Private health insurance and redistribution preferences in the Austrian health care system: A case of mere self-interest?

Background: Private health insurance (PHI) is a relevant nancing source in many health care systems, including the Austrian one where more than a third of the population possess such an insurance. The effect of PHI on redistribution preferences in health care has been an understudied phenomenon. This article examines such a relationship. Further the mediating effect of fairness attitudes within said relationship is analysed. Methods: The data for this study come from the Austrian Corona Panel Project (ACPP). OLS regression models are employed to elicit the links between the variables. Results: In this article evidence for considerable self-interest in support for redistribution in health care with matching fairness attitudes is found. Additionally, social redistribution preferences for benets and taxation, political party preferences and pandemic risk perception play a role. Conclusion: Health policy makers should carefully consider these relationships when adapting the public health care system to challenges, such as increased nancing pressure or a pandemic. This could be done by incorporating vulnerable groups and public preferences at large in an open debate about the design of the health care system, its nancing and its consequences for redistribution. The datasets generated during and/or analysed during the current study are available in the AUSSDA repository,


Introduction
Private health insurance (PHI) [1] plays a signi cant role in health policy and in many health care systems, I suggest using individual survey data to explore the effect of PHI on support for redistribution in health care. As Austria has statutory health insurance (SHI), yet a high degree of PHI, it is an ideal case to study said effects.
Insurance is a dilemma in health care rationing and priority setting because health care needs are distributed very heterogeneously across society and because such needs very often arise unexpectedly in brute luck (Brock 2007). While this dilemma has been addressed in theory, empirical research has so far not considered what people with PHI and people without PHI actually want in terms of redistribution in their health care systems. Thus, the questions asked in this study are: How do the health care redistribution preferences of people with PHI differ from those of people without PHI? And do these persons consider treatment prioritisation of privately insured as fair?
Existing Research

Redistribution in health care
Previous research has shown that possession of PHI seems to decrease the political support for the public health care systems in nations with private duplicate health insurance systems, yet not in countries with private primary health insurance (Mou 2013). Pfarr and Schmid (2016, p. 624.) nd in a German sample that "Persons who bene t from public coverage exhibit a positive willingness to pay for an extension of the coverage beyond the status quo. The others are not willing to contribute to this end". Jensen and Naumann (2016) nd different levels of support for public health care between demographic groups and between people leaning toward the political left or right in 17 European countries. Further research indicates that income affects willingness-to-pay for public health care in 29 countries included in the ISSP (International Social Survey Programme; Azar et al. 2018) and that perceptions of inequities in health care impact preferences for government provision of health insurance in the USA (Lynch and Gollust 2010). Age also matters for the preferences towards health maximisation and for egalitarian concerns in health care in a Polish study (Kolasa and Lewandowski 2015). Akkazieva et al. (2006) investigate health policy preferences of patients in Hungary. They nd that patients primarily have a preference for solidary measures in health care. Read et al. (2021) analyse an English sample and demonstrate that the majority of respondents prefer social care in later life to be paid for by the user and the state in a shared way. Lu et al. (2021) show that the UK public seems to prefer a collective approach to raising additional funds for health and social care by a public organisation. Soroka and Lim (2003) nd for the USA and UK that issue framing de nes the responsiveness of health care expenditures to public preferences. Wendt et al. (2010) compare 14 European countries and nd that higher income is associated with less desire for state involvement in health care in all studied countries, except for the Netherlands. Henderson and Hillygus (2011) show that self-interest and partisan attachments strongly affect health care attitudes in the USA. In an Icelandic sample the majority of respondents favour higher health care expenditure and health provision by the state (Vilhjalmsson 2016). Choma et al. (2018) nd that in the USA preferences for the distributive fairness perceptions of equality and need in health care mediate the relationship between political orientation and attitudes on the ACA (Affordable Care Act).
Further, personal experiences with the health care system shape attitudes towards health care policies in many countries (Larsen 2020) and in the USA speci cally (Lerman and McCabe 2017). Chavanne (2020) nds for an US sample that support for redistributive health care increases as bad luck becomes more important in causing poverty yet is unaffected as good luck becomes more important in causing wealth. Kullberg et al. (2021) do not nd evidence that voluntary health insurance, i.e. PHI, changes the willingness to pay higher taxes for public health care in Sweden. Yet, the authors do not consider redistribution in Swedish health care from a wider angle that would include anything but readiness to contribute more personally. Heap et al. (2020) demonstrate that people in the UK and USA highly prefer health over wealth, when being presented with a trade-off between lives lost to COVID-19 and disposable income losses. Barry et al. (2020) show that in the early stage of the COVID-19 pandemic the majority of US adults supported social and health policies, such as paid sick leave, universal health insurance, an increased minimum wage, as well as various unemployment support policies.
As redistribution in health care is expected to increase with public spending on public health care, health care expenditure and nance is a key part of health policy. Research with Danish survey data shows that respondents exhibit a willingness to improve the quality of the public health care system at the expense of private consumption (Gyrd-Hansen and Slothuus 2002). Empirical evidence suggests that Australians mostly care about outcome egalitarianism and not about cost per life year in their preferences for the allocation of the health budget (Richardson et al. 2012). Research with a Greek sample nds that preferences for public health expenditure allocation have more support among relatively poor citizens (Xes ngi et al. 2016). Foremmy et al. (2020) nd that preferences for health care expenditure almost doubled in the COVID-19 pandemic in Spain compared to a 2018 benchmark and that Spanish citizens would on average allocate 22.5% of the public budget to the health sector, which is considerably higher than the 15% of real public spending on health in Spain.
One debate relevant to PHI deals with two-tier medicine. "Today, that prevailing question focuses rather on what society could do for its citizens in optimizing their lives. Such an approach amounts to loading the dice in favor of redistribution" (Breyer and Kliemt 2015, p. 149). Further: "The concept of solidarity provided for a long time the ideological rationale for the transfer of nancial contributions by individuals by way of compulsory health insurance schemes or national taxation to a universal health care system that should guarantee equal access to health care for those who are in need" (Ter Meulen and Jotterand 2008, p. 191). While health care has functioned averagely well in Austria in OECD comparison (Tchouaket et al. 2012), increasing pressures towards two-tier medicine threaten such proper functioning of health care systems globally (cf. Ter Meulen and Jotterand 2008; Breyer and Kliemt 2015). Results for health care preferences in Zambia, which de-facto has a two-tier system, show that socioeconomic status has an effect on various preferences for health care quality attributes (Hanson et al. 2005). Shickle (1997) investigates public preferences for health care prioritising in the UK and develops arguments for and against certain criteria of priority setting, wherein health care professionals, managers and the general public partly exhibit different preferences. Barbosa and Cookson (2019) nd for Brazil that possession of private health insurance coverage and residing in an urban location both contribute more to health inequity than income does. Overall, PHI seems to increase health inequity in the EU (Mossialos and Thomson 2002), as failures in private health insurance markets (such as moral hazard, monopoly, adverse selection and risk selection) can result in inequality and ine ciency (Powers and Faden 2006; Thomson et al. 2020).

Fairness in prioritisation
When trying to employ purely rationality-based economics to medical priority setting one soon runs out of road (cf. Robinson 1999 Ryen et al. 2019). In the context of the COVID-19 pandemic some self-interest also seems to be present (Grover et al. 2020). To investigate treatment prioritisation of people based on their health insurance status, i.e. possession of PHI, opens a gap in the literature. What has been done so far is research on income-related health care inequalities, where ndings of 23 countries suggest that the majority of people regard income-based health privileges as unfair, which is particularly true for persons with insu cient health insurance coverage (Von dem Knesebeck et al. 2016). Moreover, lower cost barriers in health care and higher public health care spending correlate with higher perceived unfairness of health inequalities in an investigation of 28 OECD countries (Immergut and Schneider 2020).
As for priority setting preferences, evidence from Germany shows that the public only considers effectiveness without the cost component in determining whether health care insurance should pay for treatment (Diederich and Salzmann 2015). A Swedish investigation shows that citizens think the public health services should always offer the best possible care irrespective of cost, while doctors support this approach less strongly. When it comes to health care nancing, Swedish policy makers favour higher taxes, whereas doctors prefer complementary private insurance and increased patient fees (Rosén and Karlberg 2002). Also addressing the nancial aspect of prioritisation, Damm et al. (2014) show that the economic-based cost-effectiveness ratio (CER) is supported by advanced economics students signi cantly more than by law, philosophy or medicine majors in Germany. Mirelman et al. (2012) demonstrate that there are priority setting preferences for e ciency criteria over equity criteria among policy makers in four countries (Norway, Nepal, Brazil, Uganda) but not in Cuba. Thus, it appears that preferences for medical prioritisation depend on context, i.e. where and whom one asks (cf. Orton et al. 2011). Overall, there is evidence that the general public emphasises health equity (e.g. Reckers-Droog et al. 2018), while medical decision makers include e ciency and cost-effectiveness criteria more strongly (e.g. Tanios et al. 2013). However, to explicitly consider monetary contribution in prioritisation decisions is rejected by both medical professionals and laypersons, as also evidence from Switzerland shows (Krütli et al. 2016).
Theoretical Background Solidarity, equity, fairness and reciprocity in health care Prainsack and Buyx (2011; argue that solidarity is the acceptance by a person or by a group to carry nancial, practical and/or emotional costs to support others with whom they recognise similarity in a relevant aspect. This similarity can be a common goal, shared risk or an objective, as well as subjective, characteristic. Particularly, in health care such solidarity comes in three tiers: It can be interpersonal, communal on the group level or contractually incorporated into institutions. Solidarity in public health care systems with SHI is a prime example of institutionalised solidarity because citizens contribute based on their wages and salaries and receive medical treatment and care that correspond to their needs. Solidarity in an ideal state can be seen as the outcome of a 'social contract', in which personal and ethical allocation preferences of a collective are reconciled through impartial consideration, i.e. under the veil of ignorance (cf. Dolan et al. 2003). Olsen (2011) provides an account of equity in health care. He suggests that in countries with publicly funded health care two principles are paramount: Equal access for equal need and reduction of health inequalities. While the former represents the ideal of universalist welfare, the latter recognises that some health inequalities are perceived to be inherently unfair. Daniels (2001) argues that there is a social obligation to guarantee the existence of institutions, such as social insurance or subsidies for PHI, that protect equal opportunity in a rebuke to e ciency maximisation, aggregate welfare maximisation and utilitarian objectives. Yet, equity and inequity in health care remain to be normative concepts. First, as Anand (2002) argues, inequality aversion differs between different domains in health care and such an aversion ought to be higher across some population groups than others. Second, equity and equality are not the same, even though health equity incorporates the idea of equal opportunity to be healthy. Third, health equity cannot be directly measured but merely be grasped "... as the absence of disparities in health (and in its key social determinants) that are systematically associated with social advantage/disadvantage" (Braveman and Gruskin 2003, p. 256).
Fairness denotes the "... view that some important types of health inequalities are considered unfair" (Olsen 2011, p. 815). Therein, fairness also remains normative. Brock (2002) points out that 'fair' treatment priority can be given to the worst off, according to bene ts aggregation or so that fair chances can be pinned against best possible outcomes. All of these, and more, may seem fair to certain people in certain circumstances. Daniels (2007) advocates three core concerns for fairness in health that are equity, e ciency, and accountability, that should be ful lled to meet central criteria of fairness in health policy.
Reciprocity can be direct, generalised (indirect), strong, weak, conditional, unconditional, negative, and/or positive. Generally, "... a reciprocal action is modeled as the behavioral response to an action that is perceived as either kind or unkind" (Falk and Fischbacher 2006, p. 294). Framed differently, "Reciprocity demands proportionate balancing of the bene ts and burdens of social cooperation between participants…" (Viens 2008, p. 1). I argue that in its essence reciprocity should be evaluated insofar it can or cannot alleviate the 'tragedy of the commons' and whether it can establish a social norm in a given context. A scenario with identical individual incentives but absent cooperation is known as the tragedy of the commons (e.g. Gächter et al. 2017). Gross and De Dreu (2019) demonstrate that attenuation of said tragedy depends on the costs of individual solutions. If they are low, cooperation deteriorates, even if individual negative reciprocity (punishment) is available. However, centralised punishment institutions or establishment of social norms of cooperation and of generalised reciprocity [2] can solve the dilemma.
Thus, when individualism is reasonably cheap, cooperation depends on a type of governmental enforcement and/or on the establishment of a social norm that facilitates cooperation. [3] Solidarity is a collectivised practice in public health care systems with SHI. Equity in health care encompasses an ideal to aspire to and to work towards and fairness involves the individual and normative attitudes on redistribution dilemmas and decisions about allocation of scarce medical resources. In theory reciprocity, in its generalised form, can amplify or replace continuous state monitoring of and intervention in health insurance markets. However, the state often may be a required institution to generate and stabilise such generalised reciprocity through governmental constraints on resource consumption (cf. Perceived fairness: Beyond self-interest "Few individuals in the developed world will be able to afford the health care they will need in their lifetimes" (Latham 2012, p. 130). Thus some kind of health insurance is necessary to guarantee basic treatment and minimum standard of equity in health care. The fairness of possible health insurance schemes is closely tied to issues of distributive justice and consequently also relates to how health care is nanced (cf. Clark and Weale 2012).
Context has been shown to signi cantly affect which distributive justice principles people apply, i.e.
people exhibit different justice attitudes in different situations (Liebig and Sauer 2016). Van Hootegem et al. (2020) show that in health, equality is the most popular justice principle in a Belgian sample, but not when it comes to unemployment bene ts or pensions. [4] People have different justice attitudes within the health care domain as well, when high resource scarcity is pinned against medium or low resource scarcity in health-health trade-offs (cf. Scott and Bornstein 2009).
To alleviate the dilemma of scarcity between health and wealth and within health people need to form some kind of bond. They need to nd an equilibrium where they can cooperate, without levying healthrelated costs on any individual that this individual could not afford and without causing unfair costs for all other group members, while also not reducing the health of these other group members. I now provide two rationales for distinguishing SHI and PHI on grounds of fairness in order to address such a dilemma.
First, both PHI and SHI require the reciprocity [5] and shared interest of its members that are in an interdependent state with possibility of sanctions [6] . Also, both kinds of health insurance require a level of chance solidarity (Lehtonen and Liukko 2011). [7] (2020) presents solidarism as a way out of the public-private dilemma, where "Reallocation can take the form of universal insurance that secures the individual against misfortune or collective bene ts such as hospitals and infrastructure" (Kohn 2020, p. 1109). Further, there needs to be compensation for negative externalities produced by commodi cation. The solidarist position includes "… that the positive achievements are produced collectively and a critique of modes of allocation that re ect power and luck rather than justice". (Kohn 2020, p. 1113). Therefore, SHI and the public provision of health care address redistribution, which functions as a protection against misfortune. In addition, it is argued that gains are achieved collectively and should be fairly allocated, rather than according to luck and power. issues: the free rider problem and the assurance problem.
[8] Both can be solved via contracts and centralised punishments. "Overall, public provision tends to produce less e cient outcomes than private provision by allowing the possibility of severe underprovision of the public good. On the other hand, if a society dislikes income inequality and prefers progressive taxation, then public provision may deliver more desirable outcomes" (Slavov 2014, pp. 253-254). Thus, (economic) e ciency is positioned in a trade-off with equity, which further indicates that PHI is associated with a market-logic of direct reciprocity, i.e. tit-for-tat [9] , and a sense of solidarity that is limited to the interpersonal tier [10] but that does not extend to collective or contractual tiers.
Seeing health care as a (global) public (solidarity) good versus through the lens of private provision has some serious and differing ethical implications. [11] If we accept the latter, then "There are clear indications that in a market-driven medical system with poor regulation, the poor are not getting appropriate services for what they are paying" (Das 1999, p.118). And if we trust in market mechanisms for resource allocation in health care and health insurance, then we also accept the cost-effectiveness-equity trade-off and the trust in private rms to provide healthcare more e ciently than public entities. In this case market justice is preferred over social justice in health care (cf. Budetti 2008). To sum up, SHI and PHI are distinguishable on grounds of them being different reciprocal exchange types and on grounds of the division over fairness in public good provision that the two exhibit. In short, the different reciprocal exchanges and the division over fairness in public good provision ought to generate fairness differences between those, who only hold statutory insurance, and those, who possess PHI.
H2: Finding the prioritisation of privately insured persons fair increases the effect in H1.

Auxiliary explanations
The auxiliary variables employed in this study are located on the micro-level. However, they are attitudes and preferences directed towards institutions, i.e. the public health care system, the government, political parties, and the Austrian population's public health.
There is ample research on the role of trust in institutions and redistribution preferences in general welfare. Overall, higher trust is associated with higher demand for redistribution (e.g. Hetherington and Husser 2011; Yamamura 2014; Silva et al. 2016). Busemeyer (2021) nds that higher trust in the performance of the health care system and higher political trust increase support for additional health care spending in Germany. It is reasonable to expect that higher trust in relevant institutions corresponds to a stronger support of redistribution in health care as well.
H3: The higher the trust in the government and in the public health care system, the more a person favours redistribution in health care.
Some partisan theory suggests that left-wing voters tend to be people on low incomes, whereas rightwing voters are more likely high income earners (cf. Borck 2007). Additionally, left-wing parties spend more on public health care than right-wing parties when long enough in power (Herwartz and Theilen 2014). In line with this Koos and Leuffen (2020) nd that Germans with party preferences for left-wing parties are slightly more nancially solidary than supporters of mainstream parties and potential voters of right-wing parties are considerably less nancially solidary in context of the COVID-19 pandemic.
However, when it comes to medical solidarity Koos' and Leuffen's (2020) results suggest that party preferences become less distinct factors when compared to nancial solidarity. [12] So, it seems people exhibit both partisan differences overall and differences in solidarity in differential domains, where solidarity in health-related areas may be higher than in others. Hence, there may be notable differences between voters of different parties in their redistribution preferences in health care, as they might not share fundamental beliefs and expectations in the welfare state and the role of government in health care.
H4: People who would vote for a conservative, liberal or right-wing party disfavour redistribution in health care while left-leaning party voters are in favour of it.
Generally health care can be seen as a vital part of the welfare state (e.g. Moran 2000). Yet, the provision of high-quality health care can be located with the state, with private entities or with both as the privatepublic mix . Missinne et al. (2013) investigate 24 European countries and nd that support for state responsibility to provide universal health care is high, setting signi cant self-interest and ideological disposition aside. It seems plausible that people who think the state should provide high-quality health care also support redistribution in health care.
H5: The more a person supports state-provision of health care, the more they favour redistribution in health care.
The Meltzer-Richard Theorem (Meltzer and Richard 1981) suggests that the relatively poor want more government and the relatively rich want less government. The median voter, i.e. the person(s) with median income, will get the decisive vote on the size of government. Andreoli and Olivera (2020) lend support for this in nding that receiving social net bene ts results in supportive preferences for redistribution and Armingeon and Weisstanner (2021) nd that political ideology and self-interest interact when predicting redistribution preferences. Therein lies a trade-off between taxation and social spending or social bene ts. High taxes can guarantee generous social bene ts, while low social bene ts can make low tax rates feasible. I propose that from preferences about said trade-off, which concerns the role of the state and welfare more broadly, one can also derive preferences for redistribution in health care more speci cally.
H6: The more people favour social redistribution, the more they are also in favour of redistribution in health care.
Last, the COVID-19 pandemic poses a risk to public health and risk perception has for instance been linked to protective health behaviour (e.g. de Bruin and Bennett 2020). As health care systems, particularly ICUs, are in danger of being overburdened (e.g. Moghadas et al. 2020), there should be more support for redistribution in health care when perception of COVID-19 induced risk to the population is high.
H7: The higher the population-level COVID-19 risk perception of a person, the more they support redistribution in health care.
The Austrian health care & insurance system The Austrian health care system is solidaristic in the sense that people contribute on the basis of their ability (e.g. a speci c percentage of people's salaries that is taken off automatically) and receive services based on need (Ter Meulen 2017; Spahl and Prainsack 2021). Austria has a Bismarckian public health system. [13] Its core funding mechanism is compulsory social security contributions that are directly or indirectly tied to employment and to occupational status. Insured persons can include children or other family members without separate insurance. In addition, there is a growing sector of PHI for those who want services beyond the services covered by the statutory health insurance. PHI covers 'extras', such as access to private doctors, a single-occupancy room during hospital stay et cetera. Austria has very high health insurance coverage (about 99%) and its population reports few unmet medical needs (Toth 2019).
In a previous analysis Austrians exhibited the highest satisfaction with their health care system out of 14 investigated Western European countries, but Austrians also had the third lowest rank in wanting

Redistribution preferences
The dependent variable is an additive index of eight variables of preferences for redistribution in health care. These were developed for the ACPP and answers are on 5-point Likert scales. Three preferences directly grasp the relationship between public and private medicine. [15] The other ve preferences encompass equality of treatment, public spending increase on health care, personal willingness to pay higher contributions, higher contributions for cost-intensive patients, and handling of medical excess payments (user charges). More information on these preferences and the index is located in the Appendix ( Fig. A1-A5).
Private health insurance, fairness and auxiliary variables The main predictor variables are possession of PHI (dichotomous) and attitudes on fairness of prioritising privately insured persons, where the latter is on a 5-point Likert scale. Auxiliary variables are trust in government and in the public health care system. Further, party preferences are considered, wherein conservative, liberal and right wing parties are pooled, as well as left-wing parties, and also other parties and non-voters with those who would cast a void ballot and 'do not know' responses. Government responsibility to provide high-quality health care grasps the concept of state welfarism versus retrenchment. Redistribution preferences in distributive welfare are considered where high taxes and high social bene ts are pinned versus low taxes and low social bene ts. Perceived health risk is the personally assessed threat of COVID-19 to the Austrian population's public health.
All predictor and auxiliary variables are standardised. Multiple imputation is applied for all auxiliary variables and consequent listwise deletion of missings yields a nal sample size (N) of 1027 respondents.

Methods
The analyses are done using OLS regression models [16] with demographic and political sample weights to Principal Component Analysis (PCA) in OLS regressions. [17] All statistics were carried out with the Stata 16.1 software.

Results
In the demographically weighted sample 74% do not possess PHI, while 26% do. The distributions of eight redistribution preference statements show considerable variances (see Fig. 1 and in Appendix Fig.   A1), and so do the attitudes on fairness of prioritising privately insured (see Fig. 2). Fig. 1 shows that some health redistribution policies are more contested than others. Agreement to guarantee equal treatment and care for all is the most agreed upon answer, followed by support to increase public spending on health care. However, to personally pay higher contributions to nance the public health care system is rather unpopular, and so is to levy costs on patients. The largest cleavages between people that possess PHI and the ones who do not show in the three redistribution preferences that explicitly address expansion or restriction of private and respectively of two-tier medicine (for more details see Figures A1-A3 in the Appendix).
On average people with PHI exhibit lower support for redistribution in health care than persons without (see Appendix Exhibit A6). The inferential analysis shows that the possession of PHI is signi cantly and negatively associated with support for redistribution in health care (p<0.001). This effect holds true across all employed models.
When adding the variable 'fairness of prioritisation of privately insured persons' to the Baseline Model the results show that assessing such a priority setting as fair is signi cantly and negatively related to support for redistribution in health care as well (p<0.001). This also applies to the Full Model, which includes the auxiliary variables.
The Auxiliary Model only consists of the auxiliary covariates. It shows that trust in government has a signi cant and negative effect on support for redistribution in health care (p<0.05). This effect persists in the two further models as well. In addition, preferring a left-wing party is signi cantly and positively associated with such a redistribution. People that think it is the government's responsibility to provide and guarantee high-quality health care are also more in favour of redistribution of health care (p<0.01).
Preferring higher taxes and more generous social bene ts has a highly signi cant and positive effect on redistribution preferences in health care throughout all models (p<0.001). Perceiving COVID-19 to be a greater risk for the Austrian population also results in more support for redistribution in health care (p<0.01). Figure 4 shows that in the Full Model having no PHI but nding prioritisation of privately insured as (somewhat) fair results in lower support for redistribution. Possessing PHI and having the same fairness attitude yields even lower support for redistribution in health care (see also Fig. A7 in the Appendix).
In the Interaction Model it is evident that regarding the prioritisation of privately insured persons as fair reduces the support for redistribution in health care in not privately insured and even more so in privately insured people.
The GSEM yields similar results as the Full Model, with the only difference that trust in the public health care system becomes signi cantly and positively associated with redistribution preferences in health care (see Appendix Table A2). The OLS regressions with the Polychoric Principal Component Analysis (PCA) index also corroborate the initial results, with exception of the political party effects, which do not show signi cance anymore (see Appendix Table A3).

Conclusion
As H1 has been corroborated in this study, it appears that self-interest plays a considerable role in redistribution preferences in health care. However, self-interest cannot explain the whole picture. In this study fairness attitudes for prioritising privately insured persons have been considered as well. This is novel insofar usually prioritisation principles such as severity of illness, fair innings or proportional shortfall are discussed (cf. Stolk et al. 2005) and not explicit prioritisation of privately insured persons.
The second rationale besides self-interest that can explain the ndings in this study rests on reciprocity and the prosperity of health care being a public good. Dolan et al. (2003, p. 546) argue that in a social and personal perspective people ask ex ante "What value do I attach to treatment being available to a group of people amongst whom I might nd myself?" This question does not foreground mere self-interest but similarity with a shared group, namely the group of privately insured. When a person accepts to share costs to aid others with whom they recognise similarity in a relevant aspect this may be a case of solidarity (cf. Prainsack and Buyx 2011, 2017). This aspect of similarity may even be the possession of PHI, where people agree to share burdens and bene ts beyond public insurance and pay premium payments into private insurances.
While trust in the public health care system has no signi cant effects, trust in government exhibits, contrary to expectations, a negative effect on support for redistribution in health care. The most straightforward explanation for this is that at the time of the survey there was a conservative-led government coalition in Austria and, as argued, conservative-leaning persons tend to dislike redistribution. Thus, people who put more trust in the government might simply be conservative party voters.
People with a preference for left-wing parties support redistribution in health care. This might be because the left wing traditionally nds more backing in lower-income groups, who cannot afford to pay PHI premiums. It can also be because left-wing supporters are ideologically more ready to favour redistribution, irregarding of their own social stratum or status.
Persons who think high-quality health care provision is a government responsibility supported redistribution in health care as well. This indicates that people who want a strong public health care system, also want one that public resources are redistributed to potentially reduce inequity in health care.
Austrian residents who support higher taxes and more social spending or social bene ts also favour redistribution in health care. This highlights that people are consistent in their welfare choices and that health care, while having special ethical importance, does not stand out compared to social welfare and government redistribution overall.
Last, health risk to the general Austrian population has a signi cant positive effect on support for redistribution in health care. This may be because the pandemic has aggravated inequity in health care and people with a high risk perception feel this inequity needs to be addressed through increased redistribution in the public health care system.

Discussion & Policy context
Various scholars call for 'evidence-based medicine' and public engagement, also by including health care preferences of the public for issues such as prioritisation (e.g. Dolan  This study has shown that privately insured survey respondents exhibit considerable self-interest and also justify this to be fair. The implications of this are manifold. First, policy makers should be clear and explicit about whether they want to foster a medical two-tier system or not. If the answer is yes, then preferences of privately insured persons should be considered. If not, the ndings in this study are thought-provoking because privately insured people may create a demand for practises that endanger the principles of equity, e ciency and good public health for the population as a whole. In the long term PHI might also undermine solidarity in the health care system because people who possess PHI may exclude non-PHI holders in their understanding of reciprocity, fairness and solidarity, which have to be foregrounded and reinforced especially amidst a pandemic. Second, if policy makers want to safeguard existing solidaristic health care systems, they might want to actively counter the self-interest of the privately insured through dialogue and proactive policies. This  Additionally, "When PHI is a major mechanism for health care nancing, the sicker tend to purchase PHI, but when PHI is complementary or supplementary, the better off, who have the capacity to pay, may have higher tendency to purchase PHI" (Jeon and Kwon 2013, p. 75). Therefore, in social insurance systems with a statutory insurance scheme, like the Austrian one, it seems possible that PHI is purchased not only to receive supplementary services but also in the expectation to be potentially prioritised for treatment.
This generates serious pressure for health care policy makers to avoid the establishment of two-tier medicine.
Following Miller and Horowitz (2000) it would be bene cial for patients if doctors made their nancial incentives explicit and openly communicated their private practice, if existent, to patients in order to foster transparency. It would also be advantageous to health policy stakeholders and the public at large if medical priority setting standards and practises were to be made explicit.
It is a lasting concern that shifting the balance between solidarity and self-reliance in favour of the latter would decrease redistribution in health care (cf. Hinrichs 1995 bene ts from such privatization policies -e.g. reduced costs, shorter waiting times -must be weighed against the long-term damage such policies can do to countries' ability to cope with a rapidly-spreading infectious disease". Some scholars are alert when it comes to more privatisation of health care, as a strong public health care system is equipped better to manage pandemics than a weakened one (e.

Authorship
All authors contributed equally in production of this manuscript.

Funding
The project has been funded by the Austrian Science Fund (Grant P33907-G) and the Vienna Science and Technology Fund (Grant EI-COV20-006). Data collection has additionally been nancially supported by the rectorate of the University of Vienna, the Austrian Chamber of Labour and the Federation of Austrian Industries. The funders had no role in the study design, data collection and analysis, the decision to publish, or the preparation of the manuscript.

Authors' contributions
Not applicable (one author)

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Appendix.docx