East and West Germans' Attitudes and Preferences Regarding the Welfare State

Pfarr, Christian; Schmid, Andreas; Ulrich, Volker Article East and West Germans' Attitudes and Preferences Regarding the Welfare State Journal of Contextual Economics – Schmollers Jahrbuch Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Pfarr, Christian; Schmid, Andreas; Ulrich, Volker (2017) : East and West Germans' Attitudes and Preferences Regarding the Welfare State, Journal of Contextual Economics – Schmollers Jahrbuch, ISSN 2568-762X, Duncker & Humblot, Berlin, Vol. 137, Iss. 3, pp. 261-299, https://doi.org/10.3790/schm.137.3.261 This Version is available at: https://hdl.handle.net/10419/292516 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ East and West Germans’Attitudes and Preferences Regarding the Welfare State* By Christian Pfarr,** Andreas Schmid,*** and Volker Ulrich**** Abstract Do citizens from former Communist countries exhibit attitudes and preferences with regard to income redistribution that differ from those in the West? This paper seeks to answer this question for reunified Germany. The analysis uses not only survey data on attitudes but also evidence on preferences from a choice task, which forces individuals to consider trade-offs and budget constraints. While we confirm results of prior studies documenting considerable differences in attitudes, we find no differences regarding preferences between East and West Germans. When facing economic restrictions, East and West Germans behave the same. JEL Codes: D63, H23, P26 Journal of Contextual Economics 137 (2017), 261 –300 Duncker & Humblot, Berlin Journal of Contextual Economics 137 (2017) 3 *We gratefully acknowledge financial support from the German Science Foundation (DFG). We thank Peter Zweifel, Stefan Napel, Hartmut Egger, David Stadelmann, Andreas Haufler, Frank Neher, and Francesco Trebbi for very helpful discussions, suggestions, and comments, as well as two anonymous reviewers and the editorial office. Moreover, we thank the participants of the 2nd Danish Choice Modeling Days in Odense, the 28th Annual Meeting of the European Economic Association (EEA) in Gothenburg, the Meeting of the 2014 European Public Choice Society (EPCS) in Cambridge, the 2014 Annual Meeting of the German Association of Health Economics (dggö) in Munich, and the 2013 research seminar in public finance at the ifo Institute in Munich. We are also grateful to Martina Wagner for her assistance in preparing the discrete choice experiment as well as the socioeconomic questionnaire. All remaining errors are ours. We declare that we have no conflict of interest. ** Faculty of Law, Business & Economics, University of Bayreuth, Universitätsstr. 30, 95447 Bayreuth, Germany. The author can be reached at christian.pfarr@uni-bayreuth. de. *** Faculty of Law, Business & Economics, University of Bayreuth, Universitätsstr. 30, 95447 Bayreuth, Germany. The author can be reached at [email protected]. **** Faculty of Law, Business & Economics, University of Bayreuth, Universitätsstr. 30, 95447 Bayreuth, Germany. The author can be reached at volker.ulrich@uni-bay reuth.de. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 1. Introduction More than 25 years after the fall of the Iron Curtain, Germany remains a natural experiment on the persistence of the effects of socialization and ideology. Debate continues about whether attitudes and preferences with regard to the welfare state and its correlate, income redistribution, still differ between East and West Germans (e.g., Corneo 2004; Alesina and Fuchs-Schündeln 2007; Kuhn 2013). Most studies investigating the effects of socialization have relied on surveys of preferences for redistribution without controlling for tradeoffs and budget constraints. When replicating these methods with recent data, we still find the differences observed in the past. However, we provide evidence that, after controlling for economic constraints and enforcing trade-offs, East and West Germans do not differ in their preferences for redistribution. Our analysis of the effect of socialization on redistribution preferences is based on a unique dataset involving 1,538 representatively selected German citizens in 2012. The data collection was tailored to allow for detailed analyses of the citizens’stances. Their views were captured by a standard agreement question, reflecting attitudes, and by choice-task questions that account for trade-offs and budget constraints, reflecting preferences. We follow the common definition of an attitude as “a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor”(Eagly and Chaiken 1996, 269). Kahneman, Ritov, and Schkade (1999) explain that attitudes, captured by rating scale questions or similar survey questions, differ from preferences, whose meaning is rooted in economic theory. Note that attitudes and preferences do not have to overlap: individuals might desire more redistribution while not being willing to equally participate in its financing. Our contribution is threefold. First, we revisit the effects of communism on differences in attitudes toward income redistribution between East and West Germans. Second, we aim at measuring, for the first time, preferences for redistribution in East and West Germany using a discrete choice analysis that accounts for trade-offs and budget constraints. Third, we analyze whether attitudes and preferences overlap and the degree to which differences in attitudes may help to explain preference structures. The remainder of the paper is structured as follows. The next section discusses the effect of communism on individuals’attitudes and preferences. Section 3 explains the conceptual framework. Section 4 provides descriptive statistics and the econometric model specification. Section 5 presents the empirical analysis, including comprehensive robustness checks. Section 6 concludes. 262 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 2. Views on the Effect of Communist Socialization Differences in attitudes toward the welfare state between people in different countries are well documented (e.g., Alesina, Glaeser, and Sacerdote 2001; Isaksson and Lindskog 2009). Europeans, for instance, tend to support the welfare state more than U.S. citizens do. A recent literature finds that attitudes vary in particular between former socialist countries and western market economies (Alesina and Giuliano 2011; Luttmer and Singhal 2011). This literature examines the determinants of these differences, such as ideology, culture, and the effect of institutions. Various studies also show that socialism affects attitudes. For example, citizens from former socialist countries consistently show a comparatively strong desire for redistribution (Corneo and Grüner 2002; Suhrcke 2001), and this effect has persisted for two and a half decades since the fall of the Iron Curtain (Guillaud 2013). Focusing on immigrants, Luttmer and Singhal (2011) show that attitudes toward income redistribution are mainly shaped by culture, i.e. factors such as customary beliefs and values that are passed on over generations. Their results also indicate –in line with the theoretical literature (e.g., Benabou and Tirole 2006; Alesina and Angeletos 2005; Alesina, Cozzi, and Mantovan 2012) –that culture is enduring, even 20 years after immigration. Stipulating that former socialist countries have evolved differently, Corneo (2001) focuses on the so-called east-west dichotomy, exploring the differences in attitudes between U.S. citizens and East and West Germans. He finds a marked inequality aversion among East Germans, compared with West Germans and U.S. citizens, leading to a stronger desire for the state to reduce differences in income. According to Alesina and Fuchs-Schündeln (2007), German reunification can be treated as an exogenous shock. Before World War II, no relevant systematic differences between East and West could be observed. After the division of Germany in 1949, separation was increasingly enforced, culminating in the construction of the Berlin Wall in 1961. Until the fall of the wall in 1989– 1990, East Germans could not relocate to West Germany, regardless of their preferences; the political and economic system was, to a large extent, not chosen but imposed. Even individuals who did not identify with socialism were subject to its socialization, which likely shaped their stance on the welfare state and its correlate, income redistribution. In contrast, West Germans were influenced by a market economy and capitalism. This makes West Germans a good control group for East Germans. The first scholar to exploit these differences was Corneo (2004), using data for 1992 and 1999. He confirmed earlier studies, finding that socialism does impact attitudes, with partial convergence between East and West during the period of observation. In their comprehensive analysis of citizens’attitudes to- East and West Germans’Attitudes Regarding the Welfare State 263 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 ward the welfare state, Alesina and Fuchs-Schündeln (2007) cover the years 1997 and 2002, asking whether there is a feedback effect of the East German regime on citizens’attitudes. They find that, even after controlling for income, the effect of socialism prevails, making East Germans more inclined to agree that the political system is responsible for individual success, while poverty is the fault of society. The authors conclude that the influence of communism does not depend on socio-demographic characteristics such as state of residence. Rainer and Siedler (2008) and Stichnoth (2012) find supporting evidence. Kuhn (2013) uses a slightly different framing. Respondents to the survey of the International Social Survey Program of the years 1987, 1992, and 1999 were asked to estimate wages earned in several professions and to say how much people working in these professions should earn. Using the difference between these two values as a proxy for respondents’demand for redistribution, he finds this demand to be much stronger among East than West Germans. His results suggest that different inequality perceptions of East and West Germans explain some of the observed differences. Summarizing, communist socialization seems to be a crucial and persistent determinant of individuals’attitudes.The latest data available for Germany applied in an academic study are from 2005 (Rainer and Siedler 2008). Since then, attitudes of East and West Germans may have converged, as hypothesized by Alesina and Fuchs-Schündeln (2007), who predicted the year 2009 as a lower bound for convergence. Using data from 2012, we aim to close this gap. Our study also addresses an important methodological issue, i.e., distinguishing attitudes from preferences. Attitudes are measured by drawing on questions such as “To what extent do you agree or disagree with the statement, ‘It is the responsibility of the government to reduce the differences in income between people with high incomes and those with low incomes’?”(e.g., Alesina and La Ferrara 2005; Corneo and Grüner 2000).Responses to questions of this type are typically labeled as preferences (e.g., Alesina and Giuliano 2009; Corneo and Grüner 2002; Guillaud 2013). But from a microeconomic perspective, it seems questionable that agreement with such a question reflects a true preference rather than a pure attitude. This touches the fundamentals of economic theory. As McFadden notes, the “existence of underlying preferences is a vital scientific question for economists”(2001, 363). While preferences are assumed to be stable, attitudes can change and are influenced by sometimes transient and volatile affects and motivation. Put simply, attitudes and preferences are not the same. Individuals might want more redistribution while not being willing to equally contribute to its financing. Another common problem of methods that rely on respondents’agreement with certain statements is the tendency toward bias. For example, social desirability may likely result in excessive generosity when no budget constraint is 264 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 imposed (“yeah-saying”; Blamey, Benett, and Morrison 1999). Such results therefore may not predict actual decision-making. Misconceptions of this type may cause policymakers to devise proposals that fail politically. The typical framing of survey questions capturing citizens’stances on redistribution is inadequate for eliciting preferences (e.g., Fong 2006). Yet studies reaching beyond this standard approach are rare. There are exceptions, such as work by Neustadt and Zweifel (2010; 2015), Neustadt (2011), Pfarr (2012; 2013), Pfarr and Schmid (2016), as well as Pfarr, Schmid, and Mørkbak (2016). These authors estimate preferences for redistribution using a discrete choice analysis, though none of them examine the effects of communism or differences between East and West Germans. Another strand of pertinent literature that helps to overcome some methodological issues involves laboratory experiments. Ockenfels and Weimann (1999) and Brosig-Koch et al. (2011) experimentally examine differences between East and West Germany. (The latter study replicates the first one 20 years after reunification.) Their focus is on solidarity as a social norm, which may explain demand for redistribution. Their evidence indicates that East Germans show consistently less solidarity than West Germans: they are less generous toward the losers of a lottery with whom they could voluntarily share their winnings. This contrasts with the previous survey findings. Thus while they confirm the existence of differences in norms, their results present a puzzle. The results of our choice-based approach illuminate this paradox. The advantage of experiments is that they provide greater control over the decision situation and can offer real financial incentives. What’s more, compared with the participants in the two prior experiments, who are university students from three different large German cities, most of them in their early twenties, our sample is representative of the German voting age population. 3. Methods 3.1 Conceptual Framework To measure preferences in the economic sense, we apply a discrete choice approach, creating stated preferences data. This approach can be justified because redistribution of income is a nonmarketable good, and citizens rarely reveal their preferences for it. Absent revealed preferences, stated preferences may be used instead. The discrete choice analysis, developed by Louviere and Woodworth (1983) and Louviere and Hensher (1982), is rooted in decision theory and microeconomic utility theory (Lancaster 1966). Individuals derive utility not from the good per se, but from its characteristics (the so-called attributes). In the baseline version of an experimental task, participants choose repeatedly between the status quo, which remains fixed, and hypothetical alterna- East and West Germans’Attitudes Regarding the Welfare State 265 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 tives. These alternatives have the same attributes but vary regarding the attribute levels.1In our case, the individuals make a number of pairwise decisions. We define the whole redistributive system as one good with specific attributes, described below. In general terms, a utility-maximizing person will always choose an alternative if its utility exceeds that of the status quo. Utility must be treated as a latent construct that cannot be directly observed. Building on the random utility theory (McFadden 1974, 1981; Manski 1977), the indirect utility function is split into a deterministic observable component and a stochastic component. The deterministic part consists of the price of the respective alternative and its attributes as well as the income and socio-demographic characteristics of the respective individual.The inclusion of the price attribute imposes a budget constraint. According to random utility theory, only the probability that an individual chooses a specific alternative can be estimated (Louviere and Street 2000, 53). This probability indicates individuals’decision-making and comports with their preference for a given redistributive system expressed by choosing one of the proposed options. An important parameter reflecting the structure of preferences is the marginal rate of substitution (MRS). However, as indicated by microeconomic theory, the MRS depends on where it is evaluated along an indifference curve. For this reason, in many choice tasks, a fixed status quo is used as the baseline, ensuring that comparisons of utilities refer to the same reference point. Otherwise, the calculated welfare measures might be biased. Accordingly, a rational utility-maximizer chooses proposed alternative Bif this alternative offers a higher utility than the status quo S. If the individual also chooses status quo Srather than alternative A, his or her indifference curve must be located between Aand B(Vroomen and Zweifel 2011, 89). Repeated choices of this type enable estimation of the slope of the indifference curve, that is, MRSn;mbetween attributes mand n(Lacsar, Louviere, and Flynn 2007, 1741). By partially differentiating the indirect utility function, the MRS is given by MRSm n¼  Attribute parameter m Attribute parameter n ¼  ^ m ^ n ð3:1Þ where δis the estimated parameter of the respective attribute. If the denominator is specified as the price attribute, the MRS can be interpreted as marginal willingness to pay (MWTP),2that is, the MWTP for an additional unit of attri- 266 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 1For example, alternatives could be characterized by the attributes income and leisure time, the according levels being 1,000 €/1,500 €/2,000 €per month and 2 h/3 h/4 h per day respectively. 2Using Roy’s Identity, the price parameter can be interpreted as the marginal utility of income. For the formal proof, see Hanemann (1983, 544) or Telser (2002, 56). OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 bute mexpressed in units of the individual’s income. This measure of preferences will be applied in the empirical analysis. 3.2 Implementation and Survey Design Because of the hypothetical nature of the choice task, its design requires special attention. Following Bateman et al. (2002, 258), the process included intensive literature reviews, expert interviews, and group discussions involving a total of 629 students as well as three independently conducted pretests with about 40 participants each. The scenarios depicted in the choice task should replicate the social and redistributive system in Germany as accurately as possible and cover the aspects most relevant to the general population. After having identified all potential sources and recipients of social spending –among others based on the analysis of the annual financial statement of the Federal Ministry of Labor and Social Affairs (BMAS 2011) –we reduced attributes to a number manageable for the participants and phrased them in terms understood by the broad public. Both goals were pursued in the aforementioned interviews, discussions and pretests, in which participants could prioritize the different potential attributes and could suggest own aspects. The framing of the technical terms was tested and repeatedly revised. Ten attributes were found to primarily affect individuals’utility in the context of redistribution. These attributes are personal tax and social insurance contributions, the amount of redistribution as a percentage of GDP, the sociodemographic status of beneficiaries (sick persons and persons in need of care, families and children, retirees, the unemployed, the working poor) as well as the nationality of recipients (German, West European, other).3These categories are not mutually exclusive. For example, sick persons could also receive benefits due to being retirees. The attributes must then be assigned meaningful levels. Pitting an increase in the level of one attribute against a decrease of another attribute forces respondents to face trade-offs (Bateman et al. 2002, 260; Telser 2002, 39). Figure 1 presents the chosen attributes and their respective levels categorized by their substitutive relationships. For example, the attribute “retirees”indicates the share of the redistributive budget that is dedicated to this group. The status quo level for this attribute is 40%, and the two alternative levels are 30% and 45%. If the higher level is chosen, this comes at the expense of one of the other four beneficiary groups. East and West Germans’Attitudes Regarding the Welfare State 267 Journal of Contextual Economics 137 (2017) 3 3This design enables researchers to make use of this dataset for a variety of research questions. In the case of this study, the different beneficiary groups as well as the nationality of recipients are not of primary interest and only serve as controls to avoid potential bias. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Figure 1: Attributes, Labels and Levels The status quo levels replicate the actual status quo in Germany.4The attribute levels of the alternatives have to cover a sufficiently wide range to capture the individual’s potential preferences as approximated through the pretests. At the same time, the number of levels must be kept to a minimum to avoid too many alternatives, which are a result of the number of levels per attribute multiplied with each other. For most attributes, this resulted in a rather balanced number of levels below and above the status quo. However, attributes with a very low value for the status quo naturally exhibit higher levels. Keeping the minimum difference between levels at five percentage points also implies that, for the working poor and West Europeans, no level lower than the status quo exists.5Table A.1 in the appendix summarizes the means of the choice charac- 268 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 personal tax and social insurance contributions total amount of redistribution type of beneficiaries nationality of recipients tax and contribution (TC) redistribution (RE) retirees (RI) sick persons and persons in need of care (SP) unemployed (UL) families and children (FC) working poor (WP) German (DE) West Europeans (WE) other (OT) 010 20 30 40 50 60 70 80 90 100 % alternative attribute levels status quo Note: All attribute labels in Percent. 4For econometric reasons –especially to maintain the independency of attributes and individuals –all individuals must be given the same status quo. This is critical, as individual tax rates vary due to the progressivity of the German income tax. For the status quo, we thus chose the average income tax rate. As this is a strong assumption, we test for this, for example, by controlling for income levels and find no bias. 5In this particular case, this is no major concern, as the pretests showed that basically nobody wants to reduce the share dedicated to the working poor. Offering a level of zero would thus have been highly unrealistic. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 For the econometric specification, we must bear in mind that the groups of beneficiaries and recipients according to nationality each add up to 100%. The attributes sick persons and persons in need of care (SP) and Germans (DE) are the omitted reference categories to avoid perfect collinearity. Individuals’decision-making is only determined by the utility differences. Consequently, sociodemographic characteristics cancel out of the calculation of the MWTP, as they are constant between the decisions. Thus the variable communism enters the estimation equation in the guise of an interaction with the main variables of interest ΔTC, ΔTC², ΔRE, and ΔRE² to allow for a difference of MWTP between respondents with and without a communist background (Boxall and Adamowicz 2002, 421; Johnson and Desvousges 1997, 83): choice ¼0þ1TC þ2TC2þ3ðTC communismiÞþ 4ðTC2communismiÞ þ 5RE þ6RE2þ 7ðRE communismiÞ þ 8ðRE2communismiÞþ 0Xremaining attributes þ"i ð4:2Þ with srepresenting parameters to be estimated. The remaining attributes serve as controls. To simplify the interpretation, the model is specified as a difference model; that is, the value of the attributes included in the estimation reflects the deviation from the status quo.10 5. Results 5.1 Attitudes Regarding the Welfare State Table 4 presents the results for the ordered probit models designed to analyze attitudes toward the welfare state for East and West Germans, with inequality reduction as the dependent variable.11 Estimate [1] shows the effect of being an East German versus being a West German (reference category). In line with the literature discussed in section 2, East Germans tend to strongly agree with the statement that the government should reduce income differences. This difference is highly significant, suggesting that, almost 22 years after reunification, the two groups of Germans still differ in their perception of the state’s role vis-a-vis income redistribution, thereby confirming East and West Germans’Attitudes Regarding the Welfare State 275 Journal of Contextual Economics 137 (2017) 3 10 For example ΔRE = RE status quo −RE alternative . The quadratic terms are generated in the same way (e.g., ΔRE 2 = (RE status quo −RE alternative ) 2 ) to adequately reflect the quadratic deviation from the status quo. 11 The coefficients reported in the table are the total coefficients. We report corresponding marginal effects in the text when we are interpreting the size of the coefficient and compare it, e.g., to the results of Alesina and Fuchs-Schündeln (2007). OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 the results of Alesina and Fuchs-Schündeln (2007), Corneo and Grüner (2002), and Kuhn (2013). However, this finding also contradicts predictions of Alesina and Fuchs-Schündeln (2007), who hypothesize that attitudes are converging (and forecasting the year 2009 as a lower bound for convergence). We find that Eastern origin still increases the probability to voice strong agreement with the government’s responsibility to reduce inequality by 10.5 percentage points. This resembles the level reported by Alesina and Fuchs-Schündeln (2007) for favoring state intervention in the year 2002.12 Thus estimate [1] points to attitudes as being an enduring characteristic. Estimate [2] of Table 4 addresses the effect of migration. West Germans continue to constitute the reference group. By contrasting East living in West against East living in East, one can see considerable differences. The first variable is defined as individuals who have migrated to the West since reunification, and the second is defined as East Germans who have stayed on the territory of the former GDR. Coefficients of East living in East in estimate [2] are somewhat higher than those for East Germans in estimate [1] of Table 4. The coefficient of the variable East living in West fails to reach statistical significance. Attitudes of Easterners who moved west do not seem to differ from those of their West German neighbors. Moving to a West German neighborhood since reunification could have fostered convergence compared to people who stayed in the East, but individuals who migrated to the West could also constitute a self-selected group with attitudes closer to those of the West Germans from the outset.13 Relying on cross-sectional data, we cannot analyze this aspect. Due to the low number of individuals who have moved, the lack of significance could also been driven by the sample size. Thus, when both groups are summarized as East Germans, the effect of East Germans still living in the East is marginally diluted by East Germans living in the West. Due to these results, the following analysis will rely on East living in East as a proxy for communism and West Germans as the reference category.14 276 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 12 Alesina and Fuchs-Schündeln report only selected marginal effects for different subgroups of recipients. They state: “Being from the East increases the probability of favoring state intervention by between 14.5 and 17 percentage points in 1997, compared to being from the West. Between 1997 and 2002, the probability of favoring state intervention for an East German declines by between 2.3 and 6.9 percentage points”(2007, 1512). 13 This is discussed in more detail by Alesina and Fuchs-Schündeln (2007), explaining the diverging paths of these two East German groups. Most of the convergence between East Germans and West Germans between 1997 and 2002 seems driven by the East Germans who stayed in the East. So migrants either differed from the outset or had already achieved their final level of convergence in 1997. 14 In another estimate, West Germans who migrated to the East were excluded without any relevant effect. Results are available from the authors upon request. As the number of observations is very low for this subgroup, we refrain from further analysis of pertinent questions. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table 4 Results of Ordered Probit Models for Inequality Reduction inequality reduction [1] [2] [3] [4] Coeff. SE Coeff. SE Coeff. SE Coeff. SE East Germans 0.268 (0.072)*** East living in East 0.290 (0.074)*** 0.393 (0.194)** 0.319 (0.155)** East living in West 0.015 (0.202) East living East *Age1 (highest age) 0.347 (0.419) *Age2 0.153 (0.246) *Age3 –0.303 (0.235) *Age4 –0.203 (0.239) *Income1 –0.140 (0.254) *Income2 –0.000 (0.211) *Income4 –0.205 (0.222) *Income5 –0.137 (0.264) Age1 (highest age) –0.283 (0.228) Age2 0.088 (0.118) Age3 0.073 (0.113) Age4 0.051 (0.120) Income1 0.007 (0.127) Income 2 0.105 (0.126) Income 4 –0.158 (0.119) Income 5 –0.188 (0.118) LL –1,257.64 –1,256.84 –1,250.72 –1,248.15 Mc Fadden Adj. R 2 0.003 0.002 0.002 0.004 AIC 2,523.29 2,523.69 2,525.44 2,520.30 BIC 2,543.70 2,549.20 2,586.67 2,581.53 N1,215 1,215 1,215 1,215 * p < 0.1, ** p < 0.05, *** p < 0.01; Results of ordered probit models with inequality reduction as dependent variable. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. The persistence of the effects from living under a communist regime can be analyzed further by introducing age as an intervening variable. This variable reflects the amount of time spent under the GDR regime. One would hypothesize that the longer someone was exposed to communism (i.e. the older someone is), the stronger and longer-lasting the effects on their attitudes would be. We define five age cohorts15 using the youngest (age5) as the reference catego- East and West Germans’Attitudes Regarding the Welfare State 277 Journal of Contextual Economics 137 (2017) 3 15 Following Alesina and Fuchs-Schündeln (2007), the category of the eldest participants is extreme. The goal was to capture individuals who had been socialized for a long time before the imposition of communism in the East. However, this results in very small OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 ry. As can be seen from Table 4 (estimation [3]), the effect of communism remains significant even when accounting for age. No statistically significant age effect can be identified for East Germans living in the East.16 Examining the interactions of East living in the East with the respective age categories, even the difference between categories two and three fails to reach significance on the ten percent level, as do all other differences. Regarding age, our results therefore only partially confirm those of Alesina and Fuchs-Schündeln (2007). They find that West Germans are less inclined to emphasize the responsibility of the government the older they are, which is the opposite of East Germans. Their evidence suggests a strong relationship between the amount of time spent under communism and the attitude toward redistribution. Kuhn (2013) concludes that, across all age cohorts, East Germans show greater support for redistribution. But the differences between the age cohorts are small, and no levels of significance are given. Thus our results resemble those of Kuhn (2013) and contribute additional nuance to a heterogeneous literature. One could argue that the relatively long period since the end of communism in Germany overshadows the effects that were still visible in 1997 and 2002. Income is a second obvious factor that might influence individuals’attitudes regarding redistribution. Following Romer (1975), Roberts (1977), and Meltzer and Richard (1981), one could argue that low-income individuals have a strong self-interest in favoring redistribution. Income levels may also have different effects on individuals in the East and in the West. Thus estimation [4] of Table 4 accounts for individual gross income with the middle category as reference. Again, the coefficient of East living in East remains significant, which is again in line with the results of Alesina and Fuchs-Schündeln (2007). In contrast, income categories do not seem to have any significant effect. While one might see some statistically insignificant trend in the general income categories, suggesting a stronger attitude in favor of redistribution by low-income individuals, the results of the interaction terms do not suggest any pattern. Summarizing, we do find relevant differences between East Germans who stayed in the East and those who migrated to the West. While the latter group does not exhibit significant differences when compared to West Germans, we still find significant differences between East Germans who stayed in the East 278 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 numbers in the class. We thus did the same analysis with five balanced classes and with only three classes. The results remained unchanged. 16 Ai and Norton (2003) point out that the full interaction effect in a nonlinear model should be calculated using the “cross derivative of the expected value of y[…].”However, in a nonlinear model, the expected value of ydepends on the unobserved error term. This renders an exact calculation almost impossible. Thus we have tested whether the results prove robust when using a traditional OLS specification and find that the magnitude does not change significantly. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 and West Germans. This is robust when controlling for age and income and underlines the persistence of attitudes shaped under communism. 5.2 Preferences Regarding the Welfare State As outlined above, measured attitudes are prone to misinterpretation because they do not account for trade-offs imposed by a budget constraint. Thus, as Phillips, Johnson, and Maddala (2002) argue, “… the most useful measure of the strength of preferences is the monetary payment that would leave people indifferent between having a given utility change and not having the change (i.e., willingness to pay or willingness to accept).”Therefore MWTP for redistribution is the preferred measure. Estimating eq. [4.2] leads to coefficients that allow us to express preferences in terms of MWTP values.17 The results show a negative coefficient for the price attribute (ΔTC), in line with microeconomic theory, suggesting diminishing utility in the case of increasing prices. As the coefficient of ΔTC 2 is also negative, the disutility increases disproportionately with the price attribute. While the estimated utility parameter for ΔRE is positive,ΔRE 2 has a negative sign. This suggests that respondents’utility from redistribution is positive, but with a decreasing marginal return. Regarding beneficiaries of redistribution, coefficients of the estimation model indicate a preference for increasing the budget for retirees and families and children while cutting the redistributive budget devoted to the unemployed and working poor. Coefficients for ΔWE and ΔOT are negative, suggesting Germans prefer a shift in the distribution toward recipients of German nationality. For the analysis of the effects of communism, the proxy East living in East is interacted with the linear and quadratic terms of the two attributes ΔTC (tax and contributions) and ΔRE (redistribution). However, results show no significant effects of the interaction terms. In the context of this study, the price attribute is specified as percentage of income that is spent on tax and contributions. The higher this attribute, the higher the disutility because of the direct negative effect on net income. The MWTP can thus be interpreted as the willingness to give a certain number of percentage points of income for an additional percentage point of GDP devoted to redistribution beyond the status quo. For the subsequent analysis, we will refer to MWTP values for the interpretation of citizens’preferences toward redistribution. MWTP values are calculated by first estimating equation [4.2]. This permits calculation of the partial derivatives with respect to the attributes ΔRE and ΔTC according to equation [3.1]. Evaluating them at the status quo values, i.e., RE ¼TC ¼0, causes the quadratic terms to drop out of the equation. The calculation of the MWTP also East and West Germans’Attitudes Regarding the Welfare State 279 Journal of Contextual Economics 137 (2017) 3 17 The full estimation results are available in the appendix, Table A.2. OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 includes the partial derivatives pertaining to the interaction terms, evaluated at communism = 1 and communism = 0: MWTPRE TCjTC¼0;RE¼0¼  Attribute parameter of RE Attribute parameter of TC ¼  1þ3 5þ7 ¼ 0:0334 – 0:0005 0:0559 – 0:0031 ¼0:558 ð5:1Þ East Germans living in the East therefore exhibit a MWTP of 0.558 percentage points of their monthly gross income for an increase of redistribution by one percentage point of GDP.18 Standard errors are estimated using the delta method (Hole 2007). As for the reference category West German, the calculation of the MWTP reduces to –(1=3) = 0.598, which is higher than that of the East Germans, but again not significantly so. This result is surprising at first, as attitudes toward redistribution strongly differ between the two groups. While attitudes for redistribution are significantly stronger for East Germans living in the East, preferences for redistribution seem to be equal. Estimates [2] and [3] control for age and income. Since incomes in the East continue to be lower than in the West, this might explain equality of MWTP values in the face of strongly divergent attitudes. However, even when accounting for age and income in such general terms, East and West Germans fail to exhibit a statistically significant difference in their MWTP for redistribution in excess of the status quo, at least in the year 2012. Contrasting these results with section 5.1, it is obvious that attitudes toward redistribution and preferences for redistribution do not coincide. Replicating the steps of section 5.1, we make particular efforts to control for age and income. See Table A.3 for full estimation results. The MWTP values derived from these estimates and depicted in estimate [2] in Table 5 thus control for age effects, interacting age categories with the East indicator as well as with the main effect attributes ∆RE,∆RE 2 ,∆TC, and ∆TC 2 . Interestingly, there are no significant differences between the MWTPs of East and West Germans within a given age group. The only somewhat larger –yet statistically insignificant –difference can be observed for the youngest age cohort, exhibiting a somewhat higher MWTP for young West Germans compared to their East German peers. However, age cohorts do differ with respect to their MWTP in both groups. For example, setting aside the very sparsely populated age group 1, across the country, age group 2 has the highest MWTP compared to all other groups. When controlling for income (see the aforementioned procedure), the 280 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 18 1represents the parameter estimated for the attribute ∆RE (0.0334), and 3is the estimated parameter for ∆TC (−0.0559). Both values are taken from the estimation presented in Table A.3 in the appendix. The following MWTP values are derived in the same way, taking the estimated parameters from the estimation equations. (Full results for all MWTP estimates are presented in tables A.3, A.4, and A.5 in the appendix.) OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 results are very similar (estimate [3]). Within income categories, no significant differences between East and West Germans exist. Across both parts of the country, similar differences between income categories exist, but no obvious pattern emerges. In summary, when controlling for the pivotal socioeconomic characteristics age and income, no significant differences can be observed between East and West Germans. Table 5 Marginal Willingness to Pay for Redistribution Choice [1] [2] [3] MWTP SE MWTP SE MWTP SE East living in East 0.558 (0.066)*** West Germans 0.598 (0.046)*** Age1 * East 0.060 (0.530) Age1 * West 0.843 (0.378)** Age2 * East 1.197 (0.185)*** Age2 * West 1.102 (0.127)*** Age3 * East 0.581 (0.114)*** Age3 * West 0.532 (0.070)*** Age4 * East 0.378 (0.106)*** Age4 * West 0.435 (0.087)*** Age5 * East 0.218 (0.165) Age5 * West 0.418 (0.101)*** Income1 * East 0.448 (0.166)*** Income1 * West 0.478 (0.102)*** Income2 * East 1.028 (0.162)*** Income2 * West 0.792 (0.142)*** Income3 * East 0.431 (0.097)*** Income3 * West 0.645 (0.114)*** Income4 * East 0.230 (0.162) Income4 * West 0.558 (0.089)*** Income5 * East 0.618 (0.194)*** Income5 * West 0.592 (0.080)*** LL –5,559 –5,495 –5,518 Mc Fadden Adj. R 2 0.095 0.100 0.096 AIC 11,154 11,097 11,144 BIC 11,283 11,485 11,532 N9,720 9,720 9,720 * p < 0.10, ** p < 0.05, *** p < 0.01. The dependent variable choice is a binary variable that takes the value one if the individual chooses the alternative and zero if the individual chooses the status quo. The attributes ∆SP and ∆DE were omitted to avoid perfect collinearity. All attributes are differenced reflecting deviations from the status quo. Standard errors were calculated using the delta method. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. Note: see Appendix A.3 for the full estimation results. East and West Germans’Attitudes Regarding the Welfare State 281 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 While we treat attitudes and preferences as strictly distinctive features, in the economic sense, this does not mean that attitudes do not help explain preferences. Therefore the observed variable inequality reduction from section 5.1 is now added to the estimation equation, with the two categories disagree and totally disagree (which were rarely chosen by respondents, see Figure 2) merged into the variable disagree. This would imply that attitudes, while not equivalent to MWTP values, do matter in predicting them. In addition, the remaining three categories (totally agree,agree, and disagree) are interacted with East living in East, controlling for potential differences between East and West Germans. See Table A.4 for full estimation results. Results in Table 6 show that, within the group strongly agreeing with inequality reduction, there is no significant difference in MWTP between East and West Germans. Furthermore, it is not surprising that this group exhibits the highest MWTP. However, when turning toward the group with a somewhat lower agreement on this issue, the MWTP of West Germans stands out. It is not statistically different from the first group, but significantly higher than the MWTP of the respective East Germans. The MWTP values of the latter are not statistically significantly different from those of the group who disagrees with inequality reduction, which again is uniform across both parts of the country. Therefore attitudes towards redistribution seem to be connected with preferences for redistribution, but West Germans with a mild level of agreement behave very differently from their eastern compatriots. Only East Germans who have a very strong attitude have a high MWTP, while, among West Germans, those who moderately favor redistribution are also willing to contribute. This is puzzling and cannot easily be explained by the communist past. Further research may be guided by Brosig-Koch et al. (2011) and Paetzel, Sausgruber, and Traub (2014). In their laboratory experiments, the former still find differences between East and West Germans, while the latter conclude that an individual’s positive marginal willingness to pay for a reduction of inequality (i.e., an increase of redistribution) is not necessarily conditional on positive payoffs for the individual. The social preference can outweigh the individual disadvantage and shape voting decisions. This goes beyond the scope of this study, but underscores the importance of behaviors such as risk aversion, altruism, and fairness and how they intermingle with psychological concepts such as attitudes –also in combination with economic self-interest as discussed in the light of insurance motives, the Meltzer-Richard model, or the POUM hypothesis. Summarizing, there is no straightforward explanation for the result presented in Table 6. The approximation of preferences on basis of a discrete choice analysis is now well established and –in contrast to measures of attitudes –in line with economic theory. Thus these findings highlight once more that attitudes and preferences are not synonymous and that there is no evidence for a direct link between communism and preferences for redistribution. 282 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table 6 Marginal Willingness to Pay for Redistribution, by Attitudes MWTP SE East: Strongly agree with inequality reduction 0.734 (0.083)*** West: Strongly agree with inequality reduction 0.680 (0.080)*** East: Agree with inequality reduction 0.236 (0.097)** West: Agree with inequality reduction 0.634 (0.059)*** East: Disagree with inequality reduction 0.274 (0.218) West: Disagree with inequality reduction 0.221 (0.143) LL –5,542 Mc Fadden Adj. R 2 0.096 AIC 11,150 BIC 11,387 N9,720 * p < 0.10, ** p < 0.05, *** p < 0.01. The dependent variable choice is a binary variable that takes the value one if the individual chooses the alternative and zero if the individual chooses the status quo. The attributes ∆SP and ∆DE were omitted to avoid perfect collinearity. All attributes are differenced reflecting deviations from the status quo. Standard errors were calculated using the delta method. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. Note: see Appendix A.4 for the full estimation results. 5.3 Robustness Tests We test the validity of our results with a series of robustness checks. In the first part, we elaborate whether socio-economic controls change the results of the estimates. In the second part we look into issues pertaining to the special characteristics of the choice task, e.g., whether respondents understood the choice framework and could deal with the decision situation. The results discussed in section 5.1 on citizens’attitudes toward redistribution might be confounded due to omitted variables. To overcome this issue and to make our results more comparable to the ones of Alesina and Fuchs-Schün- deln (2007), we include a set of socio-demographic variables covering age, family status, education level, labor force status, gross income and experiences with unemployment. Thus eq. [2] in Table 4 is extended by these variables. The descriptive statistics and the results of an ordered probit model are presented in Table 7. All statistically significant controls reduce the probability to favor government’s responsibility for redistribution. These are in particular being married or having higher levels of education and a full-time job. In contrast to other studies, such as Alesina and Fuchs-Schündeln (2007), unemployment does not have any significant effect. East and West Germans’Attitudes Regarding the Welfare State 283 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table 7 Results of Ordered Probit Models for Inequality Reduction inequality reduction [1] Mean SD Coeff. SE East living in East 0.210 0.410 0.282 (0.081)*** East living in West 0.030 0.160 0.062 (0.200) Age 51.690 15.500 0.017 (0.016) Age 2 /100 29.120 16.260 –0.017 (0.015) Married 0.580 0.490 –0.180 (0.106)* Widowed 0.070 0.250 –0.088 (0.178) Divorced 0.130 0.340 –0.074 (0.135) Number of children 1.300 1.130 –0.006 (0.034) Secondary school certificate 0.300 0.460 –0.183 (0.095)* Vocational training 0.210 0.410 –0.275 (0.104)*** University entry diploma 0.120 0.320 –0.246 (0.115)** University degree 0.150 0.350 –0.381 (0.118)*** Full time employed 0.420 0.490 –0.185 (0.094)** Monthly gross income/1,000 1.880 1.930 0.006 (0.028) < 6 months of unemployment 0.080 0.270 0.107 (0.116) 6–24 months of unemployment 0.110 0.310 0.090 (0.108) > 24 months of unemployment 0.080 0.270 0.219 (0.143) LL –1.241 Mc Fadden Adj. R 2 0.003 AIC 2.522 BIC 2.624 N1.215 *p< 0.1, ** p< 0.05, *** p< 0.01; note: The reference category for the different types of schooling comprises individuals with no completed schooling or a lower-level secondary school certificate (Hauptschulabschluss) as the highest educational level reached. The other categories refer to a successful Realschulabschluss,Ausbildung,Abitur, and Universitätsabschluss, respectively. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. We still find a strong and positive effect of individuals with a communist past. East Germans still living in the Eastern part of Germany exhibit stronger attitudes toward redistribution, while their compatriots who migrated west do not differ significantly from their western neighbors. Comparing the magnitude of these results, we find that being from the East increases the probability of 284 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Corneo, G. and H. P. Grüner. 2000. “Social Limits to Redistribution.”American Economic Review 90 (5): 1491–1507. —. 2002. “Individual Preferences for Political Redistribution.”Journal of Public Economics 83 (1): 83–107. Eagly, A. and S. Chaiken. 1996. “Attitude Structure and Function.”In The Handbook of Social Psychology, edited by D. Gilbert, S. Fiske, and G. Lindzey, 269–322. New York: McGraw-Hill Book Co. ESOMAR. 2014. “ESOMAR/WAPOR Guide to Opinion Polls and Published Surveys.” Accessed 13 March, 2018. https://www.esomar.org/uploads/public/knowledge-and- standards/codes-and-guidelines/ESOMAR-WAPOR-Guideline-on-Opinion-Polls- and-Published-Surveys-August-2014.docx. Fong, C. 2006. “Prospective Mobility, Fairness, and the Demand for Redistribution.” Working Paper, Carnegie Mellon University. Accessed 12 March, 2018. https://pdfs. semanticscholar.org/e72a/131c00d17c352b478e8031f581a5968fa0da.pdf. Gegax, D. and L. R. Stanley. 1997. “Validating Conjoint and Hedonic Preference Measures: Evidence from Valuing Reductions in Risk.”Quarterly Journal of Business and Economics 36 (2): 31–54. Guillaud, E. 2013. “Preferences for Redistribution: An Empirical Analysis over 33 Countries.”Journal of Economic Inequality 11 (1): 57–78. Hanemann, M. 1983. “Marginal Welfare Measures for Discrete Choice Models.”Economics Letters 13 (1): 129–36. Hensher, D. A. and W. H. Greene. 2003. “The Mixed Logit Model: The State of Practice.”Transportation 30 (2): 133–76. Hole, A. R. 2007. “A Comparison of Approaches to Estimating Confidence Intervals for Willingness to Pay Measures.”Health Economics 16 (8): 827–40. Hosmer, D. W. and S. Lemeshow. 2000. Applied Logistic Regression. 2nd ed. New York: Wiley. Isaksson, A.-S. and A. Lindskog. 2009. “Preferences for Redistribution –A Country Comparison of Fairness Judgements.”Journal of Economic Behavior & Organization 72 (3): 884–902. Johnson, R. F. and W. H. Desvousges. 1997. “Estimating Stated Preferences with Rated- Pair Data: Environmental, Health, and Employment Effects of Energy Programs.” Journal of Environmental Economics and Management 34 (1): 78–99. Kahneman, D., I. Ritov, and D. Schkade. 1999. “Economic Preferences or Attitude Expression? An Analysis of Dollar Responses to Public Issues.”Journal of Risk and Uncertainty 19 (1–3): 203–35. Kuhfeld, W. F., R. D. Tobias, and M. Garratt. 1994. “Efficient Experimental Design with Marketing Research Applications.”Journal of Marketing Research XXXI: 545–57. Kuhn, A. 2013. “Inequality Perceptions, Distributional Norms, and Redistributive Preferences in East and West Germany.”German Economic Review 14 (4): 483–99. East and West Germans’Attitudes Regarding the Welfare State 291 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Lancaster, K. J. 1966. “A New Approach to Consumer Theory.”Journal of Political Economy 74 (2): 132–57. Lancsar, E. and J. Louviere. 2006. “Deleting ‘Irrational’Responses from Discrete Choice Experiments: A Case of Investigating or Imposing Preferences?”Health Econ 15 (8): 797–811. Lancsar, E., J. J. Louviere, and T. N. Flynn. 2007. “Several Methods to Investigate Relative Attribute Impact in Stated Preference Experiments.”Social Science & Medicine 64 (8): 1738–53. Long, J. S. 1997. Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: Sage Publ. Louviere, J. J. and D. A. Hensher. 1982. “Design and Analysis of Simulated Choice or Allocation Experiments in Travel Choice Modelling.”Transportation Research Record 890: 11–17. Louviere, J. J. and D. J. Street. 2000. “Stated-Preference Methods.”In Handbook of Transport Modelling, edited by D. A. Hensher and K. Button, 131–43. New York: Pergamon. Louviere, J. J. and G. Woodworth. 1983. “Design and Analysis of Simulated Consumer Choice or Allocation Experiments: An Approach Based on Aggregated Data.”Journal of Marketing Research 20 (4): 350–67. Luttmer, E. F. P. and M. Singhal. 2011. “Culture, Context, and the Taste for Redistribution.”American Economic Journal: Economic Policy 3 (1): 157–79. Manski, C. F. 1977. “The Structure of Random Utility Models.”Theory and Decision 8: 229–54. McFadden, D. 1974. “Conditional Logit Analysis of Quantitative Choice Behavior.”In Frontiers in econometrics, edited by P. Zarembka, 105–42. New York: Academic Press. —. 1981. “Econometric Models of Probabilistic Choice.”In Structural Analysis of Discrete Data with Econometric Applications, edited by C. F. Manski and D. McFadden, 198–272. Cambridge, MA: MIT Press. —. 2001. “Economic Choices.”American Economic Review 91 (3): 351–78. McFadden, D. and K. E. Train. 2000. “Mixed MNL Models for Discrete Response.” Journal of Applied Econometrics 15 (5): 447–70. Meltzer, A. H. and S. F. Richard. 1981. “A Rational Theory of the Size of Government.” Journal of Political Economy 89 (5): 914–27. Mørkbak, M. R., S. B. Olsen, S. Denver, and D. Campbell. 2012. “Will Providing a Real Economic Incentive Make Respondents Behave Differently? An Empirical Choice Experiment Investigation.”Presentation at the 2nd Danish Choice Modelling Days, University of Southern Denmark. Neustadt, I. 2011. “Do Religious Beliefs Explain Preferences for Income Redistribution? Experimental Design.”CESifo Economic Studies 57 (4): 623–52. 292 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Neustadt, I. and P. Zweifel. 2010. “Economic Well-Being, Social Mobility, and Preferences for Income Redistribution: Evidence from a Discrete Choice Experiment.” Working Paper 0909, Sozialökonomisches Institut, Universität Zürich. Accessed 12 March, 2018. https://www.econstor.eu/handle/10419/76213. —. 2015. “Is the Welfare State Sustainable? Experimental Evidence on Citizens’Preferences for Redistribution.”International Journal of Economics and Finance 7 (6): 60–79. Ockenfels, A. and J. Weimann. 1999. “Types and Patterns: An Experimental East-West- German Comparison of Cooperation and Solidarity.”Journal of Public Economics 71: 275–87. Paetzel, F., R. Sausgruber, and S. Traub. 2014. “Social Preferences and Voting on Reform: An Experimental Study.”European Economic Review 70: 36–55. Pekelman, D. and S. K. Sen. 1979. “Measurement and Estimation of Conjoint Utility Functions.”The Journal of Consumer Research 5 (4): 263–71. Pfarr, C. 2012. “Meltzer-Richard and Social Mobility Hypothesis: Revisiting the Income-Redistribution Nexus Using German Choice Data.”Discussion Paper. 10–12, University of Bayreuth. Accessed 12 March, 2018. https://eref.uni-bayreuth.de/8228/. —. 2013. Einkommen, Mobilität und individuelle Präferenzen für Umverteilung: Ein Discrete-Choice-Experiment. Beiträge zur Finanzwissenschaft, Vol. 30. Tübingen: Mohr Siebeck. Pfarr, C. and A. Schmid. 2016. “Redistribution Through Social Health Insurance: Evidence on Citizen Preferences.”European Journal of Health Economics 17 (5): 611– 28. Pfarr, C., A. Schmid, and M. R. Mørkbak. 2016. “Modelling Heterogeneous Preferences for Income Redistribution –An Application of Continuous and Discrete Distributions.”Review of Income and Wealth. Accessed 12 March, 2018. http://onlinelibrary. wiley.com/doi/10.1111/roiw.12282/abstract. Phillips, K. A., R. F. Johnson, and T. Maddala. 2002. “Measuring What People Value: A Comparison of ‘Attitude’and ‘Preference’Surveys.”Health Services Research 37 (6): 1659–79. Phillips, K. A., T. Maddala and R. F. Johnson. 2002. “Measuring Preferences for Health Care Interventions Using Conjoint Analysis: An Application to HIV Testing.” Health Services Research 37 (6): 1681–1705. Rainer, H. and T. Siedler. 2008. “Subjective Income and Employment Expectations and Preferences for Redistribution.”Economic Letters 99 (3): 449–53. Roberts, K. W. S. 1977. “Voting Over Income Tax Schedules.”Journal of Public Economics 8 (3): 329–40. Romer, T. 1975. “Individual Welfare, Majority Voting, and the Properties of a Linear Income Tax.”Journal of Public Economics 4 (2): 163–85. Sennhauser, M. 2010. “Why the Linear Utility Function is a Risky Choice in Discrete- Choice Experiments.”Working Paper. 1014, Socioeconomic Institute, University of Zurich. Accessed 12 March, 2018. https://www.econstor.eu/handle/10419/76191. East and West Germans’Attitudes Regarding the Welfare State 293 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Statistisches Bundesamt. 2011. Statistisches Jahrbuch 2011. Wiesbaden. Stichnoth, H. 2012. “Does Immigration Weaken Natives’Support for the Welfare State? Evidence from Germany.”Public Choice 151 (3–4): 631–54. Suhrcke, M. 2001. “Preferences for Inequality: East vs. West.”Innocenti Working Paper. 89, UNICEF Innocenti Research Centre. Accessed 12 March, 2018. https://www. unicef-irc.org/publications/324-preferences-for-inequality-east-vs-west.html. Telser, H. 2002. Nutzenmessung im Gesundheitswesen: Die Methode der Discrete- Choice-Experimente. Hamburg: Kovač. Train, K. E. 2009. Discrete Choice Methods with Simulation. 2nd ed. New York: Cambridge University Press. Vroomen, J. M. and P. Zweifel. 2011. “Preferences for Health Insurance and Health Status: Does it Matter Whether You Are Dutch or German?”European Journal of Health Economics 12 (1): 87–95. Appendix A. Tables Table A.1 Choice Characteristics, Attribute Levels and their Frequencies means of choice characteristics frequencies of attribute levels in percent all chosen 5 10 15 20 25 30 35 40 45 …75 80 85 90 TC 30 27 20 21 21 16 22 RE 32 31 16 19 20 23 22 RI 38 38 36 36 28 SP 35 35 33 37 30 UL 9 9 45 27 28 FC 11 11 34 32 18 16 WP 7 7 53 47 DE 83 83 16 36 26 22 WE 7 7 52 48 OT 10 9 35 31 34 N9,720 3,302 Note: in the first part of this table, the first column shows the means of the choice characteristics, and the second column presents mean values of the choice characteristics from the chosen alternatives only. In the second part, frequencies of the attribute levels are given in percentages; bold values indicate the levels of the status quo. 294 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table A.2 Estimation Results for MWTP for Redistribution choice [1] RE Probit Coeff SE TC –0.0559 (0.0023)*** TC 2 –0.0013 (0.0002)*** RE 0.0334 (0.0030)*** RE 2 –0.0009 (0.0003)*** RI 0.0145 (0.0043)*** WP –0.0219 (0.0077)*** UL –0.0197 (0.0047)*** FC 0.0107 (0.0047)** WE –0.0415 (0.0061)*** OT –0.0389 (0.0037)*** OT 2 0.0048 (0.0014)*** _cons (status quo) –0.3188 (0.0433) East living in East ×∆TC –0.0031 (0.0041) ×∆TC 2 –0.0001 (0.0003) ×∆RE –0.0005 (0.0053) ×∆RE 2 0.0007 (0.0005) MWTP East living in East 0.558 (0.066) † *** West Germans 0.598 (0.046) † *** N9,720 LL –5,559 McFadden Adj. R 2 0.095 AIC 11,244 BIC 11,283 * p < 0.10, ** p < 0.05, *** p < 0.01. The dependent variable choice is a binary variable that takes the value one if the individual chooses the alternative and zero if the individual chooses the status quo. The attributes ∆SP and ∆DE were ommited to avoid perfect collinearity. All attributes are differenced reflecting deviations from the status quo. †Standard errors were calculated using the delta method. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. East and West Germans’Attitudes Regarding the Welfare State 295 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table A.3 Estimation Results for MWTP for Redistribution, by Age and Income choice choice age [1] income [2] RE Probit RE Probit Coeff. SE Coeff. SE ∆TC –0.0656 (0.0060) *** ∆TC –0.0509 (0.0051) *** ∆TC 2 –0.0016 (0.0004) *** ∆TC 2 –0.0009 (0.0004) ** ∆RE 0.0274 (0.0075) *** ∆RE 0.0329 (0.0067) *** ∆RE 2 0.0004 (0.0006) ∆RE 2 –0.0015 (0.0006) ** ∆RI 0.0144 (0.0043) *** ∆RI 0.0144 (0.0043) *** ∆WP –0.0217 (0.0078) *** ∆WP –0.0220 (0.0078) *** ∆UL –0.0196 (0.0047) *** ∆UL –0.0198 (0.0047) *** ∆FC 0.0106 (0.0047) ** ∆FC 0.0106 (0.0047) ** ∆WE –0.0416 (0.0062) *** ∆WE –0.0423 (0.0062) *** ∆OT –0.0398 (0.0038) *** ∆OT –0.0396 (0.0038) *** ∆OT 2 0.0049 (0.0014) *** ∆OT 2 0.0048 (0.0014) *** _cons (status quo) –0.2709 (0.0764) *** _cons (status quo) –0.2352 (0.0700) *** Age1 Income1 ×∆TC 0.0286 (0.0125) ** × ∆TC –0.0079 (0.0075) ×∆TC 2 0.0014 (0.0010) ×∆TC 2 –0.0003 (0.0005) ×∆RE 0.0038 (0.0174) ×∆RE –0.0047 (0.0095) ×∆RE 2 –0.0018 (0.0015) ×∆RE 2 0.0008 (0.0008) Age2 Income2 ×∆TC 0.0231 (0.0073) *** × ∆TC 0.0090 (0.0072) ×∆TC 2 0.0006 (0.0005) ×∆TC 2 –0.0000 (0.0005) ×∆RE 0.0194 (0.0095) ** × ∆RE 0.0004 (0.0095) ×∆RE 2 –0.0025 (0.0008) *** × ∆RE 2 –0.0001 (0.0008) Age3 Income4 ×∆TC –0.0005 (0.0074) ×∆TC –0.0087 (0.0070) ×∆TC 2 –0.0001 (0.0005) ×∆TC 2 –0.0002 (0.0005) ×∆RE 0.0078 (0.0092) ×∆RE 0.0004 (0.0091) ×∆RE 2 –0.0011 (0.0008) ×∆RE 2 0.0009 (0.0008) Age4 Income5 ×∆TC 0.0078 (0.0074) ×∆TC –0.0156 (0.0071) ** ×∆TC 2 0.0002 (0.0005) ×∆TC 2 –0.0010 (0.0005) ** ×∆RE –0.0022 (0.0094) ×∆RE 0.0065 (0.0091) ×∆RE 2 –0.0010 (0.0008) ×∆RE 2 0.0011 (0.0008) East living in East East living in East ×∆TC 0.0060 (0.0104) ×∆TC –0.0289 (0.0088) *** ×∆TC 2 0.0005 (0.0007) ×∆TC 2 –0.0009 (0.0006) ×∆RE –0.0144 (0.0130) ×∆RE 0.0015 (0.0110) ×∆RE 2 0.0001 (0.0010) ×∆RE 2 0.0003 (0.0009) ×Age1 ×Income1 ×∆TC –0.0160 (0.0277) ×∆TC 0.0271 (0.0138) ** ×∆TC 2 0.0008 (0.0019) ×∆TC 2 –0.0000 (0.0009) ×∆RE –0.0139 (0.0327) ×∆RE –0.0025 (0.0172) ×∆RE 2 0.0024 (0.0026) ×∆RE 2 0.0016 (0.0013) 296 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 choice choice age [1] income [2] RE Probit RE Probit Coeff. SE Coeff. SE ×Age2 ×Income 2 ×∆TC –0.0103 (0.0132) ×∆TC 0.0265 (0.0118) ** ×∆TC 2 –0.0004 (0.0009) ×∆TC 2 0.0008 (0.0008) ×∆RE 0.0235 (0.0173) ×∆RE 0.0109 (0.0153) ×∆RE 2 –0.0001 (0.0013) ×∆RE 2 0.0006 (0.0012) ×Age3 ×Income4 ×∆TC –0.0013 (0.0128) ×∆TC 0.0318 (0.0126) ** ×∆TC 2 0.0001 (0.0009) ×∆TC 2 0.0014 (0.0009) ×∆RE 0.0149 (0.0162) ×∆RE –0.0217 (0.0160) ×∆RE 2 0.0009 (0.0013) ×∆RE 2 0.0008 (0.0013) ×Age4 ×Income 5 ×∆TC –0.0214 (0.0134) ×∆TC 0.0253 (0.0159) ×∆TC 2 –0.0020 (0.0009) ** × ∆TC 2 0.0010 (0.0011) ×∆RE 0.0170 (0.0166) ×∆RE 0.0025 (0.0200) ×∆RE 2 0.0006 (0.0013) ×∆RE 2 0.0009 (0.0015) N9,720 9,720 LL –5,495 –5,518 McFadden Adj. R 2 0.100 0.096 AIC 11,097 11,144 BIC 11,485 11,532 * p < 0.10, ** p < 0.05, *** p < 0.01. The dependent variable choice is a binary variable that takes the value one if the individual chooses the alternative and zero if the individual chooses the status quo. The attributes ∆SP and ∆DE were ommited to avoid perfect collinearity. All attributes are differenced reflecting deviations from the status quo. † Standard errors were calculated using the delta method. AIC is the Akaike information criterion. BIC is the Bayesian information criterion. LL refers to the log likelihood. Table A.4 Estimation Results for MWTP for Redistribution, by Attitudes choice attitudes RE Probit Coeff. SE ∆TC –0.0529 (0.0064) *** ∆TC 2 –0.0009 (0.0004) ** ∆RE 0.0117 (0.0083) ∆RE 2 0.0001 (0.0006) ∆RI 0.0146 (0.0043) *** ∆WP –0.0214 (0.0077) *** ∆UL –0.0196 (0.0047) *** ∆FC 0.0109 (0.0047) ** ∆WE –0.0421 (0.0061) *** ∆OT –0.0389 (0.0037) *** Table continued next page East and West Germans’Attitudes Regarding the Welfare State 297 Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 Table A4 continued choice attitudes RE Probit Coeff. SE ∆OT 2 0.0047 (0.0014) *** _cons (status quo) –0.3084 (0.0398) *** Strongly agree with inequality reduction ×∆TC 0.0029 (0.0073) ×∆TC 2 –0.0005 (0.0005) ×∆RE 0.0223 (0.0095) ** ×∆RE 2 –0.0011 (0.0007) Agree with inequality reduction ×∆TC –0.0082 (0.0071) ×∆TC 2 –0.0004 (0.0005) ×∆RE 0.0270 (0.0092) *** ×∆RE 2 –0.0012 (0.0007) * East living in East ×∆TC –0.0046 (0.0124) ×∆TC 2 –0.0005 (0.0009) ×∆RE 0.0040 (0.0160) ×∆RE 2 0.0012 (0.0013) ×Strongly agree with inequality reduction ×∆TC –0.0100 (0.0138) ×∆TC 2 0.0002 (0.0010) ×∆RE 0.0094 (0.0178) ×∆RE 2 –0.0007 (0.0014) ×Agree with inequality reduction ×∆TC 0.0128 (0.0139) ×∆TC 2 0.0010 (0.0010) ×∆RE –0.0249 (0.0180) ×∆RE 2 0.0001 (0.0014) N9,720 LL –5,542 McFadden Adj. R 2 0.096 AIC 11,150 BIC 11,387 * p < 0.10, ** p < 0.05, *** p < 0.01. The dependent variable choice is a binary variable that takes the value one if the individual chooses the alternative and zero if the individual chooses the status quo. The attributes ∆SP and ∆DE were ommited to avoid perfect collinearity. All attributes are differenced reflecting deviations from the status quo. AIC is the Akaike information criterion, BIC is the Bayesian information criterion. LL refers to the log likelihood. † Standard errors were calculated using the delta method. 298 Christian Pfarr, Andreas Schmid, and Volker Ulrich Journal of Contextual Economics 137 (2017) 3 OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57 B. Figures East and West Germans’Attitudes Regarding the Welfare State 299 Journal of Contextual Economics 137 (2017) 3 Figure B.1: Example of a Choice Situation OPEN ACCESS | Licensed under | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/jce.137.3.261 | Generated on 2023-01-16 13:33:57