TY - JOUR
AU - Keller, Ann C
AB - Abstract In the eight years since the passage of the Patient Protection and Affordable Care Act (ACA), state governments have remained critical sites of contention over the law. Intense partisan conflict over ACA implementation has raised questions about traditional theories of intergovernmental relations, which posit that federal–state cooperation depends largely on policy design. Yet, few studies have examined how partisanship, as well as other important factors, shape state policy innovations under the ACA. This article examines the ACA’s State Innovation Models (SIM) initiative. SIM is specifically geared towards incentivizing states to experiment with new models of payment and delivery that can improve health outcomes and/or reduce health-care costs. Drawing on a combination of quantitative and qualitative evidence, we find that states’ participation in SIM is shaped by partisanship, administrative capacity, and state policy legacies. Our findings have implications for future efforts at intergovernmental health reforms. Whatever else the Patient Protection and Affordable Care Act (ACA) may be, it is a window onto intergovernmental relations in a polarized age. Eight years after the law’s passage, states’ choices about whether, and how, to implement its provisions––from Medicaid expansion to insurance market regulations––are shot through with partisan conflict. Under the Obama administration, such conflicts took on a familiar form, with Republican governors and state legislators proudly announcing their opposition to “Obamacare,” refusing to create insurance exchanges and accept funds for Medicaid expansion. These events have raised important questions about the applicability of traditional theories of intergovernmental relations, which posit that federal-state cooperation depends in large part on policy design. (Gormley 2006; Bulman-Pozen and Metzger 2016; Weissert and Uttermark 2017; Kincaid 2018). Yet, while state-level partisanship has constituted a barrier to ACA implementation, it remains an inadequate explanation of how states have chosen to participate in the ACA’s intergovernmental policy innovations. As Flagg (2016) notes, the ACA’s internal complexity allowed Republican governors to hold the party’s rhetorical line against “Obamacare” while quietly negotiating Medicaid expansion through Section 1115 waivers and implementing market reforms. A closer look at ACA implementation reveals the importance of non-partisan factors—including interest-group pressure, administrative capacity, and individual state policy legacies—at shaping state policy choices (Jones 2017; Hertel-Fernandez, Skocpol, and Lynch 2016; Haeder and Weimer 2015; Callaghan and Jacobs 2014). As the configuration of party power that dominated the early years of ACA implementation begins to change, it is especially important to understand how factors other than party control of government shape state-level decision-making. Making sense of state policy choices under the ACA also requires attention to contextual factors (Béland, Rocco, and Waddan 2016). While partisan conflict over the ACA was intense overall, not all state policy choices under the ACA were equally politically salient. Furthermore, not all state policy choices required extensive action on the part of elected officials. And, as in the case of the ACA’s insurance-market reforms, some decisions were quite consistent with states’ pre-existing policy legacies. By focusing on state policy choices that are politically salient and require extensive action on the part of elected officials, existing scholarship has ignored programs within the law that focus specifically on encouraging state innovation. One such component of the Affordable Care Act was the creation of the Center for Medicare & Medicaid Innovation (CMMI). CMMI is housed within the Centers for Medicare and Medicaid Services (CMS) and is charged with testing payment and delivery models that, ideally, will improve health outcomes while controlling costs. Though CMS supported state-level demonstration projects prior to the ACA’s passage, CMMI gave these efforts both a specific home within CMS and additional funding to encourage state efforts in payment and delivery reform. CMMI is pursuing innovation efforts by partnering directly with health providers to test reform models and by funding those states willing to lead reform efforts. The latter program is called State Innovation Models (SIM) and allows states to apply for “test” or “design” grants. Test grants allow states to implement a program of reforms, while the design grants provide support for states that are still in the planning phase for future reforms. An important feature of the SIM mechanism is that, unlike the providers who partner directly with CMMI to implement CMMI-selected reforms through the model test program, states can design their own reform efforts. Through this program, many states have cooperated with the federal government to design and implement payment and delivery system reforms designed to produce better health outcomes with greater efficiency. This article aims to understand the factors that shaped state participation in the SIM initiative. We begin by reviewing the literature on intergovernmental programs and policy innovation and providing some background on the SIM initiative. We then describe our data and methods, which include a quantitative analysis of state participation in the SIM initiative and in-depth case studies of participation by three states: Arkansas, Maine, and Oregon. Our quantitative results suggest that partisanship, state administrative capacity, and policy legacies structured state participation in significant ways. Further, the case studies reveal that key stakeholders from the healthcare sector supported state participation in the program, even in states where opposition to the ACA was otherwise robust. After discussing the findings and the limitations of the study, we conclude by considering its implications for the future of intergovernmental cooperation under the ACA. The Affordable Care Act and the Politics of Intergovernmental Innovation The passage of the ACA along partisan lines, and in the face of intense opposition from conservative advocacy groups and organized business, has led political scientists to re-examine the factors that affect state adoption of intergovernmental policy innovations—specifically, federal programs or policies supported by intergovernmental grant dollars or regulatory requirements. Classic models of federalism emphasize that state uptake of intergovernmental policy innovations is conditional on incentives and policy instrumentation (Derthick 1970; Walker 2000; Posner 1998). As Gormley (2006) argues, states are more likely to engage in policy innovation when it is supported by federal grants, as long as the terms of the fiscal bargain are agreeable. Regulatory tools such as intergovernmental mandates, on the other hand, are more likely to elicit conflict and policy obstruction, as well as litigation. Yet, the implementation of the ACA casts doubt on these models. Béland, Rocco, and Waddan (2016) show that, whereas intergovernmental regulations elicited significant changes to consumer-protection laws, states left billions in federal grant dollars on the table by refusing to expand Medicaid and create state insurance exchanges. Scholars have pointed to the partisan control of state government as a key factor shaping the ACA’s anomalous pattern of intergovernmental policy innovation (Haeder and Weimer 2015; Rigby and Haselswerdt 2013; Krane and Noh 2018). Within the Republican Party, opposition to the ACA was intense, and state-level victories in the 2010 midterm elections gave Republicans an important institutional foothold for opposing the ACA. As these studies correctly note, states with a Republican governor and state legislature were significantly less likely to implement state insurance exchanges than those states controlled by Democrats. Barrilleaux and Rainey (2014) found a strikingly similar partisan pattern when explaining governors’ decisions to oppose the expansion of Medicaid. Yet, while partisanship clearly matters, other scholars have suggested that the broad partisan pattern fails to adequately capture important nuances in state policy adoption. First, several studies have highlighted the importance of interest-group mobilization. Rose (2015) shows that pressure from hospital lobbyists pushed some Republican governors to accept federal funding for Medicaid expansion—despite strong intra-party pressure not to do so. Similarly, Hertel-Fernandez, Skocpol, and Lynch (2016) find that governors were more likely to support Medicaid expansion when prominent business associations and state chambers of commerce embraced this policy. In contrast, in states where conservative activists were strongly mobilized, governors were less likely to endorse Medicaid expansion (Hertel-Fernandez, Skocpol, and Lynch 2016; Jones 2017). Drawing on insights from historical institutionalism, several studies have also noted that states’ previous policy trajectories constrained their choices in the implementation of intergovernmental policy innovations under the ACA. States with prior legacies of expanding Medicaid eligibility were quicker to expand Medicaid under the ACA. This was true even in states with a Republican governor (e.g., Rick Snyder in Michigan and Chris Christie in New Jersey) (Callaghan and Jacobs 2014). Even Wisconsin, which formally refused to expand Medicaid, took advantage of federal funds and a pre-existing Section 1115 demonstration waiver to expand its existing BadgerCare program to a larger population (Flagg 2016). Pre-existing state institutions also shaped how states engaged in policy innovation. For example, in Ohio, Governor John Kasich (R) expanded Medicaid through an administrative board, allowing conservative state legislators to avoid casting a recorded vote on the issue (Flagg 2016; Rose 2015). A fourth factor identified by prior studies was state administrative capacity. Jones (2017) suggests that states’ experience with ACA implementation depended on legislative professionalism. Importantly, because few states were in session directly after the ACA passed, outgoing Democratic governors had little ability to quickly entrench intergovernmental policy innovations prior to the end of their terms. Krane and Noh (2018) also show that states with greater administrative capacity implemented state insurance exchanges. Similarly, Callaghan and Jacobs (2014) find a moderate association between states’ bureaucratic capacity and the extent of state investment in Medicaid expansion. While most studies of ACA implementation focus on explaining variation in policy choices across the fifty states, Béland, Rocco, and Waddan (2016) note that this approach to understanding intergovernmental relations ignores important variation across programs. Indeed, as a complex policy made up of numerous reforms, understanding the ACA demands attending to “within case” variation. Hence, we might expect states’ participation in the ACA’s intergovernmental policy innovations to vary depending on the characteristics of the policy observed. For example, we might expect partisanship to account for variation in states’ adoption of intergovernmental programs when the political salience of a program is high (e.g., Medicaid expansion) or when decisions must be taken by elected officials in state legislatures (e.g., health-insurance exchanges) (Béland, Rocco, and Waddan 2016). In contrast, when programs exhibit low levels of political salience, or when state administrative agencies can accept federal funds without extensive interference from elected officials, we might expect other factors—such as interest-group mobilization, state institutional capacity, and policy legacies—to play a greater role in shaping state decisions. The SIM Initiative: A Low-Salience, Bureaucratic Policy Arena The ACA’s SIM initiative provides an ideal empirical setting to test how well hypotheses about intergovernmental policy innovation hold up in cases where political salience is low and where involvement from elected officials is more limited. Given that the federal government is the largest payer for health services in the country, federal policymakers have long been interested in policy innovations aimed at reducing cost or improving the quality of coverage. Prior to the creation of Medicaid, 1962 amendments to the Social Security Act gave the federal government the ability to approve state “experimental, pilot, or demonstration projects” under Section 1115 (codified at 42 U.S.C. § 1315). Within Medicaid, Section 1115 has enabled states to undertake innovations focused on a variety of patient populations, especially the long-term care population (Thompson and Burke 2007). Building on this legacy, Section 3021 of the Affordable Care Act consolidated federal authority for carrying out such demonstrations under Center for Medicare and Medicaid Innovation (CMMI). In addition to appropriating $10 billion for policy innovations in the first decade of ACA implementation, Congress gave the Secretary of Health and Human Services greater discretion over designing and expanding demonstration projects without additional congressional approval, pending actuarial certification of savings or quality improvement (124 Stat. 119, 389, 939 [codified at 42 U.S.C. § 1315a]). In 2012, CMMI invited state governments, with the support of the Governor’s office, to participate in cooperative agreements to design and test new approaches for improving the quality of healthcare through delivery or payment reforms. The animating idea behind the initiative is that states should be able to improve quality and reduce costs by leveraging their regulatory authority and policy expertise to act as “convener” for multi-payer initiatives and health system transformations oriented towards improving population health. First, states were encouraged to test out alternate forms of payment, specifically to move away from pure fee-for-service arrangements between payers and providers. Second, states were encouraged to experiment with healthcare delivery reforms that might improve health outcomes (Van Vleet and Paradise 2014). The current trend in delivery reform is to set up care teams (e.g., health homes or patient centered medical homes) to improve continuity and care coordination. Accountable Care Organizations (ACOs) essentially combine these two aims (delivery and payment reform) in that they are designed to provide team-based care for patients and shift some of the risks of the costs of care onto the provider organization or network. CMMI was particularly interested in proposals for ACOs, patient-centered medical homes (PCMHs) and bundled or episode-based payment. Moreover, many states were already implementing some variant of these types of reforms before the passage of the ACA (Blewett, Spencer, and Huckfeldt 2017). Given the aims of the SIM grants, state innovation plans tend to focus on expanding the role of primary care and integrating healthcare across specialists, especially for at-risk populations with high healthcare costs. States, always under pressure to control Medicaid costs, had at least this incentive to test new payment models. Moreover, states that were underperforming in terms of healthcare outcomes faced additional pressures to consider reforms (Feder et al. 2017; Van Vleet and Paradise 2014). States that had an operable state innovation plan could apply for “test” grants that would allow them to expand or deepen reform efforts and evaluate the outcomes of those efforts. States that did not have reform efforts underway could apply for grants to help them design reform efforts. In total, thirty-five states participated in at least one round of the SIM initiative. In the first round of the program, twenty-five states received nearly $300 million dollars with six states receiving test grants and nineteen receiving awards that would help them design reform plans.1 In Round 2, the Innovation Center provided $660 million in grants with eleven states receiving test awards and twenty-one states/territories receiving design grants. As states implemented these plans, CMS provided technical assistance and opportunities for peer-to-peer learning. As the program was new, CMS also sought feedback from grantees about how well or poorly the program was being administered so that CMS could improve guidance and oversight in future rounds (RTI International 2014). The case of the SIM initiative raises an interesting analytical question in that, historically, payment reform has been a highly contentious area of health policymaking. Because the program does not force states to apply for SIM grants or pursue specific types of payment reform, one might argue that it is a benign piece of federal policy. Yet, past federal investments in research that merely had the potential to change health or public health practices––for example, the Patient Outcomes Research Teams funded as part of the Agency for Healthcare Policy and Research and the CDC’s program to fund research on the risks of firearm ownership––have met with ample resistance (Sorenson, Gusmano, and Collins 2014; Keller, 2014). As Feder et al. (2017) suggest, even when political leaders are sincerely interested in payment and delivery reform, they may lack the necessary statutory or administrative tools to accomplish these goals or may be unwilling to use the tools they possess. Thus, there are also reasons to believe that—despite the relatively low salience and technocratic natures of the SIM initiative—it may not have been less politically contentious than other features of the ACA. This study aims to shed light on how factors typically associated with intergovernmental policy innovation affected states’ participation in SIM. Data and Methods We employed a combination of quantitative and qualitative data collection strategies to test hypotheses about state participation in the SIM initiative and to generate new insights into the factors that shape states’ decisions to apply for SIM grants. Both analyses evaluate leading explanations for state participation in intergovernmental policy innovations under the ACA. In our quantitative analysis, we model states’ decisions to apply for federal funds under the SIM initiative. Thus, rather than making distinctions between the “level” of state participation in SIM, we were simply interested in whether states chose to apply for any funding at all. To construct our dependent variable, we collected data from public websites on states that applied for model design and test awards during the 2012–2017 period (Toone 2014; CMMI 2017). Using these data, we constructed a binary variable indicating whether a state had applied for any funding at all during any stage of the SIM initiative. To tap into existing explanations for state participation in intergovernmental policy innovations, our model includes six independent variables (see summary statistics in table 1). First, to evaluate the state partisan control hypothesis, we included an additive four-point scale of Democratic control of state government at the beginning of the grant period (National Conference of State Legislatures 2012). We assign states 2 points for Democratic control of the governorship and one point for control of each house of the state legislature. States with unified Democratic government thus receive a score of 4 while states with unified Republican government are given a score of 0. The Republican Party’s opposition to the ACA in the years observed, we expect the relationship between Democratic control of state government and participation in the SIM initiative to be positive (e.g., Haeder and Weimer 2013; Krane and Noh 2018). Table 1 Summary statistics for all states in sample and case study states Chatacteristics Mean (SD) Oregon Arkansas Maine Democratic control 1.54 (1.64) 3 4 0 Administrative capacity 8.37 (0.17) 8.49 8.12 8.3 Waivers 2.08 (1.44) 2 5 2 Medicaid MCO penetration 62.19 (23.75) 83.4 84.5 70.5 Purchaser coalition strength 0.82 (.77) 0 1 2 Right wing network strength 0.34 (.22) 0.13 0.50 0.29 FMAP 59.41 (7.97) 62.91 70.71 63.27 Health system ranking 25.92 (15.04) 32 48 5 Chatacteristics Mean (SD) Oregon Arkansas Maine Democratic control 1.54 (1.64) 3 4 0 Administrative capacity 8.37 (0.17) 8.49 8.12 8.3 Waivers 2.08 (1.44) 2 5 2 Medicaid MCO penetration 62.19 (23.75) 83.4 84.5 70.5 Purchaser coalition strength 0.82 (.77) 0 1 2 Right wing network strength 0.34 (.22) 0.13 0.50 0.29 FMAP 59.41 (7.97) 62.91 70.71 63.27 Health system ranking 25.92 (15.04) 32 48 5 Table 1 Summary statistics for all states in sample and case study states Chatacteristics Mean (SD) Oregon Arkansas Maine Democratic control 1.54 (1.64) 3 4 0 Administrative capacity 8.37 (0.17) 8.49 8.12 8.3 Waivers 2.08 (1.44) 2 5 2 Medicaid MCO penetration 62.19 (23.75) 83.4 84.5 70.5 Purchaser coalition strength 0.82 (.77) 0 1 2 Right wing network strength 0.34 (.22) 0.13 0.50 0.29 FMAP 59.41 (7.97) 62.91 70.71 63.27 Health system ranking 25.92 (15.04) 32 48 5 Chatacteristics Mean (SD) Oregon Arkansas Maine Democratic control 1.54 (1.64) 3 4 0 Administrative capacity 8.37 (0.17) 8.49 8.12 8.3 Waivers 2.08 (1.44) 2 5 2 Medicaid MCO penetration 62.19 (23.75) 83.4 84.5 70.5 Purchaser coalition strength 0.82 (.77) 0 1 2 Right wing network strength 0.34 (.22) 0.13 0.50 0.29 FMAP 59.41 (7.97) 62.91 70.71 63.27 Health system ranking 25.92 (15.04) 32 48 5 Second, we included a variable to evaluate the role of administrative capacity. Using data from the Book of the States, we identified the average value of earnings of state and local employees (excluding employees of state education systems) in each state in the year the SIM was announced (Council of State Governments 2012). We then took the natural log of these dollar figures. Consistent with the findings in the empirical literature, we expect state administrative capacity to be positively associated with state participation in the SIM initiative. Third, we constructed two variables to tap into the policy-legacies hypothesis: Waivers: Drawing on Callaghan and Jacobs (2014), we included a count of the number of Section 1115 waivers each state had requested prior to and during the grant period (CMS 2017). In total, states requested 104 waivers, ranging from zero requests from South Dakota and Nebraska to five requests from states like Arkansas and Michigan. Since these waivers help to build up state-level expertise with federal intergovernmental grant programs, we expect this variable to have a positive association with participation in the SIM initiative. Medicaid Managed Care Organization (MCO) Penetration. Our qualitative analysis of applications for SIM funds suggested that state officials saw pre-existing Medicaid payment and contracting policies as important levers to “foster transformation to accountable care models” (RTI 2014, 3–29). Thus, we included a continuous variable measuring the percent of state Medicaid enrollees covered by a managed-care plan in 2010 (Gifford et al. 2011). We expect to see a positive relationship between MCO penetration and state participation in SIM. To assess the role of interest-group mobilization in shaping state decisions to apply for funds under SIM, we included two variables: Purchaser Coalition Strength: Preliminary qualitative research suggested that, at least in some states, larger purchaser coalitions—such as the Pacific Business Group on Health and the Maine Health Management Coalition—played an important role in stimulating state participation in SIM. Indeed, early evaluation reports from SIM suggested that governors and state officials saw purchasers as key sources of support for the development of delivery-system innovations such as primary care medical homes (RTI 2014, 20–22). Drawing on data from the National Alliance of Health Care Purchaser Coalitions (NAHCPC) and ProPublica’s Nonprofit Explorer Tool, we created an ordinal variable indicating the strength of the purchaser coalition in each of the fifty states at the beginning of the period of SIM implementation. The variable takes the value of 0 if the state contained no members of the NAHCPC in 2012. It takes the value of 1 if at the state contained at least one NAHCPC in the state in 2012. It takes the value of 2 at least one major purchaser coalition had annual revenue exceeding $1 million dollars in Fiscal Year 2012. We expect purchaser coalition strength to be positively associated with state participation in SIM. Right-Wing Network Strength: We also included an index of the strength of right-wing network organizations, as described in Hertel-Fernandez, Skocpol, and Lynch (2016).2 This index is a continuous variable that captures the state-level capacity of four conservative organizations: the American Legislative Exchange Council, Americans for Prosperity, the Foundation for Government Accountability, and the State Policy Network. We expect to observe a negative relationship between right-wing network strength and state participation in SIM. Our model includes two control variables to account for economic context and the quality of state health systems. We used states’ FY 2012 Federal Medicaid Assistance Percentages (FMAP), the formula that determines the federal share of Medicaid financing, as a proxy for state economic characteristics. The formula that computes state FMAPs is based on the relationship between a state’s per capita income and national per capita income; states with lower per capita incomes relative to national per capita income have higher FMAPs. State FMAPs ranged from 50 percent (e.g., Illinois, Minnesota) to 74.18 percent (Mississippi). To examine state health-system quality, we employed an index of statewide health system performance published by the Commonwealth Fund (2009) to measure statewide health system performance. Background interviews and research suggested that states were attentive to their own ranking on this index in the process of deciding whether and how to engage in the SIM initiative. Here we include a measure of a state’s rank on the index. A lower rank (e.g., 1) equals higher performance on the index. For example, Vermont’s superior health system performance across a variety of categories made it the lowest rank state on the 2009 index, whereas Mississippi’s poor performance made it the highest. Summary statistics on the entire dataset as well as the case-study states are presented in table 1. We supplemented the quantitative analysis with case studies of state participation in the SIM program. Our cases included three states (Arkansas, Maine, and Oregon) who received grants to test innovation models during the first year of the program. To receive a first round test grant, a state was required to have a State Innovation Plan in place—an indicator that a given state was willing and able to take a leading role in policy innovation and reform. Arkansas, Maine, and Oregon represent three of the six states that were selected to receive first round test grants. We selected these three states to capture variation in the level of participation in other, more visible ACA components. Oregon both expanded Medicaid under the ACA and setup its own state-based exchange.3 Arkansas relies on a federal IT platform to run its state-based health exchange and applied for a waiver to allow for a so-called “private Medicaid option,” in which the newly eligible Medicaid population gains coverage through the purchase of subsidized healthcare in the state's healthcare exchange. Maine chose not to construct its own health insurance exchange, defaulting to the federal exchange, and declined to expand Medicaid.3 By selecting states that varied with respect to healthcare exchanges and Medicaid expansion, we hoped to learn whether decisions about these more visible aspects of the ACA affected either decisions and/or publicity around SIM participation. In addition, by focusing on these three states, we hope to uncover factors that influenced state-level decisions to participation in health policy innovation, but that may have been overlooked in prior ACA studies that focused solely on the ACA’s Medicaid expansion and health insurance exchanges. Given the effort with quantitative data to explain variation in participating and not participation states, case selection for the qualitative interviews was not designed to capture non-participating states. Instead, we hoped to learn more about the states most ready to apply for test grants to learn what might set them apart from states that either entered at the design grant level, the pre-design level, or chose not to participate. To develop each case study, we reviewed published evaluation reports of state participation in SIM and conducted semi-structured interviews with four key informants in Arkansas, Maine, and Oregon, each of whom had been involved either in the development of their state’s plan, its implementation, or both. These interviews focused on factors affecting states’ decisions to participate in the SIM initiative, governance structures in each state to implement proposed reforms, details of state innovation plans, and successes and roadblocks in the implementation of the proposed reforms. Results State participation in the SIM initiative varied greatly. Overall, as the map in figure 1 shows, across both rounds of the SIM initiative, seventeen states participated only in the model-design phase (shown in gray). Eighteen states, shown in black, participated in the model testing phase (eight of these states also participated in the model design phase). Fifteen states (shown in white) did not participate in the SIM initiative.4 Figure 1 View largeDownload slide State participation in SIM initiative. Note: CT, IA, ID, MI, OH, RI, DE, TN participated in both model design and testing. Figure 1 View largeDownload slide State participation in SIM initiative. Note: CT, IA, ID, MI, OH, RI, DE, TN participated in both model design and testing. Our multivariate models estimate how Democratic control of state government, state administrative capacity, state policy legacies (Waivers, Medicaid MCO Penetration), and interest-group capacity (Purchaser Coalition Strength; Right Wing Network Strength) affected state participation in the SIM initiative, controlling for states’ economic and health-system characteristics (FMAP, Health System Ranking). We use logistic regression because our dependent variable is a binary measure of state participation in the program. Table 2 presents the results of a full model, as well as two reduced models which exclude one of our two measures of state policy legacies. The findings here are robust to multiple model specifications (see Supplementary Appendix table A1). To interpret the results, table 3 presents marginal effects for statistically significant variables in Model 1. For all continuous variables, we calculated marginal effects by shifting variables one standard deviation above the mean. For the party control variable, marginal effects were calculated by shifting the variable from its median value to its third quartile value. All other variables were held constant at their mean values, or in the case of Democratic Control and Purchaser Coalition Strength, their median values. Table 2 Logistic regression analysis of state participation in SIM initiative Characteristics Model 1 Model 2 Model 3 Democratic control 1.07 (.51)* 1.34 (.59)* 0.80 (.39)* Administrative capacity 12.12 (5.99)** 13.62 (5.85)* 11.78 (5.09)* Waivers 0.53 (.47) – 0.70 (.42)+ Medicaid MCO penetration 0.05 (.03)* 0.06 (.03)* – Purchaser coalition strength 0.46 (.89) 0.67 (.89) 0.75 (.85) Right wing network strength –0.97 (2.82) –0.57 (2.66) –0.13 (2.4) FMAP 0.13 (.10) 0.11 (.10) 0.22 (.10) Health system ranking –0.07 (.05) –0.07 (.04) –0.08 (.04) Pseudo R2 0.57 0.54 0.50 Correctly predicted (%) 86 90 88 N 50 50 50 Characteristics Model 1 Model 2 Model 3 Democratic control 1.07 (.51)* 1.34 (.59)* 0.80 (.39)* Administrative capacity 12.12 (5.99)** 13.62 (5.85)* 11.78 (5.09)* Waivers 0.53 (.47) – 0.70 (.42)+ Medicaid MCO penetration 0.05 (.03)* 0.06 (.03)* – Purchaser coalition strength 0.46 (.89) 0.67 (.89) 0.75 (.85) Right wing network strength –0.97 (2.82) –0.57 (2.66) –0.13 (2.4) FMAP 0.13 (.10) 0.11 (.10) 0.22 (.10) Health system ranking –0.07 (.05) –0.07 (.04) –0.08 (.04) Pseudo R2 0.57 0.54 0.50 Correctly predicted (%) 86 90 88 N 50 50 50 Note: Cell entries are coefficients from logistic regression with standard errors in parentheses. ** p < 0.01, *p < 0.05, + p <0.10. Table 2 Logistic regression analysis of state participation in SIM initiative Characteristics Model 1 Model 2 Model 3 Democratic control 1.07 (.51)* 1.34 (.59)* 0.80 (.39)* Administrative capacity 12.12 (5.99)** 13.62 (5.85)* 11.78 (5.09)* Waivers 0.53 (.47) – 0.70 (.42)+ Medicaid MCO penetration 0.05 (.03)* 0.06 (.03)* – Purchaser coalition strength 0.46 (.89) 0.67 (.89) 0.75 (.85) Right wing network strength –0.97 (2.82) –0.57 (2.66) –0.13 (2.4) FMAP 0.13 (.10) 0.11 (.10) 0.22 (.10) Health system ranking –0.07 (.05) –0.07 (.04) –0.08 (.04) Pseudo R2 0.57 0.54 0.50 Correctly predicted (%) 86 90 88 N 50 50 50 Characteristics Model 1 Model 2 Model 3 Democratic control 1.07 (.51)* 1.34 (.59)* 0.80 (.39)* Administrative capacity 12.12 (5.99)** 13.62 (5.85)* 11.78 (5.09)* Waivers 0.53 (.47) – 0.70 (.42)+ Medicaid MCO penetration 0.05 (.03)* 0.06 (.03)* – Purchaser coalition strength 0.46 (.89) 0.67 (.89) 0.75 (.85) Right wing network strength –0.97 (2.82) –0.57 (2.66) –0.13 (2.4) FMAP 0.13 (.10) 0.11 (.10) 0.22 (.10) Health system ranking –0.07 (.05) –0.07 (.04) –0.08 (.04) Pseudo R2 0.57 0.54 0.50 Correctly predicted (%) 86 90 88 N 50 50 50 Note: Cell entries are coefficients from logistic regression with standard errors in parentheses. ** p < 0.01, *p < 0.05, + p <0.10. Table 3 Marginal effects for statistically significant variables in Model 1 Characteristics Change in variable (from→to) Change in predicted probability of SIM participation (%) Democratic control 1→3 +12.58 Administrative capacity 8.37→8.54 +12.47 Medicaid MCO penetration 62.19→85.94 +9.83 Characteristics Change in variable (from→to) Change in predicted probability of SIM participation (%) Democratic control 1→3 +12.58 Administrative capacity 8.37→8.54 +12.47 Medicaid MCO penetration 62.19→85.94 +9.83 Notes: For all continuous variables, marginal effects were calculated by shifting variables one standard deviation above the mean. For the party control variable, marginal effects were calculated by shifting the variable from its median value to its third quartile. All other variables were held constant at their mean values, or in the case of the Democratic Control and Purchaser Coalition strength variables, their median value. Table 3 Marginal effects for statistically significant variables in Model 1 Characteristics Change in variable (from→to) Change in predicted probability of SIM participation (%) Democratic control 1→3 +12.58 Administrative capacity 8.37→8.54 +12.47 Medicaid MCO penetration 62.19→85.94 +9.83 Characteristics Change in variable (from→to) Change in predicted probability of SIM participation (%) Democratic control 1→3 +12.58 Administrative capacity 8.37→8.54 +12.47 Medicaid MCO penetration 62.19→85.94 +9.83 Notes: For all continuous variables, marginal effects were calculated by shifting variables one standard deviation above the mean. For the party control variable, marginal effects were calculated by shifting the variable from its median value to its third quartile. All other variables were held constant at their mean values, or in the case of the Democratic Control and Purchaser Coalition strength variables, their median value. Across all three models, Democratic Control of government is positively and significantly associated with state participation in SIM (p < 0.05). As table 3 shows, holding all other variables constant, a shift in Democratic control from one chamber of the state legislature to control of the governorship and at least one house of the legislature results in a 12.58 percent increase in the predicted probability of a state participating in the SIM initiative. Administrative Capacity was positively and significantly correlated with state participation in SIM (p < 0.01 in Model 1; p < 0.05 in Models 2 and 3). Holding all other variables constant, a shift the administrative capacity variable from its mean to one standard deviation above the mean is associated with a 12.47 percent increase in the predicted probability of SIM participation. As the results in Models 1 and 2 suggest, Medicaid MCO Penetration was positively and significantly associated with SIM participation (p < 0.05). The probability of state participation in SIM increases by 9.83 percent when MCO penetration is shifted from its mean to one standard deviation above the mean. While the coefficient for the Waivers variable is in the expected direction, its statistical significance (p < 0.1 in Model 3) disappears in the full model. The coefficients for Purchaser Coalition Strength and Right Wing Network Strength variables are in the expected direction but lack statistical significance. The results here suggest that while partisanship certainly affected states’ participation in the SIM initiative, its influence was muted when compared to other reforms with in the ACA, such as state adoptions of Medicaid expansion or insurance exchanges. Indeed, in states with high levels of administrative capacity and rates of MCO penetration, the effect of partisan control of state government is muted. In other words, while we may be seeing partisan “spillover” effects related to state Republican leaders’ opposition to “Obamacare,” partisanship is only part of the story. Indeed, state administrative capacity and policy legacies also appeared to play important roles here. To investigate how these factors, and others not captured by our models, affected state SIM participation, we turn next to evidence from case studies in three states (Arkansas, Oregon, and Maine) whose participation in SIM is correctly predicted by the quantitative model. Arkansas: Payers and Partnerships At the time of the passage of the ACA, Governor Mike Beebe, a Democrat, had been serving in Arkansas for three years. At the same time, the state voted for the Republican presidential candidates over Barack Obama by just shy of a 20 percent margin in 2008 and a 23 percent margin in 2012. Arkansas further defined itself as a purple state by electing not to expand its Medicaid program under the ACA, but to apply for a waiver from the federal government to use federal dollars that would have gone to a Medicaid expansion to pay for subsidies that would allow those who would have been eligible under a Medicaid expansion to purchase health insurance in Arkansas’s state health insurance exchange. While this plan for engaging with the ACA generated controversy in Arkansas, the plan remains in place even with the state’s subsequent governor, Asa Hutchinson. The fact that Arkansas had a governor who was a Democrat at the time of the ACA’s passage and subsequent Supreme Court decision upholding most of the Act clearly shaped the state’s decisions. At the same time, a coalition of actors in the state had already come together to address both issues of cost and quality in the state. First, in 2011, Governor Beebe approached the director of the state’s Medicaid program to try to come up with a plan to address the program’s unsustainable growth rate (Arkansas 2017). The state turned to the Arkansas Center for Health Improvement (ACHI) based at the University of Arkansas Medical School to act as a convener for stakeholders to discuss plans to improve healthcare outcomes in the state while stabilizing costs. Stakeholders included the state’s three larger payers—Medicaid, BlueCross/Blue Shield, and QualChoice as well as several provider networks. While initial conversations focused on prospective bundled payments, stakeholders ultimately decided to devise a payment reform around retrospective payment reconciliation. In 2012, stakeholders in Arkansas had already agreed to use an “episodes of care” mechanism as its approach to retrospective payment reconciliation. This involved stakeholders agreeing on a set of performance standards for a given diagnosis, evaluating the range of costs associated with care for that diagnosis for a given time period (usually a year), and penalizing or rewarding providers depending on how costly their services were and the outcomes they produced.5 That same year, CMS selected Arkansas to be part of the Comprehensive Primary Care (CPC) Initiative, which enrolled providers in payment reform efforts geared towards producing better health outcomes with more efficient uses of resources. The Centers for Medicare and Medicaid Innovation issued a request-for-proposals in August 2012 and Arkansas applied using its episodes of care payment reform and adding two additional components: (1) a patient centered medical home component geared towards preventing and treating chronic conditions and (2) a health home component designed for patients with special needs including the frail elderly and patients with disabilities. CMS awarded Arkansas a State Innovation Model award in 2013, allowing the state to expand its efforts with respect to episodes of care and launch a patient centered medical home reform in the state. Up until the receipt of the CPC and the SIM awards in 2012 and 2013, the Arkansas Center for Health Improvement had not received any federal support and relied, instead, on foundations to support its work. Once the state received the CPC and SIM awards, a key informant recounted, “That kicked everything into high gear. We would not have gone as far without that money” (Arkansas 2017). The decision to apply for a SIM grant was made when the Arkansas Department of Human Services and State Medicaid directors along with the state’s Surgeon General approached the governor to suggest applying for a SIM grant. Because Arkansas was already working hard to implement its own payment reform efforts, the decision to apply for a SIM award was “pretty much a no-brainer” (Arkansas 2017). Arkansas Surgeon General, Joseph Thompson, was heading the ACHI, which was already working with the three primary payers in the state—Medicaid, Blue Cross/Blue Shield, and QualChoice on episodes of care. That the Surgeon General was arguing for applying indicated that these major players supported the idea of federal support. Further, while healthcare provider organizations wanted to be involved in major policy deliberations, they did not ultimately oppose the decision to apply for a SIM grant (Arkansas 2017). In conversations led by ACHI with stakeholders about how to address healthcare costs, stakeholders discussed several options: (1) cut reimbursement rates; (2) limit services; (3) move to a managed care approach; or (4) pay providers using fee-for-services with the new episodes of care approach, the stakeholders agreed that the fourth option was the most palatable. When state officials found out Arkansas had been one of six states or regions to receive a SIM grant, they publicized the award widely. The mechanism itself received far less media attention that the state’s Medicaid expansion waiver, so it appears that the award was publicized primarily with providers, payers, and state legislators. Actors involved in the payment reform effort said that people in the state associated the state’s insurance exchange and Medicaid expansion with the ACA, but not the episodes of care reform nor the patient centered medical home reform. At the same time, those implementing the reform effort report that they used the award to raise their profile among stakeholders in order to try to get both more payers and more providers to participate (Arkansas 2017). The first two efforts outlined in the SIM grant for Arkansas can boast a number of successes in that the PCMH and episodes of care reforms have both produced costs savings and have attracted broad participation from payers and providers. Several factors seem to have affected Arkansas’s experience. First, there are a small number of major payers and, according to a key informant, the people who lead these organizations are familiar with one another and have common interests in improving healthcare outcomes in the state and controlling costs. Once the big three players were at the table, providers had an incentive to join the effort as well. ACHI also led efforts to pass legislation that would support payment reform efforts by requiring every clinic in the state to pay a fee to support the Patient-Centered Medical Home effort. One difficulty state implementers encountered in trying to add more payers was an assumption on the part of smaller-scale payers that, as long as the big three were demanding improvements in quality, providers would deliver those quality improvements to all of the patients in their panels, regardless of payer. This created a free-rider problem that the implementers countered with moral arguments about participation being “the right thing to do” (Arkansas 2017). The Health Homes initiative, however, was blocked by nursing homes associations. Those involved with the program argue that without the private payers (Blue Cross/Blue Shield and QualChoice) pushing for the reforms legislators were more subject to lobbying on the part of providers. The state has now pivoted to an accountable care organization approach for patients with special needs who were originally the target for health homes. The change in governorship does not appear to have changed the state’s or its stakeholders’ commitment to the SIM project. Factors that might explain the state’s engagement in the SIM project include the small number of payers who were already in conversations about how to control costs, a governance structure that brought together both public and private sector payers, and a limited set of options for covering rising healthcare costs in the state. Key informants also suggest that the poor health outcomes data in Arkansas were a motivating factor (Arkansas 2017). However, it was not clear whether Arkansas’s lower marks on health outcomes relative to the national averages created a general sense of urgency to improve care quality or whether poor health outcomes provided explicit or implicit leverage in bringing providers to the table. Certainly, with the health outcomes data widely known, it would be hard for providers to argue that the payers were getting good value for their money. Discussions with a key informant in Arkansas gave the impression that engagement with SIM was conducted at the level of interest group politics with the mobilized interests being primarily payers and providers (Arkansas 2017). Mobilized groups met through the ACHI and tended to involve legislators only at the outset when they were unsure of their inclusion in decision making. Implementers were proud of receiving the SIM grant and were vocal about receiving the grant and framing the award as evidence of Arkansas’s role as a policy leader in payment and delivery reform. Maine: Government-Led Reform in an anti-Obamacare State The politics surrounding ACA implementation in Maine look very much like those seen on the national level. Maine has experienced strong and persistent divisions between Republicans and Democrats on issues of Medicaid expansion and the construction of a state-based exchange. Maine’s Republican governor, Paul LePage, supported by local Tea Party activists, opposed both the Medicaid expansion and the establishment of a state-based insurance exchange. Democrats in the Maine legislature, on the other hand, have passed legislation to expand Medicaid on five different occasions. In vetoing a Medicaid expansion bill in 2014, LePage echoed the Republican rhetoric that has emanated from state houses in non-expansion states across the U.S. Governor LePage described the expansion’s promised savings as mere “mirages,” adding that the budgetary consequences would be “disastrous” and the expansion “ruinous” for Maine (LePage 2014; Moretto 2014a). Unlike Republican governors in Arizona, Ohio, and Nevada, who expanded Medicaid as a means of securing budget relief and boosting the state economy, LePage was not swayed by the ACA’s promise of federal Medicaid dollars. Even after Maine voters approved a ballot measure in November 2017 to expand Medicaid, LePage vowed to reject any proposal that relied on higher taxes or surplus revenues to pay Maine’s share of the expansion.6 With few exceptions, the Republican members of Maine’s legislature have supported Governor LePage’s anti-ACA policy decisions, making it impossible for Democrats to override the governor’s multiple vetoes.7 This opposition has remained despite policy concessions to address common Republican critiques and concerns regarding the ACA’s Medicaid expansion. The Medicaid expansion proposed by state legislators in 2015, for example, included a requirement that the legislature reauthorize the expansion when federal funding for the expansion population dropped below 100 percent of costs. The 2015 iteration also included an automatic opt-out if federal funding ever dropped below promised levels (Lawlor 2015; Moretto 2014b; Cousins and Shepherd 2016). This provision was a direct response to opposition that was premised on the fear that the federal government would not live up to its promise, resulting in a massive financial burden being thrust onto state budgets. Maine’s expansion legislation also contained a provision to move the entire MaineCare population into private managed care plans. Demonstrating the depth of partisan divisions, this most recent legislative attempt to expand Medicaid in Maine—an attempt that included the reauthorization provision, the automatic opt-out, and the managed care provisions—was still pronounced “dead on arrival” by Republican opponents (Lawlor 2015). Having not yet expanded Medicaid or established a state-based exchange, Maine is very much the outlier in the Northeastern United States. Despite Republican governors in Pennsylvania and New Jersey, Maine is the only state in the Northeast that has not yet implemented the ACA’s Medicaid expansion. In addition, Maine is the only state in New England not to establish a state-based exchange. Following the failure of Republicans to repeal and replace the ACA in 2017, Governor LePage stated his intention to “withdraw the state and just go do our own thing” (Thistle 2017). Evidence from other, less visible, parts of the ACA, however, suggest that Maine has neither withdrawn itself from federal health policy, in general, nor the ACA, in particular. In 2012, Maine was awarded a round 1 SIM test grant for just over $33,000,000. In financial terms, this grant represents a considerably smaller investment in the ACA compared to an expansion of Medicaid, but as one of just six states to receive such a grant, Maine assumed the position of a policy leader in this realm of ACA-related reforms. The impetus to apply for the SIM award came from the state’s Health Commissioner and leaders in MaineCare, the state’s Medicaid program. According to a key informant, the Governor’s office supported the application once the State Health Commissioner argued that reform efforts that were, at the time, only a “gleam in people’s eyes” could be tested with a SIM award (Maine 2017). SIM was also flexible, allowing states to develop their own reform plans instead of CMS instructing them to implement a prescriptive reform model. While the ability to design reforms that were particularly suited to Maine was attractive to state officials, Maine’s SIM application touted the potential for replicating Maine’s reforms and innovations in other states. Maine’s public health officials were, therefore, positioning Maine to be a leader and model for ACA-related policy innovations. In its SIM application, Maine described itself as an “incubator” for federal pilots and demonstrations, prefacing its innovation plan and test grant application on the existence of a strong foundation of existing health reform initiatives. In 2012, at the time Maine filed its SIM application, there were at least nine ongoing CMS/ACA-funded demonstrations or pilots operating in the state. One product of this legacy of reform is the creation of a “collaborative culture” and a supportive “environment for change” (Maine DHHS 2012a, 26). This environment of change and collaboration spans the public–private divide, and includes leadership and participation from a broad spectrum of payers and providers. The tradition of public–private partnerships has left Maine with what has been described as an “improvement overload” (Maine DHHS 2012b, 18). The challenge created by this “improvement overload” is one of coordination—the need to establish an organizational framework and cohesive strategy for reform. Maine’s health policy leadership envisioned the SIM grant as not only a way to build and implement new reforms, but also as a mechanisms for coordinating existing reform efforts and allowing Maine to achieve more from their ongoing reform efforts (Maine DHHS 2012a, 26–27). The SIM grant was, therefore, an effort to deepen the impact of federal-state partnerships in Maine and ensure successful and sustainable reform initiatives (Maine DHHS 2012a, 27). The governance of the test model in Maine was led by two separate bodies: the SIM Core team and the SIM Steering Committee. Members of the SIM Core come entirely from state agencies including the Health Commissioner’s Office, MaineCare, and the state’s Center for Disease Control and Prevention. Although these actors consulted with stakeholders about the goals for the SIM grant in advance of applying, the SIM Steering Committee, the body that included private sector representatives, was not convened until after the state received the federal award. Included in this group is the Maine Healthcare Management Coalition, a purchaser-led convener of employers, providers, payers and patients. Also on the steering committee are citizen representatives, HealthInfoNet, the state's health information exchange, and Maine Quality Counts, an independent group supporting patient-centered care. These private sector actors are joined on the Steering Committee by individuals from the public sector representing the state legislature, the Bureau of Insurance, MaineCare and the Maine CDC. In this respect, the governance of the Maine SIM appears to sit in between the state-run Oregon SIM model and the non-governmental AHCI. Maine’s test model has multiple components, but focuses on a few central goals. Two of its most important goals are to increase the role of primary care in managing patients’ care and to integrate both physical and behavioral health care. One of the primary mechanisms used to achieve these two goals is the creation of more centralized data platforms that allow providers across multiple sites of care to see the larger picture of care. According to a key informant, the test has been successful in a number of respects, including increased access for behavioral health providers to their patient’s medical records, which has improved care coordination and decreased over-prescribing (Maine 2017). The creation of behavioral health homes has focused on preventing diabetes in high risk populations and has shown a decrease in rates of diabetes. The effort to train healthcare leaders in how to transform delivery organizations to adopt team-based care was less successful in that the trainings were time consuming and attracted fewer participants than expected. The key informant interview did not uncover any stakeholder politics that stood in the way of Maine’s implementation of its test model (Maine 2017). Moreover, stakeholders included in the implementation process selected several aspects of the test model to continue after the final year of the SIM award. Oregon: Building on a State Policy Legacy Oregon’s decision to apply for a SIM grant, especially given support for other elements of the ACA, is less surprising than it is for Arkansas or Maine. We include this case to examine the factors that drove state decisions on SIM in states with stronger Democratic control. In November 2010, Democrat and former two-term governor Dr John A. Kitzhaber was re-elected as the governor of Oregon, continuing the streak of Democrat governors since 1987. In his first two terms, Kitzhaber created the Oregon Health Plan (the state Medicaid health plan) as a state-run managed care plan and contributed to the increase of Medicaid enrollment by 50 percent (Goldsmith and Henderson 2017). Despite this initial early success, enrollment in the OHP eventually declined significantly (LeCouteur et al. 2004) and the recession resulted in rolling back benefits. When he was re-elected to a third term in 2010 after a hiatus, he once again made Medicaid healthcare reform a centerpiece of his administration. He spearheaded the creation of Coordinated Care Organizations (CCOs)—networks of providers, payers and community organizations that cover physical health care, mental health, addiction treatment, and often dental care—that reduce costs and integrate care all under one budget.8 During his administration, he experimented with Medicaid eligibility determination and “fast track enrollment” for individuals who qualified based on participation in other state-run programs. This helped increase enrollment and reduce uninsurance rates. He received overwhelming support on healthcare reform from the legislature through his terms. From 2009 to 2011, Democrats were in control of the Oregon legislature. The 76th legislature of Oregon was nearly an even split between the two political parties, but they remained aligned on changes to the state Medicaid program (Klein, McCarthy, and Cohen 2014). A key informant during an interview said that “there was a recognition that the delivery system need[ed] to be changed” and “the legislation that placed CCOs in place was very bipartisan” (Oregon 2017). At the onset of the SIM grant application process, the Oregon legislature had already passed key legislation (HB 2009) to transform state spending and delivery of care. Among other things, HB 2009 created the Oregon Health Authority to oversee state health programs, initiated the state's Patient-Centered Primary Care Home (PCPCH) program, and outlined a plan for an Oregon Health Insurance exchange. Further legislation in 2011 and 2012 under the direction of Governor Kitzhaber led to the development of CCOs for implementation in Medicaid as cost-controlling measure and to allow for integrating care under one budget. In 2012, Oregon had also received approval for a Medicaid Section 1115 waiver bringing in $1.9 billion in return for a 2 percent reduction in per-capita Medicaid spending by year two; and a Section 2703 waiver under the ACA to integrate care in Medicaid Health Homes. While the key pieces of Oregon’s innovation under the SIM grant were placed into action prior to the receipt of the grant, the federal funding stream of $45 million allowed the state to implement the Section 1115 waiver and the use of CCOs, and to launch the Transformation Center which provides technical assistance and coordination between the different stakeholders, private and public. One interviewee mentioned that the SIM grant allowed them to roll out the plans and legislation they had already planned to do, and hasten the pace of certain goals, such as rolling out aspects of CCOs to the public employee health benefits program. The interviewee stated that the funding “expanded key elements of the CCO’s to public employees… [including] quality incentives, aligning incentive measures and keeping cost growth down, transparency in performance, [and] engaging members in [their health] care” (Oregon 2017). The grant also led to the establishment of the Office of Equity Inclusion, and investment into public and private partnerships for housing supports. The SIM grant is covered by the state at the Oregon Health Authority, which was reorganized in 2015, following the election of Democrat Kate Brown as governor. The SIM grant is managed at the state-level, but the Transformation Center that supports the CCO’s facilitates interaction between different stakeholders, and also provides training for clinicians to better implement the Triple Aim. Interviews demonstrated that the process of generating support for the CCOs came at the state level initially in terms of learning collaboratives, but soon the need lay in technical support. The public also seemed to be largely unaware of the federal financial support for CCO implementation and its ties to the ACA. Sixteen CCOs are currently operating in the state, and 90 percent of individuals are enrolled in one of them.9 Research shows that transformation into CCO’s led to a seven percent reduction in cost (McConnell et al. 2017). According to a key informant, the failures associated with Oregon’s state exchange drew public attention, while its SIM program did not, “frustration in Oregon came from our own health exchange” (Oregon 2017). The rollout of the state-based health insurance marketplace was marred by technical errors, leading the state to abandon their state website in 2014 for the federal exchange (Foden-Vencil 2014). Interviews with key stakeholders in Oregon suggest that some individuals may have enrolled in CCOs without much political antagonism perhaps in part due to the lack of association of these new models with the ACA. Like Arkansas, implementation of the SIM grant in Oregon seems to have moved quickly, largely because state leaders overcame legislative and planning hurdles prior to the state’s receipt of the funds. One interviewee suggested that reformers in Oregon were in conversation with federal officials involved in designing the SIM project such that reformers in Oregon could anticipate the potential influx of federal dollars to help with the state initiative (Oregon 2017). Indeed, the Oregon case illustrates that states may be more likely to take up intergovernmental policy innovations when stakeholders at the state level are already poised to leverage federal dollars to accomplish their goals. Discussion The results of our quantitative analysis suggest that a state’s participation in SIM is shaped by the partisan composition of state government, its prior policy investments in Medicaid managed care, and its administrative capacity. As in prior studies of ACA implementation, we found that states with Democratic-controlled legislatures and Democratic governors were more likely to participate in SIM. To some extent, the significance of the partisan variable in our quantitative model casts doubt on the notion that small-scale, low-salience grant programs will generate intergovernmental cooperation, rather than conflict (Gormley 2006). When compared to other intergovernmental components of the ACA such as the Medicaid expansion, insurance exchanges, and regulatory reforms, the SIM initiative imposed few restrictions or requirements on states and was virtually invisible to mobilized opponents of the ACA. Instead, it offered states opportunities to finance payment and delivery reforms that reflected existing private-sector trends. Nevertheless, states with stronger Republican control were significantly less likely to participate in SIM. Thus one implication of our study is that partisan polarization may impose barriers to intergovernmental cooperation, even when federal programs are designed to enable credit claiming by state-level officials. Nevertheless, partisanship appears to play a smaller role in shaping participation in SIM than it does in other areas of ACA implementation, such as the creation of insurance exchanges or Medicaid expansion. Regardless of partisanship, states with high levels of administrative capacity and a legacy of policy investments in Medicaid managed care were more likely to participate in SIM. Additionally, the quantitative model suggests that interest-group opposition to the ACA appears not to have had an impact on SIM participation. Interviews in the Republican-dominated states revealed that there was no effort on the part of those implementing SIM reforms to keep the award beneath the radar. In fact, the reverse is true in that actors in Arkansas and Maine used the award to portray themselves as leaders in health policy reform with respect to other states. This is perhaps especially surprising in Maine where a conservative governor had publicly rejected engagement with the ACA. Instead of trying to reconcile that position with applying for and receiving the SIM award, it appears that the Governor’s office never even felt the need to address the inconsistency. While one might have guessed that there was political cover for the SIM application given the presence of the Maine Healthcare Management Coalition, it does not appear that anyone pressed the supportive coalition into service on this point. Key informants from Oregon point out that they publicized their program and indicated that they are a frequent source of information for other states looking to learn from Oregon’s experience. The visibility Oregon generated around its program, however, matches its willingness to engage with other components of the ACA. Thus, while policy cannot be easily engineered to reverse partisan opposition to intergovernmental reforms, the SIM case illustrates that pre-existing policy legacies or administrative capacity may make it easier for states to participate in intergovernmental policy innovation. Key informants in Oregon and Arkansas indicated that reform efforts were planned out before SIM, but that the federal money increased either their legitimacy in pursuing reforms, expanded the scope of what they were able to accomplish, or both. These informants gave the impression that federal officials knew about existing reform efforts and used the funding to improve the chances that these states had the resources and legitimacy to move their existing reform efforts forward. Thus, those who designed the program may have been consciously trying to leverage the reforms in innovative states to encourage others to follow suit. In addition, key informants indicated that CMMI provided technical assistance to help states implement their reform efforts and created opportunities for learning across the states implementing Round 1 model grants. The structure of the SIM initiative appears to be designed so that the lessons from the lead states can be used both to encourage broader participation from other states and to inform those efforts. At the same time, it is possible that this mechanism might have encouraged states that had no reforms underway to try out new reform efforts in the state. While leaders of the SIM initiative in Maine had reform goals, the plans for how to achieve those goals were not articulated until the state and its partners sat down to write the grant application. This suggests that, even in Round 1, the SIM mechanism might have stimulated reform efforts rather than placing existing efforts on better footing. There are limits to our findings. First, our quantitative analysis does not differentiate between levels of state participation in the SIM initiative. Yet, national evaluations of SIM reveal variations in the depth and character of state implementation (e.g., Feder et al. 2017). Thus, it is quite possible that, were we to code state participation according to the intensity of investment of personnel, resources, and attention, the effect of our key variables may change. Additionally, because our quantitative analysis is cross-sectional, we do not have the ability to evaluate why states participation may have fluctuated over time. For instance, we cannot say why a state may have participated in the initiative at time T but dropped out at time T + 1. As the qualitative case studies reveal, there are clearly variables that affected state participation—including stakeholder participation and the concentration or dispersion of payers and providers in each state—but are not captured in logistic regression model. Using data to capture stakeholder participation or market power among payers and providers could allow us to add this factor to the quantitative model to shed light on how market structure might shape reform opportunities. Finally, the qualitative interviews do not include data from states that decided not to apply for a SIM grant or states that applied and dropped out. Instead, the qualitative interviews focused on a set of states that applied for and received test grants in Round 1. The article sheds light on what might be different about these states that would explain their status as early test grant recipients and explores what factors shaped their decision to apply. Future research on states that did not submit applications in either round could shed additional light on whether factors other than those captured in the quantitative model (capacity, partisanship, policy legacy, etc.) explain SIM participation. This would be especially relevant for non-participating states that our quantitative model predicted would have applied for SIM grants (FL, IN, MO, and NC). Equally, interviews with participating states that our model predicted would not participate (NH, OK, TX) could shed light on the factors that led these states to be outliers in the other direction. Our case studies are not illuminating on why Republican-dominated states that applied for and received Round 1 SIM grants experienced no partisan pushback for this engagement. One way to interpret this is that while payment and delivery reform is intensely political among the organized interests with a stake in outcomes, the nature of politicization is more pragmatic than partisan. Interest groups that have a business interest in SIM grants appear to press for a seat at the table and do not try to animate larger partisan politics or mobilize voters to undermine these reform efforts. Conclusion As of early 2018, the fate of the ACA at the national level remains unclear. Republicans in Congress failed to pass comprehensive legislation to repeal and replace the law, and only succeeded in eliminating the law’s individual coverage mandate by attaching it to a significant tax cut. Yet the ACA’s intergovernmental structure guarantees that important debates about the future of health reform will occur outside Washington, DC. Following Congress’s repeal of the individual mandate, proposals for new, more stringent mandates emerged in states like Connecticut and Maryland, where support for the ACA was strong. With encouragement from co-partisans at the Center for Medicare and Medicaid Services, states like Indiana, Kentucky, and Arkansas pursued Medicaid waivers that imposed work requirements on beneficiaries, an effort which the Obama administration had rebuffed. From these examples, one might conclude that future intergovernmental conflicts over the ACA may well retain a partisan edge. The evidence in this paper confirms that partisan control of government shapes states’ decisions about even small-scale, low-salience components of the ACA like the SIM initiative. Nevertheless, our findings also offer important lessons about the limits of partisanship as an explanation of state participation in intergovernmental policy innovations. Regardless of partisan control, states were more likely to participate in SIM when they had a high level of administrative capacity and a strong legacy of managed-care reforms within the Medicaid program. Our qualitative findings reveal that even in states where partisan opposition to the ACA was strong, entrepreneurial bureaucrats advanced SIM initiatives by uniting diverse coalitions of payers and providers. In Arkansas, the state's three largest payers worked in concert to bring providers to the table to negotiate payment reforms. In Maine, SIM efforts were driven by the Health Commissioner and MaineCare administrators and aimed at increasing care coordination, integrating physical and behavioral health, and improving the state’s data platform. In Wisconsin, where state leaders opposed the creation of an insurance exchange and the expansion of Medicaid, the state’s Medicaid program used SIM funds to carry out a new program of care integration for medically complex children.10 This study has important implications for the future of health reform in the states. In the short term, an almost unprecedented level of partisan and interest-group conflict hampered intergovernmental collaboration on the ACA. As such, it seems likely that the SIM’s incremental, low-salience character helped to insulate it from the kind of friction that met the Medicaid expansion and the implementation of insurance exchanges. Moreover, states’ participation depended on more than the ACA’s mix of “carrots and sticks.” Especially when partisan opposition to the ACA was strong, what mattered was SIM’s alignment with states’ pre-existing policy commitments and the availability of a talented administrative core. To be sure, evidence about SIM’s design and implementation hardly provides a “magic formula” for generating intergovernmental collaboration, which is necessarily contingent and context-specific. Yet, if partisan conflict over “Obamacare” continues, policymakers and activists interested in intergovernmental collaboration—even for purely instrumental reasons if not for the sake of “cooperative federalism”—may have to reconsider the scale at which innovations are likely to occur; the need to strengthen policymaking capacity in states; and the value of aligning national program goals with states’ policy demands. Beyond specific knowledge about which policy interventions improve health outcomes, policymakers need a better understanding of how the federal system works, and how to stimulate intergovernmental collaboration in an age of polarization. Supplementary Data Supplementary data are available at Publius: The Journal of Federalism online. Footnotes We are grateful to John Dinan, Shanna Rose, and two anonymous reviewers for excellent feedback on this article. Participants at the Berkeley Research Workshop in American Politics and the 2017 meeting of the American Political Science Association offered helpful comments on an earlier draft of this article. Nisha Kurani provided valuable research assistance. This research was supported by a grant from the Robert Wood John Foundation’s Scholars in Health Policy Research Program (UCSF/UCB). 1 Of the nineteen states, sixteen received Model Design Awards to support efforts to produce state innovation plans, while three other states received Model Pre-test Awards to further develop and refine state innovation plans 2 We thank Alex Hertel-Fernandez for providing us with these data. 3 Minnesota, Vermont, and Massachusetts, which were the other three states to be awarded a first round test grant, followed Oregon in both expanding Medicaid and constructed a state-based health insurance exchange. 4 This count, and our foregoing analysis, excludes the District of Columbia for purposes of comparability. The District of Columbia received a model design award only. 5 During an episode of care, providers are paid on a fee-for-service basis, but are penalized after care is given if the provider exceeds acceptable spending levels. Providers who meet performance standards and whose average costs for an episode of care fall into the “commendable” range are eligible for gainsharing. 6 At time of publication, Governor LePage has refused to implement the voter-approved Medicaid expansion and has announced a number of conditions that must be met by the legislature prior to any implementation. Among the conditions is that funding the Medicaid expansion cannot require increased taxes or the use of state’s budget stabilization fund and that Medicaid waitlists for the elderly and disabled be eliminated prior to funding the expansion (Cousins 2017). 7 Two Republican Senators, Roger Katz and Thomas Saviello, have repeatedly joined Democrats in crafting and voting for Medicaid expansion legislation. 8 For its description of its Coordinated Care Program, see the Oregon Health Authority website: http://www.oregon.gov/oha/HPA/Pages/CCOs-Oregon.aspx 9 Oregon Health Authority’s report on its CCOs can be found here: http://www.oregon.gov/oha/HPA/ANALYTICS-MTX/Documents/HST%20Annual%20Report%20-%202016.pdf. 10 We did not interview individuals involved in Wisconsin’s SIM grant. Data on its program can be found at Center for Medicare and Medicaid Innovation, Health Care Innovation Awards Round Two Project Profiles, July 2014, https://innovation.cms.gov/Files/x/HCIATwoPrjProCombined.pdf. References Arkansas . 2017 . Interview with government official. July 31. Barrilleaux Charles , Rainey Carlisle . 2014 . The politics of need: Examining governors’ decisions to oppose the “Obamacare” Medicaid expansion . State Politics & Policy Quarterly 14 4 : 437 – 460 . Google Scholar CrossRef Search ADS Béland Daniel , Rocco Philip , Waddan Alex . 2016 . Obamacare wars: Federalism, state politics, and the Affordable Care Act . Lawrence : University Press of Kansas . Blewett Lynn A. , Spencer Donna , Huckfeldt Peter . 2017 . Minnesota integrated health partnership demonstration: implementation of a Medicaid ACO model . Journal of Health Politics, Policy, and Law 42 6 : 1127 – 1142 . Google Scholar CrossRef Search ADS Bulman-Pozen Jessica , Metzger Gillian E. . 2016 . The president and the states: Patterns of contestation and collaboration under Obama . Publius: The Journal of Federalism 46 3 : 308 – 336 . Google Scholar CrossRef Search ADS Callaghan Timothy , Jacobs Lawrence R. . 2014 . Process learning and the implementation of Medicaid reform . Publius: The Journal of Federalism 44 4 : 541 – 563 . Google Scholar CrossRef Search ADS Center for Medicare and Medicaid Innovation (CMMI) . 2017 . Demonstration and Waiver List. https://www.medicaid.gov/medicaid/section-1115-demo/demonstration-and-waiver-list/waivers_faceted.html. Center for Medicare and Medicaid Innovation (CMMI) . 2017 . State Innovation Models: General Information. https://innovation.cms.gov/initiatives/state-innovations/ Commonwealth Fund . 2009 . Aiming higher: Results from a state scorecard on health system performance. http://www.commonwealthfund.org/publications/fund-reports/2009/oct/2009-state-scorecard Council of State Governments . 2012 . Book of the states . Lexington, KY : Council of State Governments . Cousins Christopher , Shepherd Michael . 2016 . How a MaineCare expansion plan differs from past 5 failures . Bangor Daily News February 22. https://bangordailynews.com/2016/02/22/the-point/how-a-new-mainecare-expansion-plan-differs-from-the-past-5-failures/. Cousins Christopher. 2017 . LePage digs in for Medicaid expansion funding battle . Bangor Daily News December 11. https://bangordailynews.com/2017/12/11/politics/lepage-digs-in-for-medicaid-expansion-funding-battle/. Derthick Martha. 1970 . The influence of federal grants: Public assistance in Massachusetts . Cambridge, MA : Harvard University Press . Google Scholar CrossRef Search ADS Feder Judy , Weil Alan R. , Berenson Robert , Dolan Rachel , Lallemand Nicole , Hayes Emily . 2017 . Statewide payment and delivery reform: Do states have what it takes? . Journal of Health Politics, Policy, and Law 42 6 : 1113 – 1125 . Google Scholar CrossRef Search ADS Flagg Robin. 2016 . Medicaid expansion: A tale of two governors . Journal of Health Politics, Policy, and Law 41 5 : 997 – 1031 . Google Scholar CrossRef Search ADS Foden-Vencil Kristian. 2014 . Oregon raises while flag over its health exchange . Kaiser Health News April 28. http://khn.org/news/oregon-raises-white-flag-over-its-health-exchange/. Gifford Kathleen , Smith Vernon K. , Snipes Dyke , Paradise Julia . 2011 . A profile of Medicaid managed care programs in 2010: Findings from a 50-state survey . Washington, DC : Kaiser Commission on Medicaid and the Uninsured . Goldsmith Jeff , Henderson Bruce . 2017 . Oregon’s high-risk, high-reward gable on Medicaid expansion . Health Affairs Blog January 10. http://healthaffairs.org/blog/2017/01/10/oregons-high-risk-high-reward-gamble-on-medicaid-expansion/. Gormley William T. 2006 . Money and mandates: The politics of intergovernmental conflict . Publius: The Journal of Federalism 36 4 : 523 – 540 . Google Scholar CrossRef Search ADS Haeder Simon F. , Weimer David L. . 2013 . You can’t make me do it: State implementation of insurance exchanges under the Affordable Care Act . Public Administration Review 73 ( s1 ): S34 – S47 . Google Scholar CrossRef Search ADS Haeder Simon F. , Weimer David L. . 2015 . You can't make me do it, but i could be persuaded: a federalism perspective on the Affordable Care Act . Journal of Health Politics, Policy and Law 40 2 : 281 – 323 . Google Scholar CrossRef Search ADS PubMed Hertel-Fernandez Alexander , Skocpol Theda , Lynch Daniel . 2016 . Business associations, conservative networks, and the ongoing Republican war over Medicaid expansion . Journal of Health Politics, Policy, and Law 41 2 : 239 – 286 . Google Scholar CrossRef Search ADS Jones David K. 2017 . Exchange politics: Opposing Obamacare in battleground states. New York : Oxford University Press . Keller Ann , Burrowes Sahai , Flagg Robin , Goldstein Gabrielle . 2014 . Credibility as Liability: Wing-clipping at the Centers for Disease Control and Prevention. Paper presented at the Annual Meeting of the American Political Science Association, August 28-31, Washington, D.C. Kincaid John. 2018 . Why coercion and cooperation coexist in American federalism. In Intergovernmental relations in transition , ed. Hamilton David K. , Stenberg Carl W. , 37 – 57 . New York : Routledge . Klein Sarah , McCarthy Douglas , Cohen Alexander . 2014 . Health share of Oregon: A community-oriented approach to accountable care for Medicaid beneficiaries. The Commonwealth Fund Case Studies of Accountable Care Systems , October 2014. New York: Commonwealth Fund. Krane Dale , Noh Shihyun . 2018 . Partisan polarization, administrative capacity, and state discretion in the Affordable Care Act. In Intergovernmental relations in transition , ed. Hamilton David K. , Stenberg Carl W. , 115 – 140 . New York : Routledge . Lawlor Joe. 2015 . Republican senators again propose MaineCare expansion . The Press Herald December 27. http://www.pressherald.com/2015/12/27/republican-senators-again-propose-medicaid-expansion/ LeCouteur Gene , Perry Michael , Artiga Samantha , Rousseau David . 2004 . The impact of Medicaid reductions in Oregon: Focus group insights. Report by the Kaiser Commission on Medicaid and the Uninsured, December 2004. LePage Paul. 2014 . Governor LePage vetoes Medicaid expansion for third time. http://content.govdelivery.com/accounts/MEGOV/bulletins/b01a5e Maine Department of Health and Human Services . 2012a . Testing the Maine Innovation Model: Project Narrative. Augusta, ME: Maine Department of Health and Human Services. Maine Department of Health and Human Services . 2012b . Maine State Health Care: Innovation Plan. Maine . 2017 . Interview with government official. August 16. Moretto Mario. 2014a . LePage vetoes Medicaid expansion, calls the effort “ruinous” for Maine’s future . Bangor Daily News April 9. http://bangordailynews.com/2014/04/09/politics/lepage-vetoes-medicaid-expansion-calling-the-effort-ruinous-for-maines-future/. Moretto Mario. 2014b . Moderate Republicans unveil Medicaid compromise that expands coverage, dramatically reforms system . Bangor Daily News February 25. http://bangordailynews.com/2014/02/25/politics/medicaid-compromise-drafted-by-two-gop-senators-to-hit-state-house-this-week/ National Conference of State Legislatures . 2012 . State Partisan Composition Map, 2012. http://www.ncsl.org/research/about-state-legislatures/partisan-composition.aspx. Oregon . 2017 . Interview with two government officials. July 28. Posner Paul. 1998 . The politics of unfunded mandates. Washington, DC : Georgetown University Press . Rigby Elizabeth , Haselswerdt Jake . 2013 . Hybrid federalism, partisan politics, and early implementation of state health insurance exchanges . Publius: The Journal of Federalism 43 3 : 368 – 391 . Google Scholar CrossRef Search ADS Rose Shanna. 2015 . Opting in, opting out: The politics of state Medicaid expansion . The Forum 13 1 : 63 – 82 . Google Scholar CrossRef Search ADS RTI International . 2014 . State innovation model (SIM) initiative evaluation: Model design and model pre-test evaluation report . Research Triangle Park, NC : RTI International . Sorenson Corinna , Gusmano Michael K. , Oliver Adam . 2014 . The politics of comparative effectiveness research: Lessons from recent history . Journal of Health Politics, Policy, and Law 39 1 : 139 – 170 . Google Scholar CrossRef Search ADS Thistle Scott. 2017 . LePage considers starting Maine health insurance plan to replace Affordable Care Act . The Press Herald March 28. http://www.pressherald.com/2017/03/28/lepage-considers-starting-health-insurance-program-to-replace-obamacare/. Thompson Frank J. , Burke Courtney . 2007 . Executive federalism and Medicaid demonstration waivers: Implications for policy and democratic process . Journal of Health Politics, Policy and Law 32 6 : 971 – 1004 . Google Scholar CrossRef Search ADS PubMed Toone Kathryn. 2014 . Trick or treating for SIM awards . Leavitt Partners October 28. https://leavittpartners.com/2014/10/trick-treating-sim-awards/ Van Vleet Amanda , Paradise Julia . 2014 . The State Innovation Models (SIM) Program: An overview. Kaiser Family Foundation Fact Sheet. http://www.kff.org/medicaid/fact-sheet/the-state-innovation-models-sim-program-an-overview/ Walker David. 2000 . Toward a functioning federalism . Chatham, NJ : Chatham House . Weissert Carol S. , Uttermark Matthew J. . 2017 . Glass half full: Decentralization in health policy . State and Local Government Review 49 3 : 199 – 214 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of CSF Associates: Publius, Inc. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
TI - Politics at the Cutting Edge: Intergovernmental Policy Innovation in the Affordable Care Act
JF - Publius The Journal of Federalism
DO - 10.1093/publius/pjy010
DA - 2018-05-19
UR - https://www.deepdyve.com/lp/oxford-university-press/politics-at-the-cutting-edge-intergovernmental-policy-innovation-in-3s6PCZsEcI
SP - 1
EP - 453
VL - Advance Article
IS - 3
DP - DeepDyve
ER -