To what extent does state intervention in the market condition how individuals subjectively experience the lives that they lead? Prevailing attempts to understand the relationship between state intervention and subjective well-being have yielded mixed empirical results. However, these differences result from omitted variable biases, not different methodological choices. Drawing on insights from the new social risk and quality of governance literatures, this article contends that the policy orientation and administrative quality of welfare state programs jointly condition the effect of state intervention on life satisfaction. State intervention exerts a strong positive effect on perceived satisfaction with life when the quality of administrative institutions is high and policy interventions focus on insuring individuals against newer, post-industrial forms of market risk. This main hypothesis is tested and confirmed against an empirical analysis of survey data taken from Wave 5 of the World Values Survey.
2011. “State Intervention and Subjective Well-Being in Advanced Industrial Democracies.” Politics & Policy 39 (2): 251-269. http://onlinelibrary.wiley.com/doi/10.1111/j.1747-1346.2011.00290.x/abstract, , and .
2010. “Self-Employed without Employees: Managing Risks in Modern Capitalism.” Politics & Policy 38 (4): 765-788. http://onlinelibrary.wiley.com/doi/10.1111/j.1747-1346.2010.00257.x/abstract.
2011. “Why People (Dis)like the Public Service: Citizen Perception of the Public Service and the NPM Doctrine.” Politics & Policy 39 (6): 997-1019. http://onlinelibrary.wiley.com/doi/10.1111/j.1747-1346.2011.00329.x/abstract.
Film Clips: Commission on the Measurement of Economic Performance and Social Progress. 2009. “Videos of Morning Sessions.” http://www.stiglitz-sen-fitoussi.fr/en/index.htm
2013. “World Database of Happiness.” http://www1.eur.nl/fsw/happiness/.
¿En qué medida condiciona la intervención estatal en el mercado la forma en la cuál los individuos perciben subjetivamente su vida? Intentos predominantes de entender la relación entre intervención del estado y bienestar subjetivo han producido resultados empíricos mixtos. Argumentamos aquí que tales diferencias resultan de sesgos omitidos de variables, no de diferentes elecciones metodológicas. Recurriendo a desarrollos de la literatura reciente sobre riesgo social y calidad de la gobernanza, este artículo plantea que la orientación de las políticas y la calidad administrativa de los programas de bienestar condicionan de manera conjunta el efecto el efecto de la intervención estatal en la percepción individual de satisfacción de vida. La intervención del estado ejerce un fuerte efecto positivo sobre la percepción de satisfacción de vida cuando la calidad de las instituciones administrativas es alta y las intervenciones de políticas se enfocan en asegurar a los individuos contra nuevas formas de riesgo posindustrial. Esta hipótesis central es probada y verificada en un análisis empírico de datos tomados del estudio Wave 5 of the World Values Survey.
The reconciliation of market forces and social pressures for equality, redistribution, and social solidarity lies at the heart of modern democratic politics. Understanding how and why societies adjudicate between these different, sometimes conflicting, imperatives has a pedigree as old and as colorful as the discipline of political science itself. Equally as important, our empirical inquiries are inextricably linked to fierce normative debates concerning the proper relationship between the state and the market. The “politics vs. markets” debate (Lindblom 1977) is central to beliefs concerning how best to achieve economic growth and prosperity, to promote social equality, to forge and strengthen the bonds of democratic governance, and to successfully compete in a global economy. Recent advances in the quality and availability of data on measures of subjective well-being 1 have enabled us to revisit this debate in novel ways by asking whether and how state intervention into the market influences the extent to which individuals find their lives to be satisfying, rewarding, and enjoyable.
Unfortunately, recent investigations provide few definitive answers to this important research question. 2 Proponents of state intervention contend that greater subordination of market processes to political control improves life satisfaction by insuring individuals against market insecurities, emancipating individuals from their captive dependence on market forces to achieve and maintain socially acceptable living standards, and by promoting equality. Skeptics, however, retort that state intervention is actually inimical to well-being because it generates various social pathologies and macro-level inefficiencies, disembeds individuals from traditional institutions of social and economic support, and inhibits the ability of individuals to act as autonomous economic agents.
Moving beyond this impasse requires new theoretical thinking about the relationship between state intervention and well-being. Key to this approach is the realization that state intervention cannot merely be measured along a scale ranging from less to more. This obscures important features concerning the substantive qualities of political interventions into the market. Interventions can reflect markedly different social and economic prerogatives, the success of which also depends crucially upon the administrative capabilities of the state. In short, previous explanations suffer from omitted variable biases. By itself, the extent of political intervention into the market is neither a necessary nor sufficient condition for the promotion or inhibition of happiness. Scholars should also consider the extent to which state intervention privileges certain kinds of market vulnerabilities over others, as well as the quality of the administrative apparatus linking policy decisions to how individuals experience political intervention into the market in their daily lives. Once these important contextual factors are accounted for, we can generate a more accurate understanding of the conditions under which state intervention is likely to produce favorable life satisfaction outcomes. This study identifies one such condition. State intervention exerts a positive impact on life satisfaction when the intervention aims to protect individuals from new, post-industrial forms of market risk and when the quality of administrative institutions is high. The present study partially replicates and then extends the findings of Flavin, Pacek, and Radcliff (2011) to precisely illustrate how these important contextual factors moderate the relationship between state intervention and life satisfaction. Hypotheses are evaluated against data taken from Wave 5 (2005-08) of the World Values Survey (WVS) using both ordinary least squares (OLS) regressions with country-clustered standard errors and hierarchical, random-effects modeling techniques.
The article proceeds as follows. The second section of this article reviews some of the major theoretical and empirical findings on the relationship between state intervention and well-being. The third section uses insights from the literatures on post-industrial market risks and governance quality to promote a more accurate theory linking state intervention to life satisfaction and identifies some testable implications of such a theory. The fourth section outlines the research strategy used in the study. The fifth section presents the results of the analysis, and the last section concludes with a discussion of the implications these findings have for the wider literature exploring the relationship between politics, markets, and subjective well-being.
A House Divided: State Intervention and Subjective Well-Being
To what extent does state intervention into the market empower individuals to lead lives that they subjectively believe to be rewarding and satisfying? Proponents contend that greater political control of the market can increase life satisfaction through several direct and indirect pathways. First of all, an active state helps insure individuals against the uncertainties and economic insecurities associated with competitive markets. The threat of job loss is an important source of economic and mental frustration for many individuals, but the state can shield individuals from these uncertainties by introducing regulatory barriers to protect employment and/or by providing them with generous economic support in the event that individuals do find themselves without a job. Similarly, the state can be seen as a bulwark against a series of neoliberal policy reforms—such as a shift away from defined benefit pension plans, reductions in unemployment benefits, and new limits on health benefits—that have otherwise forced people to assume greater personal responsibility for risks associated with unfortunate life events and economic fluctuations (Hacker 2006; Taylor-Gooby 2004). Reduced personal capacity to cope with these and similar risks is a profound source of chronic mental stress (see e.g., Brenner 1977). Unsurprisingly, higher levels of stress are frequently associated with lower levels of subjective well-being.
Second, subordinating the market to political control can increase life satisfaction by emancipating individuals from their dependency on the market as the principal mechanism through which they satisfy their needs, desires, and ambitions in life. This is expressed perhaps most forcefully by Esping-Andersen's (1990) concept of “decommodification,” which measures the extent to which individuals can maintain socially acceptable standards of living independent of the financial resources they earn from selling their own labor as a commodity on the market. This leads to the powerful observation that most individuals probably do not like being reduced to commodities that are bought and sold on domestic and international labor markets (Flavin, Pacek, and Radcliff 2011). Moreover, market processes can undermine feelings of autonomy and self-efficacy, generate frustrations over economic decision making, and disembed individuals from meaningful social relationships (Lane 1978, 2000). Political intervention into the market can thus promote life satisfaction by emancipating individuals from the grasp of the “invisible-hand” of the market.
Third, state intervention has the potential to raise life satisfaction by promoting social and economic equality. Although the empirical literature on the relationship between inequality and subjective well-being presents mixed findings, 3 some studies do find that levels of subjective well-being tend to decline as the level of income inequality rises (Alesina, Di Tella, and MacCulloch 2004; Tomes 1986). This corroborates important epidemiological findings on the positive correlation between equality and good physical and mental health outcomes (Marmot 2004; Marmot et al. 1978; Wilkinson and Pickett 2009). Health, in turn, is a strong predictor of subjective well-being (Frey 2010; Helliwell 2003). Some scholars even maintain that egalitarian social relations approach the status of a fundamental human need, the satisfaction of which is vital for improving how individuals experience the lives that they lead in a positive way (Wilkinson 1996; Wilkinson and Pickett 2009). Greater inequality, by contrast, may force individuals to engage more frequently in relative comparisons of status that, on the whole, exert a detrimental effect on life satisfaction (Fujita 2008). According to this perspective, state intervention can increase life satisfaction by actively smoothing market-generated inequalities through redistributive taxation and transfers between different income classes. The aforementioned arguments find support in several empirical studies of the relationship between state intervention and subjective well-being (Alvarez-Diaz, Gonzalez, and Radcliff 2010; Di Tella, MacCulloch, and Oswald 2003; Flavin, Pacek, and Radcliff 2011; Haller and Hadler 2006; Pacek and Radcliff 2008a, 2008b; Radcliff 2001).
Welfare opponents, however, also offer a convincing theoretical explanation of why state intervention produces more misery, not happiness. One set of arguments maintains that even the most well-intended government interventions into the market can generate many negative, unintended consequences. First, standard economic theory predicts that greater market competition will raise aggregate levels of well-being by facilitating improved economic performance. Open and competitive markets help stimulate trade, encourage innovation, and slash consumer prices. Greater state intervention into the market prevents the realization of these gains, and these inefficiencies can translate into lower rates of economic growth (Butler and Kondratas 1987; Lindbeck 1995). 4 Suboptimal growth, in turn, negatively affects levels of prosperity, employment, and funding for key social programs.
Second, state encroachment upon market processes may negatively interfere with infrastructures of social support that people call upon when confronted with challenging life situations. The use of social and interpersonal connections as resources that enhance our capabilities to deal with life challenges is well documented in the literature (Hall and Lamont 2009). Sociological studies conducted in formerly communist countries report that people frequently sought assistance from relatives, friends, and colleagues as a means of coping with resource scarcity and navigating through complex, frequently corrupt, public bureaucracies (Hann 1993; Wedel 1986). Epidemiological research also suggests that the ill recover more quickly and that individuals with denser ties to social networks generally enjoy better health outcomes than those lacking such ties (Berkman and Glass 2000). The problem is that greater state intervention can disembed individuals from, and/or reduce the effectiveness of, these important sources of social resilience. For instance, a stronger state may undermine the relative usefulness of the family as a social resource because welfare correlates with higher levels of martial instability (Buckingham 2000; Gilder 1993) and an increase in low-income, single-parent families (Murray 1984). In short, the sprawling tentacles of the welfare state undermine well-being by strangling off traditional social institutions like the church and the family to which generations of individuals have long turned for social support.
Third, the collectivization of social life that proceeds in lockstep with the advance of the state into society further reduces happiness by undermining individual autonomy and free choice (DeSwaan 1988; Veenhoven 2000). Instead of arguing that the economic choices involved in market participation generate a host of decision-making frustrations (Lane 1978, 2000) or that the market reduces individuals to the level of commodities to be bought and sold (Esping-Andersen 1990), work by Freyer (1986) suggests that market participation actually promotes well-being by empowering individuals to exercise their autonomy as economic agents. Greater state intervention into the marketplace undermines well-being by placing unnecessary restrictions on private economic choice.
The other strand of arguments linking state intervention to greater misery stems from a host of rhetorical arguments positing a linkage between political intrusion into the market and the imposition of significant “moral” costs on society. Despite the noblest of intentions, state intervention into the market generates its own set of perverse incentives. Instead of empowering individuals to improve their own aspirations, desires, and living standards, social supports from the state lock individuals into a state of perpetual dependence on the beneficence of public welfare. Strengthened by perverse financial incentives that actually make it more cost-effective for individuals to remain unemployed or to remain in low-skilled, low-wage professions, state intervention directly facilitates “cultures of dependency” (Fraser 1997; Saunders 2000). Additionally, state intervention can generate patterns of benefit-induced migration (Allard and Danziger 2000), where moving decisions—instead of being influenced by calculations regarding one's family or employment prospects—are shaped instead by the extent to which particular jurisdictions offer robust and generous social insurance benefits. Although these rhetorical claims are hard to falsify and the empirical evidence in support of these phenomena is widely disputed, the meaning is clear: greater political intrusion into the market significantly undermines subjective well-being through moral perversion. As a whole, two important empirical studies find a negative relationship between state intervention and well-being (Bjornskov, Dreher, and Fischer 2007; Ouweneel 2002), and another finds no relationship (Veenhoven 2000).
Beyond the Impasse: Size Is Not Everything
The literature clearly lacks consensus on the relationship between state intervention into the market and individual levels of subjective well-being. A reasonable explanation could stem from the fact that the empirical analysis of survey data, particularly when our dependent variable relates to something as seemingly complex as subjective well-being, yields few strong and robust empirical signals relative to the disproportionately large amount of noise these analyses tend to produce. Variation in research designs—from modeling techniques, to sample sizes and case selection, to operationalizations of the dependent and independent variables—could all help explain the divergent findings across the literature. Be that as it may, the origins of these widely divergent findings may also have to do with important theoretical oversights, not just methodological choices. The literature's preeminent focus on the overall size or extent of state intervention into the market overlooks how other factors—notably the quality of administrative institutions and the orientation of social programs—can moderate the effect of state intervention on subjective well-being. For one thing, irrespective of its substantive content, state intervention hinges on the quality of public bureaucracies. The concept of administrative quality used here refers to the way in which authority is exercised by public officials. More specifically, we are interested in the extent to which public power is exercised impartially. Following Rothstein and Teorell (2008, 170), impartial administration means that “government officials shall not take into consideration anything about the citizen/case that is not beforehand stipulated in the policy or the law … when implementing laws and policies.” It is also important to note that the impartial exercise of public authority is not conterminous with corruption. While corruption—frequently defined as the abuse of public office for private gain—implies a violation of impartiality, impartiality includes other forms of partial exercise of administrative authority. Other practices—clientelism, discrimination, nepotism, patronage, and political favoritism—also interfere with the impartial implementation of policy and law (Rothstein and Teorell 2008).
The impartiality of administrative institutions matters because it ultimately influences how individuals experience the state in their daily lives—from their use of public transportation, to the consumption of publicly provided and/or publicly financed health care, to the frequenting of public parks and recreation areas. When citizens are consistently treated professionally, respectfully, and impartially in their interactions with public servants, the procedural utility of consuming public goods and services increases (Frey 2010). People not only evaluate actions taken toward them by considering the consequences of those actions but also on the basis of how they feel treated by other people. Institutions shape the nature of those interactions by incentivizing participants to treat each other positively (or negatively) during the course of their everyday interactions. Specifically, policies regarding the provision of welfare goods and services crucially shape the interaction between policy administrators and welfare state constituencies. The extent to which the intended beneficiaries of political interventions into the market feel treated in a manner that is fair, respectful, and consistent with their prior experiences and the experiences of others who are like them improves their sense of self. Indeed, empirical research repeatedly shows that procedural utility in the consumption of public goods and services is positively correlated with subjective well-being (Frey and Stutzer 2000, 2005; Layard 2006; Ott 2011; Whiteley et al. 2010).
Opportunities for interacting with public service bureaucracies increase as the state assumes a more active role in managing and regulating markets. Under these circumstances, the procedural experiences associated with the consumption of public goods and services become increasingly important in how individuals evaluate the quality of their own lives. Therefore, the extent to which state intervention into the market increases subjective well-being hinges on the quality of administrative institutions. A robust state presence in the management of market forces can actually do more harm than good if public service bureaucracies are corrupt, unprofessional, disrespectful to citizens, and inconsistent in the application of regulatory rules and policies. Indeed, one study (Bjornskov, Dreher, and Fischer 2007) finds that state intervention into the market only exerts a positive effect on happiness at high levels of institutional quality.
Procedural experiences in the consumption of public goods and services are important, but this should not detract from the fact that the substantive content of market interventions also influences how individuals subjectively experience life. This leads to another potential theoretical oversight dogging the relationship between state intervention and subjective well-being. One of the principal mechanisms linking state intervention to greater quality of life outcomes relates to the ability of the state to insure individuals against market-generated risks. However, the concept of risk invoked by the literature is cast in a very general sense. The nature of these risks is not always clearly defined, and it is also unclear whether such risks are considered homogeneous across time and space. This latter issue is of particular concern. Political scientists have long recognized that drastic structural changes across many of the world's advanced, capitalist democracies have generated a series of new, post-industrial market risks (Bonoli 2005; Esping-Andersen 2002; Taylor-Gooby 2004), and that societies have differed greatly in their ability to respond to this challenge. Such risks include unstable employment patterns, long-term unemployment, working poverty, single parenthood, demographic ageing, the role of women in the workforce, and the related issues stemming from reconciling the demands of work and family life.
Protecting individuals from new social risks requires an array of policy instruments distinct from those used to address the traditional risks defining many societies in the immediate post-war era. A traditional concern with income replacement policies—in the form of pensions, disability payments, or unemployment benefits—competes with additional concerns about the promotion of human capital formation and the empowerment of various family types to balance the exigencies of employment with childbearing. Unfortunately, current approaches fail to explicitly model these heterogeneities of risk—both theoretically and empirically. Overall measures of government size, welfare effort, or similar indices of state intervention fail to differentiate between different types of risk. States with seemingly robust welfare capacities may be grossly misallocating those resources. The extent to which welfare resources are allocated between old and new social risks determines the extent to which particular welfare state constituencies are actually insured against relevant market risks. Greater resource misallocation suggests that more individuals go un- or under-insured, and this, in turn, should correspond to lower levels of subjective well-being.
However, this second pathway linking the allocation welfare resources and subjective well-being is clearly conditional on the quality of the administrative apparatus of the state. The implementation and enforcement of regulatory frameworks, the provision of public goods and services, and the resolution of disputes regarding the administration of benefits in a manner consistent with social policy legislation designed to privilege post-industrial forms of market risk all presume an impartial, nonpartisan, and efficient system of public administration. Public bureaucracies operate as an important transmission belt between legislative outputs and policy outcomes. However, policy opponents can more easily weaken this linkage when corruption and mismanagement plague public administration. Policy opponents can use a pliable civil service system to undermine the influence of regulators, raise legal challenges, weaken monitoring and enforcement mechanisms, and dilute sanctions for noncompliance to circumvent policies they find inconvenient or disagreeable. The literature on post-communist transitions, for example, is replete with examples of vested political and private interests capturing public institutions and exploiting them to achieve their own ends, frequently circumventing current policy statutes in the process (see e.g., Hellman 1998; Holmes 2006; Vachudova 2009). Such relationships are not unique to the post-communist realm, either. A recent study of Western European democracies finds that satisfaction with democracy, frequently regarded as a proxy for how well democracy works in practice (Linde and Ekman 2003), is associated with the rule of law, smooth regulatory frameworks, and low levels of corruption (Wagner, Schneider, and Halla 2009).
Such destructive behavior is not unique to policy opponents, either. Poorly trained, unprofessional public bureaucrats can also subvert the letter of the law when administering public goods and services designed to help insulate individuals from market and social risks. Bureaucrats can wrongfully deny eligible benefit claimants due to poor administrative practices, poor operating procedures, and/or poor quality decision making (Van Oorschot 2002). Errors in administrative judgment could stem from a lack of sufficient information or stereotyping clients on the basis of race, ethnicity, religion, and so forth. For instance, in an ethnographic study of a local social security office in Northern Ireland, Howe (1985) finds that administrators frequently treated claimants differently on the basis of their physical characteristics and social demeanors. Overall, administrators were less likely to solicit sufficient biographical and financial information from claimants when claimants appeared slovenly and/or were perceived to be rude, aggressive, and intransigent. Profiling in this manner undermines the intentions of various forms of political intervention into the market designed to help particular constituencies of the welfare state. Otherwise eligible individuals, such as single mothers seeking tax relief or the recently unemployed in search of training services, could be wrongfully denied public assistance. In the aggregate, actions such as these overstate the effectiveness of social policy regimes that would otherwise appear quite accommodating of post-industrial forms of market risk.
Combing insights from these two moderating variables suggests that state intervention should have a strong positive effect on life satisfaction when social programs are oriented toward post-industrial market risks and when the quality of administrative institutions is high. In this scenario, individuals benefit from the efficient and fair administration of social policies designed to insure them against relevant market risks. This leads to the following hypothesis:
- Hypothesis 1: State intervention will have a strong positive effect on life satisfaction when the quality of administrative institutions is high and when social policies privilege insuring individuals against post-industrial forms of market risk.
The main hypothesis of this study attempts to explore how the effect of state intervention on subjective well-being is moderated by two key variables: the quality of administrative institutions and the extent to which social policies cater to traditional versus post-industrial forms of market risk. To help isolate the effects of these key theoretical variables, the empirical approach will center on replicating and then extending the findings of a recent study by Flavin, Pacek, and Radcliff (2011) on the relationship between state intervention and life satisfaction. In their original study, Flavin, Pacek, and Radcliff find that greater state intervention, measured in a variety of ways, exerts a positive and significant effect on life satisfaction.
All individual-level data come from survey responses drawn from 15 advanced capitalist democracies in Wave 5 (2005-08) of the WVS. 5 The case selection in this study reflects key analytical interests, as well as practical considerations regarding the availability of data on important independent variables. First, the countries included in this study mirror those included in the original analysis conducted by Flavin, Pacek, and Radcliff (2011). Restricting the analysis to an identical set of cases helps increase the validity of the causal inferences drawn from the empirical analysis. This quasi-experimental approach helps ensure that any observed effects can be attributed to our key theoretical variables of interest, as opposed to the inclusion of different countries in the analysis or some combination of the two. As part of the iterative and cumulative nature of social scientific research, future research could then take up the cause of extending the arguments here to new empirical domains. The second reason for such fidelity in the case selection stems from a pragmatic concession that data limitations simply prevent us from including additional countries in the analysis. 6
The dependent variable, life satisfaction, is taken from a survey item asking respondents the following question: “All things considered, how satisfied are you with your life as a whole these days?” Responses are coded along a 1-10 scale, whereby greater values correspond with higher levels of perceived life satisfaction.
The main independent variable, state intervention, is measured using total public expenditures on social policy as a percentage of gross domestic product (GDP). Covered areas include: old age, survivors, incapacity, health, family, active labor market programs, unemployment, housing, and a residual category of other social expenditures and subsidies (OECD 2011a; Svensson et al. 2012). Admittedly, Flavin, Pacek, and Radcliff (2011) also measure the size of state intervention using three additional indicators—a country's tax revenue as a percentage of GDP, a government's consumption share of real GDP per capita, and a “social wage” measure that captures the average gross unemployment benefit replacement rate across two earning levels, three family types, and three durations of unemployment. The use of multiple indicators in this fashion reflects their contention that “[a]s important as the welfare state is, it is hardly isomorphic to the wider questions of what they [previous scholars] call ‘dependency’ on the market” (Flavin, Pacek, and Radcliff 2011, 256). This is a valid point, but limitations in the available data preclude easy empirical specification of the extent to which tax revenues or government consumption, for instance, reflect post-industrial forms of market risk—one of the key moderating variables examined in this study. Even so, this is no reason to throw the theoretical baby out with the bathwater. As long as we are aware of the potential limitations of defining state intervention solely on the basis of welfare state size, we can still produce useful insights concerning whether and how the relationship between welfare and well-being is moderated by important contextual factors.
The main moderating variables seek to capture how state intervention discriminates, or allocates resources, in favor of post-industrial forms of market risk and the quality of administrative institutions. Allocation is proxied by the New Social Risk Share (NSRS) measure, a spending variable designed to measure the extent to which public welfare expenditures cater to the alleviation of new, post-industrial risks. Following the conventions of Tepe and Vanhuysse (2010), this measure is defined as the ratio of spending on family benefits and active labor market policies (numerator) to the sum total of spending on family benefits, active labor market policies, unemployment benefits, survivors benefits, and incapacity benefits (denominator). Spending on family and active labor market policies are key new social risk programs, as they reflect new socioeconomic policy goals established across many advanced capitalist democracies. The European Union's Europe 2020 Agenda, for instance, stresses the importance of empowering European workforces through renewed investment in skills and human capital as a means of increasing productivity, competitiveness, and reducing levels of unemployment. Also key to the Agenda's flagship initiative for new skills and jobs is the reduction of labor market segregation by facilitating the reconciliation of work and family life. Additionally, some may notice that spending on public pensions, while common in many discussions of social policy, is conspicuously absent from the denominator. Pensions are deliberately omitted because old age is an inherent feature of the human condition, not something that we can easily classify as a “new” or “old” social risk. Higher NSRS values indicate that social policies cater more toward accommodating post-industrial forms of market risk.
Measures of administrative quality abound, but these indicators are plagued by many of the same substantive trade-offs and practical considerations as the theoretical constructs they claim to reflect. One obvious candidate is a measure of administrative impartiality generated from the QoG Expert Survey (Teorell, Dahlstrom, and Dahlberg 2011). The impartiality variable measures the extent to which government institutions exercise their power impartially, with the norm of impartiality, again, defined as: “When implementing laws and policies, government officials shall not take into consideration anything concerning the citizen/case that is not beforehand stipulated in the policy or the law” (Rothstein and Teorell 2008, 170). 7 The principal advantage of this measure is that it most closely approximates the concept of administrative quality used in this study. However, the national averages computed from the QoG Expert Survey only refer to the period between 2008 and 2012, as the QoG Expert Survey was first introduced in 2008. Unfortunately, Wave 5 of the WVS only covers the years 2005-08. Thus any results obtained from the use of the impartiality variable should be interpreted cautiously, particularly since the onset of the financial crisis might be associated with quite marked shifts in the quality and performance of some public bureaucracies.
To partially offset the time inconsistencies in the impartiality data, the analysis also considers another measure of administrative quality derived from the aggregate score of Transparency International's Corruption Perceptions Index (CPI) 8 as a robustness check. The CPI focuses on public sector corruption, defined as the abuse of public office for private gain. Values range between 0 and 10, whereby higher numbers indicate “cleaner” public administrative and political practices. The score reflects perceptions of the degree of corruption in different societies by business people, risk analysts, and members of the general public. Unlike the impartiality variable, aggregate CPI scores do match the country-years covered in the WVS data. However, one major shortcoming of the CPI concerns its exclusive substantive focus on corruption. These substantive concerns nevertheless do not translate into substantially different values when compared to scores from the impartiality index. The zero-order correlation between the CPI scores and the impartiality index in the data is a very reasonable 0.73. This strengthens our confidence in the suitability of the CPI data as a reasonable substitute for the impartiality data. 9
Life satisfaction is also modeled as a function of important individual- and country-level controls that could influence respondents' assessments of how satisfied they are with their lives. At the individual level, the analysis controls for the respondent's income, education, health, gender, age, church attendance, interpersonal trust, and religion. 10 At the country-level, the analysis controls for each country's GDP, national unemployment rate, and the “individualism” of each country's culture. 11 Additionally, we must confront the fact that measures of resource allocation will be partly endogenous to structural characteristics of domestic labor markets. Higher NSRS values, for example, do not constitute a priori evidence of a well-calibrated welfare state. Higher NSRS values may simply reflect high levels of female labor force participation and/or high fertility rates. A variable measuring the year in which each country experienced the onset of post-industrial risks (Bonoli 2007; Tepe and Vanhuysse 2010) is included to control for the prevalence of new social risks within domestic labor markets. Earlier post-industrial transitions imply greater functional pressures for reallocating welfare resources toward new social risks. 12
Following Flavin, Pacek, and Radcliff (2011), estimation proceeds by using a series of OLS regressions reporting country-clustered, Huber-White robust standard errors to account for between-country heteroskedasticity and within-country correlation. Although respondents answered questions concerning their perceived life satisfaction using a multi-step ordinal scale, the dependent variable is treated as continuous in the analysis. Ferrer-i-Carbonell and Frijters (2004) find that assuming ordinality or cardinality makes no difference in the analysis of subjective well-being data, and treating the dependent variable as continuous follows a convention used by many in the literature, including Flavin, Pacek, and Radcliff.
However, one of the potential limitations of this research design concerns the fact that the analysis covers a relatively small set of countries (N = 15). Conventional OLS modeling techniques can generate estimates that are subject to a high degree of sample-to-sample variability. In other words, the estimates can be overly sensitive to the random error in any given dataset (Clark and Linzer 2013), as well as heterogeneity in the sample sizes between countries 13 (Snijders and Berkhof 2008). Moreover, one of the reviewers suggested that the Nordic countries—renowned for having both high levels of happiness and high levels of governance—weigh quite heavily in the dataset and that this correlation could be rooted in other sources. While the inclusion of macroeconomic and cultural controls in the analysis helps reduce the likelihood of confounding relationships between the main independent and dependent variables, the inferential concerns raised by the small number of countries included in the analysis remain.
To ensure the validity and reliability of the results, analyses were also conducted using a two-level, hierarchical random-effects model. This approach allows factors at the levels of country (Level 2) and individual (Level 1) to explain variation in individual life satisfaction. Instead of merely correcting for the clustered nature of the data, the random-effects specification estimates random intercepts for each of the countries included in the analysis. This estimation approach is more suitable for when our target of inference is more than just the sample of countries represented in the dataset but actually includes a wider population of countries (Rabe-Hesketh and Skrondal 2008; Skrondal and Rabe-Hesketh 2004). In the case of this analysis, the estimated country intercepts are used to help generate valid inferences concerning the entire population of advanced, capitalist democracies about which we are concerned. For these reasons, a random-effects approach is a good robustness check of the findings generated from the pooled OLS regressions with country-clustered standard errors. All analyses were conducted using Stata 12.1.
Table 1 presents results on the relationship between state intervention, resource allocation, impartiality, and life satisfaction. The base model reproduces results from Flavin, Pacek, and Radcliff (2011) on the relationship between social expenditures and life satisfaction. Greater state intervention exerts a significant and positive effect on the extent to which respondents feel satisfied concerning the lives that they lead. The average effect of state intervention on life satisfaction is even robust to the inclusion of controls for the share of resources allocated toward new social risks and impartiality in Model 2. Administrative quality exerts a strong, significant, and positive effect on life satisfaction, which resonates with the findings of other studies on the relationship between government quality and subjective well-being (Helliwell and Huang 2008; Holmberg, Rothstein, and Nasiritousi 2009; Layard 2006; Ott 2011; Whiteley et al. 2010). The effect of the NSRS variable is negative. This may seem counterintuitive, as it is argued that higher NSRS values imply that social policies are better suited to insulate individuals from post-industrial forms of market risk. However, a key part of the argument is that the effect of NSRS will be conditional on the quality of administrative institutions.
|Model 1 Base||Model 2 Moderators added||Model 3 Interaction (FE)||Model 4 Interaction (RE)|
Models 3 and 4 explore the nature of these conditionalities by interacting NSRS, administrative quality, and social expenditures. Model 3 uses a conventional OLS estimation with country-clustered robust standard errors, while Model 4 employs a random-effects, multilevel model using maximum likelihood estimation to address concerns about the small number of countries included in the dataset. It was originally hypothesized that state intervention would exert a positive effect on life satisfaction when state intervention privileges post-industrial forms of market risk (i.e., a high NSRS value) and the quality of administrative institutions is high (i.e., a high score on the Impartiality Index). The interaction term is positive and significant, although a three-way interaction between three continuous variables does not lend itself to easy interpretation. To facilitate this, Table 2 presents and compares the marginal effect slopes of a one-unit increase in social expenditures on life satisfaction across four different combinations of the moderating variables—NSRS and impartiality. The high and low values of each of the moderating variables respectively correspond to a one standard deviation change above and below each variable's global mean. The four different conditions are presented in the first column of the table.
|Condition||Model 3 Slopes||Model 4 Slopes|
|1. High quality, high NSRS||.187***||.142***|
|2. High quality, low NSRS||−.100||−.053|
|3. Low quality, high NSRS||.088||.020|
|4. Low quality, low NSRS||.078***||.077***|
If Hypothesis 1 is correct, the marginal effect of a one-unit increase in social expenditures should exert a positive effect on life satisfaction when administrative quality and NSRS are high (Condition 1). This is indeed the case illustrated by the marginal effect slopes of each of the four conditions presented in the second (Model 3) and third (Model 4) columns. An increase in social expenditures exerts a significant and positive effect on life satisfaction when Condition 1 (high administrative quality; high NSRS) obtains. By contrast, increased spending exerts no significant effect on levels of life satisfaction in relatively impartial welfare states that still privilege more traditional forms of market risk (Condition 2) and relatively partial welfare states that privilege post-industrial (Condition 3) forms of market risk. Surprisingly, the marginal effect of social expenditures is also significantly positive in polities characterized by low administrative quality and outmoded forms of risk protection (Condition 4). This was not anticipated and suggests multiple conjunctural causalities (Ragin 1989) linking state intervention and subjective well-being. Even so, the effect size of Condition 1 is roughly twice as strong in magnitude as the effect size of Condition 4, and the fact that an increase in social expenditures is positively correlated with life satisfaction when both impartiality and NSRS values are high confirms the main hypothesis of the paper.
To ensure that the observed relationships are not merely artifacts of the impartiality measure, the analysis is rerun using CPI scores as a new proxy for administrative quality. This allows us to partially correct for the fact that the impartiality data used in the first analysis applies to country-years not represented in Wave 5 of the WVS. Table 3 presents the results of the analysis. Model 5 adds our two key moderating variables—CPI and NSRS. As in Model 2, the effects of administrative quality are again positive and significant, although the effects of the NSRS measure and social expenditures are now insignificant. This last finding is important as it suggests that the relationship between social expenditure and life satisfaction may vary over different combinations of our key independent variables.
|Model 5 Moderators added||Model 6 Interaction (FE)||Model 7 Interaction (RE)|
Models 6 and 7 explore how our moderating variables influence the relationships between social expenditures and life satisfaction. As before, Table 4 helps us interpret the significant and positive interaction terms between social expenditures, administrative quality, and NSRS presented in both models. From the second and third columns of Table 4, we again observe that an increase in social expenditures exerts a positive and significant effect on life satisfaction when both administrative quality and NSRS are high. Under all other combinations of these two moderating variables, increased social expenditures exert either significantly negative effects on life satisfaction (Conditions 2 and 3) or no effect at all (Condition 4). While the inclusion of a different measure of administrative quality changes the relationship between social expenditures and life satisfaction across Conditions 2 to 4, only Condition 1 is positive and significant across both analyses. This not only confirms the main hypothesis of the article, but it also helps emphasize the more general point that the relationship between state intervention and life satisfaction is conditioned by other important contextual factors.
|Condition||Model 6 Slopes||Model 7 Slopes|
|1. High quality, high NSRS||.104***||.106***|
|2. High quality, low NSRS||−.147***||−.129***|
|3. Low quality, high NSRS||−.173***||−.164***|
|4. Low quality, low NSRS||−.016||−.016|
Before proceeding, a final remark concerning model specification is in order. In their original study, Flavin, Pacek, and Radcliff (2011) conducted a second analysis of the data in which control variables for national unemployment rate and unemployment status were omitted. This modified approach helps account for the fact that these control variables could potentially mask some of the negative consequences of social expenditures on life satisfaction. To the extent that unemployment undermines life satisfaction and to the extent that increased social expenditures could be partially a response to higher levels of unemployment, controlling for unemployment could artificially inflate the positive effect of social expenditures on life satisfaction. The modified analysis also omitted the dummy variable for marital status, on the expectation that the incentives for marriage may be, themselves, partially endogenous to the generosity of the welfare state. To address these concerns, all models were re-specified and reanalyzed after dropping variables for unemployment status, national unemployment rates, and marital status. Due to space limitations, Table 5 only presents re-specified versions of all the interaction models from Tables 1 to 3. Like Flavin, Pacek, and Radcliff (2011), re-specifying all of the models yields substantively similar findings. An analysis of marginal effects slopes in Table 6 still demonstrates that greater state intervention significantly increases life satisfaction when the quality of administrative institutions is high and when social policies privilege post-industrial market risks.
|Model 8 Impartiality (FE)||Model 9 Impartiality (RE)||Model 10 CPI (FE)||Model 11 CPI (RE)|
|Condition||Model 8 Slopes||Model 9 Slopes||Model 10 Slopes||Model 11 Slopes|
|1. High quality, high NSRS||.172***||.125**||.089**||.091***|
|2. High quality, low NSRS||−.056||−.008||−.155***||−.142***|
|3. Low quality, high NSRS||.090||.014||−.145**||−.141**|
|4. Low quality, low NSRS||.081***||.079***||−.018||−.013|
Discussion and Conclusion
This study helps to explain the cacophony of conclusions circulating in the literature concerning the relationship between state intervention into the market and subjective well-being. In a sense, everyone is right; state intervention is simultaneously a force for happiness and misery. However, instead of writing off these divergent results as a function of different methodological choices, this analysis identifies in administrative quality and resource allocation two important factors that significantly moderate the effect of state intervention on happiness. Greater intervention exerts a strong, positive effect on perceived levels of life satisfaction when the quality of administrative institutions is high and intervention focuses on insuring individuals against post-industrial forms of market risk. Under these conditions, individuals benefit from the impartial administration of social policies designed to protect them from relevant sources of market risk.
These findings have some important normative implications for policy makers, as well. Because this study promotes a more nuanced picture of the relationship between state intervention and subjective well-being, policy makers should rethink increasing state intervention into the market as a panacea for empowering individuals to feel more satisfied concerning the lives that they lead. In polities plagued by inefficiencies in the administration of public goods and services and/or anachronistic social protection regimes, funneling more resources into the welfare state could actually make matters worse.
This study finds robust support for one set of contextual factors that lead to a positive relationship between state intervention and subjective well-being, but other conditionalities may likely lead to similar outcomes. Indeed, one of the analyses suggested that greater intervention also promoted life satisfaction in the absence of both contextual factors (i.e., in poorly administered welfare states privileging outmoded forms of market risk). More theorizing and empirical analysis is needed to tease out these alternative pathways. To this end, future studies would also do well to extend the arguments here to new empirical domains. One of the potential limitations of this study is the small number of countries included in the analysis. While the use of multilevel modeling techniques increases our confidence in the results, methodological controls are no substitute for testing these hypotheses against more data. Experimenting with different case selections and time periods will help increase the robustness of these findings or identify important scope conditions to the main arguments of this article. Analyses of longitudinal data would also allow us to predict how changes in levels of administrative quality and the allocation of welfare resources map onto the changes in subjective well-being individuals experience over time. 14 Finally, this study advances a relatively narrow interpretation of state intervention into the market by focusing exclusively on social expenditures. A variety of other measures—from tax revenues to government consumption statistics to regulatory measures also proxy how prevalent the state is in the economy. However, limitations in the available data preclude easy empirical specification of the extent to which these and similar indicators reflect post-industrial forms of market risk—one of the key moderating variables examined in this study.
This study makes no pretense of settling the debate as to whether greater state intervention into the market improves the extent to which individuals find their lives to be enjoyable, satisfying, and rewarding. In fact, it raises more questions than it answers. However, in doing so, the hope is that future research will continue to ask finer grained questions concerning the relationship between state intervention and subjective well-being.
The concepts of “subjective well-being,” “happiness,” and “life satisfaction” are used interchangeably throughout this article. For the purposes of this study, they all refer to the same underlying concept. A vast archive of research findings involving the use of data on measures of subjective well-being is publicly available from the “World Database of Happiness” (Veenhoven 2013).
Concerns about the suitability of “soft” concepts like “happiness,” “life satisfaction,” and “subjective well-being” for empirical, scientific research abound, but such fears are largely unwarranted. Measures of subjective well-being are valid (Costa and Mccrae 1988; Fernández and Ruiz-Belda 1995; Helliwell 2006; Lepper 1998; Moum 1996), reliable (Eid and Diener 1999; Fordyce 1988; Helliwell 2006; Kahneman and Krueger 2006; Sandvik, Diener, and Seidlitz 1993; Veenhoven 1996, 1997), and cross-nationally comparable (Inglehart 1990; Veenhoven 1993, 1996, 1997). Such measures are also quite robust to concerns about endogeneity and self-selection biasing causal inferences (Clark and Georgellis 2012; Clark et al. 2008; Gardner and Oswald 2007; Lucas 2007; Winkelmann and Winkelmann 1998). Subjective well-being indicators have not gone unnoticed by policy makers, either. Former French President Nicholas Sarkozy, dissatisfied with the use of gross domestic product per capita and other material measures of social progress, recently tasked a 25-person commission of prominent economists and other academics to search for better indicators of well-being (Commission on Growth and Development 2008). The fact that “hard-core” behavioralist economists—Joseph Stiglitz, Amartya Sen (both Nobel Prize winners), and Jean-Paul Fitoussi—endorsed the use of self-reported data on subjective well-being as a more suitable means of measuring well-being speaks to the confidence that researchers and policy makers alike should have in the validity, reliability, and comparability of subjective well-being measures (Easterlin 2010). A copy of the report is available from the Commission on the Measurement of Economic Performance and Social Progress (2009).
See Graham (2011) for a comprehensive overview of the literature.
See Kenworthy (1999) for discussion of this debate.
Countries analyzed from the WVS dataset include: Australia, Canada, Finland, France, Germany, Great Britain, Italy, Japan, the Netherlands, Norway, South Korea, Spain, Sweden, Switzerland, and the United States.
Data on two important macro-level variables, in particular—the year in which countries experience the onset of new social risks and social expenditures statistics—were not available for all the other countries included in Wave 5 of the WVS.
The impartiality variable represents an index built on expert responses to the following five items asked on the QoG Expert Survey: (1) Firms that provide the most favorable kickbacks to senior officials are awarded public procurement contracts in favor of firms making the lowest bid; (2) When deciding how to implement policies in individual cases, public sector employees treat some groups in society unfairly; (3) When granting licenses to start up private firms, public sector employees favor applicants with which they have strong personal contacts; (4) How often would you say that public sector employees today act impartially when deciding how to implement a policy in an individual case?; and (5) Hypothetically, let's say that a typical public sector employee was given the task to distribute an amount equivalent to 1,000 USD per capita to the needy poor in your country. According to your judgment, please state the percentage that would reach the needy poor. The index is constructed by weighting each item by a factor loading obtained from a principal components analysis and taking the sum of all five of the weighted items. Aggregation to the country level occurs by taking the mean value across all surveyed experts per country. Values in this analysis range between –.18 and 1.33, whereby higher values indicate more impartial administrations.
Data reported in Svensson and others (2012).
A third alternative to measuring administrative quality comes from the World Bank Governance Indicators Dataset (Kaufmann, Kraay, and Mastruzzi 2010). The dataset's aggregate indicator is commonly used in studies of the relationship between government quality and subjective well-being, but it includes data from subindicators on political stability, political accountability, and the rule of law, the combination of which is less directly relevant for our primary interest in gauging the impartiality of administrative institutions. Additionally, recent scholarship has also challenged the construct validity of this indicator (Langbein and Knack 2010; Thomas 2010). By contrast, the methodology of the CPI was generally well received in a recent independent review by the European Commission's Joint Research Center. A copy of the report is available to download at: http://cpi.transparency.org/cpi2012/in_detail/
Income is measured using a scale of incomes variable, ranging from 1-10, where 1 indicates the lowest income decile, and a value of 10 indicates the highest income decile. Decile placement is based upon a respondent's self-reported income and the relevant income distribution for the country in which the respondent resides. Education is a continuous measure that ranges from 1 and 9, whereby higher scores indicate higher levels of educational attainment. A value of 1 indicates that the respondent has no formal education, while a value of 9 indicates that the respondent has a university degree. Health reflects a self-assessment of the respondent's health status, ranging from very poor (1) to very good (5). Respondent age and age-squared are both included in the analysis, reflecting the fact an individual's level of happiness typically follows a U-shaped function. Individuals are significantly happier at the beginnings and ends of their lives than they are at the middle. Dummy variables account for a respondent's gender, marital status, employment status, interpersonal trust, and religious domination. The trust variable comes from a survey item that asks the following question: “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” The trust variable is coded 1 if the response was “most people can be trusted” and 0 otherwise. Religious dummies include variables for Protestants, Catholics, Muslims, Jews, Hindus, and Buddhists. All other confessional groups constitute the reference category.
GDP data are measured in 1,000s of U.S. dollars and are from the Penn World Tables (Heston, Summers, and Aten 2009). Unemployment rate data are taken from the OECD (2011b). Individualism captures the orientation of society along the collectivist–individualist continuum highlighted by Diener, Diener, and Diener (1995). Values range between 1 and 10. Higher values indicate more individualistic societies. The data originally come from Triandis (1989) but are reported in Diener, Diener, and Diener (1995).
This measure represents the average year in which each country in the dataset experienced the onset of three post-industrial, structural developments associated with the rise of new social risks: service employment as a percentage of total civilian employment, female employment rate, and the divorce rate. Sweden in 1970 constitutes the benchmark year (service employment = 54 percent; female employment rate = 58 percent; divorce rate = 30 percent). The new risk onset variable represents the average of the three years in which each country approached the Swedish 1970 levels of these indicators. See Bonoli (2007) for details.
The number of observations for each country are presented in Table A3 in the Appendix.
Admittedly, Veenhoven's (2000) study is longitudinal to the extent that it finds no relationship between the change in well-being and the change in the size of the welfare state between the years 1981 and 1990. However, the dependent variable only refers to the change in national-level averages of life satisfaction between two time points (1981 and 1990), and the analysis relies solely on zero-order and partial correlations that fail to control for important national-level factors. Most importantly, Veenhoven's analysis does not consider the quality of administrative institutions and the extent to which those expenditures privilege different forms of market risk.