK1983 AK1986 ID1984 ID1986 ID1985 AK1982 ID1983 AK1980 AK1985 AK1979 ID1979 ID1980 ID1981 MT1974 MT1973
IA2011 PA2015 MO2004 MO2008 MO2012 MO2005 MO2009 PA2013 IA2009 IA2012 PA2014 IA2010 MO2011 CT2015 MO2002 MO2010 MO2007 MO2006 PA2011 PA2012 IA2005 MO2001 IA2008 MO2003 NC2006 IA1996 NC2002 IA2000 NC2001 IA2004 IA2013 NC2003 GA2006 IA2014 MD2013 MD2010 NC2004 IA2002 IA2006 NC2005 GA2015 MI2007 NH1999 GA2007 GA2008 IA1998 IA2007 MI2008 GA2005 IA2003 GA2010 MD2008 KS2014 NC2000 IN2014 MD2009 KS2015 MI2006 MI2003 GA2003 KS2008 MI2001 GA2009 MI2002 NE2014 MI2005 IA1999 IA1995 RI2010 MI2000 GA2004 IN2012 IA2001 MI2004 MS2015 NH2001 NC1999 IA1985 MD2014 KS2007 MI1999 KY2015 IN2013 NC1995 LA2012 MD2004 KS2010 OK2015 LA2011 MS2014 AR2011 IA1997 CT1984 MD2007 MS2013 NH2000 LA2010 RI2008 NC1991 MD2002 MD2005 MS2012 RI2009 KS2011 LA2007 KS2013 LA2003 KS2012 ME2013 KS2009 CT1982 NC1996 NE2013 AL1993 MD2006 IN2011 LA2004 ME2014 IA1986 KS2005 LA2009 MD2003 AR2012 SC2003 AR2010 ME2015 KY2014 RI2007 CT1983 LA2008 NC1992 SC2004 CT1981 AL1994 AZ2012 MN1991 NH2002 AR2009 MD2001 KS2003 SC2002 KS2006 KS2004 LA2002 MN1990 AZ2005 MN1992 AZ2011 IN1994 KY2011 MN1989 MS2011 IN1993 MN1987 NJ2014 KY2012 NM2008 RI2002 IN1995 AZ2006 IN1996 KY2013 SC2001 AZ2004 IL1997 AZ2003 IN1991 KY1995 LA2001 IL1998 MN1988 MD2000 OR1995 AZ2001 RI2001 NJ2006 NM2007 IN1992 AZ2002 CT1980 IN1981 NJ2013 TN1983 CT1979 GA1992 NM2002 TN1988 TN1987 AZ1997 MI1982 NJ2004 TN1986 NJ2005 SC2000 SC1999 CO1995 KY1987 CO1998 LA1996 OR1996 TN1985 AL1983 AZ1992 NJ2003 MT2009 MD1999 AZ1990 ID2004 MT2010 MD1998 CO1996 AZ1998 GA1991 MT2011 CO1997 MD1997 OR1992 TN1984 NM2000 NM2001 KY1996 MD1996 OR1990 IN1982 MI1980 MO1980 MS2006 NM1994 AL1984 AR1998 MD1980 AZ1991 CT1978 NM1988 OR1989 GA1989 MS2005 OR1991 MI1981 AR2015 LA1994 MS2008 OR1988 AZ1989 NJ1996 AZ1987 AK2014 MT2012 NJ2002 NJ2001 AZ1982 AZ1988 NM1999 FL1989 GA1990 MD1995 KY1992 MT2005 ID2003 NJ2000 LA1992 MS2007 AL1981 MD1992 KY1985 MO1977 MO1979 FL1990 CT1977 MD1984 MS2004 MI1979 NM1998 AK2013 LA1995 MD1990 NM1997 OR1987 MD1991 AZ1986 AZ1985 LA1991 MD1988 MO1978 AL1982 NM1991 OR1985 AR1996 AR1995 KY1991 LA1993 AL1978 AZ1984 MD1983 MD1989 MO1976 NJ1997 NM1992 AL1979 NJ1999 AL1980 NM1990 KY1988 MD1979 MS2002 NM1995 NM1993 KY1986 NJ1998 NM1987 OR1986 NE1990 NM1989 KY1989 MD1987 MT2006 NJ1995 NM1986 NJ1994 NM1996 NV1994 OK1984 OK1991 OR1981 MS1999 MS2003 CO1978 AR1997 GA1982 ID2005 AR1992 MS2000 MT2015 OK1994 SC1982 NJ1993 OK1993 OK1992 OR1982 MO1975 HI1989 OR1983 AZ1983 ID2008 OR1980 OR1978 AZ1981 ID2006 KY1990 HI1990 OR1984 NJ1988 OR1979 NJ1987 NV1996 ID2007 CO1977 KY1982 MI1976 NE1989 OK1985 MD1976 GA1981 MS2001 MS1997 AR1991 MS1998 CO1976 NM1985 HI1987 OR1977 OK1986 AL1977 MO1974 AZ1980 MS1991 NV1995 MS1994 MI1975 IA1974 MD1975 NV1991 AZ1979 HI1988 NV1992 MS1989 HI1986 OK1983 MO1973 NJ1985 LA1980 AL1976 MD1973 MD1974 AR1990 MS1993 HI1985 KY1981 MS1992 NV1993 SC1981 MS1990 OR1976 KY1980 OK1982 NJ1986 SC1980 KY1979 LA1979 HI1976 AR1988 AR1987 IA1973 MI1974 AR1989 HI1981 LA1978 SC1979 OK1980 OK1981 NJ1978 HI1983 KY1978 AL1975 HI1982 HI1984 MI1973 OK1979 NJ1976 CO1975 AR1986 AR1985 MS1980 KY1976 LA1977 AL1974 OR1975 OK1978 NJ1977 AL1973 OK1977 LA1976 NH1974 NH1973 MS1979 KY1977 AR1983 AR1982 HI1974 AR1984 MS1978 ND1982 HI1973 NJ1975 OK1975 HI1975 AR1980 KY1975 AR1981 OK1976 NJ1974 MS1977 LA1975 ND1991 ND1992 AR1976 MS1976 ND1985 NJ1973 RI1973 OK1974 RI1974 ND1984 MS1975 ND1986 OK1973 AR1979 ND1983 MS1974 AR1975 ND1981 MS1973 NV1974 MT1977 ND1974 ND1976 NV1973 MT1978 MT1975 MT1976 ND1975 ND1973
PA2008 NY2010 PA2007 NY2009 OH2013 OH2010 OH2014 IL2007 AL2002 OH2012 OH2011 AL2008 OH2009 IL2006 IL2005 OH2008 OH2007 OH2006 IL2008 OH2005 IL2003 AL2000 FL2010 AL2001 FL2009 OH2002 AL2007 OH2004 OH2003 PA1998 AL1997 FL2014 IL2004 AL1999 AL1998 FL2006 FL2013 OH2001 PA1997 FL2005 NC1998 NY1998 PA1993 PA1994 NY1997 NC1997 PA1995 OH1998 PA1996 NY1996 NY1995 FL2004 FL2003 NY1994 NY1992 OH1997 NY1993 FL2002 NY1991 OH1996 NY1989 OH1995 NY1990 OH1994 FL1999 OH1993 FL2001 OH1992 FL2000 OH1991 CA1983 OH1989 CA1984 OH1990 FL1994 FL1993 CA1980 FL1988 NY1980 FL1987 CA1979 MA1978 CA1978 NY1979 MA1977 FL1986 MA1976 CA1977 FL1985 FL1983 IL1980 FL1982 FL1984 NY1976 IL1979 IL1978 NY1975 MA1975 FL1980 IL1977 FL1981 FL1979 NY1974 FL1976 FL1978 PA1974 FL1975 NY1973 FL1977 PA1973 FL1974 FL1973
PA2010 PA2009 PA2005 NY2007 PA2006 PA2004 PA2003 NY2004 NY2006 NY2008 PA2002 NY2005 NY2003 PA2001 NY2002 FL2015 NY2001 FL2008 FL2007 PA1992 PA1991 CA1990 NY1988 NY1987 CA1989 PA1990 PA1987 CA1988 PA1989 PA1986 CA1987 PA1988 PA1985 NY1982 CA1986 CA1985 PA1984 CA1982 NY1981 CA1981 PA1982 PA1983 PA1981 IL1985 PA1980 IL1986 PA1978 NY1978 PA1979 PA1976 CA1976 NY1977 PA1977 CA1975 PA1975 CA1974 CA1973
CA2006 CA2005 NY2000 CA2004 CA2003 NY1999 CA1996 CA1992 CA1995 CA1991 NY1986 NY1984 NY1985 NY1983
CA2015 CA2002 CA2001 CA1994 CA1993
CA2009 CA2010 CA2011 CA2012 CA2013 CA2014 CA2000 CA1999 CA1998 CA1997
CA2008 CA2007
AR1977 ND1980 AR1978 ND1977 ND1978 ND1979
10
(6) m A
m A
A
P
P
mp
mp
A
IA1994 OH2015 AL2009 AL2010 AL2015 AL2014 MO2000 AL2012 NY2015 AL2011 AL2013 AL2003 IL2015 MO1999 NY2014 NY2011 NY2013 NY2012 AL2005 AL2004 IL2013 IA1993 IL2012 IL2009 IA1988 IL2010 MD2015 IA1990 KY2006 IA1992 IL2011 KY2003 GA2014 GA2013 GA2012 IA1991 IL2014 AL2006 IA1989 KY2005 IA1987 GA2011 FL2012 KY2008 KY2009 PA1999 AL1996 PA2000 FL2011 IL2002 KY2010 KY2001 OH2000 KY2007 NC1993 NC1994 OH1999 IL2001 KY2004 AL1995 IL2000 MI1991 MI1992 NJ2009 IL1999 KY1997 IL1996 KY2000 MI1989 AR2014 KY2002 AR2013 MI1990 NJ2010 MA1992 NJ2011 MA1991 IL1995 KY1999 NJ2007 MA1990 KY1998 MA1989 NJ2012 AZ1994 IL1994 MI1988 MA1988 MA1982 MA1987 MA1985 AZ1993 AZ2000 NJ2008 FL1998 IL1992 MA1984 MI1983 IL1993 MI1987 MA1986 MA1981 MI1984 MA1983 AZ1999 KY1994 MI1985 MI1986 IL1991 MA1980 OH1988 IL1987 FL1992 FL1991 MD1994 MA1979 FL1996 FL1997 IL1990 IL1988 OH1987 FL1995 IL1989 IN1980 MD1993 KY1993 IL1984 LA1989 OH1986 LA1990 LA1987 IN1978 IN1977 MD1978 IL1982 OH1985 IL1983 OH1984 IL1981 IN1979 OH1982 NJ1991 NJ1989 MI1978 KY1983 LA1988 NJ1990 OH1981 NJ1992 OH1983 LA1985 MD1977 KY1984 OH1979 MI1977 OH1980 MS1995 IN1976 OK1988 LA1982 LA1986 OK1990 LA1984 LA1983 OK1987 OH1978 MA1974 IL1976 MS1996 OK1989 LA1981 OH1977 MA1973 IL1975 IN1975 MS1988 OH1976 IN1974 IN1973 IL1974 MS1987 MS1982 MS1981 MS1984 MS1986 IL1973 OH1975 MS1983 MS1985 OH1974 OR1974 OH1973 OR1973
AK1974 AK1973
(5) A
A
CA2008 CA2007
AR1977 AR1978
2
100
CA2000 CA1999 CA1998 CA1997
m n o
22 20 18
l_total_cost
16 14 12
MA1996 MA1995 MI2015 MI2011 MI2012 MI2014 MI2013 DC2009 DC2010 DC2011 WI2014 WI2013 OR2010 CO2015 CO2012 OR2009 DC2012 CO2011 CO2013 CO2014 OR2012 OR2011 OR2008 OR2006 OR2005 OR2007 OR2004 SC2009 OR2003 OR2002 CT1996 SC2010 CT1995 UT2011 DC2002 UT2012 CT1992 DC2008 OR2001 DC2005 DC2007 DC2003 AZ2015 CT1991 NM2009 OR2000 DC2006 DC2004 DC2001 CT1990 NM2010 CT1989 OR1999 MO1990 DC2000 TN1996 DC1999 TN1995 MO1989 NM2012 NM2011 MO1988 SC2008 DC1994 OR1998 DC1997 DC1996 DC1998 SC2007 DC1995 OK2014 MO1987 OK2013 OR1997 DC1993 NH2012 NH2011 CO1990 DC1992 MO1986 CO1989 MO1985 DC1991 CO1988 DC1990 LA2000 RI2000 OK2007 DC1989 NE2006 CO1987 LA1999 NE2005 OK2008 HI2014 NE2004 ME2004 ME2003 ME2006 NV2010 ME2002 ME2007 WV2012 CO1986 ME2005 ME2008 NE2003 HI2013 NV2009 DE2009 NE2002 CO1985 DE2010 UT1994 NC1978 ME2001 DC1987 RI1999 UT1993 NC1976 DC1988 DE2012 NV2012 NC1977 RI1997 AR2002 RI1998 CO1984 NE2001 MN1978 WV2011 NC1975 DE2011 CO1983 NV2011 ME2000 WV2014 WV2013 AR2001 MN1977 MN1976 DC1986 RI1996 DC1985 RI1995 NE2000 CO1981 CO1980 RI1992 RI1994 RI1993 CO1982 RI1991 MN1975 MN1974 NH1994 KS1996 NE1999 NH1993 CO1979 RI1990 ME1999 KS1995 RI1989 DC1984 NE1998 ME1998 DE2008 DE2006 KS1991 DC1983 KS1994 DE2007 DE2004 KS1993 DE2003 SD2015 DC1981 MN1973 WY2009 RI1988 DE2005 NE1997 DC1982 KS1992 DC1980 RI1987 ME1997 DE2002 ME1996 WY2015 KS1990 DC1977 DC1974 RI1986 WY2010 ME1991 ME1990 DC1978 NV2007 DC1979 WY2014 ME1995 KS1989 NV2008 KS1988 KS1987 RI1985 WY2013 DC1976 ME1994 DC1973 DE2001 DC1975 KS1986 RI1984 KS1985 ME1993 KS1980 KS1979 DE2000 RI1983 KS1984 KS1981 ME1992 KS1982 ME1989 KS1983 KS1978 RI1982 KS1977 RI1981 RI1980 ME1987 ME1988 NM1978 ME1986 DE1999 AZ1976 AZ1975 ME1985 RI1979 WY2004 ME1984 NM1977 SD2000 DE1998 DE1996 WY2003 WY2008 WY2001 WY2002 VT1976 WY2007 DE1997 ME1983 WY2006 RI1978 VT1975 ME1982 WY2005 DE1990 RI1977 DE1994 ME1981 VT1974 NH1976 NH1978 DE1995 DE1991 ME1980 NH1977 WY2000 DE1992 ME1976 DE1993 ME1979 ME1977 NH1975 DE1989 MT1992 ME1978 MT1990 VT1973 ME1975 SD1999 MT1991 MT1989 ND1995 DE1988 DE1987 MT1987 ND1996 ME1974 MT1988 SD1997 DE1984 ME1973 SD1998 DE1986 DE1985 DE1983 DE1980 ND1993 MT1986 DE1982 WY1998 DE1979 MT1983 SD1994 ND1994 MT1985 DE1981 SD1992 MT1984 WY1999 SD1996 SD1995 SD1993 WY1992 WY1991 DE1978 SD1975 SD1974 ID1996 SD1991 DE1976 WY1993 DE1977 MT1982 MT1981 WY1989 WY1997 WY1984 WY1990 ID1994 ID1991 ID1993 SD1978 ID1995 NV1975 ID1992 WY1985 WY1987 DE1974 WY1983 MT1980 MT1979 SD1988 SD1989 WY1988 WY1996 ID1974 NV1977 SD1977 WY1986 SD1973 DE1975 SD1987 SD1990 SD1976 WY1995 SD1980 SD1979 NV1976 WY1982 WY1994 NV1978 ID1976 ID1977 WY1981 DE1973 WY1979 WY1978 WY1980 ID1975 WY1977 ID1978
MA2010 MA2009 MA2012 MA2011 MA2008 MA2007 MA2006 MA2005 MA2004 MA2003 MD2012 MD2011 MA2002 MA2001 MA2000 NC2009 NC2010 MA1999 NC2012 NC2011 NC2013 MA1998 NC2015 NC2014 MA1997 NC2008 NC2007 MI2009 MA1994 MI2010 MA1993 MN2010 CT2010 MN2009 CT2009 MN2015 MN2012 MN2014 MN2013 MN2011 MO2015 TN2012 TN2011 CT2012 CT2011 MO2014 CO2010 CT2014 CT2013 WI2015 CT2008 CT2007 CO2009 MN2008 MO2013 CT2006 CT2005 MN2007 CT2004 VA2015 TN2005 TN2006 MN2006 CT2003 MN2005 TN2004 MN2004 CT2002 OR2015 TN2003 MN2003 CT2001 OR2014 MN2002 OR2013 CO2008 IN2009 TN2002 CO2005 IN2010 CO2007 CO2003 CT2000 MI1998 CO2004 MN2001 GA2002 CO2006 IA2015 MI1997 NJ2015 AZ2010 CT1999 IN2015 CO2002 MN2000 MI1996 MO1998 TN2001 AZ2009 MI1995 GA2001 CO2001 MI1994 NC1990 CT1998 MO1997 SC2015 UT2010 UT2009 CT1997 MI1993 MN1999 NC1989 UT2015 CO2000 AZ2013 GA2000 MD1986 TN2000 MO1996 MD1985 CT1994 NC1988 AZ2014 IN2008 MO1995 IN2006 MN1998 RI2015 IN2007 CO1999 CT1993 SC2014 UT2014 IN2005 MN1997 MN1996 MO1994 TN1999 MN1995 SC2012 UT2013 IN2004 LA2013 NC1987 SC2011 GA1999 LA2014 LA2015 MN1994 IN2003 MO1993 WA1988 MO1992 MO1991 RI2011 SC2013 TN1998 AZ2007 LA2005 RI2012 MN1993 IN2002 RI2013 GA1998 AZ2008 TN1997 LA2006 CO1994 CT1988 NC1986 RI2014 UT2008 NE2010 OK2009 WA1987 CT1987 IN2001 VA1991 VA1996 GA1997 VA1995 NC1985 NE2009 CO1993 NH2009 OK2010 TN1994 VA1992 IN2000 TN1993 VA1990 NE2015 NH2010 UT2005 UT2003 TN1992 UT2007 UT2006 WA1986 CT1986 CO1992 NM2014 RI2006 RI2005 RI2004 UT2002 VA1989 GA1996 NM2013 UT2004 TN1991 WA1985 CT1985 IN1999 NH2015 NM2015 RI2003 SC2005 GA1995 GA1994 NE2012 NH2014 OK2012 SC2006 ME2009 UT2001 CO1991 GA1993 OK2011 NH2013 TN1990 MD1982 NC1984 MN1986 NE2011 VA1987 IN1998 ME2010 UT2000 ME2012 HI2009 AL1992 MN1985 VA1988 IN1997 TN1989 AL1991 MD1981 ME2011 HI2010 NH2004 NH2005 VT2009 NH2003 OR1994 NM2006 WA1984 NC1983 AL1990 AL1989 MS2010 NH2006 UT1999 VT2010 NH2007 AL1988 NM2004 NM2003 MS2009 NC1982 NM2005 WA1983 OR1993 NH2008 MN1984 NC1981 HI2011 OK2004 NE2008 HI2005 VA1985 AL1987 HI2012 MO1984 NC1980 UT1998 VA1986 UT1997 KS2002 OK2005 OK2003 WA1982 WA1981 AZ1995 NE2007 VT2014 MN1983 OK2006 VT2013 AZ1996 NC1979 VT2011 MO1983 VT2012 HI2004 MN1982 VT2003 GA1988 UT1996 UT1995 MO1981 HI2008 VT2004 IN1990 LA1998 MN1981 MN1980 VT2007 VT2005 HI2007 VT2008 MO1982 OK2002 GA1987 HI2015 KS2001 VT2006 LA1997 VT2015 AL1985 AR2005 IN1989 VT2002 AL1986 DE2014 UT1992 AR2006 AR2007 VA1982 AR2008 VT2001 IN1988 UT1991 AR2004 NV2014 GA1986 MN1979 HI2006 VA1983 IN1987 KS2000 MT2014 OK2001 UT1990 CT1976 DE2013 VA1984 WA1976 AK2009 AR2003 DE2015 MT2013 NV2013 HI2003 TN1982 AK2010 SD2009 UT1989 HI2002 VT2000 TN1981 GA1985 NC1974 TN1980 IN1986 KS1999 SC1998 UT1988 CT1974 OK2000 KS1998 ID2009 IN1985 NH1997 WA1975 OK1999 UT1987 CT1975 HI2001 NH1996 TN1979 SD2011 NC1973 AK2012 AR2000 IA1984 NH1998 NV2015 SC1997 SD2014 AK2011 MT2008 SD2012 GA1984 NH1995 ND2009 UT1986 VT1999 SD2013 UT1985 VA1981 NH1992 SC1996 SD2010 HI2000 IN1984 KS1997 TN1978 MT2007 MT2003 OK1998 NJ1984 VA1980 VT1991 GA1983 ID2010 MT2004 ND2010 SC1994 IA1983 ND2014 SC1995 ND2011 TN1976 VT1993 NH1991 IN1983 TN1977 ID2014 ND2012 OK1997 OK1996 VT1990 VT1992 VT1998 AK2015 AR1999 IA1982 ND2013 VA1979 SC1991 ID2012 NJ1983 VT1994 VT1989 CT1973 GA1980 NH1990 IA1981 NE1996 SC1993 SC1992 TN1974 OK1995 MT2002 NE1995 UT1984 ID2013 NJ1981 TN1975 VT1997 IA1980 ID2015 CO1974 ID2011 ND2015 VT1988 GA1979 UT1983 SC1990 VA1978 HI1994 NE1993 UT1981 IA1979 NJ1982 NJ1980 NE1994 NE1992 HI1999 NV2006 NV2005 SC1989 NH1989 NJ1979 GA1978 UT1982 MT2001 VA1977 VT1987 NV2004 SC1987 HI1998 NE1991 ND2005 NH1988 NV2003 SC1988 WY2011 HI1997 VA1976 HI1995 SD2005 TN1973 GA1976 HI1996 WY2012 AR1994 VT1986 SC1986 IA1978 MT2000 SD2006 GA1977 GA1974 SD2008 NV2001 GA1975 HI1993 NH1987 NV2002 SD2004 ND2007 NH1986 VT1985 CO1973 ND2004 SC1985 UT1978 VA1975 UT1976 IA1977 LA1974 HI1992 IA1976 VA1974 AK2005 AR1993 ND2008 SD2007 IA1975 ND2003 NV2000 VT1984 NE1987 NH1985 ND2006 SD2003 HI1991 NE1988 NM1984 ND2002 SC1984 SD2002 UT1977 KS1976 UT1974 GA1973 NM1981 NM1979 MT1999 SC1983 NH1984 LA1973 VA1973 NM1983 ID2002 NM1980 UT1975 AK2008 AZ1978 KS1975 AK2004 AK2007 ND2001 NV1999 VT1983 NE1985 AK2003 AZ1974 NE1986 VT1982 KS1974 NV1998 AK2002 VT1981 UT1973 AZ1977 NH1983 VT1979 NE1984 NV1997 VT1980 NH1982 NM1982 NH1981 KS1973 AK2001 SD2001 NH1980 NM1976 AK2006 MT1998 MT1997 NM1975 VT1978 MT1996 NM1974 NE1981 NE1983 NH1979 SC1978 VT1977 HI1980 KY1974 NE1980 RI1976 NE1979 HI1979 SC1977 ID2001 MT1995 HI1978 ND1999 AZ1973 NE1982 SC1976 RI1975 MT1994 SC1975 HI1977 ND2000 SC1974 NV1990 NV1989 ID2000 AK1974 MT1993 NM1973 NE1978 AK2000 NV1987 KY1973 NE1977 NV1988 ND1997 NE1976 WV1984 WV1983 NE1974 NE1975 AK1975 AK1996 NV1986 AK1998 SC1973 ID1999 WV1982 WV1981 AK1997 AR1974 NE1973 AK1994 ND1989 NV1985 AK1993 ND1998 AK1988 AK1992 AK1999 ID1990 MT1974 ND1988 AK1995 AK1973 ID1998 ID1989 AK1991 NV1984 ND1990 AR1973 NV1980 ID1997 MT1973 ND1987 NV1979 AK1976 ID1988 NV1981 NV1983 AK1990 SD1981 AK1978 AK1977 AK1989 WY1974 ID1986 ID1985 SD1983 NV1982 WY1975 ID1987 SD1986 AK1980 ID1980 SD1982 AK1987 ID1981 ID1984 AK1981 AK1979 WY1973 ID1982 WY1976 ID1979 SD1984 ID1983 SD1985 AK1984 AK1983 AK1986 AK1982 AK1985
95
(4)
Scatter plot log of total cost and Average Appropriators Per State MA2015 MA2014 MA2013
AK1974 AK1973
ID1973
80
Av App
App p
S
F g 5 Sca e p o s o he ma n va ab es o n e es No es Sca e s be ween he ma n va ab es o n e es Da a agg ega ed a he S a e P es den a Manda e eve o he pe od 1972–2015
in states that lose PEI from one presidential election to the next lose from 2.8 to 2.3 % in funding. Second, in Table 7, I run falsification/placebo tests by adding the forward value of PEI, labeled F_PEI. A test adding F_PEI, and implying a loss of significance of the PEI’s coefficient, will cast doubt on the possibility that the results in Table 5 are based only on spurious correlation between PEI and F_PEI. Table 7 shows the same set of regressions of Table 5, but now F_PEI is added as a control variable. PEI passes convincingly the test: the size of the coefficient does not change relevantly and remains significant at 1 % level in all specifications from column (1) to (8). At the same time, the coefficient of F_PEI is trivial in size and never statistically significant. A third robustness test includes variables measuring the contestability in House and Senate elections, S_CONT and H_CONT. Descriptive statistics regarding the dataset used for this test
123
252
Public Choice (2016) 168:239–263
Table 5 Presidential importance and NIH funding baseline regressions VARIABLES
(1) OLS
(2) FE
(3) FE
(4) FE
(5) FE
(6) FE
(7) FE
(8) FE
PEI (LOG)
0.077
1.227**
1.205**
1.232***
1.058***
0.775***
0.746***
0.701***
(1.520)
(0.476)
(0.465)
(0.452)
(0.269)
(0.191)
(0.161)
(0.203)
0.012
0.011
0.008
0.000
0.001
0.000
(0.011)
(0.013)
(0.007)
(0.006)
(0.005)
(0.006)
0.000
-0.000
-0.000
0.000
0.000
(0.000)
(0.000)
(0.000)
(0.000)
(0.000)
0.415***
0.413***
0.411***
0.353***
(0.020)
(0.019)
(0.019)
(0.018)
0.274***
0.425***
0.488***
(0.045)
H/SAC N_ORGSTATE LAG_DEP STATE_FY FUNDING (LOG)
(0.045)
(0.038)
Observations
85,365
85,365
85,365
85,365
69,425
69,425
69,425
57,790
R-squared
0.000
0.852
0.852
0.852
0.876
0.877
0.878
0.880
Inst. FE
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
FY Dummies (43)
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
State FE (50)
No
No
No
No
No
No
Yes
No
District FE (435)
No
No
No
No
No
No
No
Yes
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 The dependent variable is the log transform of the total funding per fiscal year to a particular NIH performer TOTFUND (LOG). PEI is the log-transformed value of the 0–100 scale of presidential electoral importance. H/SAC is the total number of appropriators in the State (appropriators in the House plus the ones in Senate). N_ORGSTATE is the total number of institutions in the same state. LAG_DEP is the lagged value of the dependent variable TOTFUND (LOG). STATE-FY is the log-transformation of the total NIH funding to a specific state in a specific fiscal year FY. Inst. FE, represent one dummy for each performer, similarly regressions control for 43 Fiscal Years dummies and, alternatively, for 50 US states fixed effects or for the 435 electoral districts. Results are robust to replicating estimates in column (6) by dropping the 670 observations regarding about 200 institutions that are in multiple states. The estimate is 0.767*** for the PEI variable (results available under request). Institutions in the District of Columbia are omitted: this implied the exclusion of about 2165 NIH performers. When the District of Columbia is included the sign of presidential electoral importance is still relevant in size and significant at the 1 % level. However, a full comparison between these results and those in this table is not always possible, because variables on appropriators are not available. Results are available on request
are provided in Table 8, while Table 9 presents the regression results. The idea behind this test is that the electoral contestability for seats in the two chambers is correlated with presidential contestability and, at the same time, can be a driver of NIH funding for individual organizations. I collected data on about 6960 elections for the House and Senate over the 1984–2014 period. As already said, it was not possible to collect data all the way back to 1972, because details on performers’ congressional district locations before 1984 are not available. This makes impossible linking House’s electoral results to institutions’ districts.20 The inclusion of these
20 On the other hand, averaging out districts’ results at the state level would add too much noise to this variable, making it a non-trustworthy indicator of the institutional exposure to House’s contestability.
123
No
No
No
Inst. FE
FY dummies
State FE (50)
District FE (435)
No
No
Yes
Yes
0.852
85,365
No
No
Yes
Yes
0.852
85,365
No
No
Yes
Yes
0.852
85,365
No
No
Yes
Yes
0.876
69,425
0.877
No
No
Yes
Yes
No
Yes
Yes
Yes
0.878
69,425
(0.040) 69,425
(0.044)
(0.019) 0.441***
(0.019)
0.411***
(0.000)
-0.000
(0.005)
0.003
(0.009)
-0.028***
(7) FE
0.288***
(0.021)
(0.000) 0.413***
(0.000)
-0.000
(0.005)
0.003
(0.009)
-0.029***
(6) FE
0.416***
(0.000)
(0.007) -0.000
(0.012)
(0.011)
0.012*
(0.009)
-0.030***
(5) FE
-0.000
0.015
(0.010)
-0.036***
(4) FE
0.015
(0.010)
-0.036***
(3) FE
Yes
No
Yes
Yes
0.880
57,790
(0.047)
0.502***
(0.018)
0.353***
(0.000)
0.000
(0.006)
0.002
(0.008)
-0.023***
(8) FE
The dependent variable is the log transform of the total funding per fiscal year to a particular NIH performer TOTFUND (LOG). PEI is the log-transformed value of the 0–100 scale of presidential electoral importance. H/SAC is the total number of appropriators in the State (appropriators in the House, plus the ones in Senate). N_ORGSTATE is the number of performers in the state. LAG_DEP is the lagged value of the dependent variable, TOTFUND (LOG). STATE-FY is the log-transformation of the total NIH funding to a specific state in a specific fiscal year FY. PRES_LOSS in this table is a dummy equal to 1 if there has been a loss in PEI from the previous to the current presidential election. F_PEI in Table 7 is included as falsification test, and represents a state’s PEI in the next presidential mandate. Fixed effects are described in Table 5
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1
0.000
No
R-squared
85,365
Observations
STATE_FY FUNDING (LOG)
LAG_DEP
N_ORGSTATE
H/SAC
-0.035***
(0.010)
-0.066
(0.056)
PEI_LOSS
(2) FE
(1) OLS
VARIABLES
Table 6 Robustness presidential importance and NIH funding: loss of PEI
Public Choice (2016) 168:239–263 253
123
123
0.006
No
No
No
No
R-squared
Inst. FE
FY dummies
State FE (50)
District FE (435)
No
No
Yes
Yes
0.859
76,912
No
No
Yes
Yes
0.859
76,912
No
No
Yes
Yes
0.859
76,912
No
No
Yes
Yes
0.882
62,219
0.883
No
No
Yes
Yes
No
Yes
Yes
Yes
0.884
62,219
(0.034)
(0.043) 62,219
0.429***
(0.019)
0.410***
(0.000)
-0.000
(0.005)
-0.001
(0.153)
0.053
(0.167)
0.630***
(7) FE
0.271***
(0.019)
(0.020)
(0.000) 0.411***
(0.000)
-0.000
(0.006)
-0.003
(0.188)
0.083
(0.175)
0.673***
(6) FE
0.414***
(0.000)
(0.007) 0.000
(0.010)
(0.010)
0.003
(0.301)
0.322
(0.206)
0.925***
(5) FE
0.000
0.007
(0.493)
0.371
(0.341)
1.052***
(4) FE
0.010
(0.529)
0.301
(0.341)
1.005***
(3) FE
Yes
No
Yes
Yes
0.886
50,704
(0.044)
0.517***
(0.018)
0.347***
(0.000)
0.000
(0.006)
-0.003
(0.218)
0.159
(0.216)
0.568**
(8) FE
The dependent variable is the log transform of the total funding per fiscal year to a particular NIH performer TOTFUND (LOG). PEI is the log-transformed value of the 0–100 scale of presidential electoral importance. H/SAC is the total number of appropriators in the State (appropriators in the House, plus the ones in Senate). N_ORGSTATE is the number of performers in the state. LAG_DEP is the lagged value of the dependent variable, TOTFUND (LOG). STATE-FY is the log-transformation of the total NIH funding to a specific state in a specific fiscal year FY. PRES_LOSS in Table 6 is a dummy equal to 1 if there has been a loss in PEI from the previous to the current presidential election. F_PEI in this table is included as falsification test, and represents a state’s PEI in the next presidential mandate. Fixed effects are described in Table 5
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1
76,912
(0.536)
(1.718)
Observations
STATE-FY FUNDING (LOG)
LAG_DEP
N_ORGSTATE
H/SAC
0.333
-5.134***
(0.343)
(1.152)
F_PEI
1.020***
2.540**
PEI (LOG)
(2) FE
(1) OLS
VARIABLES
Table 7 Robustness PEI and NIH funding: falsification test using next period PEI
254 Public Choice (2016) 168:239–263
Public Choice (2016) 168:239–263
255
Table 8 Robustness senate and House contestability (period 1984–2014), house and senate electoral data from 1984 to 2014 Variable
N
Mean
FISCAL YEAR (NIH)
6960
1999
S_CONT
6960
69.27
S_CONT (LOG)
6960
4.210
CONGRESS
6960
106.5
W_VOTE
6960
0.680
C2_VOTE
6960
0.300
H_CONT
6960
49.30
H_CONT (LOG)
6960
3.570
SD 9.220
Min
Max
1984
2014
18.54
98.80
0.250
2.920
4.590
4.610
99
114
0.140
0.410
1
15.31
0.130 26.30 1.170
0
0.500
0
100
0
4.620
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 House and senate electoral data from 1984 to 2014 shows the summary statistics of the data collected for 6960 electoral races for the 1984–2014 period. I excluded special races. To favor the comparability of the results, contestability measures are computed in a similar fashion than the PEI measure. In particular for the Senate’s contestability measure, the percentage difference between the votes of the winning candidate (W_VOTE) and the second candidate (C2_VOTE) is subtracted to 100. In this way when the percentage difference is large (low contestability) the S_CONT = 100 - %DIFF is low. Vice versa, when the difference is low (high-contestability) the 100 - %DIFF measure will be large. Note that because in there are unopposed House races, the H_CONT measure is equal to 0 in some cases. The log-transformation in this case is log(H_CONT?1) otherwise some observations would be lost. Because House races are computed at the congressional district level, I control for District FE. For Senate I use an average of the two Senatorial results as a State level indicator. SE are robust and clustered at the state level
variables (S_CONT and H_CONT)21 in Table 9 slightly enlarges the size of the coefficient estimated for PEI. It is possible that the restriction of the reference period plays an important role in the increase of the PEI effect on funding. For example, the sub-period selected to control for Senate and House contestability exposes institutions to a specific group of presidents who were particularly effective in delivering additional funding to institutions in contestable states. This hypothesis finds support in column (7), where I estimate the same regression model used in column (6), but omitting House and Senate contestability measures. The elasticity is reduced only slightly from 0.705 to 0.693 (1 % significance level in both cases),supporting the idea that the sharp change owes mostly to the shorter time span available. Overall, the baseline findings in Table 5 are robust to three different robustness checks: the loss in PEI effect, the falsification test including F_PEI, and taking account of potential omitted variables measuring the concomitant contestability for congressional seats.22 21 H_CONT and S_CONT were constructed by collecting voting data on winners and the first of the nonelected competitors, and by subtracting the percentage difference in the two shares from 100. This created again a contestability scale varying from 100 to 0, similar to the one used for PEI. While several ways of gauging contestability are possible, I think the one adopted here has the advantage of making the three measures for House, Senate, and presidential elections easily comparable. More details on the construction of these measures are in the notes section of Table 10. 22 Results are also robust to removing outliers following several criteria: Removing residual outliers below the first percentile and above the 99th; also by removing residuals below the 5th and above the 95th percentile. In another test I remove all residuals larger than 4/N (Cook’s test). In all cases, the PEI variable returns the predicted positive sign, which is fairly stable and significant at the 1 % level. Results are available by request.
123
256
Public Choice (2016) 168:239–263
Table 9 Robustness senate and House contestability (period 1984–2014), regression results by adding house and senate contestability VARIABLES
(1) OLS
PEI (LOG)
-1.995
1.082***
1.070***
1.179***
0.957***
0.705***
0.693***
(1.597)
(0.393)
(0.384)
(0.380)
(0.277)
(0.213)
(0.219) –
H_CONT (LOG) S_CONT (LOG)
(2) FE
(3) FE
(4) FE
(5) FE
(6) FE
(7) FE
-0.106***
-0.007
-0.007
-0.007
-0.005
-0.006
(0.025)
(0.009)
(0.009)
(0.009)
(0.005)
(0.004)
–
-0.126
-0.009
-0.009
0.002
0.015
0.006
–
(0.153)
(0.053)
(0.053)
(0.054)
(0.038)
(0.032)
–
0.006
0.004
0.000
0.000
0.000
(0.013)
(0.012)
(0.009)
(0.008)
(0.008)
H/SAC N_ORGSTATE
0.001*
0.000
-0.000
-0.000
(0.000)
(0.000)
(0.000)
(0.000)
LAG_DEP
0.359***
0.357***
0.357***
(0.019)
(0.019)
(0.019)
0.324***
0.324***
STATE-FY FUNDING (LOG)
(0.068)
(0.069)
Observations
65,358
65,358
65,358
65,358
52,372
52,372
52,372
R-squared
0.007
0.863
0.863
0.863
0.883
0.883
0.883
Inst. FE
No
Yes
Yes
Yes
Yes
Yes
Yes
FY dummies
No
Yes
Yes
Yes
Yes
Yes
Yes
District FE
Yes
Yes
Yes
Yes
Yes
Yes
Yes
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 Regression results by adding house and senate contestability presents a set of six regressions that include the log transformations of H_CONT and S_CONT
These findings thus increase the support for a causal interpretation of the robust conditional correlation found between PEI and NIH funding.
3.4 Heterogeneity tests In this section, I run a battery of heterogeneity tests to see how the PEI effect changes when considering specific types of institutions or grants. Table 10 offers an abridged version of the tests by showing the estimates of interest. Table 10 shows a set of results obtained for specific types of applications (tests 2–5) and educational institutions (test 6). Each column presents the estimated elasticities obtained when running (i) a pooled cross-section regression, (ii) when adding fiscal year and institutional fixed effects to (i), and (iii) adding the usual controls used throughout the paper to (ii). The following discussion of the results refers to the specification whose results are reported in the third column. For ease of comparison with the baseline results, the first row reports also the effects obtained in column 7 of Table 5. Estimates are run for five selected subgroups of applications. The first subgroup evaluates the PEI’s effect when excluding a specific and important (almost half of the NIH extramural budget) subgroups of NIH applications, those classified as Research Project Grant Program or, shortly, the
123
Public Choice (2016) 168:239–263
257
Table 10 Summary of heterogeneity results Test
Dependent variable: NIH funding per fiscal year (log)
OLS
FE (inst. and FY)
State FE plus controls
Types of applications and Institutions (1)
Baselines
0.077
1.227**
0.775***
(2)
R01 projects excluded
0.220
1.303**
0.810***
(3)
New applications
0.885
1.374***
0.781***
(4)
Less than 4 years of support
-2.636*
0.928**
0.622**
(5)
Competitive grants and renewals
-1.948
1.516**
0.741***
(6)
Educational Institutions
0.252
0.505**
0.554**
‘‘Small’’ Institutions (based on Org. funding) Funding below median (7)
PEI
-0.023
Interaction
-0.122
1.755*** -1.822**
0.825*** -0.215
Funding in the smallest quartile (8)
PEI Interaction
-0.531
1.502***
0.591
-2.600***
0.827*** -0.855
Funding in the smallest decile (9)
PEI Interaction
-0.134 1.936
1.270**
0.775***
1.417
1.565
1.267**
0.774***
7.238***
7.270***
Funding in the smallest ventile (10)
PEI Interaction
-0.097 2.059
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 Baselines results are the ones in Table 5, columns 1, 2, and 6. Tests for subgroups of applications of the PEI effect on the log of TOTFUNDING at the performer level. Controls are H/SAC, N_ORGSTATE, LAG_DEP, and STATE-FY FUNDING. FE are Institutions and FY fixed effects. State fixed-effects added in the third column. All regressions are robust to adding state FE. In all cases SE are robust and clustered at the state level
R01 activity type grants (row 2). When excluding these types of applications, PEI’s elasticity is 0.810, slightly larger than the one found in the baseline specification. The results thus are robust when excluding R01 grant contracts, and suggest that for the remaining types of applications presidential influence might be a bit stronger. It is reasonable to expect that lines of research more rooted in the NIH funding mechanism might be less influenced by political factors, while those recently approved or supported for fewer years could be more subject to political influence. I identify these types of application in two different ways. A first test is run for ‘‘new applications’’, i.e., applications that do not have a history of previous competitive and non-competitive renewals (test in row 3). In other words, I evaluate the impact of PEI on institution-level funding only for the part of the research portfolios of those institutions that includes new contracts and grants. As expected, the elasticity coefficient is 0.781 and significant at 1 % level. Similar results are found by including only those grants that received support for four or fewer years (test 4). While data from NIH allow selecting several thresholds, the fouryear time span is chosen to harmonize the flow of funding for new research within each presidential cycle. In this case the elasticity coefficient is 0.622 and still significant at 1 %
123
258
Public Choice (2016) 168:239–263
level. I interpret these results as a confirmation of the mechanism transmitting political influence from the President. The subsequent heterogeneity test excludes all those grants subject to non-competitive renewals (test 5). The effect found previously is very similar (0.741) and has the same level of significance. The last two heterogeneity checks involve the inclusion of educational institutions only (test 6) and ‘‘small’’ institutions, defined as those obtaining total funding below the median value of funding in the sample (tests 7–10). In the first case, the effect is smaller and less significant (0.554 at 5 % significance level). In the second case, I define the group of small institutions as those falling respectively into the lower half, quartile, decile, and ventile of the distribution of funding computed at the institution level for the whole period of reference. A dummy variable identifies when the institution falls in the distribution’s category of interest. To capture the heterogeneity of the elasticity for small institutions, I then generate an interaction term between the dummy and the PEI variable. The interaction terms are not significant, with the exception of the smallest institutions in the sample, i.e., those in the bottom ventile, are considered. In this case, the interaction term of 7.27 shows that the PEI elasticity effect for these institutions is about ten times larger than the average one estimated for the whole sample. This finding also indicates that larger PEIs favor all institutions on average, but in particular the smallest ones, at least according to the definition adopted here (Table 10). Another possibility is that institutions which have been ‘‘lucky’’, and frequently exposed to large PEIs, might be able to enjoy more funding that improves the quality of research, for example, attracting better researchers. If so, the average estimate might include two effects, a short- and a long-term one. In order to identify the presence of a long-run effect, I run a series of estimates in Table 11, where I verify the effect of the interaction of PEI with two variables. The first, PEI_DECILE, is a count variable that goes from the first decile of PEI, the one with the 10 % smallest values, to the 10th decile. Thus, the first interaction PEI 9 PEI_DECILE should be positive. As columns (3) and (4) of the table reveal, the coefficient of the interaction term is negative and not significant. Furthermore, I test a model capturing the long-term effects of exposure to PEI. I calculate the variable PEI_AV_ORG. It measures the average decile of past exposure to PEI of a performer at every new presidential election. A larger level of this variable indicates that in the past the state in which the institution is located has been, on average, in a higher decile of the distribution of PEI. For example, if an institution appeared first in a state whose PEI was in the 10th decile (highest decile) of the PEI distribution during the 1980 election of President Reagan, and the same state dropped into the 5th decile during his reelection in 1984, the institution will experience a drop in its exposure to large PEI from 10 to 7.5 (=(10 ? 5)/2). This effect is not present and the coefficient of the interaction variable PEI 9 PEI_AV_ORG is trivial in magnitude, and not significant either. I conclude that the PEI effect found previously depends exclusively on the exposure to the electoral importance of a state recorded for the current presidential race.
4 Presidential electoral importance and impact on the NIH budget In this section, the elasticity estimated in the baseline results is used to implement a simple method to gauge the impact of the percentage changes in PEI on the percentage of the NIH budget allocated for different purposes. To be precise, I use the estimate of 0.775 in
123
Public Choice (2016) 168:239–263
259
Table 11 Presidential importance and NIH funding: exposure to PEI (1) PEI
(3)
(4)
1.213***
0.983**
(0.415)
(0.429) 0.016**
-0.235
-0.275
(0.007)
(0.544)
(0.519)
PEI_DECILE
(2)
PEI 9 PEI_DECILE
(6)
(7)
0.970**
1.368***
1.360***
(0.423)
(0.484)
(0.477)
0.018***
-0.067
-0.060
(0.006)
(0.513)
(0.512)
0.013
0.011
0.052
0.061
(0.117)
(0.112)
H/SAC
(5)
0.011 (0.013)
ORG_AV_PEI PEI 9 ORG_AV_PEI
(0.111) H/SAC
(0.111) 0.011 (0.013)
Observations R-squared
84,667
84,667
84,667
84,667
84,667
84,667
84,667
Inst. FE
0.857
0.857
0.857
0.857
0.856
0.857
0.857
FY dummies
Yes
Yes
Yes
Yes
Yes
Yes
Yes
State FE
Yes
Yes
Yes
Yes
Yes
Yes
Yes
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 Regressions from (1) to (4) show the interaction effect between PEI and the specific decile of PEI according to the overall distribution of PEI. Form (5) to (7) the moving average value of the PEI decile is used in order to capture the past exposure to high/low values of PEI in relationship to the overall distribution. No significant effect is captured by the interaction term. SE are robust and clustered at the state level. This test shows that there are no long-term effect of exposure to PEI
column 6 of Table 5 as the reference value and interact it with the average percentage change in PEI. The four groups of interest are presented in Tables 12 and 13. Aggregate (A) in Tables 12 and 13 shows the summary statistics of average fraction of NIH’s extramural budget that could be attributed to PEI variation by fiscal year. The column presenting the mean value for the full 43-year sample, shows that about 2.2 % of the NIH budget is allocated because of PEI’s variation. For an extramural budget of about $25 billion, this means that a bit more than $500 million is allocated through PEI at the margin. However, the maximum value reaches, for the overall effect, about 4.5 % and the increase in PEI reached a maximum of 10.03 % (aggregate A). When this effect is decomposed by looking separately at increments of PEI, the effect rises to 2.4 %, while it is 1.75 % when looking exclusively at PEI loss. A similar methodology is applied for aggregate (B), regarding institution-level budgets, the total NIH budget by president (aggregate C) and the budget by US states (aggregate D). In this last case, both the overall effect (2.73 %) and the effect of an increase in PEI (2.83 %) are the largest observed. It is worth noting that the computed overall marginal effects are arguably a lower bound, because they include states that have not been exposed to significant variations in PEI. In other words, the average values found include observations that essentially have not been exposed to significant variation in PEI. For example, the median value of the percentage
123
260
Public Choice (2016) 168:239–263
Table 12 Estimates of PEI budget incidence, marginal impacts of PEI changes on NIH budget for different aggregates Variable
Obs
(A) % of budget by fiscal years
(B) % of budget by Institution
Mean (%)
SD
Min (%)
Max (%)
Effect
43
2.40
2.68
0.35
10.03
43
1.75
0.96
0.56
3.83
PEI loss
PEI increase PEI overall
43
2.17
1.21
0.76
4.87
10,837
1.82
2.56
0.01
30.78
PEI increase
11,572
2.10
2.42
0.01
22.23
PEI loss
15,847
1.98
2.57
0.01
30.78
PEI overall
11
2.36
2.76
0.36
9.95
PEI increase
11
1.74
0.99
0.56
3.75
PEI loss
11
2.14
1.26
0.79
4.82
PEI overall
50
2.83
2.26
0.60
10.23
50
2.65
1.79
0.54
8.54
PEI loss
50
2.73
1.87
0.57
8.91
PEI overall
(C) % of budget by President
(D) % of Budget by US State
PEI increase
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 Impact of Presidential Electoral Importance on budget shares according to different unit of observations. In (A) overall NIH budget for the 43 fiscal years. The first row indicates the impact in terms of budget share due to an increase in PEI, the second row indicates the impact when only a loss in PEI is considered, the third row reports the overall effect. In (B) single institutions budgets for the whole period is considered. (C) and (D) show, respectively the breakdown by US President and by US State. The calculus of the impact uses the PEI estimate of 0.775 in column (6) of Table 5
Table 13 Estimates of PEI budget incidence, Small Institutions Variable (A) % of budget by fiscal years
(B) % of budget by Institution
(C) % of budget by President
(D) % of Budget by US State
Obs
Mean (%)
SD
Min (%)
Max (%)
Effect
36
30.40
28.28
8.54
149.30
31
29.25
10.58
8.35
51.20
PEI loss
PEI increase PEI Overall
43
28.64
11.63
8.35
54.63
223
39.50
43.75
8.24
253.39
PEI increase
446
31.45
34.04
8.05
230.91
PEI loss
656
34.06
37.60
8.05
253.39
PEI Overall
11
26.09
21.81
9.89
82.87
PEI increase
10
30.17
8.46
15.91
48.67
PEI loss
11
27.93
8.86
15.55
48.70
PEI Overall
44
44.54
41.24
9.48
213.42
PEI increase
48
48.92
41.75
9.22
230.91
PEI loss
50
48.53
43.68
9.80
222.48
PEI Overall
If not indicated otherwise US states clustered robust standard errors are in parenthesis, with p indicating the p-values and stars meaning: *** p \ 0.01, ** p \ 0.05, * p \ 0.1 For Small Institutions, the linear combination of the main estimation in row 10 of Table 10 is significant at the 1 % level (value of 8.044) and it is used to estimate the effect for the subsample
123
Public Choice (2016) 168:239–263
261
change in presidential importance. While the median value of this variable is about 2.5 % and the average is 2.77, 50 % of the observations are exposed to a value equal or below 1.52 %. In Table 13, as expected, we find that when considering small institutions, (those below 5 % of the funding distribution), the effect is about 10 times larger, and this is reflected through the average values for the main aggregates.
5 Conclusion This paper explores the role that presidential politics plays in shaping the distribution of National Institutes of Health (NIH) funding in the United States. To do this I construct a new dataset combining information from NIH applications from fiscal years 1972 to 2015, with information about presidential election results. In doing so, I have the opportunity to exploit the variability of the importance of states to presidential elections measured across 11 different presidential races. The main finding is that NIH funded institutions/performers in states where Presidential Electoral Importance (PEI) increases by 1 % receive on average 0.75 % more funding. Results are also robust to three different tests, finding that supports a causal interpretation of the PEI’s effect. The PEI effect is positive and significant for several subsamples of the dataset, namely taking into account only new funding applications, research projects supported for less than four fiscal years, and only competitive grants and renewals. Educational institutions in states with larger PEIs seem to enjoy more NIH funding as well, though the estimate of the elasticity of funding with respect to PEI is slightly smaller than the one found in the baseline regressions. The impact of PEI is strongest when the smallest 5 % of the recipient institutions are considered. In this case the elasticity is about ten times larger than the one estimated in the baseline case (Table 13). Finally, the overall impact on the NIH budget is estimated for different aggregates. The percentage of the total NIH budget allocated for each fiscal year on the basis of the cross-state variation in PEI is about 2.1 %, on average. At the institutional level, the average is 1.9 % and for each presidential administration, the average is 2.06 %. The largest effect is found when all US states, excluding outlying possessions and the District of Columbia, are considered (2.63 %). Summing up, this paper fills a gap left in the literature on the political economy of the NIH, which focused heretofore on the sole role of representation on the appropriations committees of Congress, and finds that the importance of states to presidential elections is a key, if not a predominant political factor steering NIH funding towards so-called battleground states in contests for the White House. Acknowledgments I am really thankful to the Editors of this Journal and to an anonymous referee for very useful feedback and suggestions on this paper. I am also grateful to Mark Bonica and Christis Tombazos. I would like to thank Muyang Zhang, and other participants at the Fourth Annual International Workshop on Economic Analysis of Institutions, Xiamen University, April 16th 2016. I gratefully acknowledge the cooperation of the NIH Office of Extramural Research for sharing the NIH CRISP data under FOIA#4044, and to Shiyi Qiu for excellent research assistance. Any error, omission or mistake in this paper is my only responsibility.
References Anderson, G. M., & Tollison, R. D. (1991). Congressional influence and patterns of new deal spending, 1933–1939. Journal of Law and Economics, 34, 161–175.
123
262
Public Choice (2016) 168:239–263
Arrington, L. (1969). The new deal in the west: a preliminary statistical inquiry. The Pacific Historical Review, 38, 311–316. Arrington, L. (1970). Western agriculture and the new deal. Agricultural History, 44(4), 337–353. Cameron, A. C., & Miller, D. L. (2015). A practitioner’s guide to cluster-robust inference. Journal of Human Resources, 50(2), 317–372. Chappell, H. W., Jr., Havrilesky, T. M., & McGregor, R. R. (1993). Partisan monetary policies: presidential influence through the power of appointment. The Quarterly Journal of Economics, 108(1), 185–218. Chappell, H. W., Jr., Havrilesky, T. M., & McGregor, R. R. (1995). Policymakers, institutions, and central bank decisions. Journal of Economics and Business, 47(2), 113–136. Coate, S., & Morris, S. (1995). On the form of transfers in special interests. Journal of Political Economy, 103(6), 1210–1235. Coats, R. M., Karahan, G., & Tollison, R. D. (2006). Terrorism and Pork-barrel spending. Public Choice, 128(1–2), 275–287. Couch, J. F., & Shughart, W. F. (1998). The political economy of the new deal. Cheltenham: Edward Elgar. Faith, R. L., Donald, R. L., & Tollison, R. D. (1982). Antitrust pork barrel. Journal of Law and Economics, 25(2), 329–342. Garrett, T. A., & Sobel, R. S. (2003). The political economy of fema disaster payments. Economic Inquiry, 41(3), 496–509. Gimpel, J. G., Lee, F. E., & Thorpe, R. U. (2010). The distributive politics of the federal stimulus: the geography of the American Recovery and Reinvestment Act of 2009. In Annual Meeting of the American Political Science Association, Washington, DC. Godefroy, R. (2011). The birth of the congressional clinic. SSRN 1960609. Grier, K. B. (1987). Presidential elections and federal reserve policy: an empirical test. Southern Economic Journal, 54(2), 475–486. Grier, K. B., McDonald, M., & Tollison, R. D. (1995). Electoral politics and the executive veto: a predictive theory. Economic Inquiry, 33(3), 427–440. Hegde, D. (2009). Political influence behind the veil of peer review: an analysis of public biomedical research funding in the United States. Journal of Law and Economics, 52(4), 665–690. Hegde, D., & Mowery, D. C. (2008). Politics and funding in the us public biomedical R&D system. Science, 322(5909), 1797–1798. Hunter, W. J., & Nelson, M. A. (1995). Tax enforcement: a public choice perspective. Public Choice, 82(1–2), 53–67. Kosnik, L. R. D. (2006). Sources of bureaucratic delay: a case study of ferc dam relicensing. Journal of Law Economics and Organization, 22(1), 258–288. Larcinese, V., Rizzo, L., & Testa, C. (2006). Allocating the US federal budget to the states: the impact of the president. Journal of Politics, 68(2), 447–456. Lazear, E. (1997). Incentives in Basic Research. Journal of Labor Economics, 15(1), S167–S197. Lichtenberg, F. R. (2001). The allocation of publicly funded biomedical research. Medical care output and productivity (pp. 565–590). Chicago: University of Chicago Press. Moe, T. M., & Wilson, S. A. (1994). Presidents and the politics of structure. Law and Contemporary Problems, 57(2), 1–44. Posner, R. A. (1969). The federal trade commission. University of Chicago Law Review, 37, 47–83. Reading, D. C. (1973). New deal activity and the states, 1933–1939. The Journal of Economic History, 33(4), 792–810. Reifler, J., & Lazarus, J. (2010). Partisanship and policy priorities in the distribution of economic stimulus funds. SSRN 1670161, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1670161. Ryan, M. E. (2014). Allocating infection: the political economy of the Swine Flu (H1N1) vaccine. Economic Inquiry, 52(1), 138–154. Sampat, B. N. (2012). Mission-oriented biomedical research at the NIH. Research Policy, 41(10), 1729–1741. Tullock, G. (1983). Economics of income redistribution. Boston: Kluwer. Wallis, J. J. (1987). Employment, politics, and economic recovery during the great depression. The Review of Economics and Statistics, 69(3), 516–520. Weingast, B. R. (1984). The congressional-bureaucratic system: a principal agent perspective (with applications to the SEC). Public Choice, 44(1), 147–191. Weingast, B. R., & Marshall, W. J. (1988). The industrial organization of congress; or, why legislatures, like firms, are not organized as markets. Journal of Political Economy, 96(1), 132–163. Weingast, B. R., & Moran, M. J. (1983). Bureaucratic discretion or congressional control? regulatory policymaking by the federal trade commission. The Journal of Political Economy, 91(5), 765–800.
123
Public Choice (2016) 168:239–263
263
Wright, G. (1974). The economic determinants of new deal spending. Review of Economics and Statistics, 56(2), 30–38. Young, A. T., & Sobel, R. S. (2013). Recovery and reinvestment act spending at the state level: keynesian stimulus or distributive politics?. Public Choice, 155(3–4), 449–468. Young, M., Reksulak, M., & Shughart, W. F. (2001). The political economy of the IRS. Economics & Politics, 13(2), 201–220.
123