SOCIAL TRUST IN CRIMINAL JUSTICE: A METRIC.

• Author: Kleinfeld, Joshua

INTRODUCTION 816 I. THE PROBLEM 828 A. A Critique of Existing Metrics 828 1. Crime Rates 828 2. Conviction, Arrest, and Clearance Rates 832 3. Recidivism Rates 835 4. Cost 837 B. A Cacophony of Goals 839 II. SOCIAL TRUST 843 A. A Central Concept in All of Social Science 843 B. Trust and the Nature of Crime and Punishment 848 C. An Overlapping Consensus 850 1. Communitarian Coals 850 2. Crime Control Goals 853 3. Retributive Goals 861 4. Racial Justice Goals 864 5. Liberal Goals 869 D. Incentives 871 E. Objections 875 1. Bad Social Norms 875 2. Popularity Contests and Noble Lies 878 CONCLUSION: TOWARD AN INSTRUMKNT 881 INTRODUCTION

What is the metric by which to measure a well-functioning criminal justice system? If a modern state is going to measure performance by counting something--and a modern state will always count something--what, in the criminal justice context, should it count?

Governments at present commonly focus on crime rates, recidivism rates, conviction rates, arrest rates, clearance rates, and cost. As we argue below, those metrics, while important and illuminating for certain purposes, are misleading in terms of what they are commonly taken to show and pernicious in terms of the incentives they create. Conviction rates, for example, are typically used to assess prosecutors' performance: they measure the proportion of people found guilty once prosecutors have decided to file charges. The effect is to incentivize prosecutors to focus on easy-to-prove cases rather than socially serious cases and, since the most certain conviction is a plea deal, to encourage excessive and sometimes unsavory forms of plea bargaining. (1) Another example: crime rates are commonly used to evaluate police forces, (2) yet crime rates are affected by such a complex array of societal factors beyond policing (broken families, poverty, poor mental health services, etc.) that it is perfectly possible for a community to have high crime notwithstanding good policing or low crime notwithstanding bad policing. Crime rate metrics can also incentivize harshness, since the intuitive way to prevent crime, at least in the short term, is to give offenders long sentences.

We are not the first to notice this problem, and in response to it a scholarly literature has formed, largely in empirical journals, around finding something better. Yet the question of what to measure in criminal justice is fundamentally a philosophical problem. It turns on what constitutes the excellence of a criminal system, even what constitutes justice in a criminal system--matters native to the theory of punishment. And therein lies the rub. The empiricists typically do not engage with the philosophical problem, and so construct metrics that nip around the edges of considerations that should matter. And the philosophers are commonly uninterested in problems of measurement, and so ignore the bureaucratic conditions of modern life--the fixation with quantitative metrics and, consequendy, the power of metrics to structure incentives, determine the allocation of resources, and redefine what governments care about or even, in a sense, "see." (3) Given these conditions, which prevail with particular force in the institutional and bureaucratic landscape of modern criminal systems, a good metric might have more impact on justice and even, potentially, more interest as a matter of first principles, than any statement of values abstractly understood. The real goal is to translate values into bureaucratic operating procedures; the "killer app" of modern life is the bureaucratic operationalization of values. But the philosophers are on the whole not communicating with the empiricists about what to measure.

The goal of this Article is to propose a certain concept of social trust and offer an argument as to why social trust, so understood, should be the north star in measuring the success or failure of criminal justice institutions. Social trust, as we discuss in Section II.A below, is anything but a novel concept. It refers broadly to an individual's beliefs about the general trustworthiness of other people, social institutions, and government: on a large scale, it refers to ambient levels of trust within a society toward other citizens, toward institutions, and toward government generally. It has been extensively studied in economics, political science, sociology, and psychology, and it has proven to be one of the best predictors of overall societal well-being known to social science, with remarkable and well-documented effects on economic success, effective government, levels of participation in civil society, and compliance with the law. Social trust is of particular interest now, in our politically polarized era, as it seems to be the very thing partisan polarization destroys. Others in the fledgling literature on criminal justice metrics have started to explore it or concepts related to it. (4) And yet those efforts have been hampered by an inadequate understanding of why social trust matters in criminal justice, and thus how best to conceptualize it and approach its measurement. Our goal in this Article is therefore to justify social trust as the lodestar of criminal justice metrics, and, in justifying it, to clarify the concept in ways that could contribute to future empirical work.

As we use the term in the context of criminal justice, social trust refers to (1) the level of trust a polity's members have toward the institutions, officials, laws, and actions that comprise the criminal justice system; (2) the level of trust a polity's members have, in virtue of the criminal system's operations, toward government generally (beyond the criminal justice system); and (3) the level of trust a polity's members have toward one another following incidents of crime and responses to crime. The goal of measurement in criminal justice, we submit, is to construct a tool by which to measure people's responses to the operation of the criminal system along these three dimensions. For example--and it is just an example--someone who is a crime victim, witness, community stakeholder, or even offender might get a survey in the wake of a contact with the criminal system asking whether the contact increased or decreased her sense of trust in the criminal system, in government generally, and in other people. Our central claim is that a well-functioning criminal system is one whose operation increases people's sense of trust along those three lines; a poorly functioning criminal justice system is one whose operation diminishes trust along one or more of those three lines. Marginal effects on social trust are the key measure of whether a criminal system is functioning well or poorly. Thus, in criminal justice, the output that is most important to measure and incentivize is something affective and interpersonal, not material.

These claims are not as absolute as they might sound, for two reasons discussed at length below. First, we are not arguing that other metrics should be done away with; the way social trust should "fit" into other metrics is more nuanced than that. Crime control, for example, is indispensable to measure; it just turns out that crime control depends on social trust to such an extent that, if one wants to minimize crime, there is good reason to measure trust. Second, we are not arguing for pure maximization of social trust. Pure maximization runs into problems of both the distribution of trust (for example, it can be dysfunctional to increase average trust if doing so creates radical distrust within a minority group) and non-trust-based constraints of justice (for example, scapegoating and show trials would violate basic principles of justice even if they were secret enough to increase trust in the short term). We will later argue that the distribution of trust matters, especially in conditions of an alienated minority, and that principles of basic justice constrain the maximization of trust. Those caveats in place, however, we do claim that social trust is sufficiently central to understanding what makes a criminal justice system function well that a metric based on social trust should take pride of place among all metrics.

A criminal justice metric based on social trust would have nine virtues, which we preview here and defend below, and which set it apart from any other available metric.

First, a metric based on social trust would radically alter criminal justice officials' incentives. Rather than encouraging excessively harsh policing and punishment practices in order to increase arrest or conviction rates, or excessively mild policing and punishment practices in order to cover up rising crime or curry favor with interest groups, criminal justice officials would have reason to handle crime and punishment in ways that victims, community members, and even offenders themselves find sensible given local community standards. Police officers, prosecutors, prison wardens, and others would be incentivized to care whether societal stakeholders would view their decisions as just, effective, and fair. The result, we predict, would be a criminal system that substantially reflects local views of what it means to be moral and effective in responding to crime. In short, a metric based on social trust would align official incentives with community norms.

Whether that is a good or bad thing depends to some extent on what one thinks of the local community's norms. But it does not entirely turn on that question. For one thing, the question isn't just the merits and demerits of community norms but what the alternative would be if community norms do not prevail (e.g., bureaucratic imperatives or special interest group ideologies). For another, reflecting community norms has positive downstream effects even in conditions of flawed norms. It overcomes in a single gesture the current array of bureaucratic imperatives, like certainty, punitiveness, and cost savings, that...