Welfare,innovation capacity,and economic performance: Evidence from American federalism

Abstract

Despite years of increased federal welfare spending, it is unclear how federal financial investments in welfare areas influence state economies. Based on the theory of functional federalism, this paper examines the impact of federal welfare spending on state economic performance. Employing error correction models, this research finds that state economic growth and volatility are directly observed by the long-term impact of welfare spending. The results also show that increased federal welfare spending boosts state economic performance indirectly, by motivating states to generate higher innovation capacity, especially, in knowledge-based economic development programs. This research extends our understanding of the determinants of state economic performance by addressing intergovernmental redistributive transfers, and state government’s innovation capacity.

Keywords

Economic performance functional federalism innovation capacity welfare policy

Since the 1930s, modern federalism has expanded federal responsibility, especially in the area of human services. Although the era of New Federalism has been a critical period for advocating devolution, the federal government has continued to play an active and at times a resurgent role in redistributive areas (Bovbjerg et al., 2003). Since 1960, the share of the country’s spending on welfare programs at state and local levels has steadily declined, while federal government expenditures have more than doubled in those areas, since 2000. For example, in 2007, the federal government spent $1.45 trillion (almost half of all federal spending) on redistributive programs, such as Medicaid, housing assistance, unemployment compensation, and food stamps (Miron, 2011).

In recent years, policy makers have utilized a model of federalism that shares functional responsibility for major policy issues with states. While the federal government has transferred a vast amount of funds, particularly to redistributive programs, state governments have also allocated considerable financial resources to economic development programs. The federal government’s comprehensive measure in dealing with an economic recession involved spending a significant amount of money in welfare areas, in part, to rehabilitate the sluggish economy. In contrast, state governments have remained passive in investing in welfare programs, not only because state politicians fear that their states will become “welfare magnets,” but also because they cannot afford to adequately finance large welfare programs (Peterson, 1995; Peterson and Rom, 1990). Instead, market forces and political pressure have forced state governments to focus their efforts on stimulating economic development (Brace and Mucciaroni, 1990; Eisinger, 1989). However, due to the increased financial burdens incurred by welfare programs, and limited revenue streams, many states’ resources are inadequate for stimulating innovation capacity that would further economic performance (Shin and Hall, 2018). It is reasonable to expect, therefore, that the federal government would help states increase economic performance by transferring welfare resources.

Surprisingly, scholars of American federalism have not yet systematically examined and tested the relationship between the federal effort on welfare programs and the overall economic outcomes in the American states. This research identifies expectations regarding how federal welfare spending impacts state economies, both theoretically and empirically. In this paper, I suggest that federal welfare funds spur state economic performance indirectly, by increasing slack resources that state governments can utilize, which would increase the state government’s innovation capacity in the economic development programs, especially knowledge-based strategic programs. It is also suggested that federal welfare funds can increase state economic performance directly, by improving labor productivity in ways that assists people who are temporarily living in poverty, but are eligible and capable to work, and to find employment (Aghion et al., 1999; Conley, 2010). In order to measure the intervening variable, this paper develops an index of state innovation capacity by creating composite variables that are based on overall state expenditures (both knowledge-based newer development programs and more traditional development programs). State economic performance is examined by measuring both quantitative and qualitative economic outcomes, which are operationalized by two indicators: economic growth measures (gross state product (GSP) and personal income) and economic volatility measures (a standard deviation of GSP growth rate and a standard deviation of personal income growth rate), respectively.

The results of error correction models (ECMs) show that state economic growth and volatility are directly observed by the long-run impact of federal welfare spending. Results also show that increased federal welfare spending boosts state economic growth, not state economic volatility, indirectly through increasing state innovation capacity particularly in knowledge-based economic development programs. The findings support a fundamental assumption that federal welfare funds would lead to higher economic growth and lower economic volatility within the American states.

The federal government’s functional responsibility in redistributive programs

A large number of studies examining state and local governments in America focus on the economic impacts of two broad economic policies: locational-model policy and human-capital development policy. For instance, studies that focus on the locational model—that is, on government policies that can affect the location decisions of business and industry—investigate the effect of tax incentives and government expenditures on economic growth (e.g., Brulhart and Schmidheiny, 2015; Crain and Lee, 1999; Poulson and Kaplan, 2008; Prillaman and Meier, 2014). In contrast to this, some studies focus considerable attention on how human–capital development policies, such as education, job training, and health programs impact state economic growth (e.g., Deskins et al., 2010; Feiock, 1999; Murphy and Topel, 2016). Economic growth can also be explained within policy, management, and political boundaries. Many studies have examined the effect of institutional power, management, party control, and ideological differences, on economic development policies (e.g., Bae et al., 2012; Brace, 1993; Howell-Moroney, 2008; Rigby and Wright, 2013). Additionally, a significant number of scholars have examined the relationship between the level of fiscal decentralization of American states and state economic growth (Akai and Sakata, 2002; Shin, 2018; Xie et al., 1999).

Based on the extant literature, conventional wisdom on state economic growth holds that American states have been the primary engines of economic development efforts not only because states compete with one another to provide competitive economic environments, but also because state governments can receive more information about efficient policy choices from the marketplace. However, this does not necessarily mean that the federal government remains silent when it comes to state economic performance. Saiz (2001) notes that the devolution of power from Washington to state and local governments not only produces policies of questionable efficiency, but also limits innovation. He suggests that studies of economic development should include the effect of federal government action on reducing the negative effects of interjurisdictional competition in pursuing economic development. Indeed, economic performance at the state level is “set within the context of national forces that inevitably affect state opportunities and create state economic development opportunities, as well as challenges” (Eisinger, 1989: 301).

Traditionally, American states have been reluctant to develop welfare programs. This behavior is predicated on two eventualities. First, increasing redistributive spending will attract the poor from regions where there is an increase in welfare tax rates or where fewer welfare benefits are provided. Second, businesses and firms may move to avoid the cost of redistribution that would eventually increase developmental expenses (Peterson and Rom, 1990). For this reason, the federal government has steadily increased welfare spending since the federal government began the “War on Poverty” in the 1960s (Rector and Sheffield, 2014).¹

Despite considerable evidence indicating the predominant role of the federal government regarding welfare programs, some scholars have suggested that states actually should play an active role in redistribution, though not as much as the federal government. For example, in literature examining fiscal decentralization, Pauly (1973) and Johnson (1988) argue that state and local governments can better manage welfare programs and control welfare spending because they are better informed of the diverse preferences and attitudes of the residents in their jurisdiction. More recently, based on a positive perspective of fiscal function between different levels of government, Gordon and Cullen (2012) find that states in equilibrium take primary responsibility for redistribution, regardless of the level of federal intervention. Other studies identify the conditions that state governments are more likely to increase redistributive expenditures. Bahl et al. (2002) find that state and local governments that receive more federal aid for health and welfare, have higher poverty rates, are less urbanized, and are more likely to increase redistributive spending. However, these discussions are merely suggestive. Indeed, the majority of scholarship from economists contends that which unit of governments is responsible for welfare policies should be dealt with in a way that increases the institutional efficiency of American federalism to benefit the economy in the future (e.g., Bartik, 1999; Callen, 2009; Oates, 1972).

One of the promising theories of American federalism, the theory of functional federalism, provides a potentially powerful explanation on the linkage between federal political–economic influence and state economic growth (Peterson, 1995). Functional theory suggests that each level of government focuses on the function that it can best perform based on political motivation to maximize the institutional efficiency of U.S. federalism (Anton, 1989; Oates, 1972; Peterson, 1995). Peterson (1995) argues that the federal government has focused more on redistributive policy, reallocating economic resources from the rich to the poor, while state governments have been more interested in pursuing growth policy including developmental programs. Based on the functional rationale, federal politicians have focused on welfare areas because regional economic growth can be achieved not only by transferring income through both the tax system and outlays, but by reducing the large costs of welfare projects that states are unable to adequately fund (Bohte and Meier, 2000; Labonte, 2009).² When state governments face a trade-off between equity and growth, they have strong incentives to support economic growth and to reduce redistributive spending.

In the American political context, there exists little consensus that the political goal should be transferring wealth from the rich to the poor (Witte, 1985). The intent of the federal government to motivate economic development by raising its investment in welfare programs might be justified through the merits of the social development approach that “calls for harmonizing social policy and economic development, and offers a conception of redistribution based on investments in people and communities” (Midgley, 1999: 16). Federal welfare funds can help states increase expenditures on welfare policies that are necessary but politically unpopular. If states want to spend their revenue on development policies, with welfare spending becoming a fiscal burden, then federal welfare support can lower the marginal cost of state welfare activity, and in turn, the benefits liberate states to focus more on pro-growth development policies (Peterson and Rom, 1990).

Federal welfare spending, state innovation capacity, and state economic performance

Since the 1960s, states have received an ever-growing slice of the revenue pie, which has motivated state politicians to pursue more innovation-driven economic development policies that would attract more new industries and markets than the security-oriented policies that focus on retaining existing industries and resources (Hall and Howell-Moroney, 2012). A state government’s innovation has been defined as a “program or policy that is new to the states adopting it, no matter how old the program may be or how many states have adopted it” (Walker, 1969: 881). Thus, state innovation capacity in economic development programs refers to the state government’s ability to generate or to adopt new development policies or programs. Particularly, the demand-side theory, the newer theory of economic growth, emphasizes policy innovation by suggesting that the government’s strategic goal to further economic development is focused on discovering and improving new markets, productive capacity, and the quality of jobs for indigenous industries (Brace and Mucciaroni, 1990; Mahroum and Al-Saleh, 2013). Thus, the propensity of state governments to develop new resources and to create strategic investment for their regional economic growth will incentivize states to innovate rather than direct newly available resources to traditional programs. In the contemporary economy, such traditional incentives as capital and labor for production to reduce factor cost are no longer important inducements to attracting firms and businesses (Feldman and Francis, 2004). To this end, economic policies of states have been directed by innovation capacity that stimulates the potential not only to create new economic opportunities but also to ensure economic growth (Hall, 2007; Hall and Howell-Moroney, 2012; Kwon et al., 2009; Saiz, 2001). Thus, states’ economic development policies are focused on removing social and economic hurdles that may impede policy innovation.

The theory of fiscal decentralization suggests that subnational governments promote economic efficiency by providing public goods that meet different preferences of individuals, increasing accountability of subnational officials, and allowing experimentation and innovation in the pubic-service production process (Oates, 1972; Stansel, 2005). Such benefits of a higher fiscal decentralization suggest that states with more fiscal capacity not only exercise higher innovation capacity, but secure their own source revenues, which in turn translate to economic development (Saiz, 2001; Shin, 2018). Thus, states can expect that an increase in federal welfare funds would secure the state budget on welfare programs that the states have not adequately dealt with; and thereby increase state innovation capacity to generate strategic and progressive economic development programs by lowering the cost of performing redistributive activity and increasing slack resources to be transferred into the regional economy.

As Hall and Howell-Moroney (2012: 229) note, “effective policies will be able to target key resources to their most productive uses as state policy makers strive to overcome barriers and stimulate growth while protecting society’s more vulnerable populations.” Studies in policy innovation and adoption claim that a state’s probability for innovation would be higher if the state could diminish obstacles that may hinder innovation (Berry and Berry, 1990; Mohr, 1969). When states are negligent and incapable of undertaking redistributive programs relative to economic development programs, the state government’s innovation capacity in economic development programs would depend on how effectively they could manage redistributive programs. Thus, state governments may not be able to increase innovation capacity, unless they can adequately protect society’s vulnerable low-income population, as well as the potential for quality in its labor force. As a result, federal financial assistance in redistributive areas not only increases state economic growth, but reduces economic volatility indirectly by reducing substantial obstacles caused by the state’s inability to manage welfare programs, and thereby stimulating state governments to exercise greater innovation capacity in economic development. Taken as whole, the following three hypotheses are proposed:

Hypothesis 1: Increases in federal government spending on welfare programs lead to greater state innovation capacity in economic development programs.

Hypothesis 2: Greater state innovation capacity in economic development programs leads to higher economic growth in the state.

Hypothesis 3: Greater state innovation capacity in economic development programs leads to lower economic volatility in the state.

On the other hand, an increase in federal welfare spending can directly benefit state’s economic growth and stability due to the positive effects of the welfare policy on the economy. Much evidence exists that the motivation behind federal welfare spending has changed. For example, the Family Support Act of 1988 intends to improve poor families’ economic conditions by encouraging recipients to become self-sufficient through education or training, not by providing unconditional cash benefits. The American Recovery and Reinvestment Act of 2009 was based on the Obama government’s belief that the economy would work best when the wealth is spread around (Harwood, 2011). Furthermore, by enacting the Patient Protection and Affordable Care Act of 2010, federal politicians aspired, in part, to reduce the number of uninsured Americans by creating a healthy labor supply. The current Trump administration has proposed new welfare rules with tighter work requirements that encourage the poor to return to the labor force (Restuccia, 2017).

With the changing purpose of the federal government to deal with welfare policy in the context of economic development, the federal government has also increased spending in some welfare programs that are more closely related to potential economic development and growth. Specifically, the federal government has committed to using “welfare-to-work programs,” such as job training, child care, short-term benefits, and the refundable earned income tax credit (EITC),³ by emphasizing investment in human capital and increasing economic self-sufficiency (Greenberg et al., 2009; Hansen and Imrohoroglu, 1992). These welfare benefit programs have placed a strong emphasis on training and workfare in order to help the low income and welfare recipient population reenter the labor market, and thereby increase economic efficiency (Long et al., 1981; Marr and Huang, 2014; Palmer, 1987). For example, Figure 1 presents the changing federal expenditures on specific categories in the Temporary Assistance for Needy Families (TANF) program between 1997 and 2013. The linear trend lines show that while the federal government has steadily decreased expenditures on direct assistance programs as a share of total TANF spending over time, except for the period of financial crisis between 2009 and 2010, it has gradually increased expenditures on non-assistance programs as its share, regardless of national economic conditions.⁴ There is also a well-established conceptual trade-off between welfare versus economic development at the state level. A good example is one of the largest federal welfare funds, TANF block grant, which succeeded Aid to Families with Dependent Children, in 1996. A key argument for TANF block grant is that states have diverted much of their TANF funds to other areas of the state budget by emphasizing welfare as temporary and work focused (Lurie, 2001; Schott et al., 2015). Based on Health and Human Services 2014 TANF financial data, while about 26% of combined federal and state TANF funds went for basic assistance for poor families, the rest of the funding was diverted to work programs that would reduce caseloads, further savings, and benefit potential economic growth (Schott et al., 2015). Thus, increased federal welfare spending would directly benefit state economic performance by not only increasing the developmental goal of welfare programs at the national level, but by motivating states to be engaged in a work-focused welfare system.

Figure 1.

Federal TANF expenditures by ACF-196 spending category.

The federal government can also improve regional economic conditions by distributing welfare money as investments in the functioning of needy individuals and families (Conley, 2010; Knapp and Midgley, 2010). More directly, welfare spending presents a positive income elasticity of demand, that is, it helps people find work, increases their skills, protects their health, promotes a healthy labor supply, improves standards of living in retirement, and increases current consumption levels. All of these things promote economic growth (Aghion et al., 1999; Temple, 1999). Furthermore, higher redistributive spending improves the level of social protection and social security. This promotes efficiency while substituting for the missing markets, encourages individuals to be less risk averse in doing business, and helps in the more efficient administration of societal risks (Ahmad et al., 1991; McCallum and Blais, 1987). Consequently, the close linkage between redistribution and economic growth suggests that the federal government’s investment in welfare programs is potentially beneficial to the economies of all states if they take advantage of it, thus:

Hypothesis 4: Increases in federal government spending on welfare programs lead to higher economic growth in states.

Hypothesis 5: Increases in federal government spending on welfare programs lead to lower levels of economic volatility in states.

Key variables and data

Based on the theory of functional federalism and existing arguments on redistribution and growth, federal welfare spending has a direct effect on state economic performance; it reduces welfare costs that states are burdened with, and it improves the quality of human capital in each state, and thereby increases states’ economic performance. The literature discussed above also identifies a path of indirect influence of federal welfare spending on state economic performance. An increase in federal welfare spending reduces the need for state governments to devote their own resources to redistribution, thereby freeing the states to spend more on economic development areas and increasing the states’ capacities to develop progressive economic development programs, which in turn, increase the states’ economic outcomes.

To test the above hypotheses, a panel data set that includes 50 American states for the period 1972–2010 is utilized. Since this study examines direct and indirect effects of federal welfare spending on state economic performance, it is important to design appropriate models that specify the causal relationship between independent variables, intervening variables, and dependent variables.

First, economic performance can be defined in several ways at each level of analysis. As many scholars have discussed, economic performance usually refers to economic growth or economic activity, which shows how well a state is performing economically over time (Brace, 1993; Crain, 2003). Thus, two major indicators of state economic growth (dependent variables) are employed, which have been commonly used in existing empirical studies: state income including GSP per capita and personal income per capita. GSP is the most comprehensive statistic to measure economic growth of total state income because it indicates the value of all goods and services produced in a state. Per capita personal income is also a critical indicator of a state’s economy, as it provides a measure of the individual’s value and distributional aspect of growth (Brace, 1993; Jones, 1990). Both GSP and income data were deflated to constant 2005 dollars. Data were obtained from the Bureau of Economic Analysis of the U.S. Department of Commerce.

Although traditional measures provide a direct quantitative value of state economic performance, these measures are not comprehensive. Following modern financial theory, both risk and rate of return are critical indicators used to evaluate stable and consistent asset performance (Brealey et al., 2008). Although the standard measures used to gauge state economic performance may adequately represent a degree of quantitative economic growth in individual states, a complete measure of state economic performance needs to include levels of economic growth and volatility of state economic conditions. That is, more balanced states may experience high levels of economic growth and lower levels of economic volatility (or instability). Based on the above arguments, this paper uses economic volatility in a state as the alternative dependent variable.⁵ To measure state economic volatility, a traditional measure of volatility is employed: standard deviation of state growth rate for a given time period (e.g., Kormendi and Meguire, 1985; Ramey and Ramey, 1995).⁶ The standard deviation is taken of a four-year moving average of the growth rate of per capita real GSP and the growth rate of per capita real state personal income.⁷ Higher values of these measures mean increasing growth volatility within a state. Thus, it is expected that federal welfare spending is negatively correlated with the standard deviation of the error term.

As mentioned above, state innovation capacity reflects state government’s economic development efforts to increase its strategic investment on economic development programs that would drive economic growth. Since the major argument of this paper is that states receiving more federal welfare funding would have slack resources to dedicate extra revenues to economic development, it is important to measure state innovation capacity based on the state government’s spending on economic development programs. The policy innovation literature suggests that one government innovation does not necessarily mean that innovation will occur with completely new programs (Kwon et al., 2009; Walker, 1969). On the practitioner side of economic innovation, although the government performance in innovation is usually measured by research and development (R&D) outputs, researchers also use the government’s spending on all available development programs to measure composite innovative activities (U.S. Chamber of Commerce Foundation, 2015; West et al., 2012). Based on these discussions on economic innovation, this study creates a composite variable that measures the degree to which states possess innovation capacity by increasing spending on major economic development programs. Hall (2007) suggests that innovation capacity in economic development can be better measured by using particular human or financial elements.

To determine the spending categories, this research considers two existing theories, demand-side theory and supply-side theory. These theories suggest that states have developed different economic development strategies in two major forms. First, a supply-side theory, which Eisinger (1989) described as an effort of governments to fashion a friendly “business climate” in states, supposes that the government can attract more businesses and industries by providing competitive incentives that lower capital, land, and labor costs. On the contrary, the newer demand-side theory emphasizes that strategic efforts not only discover and improve new markets, but they stimulate modern knowledge and service-based economies (Hall and Howell-Moroney, 2012). From a policy perspective, the government that favors the supply-side strategy focuses on not only supporting established firms, but also increasing the quality of infrastructure and the productivity of workers, while the government that pursues the demand-side strategy prefers to offer selective assistances to knowledge-based programs by giving priority in stimulating education and R&D (Boix, 1997; Halvorsen and Jakobsen, 2013).

Based on these two theories, the data used to measure state innovation capacity represents the two categories of the state government’s strategic economic development programs—conventional economic development programs (i.e., programs based on the supply-side theory) and knowledge-based economic development programs (i.e., programs based on the demand-side theory). Based on the supply-side theory, the variables for the pooled capacity factor analysis include state expenditures on large physical and capital programs including highway, air transit, public building, housing and community development, natural resources, and foreign trade, as well as expenditures on labor productivity, including employment security and primary and secondary education. Based on the demand-side theory, the variables include state expenditures on higher education and academic R&D expenditures by state governments. Although these selected variables may not exactly represent all available factors based on the two theories, the variables well capture the state government’s innovation capacity in major economic development programs on the basis of the existing literature and data availability. The 10 variables are collected from three major data sources—National Science Foundation, U.S. Bureau of Economic Analysis, and the U.S. Department of Education.

Each of these financial variables were divided by state population and deflated to constant 2005 dollars to control for inflation over time periods. Combining all 10 major state spending variables in economic development programs between 1972 and 2010, factor analysis was conducted to group a set of variables that are correlated with each other. The factor analysis procedure was conducted to create a state innovation capacity index in economic development programs.⁸

When all of the factor analysis procedures were completed, two common factors that represent two comparable theories—the supply side and demand side—were clearly extracted and grouped. It is important to note that grouping new variables was quite subjective. Based on the results from rotated factor matrix, shown in Table 1, the first common factor reduces the six original variables into one innovation capacity index related to more conventional economic development programs. All of the variables grouped in this common factor indicate state expenditures on infrastructure programs and labor productivity including natural resources, highway, air transit, public building, foreign commerce, and employment security. As shown, all of the variable loadings were moderately strong and regarded as state innovation capacity in conventional economy. The remaining four variables—state spending on R&D, higher education, primary and secondary education, and housing and community development—were grouped into the second common factor, which constituted state innovation capacity in more knowledge-based (new) economy. Two variables—primary and secondary education and housing and community development expenditures—loaded onto this factor are counterintuitive: primary and secondary education spending and housing and community development spending were expected to load on conventional economy as a human resource that may produce labor productivity and as a physical resource that may provide competitive incentives to industries and businesses, respectively. However, these results do not mean that the second common factor, state innovation capacity in the new economy, is not reliable. Indeed, many studies indicated that investment in primary and secondary education is a great potential source to increase economic competitiveness in the knowledge-based economies through increases in labor productivity, research, and innovation (e.g., Hanushek and Woessmann, 2012; Sahlberg, 2006). Literature in economic development also suggested that government programs to encourage housing and community development help regions not only foster innovation, but open new markets (Leigh and Blakely, 2017; Markusen and Glasmeier, 2008).

Table 1.

Factor analysis of state government capacity in economic development programs

	Component
Variables	Innovation capacity in old economy	Innovation capacity in new economy
Natural resources expenditures	0.802
Highway expenditures	0.751
Employment security expenditures	0.696
Air transit expenditures	0.659
Public building expenditures	0.588
Foreign commerce expenditures	0.482
Research and development expenditures		0.710
Higher education expenditures		0.652
Primary and secondary education expenditures		0.576
Housing and community development expenditures		0.421

Note: Extraction method: Principal axis factoring. Rotation method: Varimax with Kaiser normalization; rotation converged in three iterations.

Factor scores were then generated for each state and year in the dataset, by using the Anderson–Rubin method, which represented a 50-state innovation capacity index in old and new economic programs. The two factors are termed simply as “new innovation capacity,” and “old innovation capacity,” respectively. The results from rotated factor matrix are reported in Table 1.

The primary independent variable is the federal welfare transfer to the states in the form of federal grants, which varies by state. This federal welfare expenditure variable is measured as the per-capita federal outlays in public welfare programs in real term at 2005 constant dollars. The data are obtained from the U.S. Census Bureau’s State Government Finances data, which includes the aggregate amount spent for the support of and assistance to needy people contingent upon their need. While more than half of the federal spending consists of Social Security and Medicare, neither of these programs are included in the public welfare category. These programs were excluded because while such means-tested programs as Medicaid, food stamps, and other programs designed to assist the poor are generally considered redistributive, many experts in social policy would not accept the same label applied to social insurance, particularly Social Security and Medicare (e.g., Rudolph and Evans, 2005; Wilson, 1973). Furthermore, Social Security and Medicare are straight transfers that do not enhance the job skills or health of working-age adults.

All models include a set of control variables representing a state’s political, policy, and socioeconomic characteristics, which are commonly used among scholars who examine the determinants of state economic growth. It can be expected that state politicians make different economic development policies according to their ideological position. To measure the effect of state government ideology, a measure developed by Berry et al. (1998) is used. The ideology measure ranges from 0 to 100 with lower values indicating more conservative states. Both state-level governor’s (gubernatorial) power and legislative professionalism are considered as differences in the degree of institutional capacity that leads to variation in economic policy and performance across states. Governor’s power is drawn from the index of the governor’s institutional powers developed by Beyle (2003). The data of legislative professionalism were obtained from Squire (2007).

Several indicators of the public policies of state governments are controlled for. All models include several measures to capture state tax policies that may influence state economic policies and outcomes. The variables used in the analyses are Tax and Expenditure Limits (TELs) and supermajority requirements for raising taxes. It is expected that more stringent tax policies may limit state activity on economic development areas and decrease state economic performance. The data on the TELs are from Amiel et al. (2009). The supermajority requirements variable is the percentage of the legislature required to raise taxes or approve new taxes.⁹ The data are from the National Conference of State Legislatures. Another critical policy indicator to explain state economic performance is the level of state fiscal decentralization. The theoretical prediction of fiscal decentralization suggests that states with a higher degree of fiscal decentralization have higher fiscal capacity and thus, secure more own source revenues, which are, in turn, translated into economic development. This variable is measured by the share of state and local expenditures compared to total government spending.¹⁰

Lastly, the models include a set of state socioeconomic variables including population, poverty rate, labor force, national unemployment rate, and welfare spending. The population variable is the natural log of the total state population. It is expected that the population of a state is negatively related to state economic capacity and performance since population increases industrial land costs. The models also include poverty rate (the percentage of families below poverty income level), which is expected to be negatively related to state economic capacity and performance. The data are from the U.S. Census Bureau’s Population Estimates Program. Since the potentially strong labor productivity increases economic outcomes, the state’s labor force (the percentage of persons who are16 years old and over) is also included. The national unemployment rate (the number of unemployed people as a percentage of the labor force) is another important consideration not only because state economies usually mirror national economy, but because it controls for time-invariant effects across states. Both state labor force and national unemployment data are obtained from Bureau of Labor Statistics. Finally, the state’s own welfare spending is included to compare the effects of federal welfare funds. It is expected that state own welfare spending is negatively correlated with state economic outcomes since the increase of welfare spending at the state level would lead to a financial difficulty. The variable is measured as the per capita state own spending in public welfare programs in real term at 2005 constant dollars.

Model specification

Given the nature of panel data, there exists a set of critical problems including the presence of heteroskedasticity and the autoregressive nature of the dependent variable, which cannot be adequately dealt with by the traditional panel data estimation techniques (Beck and Katz, 1995). In order to test both the direct and indirect effects of federal welfare spending on state economic performance, ECMs are employed, estimated with ordinary least squares (OLS), and panel-corrected standard errors (PCSEs) with a state-specific fixed effects by applying first-order autocorrelation.¹¹ Diagnostic tests were conducted to see whether the employed panel data are stationary. For all variables measuring state economic performance and state innovation capacity, the results show that they violate the stationarity assumption. Thus, the ECM is best suited for dealing with the presence of unit roots and cointegration present in time series data, and allows for an intuitively useful way to explore temporal dynamics (Beck and Katz, 1995; De Boef and Keele, 2008).¹² This model deals with the fact that the direct measure of the economic growth and innovation capacity into the states are an integrated series.

The traditional method for dealing with nonstationarity is to estimate a regression with a first difference of the dependent variable. This allows the analyst to see the impact caused by the independent variables over one period. Sometimes this might be fine, but in the case of economic status, the notion that a change in a state’s welfare revenues or innovation capacity would only cause a change in economic growth/volatility in the following year is unrealistic. While an impact in state economic outcomes will be visible in the following year, the effect of welfare benefits will be more visible after a few periods because it requires a substantial time for people who receive welfare benefits to increase their skills and to find a job. There is also an argument that welfare spending is considered to be consumption and is associated with higher future unemployment in some decades (Jones, 1990; Peterson, 1995). Economists clearly suggest that economic growth is a process of the long run period of change, in which prices have time to adjust to return the economy to a full-employment equilibrium (DeLong and Olney, 2005). It is also possible that federal welfare funds produce an immediate impact on state innovation capacity. However, state-spending response to federal aid would take a little longer than expected since state expenditures on economic development programs is limited by some fiscally constraining institutions, especially TELs (McLendon et al., 2009). Thus, it is likely that the increase of state innovation capacity changes over a period of years through a partial adjustment process. Overall, it is reasonable to expect that the impact of federal welfare spending should be observed in the long run.

An ECM helps overcome this problem by observing both the immediate and long run effects by having a set of the first differenced term and the lagged term of the independent variables and a differenced dependent variable in the model. The short-run impact, captured by the differenced independent variable, is interpreted as in any other model. The estimation of the long-term effects begins with the assumption that a change in the independent variable disrupts the equilibrium between that series and the dependent variable, and that the dependent variable will eventually change to move back into equilibrium with the new level of the independent variable. The lagged level of the independent variable captures the size of that change over the entire time period in the sample. Finally, the ECM also allows us to see, through the lagged level of the dependent variable, the speed with which the dependent variable series “corrects,” or comes back into equilibrium with the independent variable. ECM, therefore, is specified as follows¹³

\begin{matrix} {Δ STATEINNOV}_{st} = β_{0} + β_{1} {STATEINNOV}_{st - 1} + β_{2} {Δ FEDWELEXP}_{st} \\ + β_{3} {FEDWELEXP}_{st - 1} + β_{k} {Δ X}_{st} + β_{k} X_{st - 1} + ɛ_{st} \end{matrix}

(1)

\begin{matrix} Δ {STATEECON}_{st} = β_{0} + β_{1} {STATEECON}_{st - 1} + β_{2} Δ {FEDWELEXP}_{st} \\ + β_{3} {FEDWELEXP}_{st - 1} + β_{4} Δ {STATEINNOV}_{st} \\ + β_{5} {STATEINNOV}_{st - 1} + β_{k} Δ X_{st} + β_{k} X_{st - 1} + ɛ_{st} \end{matrix}

(2)

where subscript s denotes states, t denotes time, and ɛ denotes a stochastic error term. Specifically, Δ denotes the first difference operator. Each equation is modeled by operating that the change in the dependent variable from year_t−1 to year t is a function of the change in the level of the major independent variable from year_t−1 to year t and its lagged value with the lagged dependent variable.

Δ X_{st}

denotes the set of control variables of state s from year_t−1 to year t, and the

X_{st - 1}

represents those of state s at time t – 1. Two models include state dummies because there are certainly unobserved factors that vary across states and affect state economic performance, but are not included as variables in the models.

The first model (1) estimates the indirect effect of federal welfare spending on state economic performance through the regression including one intervening variable—state innovation capacity in economic development programs. The dependent variable, ${Δ STATEINNOV}_{st}$ in the model (1) represents state government’s efforts to foster economic development, which denotes the first differenced level of state innovation capacity in economic development programs. The second model (2) estimates the direct effect of federal welfare spending on state economic growth/volatility by including the intervening variable—state innovation capacity in economy development areas—as the predictor in the model. $Δ {STATEECON}_{st}$ is the first difference of the state economic growth/volatility variables in state s at time t.¹⁴

Main results

Table 2 shows the results from the estimation of the ECMs that investigate the relationship between total federal welfare expenditures and state innovation capacity in economic development programs specified by new economy capacity and old economy capacity. The relationship is expected to positively affect state economic growth and to negatively affect state economic volatility. The key independent variables are the measure of a one-time change and the lagged level of per capita total federal welfare spending. Following a casual interpretation of the ECM, the coefficient on the lagged terms means the long-term relationship between the dependent and independent variables, while the differenced terms reflect the short-term adjustment. In order to look at some sense of the robustness of the results to alternative specifications, in models 1 and 3, only the differenced and lagged value of federal welfare expenditures are included, without the plethora of control variables. As shown in the results, the coefficient estimates of federal welfare spending on innovative economy capacity are statistically significant at the 0.05 level in the long-term effect. These simple models support the first hypothesis that the federal welfare spending motivates states to spend more on economic development programs, but it only works in a way that increases state innovation capacity in progressive economic development programs, not in conventional programs. That is, the effect of federal welfare spending on state government’s efforts to increase new innovation capacity is distributed across some future time periods.

Table 2.

The effect of federal welfare expenditures on the intervening variables

	State innovation capacity in economic development
	New economy		Old economy
Variables	(1)	(2)	(3)	(4)
Lag of dependent variable	−0.039***	−0.068***	−0.017*	−0.029**
	(0.014)	(0.017)	(0.014)	(0.018)
Δ Federal welfare expenditures	0.0002	0.0002**	0.00003	−0.0001
	(0.0001)	(0.0001)	(0.0001)	(0.0001)
Federal welfare expenditures_t−1	0.0001**	0.0002***	0.00001	0.00004
	(0.00004)	(0.00005)	(0.00002)	(0.00003)
Δ Government ideology		−0.0002		−0.0001
		(0.001)		(0.0004)
Government ideology_t−1		0.0002		−0.0002
		(0.0003)		(0.0002)
Δ Gubernatorial power		−0.041		0.008
		(0.032)		(0.027)
Gubernatorial power_t−1		0.008		0.013*
		(0.009)		(0.008)
Δ Legislative professionalism		−0.053		0.079
		(0.237)		(0.178)
Legislative professionalism_t−1		0.106**		0.024
		(0.055)		(0.048)
Δ Tax and expenditure limitation		−0.0003		−0.003
		(0.003)		(0.002)
Tax and expenditure limitation_t−1		−0.0001		−0.00003
		(0.0009)		(0.0006)
Δ Supermajority requirement		−0.418*		0.115
		(0.322)		(0.290)
Supermajority requirement_t−1		−0.031		0.066
		(0.155)		(0.087)
Δ Fiscal decentralization		1.213*		−0.297
		(0.826)		(0.675)
Fiscal decentralization_t−1		0.557*		0.116
		(0.354)		(0.288)
Δ Population (ln)		1.045*		−0.825
		(0.756)		(0.653)
Population (ln)_t−1		−0.037*** (0.013)		−0.031*** (0.010)
Δ Poverty rate		0.003		−0.0009
		(0.004)		(0.003)
Poverty rate_t−1		−0.0001		0.001
		(0.002)		(0.002)
Δ Labor force		1.39e-07*		3.82e-08
		(9.33e-08)		(6.58e-08)
Labor force_t−1		3.61e-09		4.40e-09*
		(4.04e-09)		(2.79e-09)
Δ State welfare expenditures		−0.0003		0.0004*
		(0.0003)		(0.0003)
State welfare expenditures_t−1		−0.0003***		−9.64e-07
		(0.0001)		(0.0001)
Δ National unemployment rate		0.010		0.006
		(0.009)		(0.007)
National unemployment rate_t−1		−0.008		−0.003
		(0.007)		(0.005)
Constant	0.016	0.069	−0.025**	0.281
	(0.022)	(0.310)	(0.012)	(0.297)

R ²	0.010	0.027	0.016	0.047
Observations	1900	1900	1900	1900

Note: Error correction models are estimated with Prais–Winsten correction for panel-specific autocorrelation; panel-corrected standard errors shown in parentheses.

p < 0.10. **p < 0.05. ***p < 0.01; one-tailed tests.

These results are statistically robust even in the models that include a set of control variables which would be critical to affect the state economic development efforts. The results from Columns 2 and 4, in Table 2, indicate that states with larger welfare aid from the federal government exercise a greater new innovation capacity, not old innovation capacity, in the short and long terms. Specifically, the lagged level of federal welfare expenditures, as presented in Columns 2, shows a positive and significant long-term relationship with a state’s economic development efforts. The actual magnitude, the so-called “long-run multiplier,” implies that a one standard deviation increase in federal welfare spending within a state leads to a 0.77 standard deviation increase in predicted a state’s new innovation capacity particularly in the long term.¹⁵ These results indicate that federal welfare aid has a much larger magnitude for stimulating states’ new innovation capacity than states’ old innovation capacity. Furthermore, the coefficients of the lagged dependent variables for the lagged measures of state new innovation capacity fall into the small absolute value between –1 and 0, suggesting that the increase in state innovation capacity occurs slowly over future time periods at a rate dictated by the error correction coefficients, β = −0.07 (±0.02). Column 2 also presents that the coefficient of the differenced term of federal welfare spending is positive and significant, suggesting that a one standard deviation increase in federal welfare spending leads to a 0.08 standard deviation immediate increase for the state new innovation capacity. Overall, the results indicate that federal welfare funds are the significant determinant that increases the state government’s innovation ability especially on the knowledge-based economic programs.

Of the statistically significant control variables, the long-term population variable is negatively correlated with the two intervening variables, suggesting that a state with a higher population is less likely to be an active participant in economic development efforts. In general, a higher population increases industrial land costs and discourages new business, which in turn reduces a state’s financial resource. The measure of state legislative professionalism is a positive and significant variable in the long term, indicating that states with higher level of legislative professionalism achieve higher new innovation capacity. The coefficient of the supermajority requirement variable is negative and statistically significant in the short run, suggesting that states with more stringent tax policy decrease new state innovation capacity by making it harder to increase revenues. As expected, fiscal decentralization is positively and significantly associated with state innovation capacity in a new economy in both the short and long terms. Labor force is particularly significant in an old economy, which supports the idea of the old supply-side theory that the growth in a traditional economy is better explained by competitive incentives on capital, land, and labor costs. Contrary to the effect of federal welfare spending, the estimated long-term coefficient for the state-level welfare spending is negatively and significantly correlated with the state innovation capacity variable, particularly in new economic programs. This result provides partial evidence that an increase in welfare spending at the state level would lead to financial difficulty and thus would not benefit state innovation capacity.

Table 3 reports results from the estimation of the ECMs that focus on the direct relationship between federal welfare spending and state economic growth. The dependent variables include two economic growth indicators: the measure of change in per capita GSP and change in per capita personal income. Looking briefly at the results from the ECM models that do not include control variables (Columns 1 and 3), it turns out that a higher federal welfare spending directly increases both GSP and income, once the persistence of the relationship is accounted for. Focusing on the relationship between the intervening variables and the state economic growth variables, while state new innovation capacity is significantly correlated with GSP in both short- and long-term changes, it is statistically significant only in the long term.

Table 3.

The effect of federal welfare spending on state economic growth

Variables	(1) GSP	(2) GSP	(3) Income	(4) Income
Lag of dependent variable	−0.137***	−0.155***	−0.072***	−0.072***
	(0.042)	(0.043)	(0.026)	(0.020)
Δ Federal welfare expenditures	0.032	0.829	−1.112*	0.129
	(1.215)	(1.214)	(0.743)	(0.502)
Federal welfare expenditures_t−1	1.912**	1.496**	0.867**	0.639**
	(0.833)	(0.619)	(0.399)	(0.268)
Δ New innovation capacity	682.828*	860.037**	247.245*	391.807***
	(483.123)	(456.570)	(172.379)	(141.681)
New innovation capacity_t−1	441.730*	304.867*	209.705**	134.950*
	(297.018)	(241.575)	(125.568)	(86.565)
Δ Old innovation capacity	1906.741***	2133.648***	388.659**	518.316***
	(662.078)	(639.763)	(193.255)	(167.589)
Old innovation capacity_t−1	779.684**	869.246**	186.199	102.010
	(459.541)	(420.850)	(185.397)	(125.160)
Δ Government ideology		3.621		1.952
		(3.063)		(2.115)
Government ideology_t−1		−1.404		−1.896*
		(2.417)		(1.379)
Δ Gubernatorial power		−488.563**		−29.679
		(217.199)		(148.695)
Gubernatorial power_t−1		−347.713**		−134.106*
		(154.203)		(88.471)
Δ Legislative professionalism		−178.492		−422.360
		(1540.197)		(1188.359)
Legislative professionalism_t−1		−1183.795		−1026.032**
		(794.832)		(505.626)
Δ Tax and Expenditure Limitation		−24.058		−0.761
		(19.685)		(8.435)
Tax and Expenditure Limitation_t−1		−31.247***		−1.750
		(10.091)		(4.095)
Δ Supermajority requirement		−1157.224		−478.289
		(2060.239)		(1378.719)
Supermajority requirement_t−1		2538.527**		430.989
		(1089.923)		(554.428)
Δ Fiscal decentralization		−6066.189		−3393.691
		(6133.553)		(3571.499)
Fiscal decentralization_t−1		−3245.194		−1753.798
		(4652.694)		(2722.696)
Δ Population (ln)		13641.6		11798.59***
		(14593.2)		(4603.798)
Population (ln)_t−1		1156.348*		124.593
		(597.347)		(249.439)
Δ Poverty rate		−84.079***		−25.019**
		(23.056)		(14.773)
Poverty rate _t _– ₁		−52.998**		−16.456
		(27.416)		(16.665)
Δ Labor force		0.002**		0.001***
		(0.001)		(0.0005)
Labor force _t _– ₁		0.0002***		0.0001***
		(0.0001)		(0.00003)
Δ State welfare expenditures		−1.980		−1.238
		(3.330)		(1.146)
State welfare expenditures _t _– ₁		−7.270***		−2.143***
		(2.523)		(0.660)
Δ National unemployment rate		−504.976***		−418.133***
		(64.341)		(56.795)
National unemployment rate _t _– ₁		8.173		−46.312
		(58.606)		(46.800)
Constant	3548.003***	−9765.182	1794.54***	2308.749
	(1008.06)	(8654.916)	(577.314)	(3576.729)

R ²	0.104	0.241	0.061	0.359
Observations	1900	1900	1900	1900

GSP: gross state product.

Note: Error correction models are estimated with Prais–Winsten correction for panel-specific autocorrelation; panel-corrected standard errors shown in parentheses.

p < 0.10. **p < 0.05. ***p < 0.01; one-tailed tests.

As shown in Columns 2 and 4, the impact of federal welfare funds on state economic growth remains the same even in the ECM models that include a set of control variables. The lagged measure of aggregate federal welfare spending, which indicates the long-run relationship to GSP, is significant (at the 0.05 level) and positive. This suggests that GSP per capita will continue to change a total of $9.65 following a one dollar change in federal welfare spending. Based on the error correction rate (–0.16), GSP per capita will change $1.49 after one year, another $0.23 after two years, $0.04 after three years, and so on until the two series come back to equilibrium. The positive long-run effect of federal welfare expenditures, as shown in Column 4, continues to show significant correlation with personal income, suggesting that the total effect of federal welfare spending on per capita income, as the long-run multiplier of $8.88, is distributed over future time periods.

In addition, Columns 2 and 4 consider the potential relationship between the intervening variables and state economic growth. As expected, the state government’s innovation capacities in both new and old economic development programs are positively and significantly correlated with state economic growth. Specifically, a state’s new innovation capacity has positive and significant short- and long-term effects on changes in both per capita GSP and per capita income, while its old innovation economy capacity has a positive and significant short- and long-term impact only on change in per capita GSP.

Turning to other alternative measures of state economic performance, Table 4 in Appendix 1 shows the results from two models that include state economic volatility as the dependent variable, which is computed as the four-year moving standard deviation in GSP growth rates and personal income growth rates, respectively. First, the results from both the ECMs without control variables and with control variables are not significantly different, although in the reduced model (3) that measures state income growth volatility, the coefficient of the long-term federal welfare expenditures variable is not statistically significant.

Focusing on interpreting the statistical results from the ECMs including a set of control variables (Columns 2 and 4), federal welfare spending per capita has a negative and significant short- and long-term effects on both GSP and income growth volatility, suggesting that a $1 increase in aggregate federal welfare spending per capita will result in a 0.002 immediate decrease in the four-year moving standard deviation in GSP growth rate and a 0.001 immediate decrease in the four-year moving standard deviation in personal income growth rate, respectively. The results also indicate that an increase in federal welfare expenditures disturbs the long-run equilibrium of state economic volatility, suggesting that a $1 increase in per capita total federal welfare spending decreases state economic volatility, calculated by GSP growth rate, by a total 0.003 four-year moving standard deviation over future time periods. This impact is almost identical to the model that includes economic volatility measured by personal income growth rate. In these models, the long-term effects of federal welfare funds are felt at a rate of roughly around 40% each year, so it takes relatively long to be fully observed. Overall, the negative impact of federal welfare expenditures on state economic volatility is quite substantial in both the short and long term.

When the intervening variables are considered, the coefficients for the short- and long-term new innovation capacity variables are negative, but only its short-term impact is statistically significant. Contrary to the expectation, however, the long-term effects of old innovation capacity are positive and statistically significant in the models of both GSP and income growth volatility, meaning that states with a higher innovation capacity in traditional economic development programs are associated with higher economic volatility. This result, however, is understandable given the limitations of traditional economic development programs. Indeed, the micro-level studies in economics have identified that the financial and social outcomes of infrastructure investment are strikingly inefficient, and thus the increase of infrastructure spending would disrupt economic stability in the long run (Ansar et al., 2014; Flyvbjerg et al., 2009).

In addition, in order to show the total effect of federal welfare expenditures on state economic performance over time, presented here is the change of the median lag length in state economic growth and volatility from a $1 change in federal welfare spending during different periods of time. Figure 2 shows that the effect of federal welfare funds in the long run followed by the substantial changes in per capita GSP for the current year and the following year. The effect of federal welfare spending on economic growth based on per capita income is not immediate; its effect is mostly felt the following year, and then marginally decayed over four lags. These results suggest the positive effect of federal welfare spending on state economic growth appears immediately but continues into the future. When looking at the over-time effects of federal welfare spending on state economic volatility (Figure 3 in Appendix 1), the effect of federal welfare outlays on GSP growth volatility occurs quite immediately (98% of the effect at time t). Similarly, the data shows that federal welfare expenditures on income growth volatility have the largest shift during the first year. As a result, the effect of federal welfare spending on state economic volatility is accentuated over short lags: federal welfare spending takes shorter to adjust back to equilibrium in the volatility measures than in the growth measures.

Figure 2.

Estimated lag distributions for a change in the long-term direct impact of federal welfare spending on state economic growth.

Conclusion

This paper has examined how federal welfare spending influences state economic performance. The results show that federal welfare expenditures allocated to states do affect state economic performance directly and indirectly. While increased federal spending on welfare programs stimulates state economic performance directly, it influences state economic performance indirectly through the positive association with state economic development efforts. Federal aids on welfare programs directly contribute to better state economic performance because larger welfare spending reinforces the multiplier effects of redistribution that would stimulate further economic benefits such as business activity, jobs, and income. The central implication of the indirect effect of federal welfare assistance is that federal financial assistance on redistributive programs will produce larger regional economic performance by motivating states to generate progressive economic development programs.

Specifically, the empirical results suggest that federal welfare spending has a significant upward impact on state economic growth measured by per capita GSP and per capita personal income, while it has a significant downward impact on state economic volatility measured by the four-year moving standard deviation of both GSP and personal income growth rates. These direct effects of federal welfare spending are particularly noteworthy in the long term. For the indirect effects of federal welfare spending on state economic performance, the results indicate that a larger federal welfare spending policy indirectly increases state economic growth throughout by enhancing state innovation capacity especially in the knowledge-based economic program. This indirect impact is also robust in the long run. Taking these findings into account, this research suggests two distinct but interrelated conclusions.

First, this paper suggests that a focus on functional federalism could contribute to our understanding of economic growth in the American states. Although the theory of functional federalism does not suggest any causal linkage between a larger role of the federal government on welfare programs and the state government’s straightforward effort to enhance economic growth, the theory suggests potentially that the former can further the latter due to the institutional efficiency of American federalism based on the doctrine of cooperative federalism. This paper has made a large effort to identify this causal relationship, and thereby extending the existing functional theory by emphasizing that the comparative advantage of each level of government can benefit regional economic growth and stability. Utilizing evidence-based analysis to examine the state government’s economic performance, this research identifies the institutional efficiency of the functional federalism in that the federal government’s investment on welfare areas increases state economic growth and stability. It also suggests that the functional mechanism of American federalism successfully works in a way that a higher level of government takes a substantial responsibility over policy areas that lower levels of governments are not fully equipped to manage effectively. That is, a powerful intergovernmental externality (i.e., the increased federal welfare funds to states) has increased state economic performance through reducing the welfare burden of state governments while securing a greater policy innovation on economic development.

Second, this analysis suggests that means-tested welfare programs of the federal government have a direct positive impact on state economic growth because the federal government has developed welfare programs that make a positive contribution to economic growth. This is particularly true when we see the changing federal goal on welfare policy, which supports the poor conditionally to return them into a labor market. Midgley (1999) claims that welfare policies for economic development purpose have garnered a larger support from liberals, conservatives, and the general public regardless of its view on welfare issues. As a result, the federal government has increased strategic responsibility at least for welfare policies, but its goal has been developmental in intent so as to gain political acceptance among the federal and state governments (Rom, 1989). The findings from this research are straightforward: the direct effect of federal welfare spending is substantial and continue to be an important intergovernmental policy externality that would sustain long-term state economic development and growth.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2018S1A3A2075531).

Notes

Appendix 1

References

Aghion

Caroli

Garcia-Penalosa

(1999) Inequality and economic growth: The perspective of the new growth theories. Journal of Economic Literature 37(4): 1615–1660.

Ahmad

Dreze

Hills

et al. (1991) Social Security in Developing Countries, Oxford, UK: Clarendon Press.

Akai

Sakata

(2002) Fiscal decentralization contributes to economic growth: Evidence from state-level cross-section data for the United States. Journal of Urban Economics 52(1): 93–108.

Amiel L, Deller S and Stallmann J (2009) The construction of a tax and expenditure limitation index for the US. Department of Agricultural and Applied Economics. Staff Paper No. 536. University of Wisconsin-Madison, Agricultural and Applied Economics, pp. 1–32.

Ansar

Flyvbjerg

Alexander

et al. (2014) Should we build more large dams? The actual costs of hydropower megaproject development. Energy Policy 69(June): 43–56.

Anton

(1989) American Federalism and Public Policy: How the System Works, New York: Random House.

Bae

Moon

Jung

(2012) Economic effects of state-level tax and expenditure limitations. Public Administration Review 72(5): 649–658.

Bahl

Martinez-Vazquez

Wallace

(2002) State and local government choices in fiscal redistribution. National Tax Journal 55(4): 723–742.

Baron

Kenny

(1986) The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology 51(6): 1173–1182.

10.

Bartik TJ (1999) Federal policy toward state and local economic development in the 1990s. In: Norton RD (ed) The Millennial City: Classic Readings on U.S. Urban Policy. Stamford, CT: JAI Press, pp. 235–251.

11.

Beck

Katz

(1995) What to do (and not to do) with time-series cross-section data. American Political Science Review 89(3): 634–647.

12.

Berry

(1990) State lottery adoptions as policy innovations: An event history analysis. American Political Science Review 84(2): 395–415.

13.

Berry

Ringquist

Fording

et al. (1998) Measuring citizen and government ideology in the American states, 1960–1993. American Journal of Political Science 42(1): 327–348.

14.

Beyle

(2003) The governors. In: Gray

Hanson

(eds) Politics in the American States: A Comparative Analysis, 8th ed. Washington, DC: CQ Press, pp. 194–231.

15.

Bohte

Meier

(2000) The marble cake: Introducing federalism to the government growth equation. Publius: The Journal of Federalism 30(3): 35–46.

16.

Boix

(1997) Political parties and the supply side of the economy: The provision of physical and human capital in advanced economies, 1960–90. American Journal of Political Science 41(3): 814–845.

17.

Bovbjerg RR, Wiener JM and Houseman M (2003). State and federal roles in health care: Rationales for allocating responsibilities. In: Holahan J, Weil A and Weiner JM (eds) Federalism and Health Policy. Washington, DC: Urban Institute Press, pp. 25–58.

18.

Brace

(1993) State Government and Economic Performance, Baltimore: The Johns Hopkins University Press.

19.

Brace

Mucciaroni

(1990) The American states and the shifting locus of positive economic intervention. Policy Studies Review 10(1): 151–173.

20.

Brealey

Myers

Allen

(2008) Principles of Corporate Finance, Boston: McGraw-Hill/Irwin.

21.

Brulhart

Schmidheiny

(2015) Estimating the rivalness of state-level inward FDI. Journal of Regional Science 55(1): 139–148.

22.

Callen Z (2009) The seams of the state: Infrastructure and intergovernmental relations in American state building. PhD Thesis. University of Chicago, Chicago, IL.

23.

Charemza

Deadman

(1992) New Directions in Econometric Practice: General to Specific Modeling, Cointegration, and Vector Autoregression, Northampton, MA: Edward Elgar.

24.

Congressional Research Service (2012) CRS report: Welfare spending the largest item in the federal budget. Available at: http://www.theccwr.org/pdfs/D_CRS-Welfare-Spending-Report.pdf (assessed 25 November 2017).

25.

Conley A (2010) Social development, social investment, and child welfare. In: Midgley J and Conley A (eds) Social Work and Social Development: Theories and Skills for Developmental Social Work. New York, NY: Oxford University Press, pp. 31–54.

26.

Crain

(2003) Volatile States: Institutions, Policy, and Performance of American State Economies, Ann Arbor: University of Michigan Press.

27.

Crain

Lee

(1999) Economic growth regressions for the American states: A sensitivity analysis. Economic Inquiry 37(2): 242–257.

28.

De Boef

Luke

(2008) Taking time seriously. American Journal of Political Science 52(1): 184–200.

29.

DeLong

Olney

(2005) Macroeconomics, New York: McGraw-Hill.

30.

Deskins

Hill

Ullrich

(2010) Education spending and state economic growth: Are all dollars created equal? Economic Development Quarterly 24(1): 45–59.

31.

Eisinger

(1989) The Rise of the Entrepreneurial State, Madison: University of Wisconsin Press.

32.

Feiock

(1999) Development policy competition and positive sum growth: Incentive competition and its alternatives. International Journal of Economic Development 1(3): 238–255.

33.

Feldman

Francis

(2004) Homegrown solutions: Fostering cluster formation. Economic Development Quarterly 18(2): 127–137.

34.

Fisher

(2015) The crucible of debt and welfare spending: Evidence from state-level welfare expenditures in the United States. Review of Political Economy 27(2): 201–217.

35.

Flyvbjerg

Garbuio

Lovallo

(2009) Delusion and deception in large infrastructure project. California Management Review 51(2): 170–193.

36.

Gordon

Cullen

(2012) Income redistribution in a federal system of governments. Journal of Public Economics 96: 1100–1109.

37.

Greenberg

Deitch

Hamilton

(2009) Welfare-to-Work Program Benefits and Costs: A Synthesis of Research, New York: MDRC.

38.

Hall

(2007) Developing historical 50-state indices of innovation capacity and commercialization capacity. Economic Development Quarterly 21(2): 107–123.

39.

Hall

Howell-Moroney

(2012) Poverty, innovation capacity, and state economic development in the knowledge economy: Evidence from the U.S. Growth and Change 43(2): 228–251.

40.

Halvorsen

Jakobsen

(2013) Democrats, republicans—Or both? An empirical analysis of the effects of the composition of state governments on FDI, 1977–2004. International Interactions 39(2): 167–191.

41.

Hansen

Imrohoroglu

(1992) The role of unemployment insurance in an economy with liquidity constraints and moral hazard. Journal of Political Economy 100(1): 118–142.

42.

Hanushek

Woessmann

(2012) Do better schools lead to more growth? Cognitive skills economic outcomes, and causation. Journal of Economic Growth 17(4): 267–321.

43.

Harwood J (2011, November 28) The caucus: ‘Spreading the wealth’ in Democrats’ favor. The New York Times. Available at: http://query.nytimes.com/gst/fullpage.html?res=9405E5DF1230F93BA15752C1A9679D8B63 (accessed 30 November 2017).

44.

Howell-Moroney

(2008) A mixed method look at state growth management policy. American Review of Public Administration 38(3): 339–361.

45.

Johnson

(1988) Income redistribution in a federal system. American Economic Review 78(3): 570–573.

46.

Jones

(1990) Public policies and economic growth in the American states. Journal of Politics 52(1): 219–233.

47.

Knapp J and Midgley J (2010) Developmental social work and people with disabilities. In: Midgley J and Conley A (eds) Social Work and Social Development: Theories and Skills for Developmental Social Work. New York, NY: Oxford University Press pp. 87–104.

48.

Kormendi

Meguire

(1985) Macroeconomic determinants of growth: Cross-country evidence. Journal of Monetary Economics 16(2): 141–163.

49.

Kwon

Berry

Feiock

(2009) Understanding the adoption and timing of economic development strategies in U.S. cities using innovation and institutional analysis. Journal of Public Administration Research and Theory 19(4): 967–988.

50.

Labonte M (2009) The size and role of government: Economic issues. CRS Report for Congress. RL 32162. Available at: https://digitalcommons.ilr.cornell.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsredir=1&article=1640&context=key_workplace.

51.

Leigh

Blakely

(2017) Planning local economic development, Thousand Oaks, CA: Sage Publication.

52.

Long

Mallar

Thornton

(1981) Evaluating the benefits and costs of the job corps. Journal of Policy Analysis and Management 1(1): 55–76.

53.

Lurie

(2001) State implementation of TANF: Where do we stand in 2001? Evaluation and Program Planning 24: 379–388.

54.

Mahroum

Al-Saleh

(2013) Demand-led related diversification: An innovation policy approach to economic diversification and development. Science and Public Policy 40(3): 406–418.

55.

Markusen

Glasmeier

(2008) Overhauling and revitalizing federal economic development programs. Economic Development Quarterly 22(2): 83–91.

56.

Marr

Huang

(2014) Earned Income Tax Credit Promotes Work, Encourages Children’s Success at School, Research Finds, Washington, DC: Center on Budget and Policy Priorities.

57.

McCallum

Blais

(1987) Government, special interest groups and economic growth. Public Choice 54(1): 3–18.

58.

McLendon

Hearn

Mokher

(2009) Partisans, professionals, and power. Journal of Higher Education 80(6): 686–713.

59.

Midgley

(1999) Social Development: The Developmental Perspective in Social Welfare, Thousand Oaks, CA: SAGE Publications.

60.

Miron

(2011) Rethinking redistribution. National Affairs 9(Winter): 70–85.

61.

Mohr

(1969) Determinants of innovation in organizations. American Political Science Review 63(1): 111–126.

62.

Montgomery JM, Nyhan B and Torres M (2018) How conditioning on posttreatment variables can ruin your experiment and what to do about it. American Journal of Political Science 62(3): 760–775.

63.

Murphy

Topel

(2016) Human capital investment, inequality, and economic growth. Journal of Labor Economics 34(2): S99–S127.

64.

Oates

(1972) Fiscal Federalism, New York: Harcourt Brace Jovanovich.

65.

Oates

(1985) Searching for leviathan: An empirical study. American Economic Review 75(4): 748–757.

66.

Palmer

(1987) The next decade: The economic, political, and social context of employment and training policies. Review of Policy Research 6(4): 685–694.

67.

Pauly

(1973) Income redistribution as a local public good. Journal of Public Economics 2(1): 35–58.

68.

Peterson

(1995) The Price of Federalism, Chicago: University of Chicago Press.

69.

Peterson

Rom

(1990) Welfare Magnets: A New Case for a National Welfare Standard, Washington: The Brookings Institution.

70.

Poulson

Kaplan

(2008) State income taxes and economic growth. Cato Journal 28(1): 53–71.

71.

Prillaman

Meier

(2014) Taxes, incentives, and economic growth: Assessing the impact of pro-business taxes on U.S. state economies. Journal of Politics 76(2): 364–379.

72.

Ramey

(1995) Cross-country evidence on the link between volatility and growth. American Economic Review 85(5): 1138–1151.

73.

Rector R and Sheffield R (2014) The War on Poverty After 50 Years. Backgrounder. Report no. 2955, 15 September. Washington, DC: The Heritage Foundation.

74.

Restuccia A (2017, October 2) Trump administration weighing executive order on welfare. Politico. Available at: https://www.politico.com/story/2017/10/02/trump-white-house-welfare-low-income-executive-order-243376 (accessed 30 November 2017).

75.

Rigby

Wright

(2013) Political parties and representation of the poor in the American states. American Journal of Political Science 57(3): 552–565.

76.

Rom

(1989) The family support act of 1988: Federalism, developmental policy, and welfare reform. Publius 19(3): 57–73.

77.

Rudolph

Evans

(2005) Political trust, ideology, and public support for government spending. American Journal of Political Science 49(3): 660–671.

78.

Sahlberg

(2006) Education reform for raising economic competitiveness. Journal of Educational Change 7(4): 259–287.

79.

Saiz

(2001) Politics and economic development: Why governments adopt different strategies to induce economic growth. Policy Studies Journal 29(2): 203–214.

80.

Schott L, Pavetti L and Floyd I (2015) How states use federal and state funds under the TANF block grant. Center on Budget and Policy Priorities. Available at: https://www.cbpp.org/research/family-income-support/how-states-use-federal-and-state-funds-under-the-tanf-block-grant (assessed 30 November 2017).

81.

Shin G (2018) Corporate tax policy and multinational corporations in the American states: Exploring the intervening effect of local fiscal decentralization. Journal of Urban Affairs. 40(5): 679–704.

82.

Shin G and Hall JL (2018) Exploring the influence of federal welfare expenditures on state-level new economy development performance: Drawing from the diffusion of innovation theory. Economic Development Quarterly 32(3): 242–256.

83.

Squire

(2007) Measuring state legislative professionalism: The squire index revisited. State Politics & Policy Quarterly 7(2): 211–227.

84.

Stansel

(2005) Local decentralization and local economic growth: A cross-sectional examination of U.S. metropolitan areas. Journal of Urban Economics 57(1): 55–72.

85.

Temple J (1999) The new growth evidence. Journal of Economic Literature 37(1): 112–156.

86.

U.S. Chamber of Commerce Foundation (2015) About states innovate. Available at: https://www.uschamberfoundation.org/enterprisingstates/assets/files/Executive-Summary-OL.pdf (accessed 30 November 2017).

87.

Walker

(1969) The diffusion of innovations among the American States. American Political Science Review 63(3): 880–899.

88.

West D, Friedman A and Valdivia W (2012) Building an innovation-based economy. Governance Studies at Brookings. Washington, DC. Available at: https://www.brookings.edu/wp-content/uploads/2016/06/InnovationBased-Economy.pdf (accessed 30 November 2017).

89.

Wilson

(1973) Political Organization, New York: Basic Books.

90.

Witte

(1985) The politics and development of the federal income tax, Madison: University of Wisconsin Press.

91.

Xie

Zou

Davoodi

(1999) Fiscal decentralization and economic growth in the United States. Journal of Urban Economics 45(2): 228–239.