Rhetorical Metrics

The theoretical argument of this article is based on three insights. First, regulators live simultaneously in the worlds of normative expectation and technical specification, and they seldom consider one world without thinking about the other. Thus, regulatory metrics are not limited to providing technical guidance to policy determination: the role of numbers in regulation is constructed in sociotechnical deliberations. Second, the menu of social and technical issues deliberated in regulation is patterned after prevailing institutional models. Therefore, legitimating new rules that do not fit a prevailing model requires displacing that model in order to establish new sociotechnical legitimacy. Third, I argue that this can be done through the use of rhetoric.

Regulation Before Sociotechnical Deliberations

U.S. finance capitalism was of minor importance until after the Civil War, when monetary stability and the institutionalization of a network of stock exchanges caused a significant increase in the volume and variety of securities issuances.^{Footnote 4} The stock market crash of 1893 brought in its wake a cascade of corporate bankruptcies and violent labor conflict. Twenty years before the foundation of the Federal Reserve, legislators were at the forefront of designing rules of market control.^{Footnote 5} Unfortunately, Congress did not possess an administrative capacity to develop technical standards. And even if it did, metrics and their promises of “mechanical objectivity” [Porter Reference Porter1995: 149] were not yet a natural language among U.S. regulators who preferred to delegate the careful analysis of the facts to the judgment of investors. The Progressive Era thus offers an experimental case in which to observe the development of regulation before the establishment of sociotechnical deliberations.

Securities exchanges generated important questions of a general regulatory nature [Stäheli Reference Stäheli2013: 43-92]. Legislators recognized that securities markets offered financing opportunities for commerce, but they did not ignore the problems caused by a distinct class of speculators interested not in the securities’ underlying attributes but in the prospect of obtaining a profit from changes in prices. As speculation began to disrupt economic welfare, legislators focused on one critical question that continues to confuse regulators today: what is a sound investment and what model of regulation should be used to distinguish permissible from impermissible transactions?

The core of the debate was the problematic reality of securities: they bore no relation to precious metals of “intrinsic” value, and had the value that exchange participants gave them. In a statement that captured the frame of mind of many legislators, Nevada Senator John Jones noted that “value, being a conception of the mind, cannot be intrinsic or inherent. This cannot be difficult to perceive when we bear in mind the teaching of science, that color does not reside in the object in which we appear to see it, but is an attribute of the eye itself, and that sound is not a quality of bodies, but a property of the human ear” [U.S. Congress Reference Congress1893: 10]. This mental journey into the meaning of value implied that, just like color or sound, financial value could be assessed not in the intrinsic attributes of securities objects but in the character of the human subjects. Based on that cognitive frame, legislators embarked on an effort to regulate the trading public, and policy became a matter of sorting between financially literate traders having an “eye” from those unfit or unskilled to scientific conduct.

The insistence placed upon financial literacy also pervaded scholars’ framing of exchange value. Max Weber, assessing the economic function of the stock exchange, noted that “It is all a matter of the persons involved” [(1894) Reference Weber2000b: 332, emphasis in original]. He praised the virtues of the “strong hands” and the “large-scale capital-holders” on financial stability and advocated taking exclusionary measures against the “horde of small speculators armed with practically nothing beyond good lungs, a little notebook, and a pencil” [(1894) Reference Weber2000a: 367].

In Europe, state authorities passed laws mandating the exclusion of small speculators from exchange floors.^{Footnote 6} But calls to regulate the trading public were nowhere more controversial than in the U.S. for political reasons related to industry structure. Because of a ban on branch banking, the U.S. banking sector was a loose network of small banks, most of them located in rural areas [Bordo, Redish and Rockoff Reference Bordo, Redish and Rockoff2015]^{Footnote 7}. These were viewed by the financiers of the urban northeast as paragons of provincialism and inefficacity. Such critics were well deserved since they often operated despite elementary rules of prudence. But the model of financial literacy ran counter to a strong Jacksonian tradition of democratic inclusion: agrarian and antitrust interests in Congress saw risk-taking as a fundamental right in a free market economy, and opposed any forms of public control of securities exchanges [Hawley (Reference Hawley1966) 2015: 312]. As a result, entry controls were pursued by the stock exchanges themselves as measures of self-regulation.^{Footnote 8} The private enclosure of exchange floors expelled small and amateur traders from exchange floors but it did not stop them from speculating: despite private membership controls, the “small hands” continued to finance securities purchases on margin in call loans to New York brokers.

During the Panic of 1907, the market value of securities dropped 50%, destroying networks of trust, with dramatic consequences for depositors. Large volumes of watered stocks and fictitious securities brought heavy losses, hurting all investors irrespective of literacy or size [O’Sullivan Reference O’Sullivan2016: 189-230]. Unsound securities attributes pointed to elements of issuer-risk that were not directly about the education of traders. Nevertheless, public authorities continued to emphasize behavioral remedies to financial instability. New York Governor Charles Hughes reaffirmed calls to distinguish transactions “carried on by persons of means and experience [from those] carried on by persons without these qualifications” [Hughes Committee 1909: 4]. Unfortunately, the regulation of the trading public failed to secure political support, thanks in part to a public campaign orchestrated by the NYSE portraying America as a nation of investors [Ott Reference Ott2009]. But even if it did, it was a normative project without a concrete course of action: financial literacy resisted standardization, as it was virtually impossible for public authorities to assess with a degree of certainty how knowledgeable a trader was. Meanwhile, the risk attributes of securities were abstract notions and remained unregulated. Expertise at the level of tradable objects was developed for the first time by private statistical agencies, affording them an important regulatory role before the advent of the Federal Reserve in 1913.

Financial expertise experienced a drastic reorientation when private firms began to condense information into forecasts. Like legislators, forecasters reckoned that the trading public consisted of a majority of small traders with low management resources and little knowledge of financial risk. But in spite of calls to regulate the trading public on a hypothetical criterion of literacy, forecasters substantiated the growing popular perception after the Panic of 1907 that, since the illiterate speculator could not be inculcated or prevented from trading, the lesser evil was to “accept the services of professional speculators to enable him to speculate.”^{Footnote 9} Among forecasters were rating agencies like Moody’s Investors Service. In 1909, John Moody began to rate the risk incorporated in the bonds issued by railroad companies, borrowing the rating format from mercantile agencies [Olegario, Reference Olegario2006].^{Footnote 10} The informational reach of rating forecasts was centered on businesses and their securities issues. A rating forecast appraised all local environmental attributes defining the “intrinsic” quality of a security on a scale from “AAA” to “D”. Investors used rating-implied measures of intrinsic quality to identify instances of over- or under-valuation in securities pricing.

Circa 1910, the rating business was established as a profession and an industry [Friedman Reference Friedman2013]. Ratings quickly became popular as a way of ascertaining economic facts and maintaining transparency in lending and borrowing transactions. However, it is unclear how much investors’ use of them was based on trust or reputation. Investment recommendations were black boxes: rating firms did not explain how they arrived at judgment, making reverse engineering impossible. And since they did not publish records of performance, investors could not assess forecasting accuracy. Presumably, the source of investors’ interest in ratings was more cultural and symbolic than based on their reputation as providers of accurate information.^{Footnote 11} Accuracy notwithstanding, ratings possessed many desirable properties. They provided simple, visual representation of what people understood only impressionistically. Nature-evoking notions like intrinsic value grounded investment decisions in a reassuring principle of stability, beyond the seemingly chaotic behavior of the market. In this way, ratings talked and possessed discursive significance. In particular, intrinsic value evoked the value attributes of gold. But, unlike gold, whose value inhered from nature, the intrinsic value of bonds derived from a judgment expressed on a rating scale.

Ratings shaped investment decisions and, in so doing, they performed a regulatory role. But their capacity to standardize such decisions was limited in the absence of an external source of ratification. In this instance, private metrics had only latent regulatory authority because the rating industry was not backed by any structures of legitimacy on what counted as credible knowledge. Unregulated and unratified, the rating industry developed rival standards of risk: a rating assessment by Poor’s was not commensurable with a rating by Moody’s, which was difficult to compare with a Fitch rating. Private metrics possessed currency only within the narrow perimeter of their subscription base, thus weakening their capacity to standardize markets.

To summarize, the Progressive Era testifies to the difficult rise of regulation when policy intents and technical standards develop separately rather than interactively in sociotechnical debates. The model of financial literacy endorsed by legislators was circumscribed to behavioral injunctions directed at investors. Based on an appeal to moral responsibility, this model was restricted to being purely platonic, without any impact on securities transactions. Forecasters’ metrics too failed to normalize exchange because they lacked authority. A co-productionist perspective suggests that technical resources are implicated in the development of normative projects. So why didn’t legislators develop metrics to legitimate action at the level of securities? This is unclear without an institutional perspective. The prevailing model of financial literacy narrowed down the scope of legitimate action to the level of individual traders. Without statistical resources to clutch attention to the level of financial instruments, legislators did not see that securities in circulation had different merits. Hence securities were unregulated.

The Progressive Era legacy of failure would have important consequences on future courses of action. After the Federal Reserve Act of 1913, regulators became an actor in statistical production, devising their own metrics to ascertain and regulate financial developments. Metrics directed regulators to consider models of policy engagement at broader levels, beyond the person of traders. But regulators would also discover the meaning of developing statistical capacity in a context where private agencies were the prime movers, giving them an influence in future regulatory debates. The following section discusses the interactional context of science and society during the formative period of the Federal Reserve System (hereafter the “System”).

Sociotechnical Deliberations Without Procedures

The System was established with the purpose of furnishing currency via the rediscounting of bank loans arising from financing trade or production (“real bills” in economic parlance). The founders expected to adjust the monetary supply according to the productive needs of commerce. Yet, a major concern was the discovery that banks engaged in speculative trading using Federal credit, thus linking their own balance sheet position with that of the System. Speculative use of Federal credit was a major cause for concern, but regulators were divided as to what solutions the System could offer.

The debate on speculation polarized into two positions, each reflecting a distinct model of regulation. The “qualitative” position prescribed direct controls of banks’ investments to ensure that Federal lending was used to finance the productive needs of commerce. In contrast, the “quantitative” perspective favored making changes in the discount rate and government bond sales in the open market to regulate the cost and availability of credit and deter speculative absorption of Federal credit. The qualitative stance of regulation was put forward by the Federal Reserve Board (hereafter the “Board”) in continuity with the real bills doctrine, the cornerstone of the Federal Reserve Act.^{Footnote 12} It received the backing of the System’s founders, powerful figures in Congress such as Carter Glass, and much of the financial press. Advocates of the quantitative doctrine congregated at the Federal Reserve Bank of New York (hereafter “New York”) governed by Benjamin Strong. Among them were prominent economists such as John Maynard Keynes and Gustav Cassel who praised the countercyclical effects of open market operations. Bankers also tended to back New York, not because they fully adhered to the theoretical underpinnings of the quantitative position, but because they considered the Board a threat against the regional character of the System. Patterned into this doctrinal exchange was a political conflict over decision-making: the Federal Reserve statutes did not clarify who from the Board and New York was the primary locus of power [Meltzer Reference Meltzer2003: 137-138].

The fierce battle waged by the champions of each camp started after the war and continued unabated until the dawn of the Hoover administration. This debate was crucial because each camp developed distinct cognitive frames about the nature of financial risk and rival conceptions of market control. In this contest over regulatory models, the level at which regulation was exercised was the most contested issues. The Board urged action at the bank level to gain more active control over banks’ securities investments. Unfortunately, the legal basis for direct supervision was specious under the 1913 Act and, in any event, the Board lacked the sort of granular information at the level of bank units that qualitative action required. New York, on the other hand, viewed speculation as an issue of secondary order to the quantity of money in circulation, which it proposed to regulate on the basis of sectorial-level assessments of financial conditions. This sectorial approach afforded the regulation at-a-distance of a wide spectrum of financial problems but, according to its critics, it also made difficult the incarnation of a policy engagement at the level of the thousands of bank units and their securities transactions.

This level-perspective on regulation indicates that speculation could not be read in the terms of either of the two models. Each camp relied on an incomplete policy stance. The Board and New York justified their respective positions on speculation based on something the other did not see or did not want to see. As historians of the Federal Reserve have noted, this conflict paralyzed policy from the mid-1920s and was a major cause of the Great Depression [Friedman and Schwartz Reference Friedman and Schwartz1963: 254-265]. While entirely accurate, this representation is also partial: the same regulators who were pulled into the policy standoff also sought to find innovative ways to duck and dilute it. Perhaps the most interesting sociological aspect of the dispute was to provide fertile ground for regulators to invent and justify new policy paths under significant pressure.

In this instance, creative conflict resolution was not pursued procedurally à la Jasanoff in official sites of deliberation. During the interwar period, the System was young in years and confronted the usual teething troubles associated with new ventures. Because the nexus of power was untested and untried, actors acted outside role identities and most decisions were taken outside institutional worksites. Official bodies like the Open Market Investment Committee served more as sites to voice disagreements and wage conflicts over decisions already taken than as forums to build consensus. As for advisory committees, they were few and their function was not to generate research but to give private interests a vehicle for lobbying regulation.^{Footnote 13} Without a stable platform to unify debates, each camp began to construct its own deliberative sites and forums, in the hope of shaping the regulatory debate.

In the following section, I analyze how each camp laboring under the paradox of its own models engaged in a rhetorical exchange about metrics. The mobilization of metrics was key to resolving debates over competing projects. In this exchange, the strength of rival models depended on how well these were articulated and sustained in metrics. Those with metrics had leverage: while the Board struggled to lay down its qualitative position into technical standards, New York successfully stabilized quantitative policy into innovative metrics to become the dominant central banking authority of the 1920s.

A Rhetoric Of Metrics

In the early 1920s, the System began developing an independent research activity. Statistical capacity was an integral component of the regulatory-building process. And, in many ways, the buildup of research activities became patterned after the conflicted exchange discussed above. The Board and New York had their own team of statisticians organized into distinct research divisions. Statisticians worked in intimate connection with regulators, and they seldom considered scientific projects in any abstract or theoretical fashion. The development of metrics was part of an entrepreneurial effort to justify institutional deviance. Metrics provided cognitive clutch for use by regulators in articulating the relevance of their respective models and creating the impetus for change. Regulators derived from them three actionable arguments to persuade constituencies of the desirability and relevance of contested decisions.

The Ratification Of Rhetorical Change

After 1936, the OCC, FDIC, RFC and the twelve reserve banks used ratings in their daily work. Their rating requirements often differed, creating problems of cross-agency collaboration. Even more troublesome was the fact that rating firms employed distinct rating scales, making it difficult to compute aggregated measures of risk. In the summer months of 1937, regulators launched an investigation into the methods governing the production of ratings. This effort was coordinated from the Board in Washington by Leo Paulger and Glenn Goodman.^{Footnote 20} Paulger made arrangements to have representatives of Standard, Fitch, Poor’s, and Moody’s discuss ratings with Goodman. A year later, Fitch Vice President Henry Clancy called the outcome of this series of informal meetings the “Washington Ruling” (Paulger to unknown recipient, 1938, August 25).^{Footnote 21} The so-called ruling was the first comprehensive attempt to regulate ratings, some 40 years before the Securities Exchange Commission devised the Nationally Recognized Statistical Rating Organization license. The details of the ruling were never published in an official promulgation. This explains why it has been entirely forgotten, although it contributed significantly to making rating agencies what they are today.

The meetings reports are instructive of the inner workings of the rating industry during the interwar period. Poor’s and Standard said that they used weights to compute ratings but provided no indication about where and how they were placed. This led Goodman to suspect that their weighting was partial rather than comprehensive (Goodman to Paulger, 1937, Oct. 22). Moody’s rejected any such comprehensive approach: “it [has] been found impossible to rate securities on the basis of a fixed formula” (ibid.). Fitch stated that the determination of its ratings was the expression of “opinion concerning the standing of the issue with respect to others in the field” (Goodman to Paulger, 1937, December 12). Goodman also reported that rating agencies hesitated on the attitude to adopt regarding regulation. All four agencies complained that RRR had created unfortunate ties between agencies and bankers. Standard complained that it placed a “serious responsibility” upon the agency and decreased its capacity to operate autonomously (ibid.). In contrast, Moody’s praised rating-implied regulations since they generated more business.

Although they now had in mind Harold’s findings that rating agencies were poor forecasters (Goodman to Paulger, 1937, April 8), Paulger and Goodman barely challenged agencies’ officials on the matter of rating methodology. Their chief interest was to make rating scales uniform by standardizing rating agencies’ interpretations of the investment limits set by regulators.^{Footnote 22} Poor’s had a rating scale with more grades than its competitors, and made their first seven grades eligible according to regulatory rules (according to the OCC, the first four were eligible). This scale afforded a finer grain of evaluation. However, Poor’s also recognized that the deviation from regulatory recommendations made its ratings unpopular among bank examiners. Poor’s eventually dropped its A*****, A**** and A*** grades, and advised purchases of bonds in the first four grades. Standard and Moody’s also adjusted their recommendations, having previously advised purchasing bonds no lower than “A” (third grade). Only Fitch already complied with regulatory requirements. The standardization of rating scales was achieved just in time for the adoption of the Uniform Agreement on Bank Supervisory Procedures of June 27, 1938. In the examinations of the OCC, reserve banks, FDIC, and RFC, the agreement distinguished for the first time between investment-grade (authorized for purchase) and speculative (prohibited) bonds. This distinction would have an enormous influence on future financial developments.

The discussion continued throughout the summer of 1938. Rating agencies and regulators agreed on a majority rule to make ratings conclusive when at least three agencies agreed (Fitch to Paulger, 1938, August 29). Rating firms also accepted to mark the regulatory status of bonds in their manuals with special symbols X, Y, Z (Standard to Paulger, 1938, August 31). But a major disagreement occurred when Paulger asked agencies to compute the eligibility status of rated bonds in order to have regulatory requirements continuously updated by the rating agencies themselves rather than by the regulators. The proposal implied that agencies would swap information on rating changes, something that they refused because of the proprietary nature of information (Moody’s to Leavitt, 1938, June 25). Rating firms refused to become agencies working for regulators. They considered they had more to lose than to gain from merging with government since they could as well remain independent agencies and reap private profits from assuming a governmental role.

The Washington Ruling gave the rating industry its modern shape. Its impetus was regulators’ assessment that if metrics did not originate in regulation, they still had to bear its endorsement. One legacy of the ruling was to give monopoly privileges to Moody’s, Fitch, Poor’s, and Standard (the latter two merged in 1941), thereby narrowing the competitive dynamics of the rating industry. Ratings not hardwired in regulation fell into disuse. For instance, the American Bond Quotation Service, a Chicago-based rating firm, did not receive regulatory recognition and was put out of business in 1938 (Paulger to unknown recipient, 1938, September 2). The ruling also weakened alternative metrics like barometers. Barometers were not inferior to ratings: in fact, Babson had accurately predicted the market crash of 1929 [Friedman Reference Friedman2013: 43]. But barometers were calibrated to provide guidance on market cycles, a sectorial brand of metrics which became irrelevant after the Federal Reserve began to publish the variations of their price indexes. Barometers and other private metrics not hardwired in regulation fell into disuse. With barometers crowded out and ratings incorporated, regulators shrunk the dynamic ecology of private statistical firms that existed until the late 1920s. This is evidence of the paramount influence of regulatory practice on the industry of expert knowledge.

Sticky Rhetoric: The Imprint Of History On Current Debates

The swift rating colonization of U.S. regulation was the culmination of half a century of public debates about the stock exchange. This high point, which was only reached after a long preparation, was to have immense consequences on future financial developments. “The history of economic development is the history of regulation”, writes David Levi-Faur [2005: 14]. I extend this claim by suggesting that the history of regulation is in part the history of the influence of quantification on regulation.

The objective of this research was to understand the significance of metrics in regulation. It suggests the following pattern. First, regulation embeds social and technical developments into processes of co-production and co-evolution. Regulation acquires legitimacy in sociotechnical deliberations. Second, the deliberative context of science and society varies according to the degree of stability of the bureaucratic exchange in which regulation is debated. In routine contexts, regulators accommodate social and technical aspects following steady scripts incorporated in deliberative procedures. Sociotechnical interactions are reproductive of prevailing models of regulation. In unsettled environments, models are contested, and regulators engage in entrepreneurial efforts to manipulate extant models and legitimate institutional deviance. Such efforts often generate too much conflict to be absorbed in formally-constituted worksites, along the procedural lines described by Jasanoff. Here the mobilization of symbolic resources is required to make comprehensible the desirability and relevance of change. Third, I claim that this can be accomplished through the use of rhetoric. Regulators use metrics as rhetorical resources to exploit contradictions and ambiguities in extant models and justify creative paths of change. Ultimately, rhetoric is not suboptimal or irrational but rather a vital mechanism through which important policy decisions are taken in contexts of uncertainty.

Using the building of early U.S. securities regulation as an empirical case, this article makes four contributions to the social studies of regulation and quantification. First, it challenges the separatist perspective that characterizes many approaches of quantification in policy. The Progressive Era legacy of policy failures recalls that regulation works best when it is debated and enforced against metrics. In this sense, the general distrust of quantification that often characterizes critical legal studies is misplaced. For instance, Supiot’s claim [Reference Supiot2007: xi] that “calculating is not thinking” misses that metrics often provide a bedrock for regulatory reasoning. This research also suggests that economists have much to learn from regulators whose use of numbers rarely conforms to utility calculus, a too narrow perspective to serve as any guide for a pragmatic analysis of the practice of regulation. Quantification does more than provide neutral guidance to policy determination. It can evoke the development of innovative policy projects that were once unthinkable. This endogenous role of numbers also recalls that regulation is more than a technical achievement which warrants cost-advantage analysis. It is a normative activity whose perspective arises from the standpoint of market construction rather than stationary passivity within fixed market equilibria.

The second key insight of this paper is to emphasize the rhetorical value of metrics in policy debates. Examining the role of metrics in the building of securities regulation requires us to take seriously the rhetorical capabilities of metrics, as seriously as we take seriously their rationality. Rhetorical metrics combine two elements: cognitive clutches and actionable arguments. With the mechanical metaphor of “clutching,” I have sought to explain how metrics (in the context of this article, price indexes and credit ratings) help regulators articulate the desirability and relevance of change. Clutching cognition is integral to regulation where actors typically develop competing claims over the nature of risk and over the right level of policy action. During the interwar period, rival models of regulatory ordering of markets were available, each involving distinct costs and benefits. Regulators used metrics as rhetorical resources to clutch attention at their preferred levels. Today, the question of the levels of policy also permeates the conversation about environmental protection, food safety and the regulation of trade and finance. Many studies emphasize “cross-scale” governance, or the necessity to distribute regulation across policy sites and platforms [Lemos and Agrawal Reference Lemos and Agrawal2006]. This research suggests that the issue of effective regulation is not just about bridging geographic levels (the local and the global for instance); it is also about finding innovative ways to coordinate a plurality of policy targets at once so that relationships between different agents of risk (individuals, firms, and objects in circulation) are addressed comprehensively.

Clutching cognition is often insufficient to trigger a concrete course of action, until proponents of change find an actionable argument to make comprehensible the need for actual change. The use of persuasive language is key to ensuring the success of cognitive deviance. A situated perspective on the interpretative resources that policy-makers mobilize to resolve local conflicts is, therefore, key to understanding regulatory change. This article emphasizes three actionable arguments derived from metrics: confrontational, conciliatory, and interventionist. Although I suspect that this contribution on rhetorical change can find generalization in other policy domains, this remains an empirical question.

The third contribution concerns the history of quantification. This article presents a perfectly Porterian story, which even matches Porter’s chronology: at the turn of the 20th century, regulators resisted the use of metrics and instead entrusted investors to perform risk analysis. At the end of the 1920s, regulators abandoned their reliance on investors’ judgment in the name of standard metrics which became crucial reference points in regulation. Metrics had a legitimating effect on regulation: they gave new and ambitious regulatory projects a line of defense against charges of partiality and subjectivity. If this article broadly aligns with a Porterian account of the spread of mechanical objectivity in policy, its rhetorical approach offers precision to understand how this process occurred.

The first insight relates to Porter’s claim [1995: 17] that numbers “create new things and transform the meanings of old ones.” Here, Porterian studies of quantification should be brought into more explicit dialogue with institutional theory––a suggestion also made by Espeland [Reference Espeland1997: 1121]––to understand how the constitutive role of numbers is enabled in specific and historically-situated contexts. This article suggests that whether and how metrics transform regulation varies across deliberative contexts: in stable contexts, the preferences of regulators are formed and quantification is mostly reproductive of prevailing institutional models of government; but in unstable policy environments, models are contested, and regulators may use quantification as a symbolic resource to legitimate institutional change. Ultimately, by bringing in institutional theory, historians of quantification could become more appreciating of the validity and limits of the claim that “numbers create new things.”

Second, I suggest that studies of quantification ought to address more explicitly the question of rhetoric to unpack the symbolic role of metrics in institutional. This article’s rhetorical perspective can be understood as an effort to take seriously Porter’s claim [1995: viii] that numbers are “strategies of communication” with independent effects on policy-making. This communicative approach can be specified further along the rhetorical lines highlighted in this article. The two rhetorical components presented above are helpful in gaining perspective on precisely how numbers are communicated and with what policy effects. The component of cognitive clutch suggests that regulators can acquire metrics as rhetorical resources to expose contradictions in prevailing models. Actionable arguments are helpful to understand how the need for institutional change is communicated: either through confrontational or conciliatory arguments or through a rhetoric of intervention against the cost of inaction. Ultimately, numbers as strategies of communication can lead to fundamentally different ways of communicating trust and objectivity.

The last contribution of this research concerns the imprint of rhetorical decisions on current debates. The 2008 financial crisis revealed many problems plaguing rating production including conflicts of interest and profit-driven behaviors. In the run-up to the financial crisis, these structural problems resulted in rating firms bestowing generous ratings to debt instruments that were defaulting at a much higher rate than anticipated [MacKenzie Reference MacKenzie2011]. U.S. and European lawmakers called for the elimination of ratings from financial regulation. But despite strong political commitment, progress in that direction has been limited. Why superficial metrics like ratings continue to enjoy official regulatory status has proved a constant source of amazement for scholars. Considerations of accuracy did not drive the institutionalization of RRR (regulatory reliance on ratings) any more than they are today the sole factor in the regulatory debate. If so, regulators would have removed ratings a long time ago.

The sociohistorical analysis developed in this article offers an analytical perspective on the stickiness of rhetorical decisions. The difficult removal of ratings has first to do with their significance as regulatory metrics. Ratings have crossed the 20th century as standardized metrics standardizing exchange. Today, ratings pattern the market conduct of firms and individuals, making it difficult for regulators to remove them from provisions without at the same time weakening regulation. Thus, ratings verify the classical claim of path-dependency made by David [Reference David and McCloskey1970] that technologies, however artificial or superficial they might be, become “locked in” once used by everyone. Another explanation of the slow progress towards reform is the cognitive mechanism of clutching: removing ratings from regulation forgoes the capacity of regulators to maintain cognition at the level of financial objects. A major concern among regulators is the loss of granularity for assessing different levels of credit risk [Soroushian Reference Soroushian2016]. Without ratings, regulators are left with the alternative option of relying on banks’ internal risk assessments. This proposal revives “associationalism,” the policy stance favored under Hoover which prescribed voluntary partnerships between government and business in lieu of coercive regulation. As this article recalls, entrusting investors to carefully investigate their risk position is often assuming too much. Despite their inaccuracy having long been exposed, ratings continue to express important choices which have long been forgotten. By emphasizing metrics as a mere technical matter, critics of ratings miss the rhetoric of stability that ratings offer, and that is hard to resist even in the face of ratings’ technical limitations.

This study has shown that metrics have social and technical meanings in regulation. One must be sensitive to the sociohistorical context in which they acquired regulatory significance. Metrics assumed an important rhetorical role during the building of modern financial regulation. Today, they continue to have rhetorical value, beyond any consideration of efficacy. For the students of regulation and quantification, the conclusion is that the interactional context of science and society is key to analyzing regulatory outcomes and examining the standard metrics that regulators use to standardize human activities.