The Welfare Mix and Trust in Government

How might the share of private financing in distinct types of welfare states—or, in our case, health systems—influence citizen trust in political institutions? Building on the scholarly literature on trust, risk, and the welfare state, we argue that private financing of health care can shape trust in political institutions through two interrelated pathways: via perceptions of risk and via evaluations of the performance of the health care system.

Trust in government (and a closely related concept, confidence in political institutions) has been found to emanate from a positive evaluation of the performance of public institutions. ^{Footnote 5} Assuring the basic welfare of citizens is a central—perhaps the central—task of public institutions in regulated market economies such as those in the rich democracies. But in none of these countries do public institutions take on sole responsibility for financing and delivering the totality of the social goods that make up the welfare “state.” Governments in regulated market economies play both direct and indirect roles in social service financing and provision: providing revenues to pay for cash transfers and deliver public services, but also regulating private markets for social goods, contracting out, and the like. Thus, while citizen assessments of government performance are influenced by how well the government provides welfare services, ^{Footnote 6} satisfaction with those functions of the welfare “state” that are left to private actors likely also influence citizens’ evaluations of public institutions. ^{Footnote 7} In other words, trust in government is likely to depend not only on citizens’ satisfaction with the performance of public welfare institutions, but also on their satisfaction with privately financed or provided social services that make up the other part of the welfare “state.”

Satisfaction with the welfare state, and ultimately trust in government, is likely to be influenced not only by the quality of services, but by the degree to which welfare states perform what is arguably their core function: to provide insurance against risk. Prior research on public opinion regarding social policy has found that citizens’ satisfaction with welfare institutions depends in part on their perceptions of how well these institutions protect them from life-course risks such as unemployment or disability—or, conversely, leave them more exposed, through greater “individualization of risk.” ^{Footnote 8} To the extent that distinct mixes of public and private financing of the welfare state differently mediate individual experiences of risk to shape attitudes toward welfare policies, they should in turn affect levels of trust in government.

Previous research, however, does not reveal the precise shape of the relationship between individuals’ exposure to risk and their trust (or lack of trust) in government. On the one hand, individuals with greater objective exposure to risk have been found to be more supportive of government social programs. ^{Footnote 9} Citizens who need public support may develop a positive view of the government institutions on which they depend, and a more skeptical view of markets, which leave them exposed to risk. ^{Footnote 10} Increased trust in government may thus arise in settings with high levels of private financing because those at risk appreciate the government’s actions to buffer market risks and trust the state to act as a provider of last resort.

On the other hand, trust in government may be reduced in settings with high levels of private social provision, particularly among those who believe that their government does not do enough to protect them from the risks to which the market exposes them. For example, large out-of-pocket payments for health care services have been found to reduce confidence in the health care system, particularly among lower-income citizens. ^{Footnote 11} Moreover, poverty has been linked to lower evaluations of health care quality in Europe, particularly in countries with higher perceived levels of corruption. ^{Footnote 12} The combination of greater need with disappointment in the face of the failure of government to meet expectations may, in other words, reduce trust in government institutions. If this is the case, citizens at risk may simultaneously expect more from their governments and, in more privatized systems, express lower political trust.

In sum, when government delegates the task of insuring against social risk to the private sector, individuals may feel—and indeed may be—less well protected from risk than they would in a comprehensive, public welfare state. Health systems with more private financing and provision have been found to provide weaker protections to ensure that all citizens are buffered against the risk of needing care that they cannot afford. For example, disparities in health care utilization by socioeconomic status have been found to be greater where the share of private spending is higher. ^{Footnote 13} The question is whether this greater exposure to risk in more private systems leads to greater or lesser trust in government.

The logic we have laid out here suggests that variation in the share of private financing of health care may be associated with varying levels of trust in government because private financing amplifies exposure to risk. But previous research has found that those who are more trusting of the state ex ante ^{Footnote 14} or who have more positive evaluations of the quality of their government ^{Footnote 15} may be more likely to call upon the state to insure them against risk, leading to higher levels of state provision of social welfare goods. In assessing the potential causal impact of private financing of health services on trust in government, then, we must at a minimum take into account preexisting levels of trust in government. In our further analysis we will employ several strategies to establish a plausible direction of causation and rule out other factors or possible reverse causation as the source of our results. However, it is worth raising at the outset a more conceptual motivation for our hypothesis that private financing influences trust in government, rather than (or in addition to) the other way around.

Before they affect government policy, individual-level attitudes—even when aggregated at the national level—are normally first filtered and shaped by institutional and electoral configurations. ^{Footnote 16} Hence, if we think that the causal arrow goes from attitudes (citizen trust) to policy (the role of private versus public actors in welfare regimes) we must assume that country-specific political and electoral institutions don’t make much of a difference in how individual-level preferences are aggregated and expressed. The causal direction we propose—that government policy, mediated by risk, leads to individual preferences—instead requires only the more modest hypothesis that similarly situated individuals (on average and ceteris paribus) experience the same psychological state (lack of trust) in response to the same stimulus (more market, less state).

The pattern of recent reforms to health care systems also suggests that current levels of private financing are at least partly a product of exogenous forces. While a generalized and growing public mistrust in government may have facilitated the trend toward more private financing, ^{Footnote 17} in western European countries fiscal pressure and a desire to control rising health care costs have been prominent motivations for expanding out-of-pocket payments in health care. And regardless of the motivations, health system reforms in the past twenty years have tended to be a top-down process, driven by political elites and policy experts rather than citizens. ^{Footnote 18} In eastern Europe, private financing and provision were introduced in the wake of a wholesale regime collapse that was caused only in the broadest of senses by declining trust in government. In sum, we see good reasons to believe that private financing of health care affects trust in government at least as much as the reverse. Before proceeding to the analysis, we lay out the steps we propose to link health care financing arrangements and citizen attitudes towards governments.

The Pathway: Health Care Financing, the Experience of Risk, and Trust in Government

The literatures reviewed earlier suggest that the public/private mix of financing of health care may affect trust in political institutions indirectly through a multistep pathway: the extent of private financing of health care at the country level interacts with individuals’ market positions to produce varying risks for the individual of having unmet health care needs. This risk in turn influences individuals’ evaluations of government performance (in this case assessment of the functioning of the health care system), which finally affects their level of trust in government. This set of relationships is diagrammed schematically in figure 1.

Figure 1 Steps linking health care financing to trust in government

Drawing on our earlier discussion, the net effect of this pathway is likely to be that, at the country level, higher levels of private financing are associated with lower trust in government. But what happens in between to produce this negative association? Here we go through the logic of our argument in reverse, from the most proximate to the most distal relationship in the proposed pathway linking private financing of health care and political trust.

In line with the standard “performance” hypothesis in the literature on trust in government, we expect that a positive evaluation of the performance of the health care system is associated with higher trust in government institutions.

Next, building on the insight that individual experiences of risk affect attitudes towards social policy, citizen satisfaction with social welfare programs should be a function of the extent of insurance against risk that the program provides. The most important risk that a health care system insures against is having an unmet health care need—that is, needing medical care but not being able to afford or access it. As a result, respondents at high risk of having unmet health care needs should be less satisfied with the health care system. Both objective risk factors (such as low income) and perceived risk are expected to have this effect. Because private financing mechanisms may do less than public ones to buffer risks created by the activity of markets, a higher share of private financing should amplify the effects of risk on evaluation of the health care system.

The relationship between objective risk factors such as low income and subjective, perceived risk is likely to depend on a host of individual-level predispositions. Nevertheless, all other things being equal, we expect that respondents with lower household income will perceive a higher risk of unmet health care needs. However, and less trivially, we expect the objective risk generated by low income to create an amplified perception of risk in the context of more private health care health care financing. Hence, private financing of health care should moderate (in a statistical sense) the effect of household income on perceived risk. That is, low-income individuals should feel more at risk of not getting the health care they might need in countries with greater shares of private financing in their health care systems.

Taken together, these multiple steps along the pathway we propose suggest that in countries with higher levels of private health care financing, citizens will on average express lower levels of trust in government. This relationship should come about, we argue, because those citizens who are more exposed to market forces, by virtue of having low incomes in a more privatized system, will feel more at risk of having unmet health care needs and hence less satisfied with the performance of the health care system. In turn, these citizens should express less trust in their governments, reducing the average level of trust in government in countries with more private health care financing.

Evidence for the Pathway between Private Health Care Financing and Reduced Trust in Government

To evaluate the various steps in our argument, we bring together individual- and country-level data from multiple sources in a hierarchical linear model. Our data describe survey respondents nested in countries with health care systems characterized by varying levels of private health care provision. The small number of countries in the dataset does not allow for a great deal of statistical power when addressing hypotheses about country-level factors, but the incorporation of individual-level data nested within countries enables us to suggest the pathways through which private provision of health care may affect individual attitudes toward state institutions. The hierarchical model also allows us to adjust for differing average levels of trust across countries, and to study the conditioning effects of state-level variables on individual attitudes through cross-level interaction terms. All analyses were conducted in R 2.15 using the LME4 package. ^{Footnote 19}

The English-language versions of the question and response options, as well as descriptive statistics for all variables used in this analysis, are reported in the “Data and Methods” appendix. Our key outcome of interest, measured at the individual level, is trust in government, as indicated by the mean level of trust (ranging from 0 to 10) in five government institutions: the parliament, legal system, the police, politicians, and political parties (Cronbach’s alpha = .905). ^{Footnote 20} Data are from 48,867 respondents in twenty-five European countries to the 2008 wave of the European Social Survey. This composite variable is a commonly used measure of trust in political institutions in countries outside of the United States. ^{Footnote 21} It reflects a comprehensive view of trust in government rather than attitudes toward a single or a narrow set of state institutions, and has the further advantage of distinguishing political trust from evaluations of institutional performance or evaluations of government officials. ^{Footnote 22} Mean trust scores at the national level vary dramatically across different country groupings, with the postcommunist countries exhibiting the lowest values and the Nordic countries having the highest trust (see table 1 and appendix, figure A1.) For example, average trust was 1.7 in Ukraine and 6.6 in Denmark.

The key country-level feature of health policy whose association with trust in government we wish to examine is private financing of health care. In most of Europe, health care is provided by a complex mix of public and private actors: public and private hospitals and clinics, salaried health care providers working in government-run health facilities, groups of private practitioners working out of their own consultancies, self-employed professionals under contract to public health systems, and the like. By contrast, the share of private financing of health care is relatively obvious to individuals: private payments for health insurance premia, unofficial payments to doctors and hospitals in exchange for expedited care in nominally state-run systems, and out of pocket payments for medications and doctor visits are relatively visible to the end user. (In fact, co-payments are utilized in some settings precisely to make it plain to citizens that health care costs are partly their private responsibility, and hence discourage overuse.) Because private financing is more likely to be visible than private provision, we measure the “privateness” of health care systems in European countries as the private share of health care financing, rather than the public versus private nature of health care service provision. ^{Footnote 23}

We measure private financing as the 2006 measure of private sector expenditure on health as a percentage of total health expenditure in the WHO’s European Health for All Database (http://data.euro.who.int/hfadb/). ^{Footnote 24} Values for the countries in our study are presented in table 1. In most countries in Europe, general tax revenues or mandatory social insurance contributions collected and aggregated by quasi-governmental actors provide the majority of the funds that finance health care services. Private financing of health care accounts for between 13 percent (Czech Republic) and 45 percent (Bulgaria) of health care financing in European countries, and takes three main forms: premia for voluntary or obligatory private health insurance; cash payments by patients for goods and services not fully reimbursed by public or social insurance schemes (co-payments); and out-of-pocket payments for goods and services that fall outside of the insured system. Many of Europe’s health care systems have come to include more private financing since the 1980s. ^{Footnote 25,Footnote 26} Conjunctural fiscal pressures, combined with secular growth in the volume and cost of medical care, have encouraged governments to create a larger role in the financing mix for co-payments and out-of-pocket payments. ^{Footnote 27}

It could be argued that some types of private financing have a more meaningful effect on political trust than others, so it is a mistake to aggregate out-of-pocket payments and private insurance payments. It might be, for example, that the out-of-pocket component of private financing has a greater impact on attitudes than do private insurance payments, especially those made by employers on behalf of employees: out-of-pocket payments are likely more tangibly and directly linked to health care even if, in a purely economic sense, the two are equivalent as costs are ultimately pushed on to employees. We therefore examined the empirical breakdown of private funding between private insurance and out-of-pocket payments. In countries with higher levels of private financing, out-of-pocket spending is correspondingly higher. (Pearson’s r between out-of-pocket spending as a share of total health expenditure and private financing is very high, r = .87) This is in part due to the fact that for the vast majority of cases, private financing is predominantly covered out-of-pocket and private insurance plans typically account for a smaller or, frequently, negligible share of costs. Not surprisingly, analyses using the out-of-pocket measure produce virtually the same results as those using private financing.

Whatever type of private financing is dominant, and despite the recent convergence on a more significant role for private financing, however, health care systems in Europe still vary widely in the extent to which private insurance, co-payments, and out-of-pocket payments supplement government spending on health care. ^{Footnote 28} Part of this variation can be traced to different fundamental models of health care provision employed in different countries, which have implications for sources of financing. In countries with a national health service (NHS)—for example, the United Kingdom, Sweden, Portugal—general government revenues finance the bulk of health care, which is provided free or at a minimal cost to patients at the point of service. Social insurance health care systems, found in many continental and east-central European countries (e.g., Germany, France, Poland, Czech Republic), are financed mainly via mandatory contributions by employers and employees into state-regulated social insurance funds. Switzerland, the Netherlands, and Slovenia have private health insurance systems, in which citizens or their employers purchase insurance on (heavily regulated) private markets. Finally, a fourth group of countries relies heavily on out-of-pocket payments to supplement less robust government-run provision (Ukraine, Latvia) or insurance-based payment systems (Greece, Bulgaria, Russia). Broadly speaking, private financing is most prominent in those health care systems with highly developed private insurance markets, and where de facto out-of-pocket payments are necessary to guarantee access to care. However, there is substantial variation in the extent of private financing even among NHS and social insurance systems, which may not cover all services or medications.

Each of these different forms of private financing might be expected to have different effects on health care systems and patients’ experiences with them. However, in the most comprehensive study to date of the effects of different forms of private financing on health care systems and public opinion, Tuohy, Flood, and Stabile ^{Footnote 29} find no systematic relationship between the growth of different types of private financing (increases in a parallel private sector, increases in co-payments, reductions in coverage by the public system, and creeping privatization caused by shifts in care venues from largely public hospital systems to private outpatient facilities) and political support for the health care system. They do, however, find a consistent relationship between “changes in the public/private balance of health finance and political support for the public system” even across health systems with different types of private financing. ^{Footnote 30} For this reason, and because of the lack of precision in some breakdowns of different types of private financing, we operationalize private financing to include all spending on health care that comes directly from users rather than from general government revenues or earmarked payroll taxes.

To understand the relationship between health system characteristics and trust in government, we carried out a series of statistical analyses in which the outcome variables were each of the steps in the pathway we proposed in the previous section and depicted in figure 1. Before presenting the results of those analyses, it is valuable to examine these key variables descriptively.

One key measure of risk is income, which we expect to be associated with both perceived risk of having unmet health care needs, and health care system evaluation. We measure income using the self-reported adequacy of the respondent’s household income. The responses range from “living comfortably on present income” to “finding it difficult on present income” on a 4-point scale. As figure 2 shows, the proportion of individuals who find it “very difficult” to make ends meet varies substantially in the 25 European countries in the study. We use this subjective measure of income adequacy in part because it contained much more complete data than the numerical household income measure in the ESS survey. It is also valuable because it captures not only the objective market position of a household but also the subjective feeling of risk that may be associated with it. ^{Footnote 31}

Figure 2 Variation in the proportion of respondents reporting “Very Difficult” circumstances in living on their present income in 25 European countries

Source: ESS (2008).

The amount of risk that respondents reported feeling in the context of their health care system was captured by responses to the question “During the next 12 months, how likely is it that you will not receive health care you really need if you become ill?” Responses to this question about the risk of having unmet health care needs ranged from 1 (not at all likely) to 4 (very likely). ^{Footnote 32} Approximately 25 percent of all respondents reported that they were likely or very likely to not receive needed medical care in the coming year. Average perceptions of health care risk tend to be higher in the postcommunist countries. Ukrainians present the bleakest assessments of their risk, with an average response of 3 on this 4-point scale.

Finally, citizens’ evaluations of the performance of their country’s health care system ranges from “extremely bad” (0) to “extremely good” (10). In general, respondents in the postcommunist countries reported worse opinions of their health care systems, with the lowest value in Ukraine (2.4). More favorable assessments were reported in western European countries, with Belgian respondents having the highest average score (7.4). Respondents in Norway, Sweden, the United Kingdom, France, Spain, the Netherlands, Finland, and Switzerland also gave relatively high marks to their health systems, with average values of 6 or more on the 11-point scale. Among western European countries, Greek respondents reported the least favorable assessment of their health care system (3.3), with only Ukraine and Bulgaria in the full sample exhibiting lower average health care system performance scores (refer to table 1).

Our analyses also controlled for factors at the country level aside from health care system financing that are likely to influence a citizen’s trust in government. The small number of country units (twenty-five) limited our ability to control for all factors at the level of the polity that might be associated with trust in government. Hence, we include indicators for only the most important country-level variables that could, if omitted, bias the estimates of the relationship between private health care financing and trust in government.

One key control variable at the country level is for welfare state type. We used Castles and Obinger’s ^{Footnote 33} groupings, which were based on multidimensional cluster scores summarizing distinctive patterns of public policy and socioeconomic arrangements circa 2000–2005. These reflect, crucially, historical levels of trust in government and income inequality, which Rothstein and Uslaner ^{Footnote 34} argue affect both subsequent welfare state organization and trust in government. ^{Footnote 35} Including indicators for Nordic, English, and Postcommunist welfare families hence allows us to control at least partially for the effect on trust in government of long-standing patterns of societal and welfare state organization. ^{Footnote 36} In the final models we included dummy variables for Nordic and Postcommunist welfare state families, which had statistically significant effects on either trust or other important parameters in exploratory modeling.

We also maximize statistical power by allowing welfare family groupings to stand in for other variables related more generally to state and economic capacity with which welfare families were highly correlated. These included per capita GDP, state capacity, government effectiveness, prevalence of corruption, and per capita spending on health care, all of which are highly correlated with each other and with membership in the postcommunist welfare state family (refer to the appendix, table A2).

In preliminary analyses we also included as controls measures of income inequality, health care expenditures as a percentage of total government spending, and ethnic and cultural fractionalization, ^{Footnote 37} which may be associated with poor public goods provision. ^{Footnote 38} All of these variables were discarded after showing no significant associations with average level of trust in government in multivariate analysis at the country level.

At the individual level, we controlled for factors that may have a direct impact on trust in government: interpersonal trust and overall life satisfaction; ^{Footnote 39} interest in politics; assessment of the state of the economy; membership in a group that is subject to discrimination; and whether the respondent voted in the last election for a party that is included in the current government. ^{Footnote 40} To construct this last measure, we cross-referenced the respondents’ reported closeness to and vote for a given party with the composition of governing coalitions at the time of the administration of the survey. Individuals who favor state intervention may also be more likely to trust political institutions in general. To explore the possibility that this kind of ideological orientation colors experiences with the health care system, we also include a variable Pro-state Ideology that captures respondents’ views on the appropriate role of the government in the economy. ^{Footnote 41}

Another important set of controls is for factors that would be likely to affect individuals’ interactions with the health care system apart from the privateness of health system financing. Individuals in ill health are likely to use health care services more frequently. Self-reported health status (the variable labeled Sick) is measured in the ESS on an ordinal scale ranging from 1 (very good) to 5 (very bad). ^{Footnote 42} Similarly, utilization of health care varies across individuals, with age, gender, educational attainment, and employment status all affecting the likelihood of receiving care even after controlling for medical need, and even in countries with high levels of health equity. ^{Footnote 43} Hence, we included covariates tapping these social determinants of health care utilization in our initial analyses. ^{Footnote 44} Measures of educational attainment, unemployment, and gender were excluded from the final model as their association with trust was not statistically significant and their exclusion did not affect the size or significance of other coefficients. ^{Footnote 45}

Results of Multilevel Modeling Confirm Link between Private Financing of Health Care and Lower Trust in Government

Having assembled our multi-level dataset, what did we find? To preview, the key result is that inadequate household income and perceived risk of having unmet health care needs are both associated with less positive evaluations of the health care system, and these relationships are heightened in the context of private health care financing. This negative assessment of the health care system, as expected, reduces trust in government. The discussion that follows proceeds in several stages in order to trace the multiple steps along the pathway linking private financing of health care and trust in government. In all of our models, the vast majority (93–94 percent) of the variance in expressed trust in government is explained by variables at the individual level, which is not surprising for a model of individual-level opinion within a relatively similar set of countries. In the model predicting health care system evaluations (Model 4), however, the portion of the variance explained by country-level factors increases to about 11 percent, indicating systematic differences across national health care systems.

As a first step in the analysis, we ran models predicting trust in government using a variety of national- and individual-level covariates. Table 2 shows the results of our first model, with trust in government as the dependent variable, and individual-level covariates. Because the individuals are clustered within countries, we use panel-corrected standard errors, but we do not otherwise account for differences across country contexts. For the most part, the results are as expected from existing studies of trust in government. Respondents who are happier, more trusting of others, more satisfied with the state of the economy, and more interested in politics display high levels of trust in government institutions. Conversely, we find that the 6 percent of the overall sample who identify as members of a group subject to discrimination, and individuals in poor health, report less trust in government. Without further specification of the country context within which individuals are situated, we do not detect a strong effect on trust in government of a pro-state orientation. Perceptions of the adequacy of household income have a statistically significant effect on political trust in the single-level model: a one-unit increase in self-reported income sufficiency (on the 4-point scale) was associated more than a 0.2-point rise in trust. However, as the next model shows, the coefficient on income is no longer statistically significant once country-level variables are introduced, suggesting that the effects of household income on trust in government are also contingent on nationally-specific factors.

Table 2 OLS model predicting trust in government

* p<.05; **p<.01; ***p<.001.

Panel-corrected standard errors reported below coefficients.

Table 3 reports standardized coefficients from multilevel analyses. Model 2 uses the same individual-level variables as in Model 1, but now in a hierarchical context. The coefficients we present are standardized regression coefficients (betas), scaled using grand mean centering. (Unscaled coefficients are reported in the appendix, table A3.) The model also now includes the two key indicator variables at the country level: for private financing of health care, and for welfare state families (which, as explained earlier, proxy for variation in a number of country-level factors likely to affect trust in government). After controlling for these country-level factors, the effects of the individual-level variables are consistent with the results in the single-level Model 1, with the exception of income as noted above. At the country level, citizens in postcommunist welfare states exhibit low levels of trust in government, as expected. Consistent with the arguments we have presented, our private financing variable is negatively associated with trust in government, and statistically significant despite the small number of country units (p ≤ .05, N=25). However, this result does not hold when one or more countries are excluded. For example, when we dropped Bulgaria and Ukraine from the analyses, private financing is no longer a significant predictor of trust in government. ^{Footnote 46} Alternatively, excluding Switzerland, a country with both high private financing and trust, yields a stronger association between these two variables. The low degrees of freedom at the country level probably account for the volatility of the relationship between private health care financing and trust.

Table 3 Standardized coefficients, multilevel models of trust, risk, and health system evaluation

* p<.05; **p<.01; ***p<.001. Standard errors reported below coefficients.

Although private financing may not exert a strong direct effect on trust in government, the pathway we show in figure 1 suggests that it may condition the impact of other factors on this outcome. To probe the plausibility of these interactions, we next modeled the effects of private financing on perceived risk, as indicated by the respondent’s concern that her medical needs will not be met, and on her evaluation of the health care system (refer to table 3).

In Model 3, we examine the effects of income on perceived risk of unmet health care needs in the context of different levels of private financing. Respondents in postcommunist countries perceive higher levels of health care risk than those in other countries. Similarly, the coefficients on individual-level covariates predicting perceived risk are generally consistent with expectations. Lower-income individuals express higher levels of perceived risk (H5), with private financing amplifying the association between low income and perceived risk (H6).

Figure 3 illustrates this conditional relationship graphically. It depicts the results of simulations showing the substantive effects of income on perceived risk of having unmet health care needs at different levels of private financing, and the uncertainty associated with these estimates. The perceived risk gap between low and high income individuals increases with greater levels of private financing.

Figure 3 Perceived risk of having unmet health care needs, by income, and at varying levels of private health care financing

Next we explored the plausibility of the second step in the pathway by assessing how private financing affects evaluation of the health care system’s performance (Model 4). Previous studies ^{Footnote 47} have generated mixed findings regarding the relationship between private financing and satisfaction with health care systems in European countries. In our model, the country-level private financing variable is negatively associated with health care system evaluations, and statistically significant (p <.05), despite the small number of country units.

Due to the presence of two cross-level interaction terms, with income and perceived risk, it is not possible to directly interpret the magnitude of the coefficient of private financing. Since risk is a component of the interaction term with private financing, interpretation of its effect must include both the main and interactive effect with private financing. We can simulate the substantive effects of perceived risk on health care system evaluations at different levels of private financing. Figure 4 depicts the expected values of health care system evaluations from simulation draws based on Model 4 with 95 percent confidence intervals, varying the level of financing across the observed range at maximum and minimum values of risk. All other variables were held constant at their mean.

Figure 4 The association between risk and health care system evaluations at varying levels of private health care financing

The effect in these simulations of perceived risk is to depress health care system evaluations everywhere; higher levels of perceived risk are associated with less favorable assessments of the health care system. However, as the share of private financing increases, health care system evaluations across different levels of risk converge due to the steep decline in health system satisfaction in the lower risk category.

As a final step in the analysis of the causal pathway, Model 5 returns to the relationship between private financing of health care and trust in government, now incorporating the hypothesized pathway linking the two variables by including our perceived risk and performance evaluation variables. We expected private financing to mediate the association between objective risk factors and performance evaluations, on the one hand, and trust in government, on the other hand. We found a weakly significant direct negative relationship between private financing and performance evaluations in Model 4, but the strong effects of the cross-level interaction term in Model 3 suggests that a more indirect effect is also plausible. In Model 5, we find that health care system evaluations are a key driver of trust in government (H2). Performance evaluations have a large positive association with trust in government—an effect size second only to that of satisfaction with the economy and even greater than that of interpersonal trust. As the standardized coefficients indicate, perceived risk of not receiving needed health care in the future has a weaker effect on trust, which conforms with our expectation that the relationship between perceived risk and trust in government is mediated by evaluations of the performance of health systems. The potential effect of private financing thus appears to operate through performance evaluations and risk perceptions.

What Our Results Do and Do Not Show

Our findings are consistent with the hypothesis that private health care financing results in decreased trust in government, and suggest that it does so through a multistep pathway. Lower income individuals perceive themselves to be more at risk of not being able to access medical care when they need it, which in turn depresses their evaluations of their country’s health care system. To be sure, other demographic characteristics also affect perceptions of health care risk: older and sicker Europeans, those living in the postcommunist east, and those who belong to groups that experience discrimination perceive themselves to be most at risk of unmet health care needs. But low income is an important predictor of risk; and a large share of private health care financing in the national health care budget amplifies the effects of perceived risk, producing even greater dissatisfaction with the health care system. In turn, poor health system evaluations reduce trust in government institutions (consistent with theories that locate the source of political trust in citizens’ assessments of the performance of government institutions).

Could it be that we have the causal ordering wrong? That is, are countries in which citizens have low trust in government and more negative evaluations of their health care systems also more likely to have higher private financing, perhaps because their citizens trust markets more than the state to engage in social provision? Our empirical strategy suggests that this is unlikely. We start with individual-level factors such as income, health status, and minority status and then trace a relationship through risk, satisfaction with the health care system, and eventually trust. Unless it is the case that trust causes income differences, sickness levels, or minority status, this progression of the argument ensures that trust is for the most part exogenous.

The contrast between two country cases in our analyses illustrates this point. If low trust in government led to greater private financing, then we would predict that in Switzerland, with its high levels of private health care financing, citizens would have, on average, high perceived risk, low satisfaction with the health care system, and low trust in government. But none of this is true: Switzerland has one of the highest mean levels of health care system satisfaction, lowest perceived risk levels, and highest mean trust levels. One reason for this is that Switzerland has a higher proportion of economically secure individuals who have less need for a social safety net than in other countries. Not coincidentally, it also has the lowest percentage of sick people of any country in our sample of 25 European countries. What this indicates is that the general relationship between financing and trust is conditional on the distribution of risk in the country. In Switzerland, trust is very high, even though financing is highly privatized, because there are very few people whose exposure to health care markets makes them feel at risk of having insufficient income or unmet health care needs. In Bulgaria and Ukraine, however, which have almost exactly the same level of private health care financing as Switzerland, citizens have poor evaluations of their health care system and low levels of political trust, in part because the average citizen faces much higher objective levels of risk. Figure 5 illustrates the contrast between Switzerland, on the one hand, and Bulgaria and Ukraine, on the other hand.

Figure 5 Private financing and mean levels of risk of Have Unmet Health Care Needs across income categories in Bulgaria versus Switzerland

Source: ESS (2008).

As the figure shows, despite being exposed to similar levels of private financing in their respective health systems, more Bulgarians on average report higher levels of risk of having unmet health care needs at every level of perceived income adequacy than their Swiss counterparts. This contrast underscores the point that risk is in part driven by individual-level factors, such as income, and therefore it very much depends on the distribution of income in a given country (refer to figure 2). Much larger portions of the population in Bulgaria and Ukraine experience difficulty making ends meet than in Switzerland. Income distributions shape levels of risk, which in turn affect trust.

Another reason we are confident that we have the direction of causality right concerns the relationship between citizen trust in government and preferences for statist policies. If health policies (including the level of private financing) were shaped mainly by pressure from below, such that low trust in government led to more limited government activity, Eastern European countries would have highly private healthcare systems. These systems do have high levels of private financing of health care. But the main reason for this is not bottom-up pressure from citizens, but rather the exigencies of fiscal policy and, often, pro-market pressure from international financial institutions. ^{Footnote 48} Support for more redistribution and statist policies is in fact generally higher in countries like Bulgaria and much of the rest of Eastern Europe than in Western Europe (refer to figure A2 in the online appendix). But the main point here is that the link between citizen preferences and policy outcomes is not clear-cut: many factors—e.g., aggregation and representation of public interests, collective action problems, policy elites, interest group pressure, fiscal constraints, pressure from external actors like international financial institutions—stand between individual policy preferences and policy outcomes. ^{Footnote 49,Footnote 50}

The structure of our data does, however, impose some important limitations on our analysis that are worth mentioning. First, we are unable to apply formal tests of mediation to the proposed pathways linking private health care financing with decreased trust in government. Such tests would be possible in a multilevel setting only if we assigned arbitrary values to dichotomize our treatment variable, Private Financing. ^{Footnote 51} Since we are primarily interested in the effect of risk and income at various levels of private financing, we chose the more defensible alternative of modeling each step in the causal chain separately, using cross-level interaction terms. Further, we followed recommended practice and used simulation based on Markov chain Monte Carlo (MCMC) draws from the posterior distribution to establish statistical significance. ^{Footnote 52}

A further limitation of our cross-sectional data is that we cannot make inferences about the effects on trust in government of changes in the composition of private health financing. Our data are about the consequences of privateness, not privatization, and may lead to faulty inferences about the effect of shifts in private financing generated by policy changes like the introduction of the ACA. This is an inherent problem with using cross-sectional analysis to think about the consequences of policy change: One runs the danger of reading the future sideways, to paraphrase Arland Thornton’s ^{Footnote 53} memorable formulation.

Despite these limitations, however, we believe that our findings are useful in advancing knowledge about public opinion regarding social policy provision, and about the sources of political trust. Our claims and empirical findings add to debates about the determinants of trust in government by linking political trust to citizen experiences of the welfare state, and in particular the way that the welfare state conditions risk. Previous research has found that the structure and performance of welfare state institutions may promote social trust. ^{Footnote 54} Similarly, an established literature has shown that government performance is an important predictor of citizens’ political trust. Our findings confirm this relationship, while highlighting linkages between trust in government and evaluation of the health system, a relationship that has received relatively little attention in research on political trust. ^{Footnote 55} The pathway that we identify linking private financing to political trust via perceived risk also suggests that trust may have a partly prospective basis. In formulating their opinions about their governments, citizens not only look to their past interactions with welfare state institutions and lived socioeconomic experiences, but also draw on hopes and expectations about which actors will best enable them to meet their future health care needs.

Neoliberal Politics and Political Trust in Comparative Perspective

What does this analysis tell us about how citizens are likely to respond to ongoing reforms in health care systems in Europe and the United States? In recent decades, health care systems in Europe and in much of the world have become increasingly privatized, and the economic crisis has amplified this trend. ^{Footnote 56} Our analyses suggest that the growing privatization of health care in Europe is likely to have political ramifications. When states do less to ensure the basic health care needs of their populations, especially the needs of members of society who are at greatest risk, citizens may come to place less trust in government institutions.

Conversely, in the United States, the Affordable Care Act has expanded the role of government, both state and federal, in regulating, administering, and financing health care. The reform occurred in part because a winning coalition of political actors came to believe that the dominance of private health insurance provision in the U.S. health care mix had led to unacceptable levels of risk for too many citizens. In order to pass and to implement the reform, the Obama administration has had to overcome a preexisting burden of public distrust in government ^{Footnote 57} —a distrust that has been actively cultivated by political actors keen to further reduce the role of government. Can an expansion of the government’s role in health care contribute to decreasing this reservoir of distrust?

Our analysis of data from European countries does not reveal any tendency for high-income individuals or those who perceive little risk of having unmet health needs to be less trusting in countries where the government’s role in health care financing is larger. In other words, while public financing of health care is associated with higher levels of trust among low-income individuals, it is not associated with lower levels of trust among high-income individuals. So expanding the state’s role in health care financing via the ACA could increase trust in government among those who benefit most from the larger public role, without risking alienating higher-income citizens.

On the other hand, there are reasons to be cautious about generalizing these results. More privileged Americans may differ from their European counterparts in important ways. For example, Hetherington ^{Footnote 58} argues that in the United States, higher-income people, who stand to lose from redistributive policies, are more likely to express distrust in government when it attempts to undertake such policies. Furthermore, while the ESS data show that overwhelming majorities of European citizens, including upper-income ones, agree that it is mostly or entirely the role of government to ensure adequate health care for the sick, the most recent Gallup poll found that less than half of Americans thought the federal government ought to be responsible for making sure that all Americans have health care coverage. ^{Footnote 59} There may, then, be an upper limit to the amount of trust that a reform such as the ACA can foster in the United States, given overall preferences for a more limited role of government.

Perhaps more importantly, if the implementation phase of the reform leaves significant populations exposed to the risk of unmet health care needs, a substantial “trust dividend” is unlikely to ensue from the passage of the ACA. For example, low-income citizens living in states that have declined to expand access to Medicaid, and individuals and families for whom the federal subsidies are too low to make the purchase of insurance on the exchanges affordable, may well continue to distrust government, at a minimum at the state level. However, our findings suggest that if the reform can make citizens who were previously exposed to a great deal of risk feel more secure, this may indeed increase trust in government, which in turn may make further reform possible.

The ACA is the largest expansion of the welfare state in the United States in decades. It signals a shift from a largely private system of health insurance for the working-age population to one with significant public regulations and subsidies, albeit largely using private insurance. Its implementation is an opportunity to study a number of key questions in political science, including the ramifications of welfare policies for trust and distrust in government, as we have done. With twenty-five Republican-controlled states at least initially refusing to participate in the Medicaid expansion allowed under the ACA, the situation in the United States is in some ways more akin to the nations of Europe than it is to a single nation. This offers an opportunity to replicate our study more directly, by comparing trust in government and support for public and private health care programs across the states.

European countries have embarked on a policy experiment moving in the opposite direction to that of the United States, moving from public to more privatized forms of health care. European governments should also take note: they risk further eroding the public’s trust if they do not at the same time work to reduce income inequality and exposure to market risks among the most vulnerable members of their societies.

This brings us to our final point. For thirty years, beginning with the widely circulated report for the Trilateral Commission by Michel Crozier, Samuel Huntington and Joji Watanaki, ^{Footnote 60} political scientists have worried about an apparent secular decline in trust in government in the rich democracies. ^{Footnote 61} The main storyline that has emerged to explain this decline in trust centers on the declining capacity of the state to deliver to “system outputs,” ^{Footnote 62} including public social goods, deemed satisfactory by its citizens. ^{Footnote 63} A number of theories have been offered to explain these increasingly unsatisfactory outputs. For example, the “overload” theorists of the 1970s ^{Footnote 64} attributed poor performance to activist governments trying to do too much, while social capital theorists argued that a decline in social trust undermines institutional performance. ^{Footnote 65}

We think that this main storyline pays too little attention to the context within which the decline in trust has emerged. This context is characterized, above all, by the gradual acceptance throughout the rich industrialized democracies of a neoliberal economic paradigm calling for a reduced role of the state in guaranteeing citizens protection from market risks. This has resulted in growing income and wealth inequalities in all of these countries, generating a concomitant growth of feelings of insecurity among a larger share of the population.

To be clear, we are not arguing that the withdrawal of the state from social provision necessarily causes trust in government to fall. Nor are we arguing that inequality per se results in declining political trust. ^{Footnote 66} What we are saying is that the effect of private financing of public social goods on trust in government is conditional on the larger environment—which, crucially, includes how risks are distributed in society. And when social risks are distributed unequally, more private provision of social goods may indeed reduce citizens’ trust in government. If the social context is one of widespread abundance and security, even wholesale privatization of social service provision may do little to hurt citizens’ relationships with their governments. But where neoliberal policies and increasing inequality have made larger numbers of citizens more economically insecure, further withdrawal of the state is likely to promote distrust in government.

Political scientists are at our most comfortable when we are focused on “getting the institutions right,” whether presidential or parliamentary systems, ^{Footnote 67} the rule of law, ^{Footnote 68} or political institutions to help politicians “get to yes.” ^{Footnote 69} This study does not support a confident institutional recommendation. In fact, it leads us to think that that this approach may be limiting, because it focuses our attention in the wrong place. We began this project looking for a story about a relationship between institutions for social provision and trust in government. And indeed, we could recommend that policy makers expand public social provision to increase political trust. It wouldn’t hurt, even in countries like Switzerland; and in many places it would help. But the real policy lesson, we think, is much more complicated. Neoliberal economic policies, the rise of inequality and precariousness, and the decline of political trust in rich industrialized democracies are deeply intertwined. And what policy makers must hear from us is that they jeopardize faith in their own governments if they fail to address the root causes of the growing inequalities that expose ever-larger numbers of their citizens to insecurity and risk.