Population aging and its profound consequences puts the sustainability of pension systems in industrialized Western democracies at risk. Almost all pensions systems in these countries are primarily based on a pay-as-you-go (PAYG) principle. Hence, changes in pension benefits or contribution rates are immediately felt by the participants of the system. Any change in a pension system will thus involve winners and losers. Pension reforms are obviously a very interesting and difficult to analyze policy field because they involve policy changes that don't constitute mere Pareto improvements nor do they easily allow the implementation of simple compensation schemes for the distributional losers of a reform. To model the impact of population aging on pension systems and their reform thus requires a profound understanding of voters' preferences and of the political environment in which such an endeavor is undertaken. This holds for any type of pension reform, regardless of whether the reform is parametric (i.e.\ changes in the contribution rate, benefit levels or eligibility rules) or non-parametric (partial or full transitions to prefunded systems). Building on the current state of research on pension systems and their reform, this dissertation sets out to, first, review existing political economy models showing that pension policy is mainly a political problem. Second, it will prove that any pension reform is by and large a redistributional policy shift. Third and foremost, this dissertations develops a political economy model of a pension system that explicitly considers the political preferences of different age groups and also takes into account some of the underlying political incentives generated by the size of the pension system and the political environment. Hence, pension policy is examined not in isolation but in view of the fiscal trade-offs made with respect to other public policy goods. The resulting simple overlapping- generations model allows for an easy derivation of the policy preferences of different age groups. This alone, however, would be insufficient when analyzing the prospects of a pension reform, since individual preferences need to be aggregated through an electoral process. The existing political economy literature has mainly analyzed this aggregation process by assuming the existence of a direct democracy with popular referenda or by analyzing rather simple settings of representative democracy, without properly taking into account the impact of different types of electoral institutions. This dissertation sets out to address this very shortcoming. In the research tradition of modern political economy, it attempts to offer a theoretical analysis by focusing on voters' policy preferences and the way in which these preferences are aggregated through domestic electoral institutions. This focus allows to derive different scenarios for the feasibility and direction of pension reform. Finally, some hypotheses derived from the model are tentatively tested using cross-national survey data of 21 countries provided by the International Social Survey Programme. Following the introductory chapter that puts this thesis into a theoretical and empirical context, chapter 2 examines the premise that population aging has severe economic consequences and strains existing pension systems. This is empirically fleshed out by reviewing current demographic and economic long term projections provided, among others, by the United Nations and the European Commission. It is shown that aging has not only severe domestic and international macroeconomic consequences but also leads to strong imbalances in current pension systems. As a result, in many countries the present expected value of future benefits is by no means matched by the present expected value of future revenues. Chapter 3 reviews the existing formal political economy literature on the existence and size of public pension systems, thereby also introducing the concept of overlapping-generation models and explaining the necessary notation. This review shows that pension systems can only be understood by looking at the politics involved. Assuming, as part of the literature does, that pension policy is decided in a direct democratic referendum represents a valuable first attempt in understanding the underlying political processes. These kinds of models show that coalition building might be important. The literature has suggested that retirees and older workers may conspire, and that an elaborated inter-generational punishment mechanism ensures sustainability of the system once it has been introduced. Alternatively, it has been proposed that intra-generational inequality leads poor workers to team up with pensioners to push for public pension provision. Another factor could be the existence of altruism that induces workers to consider the well- being of the old, which in turn affects their voting behavior. The chapter then shows that although being a valuable first step, the explanatory power of these approaches is not fully satisfactory. Their Achilles' heel is the assumption of a direct democracy. Clearly, pension politics in Western countries takes place within the confines of a representative democracy. Some models have explored this possibility. In one approach, more than one party could wield influence by using the threat of a veto. Even more importantly, workers and retirees might influence policy-making beyond the individual act of voting. Hence, pensioners could be able to gain privileged access to a government through lobbying and campaigning. If their thus defined political power is greater than the power of workers, they may be able to push for intergenerational transfers against the will of the young. One possible explanation for the greater political leverage of the old could come from the fact that they are more 'single-minded' and thus ideologically more homogenous. Employing a probabilistic voting model suggests that in this case electoral competition will induce office-motivated political parties to favor the old in their electoral programmes. However, beyond these explanations, political modeling has not gone much further in examining the political rationale of public pension provision in a representative democracy. In particular, despite the use of probabilistic voting models, the possible impact of different electoral institutions has not been sufficiently analyzed in the literature. Chapter 4 then clarifies what I mean by 'pension reform'. It turns out that any change to a pension scheme, be it parametric or non- parametric, is redistributional. Any reform will involve winners and loser and is therefore a politically very contested issue. As a result, discussions about whether to completely replace a PAYG system with a prefunded scheme are often misleading. Neither of these two systems Pareto-dominates the other. Transition from one to the other is always redistributional and both types of schemes can be reformed to deal with the consequences of aging. Chapter 5 first develops a three-period overlapping-generations model to derive policy preferences of individuals with respect to a public PAYG pension scheme, taking the preferences for expenditures on other policy goods into account. The results of the model are that, (1), preferences for a big public PAYG system increase with age; and, (2), older workers policy preferences depend on population growth and on the size of the existing pension scheme. Based on these results, I then proceed to examine the political process. If voters' preferences are aggregated through a direct referendum, then the prevailing population growth rate is decisive, for it determines which age group contains the pivotal median voter. If the median was a young worker, the public scheme would be completely replaced by a prefunded system of individual saving. If retirees were in a majority, the PAYG scheme would be preserved and any fiscal imbalance in the system would be resolved through changes in the contribution rate. Finally, if the median voter was an old worker, then reform outcomes depend on the size of the existing pension scheme. If it is generous, old workers are in favor of keeping it. Depending on its prevailing size relative to what old workers consider as optimal, they may favor a reduction in its overall generosity in favor of more prefunding. If the existing PAYG system is small, old workers prefer its complete replacement with a prefunded system of private saving. If policy preferences are aggregated within the framework of a representative democracy, then the electoral system becomes important. This is analyzed in the framework of a probabilistic voting model. Now the number of swing voters turns out to be the decisive variable. Under proportional representation, the group with the highest number of swing voters will find their policy preferences catered to by the parties vying for office. This need not be the case under majoritarian elections though. Here the number of swing voters in the swing district becomes crucial. Therefore, under certain conditions a pension reform to reduce the PAYG pillar and increase individual prefunding may be feasible under proportional representation but not under majoritarian elections. The latter therefore may allow smaller groups (i.e.\ groups with fewer politically important swing voters) to determine the pension policies that parties propose. Chapter 6 explains that, due to data limitations, a thorough econometric test of the reform scenarios and the impact of electoral institutions is not feasible. This chapter does test the two hypotheses related to individual policy preferences, however. Employing logit and ordered-logit analyses of cross-national survey data of 21 countries show that, as hypothesized, age is positively associated with preferences for more public PAYG pension spending. Retirees are much more likely to favor large pension systems than young workers. As predicted by the model, old workers' preferences are more likely to be aligned with the interests of pensioners in countries with generous public schemes and, regardless of the existing size, when population growth is positive. In countries with negative population growth rates, the odds are much higher that old workers join young workers in preferring a smaller pension system. With respect to pension reform, these results suggest that retirees are the most likely to oppose cuts in public pensions, whereas young workers are the most likely to favor such changes. The attitude of old workers ultimately depends on the degree of population aging and the size of the existing pension system. As a result, the main implications of the model are tentatively confirmed by the data. Finally, chapter 7 summarizes premises, hypotheses and results of the thesis. It also offers some thoughts on future avenues of research and future developments of pension systems in industrialized countries.
Die Alterung der Gesellschaften in westlichen Industrieländern erhöht den Anpassungsdruck für die bestehenden umlagefinanzierten Rentensysteme. Da diese Form von Alterssicherungssystem die aktive mit der in Ruhestand befindlichen Generation verbindet, hat jede mögliche Reform unmittelbare verteilungspolitische Auswirkungen. Rentenreformen sind somit ein analytisch besonders interessantes Politikfeld, da sie keine Pareto-Verbesserung darstellen, sondern es immer verteilungspolitische Gewinner und Verlierer gibt. Dies gilt sowohl für parametrische wie auch nicht-parametrische Reformen. Trotz eines recht gleichförmigen Anpassungdruckes auf die Alterssicherungssysteme westlicher Industrieländer können wir jedoch unterschiedliche Anpassungsreaktionen beobachten, denn Richtung, Ausmaß und politisch erfolgreiche Durchführung von Rentenreformen unterscheiden sich erheblich. Ziel dieser Dissertation ist es, aus der Perspektive der modernen Politischen Ökonomie einen Erklärungsbeitrag zu liefern, unter welchen Bedingungen Reformen eher möglich sind und welche Formen der Anpassung zu erwarten sind. Nach einer kurzen empirischen Illustration der ökonomischen und finanziellen Konsequenzen von gesellschaftlichen Alterungsprozessen, führt die Arbeit zunächst in die bestehende politik-ökonomische Literatur ein. Es zeigt sich in dieser Forschung deutlich, dass Politik eine entscheidende Rolle für die Erklärung von Existenz und Größe von Rentensystemen spielt. Es wird jedoch auch deutlich, dass der politische Aggregationsprozess individueller Wählerpräferenzen größtenteils vernachlässigt oder sehr vereinfacht dargestellt wird. Besonders der Einfluss von elektoralen Institutionen ist bisher nicht ausreichend gewürdigt worden. In einem nächsten Schritt wird gezeigt, dass jede Art der Reform redistributiv ist und somit neben den Gewinnern immer auch Umverteilungsverlierer produzieren wird. Kern der Dissertation ist anschließend die Entwicklung eines dreistufigen Modells überlappender Generationen (OLG). Dieses erlaubt die formale Analyse individueller politischer Präferenzen. Es zeigt sich, dass politische Präferenzen für ein großzügiges umlagefinanziertes Rentensystem mit dem individuellen Lebensalter zunehmen. Als besonders interessant erweisen sich die Ergebnisse für ältere Arbeitnehmer. Deren Zustimmung oder Ablehnung eines staatlichen Umlagesystems hängen vom Grad der gesellschaftlichen Alterung und der Größe des bestehenden Rentensystems ab. Nach der Analyse der Wählerpräferenzen wendet sich die Arbeit dem politischen Aggregationsprozess zu. Während in einer direkten Demokratie das Alter des Medianwählers ausschlaggebend ist, tritt in repräsentativen Demokratien die Zahl der Wechselwähler (swing voters) in den Vordergrund. Anhand eines probabilistischen Wahlmodells lässt sich zeigen, dass unter bestimmten Bedingungen die Art des Wahlsystems einen großen Einfluss auf die Realisierung von Rentenreformen haben kann. Eine vergleichende Analyse von Mehrheits-- und Verhältniswahlsystemen ergibt, dass in bestimmten Konstellationen große, nicht-paramtetrische Reformen in Verhältniswahlsystemen wahrscheinlicher sein könnten. Die Arbeit schließt mit einigen empirschen Überlegungen und Tests. Die theoretischen Vorhersagen bezüglich des politischen Aggregationsprozesses und dessen Auswirkungen auf Rentenreformen lassen sich nicht statistisch prüfen. Jedoch können die Hypothesen zu individuellen Reformpräferenzen anhand internationaler Surveydaten ökonometrisch getestet werden. Unter Verwendung von Umfragedaten aus 21 westlichen Industrieländern lassen sich die Hypothesen mit Hilfe von Logit- und ordinalen Logitschätzungen tentativ bestätigen. Lebensalter, Alterungsprozess und Größe eines bestehenden umlagefinanzierten Rentensystems haben einen statistisch belegbaren Einfluss auf individuelle Reformpräferenzen.
