Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/ou_press/understanding-the-shift-from-micro-to-macro-prudential-thinking-a-GZutgWr2Bf
- Publisher
- Oxford University Press
- Copyright
- © The Author(s) 2017. Published by Oxford University Press on behalf of the Cambridge Political Economy Society. All rights reserved.
- ISSN
- 0309-166X
- eISSN
- 1464-3545
- DOI
- 10.1093/cje/bex056
- Publisher site
- See Article on Publisher Site
Abstract
Abstract Macro-prudential thinking and its emphasis on endogenously created systemic risks, marginal in the banking regulation discourse until the mid-2000s, has become central post-crisis. This paper analyzes this intellectual shift by using discourse and citation-network analysis of the most-cited scholarly works on banking regulation and systemic risk from 1985 to 2014 and six in-depth interviews with central scholars. It demonstrates that the predominance of formalism, particularly partial equilibrium analysis, in studies on banking regulation pre-crisis impeded economists’ engagement with the endogenous sources of systemic risks discussed in the systemic risk sample. These studies excluded observed phenomena that could not be accommodated in mathematical models, largely ignoring contributions based on historical and practitioners’ styles of reasoning. Post-crisis, while informal analyses gain prominence in studies on banking regulation, attempts to conceptualize systemic risk as a negative externality indicate the persistence of formalism and equilibrium thinking and its concomitant epistemological limitations that stymie theoretical progress. 1. Introduction Since the outbreak of the financial crisis in 2007, the macro-prudential policy paradigm has gained increasing prominence (BoE, 2009; Bernanke, 2011; Bisias et al., 2012; Baker, 2013, 2015). The pre-crisis consensus of focusing solely on the risk management of individual banks, based on the idea that ‘for the financial system to be sound it is necessary and sufficient that each individual institution is sound’ (Borio, 2009, p. 33; Crockett, 2000), has been discredited. Instead, driven by the ruptures of the global financial crisis, policymakers have increasingly adopted the macro-prudential approach which focuses on the systemic risks generated endogenously within the financial system by the collective behavior of financial institutions (Crockett, 2000, p. 3) and attempts to maintain the stability of the financial system as a whole through controlling these risks (Hanson et al., 2011; Bisias et al., 2012). While some literature has acknowledged the difficulties of measuring systemic (tail) risks, which are often hidden from sight (Danielsson et al., 2012; SRC, 2015, p. 21), many new measures and models are being currently developed to deal with the risks emanating from contagion and the financial cycle (Smaga, 2014; Bisias et al., 2012). The trend towards macro-prudential policy interventions seems irreversible (Baker, 2015); however, it is not definite that a fully formed, coherent and effective variety of macro-prudential policy will become a common basis for the activity of central banks and regulators. In particular, this shift in the outlook on financial regulation has not been accompanied yet by a consensus on the appropriate measures and policy instruments (Baker, 2013; Claessens and Kodres, 2014). Despite the importance of this micro- to macro-prudential regulatory shift, the dynamics of the intellectual shift that underlies it and the reasons that prevented this shift to take place prior to the crisis in the academic economic discourse have not been addressed systematically. Prior work has shown how the shift in economic theory from public to private interest regulation since the late 1960s has undergirded the decline of macro- and the growth of micro-prudential regulation in the following decades (Harnay and Scialom, 2015). Our work carries this research forward by focusing on the evolution of the economic discourse on banking regulation since the late 1980s, contrasting it with the literature on systemic risk as the concept that undergirds macro-prudential regulation (Bisias et al., 2012). Investigating the intellectual reasons that impeded the shift from micro to macro-prudential regulation, we identify the epistemological barriers that prevented the economic discourse on banking regulation to adopt a macro-prudential regulatory perspective pre-crisis. Investigating the micro-macro shift also allows us to point to the persistence of these reasons in the current conjuncture post-crisis. The production of economic knowledge in academia influences policymaking in financial regulation, particularly, due to the ongoing technocratization of the latter (Goodhart, 2011; Marcussen, 2013). By focusing on the failures of the economic discourse that has fed into the policy sphere, our study complements prior studies that focused on the policy sphere alone (Baker, 2013, 2015; Seabrooke and Tsingou, 2009, 2014). While these studies point to the importance of groups that act as carriers of ideas, their institutional standing as well as the alliances they enter into, they rather ignore the dynamics in the sphere of knowledge production itself. In this paper, to give these studies a better grounding in their discursive context, we investigate the sphere of economic discourse, i.e. the place where the cognitive models with which regulators were seeking to optimize regulation originated (Black, 2013; Gigliobianci and Giordano, 2012).We therefore analyze the most prominent writings of economists before the crisis on banking regulation, as they provided the intellectual input for the interpretative framework, the ‘policy paradigm’ (Hall, 1993, p. 279) in which regulatory policy was to be enacted. Our analysis thus seeks to contribute to a better understanding of the role of economics in pre-crisis regulatory failures. The structure of this paper is as follows: Section 2 describes our method and samples. In Section 3, we conduct the discourse and network analysis of our samples and complement it with findings from six expert interviews conducted with scholars central for the evolution of the systemic risk discourse pre- and post-crisis. Section 4 reflects on our findings and concludes. 2. Method and data description We analyze the evolution of economic thinking about banking regulation and systemic risk in a longitudinal perspective through content and citation network analysis for the period from 1985 to 2014. We collect the 10 top-cited scholarly works per five-year period on banking regulation. To understand whether and if so why this discourse has failed to take into account systemic risk prior to the crisis, we also trace the economic discourse on systemic risk for the same period, using a similar procedure. We consider the highest-cited papers to be a good representative of the most influential and well-established ideas in the economic discourse on banking regulation in the years following their publication, during which these top-cited articles have attracted most of their citations. Using the date of publications of these scholarly works, we divided the sample into six periods (1985–1989, 1990–1994, 1995–1999, 2000–2004, 2005–2009 and 2010–2014). To collect the top-cited works in each period, we used Google Scholar1 and searched for the following search terms in the ‘title’ search: Banking Regulation, Bank Regulation, Financial Regulation, Microprudential Regulation, Micro-Prudential Regulation, Microprudential, Macroprudential Regulation, Macro-Prudential Regulation, Macroprudential and Banking Law. We have also used the following search terms in the whole article search in Google Scholar: Banking Regulation and Bank Regulation. For the systemic risk sample, we searched for titles in Google Scholar that include any of these terms: Systemic Risk, Financial Contagion, Bank Contagion, Banking Contagion, Banking Crisis, Financial Crisis, Bank Crisis, Financial Stability and Financial Fragility. In the whole article search function, we searched for Systemic Risk, Financial Contagion, Financial Crisis and Banking Crisis. In order to make sure that we did not miss any of the top-cited scholarly works, we replicated the search above using the Web of Science database. The above search resulted in a generic sample of the top-cited scholarly works that touch upon banking regulation and/or systemic risk, which we then refined by excluding the works that do not address banking regulation or systemic risk directly as being their key theme. However, we have not excluded the review articles that reached the top-cited scholarly works in our banking regulation sample because they function as excellent proxies for the contemporary economic discourse on banking regulation; they reflect the economic discourse at the time of their publication and have influenced future research on banking regulation in the period following their publications. The final aggregate sample on systemic risk and banking regulation has 114 resources because six resources appear in both samples, as these resources address directly both banking regulation and systemic risk. The number of citations functions as a good proxy for the established ideas in the economic discourse unless some of the scholarly works are highly cited because their advocated ideas are rejected. Investigation of the banking regulation sample confirmed that the top-cited works in each period share the same ideas. This qualitative finding is further confirmed by the hubs of the banking regulation network we identified using citation network analysis (see fig. 1 below). The only outliers prior to the crisis were the articles of Borio (2003) and Jimenez and Saurina (2006), and these articles received most of their citations post-crisis (see below). In the systemic risk sample, ideas are more dispersed; however, contestation only plays a limited role, as there is only very limited cross-citation in the systemic risk sample (see below). Fig. 1. View largeDownload slide Banking Regulation Sample Network. Colours correspond to the colours in the Tables 1 and 2. Banking regulation sources are depicted as circles. Sources common to both samples are depicted as triangles. In this figure and the following, academic papers cited within the discourse analysis are depicted by author name; the others are depicted by their place in the table. Fig. 1. View largeDownload slide Banking Regulation Sample Network. Colours correspond to the colours in the Tables 1 and 2. Banking regulation sources are depicted as circles. Sources common to both samples are depicted as triangles. In this figure and the following, academic papers cited within the discourse analysis are depicted by author name; the others are depicted by their place in the table. Table 1. Banking regulation sample Both samples were collected in May 2014. In both samples, we refer to each resource by using the authors’ family name, date of publication and the first few words from the title. The number in brackets beside each resource refers to the citations received by the relevant resource as collected in May 2014. Scholarly works using the historical approach are highlighted in green, those using practitioners’ discourse are highlighted in red and those using informal theoretical analysis are highlighted in black. Finally, scholarly works using quantitative/formal methods, whether theoretical or empirical, are highlighted in blue. View Large Table 1. Banking regulation sample Both samples were collected in May 2014. In both samples, we refer to each resource by using the authors’ family name, date of publication and the first few words from the title. The number in brackets beside each resource refers to the citations received by the relevant resource as collected in May 2014. Scholarly works using the historical approach are highlighted in green, those using practitioners’ discourse are highlighted in red and those using informal theoretical analysis are highlighted in black. Finally, scholarly works using quantitative/formal methods, whether theoretical or empirical, are highlighted in blue. View Large Table 2. Systemic risk sample View Large Table 2. Systemic risk sample View Large Once we compiled the data, we used qualitative discourse analysis (Mayring, 2010) of the top five cited papers per period in each sample (30 resources per sample) to analyze the styles of reasoning, and the treatment of systemic risk in both samples. We focused on the origins of disturbances and chains of contagion in our analysis, the sources and propagation mechanisms of systemic risk. For the purpose of our discourse analysis, we distinguished between three sources of systemic risk that we distilled from the literature (Smaga, 2014): bank runs, contagion and financial cycle. Individual bank runs caused by exogenous shocks are seen as the predominant source of systemic risk in the traditional neoclassical literature starting from Diamond and Dybvig (1983). Contagion/propagation risk, while exogenously generated, already has an endogenous element as contagion channels endogenously amplify the shock across the financial system (Smaga, 2014). Finally, the financial/credit cycle is endogenously generated through the process of risk accumulation over time. The type of systemic risk underlying banking regulation is the crucial difference between the micro- and macro-prudential approach (cf. Crockett, 2000), and it thus allows us to pinpoint which kind of understanding of systemic risk dominated each period of our samples. Furthermore, in our discourse analysis we focused on the particular styles of economic reasoning/methods underlying each scholarly work in our samples (for an indicative list of these styles, see Crombie [1994]). Based on the in-depth discourse analysis, we also coded the remaining 54 sources in terms of their discursive style and then applied citation network analysis to the aggregate sample of 114 sources. Network analysis is used to corroborate the findings of the discourse analysis, analyze the interrelations among both samples and identify the central and authoritative papers in both samples. Here we draw on the distinction between hub measures and authority measures in citation network analysis.2 Considering that citation network analysis as well as discourse analysis of scientific writings cannot reveal the way economists interact with the rules of discourse that structure the observed patterns nor reflect the institutional context within which they occur, we have also interviewed six leading economists, four of which are authors of some of the most-cited papers in our samples,3 to corroborate and explain further the findings of discourse and network analysis. In the following, we present our findings: first those common to both samples, then findings relevant to the banking regulation sample, findings relevant to the systemic risk sample, and finally, findings relevant to the relation between both samples. 3. Discourse and citation network analysis of our two samples: findings 3.1 Findings common to both samples Our discourse analysis shows that some styles of reasoning in social sciences such as mixed methods and computational agent-based modelling are not adopted in both samples. The styles we observe in our samples can be categorized under two broad styles of reasoning: informal and formal analysis. We distinguish informal analysis further into three types: historically, theoretically and practically inspired discourses. With the exception of practitioners’ style of reasoning, the styles of reasoning we observe in our samples (reported in Table 3 below) are well established and sufficiently distinct styles of reasoning in social sciences (Morgan, 2012). Practitioners’ style of reasoning, however, does not easily fit into this well-established list. Rather, it seems akin to an engineering style of reasoning; it formulates its research question as a policy problem, and it uses a combination of empirics, informal and formal models, insights from historical cases, intuitions and judgments pragmatically for addressing this policy problem.4 Table 3. Different styles of reasoning/discourses observed in our samples Style of Reasoning Research Question Method Informal Analysis: Historical To test theories with the help of history and to develop theories from historical observations -Descriptive statistics and informal modeling -inductive Informal Analysis: Practitioners/technocrats Find a solution to a policy concern Descriptive statistics, informal theoretical analysis (eclectic) Informal Analysis: Theoreticians To explain and predict system behavior Informal theoretical analysis: develop and engage critically with economic concepts and theories without models, apply to regulation Formal Analysis: Quantitative Approach To explain and predict system behavior Mathematical modeling and econometrics Style of Reasoning Research Question Method Informal Analysis: Historical To test theories with the help of history and to develop theories from historical observations -Descriptive statistics and informal modeling -inductive Informal Analysis: Practitioners/technocrats Find a solution to a policy concern Descriptive statistics, informal theoretical analysis (eclectic) Informal Analysis: Theoreticians To explain and predict system behavior Informal theoretical analysis: develop and engage critically with economic concepts and theories without models, apply to regulation Formal Analysis: Quantitative Approach To explain and predict system behavior Mathematical modeling and econometrics View Large Table 3. Different styles of reasoning/discourses observed in our samples Style of Reasoning Research Question Method Informal Analysis: Historical To test theories with the help of history and to develop theories from historical observations -Descriptive statistics and informal modeling -inductive Informal Analysis: Practitioners/technocrats Find a solution to a policy concern Descriptive statistics, informal theoretical analysis (eclectic) Informal Analysis: Theoreticians To explain and predict system behavior Informal theoretical analysis: develop and engage critically with economic concepts and theories without models, apply to regulation Formal Analysis: Quantitative Approach To explain and predict system behavior Mathematical modeling and econometrics Style of Reasoning Research Question Method Informal Analysis: Historical To test theories with the help of history and to develop theories from historical observations -Descriptive statistics and informal modeling -inductive Informal Analysis: Practitioners/technocrats Find a solution to a policy concern Descriptive statistics, informal theoretical analysis (eclectic) Informal Analysis: Theoreticians To explain and predict system behavior Informal theoretical analysis: develop and engage critically with economic concepts and theories without models, apply to regulation Formal Analysis: Quantitative Approach To explain and predict system behavior Mathematical modeling and econometrics View Large These different discourses display a particular relationship to the conception of systemic risk and its analysis, as we will document below. 3.1.1 Practitioners. For practitioners, there is a non-problematic relationship to systemic risk. It is their major concern; concepts such as contagion represent an empirical reality they have to deal with, even allowing them to overstep the legal boundaries of their mandate (Brimmer, 1989). Practitioners’ ease with the concepts of systemic risk and contagion also was an important source of legitimacy in the literature on systemic risk. Especially in the early literature (e.g. in the second period), one finds references by academics to practitioners to justify their theoretical work (e.g. on contagion, see Kaufman [1994]; also Bernanke and Gertler [1990]). The style of practitioners, as found for example in the work of Kaminsky and Reinhart (1999, 2000), is to attempt to observe patterns in the data and to develop better forecasting of future events with its help. The practitioners’ approach has been well summarized by John B. Taylor (2009): ‘Following an approach to policy advocated by David Dodge [the former Governor of the Bank of Canada] throughout his distinguished career in public service, I try to use empirical evidence to the maximum extent possible and explain the analysis in the simplest possible terms, including by using a series of illustrative graphs’ (p. 3, emphasis ours). 3.1.2 Historical discourse. Scholars using historical reasoning/discourse operate inductively based on the patterns they find in the historical data. Kindleberger (1988), for example, seeks to develop models that fit these patterns. Other sources, such as Calomiris and Gorton (1991), focus on the historical origins of banking panics in the USA (that is, multiple bank runs and contagion effects) and discriminate between different theoretical models using this data. They thereby represent an old style of economics which was eclipsed by the rise of formal economic model building (cf. Kindleberger, 1988). Remarkable in these sources is the use of simple flow charts, comparing countries over long periods of time as points of departure for theoretical reasoning. 3.1.3 Informal analysts. Informal analysts such as Borio (2003) and his group at the Bank of International Settlement (BIS), Brunnermeier (e.g. 2008) or Minsky5 (1992) seek to informally develop disequilibrium and endogenous risk models without being constrained by mathematical models. While they subject themselves to the rigor of mathematics where possible, they can work out the implications of financial cycle, and propagation risk without being constrained by formal models. It allows them to deal with more complex theoretical assumptions and thus to develop a broader picture of financial market developments. Informal analysts (such as Minsky) and historians (such as Kindleberger) relate to longer-term empirical facts, which allow them to acknowledge the existence of repeating cycles. 3.1.4 Formal analysis. In contrast, formal analysts rely primarily on mathematical models. That means that concepts only exist if they can be modeled. In the first four periods of both samples, mathematical modeling uses comparative statics based on partial equilibrium. Formal analysts seek to explain financial fragility itself, but not how it can work over a cycle, basing themselves on exogenous shocks rather than endogenous build-up of risk (e.g. Bernanke and Gertler, 1990). Models are made simple in order to keep them mathematically tractable. We observe a shift in methods and modeling techniques over the six periods covered, from partial equilibrium analysis to general equilibrium analysis and network analysis, both of which are better suited for thinking about systemic risk.6 This shift occurs around 2000 in the systemic risk sample, exemplified by the central work of Allen and Gale (2000). There is also a shift from comparative static analysis to dynamic analysis, as the time dimension of risk is included when analyzing systemic risk (Borio, 2003). These shifts in the modeling techniques towards network analysis and dynamic analysis are the preconditions for the fundamental post-crisis shift. It allows for a broadening of real-world phenomena that can be studied, such that mathematical modellers can more directly engage with systemic risk-related concepts, such as contagion. 3.1.5 The distribution of the different styles in both samples. When looking at the distribution of formal/informal discourses in the two samples, distinct differences emerge. Formal analysis dominates the banking regulation sample with a share of 61.64%. Conversely, the informal discourses dominate the systemic risk sample with a share of 68.26% (see Table 4). Table 4. Distribution of styles of reasoning in both samples Banking regulation Sample Formal (Theoretical and (econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 7 3 0 0 1990–1994 5 4 1 0 1995–1999 8 2 0 0 2000–2004 8 2 0 0 2005–2009 7 2 0 1 2010–2014 2 8 0 0 Total 37 21 1 1 %share 61,64% 35% 1,68% 1,68% Systemic Risk Sample Formal (Theoretical/econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 2 1 5 2 1990–1994 2 4 2 2 1995–1999 3 2 1 4 2000–2004 3 6 0 1 2005–2009 3 3 2 2 2010–2014 6 3 1 0 Total 19 19 11 11 %share 31,6% 31,6% 18,33% 18,33% Banking regulation Sample Formal (Theoretical and (econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 7 3 0 0 1990–1994 5 4 1 0 1995–1999 8 2 0 0 2000–2004 8 2 0 0 2005–2009 7 2 0 1 2010–2014 2 8 0 0 Total 37 21 1 1 %share 61,64% 35% 1,68% 1,68% Systemic Risk Sample Formal (Theoretical/econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 2 1 5 2 1990–1994 2 4 2 2 1995–1999 3 2 1 4 2000–2004 3 6 0 1 2005–2009 3 3 2 2 2010–2014 6 3 1 0 Total 19 19 11 11 %share 31,6% 31,6% 18,33% 18,33% View Large Table 4. Distribution of styles of reasoning in both samples Banking regulation Sample Formal (Theoretical and (econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 7 3 0 0 1990–1994 5 4 1 0 1995–1999 8 2 0 0 2000–2004 8 2 0 0 2005–2009 7 2 0 1 2010–2014 2 8 0 0 Total 37 21 1 1 %share 61,64% 35% 1,68% 1,68% Systemic Risk Sample Formal (Theoretical/econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 2 1 5 2 1990–1994 2 4 2 2 1995–1999 3 2 1 4 2000–2004 3 6 0 1 2005–2009 3 3 2 2 2010–2014 6 3 1 0 Total 19 19 11 11 %share 31,6% 31,6% 18,33% 18,33% Banking regulation Sample Formal (Theoretical and (econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 7 3 0 0 1990–1994 5 4 1 0 1995–1999 8 2 0 0 2000–2004 8 2 0 0 2005–2009 7 2 0 1 2010–2014 2 8 0 0 Total 37 21 1 1 %share 61,64% 35% 1,68% 1,68% Systemic Risk Sample Formal (Theoretical/econometrical) empirical Informal Theoretical Historical Practitioners’ discourse 1985–1989 2 1 5 2 1990–1994 2 4 2 2 1995–1999 3 2 1 4 2000–2004 3 6 0 1 2005–2009 3 3 2 2 2010–2014 6 3 1 0 Total 19 19 11 11 %share 31,6% 31,6% 18,33% 18,33% View Large In particular, there is a strong difference in the importance of the practitioners’ and the historical styles in both samples. With a share of 1.68% for each style of the sample, receiving a total of 7 citations (historical) and 2 citations (practitioners) from the other sources, these styles are virtually absent in the banking regulation sample. In contrast, both have a share of 18.33% in the systemic risk sample and 26 and 31 citations, respectively, showing that these sources with these two styles have a considerable influence in the systemic risk sample (see Table 5). Table 5. Number of citations per style per sample Style/Sample Banking Regulation Sample Systemic Risk Regulation Sample Historical 7 26 Informal 55 12 Practitioner 2 31 Formal 131 41 Style/Sample Banking Regulation Sample Systemic Risk Regulation Sample Historical 7 26 Informal 55 12 Practitioner 2 31 Formal 131 41 View Large Table 5. Number of citations per style per sample Style/Sample Banking Regulation Sample Systemic Risk Regulation Sample Historical 7 26 Informal 55 12 Practitioner 2 31 Formal 131 41 Style/Sample Banking Regulation Sample Systemic Risk Regulation Sample Historical 7 26 Informal 55 12 Practitioner 2 31 Formal 131 41 View Large This difference in the importance of the practitioners’ and historical style of reasoning becomes even more visible when one analyses density distribution of citations, that is, the likelihood that the sources in each period of our both samples would cite sources that use these particular styles of reasoning in the aggregate sample (see Table 6). Whereas sources using formal and informal theoretical analysis are most likely to be cited in the banking regulation sample, it is sources using practitioners or historical style that are most likely to be cited in the systemic risk sample. The overall likelihood of citing a practitioners’ source is twice as high in the systemic risk sample as in the banking regulation sample. In contrast, the overall likelihood of a formal source being cited is more than twice as high in the banking regulation sample as in the systemic risk sample. Table 6. Density distribution between Periods and Classifications for the banking regulation and the systemic risk sample. Density goes from periods towards classification Period/Classification Banking regulation Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0,017 0 0 1990–1994 0,077 0,046 0,038 0 1995–1999 0,109 0,076 0,022 0,056 2000–2004 0,106 0,116 0,03 0,095 2005–2009 0,054 0,024 0,008 0,017 2010–2014 0,055 0,071 0,033 0,017 Total 0,0668 0,058 0,0218 0,031 Period/Classification Systemic risk Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0 0 0 1990–1994 0,038 0,015 0,087 0,075 1995–1999 0,045 0,047 0,022 0,067 2000–2004 0,058 0,03 0,045 0,178 2005–2009 0,029 0,014 0,038 0,05 2010–2014 0,016 0,008 0,02 0,008 Total 0,031 0,019 0,035 0,063 Period/Classification Banking regulation Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0,017 0 0 1990–1994 0,077 0,046 0,038 0 1995–1999 0,109 0,076 0,022 0,056 2000–2004 0,106 0,116 0,03 0,095 2005–2009 0,054 0,024 0,008 0,017 2010–2014 0,055 0,071 0,033 0,017 Total 0,0668 0,058 0,0218 0,031 Period/Classification Systemic risk Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0 0 0 1990–1994 0,038 0,015 0,087 0,075 1995–1999 0,045 0,047 0,022 0,067 2000–2004 0,058 0,03 0,045 0,178 2005–2009 0,029 0,014 0,038 0,05 2010–2014 0,016 0,008 0,02 0,008 Total 0,031 0,019 0,035 0,063 View Large Table 6. Density distribution between Periods and Classifications for the banking regulation and the systemic risk sample. Density goes from periods towards classification Period/Classification Banking regulation Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0,017 0 0 1990–1994 0,077 0,046 0,038 0 1995–1999 0,109 0,076 0,022 0,056 2000–2004 0,106 0,116 0,03 0,095 2005–2009 0,054 0,024 0,008 0,017 2010–2014 0,055 0,071 0,033 0,017 Total 0,0668 0,058 0,0218 0,031 Period/Classification Systemic risk Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0 0 0 1990–1994 0,038 0,015 0,087 0,075 1995–1999 0,045 0,047 0,022 0,067 2000–2004 0,058 0,03 0,045 0,178 2005–2009 0,029 0,014 0,038 0,05 2010–2014 0,016 0,008 0,02 0,008 Total 0,031 0,019 0,035 0,063 Period/Classification Banking regulation Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0,017 0 0 1990–1994 0,077 0,046 0,038 0 1995–1999 0,109 0,076 0,022 0,056 2000–2004 0,106 0,116 0,03 0,095 2005–2009 0,054 0,024 0,008 0,017 2010–2014 0,055 0,071 0,033 0,017 Total 0,0668 0,058 0,0218 0,031 Period/Classification Systemic risk Sample Formal Informal-theoretical Historical Practitioners’ Style 1985–1989 0 0 0 0 1990–1994 0,038 0,015 0,087 0,075 1995–1999 0,045 0,047 0,022 0,067 2000–2004 0,058 0,03 0,045 0,178 2005–2009 0,029 0,014 0,038 0,05 2010–2014 0,016 0,008 0,02 0,008 Total 0,031 0,019 0,035 0,063 View Large These differences in the distribution and the references to the different styles of discourse had a profound impact on the evolution of thinking about financial crises and systemic risk in the two samples, as we will show below. 3.2 Findings relevant to the banking regulation sample As stated above, the banking regulation sample is dominated by formal reasoning, following mostly a partial equilibrium approach from 1985 to 2005. Starting from period five in the banking regulation sample, network analysis starts to appear as a style of reasoning. Scholarly works in the first period of the banking regulation sample (1985–1990) emphasized bank runs as the source of systemic risk and insolvency risk as the economic rationale for banking regulation. For addressing these regulatory problems, these scholarly works emphasized deposit insurance and capital requirements as the main banking regulatory instruments. They discussed the desirability of risk-sensitive capital regulation, as non-risk-related capital ratios may induce excessive risk-taking by changing the composition of the asset side of the balance sheet. There was almost consensus that deposit insurance is almost the only solution for bank runs except in the case of money demand shocks, where lender of last resort or interbank markets can be appropriate alternatives. In this period, the term ‘systemic risk’ has almost never been mentioned, and the contagion and financial cycle forms of systemic risk were not identified as key regulatory problems. The works of the second and third periods of our banking regulation sample (1990–2000) are a natural extension of the works in the first period. The main new issue in the third period is debate regarding risk-based capital regulation, how to calculate it and its effects on the risk-taking incentives of banks. In line with this finding, papers with high hub values in the banking regulation sample (Bhattacharya et al., 1998; Goodhart et al., 1998; Santos, 2000; Barth et al., 2004) focus on the regulation of single institutions (with some minor exception by Goodhart et al. [1998]). Systemic risks are usually derived from the financial systems’ vulnerability to suffer bank runs, based on information asymmetries (cf. Bhattacharya et al., 1998), which in turn justifies deposit insurance, but only few sources engage multiple bank runs, that is panics (cf. Calomiris and Gorton, 1991). In sum, the three first periods in the banking regulation sample (1985–2000) that were dominated by formal analysis witnessed very little theoretical developments. In the fourth period, authors question the previous consensus on deposit insurance as the optimal tool for addressing bank runs and panics (the only form of systemic risk tackled in banking regulation sample until this period), pointing to implicit deposit insurance as a more efficient risk management instrument. In addition, they controversially debate the effects of banking liberalization and banking deregulation. Finally, we observe the emergence of macro-prudential regulation in the fourth period of our sample (Borio, 2003). However, this article has attracted most of its citations during the financial crisis years (143 citations from 2003 until 2007, with an average of 28.6 citations per year; 542 citations from 2008 until September 2015, with an average of 67.75 citations per year). The fifth period (2005–2009) of our sample is the period in which the shift from micro- to macro-prudential regulation occurs. In this period, three of the top-cited articles explicitly tackle issues related to macro-prudential regulation (Acharya, 2009; Jimenez and Saurina, 2006; Brunnermeier et al., 2009). The one study pre-crisis (Jimenez and Saurina, 2006) is written by two practitioners from the Bank of Spain and shows that credit growth in booms leads to increase of credit risk and higher percentage of non-performing loans. While the paper first appeared in January 2005, it received most of its citations post-crisis (27 citations over the three years preceding the crisis, with an average of 9 citations per year, and received 357 citations from 2008 until September 2015, with an average of 44.6 citations per year). This confirms that the shift to macro-prudential regulation took place post-crisis (a finding confirmed in our expert interviews). Pre-crisis, such thinking was largely confined to a few marginal groups and was largely irrelevant to central banks. As one interviewee, a prominent British economist, stated: the central bank governors and the majority of academics thought that the system was self-equilibrating because of EMH and all of that. And that remained the case until the sky fell in in 2007–08, when they realized it wasn’t. An American economist stated: ‘It was quite a small group that really worked on these things in the early 2000s and I don’t think … with the exception of the Scandinavian central banks, so in particular the Norwegians and especially the Swedes, these things weren’t … of that much interest. … I mean it was on the radar screen because of problems in Scandinavia and, you know, the Asian crisis, but it wasn’t … it wasn’t mainstream … in most central banks, the thought was this was something that was relevant for emerging economies more than advanced economies.’ The works in the fifth period exemplify an important shift from the micro-prudential approach of the previous four periods (1985–2005). Research on banking regulation in this period has shifted from focusing on protection of investors to protection of markets, from focusing on analysis of individual banks to the analysis of interaction of multiple banks, and from static analysis, mainly taking the form of comparative statics, to a dynamic analysis of risk and credit/financial cycles. Further, banking regulation research has shifted from partial equilibrium analysis and representative-bank models to network analysis and general equilibrium models. Most importantly, it has further shifted from the understanding of systemic risk based on bank runs to emphasize contagion and financial cycle sources of systemic risk. The sixth period of our sample (2010–2014) is an exploratory phase of the macro-prudential regulatory paradigm. The top five cited articles in this period relate to macro-prudential regulatory questions. Scholars begin to use the cognitive perspective of the macro-prudential paradigm to pose new questions and to provide new answers to old questions. Substantively, the top-cited articles began to provide a rationale for old regulatory instruments such as capital and liquidity ratios based on the contagion and financial cycle conceptions of systemic risk. As a result, they have made proposals for the amendment of these regulatory instruments to fit the macro-prudential perspective, through inter alia, counter-cyclical capital ratios, high-quality capital and higher capital ratios. The economic rationale of macro-prudential banking regulation in this period relies on credit cycles, transmission channel of risk-taking and the contribution of the financial institutions to credit crunch and fire sales. Methodologically, scholars began to advocate a movement from partial equilibrium to a more focus on general equilibrium and network analysis. Despite these important changes on the methodological and substantive fronts, these changes have been formulated in the standard neoclassical language of market failures, where systemic risk itself has been conceptualized as a negative externality, to which individual banks contribute. The works of this period indicate that the endogenous model of systemic risk, which has been Minsky’s initial idea (1992) and emphasized later by Borio (2003), seems to be the most challenging concept to model in comparison to systemic risk as a propagation risk (Galati and Moessner, 2013). As scholars began to take macro-prudential regulation seriously, the common trait of studies at this period is that these studies are exploratory (Borio and Zhu, 2012; Haldane and May, 2011; Hanson et al., 2011) or of a review nature (Galati and Moessner, 2013). It is astonishing to find that the top five cited articles in the area of banking regulation are informal, given the highly formalized nature of the economic literature beforehand. As the new regulatory paradigm is unfolding, these papers explore new territory and set the stage for future research. The exploratory informal nature of this period shows that informality may be required, particularly at exploratory phases of research. The financial crisis was required in order to lower the formalism barrier in economics with informal works being highly cited, setting the stage for more formal works to come, while illustrating the failures of existing formal models. As such, informal thinking indicates to the temporary limitations of formalism. Formal theoretical works of our sample prior to the crisis were not able to engage with financial cycle source of systemic risk and thus failed to discuss or recommend any macro-prudential regulatory interventions. This internal closure is confirmed when looking at the banking regulation sample from a citation network analysis (see fig. 1 below). It has a density of 0,048 and is thereby almost twice as dense as the entire network and the systemic risk sample, indicating a rather visible community of banking regulation scholars. The fact that scholars who were organized around a clear paradigm, cross-citing each other, can be seen from the fact that of the 60 sources of the banking regulation sample, it has only five papers which are not cited within the sample and only one source that is cited only once. Its biggest hubs are the scholarly works of Santos (2000) and Freixas and Rochet (1997), which are mathematical in nature. The five biggest authorities in the aggregate sample (Dewatripont and Tirole, 1994; Berger et al., 1995; Keeley, 1990; Jacklin and Bhattacharya, 1988; and Kim and Santomero, 1988) are from the banking regulation sample, four of which are formal, and all share the focus on micro-prudential regulation. Dewatripont and Tirole, while being informal in style, are rather conventional in content as they only engage with bank runs as the causes of systemic risk and focus on the corporate governance principal-agent analysis for rationalizing banking regulation. But why were formal models prior to the crisis ignoring the financial cycle as a source of systemic risk? One reason provided by the interviews is that economists and policymakers were operating on a belief system represented by a dominant paradigm whose main theoretical elements were rational expectations, the efficient capital markets hypothesis and self-equilibrating markets (interview with British Academic, formerly BoE, ECB economist, Fed economist). This academic thinking had a strong impact on the way that central bankers approached the issue of bubbles and liquidity problems. As the American economist explained: There’s a very famous paper by Goodfriend and King, who argued in the early 80s that in modern efficient markets liquidity wouldn’t be a problem because you can always borrow against good assets … people had very much this attitude, so liquidity just wasn’t a problem. The ECB economist7 gives another very strong example: [An example is] rational bubbles literature … Bubbles are actually the practical jargon for this endogenous build-up and unravelling of widespread imbalances … in the rational bubbles literature, in macromodels, there is a friction that ensures that the bubbles are optimal; the welfare is increased by the existence of bubbles … And of course in a world where this is the only academically recognized theory on bubbles... yes and now there are umpteen papers, which then say: yeah, yeah, yeah, yeah, all the practitioners always talk about bubbles and how bad financial crises are, but there is no scientific foundation for that ... this is incredible … if you just imagine, before the crisis this was the dominant literature about bubbles. According to this paradigm, credit cycles do not exist, and if they exist, they would be efficient responses to exogenous shocks. As he went on to explain, central bankers would abstain from building early warning systems and measurement for bubbles not only because of measurement problems but also because bubbles were regarded as an economically unsound concept. Our discourse analysis gives further important examples that corroborate the existence of this dominant paradigm. For example, the deregulation debate in the third period of the banking regulation sample and the belief in the stabilizing effects of inter-financial institutions connections (e.g. inter-bank markets) arose indirectly out of a belief in the efficient markets hypothesis. In addition, pre-crisis, the discourse analysis reveals the absence of behavioural finance; our banking regulation sample reflects a strong commitment to rationality assumption. This dominant framework then contributed to the failure of the theoretical formal banking regulation discourse to engage with the endogenous financial cycle type of systemic risk in our systemic risk sample. 3.3 Findings relevant to the systemic risk sample In contrast to the banking regulation sample, there is less thematic cohesion in the systemic risk sample. In the first period of our sample, systemic risk is discussed in historical terms in relation to the Great Depression and its lessons for potential deregulation in the present, where monetarists (Schwartz, 1987) point to price-level instability as the appropriate focus of regulatory action and Minskian scholars such as Kindleberger (1988) to fads in financial markets. The outliers to this debate are the formal analysis of Mankiw (1986) and Taylor and O’Connell (1985). A focus on contagion is also present in practitioners’ discourses (Brimmer, 1989), justifying Fed interventions in markets through concerns over possible contagion effects. In the first two sample periods, systemic risk is used as an intuitive concept related to financial crises. However, it does not receive a systematic definition until 1996 (Rochet and Tirole, 1996, p. 733), and only after the crisis does it crystallize into a measurable format (e.g. Acharya, 2009). Overall, the contagion effects of individual bank runs that cause banking panics are seen as a main component of systemic risk in all the sources (e.g. Bhattacharya and Thakor, 1993, p. 26; Kaufman, 1994), which leads to a strong coupling of the understanding of systemic risk and problems of liquidity (e.g. Freixas and Rochet, 1997; Allen and Gale, 2000). While the notion of contagion is the centrally shared concept in the entire sample, Kaufman (1994) in the mid-1990s complains of a dearth of empirical studies, replaced by ‘casual empiricism’, or just-so stories (see also Rochet and Tirole, 1996, p. 734). This dearth of empirical work in our sample is only overcome with Kaminsky and Reinhart (1999), who study the interrelationship between banking and foreign exchange crises empirically. A major analytical step forward is taken by Allen and Gale (2000), as they demonstrate the capability for using network analysis as a conceptual and analytical tool to take into account the interbank deposit market. They are thus able to point to the structural factors which can turn a liquidity shock into a financial crisis, questions which are further developed in the coming 14 years (see, e.g. Gai and Kapadia, 2010). Once the linkages of banks in terms of assets and liabilities are included in models, the problem of contagion can be modelled (Kaufman, 1994; Allen and Gale, 2000). This allows for models of networks of banks linked via common exposure to assets with the danger of joint overexposure, where systemic risk can be conceptualized as growing endogenously (Acharya, 2009). De Bandt and Hartmann’s review of the work on systemic risk (2000) reflects this partial endogenization of systemic risk rather well. By focusing on the characteristics of the financial system as a whole that make it more vulnerable to systemic risk than other sectors, the interconnection of financial institutions is emphasized and contagion is placed at the heart of the concept of systemic risk (p. 8). In this way, they develop a broad concept of systemic risk that integrates systemic events in banking and financial markets as well as the payment and settlement system. At this point, the literature then mostly relates to the triggers and amplifying mechanisms operating during financial crises, but not to their causes. Around 2000, the question of whether systemic risk is growing endogenously or exogenously is rather answered in favour of the latter, while at the same time modelling internal amplifying mechanisms. The endogenous position, which is strongly connected to the notion of financial cycles, booms and busts that are driven by self-reinforcing euphorias and panics (e.g. Kindleberger, 1988; Calomiris and Gorton, 1991; Minsky, 1992; Kaminsky and Reinhart, 1999; Borio and Lowe, 2002), only gains the upper hand after the financial crisis (Bisias et al., 2012). The reason for that is given by the ECB economist: And that (financial cycles), that was already in our old survey, but it was underexposed, since the literature was not there yet. The only ones who understood [systemic risk] actually used an empirical and a historical perspective. ... There was no real quantitative literature on that and that is why it was underexposed in academia. If you don’t have models, empirical or theoretical, which really deal with it, then those will not be recognized as relevant by the academic literature. Our interview with a prominent US economist working on systemic risk confirms both that the scholars who employed the informal historical approach were able to see a credit cycle and systemic risk as a historical regularity of capitalism, but at the same time were largely ignored: Financial crises started out as something people historians were very interested in … my colleague … Gary Gorton … started out with many papers on the history of crises and so on and his work is tremendously important and yet until fairly late it was not [recognized] … On the other hand, mathematical modellers up until 2007 remain agnostic on this issue, as they undertake a simple comparative statics analysis using asymmetric information (e.g. Bernanke and Gertler, 1990) or a structural comparison of network structures (Allen and Gale, 2000) rather than a long-run cyclical analysis to analyse financial fragility. This does not mean that these sources did not take their inspirations from historical sources. As a mathematical American economist put it, … One of the things which we try to stress is that small shocks can have enormous effects and this is a theme that goes back to Kindleberger’s work. It is historical. The difference then rather relates to method and the constraints that stem therefrom. This different relationship to cycles between formal and informal analysis can very well be observed in the contribution by Bernanke and Gertler (1990), in which they seek to define the term ‘financial fragility’. They develop a mathematical theory of how sudden credit squeezes can occur in the economy, based on an unexpected shock to the system. Using exogenous shocks to vary the leverage of borrowers and including moral hazard concerns allows them to generate more or less financial fragility in their model. Compared to the contributions of Minsky, Brimmer and Kindleberger of this period, what is remarkable is the different scope of the papers. Whereas the latter speak of cycles and longer-term regularities, Bernanke and Gertler can only provide a mechanism of an economic shift in a snapshot style. Questions of the financial cycles, of booms and busts, are instead placed centre stage by informal analysts such as Kindleberger (1988) or Borio and Lowe (2002), but only received an empirical investigation starting with Reinhart and Rogoff (2008), who undertook simple historical analyses of the run-up to financial crises using charts and simple descriptive statistics. Given the simplicity of the analysis, this prior empirical gap in the literature on financial cycles can thus not be explained with mathematical difficulties. Rather, it can be accounted for by the devaluation of historical approaches in the field of financial economics, as detailed above. Whereas the style of the sources in the sample is predominantly informal in nature, this changes abruptly in the final period, when six of the 10 sources are pursuing formal mathematical analysis to provide analytical measures for systemic risk. Overall, there is a certain lack of thematic cohesion in the systemic risk sample, evidence of which is provided by the low density of the citation network for systemic risk (0.026 vs. 0.048 for banking regulation, Fig. 2 below), signalling a fragmented sub-area of research on systemic risk and the lack of a visible well-connected sub-community of scholars. The only hub of the systemic risk sample network is De Bandt and Hartmann (2000), connected to the three authorities in the systemic risk sample (Allen and Gale, 2000; Rochet and Tirole, 1996; and Freixas et al., 2000), which are all formal. There are ten papers which receive no citation within the sample and six sources which only receive one citation. Fig. 2. View largeDownload slide Systemic Risk Sample Network. Colors correspond to the colors in Tables 1 and 2. Systemic risk sources are depicted as squares, while sources common to both samples are depicted as triangles. Fig. 2. View largeDownload slide Systemic Risk Sample Network. Colors correspond to the colors in Tables 1 and 2. Systemic risk sources are depicted as squares, while sources common to both samples are depicted as triangles. 3.4 Findings relevant to the relation between both samples We have observed a strong disjunction between the systemic risk sample and the banking regulation sample prior to the crisis. Although some of the important ideas regarding contagion and the financial cycle as sources of systemic risk underlying the macro-prudential regulation were discussed starting from the first period in the systemic risk sample (e.g. Kindleberger, 1988), these issues have only appeared, peripherally, in the fourth period of the banking regulation sample (2000–2004), and taken hold starting from the shift from micro- to macro-prudential regulation period (2005–2009). Pre-crisis, scholars in the banking regulation sample simply did not engage with the informal discourse on systemic risk, citing it only to justify their own work. In order to corroborate the thematic disjunction between the two samples, we analyzed the overall network formed by the two samples. Overall, the number of citations linking the two samples is 60. Citations primarily go from the banking regulation sample to the systemic risk sample (42 vs. 18). The density between the samples is 0.016, compared to the overall density of the network of 0.026, indicating a lower connectivity between the two samples than in the overall network and inside the two samples. A closer look at the incoming citations received by the works in the systemic risk sample from the banking regulation sample shows the pre-crisis disjunction in more detail. Out of the 42 citations, 23 appeared from 2008 onwards (after the paradigm shift) with an average citation of 2.875 per year, leaving 19 before the crisis with average citations of 0.826 per year. Four citations are made by Borio (2003), a macro-prudentially oriented paper that is included in the banking regulation sample due to its post-crisis popularity. An analysis of the 15 genuine citations by 10 sources (portrayed on the right-hand side of Fig. 3 above) showed that the only case in which a concept/result from the systemic risk sample was adopted by the banking regulation sample was by Hellmann et al. (2000), when quoting both works by Demirguec-Kunt and Detragiache (1998A, 1998B) which show an increasing frequency of financial crises in the past decades. The paper by Hellmann et al. (2000) then seeks to provide a rationale for this occurrence, linking it to capital market liberalization.8 The other 13 citations are either in literature reviews, cited in passing or as historic evidence of past crises. Fig. 3. View largeDownload slide Citations from the banking regulation sample to the systemic risk sample pre-crisis Fig. 3. View largeDownload slide Citations from the banking regulation sample to the systemic risk sample pre-crisis This disjunction can be linked to the fact that the endogenous sources of systemic risk, although discussed more seriously in the systemic risk sample, have not been well conceptualized or quantified in the early three periods. The formal nature of banking regulation research at that time then might explain why due to its lack of formalization these ideas were not taken up. This mismatch between the predominant styles in the two discourses is confirmed by the dominance of informal theoretical analysis in the last period of the banking regulation sample, at the same time that the formal analysis in the systemic risk sample becomes predominant. The macro-prudential analytical framework cannot be included in the formalized discourse on banking regulation without an underlying formalization and quantification of systemic risk. This drive for formalism and quantification can on the one hand be explained by the need for academics to propose models in order to gain status and recognition, as pointed out both by the ECB economist and the American economist. But it can further be explained by the fact that academics cannot convince policymakers about any policy changes without formal models and quantification (interview with a British economist working on systemic risk). This reliance of politicians and policymakers on unequivocal policy advice based on models is deeply enshrined in the convictions held among practically oriented economists, expressed in commonplaces such as that ‘one cannot manage what one does not measure’ (Bisias et al., 2012, p. 1). Particularly, models-based policies enhance the legitimacy in the face of policymakers (interview with FED economist).9 Given the absence of such formalization, scholars try to develop it, while texts in the banking regulation sample informally explore macro-regulatory issues, awaiting such formalization. In this sense, the recent developments in both samples signal a rapprochement of the two discourses. However, the use of mathematical models may well continue to be a problem, caused by the incremental nature of many of the mathematical economists’ contributions. This can be seen from a comparison of formal and informal analysts. Informal analysts could think about and discuss propagation/contagion and financial cycle as sources of systemic risk, whereas quantitative scholars could not establish a conversation without a model, and thus were much more reluctant to include these concepts in their discourse. In this sense, mathematical models constituted epistemological barriers and constrained theoretical progress, by excluding observed phenomena, which could not yet be accommodated in mathematical models. 4. Conclusion This paper traced the evolution of the economic discourses on systemic risk and banking regulation to better understand the shift from micro- to macro-prudential banking regulation. Our samples show that the informal-theoretical, historical and practitioners’ subsets of the economic discourse on systemic risk were able to discuss the contagion and financial cycle forms of systemic risk whereas the formal theoretical subset of the economic discourse, particularly of the banking regulation sample, overlooked these types of systemic risk almost completely pre-crisis. One evident reason was the rather late conceptualization, measurement and operationalization of systemic risk, which did not make it amenable to the predominantly formal analysis in the banking regulation sample. A second one is the impact of the initial model of bank runs operating with one representative bank (Rochet and Tirole, 1996, pp. 733ff). This model made the analysis of contagion channels other than the informational channel by definition impossible, as the tractability of the Diamond/Dybvig model from 1983 was achieved by mathematical simplification. Unfortunately, Kindleberger’s remark that ‘it is not evident that the historical record should be set aside in favour of easier mathematics’ was ignored (1988, p. 91). This communicative closure in mathematical economics, where concepts only exist if they can be included in a mathematical model (a trade-off between rigor and relevance extensively discussed in the economic methodology literature; see, e.g. Backhouse, 1998; Blaug, 2009), then partially explains our finding regarding the relation between both samples. But the failure of the discourse on banking regulation to engage with these sources of systemic risk pre-crisis cannot solely be attributed to being locked into formalism, as the literature was locked into a specific form of modeling, mainly non-structural partial equilibrium analysis. Formal network analysts were able to pinpoint endogenous sources of systemic risk prior to the crisis using mathematical means (most prominently Allen and Gale, 2000). Formalism in economic discourse becomes particularly constraining when it only includes specific and very limited number of approaches. Our analysis thus substantiates the failure of equilibrium thinking of neoclassical economics in capturing disequilibrium processes, particularly crises (cf. Arthur, 2013). Moreover, as the interviews demonstrate, this failure was also due to the fact that the dominant paradigm in financial economics was based on rational expectations, the efficient market hypothesis and self-equilibrating markets. This paradigm produced a literature declaring bubbles to be rational and therefore devalued their examination. This devaluation was also linked to the non-mathematical, historical research on cycles. This brings us to explore why practitioners’ and historical approaches were able to engage with contagion and financial cycles as sources of systemic risk. We suggest that these discourses enable scholars to gain forms of knowledge that the formal theoretical approaches may fail to produce. Being guided by empirical data, historical and practitioners’ approaches do not interpret the data by using the theoretical lens of partial or general equilibrium analysis. Rather, once they perceive the system to be repeatedly in disequilibrium as revealed in the regularities in the data, they begin to think eclectically about the sources of this risk, how it accumulates and how it could be addressed. These informal approaches, being empirically oriented, allow scholars to see what theoreticians of equilibrium models can hardly see. Freed from the neoclassical theoretical assumptions and the dominant paradigm in pre-crisis financial economics, scholars can proceed on the basis of hypotheses that are inconsistent with neoclassical theories (such as endogenous risk) to address these anomalies. This is the major strength of these approaches over formal theoretical approaches, particularly when they either cannot see or accommodate these anomalies into their models. On the other hand, as our citation network analysis indicates, the informal approach seems to impede consecutive, mutually oriented research that operates in the puzzle-solving (incremental) mode of normal science (Kuhn, 1962) that we could observe by using network analysis in the banking regulation sample. A further reason for the difficulties of integrating the notion of systemic risk in banking regulation studies is that it postulates a level that extends beyond the individual, where risks may accumulate independent of or maybe even because of rational action of actors at the micro-level (Baker, 2013), often because it is not directly observable (SRC, 2015, p. 21). This contradicts the neoclassical paradigm that assumes macro-level order based on rational action at the micro-level (Harnay and Scialom, 2015). The concept of systemic risk therefore is hardly reconcilable with the neoclassical theoretical frame. This explanation is consistent with the theorization of systemic risk as a negative externality in the fifth and sixth phase of the banking regulation sample, which tries to subsume the concept of systemic risk within the neoclassical market failure paradigm and measures it (Acharya et al., 2010), rather than rethinking the concept as an emergent property of the processes of the financial system (Haldane and May, 2011). This is particularly dangerous when modelling and quantification gives a misleading sense of certainty to regulators, while endogenous, hidden risks accumulate (interview with prominent British economist working on systemic risk, in SRC, 2015). In this respect, the Post-Keynesian approach to financial markets that endorses ‘a behavioral view of financial actors’ in contrast to ‘rational expectations’, and embraces ‘radical uncertainty’ in contrast to ‘measurable and quantifiable risk’ (Argitis, 2013; Asensio, 2013), can function as an important antidote. Challenging such notions of measurability and exclusive reliance on formal equilibrium modelling techniques is vital if the macro-prudential project is not to become too narrowly conceived. There will always be a genuine level of uncertainty rather than probability distribution ‘risk’, which is a challenge to model. Proprietary information, regulatory arbitrage and the evolution of new practices in shadow-banking make the reality one that requires a precautionary principle. Yet, this will not suffice. As the Post-Keynesian scholarship points out, the increasing financialization of the economy increases both the contagion and financial cycle sources of systemic risk (Gabor, 2015, 2016). This suggests that definancialization might be a necessary complement to or component of macro-prudential regulation; this deserves close attention in future research. In conclusion, our discourse analysis and the in-depth interviews have shown that formal theoretical analysis, coupled with the exclusion of practitioners’ and historical approaches from banking regulatory studies, have impeded the evolution of macro-prudential thinking in the dominant literature on banking regulation prior to the crisis. The economic discourse on banking regulation pre-crisis, dominated by formalism, was driven by problems that could be modelled given the predominant formal methods, rather than being concerned with the problems practitioners face. The fact that the post-crisis samples on banking regulation have been exploratory and informal in nature shows that informalism was needed for developing macro-prudential thinking. Certain forms of formalism, particularly non-structural partial equilibrium analysis, intensified the obstacles to the evolution of a macro-prudential paradigm prior to the crisis. The sharp rise of formal studies in the last period of our systemic risk sample signals a shift to formalism at the cost of informal styles of reasoning. The danger is that scholars, not cognizant of the methodological limitations that prevented them from engaging with the financial cycle in the first place, now reproduce these limitations when tackling them post-crisis. This paper investigated the dominant economic discourse, as represented by published scholarly work, but has not fully captured the way it interacts with the actual processes of policymaking itself, the organizational routines and policy paradigms which drive it. Future research is needed to investigate the degree of correlation of both discourse and policymaking and their interrelations, particularly the channels through which (academic) economic discourse influences policymaking. Here, our interview material already provides important hints as to the (de-)legitimizing function of academia in the pursuit of regulatory projects (such as early warning systems). Future research is also needed to better understand the drive for formalism in the economic discourse which, as briefly highlighted in this paper, is based on the need to provide unequivocal numbers to policymakers. This interaction of particular forms of knowledge production with policymaking spheres is then the next step in the research agenda to understand what impacts the regulatory outlook on financial markets. Bibliography Acharya , V. V . 2009 . A theory of systemic risk and design of prudential bank regulation , Journal of Financial Stability , vol. 5 , 224 – 55 Google Scholar CrossRef Search ADS Acharya , V. V. , Pedersen , L. H. , Philippon , T. and Richardson , M . 2010 . ‘Measuring Systemic Risk’ , NYU Stern School of Business Working Paper Allen , F. and Gale , D. M . 2000 . Financial contagion , Journal of Political Economy , vol. 108 , 1 – 33 Google Scholar CrossRef Search ADS Asensio , A . 2013 . The Achilles’ heel of the mainstream explanations of the crisis and a Post Keynesian alternative , Journal of Post Keynesian Economics , vol. 36 , no. 2 , 355 – 80 Google Scholar CrossRef Search ADS Argitis , G . 2013 . The illusions of the ‘new consensus’ in macroeconomics: a Minskian analysis , Journal of Post Keynesian Economics , vol. 35 , no. 3 , 483 – 505 Google Scholar CrossRef Search ADS Arthur , W. B . 2013 . ‘Complexity Economics: A Different Framework for Economic Thought’ , Santa Fe Working Paper 2013-04-012, available at http://www.santafe.edu/media/workingpapers/13-04-012.pdf [accessed 10 August 2015 ] Backhouse , R. E . 1998 . If mathematics is informal, then perhaps we should accept that economics must be informal too , Economic Journal , vol. 108 , 451 Google Scholar CrossRef Search ADS Baker , A . 2013 . The new political economy of the macroprudential ideational shift , New Political Economy , vol. 18 , no. 1 , 112 – 39 Google Scholar CrossRef Search ADS Baker , A . 2015 . Varieties of economic crisis, varieties of ideational change: how and why financial regulation and macroeconomic policy differ ?, New Political Economy , vol. 20 , no. 3 , 342 – 66 Google Scholar CrossRef Search ADS Bank of England . 2009 . ‘The Role of Macroprudential Policy’ , Discussion Paper Barth , J. R. , Caprio , G. and Levine , R . 2004 . Bank regulations and supervision: what works best ?, Journal of Financial Intermediation , vol. 13 , 205 – 48 Google Scholar CrossRef Search ADS Berger , A. , Herring , R. J. and Szegö , G. P . 1995 . The role of capital in financial institutions , Journal of Banking & Finance , vol. 19 , 393 – 430 Google Scholar CrossRef Search ADS Bernanke , B. S . 2011 . ‘Implementing A Macroprudential Approach to Supervision and Regulation’ , Resource document, 47th Annual Conference on Bank Structure and Competition , Chicago, IL , http://www.federalreserve.gov/newsevents/speech/bernanke20110505a.htm [accessed 5 May 2014 ] Bernanke , B. and Gertler , M . 1990 . Financial fragility and economic performance , Quarterly Journal of Economics , vol. 105 , 87 – 114 Google Scholar CrossRef Search ADS Bhattacharya , S. , Boot , A. W. A. and Thakor , A. V . 1998 . The economics of bank regulation , Journal of Money, Credit and Banking , vol. 30 , 745 – 70 Google Scholar CrossRef Search ADS Bhattacharya , S. and Thakor , A. V . 1993 . Contemporary banking theory , Journal of Financial Intermediation , vol. 3 , 2 – 50 Google Scholar CrossRef Search ADS Black , J . 2013 . ‘Seeing, Knowing, and Regulating Financial Markets: Moving the Cognitive Framework from the Economic to the Social’ , Resource document, LSE Law , Society and Economy Working Papers 24/2013, http://www.lse.ac.uk/collections/law/wps/WPS2013-24_Black.pdf [accessed 10 September 2014 ] Bisias , D. , Flood , M. , Lo , A. W. and Valavanis , S . 2012 . A Survey of Systemic Risk Analytics , Office of Financial Research, US Department of the Treasury Blaug , M . 2009 . The trade-off between rigor and relevance: Sraffian economics as a case in point , History of Political Economy , vol. 41 , 2 Google Scholar CrossRef Search ADS Borio , C . 2003 . Towards a macroprudential framework for financial supervision and regulation ?, CESifo Economic Studies , vol. 49 , 181 – 215 Google Scholar CrossRef Search ADS Borio , C . 2009 . Implementing the macroprudential approach to financial regulation and supervision , Financial Stability Review , vol. 13 Borio , C. and Lowe , P . 2002 . ‘Asset Prices, Financial and Monetary Stability: Exploring the Nexus’ , BIS Working Papers No. 114 Borio , C. and Zhu , H . 2012 . Capital regulation, risk-taking and monetary policy: a missing link in the transmission mechanism ?, Journal of Financial Stability , vol. 8 , 236 – 51 Google Scholar CrossRef Search ADS Brimmer , A. F . 1989 . Distinguished lecture on economics in government: central banking and systemic risks in capital markets , Journal of Economic Perspectives , vol. 3, no. 2, 3 – 16 Brunnermeier , M . 2008 . ‘Deciphering the Liquidity and Credit Crunch 2007–08’ , National Bureau of Economic Research Working Paper 14612 Brunnermeier , M. , Crockett , C. , Goodhart , C. , Hellwig , M. , Persaud , A. D. and Shin , H. S . 2009 . The Fundamental Principles of Financial Regulation: Geneva Reports on the World Economy , London and Washington, DC , Centre for Economic Policy Research, Brookings Institution Press Calomiris , C. W. and Gorton , G . 1991 . The origins of banking panics: models, facts and bank regulation , pp. 109 – 74 in Hubbard , R. G. (ed.), Financial Markets and Financial Crises , Chicago , University of Chicago Press Colander , D . 2004 . ‘From Muddling Through to the Economics of Control: View of Applied Policy from J. N. Keynes to Abba Lerne’ , Middlebury College Economics Discussion Paper No. 04-21 Claessens , S. and Kodres , L . 2014 . ‘The Regulatory Responses to the Global Financial Crises: Some Uncomfortable Questions’ , IMF Working Paper, WP/14/46 Crockett , A . 2000 . ‘Marrying the Micro- and Macro-Prudential Dimensions of Financial Stability’ , Speech at the Eleventh International Conference of Banking Supervisors Crombie , A . 1994 . Styles of Scientific Thinking in the European Tradition , 3 vols ., London, Duckworth Danielsson , J. , Song Shin , H. and Zigrand , J.-P . 2012 . Endogenous and systemic risk , NBER chapters, 12054, National Bureau of Economic Research , Cambridge, Massachusetts De Bandt , O. , and Hartmann , P . 2000 . ‘Systemic Risk: A Survey’ , European Central Bank Working Paper Series No. 35 Demirguec-Kunt , A. and Detragiache , E . 1998a . The Determinants of Banking Crises in Developing and Developed Countries . IMF Staff Papers , vol. 45, no. 1. Demirguec- Kunt , A. and Detragiache , E . 1998b . Financial Liberalization and Financial Fragility . in Boris Pleskovic and Josteph e . Stiglitz (eds.), Annual World Bank Conference Development Economics, 1997 . Washington, DC , World Bank , 1998 , pp. 303 – 31 . Dewatripont , M. and Tirole , J . 1994 . The Prudential Regulation of Banks , Cambridge, MA , MIT Press Diamond , D. W. and Dybvig , P. H . 1983 . Bank runs, deposit insurance, and liquidity , Journal of Political Economy , vol. 91 , 401 – 19 Google Scholar CrossRef Search ADS Freixas , X. , Parigi , B. and Rochet , J.-C . 2000 . Systemic risk, interbank relations, and liquidity provisions by the central bank , Journal of Money, Credit and Banking , vol. 32 , part 2: What should central banks do?, 611 – 38 Google Scholar CrossRef Search ADS Freixas , X. and Rochet , J.-C . 1997 . Microeconomics of Banking , Cambridge, MA , MIT Press Gabor , D . 2015 . The IMF’s rethink of global banks: critical in theory , orthodox in practice , Governance, vol. 28 , no. 2 , 199 – 218 Gabor , D . 2016 . Rethinking financialization in European banking , pp. 81 – 100 in Cozzi , G. , Newman , S. and Toporowski , J . (eds.), Finance and Industrial Policy , Oxford , Oxford University Press Google Scholar CrossRef Search ADS Gai , P. and Kapadia , S . 2010 . Contagion in financial networks , Proceedings of the Royal Society , vol. 466 , no. 2120, pp. 2401–23 . Google Scholar CrossRef Search ADS Galati , G. and Moessner , R . 2013 . Macroprudential policy—a literature review , Journal of Economic Surveys , vol. 27 , pp. 846 – 78 Gigliobianco , A. and Giordano , C . 2012 . Does economic theory matter in shaping banking regulation? A case-study of Italy (1861–1936) , Accounting, Economics, and Law , vol. 2 , no. 1 Goodhart , C . 2011 . The Basel Committee on Banking Supervision: A History of the Early Years 1974–1997 , Cambridge , Cambridge University Press Google Scholar CrossRef Search ADS Goodhart , C. , Hartmann , P. , Llewellyn , D. , Rojas-Suarez , L. and Weisbrod , S . 1998 . Financial Regulation: Why, How and Where Now? , New York , Routledge Haldane , A. G. and May , R. M . 2011 . Systemic risk in banking ecosystems , Nature , vol. 469 , 351 – 55 Google Scholar CrossRef Search ADS Hall , P. A . 1993 . Policy paradigms, social learning, and the state: the case of economic policy making in Britain , Comparative Politics , vol. 25 , 275 – 96 Google Scholar CrossRef Search ADS Hanson , S. , Kashyapand , A. and Stein , J . 2011 . A macroprudential approach to financial regulation , Journal of Economic Perspectives , vol. 25 , 3 – 28 Google Scholar CrossRef Search ADS Harnay , S. and Scialom , L . 2015 . The influence of the economic approaches to regulation on banking regulations: a short history of banking regulations , Cambridge Journal of Economics , vol. 40 , issue 2 , pp. 401 - 26 . Hellmann , T. F. , Murdock , K. C. and Stiglitz , J. E . 2000 . Liberalization, moral hazard in banking, and prudential regulation: are capital requirements enough ?, American Economic Review , vol. 90 , 147 – 65 Google Scholar CrossRef Search ADS Jacklin , C. and Bhattacharya , S . 1988 . Distinguishing panics and information-based bank runs: welfare and policy implications , Journal of Political Economy , vol. 96 , 568 – 92 Google Scholar CrossRef Search ADS Jimenez , G. and Saurina , J . 2006 . Credit cycles, credit risk, and prudential regulation , International Journal of Central Banking , vol. June, 65 – 98 Kaminsky , G. L. and Reinhart , C. M . 2000 . On crises, contagion, and confusion , Journal of International Economics , vol. 51 , 145 – 68 Google Scholar CrossRef Search ADS Kaminsky , G. L. and Reinhart , C. M . 1999 . The twin crises: the causes of banking and balance-of-payments problems , American Economic Review , vol. 89 , no. 3 , 473 – 500 Google Scholar CrossRef Search ADS Kaufman , G. G . 1994 . Bank contagion: a review of the theory and evidence , Journal of Financial Services Research , vol. 8 , 123 – 50 Google Scholar CrossRef Search ADS Keeley , M. C . 1990 . Deposit insurance, risk, and market power in banking , American Economic Review , vol. 80 , 1183 – 1200 Kim , D. and Santomero , A. M . 1988 . Risk in banking and capital regulation , Journal of Finance , vol. 43 , 1219 – 33 Google Scholar CrossRef Search ADS Kleinberg , J . 1999 . Authoritative source in a hyperlinked environment , Journal of the ACM vol. 46 , no. 5, 603 – 32 Google Scholar CrossRef Search ADS Kindleberger , C. P . 1988 . The International Economic Order: Essays on Financial Crisis and International Public Goods , Hempstead , Harvester Wheatsheaf Kuhn , T. S . 1962 . The Structure of Scientific Revolutions , Chicago , University of Chicago Press Levieuge , G . 2009 . The bank capital channel and counter-cyclical prudential regulation in a DSGE model , Recherches economiques de Louvain , vol. 75 , June Mankiw , N. G . 1986 . The allocation of credit and financial collapse , Quarterly Journal of Economics , vol. 101 , 455 – 70 Google Scholar CrossRef Search ADS Mayring , P . 2010 . Qualitative Inhaltsanalyse , Wiesbaden , VS Verlag für Sozialwissenschaften Google Scholar CrossRef Search ADS Marcussen , M . 2013 . The triumph and despair of central banking , pp. 19 – 35 in Kauppi , N. and Madsen , M. R . (eds.), Transnational Power Elites: The New Professionals of Governance, Law and Security , Routledge , Routledge Studies in Liberty and Security Minsky , H. P . 1992 . ‘The Financial Instability Hypothesis’ , Jerome Levy Economics Institute Working Paper Morgan , M. S . 2012 . The World in the Model: How Economists Work and Think , Cambridge and New York , Cambridge University Press Google Scholar CrossRef Search ADS Reinhart , C. M. and Rogoff , K . 2008 . This time is different: a panoramic view of eight centuries of financial crises , Annals of Economics and Finance , vol. 15 , 1065 – 1188 Rochet , J.-C. and Tirole , J . 1996 . Interbank lending and systemic risk , Journal of Money, Credit and Banking , vol. 28 , part 2: Payment systems and public policy risk, efficiency, and innovation, 733 – 62 Google Scholar CrossRef Search ADS Santos , J . 2000 . Bank capital regulation in contemporary banking theory: a review of the literature , Financial Markets, Institutions & Instruments , vol. 10 , 41 – 84 Google Scholar CrossRef Search ADS Schwartz , A. J . 1987 . Real and pseudo-financial crises , pp. 271 – 88 in Schwartz , A. J . (ed.), Money in Historical Perspective , Chicago , University of Chicago Press Google Scholar CrossRef Search ADS Seabrooke , L. and Tsingou , E . 2014 . Distinctions, affiliations, and professional knowledge in financial reform expert groups , Journal of European Public Policy , vol. 21 , no. 3 , 389 – 407 Google Scholar CrossRef Search ADS Seabrooke , L. and Tsingou , E . 2009 . ‘Revolving Doors and Linked Ecologies in the World Economy: Policy Locations and the Practice of International Financial Reform’ , Centre for the Study of Globalisation and Regionalisation Working Paper 260/09 Smaga , P . 2014 . ‘The Concept of Systemic Risk’ , Systemic Risk Centre Special Paper No. 5, London School of Economics and Political Science, August, available at http://ssrn.com/abstract=2477928 [accessed 13 September 2015 ] Systemic Risk Council (SRC) . 2015 . Identifying Systemic Risks , in Systemic Risk: What Research Tells Us and What We Find Out , LSE–Systemic Risk Council , July, 21ff Taylor , J. B . 2009 . ‘The Financial Crisis and the Policy Responses: An Empirical Analysis of What Went Wrong’ , Resource document, Social Science Research Network , http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1327252 [accessed 8 August 2014 ] Taylor , L. and O’Connell , S. A . 1985 . A Minsky crisis , Quarterly Journal of Economics , vol. 100 , 871 – 85 Google Scholar CrossRef Search ADS Footnotes 1 We decided to use Google Scholar rather than Web of Science to avoid a selection bias caused by the focus on academic articles published in peer-reviewed academic journals. Thereby we could also include books and papers published by practitioners. 2 Scholarly works obtain a high hub value if they cite a lot of highly cited papers in the sample, whereas authorities are those scholarly works that are highly cited by other important papers. These measures reflect the importance of a paper in the network as the number of the out- and in-going ties are weighted with the importance of the paper to which a reference is made (Kleinberg, 1999). Publications with high hub values can be analysed as synthesizing the current state of research, which is confirmed by the fact that most of the publications with high hub values in our sample are literature reviews (DeBandt and Hartmann, 2000; Bhattacharya et al., 1998; Santos, 2000). The authorities are considered as experts by the authors that form part of the network itself, usually because they contain novel, fundamental concepts for the literature on the topic. 3 Two of the authors are not among the top 10, but we have included them as early and persistent voices of macroprudential concerns since the early 2000s, as corroborated by scholarly works from the time. 4 For a defence of this style of reasoning as a guiding principle to economic policymaking, see Colander (2004). 5 Similar to Borio (2003), Minsky’s piece from 1992 titled ‘The financial instability hypothesis’ enters our top-cited articles in the second period of systemic risk sample only when we include the citations it receives post-financial crisis. If not, this piece cannot reach our top-cited articles. This piece shares the informal discourse with the pieces of Borio, Brunnermeier and Kindleberger. 6 The analysis of banking regulation within DSGE models has become more pronounced post-financial crisis (Levieuge, 2009). Although this in principle presents progress, as financial markets are now explicitly included in models of the economy, it still bears all the problems associated with equilibrium analysis: the neglect of evolutionary change and the presumption of rational actors with full or incomplete information (Arthur, 2013). In a sense, the literature risks to move from one unrealistic, mathematically constrained view on financial markets to a different one. 7 The interview of an ECB economist was conducted in German. All the quotes from this interview are therefore translated into English from German. 8 These sources occur after the East Asian Financial Crisis, and point to the increasing relevance this event bestowed upon such analyses. 9 However, even in such a mind-set, the degree of belief involved in such posturing must not be overlooked, a fact a BIS economist highlights when referring to the ontological status of the financial cycle: ‘both business cycle and financial cycle are difficult to measure. If you don’t believe in one, then you should not believe in the other’ (emphasis added). © The Author(s) 2017. Published by Oxford University Press on behalf of the Cambridge Political Economy Society. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
Journal
Cambridge Journal of Economics – Oxford University Press
Published: Oct 20, 2017