1 Introduction

How to explain developments in social policy is a matter of longstanding debate. As Baldwin noted in 1990 (36), “[s]cores of theories compete to explain why it [the welfare state, FN/JK] exists at all, dozens of comparative analyses account for its variations, legions of narratives detail how individual examples contradict or confirm general hypotheses”. Not surprisingly, more than 30 years later social policy research continues to be distinguished by its plurality of approaches. Many of them have been developed to capture developments in OECD welfare states. Yet, since social policy scholars are now turning more to researching the global dynamics of social policy, the question of to what extent these existing social policy approaches can be applied to countries and regions beyond the OECD world comes to the fore (Kpessa and Béland 2013; Lavers and Hickey 2016; Veit et al. 2017; Schmitt et al. 2020). At the same time, aspects that have long been less evident in social policy research become more visible when studying the global dynamics of social policy, for example the role of transnational policymaking (Schmitt 2020; Leisering 2019) or autocracies (Mares and Carnes 2009; Knutsen and Rasmussen 2018; Eibl 2020). These approaches can, in turn, also inform analyses of countries that are known to be the “usual suspects” in social policy research. As a result, the list of approaches for social policy research continues to become longer.

When studying the global dynamics of social policy through qualitative case studies, this plurality can come as a challenge: How can we—in a meaningful way—compare social policy developments in very different countries and regions, relating findings from such arguably different countries as, for example, Uganda, Turkey, and Malaysia, or Bolivia and South Korea? How can we compare developments within old-age provision to those in employment policy? And is it possible to liken developments from the nineteenth century or colonial times to social policies in the twenty-first century? Which analytical level is suitable for such endeavours? Given this background, this edited volume seeks to find new ways of explaining social policy. It introduces causal mechanisms as the key concept of such a new explanatory approach. The key argument that we present throughout this book is that causal mechanisms can generate explanations that can complement, expand, deepen, and, in some cases, even correct analyses that rely on established approaches to social policy.

Causal mechanisms facilitate the identification of causal chains and patterns, but not at the level of political systems, regimes, structures, policy fields, or long-term processes. Rather, causal mechanisms trace smaller regularities at the level of sequences, which can be observed in political processes at different times, in different places, and on different topics. Importantly, these kinds of regularities enable a more than descriptive access to the development of social policy. To put it in the words of Charles Tilly: “great social regularities do not occur at the level of whole structures, full sequences, or total processes but in the detailed social mechanisms that generate structures, sequences, and processes” (Tilly 2008, 123). Thus, the conceptual advantage of causal mechanisms is that they provide the basis for a decomposed form of comparison at sub-levels of political processes, since processes can be broken down into smaller units and sequences can be analysed in a far more differentiated manner. What this mechanism-based approach does not imply, importantly, is that the mechanisms identified need to be integrated afterwards into one particular approach to social policy.

A focus on causal mechanisms can thus especially (but not only) promote qualitative approaches in social policy research. A quantitative research strategy has been found to be attractive for researchers because of its use of extremely sparse ontological assumptions about the social and political world. Whatever factor might play a role in explaining social policies is turned into a variable that can be measured at different scales. This contributes to the extreme variability and adaptability of quantitative research, but interpreting its results coherently becomes difficult. For qualitative research, which draws its momentum from a critique of the quantitative, probabilistic, and correlation-based approach (George and Bennett 2005; Goertz and Mahoney 2012), imitating such a theoretical and ontological approach can become a serious obstacle to further developing social policy research.

Until now, research on social policy has only rarely been combined with an approach that explicitly uses causal mechanisms as a theoretical or methodological concept. Diffusion research has relied on a well-known set of mechanisms to analyse social policy developments (Obinger et al. 2013), but more often than not these mechanisms are not clear when it comes to the key requirement of mechanism-based research, namely to detail the basic elements of a causal process and how they are linked to produce an outcome (Hedström and Ylikoski 2010). Similar points of critique could be raised for mechanisms of gradual institutional change (Streeck and Thelen 2005). In recent years, however, mechanism-based research has gained traction. Obinger, Petersen, and Starke (2018) focus on a set of mechanisms that explain social policy expansion in the context of war, thereby distinguishing the phases of war preparation, mobilisation, and the post-war period. Leisering (2019) introduces the mechanisms of “cultural linkages”, “theorisation”, and “quantification” as three mechanisms of global social policy diffusion. And a recent issue of Social Policy & Administration applies a mechanism-based approach to studying transnational social policy dynamics in the Global South (for an overview see Kuhlmann and ten Brink 2021). We argue that there is great potential for social policy research in continuing such efforts, especially for comparisons between case studies of historically widely divergent situations, different fields of social policy, and very different countries.

In this introductory chapter, we will present an approach to causal mechanisms that is modular, process-oriented, and actor-centred. It guides the analysis in this book’s chapters. We proceed as follows: First, we root our approach to mechanism-based research more broadly in the history of causal mechanisms as a social sciences concept by distinguishing four central strands of research. Subsequently, we present our approach to causal mechanisms which includes the idea of modularisation as well as a distinction between elementary and complex causal mechanisms. Thereafter we discuss the question of what social policy research can gain from a mechanism-based approach, further detailing our key argument that causal mechanisms can improve social policy research in a number of ways. Finally, we give a brief overview of the chapters in this volume.

2 Causal Mechanisms as a Concept in the Social Sciences

Contributions to the literature on causal mechanisms usually include a definition of a causal mechanism, and the growing literature on the topic has resulted in many—sometimes complementary, sometimes competing—attempts to define the term. In fact, it is striking that the existing literature on causal mechanisms by no means offers one single definition of the term. Already in 2001, Mahoney listed 24 different definitions of causal mechanisms (Mahoney 2001, 579), and extracted 3 different understandings of the term (Mahoney 2001, 578–82). In their review article on causal mechanisms from 2010, Hedström and Ylikoski (2010) listed nine “alternative definitions”. While diverse definitions of single concepts can be considered a scientific problem, it is important to note that a standardised definition of causal mechanisms would not solve all of the (admittedly many) challenges that conceptual work on causal mechanisms is struggling with. In a similar vein, Hedström and Swedberg have argued that “it is not so much the definition per se that is important, as the type and style of theorizing it encourages” (1996, 299). In the following, we will outline the history of the concept of causal mechanisms in the social sciences (for an even broader history, see Glennan and Illari 2018). This will help us to elaborate on the main theoretical options that we have when conducting mechanism-based research, and to lay the foundation for further advancing the conceptual debate. Looking at the multitude of reflections on social and causal mechanisms, we will differentiate between four strands of the discussion since the 1970s, before situating ourselves within this debate.

2.1 First Strand: A Methodology of Qualitative Research

The concepts of “causal mechanism” and “process tracing” delineate more recent debates in political science on a specific methodology of qualitative research (in particular: Beach and Pedersen 2019; Bennett and Checkel 2015; Goertz 2017; Goertz and Mahoney 2012). The main rationale for creating a new qualitative research methodology was the attempt to refute an original, case study-oriented approach to research, which was prominently articulated in Designing Social Inquiry (King et al. 1994). Here, Gary King, Robert O. Keohane, and Sidney Verba postulated one causal inference logic for both quantitative and qualitative research. Studies that do not follow this universal logic would not be—so the not only implicit reasoning—valuable in scientific terms. The “imperialism” of such an approach triggered, with a certain time lag, an intense counter-movement (Brady and Collier 2004; George and Bennett 2005; Gerring 2006; Ragin 2000), which finally resulted in the book A Tale of Two Cultures. Qualitative and Quantitative Research in the Social Sciences (Goertz and Mahoney 2012) as a manifesto for a separate methodology for qualitative research (see also Goertz 2017). By moving towards an independent methodology for qualitative research, the concept of causal mechanisms became central. This led researchers to search for an alternative understanding of causality and analyses that are not probabilistic. In addition, authors working more in a historical tradition used the term “process tracing” for political science and thus approached the concept of mechanisms from yet another angle (Beach and Pedersen 2019; Starke 2015; Trampusch and Palier 2016)—although the understandings of process tracing as a method are as various as the definitions of the mechanisms that it aims to trace (see the compilation of definitions in Trampusch and Palier 2016).

In their joint work, Goertz and Mahoney (2012) define causal mechanisms in purely methodological terms, without referring to the social theory literature in greater detail:

Instead, we can understand causal mechanisms to mean the intervening processes through which causes exert their effects. We propose that any relatively well-developed theory will provide a discussion of causal mechanisms. This is equally true for theories tested in the quantitative and qualitative research traditions: they propose ideas about the causal mechanisms that link independent variables to dependent variables. (Goertz and Mahoney 2012, 100)

Notably, the basic elements of research are still variables:

For the purposes of illustrating process tracing here, I use the term mechanism to refer to a factor that intervenes between a cause and outcome. I treat mechanisms in the same way as causes and outcomes; they are particular events or specific values on variables. Mechanisms are different from causes and outcomes because of their temporal position: they stand between a cause and outcome in time. Thus, in the expression XM⇢ Y, the letters refer to events or specific values on variables, with X being treated as the cause, M as the mechanism, and Y as the outcome. (Mahoney 2015, 206)

Mechanisms are thus variables that can be arranged chronologically between cause and outcome, that is a certain subgroup of intervening variables, whereby in the graph X ⇢ M ⇢ Y, the arrows are the interesting elements which have not yet been sufficiently clarified (Goertz 2017, 32). Summing up, in this understanding variables are explained in terms of their temporal position in a causal chain, which gives the term mechanism a purely analytical status.

2.2 Second Strand: Generative Mechanisms in Critical Realism

The lesser known yet oldest strand of a mechanism concept in the social sciences can be found in critical realism, which continues to exert a strong influence in Anglo-Saxon sociology. Roy Bhaskar is generally regarded as the founder of this approach. In his early work (1975), he used the term “generative mechanisms” to indicate what science should explore. Critical realism emerged as a theory of science that opposed the search for general laws that had previously been common in the social sciences (the so-called Hempel–Oppenheimer model). However, the roots of this approach go back to debates on the philosophy of science in the field of linguistics in the 1960s, which were then dominated by Noam Chomsky’s approach to a “transformational grammar” (Chomsky 1965). Mechanism is a term that Chomsky uses frequently in his work. One of Chomsky’s central considerations, however, was the critique of the behaviouralist scheme of stimulus and response. According to his argumentation, reaction mechanisms are not triggered by external stimuli. Rather, there are generative capacities in individuals which ensure that they can generate grammatically correct sentences. The corresponding mechanisms are therefore not reaction mechanisms to environmental stimuli, but generative mechanisms. The reflections from linguistics became a research paradigm in the social sciences.

I have argued that the causal structures and generative mechanisms of nature must exist and act independently of the conditions that allow men access to them, so that they must be assumed to be structured and intransitive, i.e. relatively independent of the patterns of events of men alike. (Bhashar 1975, 56)

For critical realism, causal structures or generative mechanisms are units of a not directly perceivable world that generates the events that happen and that people can experience. This is their causal power. Today, Margaret Archer (1995, 2015), in the explicit tradition of Roy Bhaskar, includes the concept of “generative mechanisms” and “causal power” as an integral part of her discussion of the sociological micro-macro or structure-agency problem and the justification of a “morphogenetic sociology”.

In these recent contributions, the structuralist roots of the concept of generative mechanisms become effective. It is not individuals who play the decisive role here, but structures and macro-phenomena that have their own causal power. To date, such discussions have not played a major role in the political science literature. Our conception of mechanism-based explanation picks up the notion that action is not triggered by external stimuli, but by individual capacities. However, we strictly reject the structuralist orientation of this strand of theory building.

2.3 Third Strand: Analytical Sociology

The third strand, dating back to the late 1970s, is the strand of “Analytical Marxism” and “Analytical Sociology”. Initially it also dealt with methodological questions but was diametrically opposed to the structuralist conceptions of critical realism. The initial search for clarity and precision—therefore the term “analytical” (Roemer 1986)—turned into a primarily methodological debate with game theory and the rational choice models as a more suitable basis for the social sciences, which also had the advantage of being able to provide a microfoundation (van Parijs 1993, 70–85) for Marxist economic theory. In this dispute between functionalism, structuralism, and rational choice explanation (a variable-based methodology did not play a major role at that time), the concept of causal mechanisms evolved (Elster 1983, 1986, 1989). In Nuts and Bolts for the Social Sciences (Elster 1989), the concept of mechanisms was a kind of compromise in view of the immanent difficulties of Elster’s attempt to think rational choice theory, its anomalies, and limits within a unified framework. Elster also transcended the mere methodological discussion by turning to single elements of rational choice theories and social psychological theories as well as classical sociological concepts such as social norms. This contributed to attempts to develop a toolbox of causal mechanisms (Elster 1989, 1999, 2015, 2017). Elster thus paved the way for a more social theoretical development of the concept of causal mechanisms, which eventually led to analytical sociology. In an early text, Elster describes the role of mechanisms in social science explanations:

To explain is to provide a mechanism, to open up the black box and show the nuts and bolts, the cogs and wheels of the internal machinery. […] The role of mechanisms is twofold. First, they enable us to go from the larger to the smaller: from molecules to atoms, from societies to individuals. Secondly, and more fundamentally, they reduce the time lag between explanans and explanandum. A mechanism provides a continuous and contiguous chain of causal or intentional links; a black box is a gap in the chain. (Elster 1983, 24)

Central to Elster are, first, the microfoundation, and second, closing the temporal gap between cause and effect. Later, Elster moved away from this definition of the role of causal mechanisms in favour of an interpretation that is more strongly based on regularities (Elster 1999, 2015). Our version of a mechanism-based approach presented here, however, more closely follows Elster’s early understanding of mechanisms.

Beginning with Hedström and Swedberg’s anthology (1998) and Hedström’s Dissecting the Social (2005), a broader stream of research developed that claimed the name “Analytical Sociology”. Apart from Jon Elster, it built on the work of Raymond Boudon, James Coleman and Thomas Schelling and their contributions to the micro-macro issue, while maintaining an action-centred theoretical foundation for social science (Hedström 2005, 6). Additionally, scholars within the analytical sociology tradition argued that referring to a rationalist theory of action provided a far more viable justification for middle-range theories as advocated by Merton: “Building upon the foundations laid by them, an analytical middle-range approach to sociological theory can be developed that avoids the somewhat empiricist and eclectic tendencies of Merton’s original middle-range approach” (Hedström 2005, 8–9).

2.4 Fourth Strand: Historical Sociology and Historical Institutionalism

James Mahoney is a leading researcher in the first strand of literature on the methodology of qualitative research. However, his methodological work is anchored in the tradition of historical institutionalism, which brings together influences from Marxism and historical sociology. In the historical institutionalist work by Theda Skocpol, Dietrich Rueschemeyer, John Stephens, Evelyne Huber, Wolfgang Streeck, Peter Hall, and Kathleen Thelen—to name prominent researchers within this research strand—the value of a mechanism-based explanation is tested on major macrosociological issues. Charles Tilly occupies a special position in this line of research, because his methodological considerations evolved during the reflection on his own wide-ranging historical research on revolutions, state development, and protest events. Here, research practice informs and shapes the methodology. This priority of empirical research also leads to lists of relevant causal mechanisms (such as environmental mechanisms, cognitive mechanisms, and relational mechanisms), especially in Tilly’s collaborative work with Doug McAdam and Sidney Tarrow in Dynamics of Contention (McAdam et al. 2001). This connection between political and historical sociology comes perhaps closest to a concept of mechanism that could tackle central questions of political science.

The work of Renate Mayntz and Fritz W. Scharpf on “Actor-centred Institutionalism” can also be considered to belong to this institutionalist strand, as their approach is also concerned with middle-range theories and explanations that consider collective actors. In his book Games Real Actors Play: Actor-centered Institutionalism in Policy Research (1997), Scharpf strongly relies on game theory, but he combines it with considerations from governance theory to look for meso-level mechanisms as elements for modular explanation. Thus, explanations are based on combining different causal mechanisms as modules. It follows from this idea of modularisation that basic mechanisms can be combined into more complex causal mechanisms (see below). For Renate Mayntz (Mayntz 2004, 2017, 2019, 2020), methodological questions play a greater role, as do the sociological debates on the macro-micro problem, whereby the meso-level of organisations and collective actors is of great importance to her. Mayntz raises doubts about the ability to solve all relevant questions based on methodological individualism, as well as scepticism towards system theoretical approaches (Mayntz 2017). She has long been interested in individual events, especially those that seem surprising when measured against everyday expectations. Importantly, Mayntz (2019) demonstrates that social science cannot be preoccupied with producing general theories. In fact, there are good reasons to study how certain events unfolded in a particular single case. Ideographic, descriptive investigations which go into detail are therefore legitimate, and studying a single case can make perfect sense. The path to modular explanations and acknowledging that also single political events, when remarkable, should be studied and explained in single cases studies, are methodological allusions that go beyond references to the rational choice paradigm.

3 A Modular and Actor-Centred Conception of Causal Mechanisms

The development of a process-oriented, actor-centred concept and a modular approach to causal mechanisms takes up ideas from three out of these four strands of research on causal mechanisms. Only the structuralist tradition of generative mechanisms is not compatible with our more actor-centred understanding of causal mechanism, as well as our process orientation.

First, the process orientation is elaborated in the methodological considerations in qualitative research on mechanisms and process tracing. However, with a focus on new testing procedures to check the validity of results, this strand risks turning into a discussion among methodology specialists with only limited effects on the discipline as a whole. Therefore, it is important that researchers develop a clear idea of which mechanisms could be identified, and what significance these mechanisms can have for research in their field. Second, the focus on actors and enriching the understanding of causal mechanisms with social theory follows analytical sociology and actor-centred institutionalism. And finally, the idea of modular explanation, which turns out to be extremely relevant for research practice, is based on Elster’s and Scharpf’s considerations. Closely related to this is the idea that social science research must also be able to explain individual cases without withdrawing from comparative research. Here, our approach follows the course of historical institutionalism and the methodological considerations by Mayntz. It is modularisation that makes the explanation of individual cases fruitful for comparative purposes, at the level of particular mechanisms.

Process Orientation: We follow the strand of qualitative research in political science with regard to its strong process orientation, as embodied in particular in the literature on process tracing. The basic elements of political processes are events. Events are spatiotemporally determined phenomena that can be distinguished from states as more permanent properties of entities. In contrast to objects and their states, events have something that is momentary: “Events prefer to pass away. On the other hand, every event brings about a total change in past, present, and future—simply because it gives up the quality of being present to the next event and becomes a past for it (i.e., for its future)” (Luhmann 1995, 287). The term causal chain refers to the chronological sequence of causally relevant events that occur between an initial state and the outcome, for example a political decision of a public authority. Acknowledging causal relevance is the first step in the analysis of a process. If researchers can demonstrate from the data that the individual and collective actors who were involved in a process perceive certain events as the reason for their own activities, they can assume causal relevance. By collecting additional data, researchers must show in the course of the further research process whether the presumed causal relevance can be confirmed or not. The starting point of an analysis, the initial state, is determined by the research interest, the result of previous scientific analyses, or particular expectations about causal relationships. The end point of a causal chain is the outcome. Like the initial state, it is determined by the researcher’s interest and previous knowledge about the respective case. A single element in the causal chain is called a causal link.

Focus on Actors: In mechanism-based research, a highly debated question is how mechanisms unfold and what their key “entities” and “activities” are (Machamer et al. 2000). Our approach to the entities of mechanisms is actor-centred. Rather than socio-economic conditions, institutionalised rules or other supposed determinants of policy developments, we consider policy actors, their actions, and interaction as the key drivers of social policy. Therefore, our conception of causal mechanisms focuses on the activities of actors and the causal relationships between these activities. Identifying events that represent activities is central to this theoretically guided analysis of mechanisms. An activity is an event that can be attributed to a specific individual or collective actor. In organisations and other forms of collective actors, actions are not those of an organisation as a collective entity or those of all its members. Rather, the interpretations, preferences, and goals of action stated by representatives or elites of these collective actors are recognised as those of their organisation. For a detailed understanding of inner-organisational developments, it is therefore necessary to refer to the individual level. For analysing political decision-making processes, it is of great importance whether actions are attributed to a collective actor or if they express the intention of an individual member of an organisation. Activities can be explained by elementary causal mechanisms (see below).

Of course, events that cannot be attributed to individual or collective actors may also be highly relevant. These events are usually the result of many people’s activities, but they cannot be attributed to specific collective or individual actors. Examples of such events include election results, price relations, and income distributions. They describe the results of the interplay of activities of an (un)determinably large number of people who do not represent a collective actor. To explain such a subsequent event, for example an election result, identical mechanisms must be assumed for a large number of persons, or different mechanisms for different groups of persons.

Another question is whether there might be mechanisms that are determined by social situations, such as a crisis, an economic downturn, a long unresolved and contentious issue, or the pressure of certain problems. The strongest argument here might be that situations themselves generate adjustments. They “force” actors to act in a certain way and impose a certain, sometimes very limited scope of action on them. However, an actor-centred understanding would focus on how actors interpret the situation, and not on the situation itself. Although the situation exists independent of the actors’ interpretation, it can only engender further activities via actors’ perceptions and understandings of the situation. Each reaction to a situation is based on the perception, interpretation, and action orientation of those actors who find themselves in the situation and have to deal with it.

For applying an actor-centred approach to causal mechanisms in comparative analyses, one way forward can be to distinguish types of actors, for example heads of government, trade unions, medical associations, or conservative parties. Building such a typology of actors can be challenging, especially when dealing with very different periods of time, regions, and policy fields, as in the present volume. Here, the research purpose and the scope of comparison determine the typology’s degree of abstraction. The clustering of causal mechanisms according to different types of actors is one aim of the concluding chapter (Chap. 14).

Modularisation: A modularised explanation focuses on the idea that a single causal link can be explained by elementary causal mechanisms, that sequences of such causal links can be explained by complex causal mechanisms, and that the entire causal chain between the initial state and the outcome can be explained by a combination of several complex causal mechanisms. Elementary causal mechanisms start at the level of individual and collective actors and comprise only one causal link, namely the production of activities (demands, programmes, protests, decisions, etc.). Complex causal mechanisms comprise several successive steps and are composed of a sequence of activities, which in turn can be explained by elementary causal mechanisms. In the following, we start by presenting modularisation at the level of elementary causal mechanisms.

Elementary Causal Mechanisms: An elementary causal mechanism designates a specific form of perception, interpretation, and action orientation. These are understood in a very broad sense. Perception and interpretation encompass all forms of attention that are directed towards something, and the processing of what is perceived. Here, cognitive aspects with descriptive, explanatory, and prognostic elements have to be considered, as well as evaluative and normative elements. Taken together, perception, interpretation, and action orientation can explain an activity undertaken by an individual or collective actor. The following typology of elementary causal mechanisms starts from the existing literature in sociological theory. Similarities to Max Weber’s four-fold division of action types are present (instrumentally rational, value-rational, affective, and traditional) as well as references to the models of Homo Oeconomicus, Homo Sociologicus, the emotional man, and the (Goffman’s) identity claimant (Schimank 2016; Little 2016). The following list of six elementary causal mechanisms gives a first impression of how a toolbox of elementary causal mechanisms could be developed. They are characterised by a coupling of (well-known) action models with a perception-interpretation component:

  • Calculatory orientation (also rational calculation) is a mechanism that combines instrumental rationality with a form of perception in which a cognitively (rather than emotionally) oriented situational analysis, with an emphasis on the causalities that will be effective in the future, is associated with an assessment employing categories of benefits and costs.

  • Norm orientation refers to a mechanism that corresponds to the traditional Homo Sociologicus, which is guided by compliance with valid social norms and is associated with a form of perception and interpretation that is determined by evaluative categories (values and norms).

  • Normatively embedded calculatory orientation is a mechanism that initially follows from norm orientation, but within the limitations provided by norms, actors apply rational calculation and a form of perception and interpretation that combines cognitive and evaluative moments.

  • Reflective orientation (or rational reflection) is a causal mechanism in which all components of one’s own perception and action orientation are examined to ensure that they can be argumentatively justified and are therefore open to further debate.

  • Emotional orientation is a causal mechanism that combines an emotional and evaluative form of perception and interpretation with the action orientation ensuing from the currently dominant feelings.

  • Finally, comparative orientation is a causal mechanism in which comparison with others determines one’s own preference formation. Different objectives are conceivable (not being worse off than the average, being the best within a group, being different), which can be combined with more cognitive or more emotional forms of perception and interpretation.

These six elementary causal mechanisms might serve as basic building blocks for explaining political processes on the micro-level.

Complex causal mechanisms: Elementary causal mechanisms explain causal links between two activities. However, a political decision-making process usually comprises a multitude of causal links. Reconstructing all causal links between an initial state and an outcome might, to a certain extent, be feasible in individual case studies. However, it is very likely that the number of causal links is so high that comparisons with other cases become difficult. Instead of comparing decision-making processes on an elementary level, research can therefore also focus on identifying sequences in a political process. Such sequences can be explained by complex causal mechanisms. Complex causal mechanisms comprise several successive steps and are composed of a sequence of activities, which in turn can be traced back to elementary causal mechanisms.

The outcompeting mechanism as a complex causal mechanism, which is further detailed and empirically traced in Chaps. 3 and 4, can serve here to illustrate our theoretical considerations. It explains how competition between political parties in a democratic setting leads to social policy expansion. Depending on the analytical level of the analysis, several elementary causal mechanisms can be identified within this complex causal mechanism, such as a calculatory orientation of party elites to push for social policy expansion in the first place, or a norm orientation of party elites who consider social policy expansion appropriate due to fairness considerations. Likewise, with regard to voters who vote for the party that is suggesting social policy expansion, elementary causal mechanisms might include a calculatory orientation as well, with voters thinking that they would benefit in material terms. Or they might—like the party elites—hold a norm orientation, a comparative orientation, assessing social policy proposals with regard to other groups or policies, or possibly even an emotional orientation, linking social policy expansion to feelings of pride or the feeling of being acknowledged in the policy process. The elementary causal mechanisms that are at play within a complex causal mechanism need to be identified in empirical research. This can often mean a time-consuming research process. Therefore, analyses at the level of complex mechanisms should be a priority for social policy research. We argue that the number of complex causal mechanisms can, in principle, be infinite. A first attempt at bringing together complex causal mechanisms relevant to social policy research will be made in Chap. 14.

Summing up, by distinguishing between elementary and complex causal mechanisms, policy processes can be disentangled into individual steps and sequences that lead to a certain effect, which can thus be analysed in more detail. While elementary causal mechanisms are mostly in the realm of social theory, complex causal mechanisms are highly relevant for social policy research, which is why they are also the primary focus of the case studies in this book. Understood as distinct process sequences, they are located at an abstraction level that can be identified in different countries, at different points in time, and in different policy areas, allowing for a modular explanation.

We illustrate our modularised, process-oriented, and actor-centred approach of mechanism-based explanation in Fig. 1.1.

Fig. 1.1
An illustration of a chain from the initial state follows activities and events with complex mechanisms at the center with thick arrows and small arrows between activities. It begins from t subscript 0 to t subscript d e c. The legend is given on the right.

Causal mechanisms and causal chain. (Source: Own presentation)

4 What Can We Expect from a Mechanism-Based Approach to Social Policy Research?

By combining a process orientation, a focus on actors, and modularisation, the concept of mechanisms offers new perspectives for the entire field of social policy research. Working with causal mechanisms can be an asset in macro-quantitative comparative social policy research as well as in case study-centred work on individual countries or social policy programmes. It can function as a complement, an expansion, it can add depth or even be a corrective to existing research approaches.

The results generated in quantitative studies by using pooled time series analysis, logistic regression or other methods are often not easy to interpret because there is a time gap between measuring the independent variables and the dependent ones (Schmitt 2019, 356). The significance of a causal relationship can only be assessed if several intermediate steps are assumed, for which quantitative studies, however, do not provide any evidence. Macro-quantitative social policy research can thus be complemented by mechanism-based studies examining these intermediate steps, which are assumed in the temporal gap between causes and effects but treated as a black box in quantitative research. Even if not all cases included in the quantitative studies will (or should!) be examined in this way, mechanism-based studies indicate which intermediate steps or which causal chains one should be able to expect between independent and dependent variables. It would thus be possible to improve the interpretation of causal effects through more precise knowledge of causal chains and mechanisms in individual cases.

Macro-quantitative studies rely on data on the relevant variables, which are often available only for a rather limited number of countries, and for a limited period of time. When aiming to analyse the historical dimension of social policy, or to include further countries of the Global South, this line of research often experiences limits of data availability. By adopting an approach that is more strongly oriented towards individual social policy programmes and decisions, however, it is possible to expand the scope of social policy research, as there might be historical accounts or reports on these decision-making processes and the actors that were involved, and in some cases also files. These non-quantified data can then be used. To ensure that this does not lead to a transition to merely descriptive-narrative research, a mechanism-based approach that ensures comparability between individual studies is central. In this way, it becomes possible to identify whether causal chains exist that can also be observed in countries or during periods for which sufficient quantified data are available.

The actor-centredness and modularisation of a mechanism-based approach also facilitates closer scrutiny of the micro-level of social policy processes and so can provide a microfoundation for previous theories. This enables researchers to add depth to the results of social policy research. In particular, this is important for qualitative social policy research, particularly for case study design. Case studies can directly focus on identifying causal mechanisms, which requires that process tracing is applied systematically. Together with a cross-case classification of the relevant actors in social policy, comparisons between case studies can be facilitated. If certain types of actors are referenced, and the same procedure for identifying causal mechanisms is applied across studies, comparisons across very different cases are possible. However, these comparisons are not about generalisation. The question is not about regularities between variables, but about specific sequences in causal chains that can be explained through the same complex causal mechanisms. It is not cases that are being compared, but (occurrences of) complex causal mechanisms. Thus, beyond a logic of regularities, references can be discovered between social policy developments in very different fields, countries, and historical periods.

A final benefit of a mechanism-based approach is to correct existing approaches and theories. The extant repertoire of theoretical approaches often leads to a tendency to use case studies to “confirm” or “reject” respective approaches. Scholars either select certain concepts from existing approaches because they consider them to have great explanatory power, or they even take the assumptions of certain approaches for granted prior to empirical investigation. The downside of this is that a comprehensive comparative perspective on the level of case studies is made rather difficult. To advance social policy research that is able to cover different regions, sectors, and time periods, causal mechanisms can also be used as a new explanatory approach that is able to overcome some of the weaknesses of case-centred research on social policy. Many of the existing approaches focus on specific actors and/or variables. Yet, empirical results from social policy research certainly show one thing: There is not one valid explanation for (almost) all social policy programmes and countries that is only based on the factors highlighted in a single approach. What is more, no approach is superior to others. Rather, what social policy research shows is that approaches are being confirmed in some cases, while being disconfirmed in others. Against this background, the mechanism-based approach and modularisation establish a kind of “meta-level” that allows the combination of existing approaches. Moreover, by introducing mechanisms that were not included or focused on in existing theoretical approaches, it also contributes to correcting some of their “blind spots”.

5 Overview on the Chapters of This Book

The individual chapters of this edited volume analyse social policies from very different countries around the globe in both single and comparative case studies. The country chapters are structured into four parts, dealing with social policies in Asian countries (Part II), African countries (Part III), European countries (Part IV), and Latin American countries (Part V) (following the distinction of geographic regions from the United Nations Statistics Division). What is more, the chapters not only cover different countries, but also different fields of social policy, such as old-age provision, health, unemployment, work injury, long-term care, and social assistance. Many chapters focus on the development of social insurance institutions, yet the volume also includes chapters on non-contributory forms of social policy, such as social assistance or public health. In sum, the chapters thus demonstrate the great explanatory power of the mechanism-based approach that we have outlined in the preceding sections for different fields and institutional arrangements within the realm of social policy. In most analyses, complex causal mechanisms are clearly in the foreground. Yet, the level of detail with which the mechanisms are spelled out depends, among others, on the time span covered by the empirical analysis, the number of countries, as well as the fields and programmes that were analysed.

This section gives a short introduction to the different chapters, especially with regard to the countries and social policies that they focus on. The mechanisms that the authors identify in the different chapters are presented and discussed in the concluding chapter of this volume (Chap. 14).

Following this introduction, the second part of the book deals with the development of social policies in Asian countries. Ten Brink, Müller, and Liu analyse the development of social protection schemes for urban workers in China, which were adopted as part of longer-term reform policies. More specifically, they trace the complex causal mechanisms that can explain the introduction of the Urban Employees’ Basic Pension Insurance in 1997, the Urban Employees’ Basic Medical Insurance in 1998, and the Work-Related Injury Insurance in 2004 (Chap. 2). Kuhlmann and Nullmeier analyse the development of two types of contribution-based pension systems, focusing on the cases of South Korea and Vietnam (who have a social insurance scheme) and Sri Lanka and Malaysia (who rely on national provident funds). Despite well-known problems in all four countries—most importantly with regard to effective coverage—the systems have been maintained or even expanded since their introduction, which the authors explain by several complex causal mechanisms (Chap. 3). Öktem zooms in on the development of the unemployment insurance programme in Turkey, which was established in 1999. He focuses on the complex causal mechanisms that have transformed the programme’s initial focus from unemployment protection to active labour market policies (Chap. 4). Finally, Heinrich, Isabekova, and Pleines apply the mechanism-based approach to countries in the post-Soviet region—thus covering both Asian and European countries—and trace the complex causal mechanisms that contributed to the introduction of mandatory health insurance in some countries within that region after the end of the Soviet Union (Chap. 5).

In the third part of the book, we turn our attention to social policies in African countries. Focusing on the highly fragmented and exclusionary social insurance schemes in Tunisia and Uganda, Thyen and Schlichte show that their roots can be traced back to decolonialisation, and analyse the complex causal mechanisms that were at play throughout this process (Chap. 6). In contrast, Devereux’s analysis does not focus on single African countries, but illuminates the role of international development agencies in promoting social protection policies, and the strategies these actors use to convince African policymakers of their ideas, focusing on one complex causal mechanism (Chap. 7).

The fourth part of the book focuses on social policies in European countries. Analysing the reform process in the healthcare system in Croatia in the early 1990s, Malinar identifies the complex causal mechanisms that contributed to the development of a hybrid healthcare system that can also be understood as a counter-reaction to the previous communist system. Notably, Malinar’s analysis also zooms in on some elementary causal mechanisms (Chap. 8). Like Malinar, Druga focuses on healthcare reform processes in the early 1990s, analysing the Albanian case. In her analysis, she focuses especially on one complex causal mechanism that characterises the interaction between Albanian policymakers and the World Bank as a transnational actor, while also zooming in on some elementary causal mechanisms (Chap. 9). Safuta, Noack, Gottschall, and Rothgang analyse the processes that followed the introduction of long-term care insurance in Germany in 1995–1996. Focusing on the crucial role that migrant workers play here, the authors trace the complex causal mechanisms that contributed to a specific form of migrantisation within the field (Chap. 10).

In the fifth part, social policies in Latin American countries are in the foreground. In their historical analysis, González de Reufels and Huhle show that the four Latin American Medical Congresses (1901–1909), as transnational events for the medical profession, played an important role for establishing public health policies in Chile and Uruguay, and highlight the complex causal mechanisms that were at play in this process (Chap. 11). Health is also the focus of Sirén’s analysis. His chapter traces the complex causal mechanisms within the political process on universal health coverage in Bolivia, which unfolded against the background of a highly fragmented and exclusionary healthcare system (Chap. 12). In the final chapter of this part, Barrientos adopts a mechanism-based approach to study whether conditional income transfers in Latin American countries lead to political responses by the recipients of these transfers and, subsequently, better political inclusion (Chap. 13).

In the concluding chapter (Chap. 14), Kuhlmann and Nullmeier draw together the mechanisms that were traced in the individual chapters, and present a structured compilation of the complex causal mechanisms that have been identified throughout the volume. Moreover, they outline future research avenues for mechanism-based approaches in social policy research.