Why preferences and institutions change: A systematic process analysis of credit rating in Germany



This article argues that a within-case analysis of the causes and patterns of the institutionalisation of rating in the German financial system offers fresh insights into change in the major socioeconomic institutions of advanced capitalism. Using the method of systematic process analysis, the article explores the expansion of credit rating in the German banking system from three perspectives: historical (power), sociological (diffusion) and behavioural institutionalism (prospect theory). It demonstrates that the proliferation of credit rating resulted from a change of preference on the part of large banks. With Germany as a least likely case for successfully implementing rating, the study's main lesson is that institutional analysis may benefit from incorporating behavioural institutionalism into the analysis of preference change because this cites economic motivations as causes of preference shifts and institutional changes.


The ‘near ubiquity’ (Carruthers & Kim 2011: 252) of credit rating is among the most interesting and challenging features that growing financial integration has brought with it.1 Rating can be performed as external rating, as a service provided by private, profit-seeking credit rating agencies. Banks, insurance companies, pension funds and investment funds can also use internal rating models. Yet external and internal ratings constitute significant changes to financial systems in coordinated market economies. In the case of banks' credit decision making, for example, rating replaces personal contacts with a ‘standardized algorithmic exercise’ (Carruthers & Kim 2011: 252) that is performed by the internal rating divisions of banks or by external rating agencies. As a result, instead of networks, firm-hierarchy and markets coordinate these financial activities.

The spread of rating is not a self-evident phenomenon of modern finance. To understand how rating transforms the governance of financial activities, we therefore need to understand why and how financial market actors have adopted it. This adoption is particularly puzzling in coordinated market economies where rating replaces non-market institutions for the purpose. This article studies the use of rating by large German private banks in the period between the 1970s and 2012. It argues that the way financial market actors frame the decision-making environments may trigger shifts in their preferences about the governance of their financial activities, and thus may induce institutional change. The study investigates the German rating trajectory from the perspective of three prominent approaches of institutional analysis: historical (power), sociological (diffusion) and behavioural (prospect theory).2 The article's most important insight is that behavioural institutionalism (prospect theory) is better in helping us to understand what motivates German banks' preference shifts.

Germany's banking system was selected because it fulfils the conditions of the least likely case study (Eckstein 1975; Levy 2002: 442). Until the 1980s, Germany was a prototypical case of a coordinated market economy with a financial market regime governed by non-market institutions (Deeg 2001; Lütz 2002). Thus, it is a hard case for successfully implementing rating practices because external rating (internal rating) means governing risk management and corporate financing through the market (firm-hierarchy) instead of through the banks' networks. At first sight, the proliferation of rating fundamentally contradicts the old German trajectory and we expect that German financial actors would be reluctant to introduce internal rating or to rely on the services of external rating agencies. However, the opposite occurred. At the beginning of the 1980s, German banks began to use external rating agencies and internal rating models to assess creditworthiness and, over time, rating became more and more institutionalised. According to Eckstein (1975), ‘least likely’ cases are particularly helpful in confirming theoretical expectations. Least likely cases allow ‘Sinatra inference’ (Levy 2002: 442), which, when applied to rating, suggests that if we can explain why rating was incorporated in Germany, it is more likely that our explanation will be corroborated for other countries with bank-based financial systems, such as Austria, Belgium, France, Italy, Japan, Portugal and Spain (Demirgüç-Kunt & Levine 1999: 22).3

In methodological terms, the study uses systematic process analysis (Hall 2008), which compares the historical record of the emergence and institutionalisation of rating in Germany with the expectations of rival analytical approaches. Systematic process analysis requires deep empirical foundations because rival empirical expectations are tested against each other by documenting conforming and non-conforming observations (Hall 2008: 314). For this reason, I have double-checked basic information and collected data from three different types of sources: academic and professional publications on rating; newspaper articles about German financial market deregulation; and interviews with experts who represent the largest German private banks and financial authorities.4 At the request of our interviewees, the matters discussed are used and quoted in the text anonymously.

The next section of this article explains how different institutional approaches conceptualise institutions and preference change and why my study investigates the puzzling German rating trajectory from the perspective of historical, sociological and behavioural institutionalism. Following the logic of systematic process analysis (theory formation, deriving predictions, making observations and drawing conclusions; Hall 2008), the three sections after that discuss the empirical expectations concerning the causes and patterns of rating in Germany derived from historical, sociological and behavioural institutionalism. These empirical expectations are compared with the historical record of rating in Germany. In the final section I discuss the implications of my findings.

Institutionalisms and preferences

This is not the first study of rating in political science (e.g., Sinclair 2005; King & Sinclair 2003; Abdelal 2007; Carruthers 2010; Pagliari 2012), but it is the first in-depth case study of a European national trajectory and the first analysis that investigates the proliferation of rating systematically from the perspective of institutional theory.

This article assumes that the clue for understanding institutional change is the exploration of preference change. The reason for this is that material interests change little and preferences translate material interests (such as the profit interest of banks) into action. Following Frieden (1999: 39), Vogel (1999: Footnote 1), Hall (2005: 130–131), Woll (2005: 8) and others, I claim that there is a difference between material interests and preferences and that it is useful to assume interests are fixed and to derive them from economic theory. Interests (which can be multiple) are the (mostly economic) values and benefits that actors pursue; their preferences define the way they order the possible outcomes of their behaviour. In this study, ‘interests’ are defined as banks' profit-seeking motives. Banks' preferences reflect their ranking of the governance mechanisms that they can use in the assessment of creditworthiness to maximise their profits – namely how they order the use of networks, firm-hierarchy and the market.

The focus on preferences means that, in contrast to most other institutional analyses, which, following Hall and Taylor (1996), use the trinity of rational choice, historical and sociological institutionalism, this article replaces the rational choice approach with ‘behavioural institutionalism’. The reason for this is that rational choice institutionalism has problems in analysing preference change and processes of preference formation. Historical, sociological and behavioural institutionalism treat preference change as an empirical question because they see the formation of preferences as endogenous. Rational choice institutionalism views preferences as exogenous and given, and it is only an exogenous shock (thus a heavy and rapid systemic transformation) that is expected to change them (e.g., Hall & Taylor 1996: 948–951; McDermott 2004: 292; Hall 2005: 130–132; Fioretos 2011b: Table 1; for an exception, see Greif & Latin 2004). In this approach, ‘institutional change happens only when ceteris is no longer paribus, that is, when shocks exogenous to the system alter the context’ (Hall 2010: 205, emphasis in original; similarly, see Mahoney & Thelen 2010: 6).5

Table 1. Differences between historical, sociological and behavioural institutionalism
 Historical institutionalismSociological institutionalismBehavioural institutionalism
(power)(diffusion)(prospect theory)
InstitutionsInstitutions reflect power relationships and are instruments of power redistributionInstitutions not only specify ‘what one should do’, but also ‘what one can imagine oneself doing’ (Hall & Taylor 1996: 948)Institutions are choice architectures
Endogenous preference changeConditioned by existing institutions, power struggles and strategic interactionConditioned by cognitive and normative dimensions of existing institutions and decision-making situationsConditioned by actors' framing of relative losses and gains

We have to note that the three strands of institutionalism – historical, sociological and behavioural – share two points. First, they contend that there is an interactive relationship between actors, institutions and the decision-making situation. And second, they analyse processes of preference formation in an inductive manner and allow for endogenous preference change as a driver of institutional change. However, despite these commonalities the three institutionalisms are nevertheless quite distinctive in their understanding of institutions and how endogenous preference change comes about (see also Table 1). While historical institutionalism has a political and historical understanding of actors' decision making and preferences and sociological institutionalism highlights normative and cognitive dimensions, behavioural institutionalism views institution-building and endogenous preference change from the perspective of economic decision making.

Historical institutionalism conceptualises institutions from a power-distributional perspective and views preferences as conditioned by existing institutions and power struggles as well as strategic interaction (Thelen & Steinmo 1992: 2, 9–10; Hall 2010: 217; Jackson 2010: 68).6 Sociological institutionalism follows a ‘cultural approach’ by stressing the cognitive dimension of institutions and preferences (Hall & Taylor: 1996: 948). Assuming the logic of appropriate behaviour (Hall & Taylor 1996: 947–949), institutions not only specify ‘what one should do’ but also ‘what one can imagine oneself doing’, with the result that cognitive and normative dimensions also affect actors' preferences (Hall & Taylor 1996: 948). Behavioural institutionalism loosens the ‘calculus’ approach of rational choice institutionalism, which, according to Hall and Taylor (1996: 939–940), assumes that actors maximise their material interests by a specific and fixed ‘preference function’ and defines institutions as coordination mechanisms sustaining particular equilibria (similarly, see Mahoney & Thelen 2010: 6). Instead of stringent calculus, behavioural economics assumes bounded rationality – thus, that rationalist calculations are limited by psychological effects such as ‘loss aversion’ and ‘framing effects’ or ‘hyperbolic discounting’ (Shepsle 2006: 333–334; similarly, see Posner 1998: 1533; Levi 2009: 131). As a result, in contrast to rational choice institutionalism, and in line with sociological and historical institutionalism, behavioural economics (prospect theory) acknowledges that endogenously generated preference shifts are possible (McDermott 2004: 292): Preferences and their formation are not deductively conceived but, as in historical and sociological institutionalism, they are treated as empirical questions. What distinguishes behavioural institutionalism from historical and sociological institutionalism is that this approach claims it is actors' framing of relative losses and gains which make their preferences change. Historical institutionalism views preference change as a function of existing institutions, strategic interaction or constellation of actors while sociological institutionalism points to the effect of appropriate behaviour. Behavioural institutionalism by contrast links endogenous preference change to how actors frame the relative losses and gains of their behaviour. Thus, behavioural economics allows us to investigate economic reasons for preference shifts.

In accordance with the logic of systematic process analysis, the next three sections compare the empirical expectations of the three institutionalisms with the historical record of the proliferation of rating among German banks.

Historical institutionalism

When historical institutionalism sees institutions from a power-distributional perspective, the idea that preferences are conditioned by the decision situation and existing institutions is emphasised (Thelen & Steinmo 1992: 2; Hall 2005: 150; Woll 2005: 5; Farrell & Newman 2010: 611, 615; Fioretos 2011a: 12). Historical institutionalism argues that ‘institutions [are] above all else distributional instruments laden with power implications’ and therefore ‘fraught with tensions’ (Mahoney & Thelen 2010: 10). Consequently, historical institutionalism expects processes of institutionalisation and change to be strongly affected by conflicts about resources (Mahoney & Thelen 2010: 7–9). Actors exhibit path-dependent behaviour and defend the power relationship which reproduces the institution. Finally, historical institutionalism also acknowledges that, in cases where we observe neither abrupt change nor path dependency, institutions may change incrementally. Such change results from actors' creative handling, because in practice institutional rules leave room for the interpretation and enforcement of rules (Streeck & Thelen 2005; Mahoney & Thelen 2010). To be sure, that does not mean that historical institutionalism always expects incremental change and excludes transformative change, but newer concepts (e.g., Streeck & Thelen 2005; Mahoney & Thelen, 2010) seek to reconcile incremental and transformative change by conceptualising the one as being an endogenous source of the other.7

For our analysis of the rise and expansion of rating, historical institutionalism implies the following two empirical expectations. First, we expect there is path dependency so that the policy positions and preferences of banks concerning rating will have been strongly affected by the special characteristics of the German financial system (as it was in the 1970s and 1980s). Against the background of Germany's old financial bank-oriented market regime (e.g., Lütz 2002), historical institutionalism expects German political and economic actors to be averse to the rise of market mechanisms because they are afraid of losing power. The second expectation is that the incorporation of rating into German financial market regulation may be a result of incremental institutional change, which can be explained by the characteristics of the old institutions. Based on Streeck and Thelen (2005) and Mahoney and Thelen (2010), we may argue that German financial market actors may have incrementally altered the financial market regime because they used the room for interpretation and enforcement that extant institutions provided.

Indeed, as predicted by historical institutionalism, in the very beginning German banks and industrial companies showed path dependent behaviour and were rather sceptical about external rating agencies and American financial market practices. This is still partially true (Engelen 2004). At first sight, the most significant evidence for this reluctant behaviour is delivered by the role Germany played during the Basel II8 consultation process. When in 1999 the first consultation paper (CPI) of Basel II recommended the use of external rating, the German government, banks and Mittelstand strongly opposed it and interpreted the initiative as an ‘attack’ on the German Hausbanken-System (Lütz 2002: 198). Consequently, the German government under Chancellor Gerhard Schröder vetoed Basel (see also Economist 2001; Woods 2005: 135; also confirmed by Interviewee 6). As a result of this opposition by the big German private banks and the German government (Lütz 2002: 199), as well as through the lobbying activities of the Institute of International Bankers (Carver 2000: 2; Lall 2012: 621–623), a ‘shift to an internal-based approach’ took place in June 2000 (Tarullo 2008: 104), which is codified in the second consultation paper (CPII).9 The acknowledgement of internal rating by CPII also led to the weakening of the opposition to Basel by the German Mittelstand.10 Lütz (2002: 199) even contends that the shift to internal rating was a ‘German negotiation success’. Similarly, in 1999, the Frankfurter Allgemeine Zeitung (FAZ) said of the Basel II consultation process: ‘The German banks do not want to hand over the field to the US rating agencies without a fight and are putting more weight on their own internal rating models.’11 Already in November 1999, the Bankenverband suggested to the Basel Committee a model of how Basel could use the internal rating of the banks as a form of supervision.12 The Basel episode indicates scepticism by large German banks which fits with the path dependency expectation of historical institutionalism. However, Basel also demonstrates that German banks defended their internal rating methods. Their aim was to avoid public regulation favouring external rather than internal rating. In this context it is also important to note that banks had already altered their risk management policies before Basel II had legitimated the use of internal rating models in banking supervision. Deutsche Bank was the first German bank (Interviews 1, 3), which developed statistical internal rating models through which it automatically checked clients' creditworthiness: A turning point for Deutsche Bank was about 1992 (Interview 1), when it began to feel like one of the most important international asset managers.13

The Basel II episode reveals that financial actors also quickly came to terms with rating, meaning that scepticism did not turn into obstructive behaviour but into the development of strategies which acknowledged the role of agencies and rating. Further evidence for this incremental adjustment is suggested not only by the fact that banks (and industrial companies) used external rating in order to enter the commercial paper market but also by the Deutsche Bank's initiative of founding a European external rating agency at the beginning of the 1990s (Engelen 2004). According to one of the experts interviewed, it was an industrial company (Siemens), which in order to undertake business activities in the United States, became the first rated German firm (Interview 2).14 In the 1980s, German banks followed suit and hired external rating agencies to confirm their credit worthiness and financial strength as they wanted to acquire fresh capital in the commercial paper market.15 In the early 1990s, the big German industrial companies further intensified their efforts using the commercial paper market as a source of external financing.16 Very quickly rating proliferated. While in 1985 three of the 13 German banks doing business in the United States had ‘Triple A’ ratings,17 by the mid-1990s of 122 German financial and non-financial firms rated by Moody's, 13 German banks enjoyed a ‘Triple A’ rating (Berblinger 1996: 29, 40). At the same time banks were intensifying their internal rating models. In 1991, for example, Dresdner Bank introduced an internal rating for securities (Wertpapierrating).18 Furthermore, banks then started developing internal rating models in order to automate the assessment of creditworthiness of borrowers in the lending business (Interviews 2, 6).

Historical institutionalism expects these practices as a result of incremental institutional change, endogenously prompted by the character of the old institutions. The old institutional rules leave room for interpretation and enforcement so that actors alter the existing banking and corporate financing regime by their creative handling of old institutions. Based on the works of Streeck and Thelen (2005) or Mahoney and Thelen (2010), the development of internal rating models by banks could be conceptualised as institutional layering, in view of the ‘introduction of new rules on the top of or alongside existing ones’ (Mahoney & Thelen 2010: 15). The problem, however, is that the old institutional rules were not characterised by a low level of discretion in interpretation and enforcement – which is, according to Mahoney and Thelen (2010: 19), a precondition for layering. Rather the reverse was the case. In the 1980s the German banking and financial market regime was characterised by self-regulation and networks (Deeg 1999; Lütz 2002), leading to a high level of discretion in the interpretation and enforcement of extant rules. Consequently, the historical institutionalist way of reconstructing the rise of rating as a result of change that is endogenous to old institutions is limited in its ability to specify the exact causes of banks' preference shifts.

To sum up: Historical institutionalism has limits in helping us to understand fully the rise of rating in Germany. As predicted, the process was contentious, but actors came to terms with rating and incorporated it into their financial market practices. However, this adaptation was not the result of actors' creative handling of extant institutions. Consequently, historical institutionalism has weaknesses in explaining the proliferation of rating because it overvalues the direct effect of old institutions and their characteristics on adaptation processes and preference change. The precise conditions under which banks have changed their preferences in assessing creditworthiness are not fully captured by this approach. Therefore, it is important to explore how the historical record fits with sociological and behavioural institutionalism.

Sociological institutionalism

A sociological constructivist approach replaces politics and the effects of extant institutions (in the form of path dependency or the level of discretion in the interpretation of rules) with the logic of appropriateness and cognitive dimensions. It highlights the importance of norms, ideas or knowledge, represented, for example, by the concept of ‘diffusion’, which is the ‘socially mediated spread of practice within a population’ (Strang & Meyer 1993: 487). In organisation theory – where sociological approaches of institutional change by diffusion are widely acknowledged and from which most diffusion approaches stem – diffusion is viewed as being the result of various mechanisms – namely coercive, mimetic or normative mechanisms (DiMaggio & Powell 1983; Campbell 2010: 96). Actors' behaviour is conceptualised through the lens of the logic of appropriateness (Campbell 2010: 95), which frames what one should do. Whereas coercive isomorphism occurs when social practices are imposed by ‘formal and informal pressures exerted on organizations by other organizations on which they are dependent and by cultural explanations in the society within which organizations function’ (DiMaggio & Powell 1983: 150), mimetic isomorphism results from ‘standard responses to uncertainty’, which means modelling the social practices of others (DiMaggio & Powell 1983: 150–151). Normative isomorphism occurs through professionalisation and is diffusion by the ‘collective struggle of members of an occupation to define the conditions and methods of their work’ (DiMaggio & Powell 1983: 152). Please note that according to Campbell (2010: 96) it is likely that combinations of these mechanisms are at work when institutional change by diffusion occurs.

For our analysis of the rise and expansion of rating, sociological institutionalism implies the following three empirical expectations. First, rating has become institutionalised by coercive measures: it was imposed by the Basel II agreement or the EU Directives which implemented Basel II in the EU (coercive mechanism). Second, rating became institutionalised because financial actors adapted to the American model of financial market regulation: German banks turned to rating practices, their modelling and exchange because they experienced the use of rating in the American market as a situation of uncertainty and reacted to this with the standard ‘liberal’ response of external rating (mimetic mechanism). As industrial companies began using rating, banks became even more uncertain. The third empirical expectation is that the diffusion of rating to Germany was driven by the professionalisation of the rating industry itself through setting standards and playing a self-regulatory role. Banks and firms implemented these standards in order to behave appropriately.

Are these empirical expectations confirmed by the historical record? The Basel II agreement and its enactment was certainly a kind of catalyst for the expansion of rating among banks (Interviews 2, 3) but – as shown above – banks used rating much earlier than Basel. In addition, according to various surveys, overviews and studies (e.g., Adams et al. 1999: Table A6.2; BIS 2000: 42, Table 1; Cinquegrana 2009: 5), there was also no national licensing of credit rating agencies in German financial market regulation before the Basel II agreement was implemented in the EU in 2006.

Not only did a top-down explanation fail, but we also find only limited evidence confirming the existence of mimetic mechanism, with banks adopting models and the social practices of rating agencies. This fact was explicitly confirmed by several interviewees. Banks regularly compare their internal rating classifications with those published by the external rating agencies (Interview 2), but in the case of credit business German banks do not rely on external rating agencies (Interview 1). They also do not adapt their internal procedures to those used by external rating agencies. There are ‘Chinese Walls’ between them and rating agencies (Interview 1). Banks also view their own internal rating models as more accurate than those of the agencies because they can rely on large historical datasets that agencies do not have (Interview 4). Internal and external ratings also use different models – point-in-time (internal) versus through-the-cycle (external) – so that adaptation also technically makes no sense (Interview 1). Through-the-cycle rating models, which external credit rating agencies use, and point-in-time rating, which banks apply, have very specific strengths and weaknesses in assigning creditworthiness and risks (Hamilton et al. 2011). Most importantly, they are two fundamentally different methodologies for calculating probabilities of default because they weigh short-term and long-term credit risks differently (Hamilton et al. 2011: 31).

Banks' use of internal rating models in the daily retail business should be distinguished from the use of internal rating in banking supervision which Basel II implemented. However, also with reference to Basel II, there is only limited evidence for mimetic or coercive adaptation. To comply with Basel II, banks are obliged to fulfill various very detailed requirements and the specific internal rating methods they use have to be approved by the BAFIN (2007). Yet despite this ‘coercive’ aspect, it was not these requirements that caused banks to introduce internal rating because – as discussed above – such practices developed long before Basel II. We also have to note that when banks lack historical default data of their own they may adopt the credit assessments of external rating agencies to cross-check their own validations (Deutsche Bundesbank 2003: 61; BCBS 2004: 461–463). Thus, in this very special case, banks calibrate their internal rating validations and external rating grades. When they map their internal rating scales on the scales used by external rating agencies banks have to document this procedure, which is also supervised by the BAFIN (Interviews 1, 3). At first sight this may be interpreted as mimetic behaviour reinforced by normative regulation. But again this adaptation did not induce banks' preference shifts because banks developed their point-in-time rating models long before internal rating models became customary in banking supervision.

How does the expectation of professionalisation and of appropriate behaviour by banks fare? There is indeed some professionalisation in the German rating industry. On the one hand, in the late 1990s several educational programmes started to train professional rating analysts and advisers (Sinclair 2005: 134).19 On the other hand, there is good evidence that private advisers strongly lobbied banks and firms to introduce ratings in order to establish a market for rating advice for their own benefit (Interview 2). However, Kerwer (2005: 472) argues that self-regulation is totally absent in the case of rating and that professionalisation of the rating industry is very limited compared with other domains such as auditing and accounting. Whereas auditors and accountants have developed independent standards by which public authorities can assess their work, there are none for rating. Without standards, banks need not behave appropriately.

To sum up: As with historical institutionalism, sociological institutionalism fails to deliver a fully convincing explanation for the spread of rating. There is limited evidence for the occurrence of coercive, mimetic or normative mechanisms, but our evidence suggests that these mechanisms were not the ultimate causes of why banks shifted to rating. Consequently, it seems plausible to argue that the weakness of sociological institutionalism in explaining banks' behaviour is that it overvalues the effect of actors' emulative behaviour on the proliferation of rating. Let us now turn to behavioural institutionalism.

Behavioural institutionalism

Our third option is behavioural institutionalism (prospect theory). Prospect theory stems from ideas in the literature on behavioural economics which loosen the calculus assumption of the rational choice approach and assume cognitive limitations on rational choices (Kahneman 2003). Although there is a variety of approaches in behavioural economics, it is reasonable to focus on prospect theory in this article for two reasons. The first – as already mentioned – is that studying preference change takes centre stage in this approach. The second is that by now this strand of behavioural economics has become widely accepted and applied in political science (e.g., Weyland 1996; Levy 1997; McDermott 2004), particularly in studies of foreign policy and international relations.

Developed by the economists Kahneman and Tversky (1979) through experimental research in economic decision making, prospect theory is a descriptive model of economic behaviour under risk. It argues that economic decision making is a choice between relative gains and relative losses, rather than between final sets of outcomes, and that relative losses count more than relative gains (Kahneman & Tversky 1979: 274). Losses and gains are defined relative to a reference point. The assumption of a reference point distinguishes prospect theory from other expected-utility approaches (Kahneman & Tversky 1979: 274; Levy 1997: 89). The assumption is also viewed as the central analytical weakness of prospect theory because the reference point is dependent on the subjective framing of the actors (Levy 1997: 92, 100; McDermott 2004: 309). Yet, in our context, and following McDermott's (2004: 309) advice, we may assume that actors (e.g., banks) define the old status quo – that is, the governing of financial activities by banking credits and networks – as their reference point (see also Weyland 1996: 189). Putting this into the words of Fioretos (2011a: 12): ‘[E]conomic actors respond differently to prospective changes [in their external environment] depending on whether they frame departures from the status quo … as entailing losses or gains.’20

For our study, behavioural institutionalism implies the following empirical expectation. The institutionalisation of rating in the German financial market results from a preference change by large banks because they framed changes in their environment as losses relative to the status quo (their reference point). Faced with these changes, large banks experienced their old business model (commercial banking and banking credits) as a source of great loss. Consequently, they transformed their preferences concerning corporate financing and the assessment of financial strength. They entered risky capital markets and began using the services of rating agencies and therewith deviating from their old business models and the trajectory of the financial market regime.

How does this theoretical expectation fit with the observations? What exactly pushed banks so that they made losses? Why did banks consider themselves likely to make losses if they retained their old business model? There were three changes in the environment which indicate that banks thought their old way of assessing creditworthiness was a liability and a cause of declining profits. The losses prompted banks to use ratings. The three situations were: the downturn in lending business; the entry of rating agencies into German and European capital markets; and the banks' economic problems plus the need to restructure their information technology systems. As predicted by prospect theory, the historical record shows that banks took these situations seriously and reacted to them by fundamentally changing their preferences.

In the 1970s, when large industrial companies began replacing bank credits with bonds, share issues and internal reserves, banks heeded the warning bell by shifting to investment banking, which forced them to be rated. Rising interest rates in the wake of the two oil crises made bank credit expensive and this caused industrial companies to search for alternative sources of finance. The successive decline of profits made banks believe that they would be worse off if they retained their old business model based on lending than if they changed to investment banking – which implied using ratings. According to Deeg (1999: 80–87), in the early 1970s large non-financial firms increasingly relied on retained profits and depreciation to finance investment. Deeg (1999, 2005) refers to the study by Welzk (1986: 67), who found that, with the two oil crises and the rise in interest rates, large companies in the chemicals, automotive and electrical engineering sectors increasingly used retained profits (which were tax-free and intended for later pension payments) as an internal financing source instead of external bank credits. For example, the share of bank credits in the total working capital of the AGs (Aktiengesellschaften) in the electrical engineering sector declined from 16.3 per cent in 1974 to 2.8 per cent in 1984, in vehicle construction from 13.9 to 4.0 per cent, in engineering from 18.9 to 9.6 per cent, and in chemicals from 18.4 to 4.0 per cent (Welzk 1986: 67).

The changed financial behaviour of industrial companies, as well as the fact that non-financial firms were increasingly busy in lending capital to other firms (Deeg 1999: 85–86), ‘undermined’ commercial lending by banks (Deeg 2001: 22). Raw numbers clearly indicated to the large private banks the source of their losses; for example, the market share of the big three in loans to manufacturing in 1982 was 18.2 per cent, one-third lower than the 28.4 per cent in 1974 (Deeg 2001: 21). In addition, as already mentioned, German industry discovered bonds and equity financing. Interviewees explicitly named the activities of Siemens in the 1970s in the United States and the change to investment banking as causes of the expansion of German rating practices (Interviews 2, 5). Later, in 1993, the first German industrial company (Daimler Benz AG) registered at the New York stock exchange and switched to the American GAAP (generally accepted accounting principles) standard. This shift in financing strategies from bank credits to internal reserves, bonds and share issues challenged the large commercial banks as they were pressed to alter their business model from the traditional Hausbank policy to more engagement in investment banking (Deeg 1999: 86–91).21

Consequently, large banks demanded a fundamental deregulation of the German financial system and were actively engaged in its further internationalisation (Deeg 1999: 86; 2001: 23) in order to open the door to complex financial instruments such as asset-backed securities (ABS) and other derivatives. Deeg (2001: 22) views the industrial companies' shift away from the commercial banks as an endogenous cause of change in the system. In response to this challenge, the large commercial banks altered their strategy and gravitated toward meeting the demands of foreign external investors. Deeg (2001: 24) contends that financial integration and liberalisation driven forward by the United States or the European Union have become important for Germany, even though banks and firms were already ‘moving in this direction’ and altering their preferences. Our interviewees confirmed that industry's changing strategies, as well as increased internationalisation, disintermediation and securitisation of financial services, prompted banks and industry to enter risky capital markets and turn to rating (Interviews 2, 5).22

Disintermediation and securitisation of financial markets drew external rating agencies into German and European financial markets and their activities caused banks to worry about loss because they were afraid of losing confidential information and data to external agencies (Interview 2). They did not want their companies assessed externally, for example by American agencies (Sinclair 2005: 133, Footnote 71). In 1988 a significant turning point in banks’ credit policies was the initiative of the Deutsche Bank and Rolf Breuer (chairman at that time) to draw together the most important German banks and industrial companies into the Working Group Rating (Arbeitskreis Rating) and later the Project Team (Projektgesellschaft) to develop alternatives to dependence on American rating agencies (Sinclair 2005: 133–134; confirmed by Interviewee 2).23

Breuer was convinced that the use of issuer rating is necessary for the success of the European capital market project and that the more German companies use the capital market for corporate financing, the more they are forced to rate their securities (bonds or commercial papers).24 According to one interviewee, the main motivation of the private banks which backed Breuer's rating initiative was that they were against any policy that ‘socializes [in the sense of reveals] any knowledge about creditworthiness’ (Interview 2) to external rating agencies. Banks feared losing their authority over assessing creditworthiness and viewed the expansion of Standard & Poor's and Moody's as a battle over knowledge and confidential data (Interview 2). As they realised that their project to establish an external rating agency had failed, they increased their encouragement for developing internal rating models (Interview 2). In this sense, we can argue that due to the differences in methods of measuring credit risks between Germany, the United States and the United Kingdom,25 German banks opted for firm hierarchy instead of the market mechanism and outsourcing the service to a European external agency. Oliver Everling, the rating projects manager, argued: ‘Those who supported our project to establish a European rating company saw Europe's total dependency on America's rating monopoly coming’ (quoted in Engelen 2004: 65). In line with Anglo-Saxon bank practices, Deutsche Bank also introduced compliance officers as a reaction to scandals, introduced a new salary system, and put ‘Chinese Walls’ between different divisions.26 The Commerzbank followed suit at the end of the 1990s (Interview 3). It was the 6. KWG-Novelle (KWG: Gesetz über das Kreditwesen, German Banking Act) that further catalysed internal rating expertise (Interviews 1, 3) as this reform allowed German banks to use internal and statistical models (developed originally by rocket scientists) to assess their equity capital.27 This evidence suggests that German banks changed their preferences as a reaction to the challenge of external rating agencies. They recognised that sticking to their old practices was a lost cause and decided to incorporate rating in their business model. The benefits to banks of intensifying internal rating were that they retained their valuable clients' historical data and kept them secret.

The third situation, which made banks frame the status quo as a liability and the change to rating as a benefit, was that they had been experiencing fundamental economic problems since the early 1990s and were feeling the need to restructure by improving their information technology systems and dismissing some of their employees. Banks' economic problems generated by the downturn of lending to industry pushed them to use credit rating in order to attract external investors (Standard & Poor's 2002) and to close branches and dismiss staff (Interviews 2, 3). Concerning this last aspect, it is important to note that with German unification the banks acquired former GDR banks and built up too many, too expensive branches in the former GDR. Already in 1992 this expansion put banks under immense pressure to bring in rationalisation measures.28 The relative loss from maintaining personal ties in assessing customers' creditworthiness was here compounded by escalating labour costs. It was even the case that rating advisers recommended that banks enforce internal rating practices as a means of dismissing employees (Interview 2). This was reinforced when massive structural problems reappeared in 2000 and banks were challenged not only by international competition, but also by low profits in their lending businesses (Standard & Poor's 2000, 2001, 2002).29 Low interest rates and the high costs of an oversized workforce exerted pressure on banks to improve their information technology systems by switching to internal rating models (Standard & Poor's 2000). Likewise, the retreat of banks from lending increased credit costs for industrial companies, prompting industry to ‘consider the need for a credit rating in order to gain access to alternative sources of funding in capital markets’, as, tellingly, Standard & Poor's (2002) writes.

As predicted by prospect theory and confirmed by our interviewees (Interviews 2, 3, 5), confronted with these losses banks and industry made risky choices. They entered capital markets and used rating and rating agencies to meet their financial needs as well as an alternative means of assessing creditworthiness and financial strength. The ‘maladaptation costs’30 were too high. The gains and benefits created by entering the capital market business and introducing internal as well as external rating were depicted as being much higher than the costs of focusing on credits, losing confidential clients' data to rating agencies and the need for restructuring due to fundamental innovations in telecommunications and data processing.


In Germany, the proliferation of rating in the banking system has been not so much politically driven, but came as the result of changed economic practices of companies and banks. Viewed from the point of view of behavioural institutionalism, the German banking system changed preferences about rating because of the fear of losses. Perceiving these losses to be caused by their old business strategies (commercial banking and banking credits), banks shifted to risky capital markets and, consequently, started using the services of rating agencies. Faced with losses on their old business model, banks also began using internal rating models.

What are the broader lessons of this article for theories of institutional change? First, by presenting conforming and non-conforming observations this study documents that historical institutionalism (power), sociological institutionalism (diffusion) and behavioural institutionalism (prospect theory) make different assumptions about the causes of banks' preference changes. Historical institutionalism provides a sound micro-foundation for the explanation of change by highlighting power and its distribution as a mechanism of path dependent behaviour or by pointing to actors' creative handling of institutions due to the available room to manoeuvre. However, in explaining the rise of rating in the German banking system, this approach overvalues the significance of the direct effect of the characteristics of old institutions on preference change. Furthermore, it seems reasonable to argue that the sociological concept of diffusion overvalues the effect of emulative behaviour. Finally, behavioural institutionalism (prospect theory), which views actors' framing of relative losses and gains as reasons for preference and institutional change, seems to account more effectively for the proliferation of rating in the German banking system.

Second, the article also shows that it makes sense to take conceptual and ontological differences between different strands of institutionalism seriously. In the institutional literature there is also a debate about commonalities and overlaps between the various institutionalist approaches (e.g., Mahoney & Thelen 2010: 32; Hall 2010; Jackson 2010). For example, Jackson (2010: 63, 70) argues that rational choice, sociological and historical institutionalism (as well actor-centred institutionalism) share a conception of actors and institutions as ‘being mutually constitutive’ and ‘co-generative’. Although my analysis confirms these overlaps, it also highlights the value of acknowledging the conceptual differences. On the one hand, recognising differences is obviously a first step in finding ‘points of tangency’ (Hall 2010: 220) between different institutionalist strands. On the other hand, this study documents that one clear ontological difference between historical, sociological and behavioural institutionalisms is how they conceptualise the effect of political, sociological and economic motivations on preference changes.

Third, and consequently, this article has also shown that behavioural institutionalism is not superior to the other two institutionalisms in the analysis of preference and institutional change. All three approaches are actor-centred and consider effects of the decision situations to allow for endogenous preference change as a driver of institutional change. But confronting the three institutionalisms helps us to see the benefits of synthesising the different strands of institutionalism. One of these productive intersections may be the question of whether and how the framing of relative losses and gains impacts on the effect of power or emulative behaviour on preference and institutional change and vice versa (similarly, see Levy 2003: 226). If institutionalist analysis solves this problem, an integrative and synthesised political-economic institutionalist approach which incorporates historical, sociological and economic reasoning in the analysis of endogenous preference may be developed. In the study of political decisions about financial market integration, which are obviously linked to economic effects, such an integrative perspective is particularly significant.


I thank the four autonomous reviewers and the editors of this journal for helpful comments.


  1. 1

    Please note that publications by official organisations and rating agencies to which this article refers are collected in the online appendix. The appendix is available at the EJPR webpage.

  2. 2

    With reference to sociological institutionalism, my analysis focuses on the diffusion approach as this framework explicitly refers to change in institutional practices (Campbell 2010: 95). It is the study of Orfeo Fioretos (2011a) which inspired me to think about the use of prospect theory. Please note that Fioretos (2011a: 13) views ‘behavioural institutionalism’ as a variety of historical institutionalism, which incorporates insights from behavioural economics (thus prospect theory). However, the present study defines ‘behavioural institutionalism’ as a distinctive institutional approach, which is based on behavioural economics (on this concept, see also Ball 1998: 1509). My focus on the power explanation of historical institutionalism is explained in the online appendix.

  3. 3

    Least likely case studies select hard cases where it is unlikely that the theory is confirmed. Levy (2002: 442) calls this ‘ “Sinatra inference” – if I can make it there I can make it anywhere’. In our qualitative case study this ‘anywhere’ is to be located in similar systems – thus other countries with bank-based financial systems.

  4. 4

    I systematically searched the most important German business newspaper Frankfurter Allgemeine Zeitung (FAZ) for articles about rating and credit rating agencies from 1949 until June 2012. Furthermore, I conducted seven interviews with leading risk research experts at the headquarters of major German banks and insurances, and with rating experts and chief executives of the Federal Financial Supervisory Authority (BAFIN), as well as with private rating advisers (May and August 2012). In all, 12 people were interviewed.

  5. 5

    In addition, the results of Deeg (1999, 2001: 11, 15) suggest it is very reasonable to assume that the ‘initial motor’ for the liberalisation of the German financial system stemmed from the endogenous preference change exhibited by large banks in response to the changed financial behaviour of German industrial companies.

  6. 6

    Please note that sometimes historical institutionalism also incorporates rationalist and sociological arguments. I explain in the online appendix why I exclude these strands.

  7. 7

    The dependent variable of this article is not the degree and pattern of change (incremental versus transformative) but institutional and preference change, in order to understand why and how financial market actors have adopted rating as a governance mechanism. Therefore this discussion is not explored further.

  8. 8

    Basel II introduced two alternative, new regulatory requirements for credit risks of banks. The standardised approach bases assessments on external rating agencies and the internal-rating-based (IBR) approach acknowledges internal rating models (Lall 2012).

  9. 9

    On the influence of the big American banks and their lobbying for external rating, see also Woods (2005: 157) and Lall (2012).

  10. 10

    Kreditwesen, 5 February 2001, p. 0174.

  11. 11

    FAZ, 15 October 1999, p. 13.

  12. 12

    FAZ, 17 November 1999, p. 29.

  13. 13

    FAZ, 25 February 1992, p. 24.

  14. 14

    In 1975 the first comprehensive article on rating was published in the FAZ (Jürgensen 1975).

  15. 15

    FAZ, 10 March 1980, p. 14; FAZ, 6 September 1986, p. 16; FAZ, 12 May 1987, p. B26.

  16. 16

    FAZ, 28 April 1992, p. B9.

  17. 17

    FAZ, 10 September 1985, p. 16.

  18. 18

    FAZ, 16 March 1991, p. 18.

  19. 19

    FAZ, 14 September 1999, p. 62.

  20. 20

    For a further explanation of why behavioural institutionalism allows for preference change without exogenous shocks, please see the online appendix.

  21. 21

    FAZ, 1 October 2001, p. 31.

  22. 22

    The gains and benefits (in the form of profits) that large German banks achieved by entering capital markets are widely documented in the studies by Deeg (1999, 2001, 2005) and Lütz (2002).

  23. 23

    FAZ, 11 June 1990, p. 17; FAZ, 18 July 1991, p. 18.

  24. 24

    FAZ, 18 July 1991, p. 18.

  25. 25

    See also FAZ, 15 October 1999 on p. 13, with reference to Basel II.

  26. 26

    FAZ, 28 January 1992, p. 14; Wirtschaftswoche, 29 May 1992, p. 121.

  27. 27

    FAZ, 27 June 1996, p. 14.

  28. 28

    FAZ, 12 January 1992, p. 6; Horizont, 7 February 1992, p. 15; FAZ, 3 September 1992, p. 17; FAZ, 26 November 1992, p. 19.

  29. 29

    Handelsblatt, 11 December 2003.

  30. 30

    I borrow this term from Fioretos (2011a: 29).