If two features define contemporary capitalism, they are first the tendency of each individual to increasingly bear alone the risks associated with living in a market society, and second the enmeshment of individuals in a totalizing system of surveillance constructed from our proliferating personal data. Two remarkable recent books demonstrate that these apparently novel features of 21st-century capitalism in fact have deep historical roots, and that their histories are intertwined in consequential ways.
But not only do these histories intertwine, they also inform each other, inviting the reader to consider what it means to be an “individual” in contemporary financialized capitalism. Dan Bouk’s engrossing history of the life insurance industry examines professional risk makers’ growing reliance on numbers (particularly as encoded in statistics) to narrate individual lives, rendering individual experience with ever greater precision. In Bouk’s telling, the rise of the actuary in the making and marketing of risk seems to move us closer to actually knowing the individual as a unique person—the “statistical individual”—if only to more closely calibrate contributions to insurance risk pools. Lauer’s captivating history of credit reporting appears to tell the opposite story, as quantitative techniques used to price risk in credit markets increasingly lift the individual from her particular circumstances, resulting in a disembodied, abstract self that Lauer describes as one’s “financial identity.”
Sorting out the discrepancies between these two accounts is not only of historical interest, because these two versions of the individual compete in accounts of contemporary society in the digital age: Do algorithms render us more transparent—and therefore known as singular individuals—to the systems that collect and process our personal data, or do they dissolve our individuality in ways that violate human subjectivity and personhood? I will argue in this essay that developments in insurance and credit markets over the last century raise the question of whether persons can (or should) be known as unique individuals by systems of risk pricing. I will suggest that the history of risk pricing in life insurance and consumer credit scoring reveals the quest to apprehend the individual qua individual, but that the object of this quest remains elusive. In fact, the elusiveness of the person may be a constitutive feature of financial capitalism.1
The Individualization of Risk in Life Insurance Markets
Dan Bouk’s meticulously researched account of the development of the American life insurance industry in the late 19th and early 20th centuries centers on a dialectic that defines the history of risk: on the one side, the tendency of risk makers to anchor knowledge of the individual in the “average” or “typical” experience; on the other, the tendency to assess risk by using information ever more precisely calibrated to each individual’s unique life story. The distinction between these two orientations relates to larger questions regarding how broadly—or narrowly—risk is shared in capitalist society, with the first approach hewing to age-old principles of social solidarity and the second anticipating a future in which risk is fragmented into discrete parcels and borne by each individual alone.
Much of the narrative in How Our Days Became Numbered hinges on the contest between these two approaches, with the invention of the “substandard policy” arguably the key moment in this drama. An important contribution of Bouk’s account of the development of the substandard policy is to make clear that deeply entrenched anxieties about racial difference propelled the individualization of risk in insurance markets. In postbellum America, life insurance was largely restricted to Northern cities, where major life insurance companies competed for the business of relatively affluent white men. When it became evident that the insurance industry could no longer survive on such a narrow foundation, life insurers began actively promoting insurance sales in Southern states. In addition, life insurance companies led by Metropolitan, Prudential, and John Hancock introduced so-called “industrial insurance” policies aimed at the working class.
Insurers were not prepared for the influx of African Americans who applied for these new policies, however, and they responded to the paucity of reliable mortality data for nonwhites by arbitrarily reducing the standard benefit offered to African American policyholders. When this practice ran afoul of an antidiscrimination statute passed in Massachusetts in 1884 (to be followed in short order by a wave of similar legislation in other Northern states), most companies simply stopped soliciting the business of African Americans.
With the suspension of African Americans from insurers’ risk pools, a challenge was laid down: How would insurers balance the pressing need to expand their customer base with the equally pressing need for reliable data describing the mortality experience of populations previously excluded from the actuary’s calculations? The search for better data was on, inviting some rather unconventional methods. Prudential enlisted a rogue statistician named Frederick Hoffman to compile data on African American mortality from varied local sources, including city auditors, health boards, prisons, and charities.2 (Hoffman was equally assiduous in conducting research on the mortality of Southern whites, lugging a typewriter from one Southern cemetery to the next in order to record mortality data from weathered gravestones.) While actuaries collected far-flung statistics, medical examiners honed an ever more invasive battery of tests to rule out—or increasingly rule in—risks, relying on extensive family histories, precision measurements of height and weight, newfangled medical devices such as the stethoscope, and (even more revolutionary) the chemical analysis of urine.
As statisticians calculated and doctors prodded, a meticulous medical director at New York Life named Oscar Rogers combined these approaches, conducting a systematic study of 25,000 individuals who had previously been rejected for insurance coverage. Rogers’s objective was to price “impairments”—particular risk factors such as a history of tuberculosis or employment in a hazardous industry that changed one’s expected mortality—so that some portion of the approximately 15 percent of applicants who were typically excluded from insurance coverage could be included in insurance risk pools (at a higher premium). As the fruit of Rogers’s labor, the “substandard policy” was born in 1896, accounting for 10 percent of New York Life’s business within a decade. Notably, this pricing technology directly encoded race by treating African Americans as “substandard” risks by default.
With the invention and diffusion of the substandard policy, Bouk catches the first glimpse of the emergence of a new risk paradigm that would seek to calibrate individual experience ever more precisely. Under this new paradigm, risk was personalized, enclosed in boxes more closely fitted to the individual’s past, present, and likely future. Personalization meant bringing more information to bear on each individual’s life story—rather than assessing risk across large undifferentiated groups, calculations were made within groups distinguished by body type, family history, occupation, and assorted indications of lifestyle. The actuarial profession parceled risk into smaller and smaller packages, insisting that each individual bear more fully the cost she contributed to insurers’ risk pools. In this regard, Bouk’s history demonstrates that the individualization of risk that we typically think of as a signature feature of contemporary neoliberalism was already strongly in evidence early in the 20th century.3
The increased individualization of risk brought with it expanded possibilities for surveillance. Critical in this regard was the development of new technologies for processing and storing data such as the index card, which gradually displaced the ledger book as the foundation of modern business recordkeeping. Companies such as New York Life, Equitable, and John Hancock initially began sharing index cards containing information about individuals who had been rejected from life insurance coverage—a black list, of sorts—but later circulated cards for individuals, including policyholders, with “impairments” of any kind. Ultimately, access to augmented data assisted insurers in making the shift from prediction to control—i.e., from attempting to gauge risks to engineering them. A heart lesion, excess body weight, or a history of tuberculosis no longer provided a rationale for rejecting an individual from coverage, but rather offered an inducement to produce healthier lives. The preventative check-up became a staple of insurance practice in the early 20th century, just as the Fitbit would offer insurers the opportunity to shape bodies into more appealing risk profiles a century later.
Ultimately, this new risk paradigm accelerated the quest for the “statistical individual,” the enigmatic figure, at once singular and generic, who occupies center stage in Bouk’s narrative. In seeking the statistical individual, risk makers relied on statistical techniques to attempt to determine a precise, unique value or price for each individual. Insurers had already taken an important step in this direction when they created the substandard policy, placing select individuals into narrowly defined risk classes that better approximated each individual’s actual cost to insure. But even as actuaries constructed increasingly refined risk classes, individuals were nevertheless given the average price for the group or class to which they were assigned, obscuring individual variation around this average.4 In this regard, there was something deeply paradoxical about relying on the “law of averages”—the basic rule in insurance markets5—to better assess individual qualities and characteristics.6
The form of individualization represented by numerical rating generated significant resistance from actuaries who argued that this method of pricing risk undermined risk sharing.
But if in this sense the substandard policy offered only a partial realization of the “statistical individual,” more radical possibilities were already visible on the horizon of insurance practice. Most notably, Oscar Rogers’s frenetic calculations of how particular “impairments” changed one’s risk profile, elevating or reducing the likelihood of mortality, offered insurers a novel way of approaching risk. The numerical rating method—the name eventually given to this approach—evaluated each of these “impairments” (e.g., a heavy build, a hazardous occupation, presence of a heart lesion, etc.) as independent risk factors, tallying and subtracting points in order to arrive at an individual score for each person. Unlike traditional actuarial methods—in which overweight coal miners with heart lesions were treated as a group, with each individual member of this group given the same exact price—the numerical rating method allowed insurers to evaluate risk factors one at a time, potentially relieving actuaries of the necessity of constructing risk classes.7 With the arrival of numerical rating, the “statistical individual” broke free of the group and struck out on her own.
Or so it appeared. When Rogers developed his novel method of scoring and tallying individual risk factors in order to price substandard policies, he intended to extend this method to standard risks as well, so that individuals with more “normal” expectations of mortality could also receive a unique price reflecting their precise risk.8 But this didn’t occur in Rogers’s own time, and it hasn’t occurred subsequently. In fact, the form of individualization represented by numerical rating generated significant resistance from actuaries who argued that this method of pricing risk undermined risk sharing—the basic principle of insurance as a social institution.9 The result is that for the vast majority of individuals who qualify for standard policies (in recent years, around 95 percent of all life insurance policyholders10), premiums are assessed on individuals who are presumed to represent the average risk of the group or class to which they are assigned. In this regard, the pricing of risk in life insurance markets is still largely a product of group membership, rather than a shifting aggregation of personal traits that cumulate, one by one, into a life.
Credit and the Invention of “Financial Identity”
Notwithstanding this group-identity aspect of insurance pricing, Bouk treats insurers’ growing reliance on increasingly sophisticated statistical techniques as moving risk makers closer to more fully apprehending the subject of actuarial calculations as a unique person. Josh Lauer’s riveting history of the credit reporting industry tells in many respects a contrary story. Here the introduction of quantification into the assessment of credit risk resulted in a more abstract and distant representation of the individual, a transformation of a concrete person into a “financial identity,” flesh and blood transfigured into paper. These disparate accounts in some ways reflect their different starting points, in the insurance and credit industries, respectively. If insurers’ efforts to individuate and differentiate risks represented a deviation from a longer history in which individual particularity was subordinated to overarching trends (i.e., “the law of averages”), credit men operated in a business that had embraced particularity from its earliest days. If anything, credit reporting before quantification was too personal, representing little more than a glorified gossip circle.
Lauer’s account begins with the establishment of the commercial credit rating industry, by a failed investor named Lewis Tappan. In the 1840s, Tappan turned his entrepreneurial energies (and questionable business ethics) toward collecting and selling information circulating in local business networks to potential creditors. Creditors had always drawn on such information, paying close heed to talk at the local tavern as well as traffic at the courthouse, but as commerce became increasingly extra-local in nature, there was a growing need to systematize these seemingly idiosyncratic sources. Tappan created a centralized subscription service with local attorneys serving as correspondents who filed credit reports in which the objective was to “extract and reproduce the individual’s local reputation for a national audience.” Eventually, credit rating agencies supplanted these rambling narrative reports with published reference books that provided abbreviated numerical ratings. Quantification was underway, and with it the creation of an increasingly disembodied financial subject.
Our Metrics, Ourselves
Lauer emphasizes the history of commercial credit rating because creditors operating in consumer markets imported key features of the reporting infrastructure developed for commercial use, including Tappan’s system of surveillance. The goal of consumer credit reporting was to provide a comprehensive portrait of the financial habits of the consumer. As with their commercial counterparts, consumer reporting agencies distilled this portrait into a compact summary published in a rating book (later an index card on file at the credit bureau). But unlike commercial credit rating, this portrait drew primarily on ledger data reflecting merchants’ cumulative experience with a given customer. The use of ledger data separated consumer reporting from the sometimes questionable sources relied on by commercial credit correspondents, making consumer credit reporting seem more reputable and less biased, “a neutral transcription of an individual’s own financial behavior, for which he or she alone was responsible.”
One notable aspect of these early credit reporting systems was the involuntary nature of credit surveillance. When a consumer made a purchase, she was entered into a ledger, transformed into data, held in the expansive filing systems of credit bureaus (and increasingly in the files of the credit departments of major retailers, as well). There was no opting out of this information network (unless one opted out of consumer society altogether), nor was it possible to track or monitor one’s movement through this system. Although we might assume that “surveillance capitalism” is a product of computerization,11 Lauer’s account convincingly shows that personal data circulated on the basis of a rudimentary set of office technologies in the first half of the 20th century—card filing systems, teletype machines, telephones, and pneumatic vacuum tubes. Even targeted sales promotions could be carried out on the basis of consumer information efficiently stored—and quickly retrieved—on punch cards. Long before computer algorithms scanned the text of a customer’s email in order to coax her next purchase, a discerning credit manager could make similar inferences about consumer buying habits and send an encouraging letter inviting select clients to the ready-to-wear department to personally experience the new inventory of spring coats.
This is not to say that computerization left practices in the credit industry undisturbed. Instead, Lauer’s account suggests, the introduction of computers into credit reporting represented a “seismic event” that reoriented the credit industry in a number of important respects. First, reliance on computers moved credit reporting toward even greater compression of individual experience. Cryptic codes that could be incorporated with a few keystrokes into programming languages replaced more “subjective” forms of evaluation, particularly those that hinged on the all-important personal interview. Quantification decreased the time necessary to process a loan and economized on data storage, but it also “pushed the concept of credit risk further toward impersonality and abstraction.” Second, the introduction of computers facilitated the widespread adoption of credit scoring technologies to price risk, beginning in the 1960s. Credit scoring relied on statistical techniques that allowed creditors to separately weight discrete variables in order to calibrate how much each factor added to the risk of default on a loan. A creditor then tallied an individual’s credit score by summing across all of these variables: those with scores above a given threshold received credit; those below this cutoff were denied credit. This simple decision rule vastly simplified the credit manager’s task, and ultimately enabled the extension of credit to higher- as well as lower-risk customers.
Even as risk makers devise ever more penetrating systems of individual surveillance and control, the status of the “person” who is enmeshed in these systems is uncertain.
Notably, credit scoring bore an affinity to the numerical rating method introduced in insurance in the opening decades of the 20th century. In fact, Bouk’s and Lauer’s accounts demonstrate that there was considerable cross-fertilization in the development of technologies for rating risk between insurance and credit over the course of the 19th and 20th centuries. In the case of credit scoring, the resemblance to methods of insurance pricing was not accidental, since creditors both envied and emulated the precision of actuarial science in constituting the “statistical individual.” But as we have seen, two different versions of the “statistical individual” inhabited insurance markets. One of these individuals clung tightly to the “average” mortality of the group; the other jettisoned the group and asserted her own unique life experience. The latter individual—constituted by tallying up a series of independently varying risk factors, irrespective of membership in any larger group or class—found a more comfortable home in credit markets than in insurance.
But rather than seeing the development of credit scoring as marking the ultimate triumph of the individual in markets for risk, Lauer discerns something else altogether. Notably, the first generation of credit scoring models had used retailers’ own loan files to build customized scoring systems specific to a particular merchant. Beginning in the 1980s, however, Fair, Isaac, and Company began to market a “generic” (FICO) score based on national credit bureau data, and knowledge of the financial subject became dependent on large statistical samples dislodged from time and place. As a result, the most “personal” representation of the financial subject—a unique score based on her particular values on a series of variables deemed relevant for determining creditworthiness—was placed at the farthest remove from lived social experience.
Rather than the individualization of risk, then, Lauer gives us an account of its abstraction: the removal of credit risk from where it had typically resided—in complex, messy human relationships—to a more sterile existence encoded in statistical data. Of course, this very process of abstraction also meant that personal information could travel more easily between contexts—credit, employment, and insurance markets—with the result that the financial subject is now observed and recorded in a domestic surveillance regime that has far surpassed its early 20th-century predecessor in its reach.
These two accounts, then, converge on a seeming paradox: even as risk makers devise ever more penetrating systems of individual surveillance and control, the status of the “person” who is enmeshed in these systems is uncertain. Bouk’s history of life insurance traces the quest for the “statistical individual” in the development of novel techniques for pricing risk in the early 20th century, but, as the author notes, the object of this quest ultimately proved elusive. Ironically, treating a life insurance policyholder as an “individual” in most instances means giving that person the average value of whatever group or class to which she is assigned. The individual’s particularity—seemingly, the very essence of what it is to be an individual12—is eclipsed.
A century later, as Lauer’s account of consumer credit scoring makes clear, risk pricing techniques have evolved—but the scored self seems no more an “individual” than does the actuarial self.13 If anything, the “personalization” of capitalism enabled by extensive data mining has even more completely effaced the person; fragments of data are assembled as a “person” in one moment (as long as a transaction endures), and abruptly reconfigured in another assemblage in the next.14
We are left with the startling conclusion that as financial capitalism constructs totalizing systems of surveillance, the flesh-and-blood person who is surveilled may in a certain sense escape the grasp of these systems. Whether this points to hidden potentials for liberation or simply the hollowing out of human subjectivity in new technologies of risk pricing is a question that is unresolved in this still unfolding story of risk and its quantification.
I gratefully acknowledge helpful comments I received on this essay from Daniel Hirschman, Ben Platt, and Caitlin Zaloom.
This article was commissioned by Caitlin Zaloom.
- See Greta R. Krippner and Daniel Hirschman, “The Person of the Category: Pricing Risk and the Politics of Classification,” unpublished manuscript, Department of Sociology, University of Michigan; Liz Moor and Celia Lury, “Price and the Person: Markets, Discrimination, and Personhood,” Journal of Cultural Economy, vol. 11 (2018). ↩
- Paradoxically, Hoffman’s extensive compilations of statistical data were mobilized in support of his controversial thesis that the African American race was headed for extinction, justifying the continued exclusion of blacks from insurance coverage rather than their inclusion at higher premiums. See Frederick L. Hoffman, “Race Traits and Tendencies of the American Negro,” Publications of the American Economic Association, vol. 11 (1896). ↩
- See Jacob S. Hacker, The Great Risk Shift: The Assault on American Jobs, Families, Health Care and Retirement and How You Can Fight Back (Oxford University Press, 2006). ↩
- Insurers’ fixation on average values was deeply embedded in insurance practice: because actuaries can’t know the future, they have to form expectations, which can only be done on the basis of group averages. In other words, no insurer can accurately predict when a particular individual will die, only that some number of persons in a reasonably sized group will die over some period of time. See Greta R. Krippner and Daniel Hirschman, “Undoing Difference: Risk Classification and Gender Discrimination in Auto and Life Insurance,” unpublished manuscript, Sociology Department, University of Michigan. ↩
- Jonathan Levy, Freaks of Fortune: The Emerging World of Capitalism and Risk in America (Harvard University Press, 2012). ↩
- In fact, feminist organizations that, beginning in the 1970s, contested gender discrimination in pensions and insurance considered it illegitimate to rely on group averages to assess an individual’s risk of mortality, illness, or accident—a stance largely consistent with the civil rights tradition in American law. See Mary Heen, “Nondiscrimination in Insurance: The Next Chapter,” Georgia Law Review, vol. 49 (2014); Krippner and Hirschman, “Undoing Difference.” ↩
- Oscar Rogers, “Medical Selection and Substandard Business,” Proceedings of the Association of Life Insurance Medical Directors of America (1907). ↩
- Rogers, “Medical Selection and Substandard Business,” p. 83. ↩
- See Martin Lengwiler, “Double Standards: The History of Standardizing Humans in Modern Life Insurance,” in Standards and Their Stories: How Quantifying, Classifying, and Formalizing Practices Shape Everyday Life, edited by Martha Lampland and Susan Leigh Star (Cornell University Press, 2009); Greta R. Krippner, “From Contract to Status? The Persistence of Gender Discrimination in Insurance Markets,” unpublished manuscript, Sociology Department, University of Michigan. ↩
- Michael W. Kita, “The Rating of Substandard Lives,” in Medical Selection of Life Risks, 4th ed., edited by R. D. C. Brackenridge and W. John Elder (Palgrave Macmillan, 1998). ↩
- Shoshana Zuboff, “Big Other: Surveillance Capitalism and the Prospects of an Information Civilization,” Journal of Information Technology, vol. 30 (2015). ↩
- Émile Durkheim, The Division of Labor in Society, translated from the French by W. D. Halls (Free Press, 1984). ↩
- I am indebted to conversations with Daniel Hirschman for this point. ↩
- See Krippner and Hirschman, “The Person of the Category”; Moor and Lury, “Price and the Person.” ↩