U.S. v. Kordel, Parallel Proceedings, and the Value of Statistical Freedom

Introduction

Suppose that the Securities and Exchange Commission (SEC) charges a defendant with a financial violation. Suppose further that the Department of Justice (DOJ) is also currently investigating the matter for potential white-collar criminal prosecution. Often, in scenarios such as this one, without direct evidence of a future criminal proceeding, a district court judge will deny a defendant’s motion to stay the civil proceeding until after the disposition of a potential criminal proceeding surrounding the same facts. Forced to continue the civil proceeding, the defendant must respond to various informational requests, including interrogatories and subpoenas. However, if the defendant refuses to comply with the requests, a judge or jury may draw an adverse inference against them. Defendants who face potential parallel civil and criminal proceedings (hereafter “parallel proceedings”) tied to the same actions often face a difficult question: “Should I invoke my Fifth Amendment privilege against self-incrimination?” However, if they choose not to invoke their Fifth Amendment privilege, they will be barred from doing so should the criminal action get filed. This forces the defendant into a double-bind: they must choose between potentially incriminating themself and potentially losing or raising the costs of their civil case. Although some judges, acknowledging the impossibility of this double-bind, will not enforce an adverse inference finding, many will.

This fact pattern mirrors what happened in United States v. Kordel (1970). In Kordel, the Food and Drug Administration (FDA) commenced a civil action against Detroit Vital Foods, Inc. (DVF) for violations of the Food, Drug, and Cosmetic Act. As part of the civil litigation, the FDA submitted extensive interrogatories pursuant to Rule 33 of the Federal Rules of Civil Procedure. After receiving these interrogatories, DVF filed for a motion to stay the civil proceedings. It did so in fear that its employees would face future criminal prosecution and that DVF’s answers to the interrogatories would be used in said prosecution. The district court denied the motion on the basis that the defendants “did not conclusively indicate the Government would institute a criminal proceeding.” Faced with the prospect of virtually automatically losing the civil case if they refused to comply, the company’s president and vice president responded to the interrogatories. Later, however, DVF’s fears came to fruition. Months after answering the interrogatories, the president and vice president of DVF were indicted and later convicted using evidence from those interrogatories. The Supreme Court upheld this conviction, though enunciated standards for when parallel proceedings may be unconstitutional.

This Essay does not take a position on the ethics or constitutionality of parallel proceedings. Instead, it proposes using the dilemma defendants face in parallel proceedings as a way to measure the Value of Statistical Freedom (VSF). The VSF (sometimes called the Value of Liberty) can be thought of as an individual’s willingness to pay to not be in prison. In many ways, the VSF is spiritually similar to the far more prevalent “Value of Statistical Life” (VSL). The VSL, which measures the willingness to trade money or wealth in exchange for an increase in the mortality probability, is often used by policymakers and academics in cost-benefit analyses. However, unlike the VSL, the VSF has received very little research. Calculating a VSF number is difficult because a vast majority of defendants are represented by public defenders or assigned counsel. Additionally,  there exists some skepticism about whether incorporating the utility of prisoners when conducting cost-benefit analyses for incarceration policy should be done at all.

The unique trade-off provided by parallel proceedings, where a defendant trades off changes in the expected loss in a civil lawsuit for change in expected prison time, provides an opportunity to measure the VSF. In its purest form, this trade-off approximates the defendant’s willingness to pay to not go to prison. Furthermore, as detailed later in this Essay, this setting is insulated from many of the flaws innate to the VSL and VSF literatures. For example, defendants in parallel proceedings are often not credit-constrained.

The rest of this Essay is structured as follows: Part I provides an overview of the VSL literature. Part II then details the current state of the VSF research, identifying similarities with the VSL and detailing the common critiques of the current measurements. Part III then delves into parallel proceedings, explains how they can be leveraged to analyze the VSF, and illustrates the unique advantages of the setting.

I. The Value of Statistical Life

Since Professors Richard Thaler and Sherwin Rosen’s seminal paper, The Value of Saving a Life: Evidence from the Labor Market, the Value of Statistical Life has become an invaluable tool in the policymaker’s toolbelt. Although putting a monetary value on a human life may seem crude, it can help policymakers determine whether a public safety project is worthwhile. For example, the VSL can help answer questions such as “should a community build a traffic light in the middle of an intersection?” or “should the government institute an environmental policy reducing pollution, even if it costs companies billions of dollars?”

To measure the VSL, researchers look for scenarios where individuals trade off wealth for a higher or lower mortality rate. For example, in Professors Kyle Greenberg, Michael Greenstone, Stephen P. Ryan, and Michael Yankovich’s paper, The (Very) Heterogeneous Value of a Statistical Life: Evidence from U.S. Army Decisions, the authors analyze army soldiers’ reenlistment decisions. In the reenlistment process, soldiers choose between jobs of varying mortality rates, where deadlier jobs are often associated with higher lump-sum cash bonuses provided by the government. Similarly, in Market Versus Policy Responses to Novel Occupational Risks,  Professors Robert J. Cramer, Elissa Phillip Gentry, and W. Kip Viscusi look at how market wages responded to heightened mortality rates during the beginning of the COVID-19 pandemic. In both papers, the authors examined the amount an employer needed to pay an employee to compensate the employee for a heightened mortality risk. Once these numbers are identified, calculating the VSL is simple: divide the increase in the payment from before and after the mortality rate change by the change in mortality rate.

However, measurements of VSL come with various critiques. Two of the most common critiques are that 1) these studies do not tend to capture the VSL for all individuals but instead capture the VSL for only nonrepresentative groups and 2) many of the calculations can be biased by behavioral or economic constraints. Both can lead to underestimates of the true VSL. Taking these criticisms in turn, criticism 1 points out that these studies often focus on low-earning individuals. If VSL is heterogeneous and correlated with income or wealth, these studies underestimate the actual value of a life by selectively studying low-earning communities. Of course, this also leads to the existential question of whether policymakers should use a number that inherently values wealthy individuals’ lives at a higher value. A similar but distinct critique, criticism 2, notes that these equilibrium numbers only capture those who opt into the decision and those with the economic capacity to make these trade-off decisions. If individuals have a maximum they can pay to offset health risks or if they value their lives so highly that they are not found in the data at all (a selection effect), VSL estimates are deflated. These criticisms emphasize the need to scrutinize how these numbers are calculated and leveraged for policy decisions. Furthermore, they provide similarly necessary insights into the Value of Statistical Freedom. 

II.  The Value of Statistical Freedom

The success of the Value of Statistical Life literature has created a similar appetite for parallel numbers in other contexts. Notably, recent academic work has begun to port the methodology to the incarceration context. The novelty of this application is illustrated by the fact that the literature has yet to settle on a precise term: some scholars call it the “Value of Liberty,” others call it the “Value of Freedom.” To capture its similarity to VSL, however, I will call it the “Value of Statistical Freedom.” One explanation for the slow adoption of the methodology is that there are fewer opportunities for analyzing the VSF, as it is rare that an individual can choose to pay money to avoid prison. However, clever scholars have found ways to attack the proposition over the last decade.

The first measurement of the VSF came in 2011, many decades after Thaler & Rosen’s seminal work. In their paper, Optimal Bail and the Value of Freedom: Evidence from the Philadelphia Bail Experiment, Professors David Abrams and Chris Rohlfs derived a VSF of about $1,050 per 90 days, about $4,258.33 per year in 2003 dollars, using arrested individuals’ willingness to pay cash bail. This enterprising study, however, has been criticized for having similar flaws to the VSL literature. Many incarcerated individuals are credit-constrained and cannot access the money necessary to pay their bail, even if they would prefer to do so. Relatedly, jailed individuals are disproportionately poor, meaning that average measurements are likely underestimations of the VSF. In light of these criticisms, researchers Megan Stevenson and Sandra Mayson use a survey-methodology strategy in their paper, Pretrial Detention and the Value of Liberty, to illustrate incarceration’s unjustifiably cost. Notably, however, they abstract away from a dollar figure due to problems translating to monetary costs.

Although considerable in-roads have been made in achieving a VSF number, the current research has yet to provide a real-world measurement insulated from the prominent biases that lead to underestimation. Specifically, there is no real-world study that analyzes a trade-off between wealth and prison time for non-credit-constrained or disproportionately indigent individuals.

III.  Getting the Value of Statical Freedom From Parallel Proceedings

By allowing the government to force parallel proceeding defendants to make the trade-off between expected civil loss and expected jail time, Kordel opened the door for a unique setting to analyze the Value of Statistical Freedom. Suppose a defendant is denied a stay in a civil proceeding, and future criminal litigation that would result in incarceration is possible. In that case, they are forced to make a difficult decision: Do they participate in a civil proceeding and answer the interrogatories, subpoenas, and other investigatory actions? Or do they plead the Fifth and lessen their expected jail time? Notably, asserting the Fifth protects an individual in a criminal proceeding but allows a judge or jury to make an adverse inference finding in the civil proceedings. Furthermore, when a civil defendant does not cooperate with a governmental agency, the government is often instructed to pursue harsher penalties, upping their dollar exposure.

Needless to say, a defendant making this choice finds themselves in a double bind: they must choose between potentially incriminating themself and potentially losing or exacerbating their civil case. Translating this to economics, we can see that this trade-off is equivalent to trading off expected prison time with expected payment in a civil case. Here, we can analogize these numbers to those in the VSL context; expected prison time is analogous to the mortality rate, and expected payment in the civil case is the equilibrium payment (the compensating risk differential). The VSF then can be measured as equilibrium payment divided by the expected change in prison sentence.

Conclusion

This setting has the advantage of explicitly correcting for the weaknesses of previous studies. It involves real-world high-profile cases, where the defendants are often highly sophisticated individuals with significant resources. This allows scholars to capture a different group of criminal defendants, analyzing a VSF number when defendants are wealthy and not credit-constrained. Additionally, the resources available to these defendants likely allow them to more precisely estimate the expected prison time and civil damages payments, helping researchers get around any behavioral mechanism that may poison the estimate.

Note that this proposed methodology has its own drawbacks as well. Unlike a survey or the bail context, the universe of potential observations is likely far smaller. Furthermore, if the set of defendants making up the sample is disproportionately wealthy, one applying the method may instead calculate an overestimation of the VSF. The most significant advantage of this methodology is that it complements the existing research; whereas previous research may underestimate, this scenario captures the other side of the spectrum.

Ultimately, parallel proceedings raise interesting constitutional questions for defendants. While this line of inquiry is undoubtedly important, the current state of the law allows for a unique opportunity to measure the Value of Statistical Freedom. With decades of parallel proceedings, where defendants have been forced into choosing between potentially self-incriminating and potentially exacerbating their civil case, Kordel has provided future researchers the opportunity to provide empirical insight to the VSF and measure a number that can help lawmakers tailor incarceration policy for years to come.

* * *

George Vojta is a J.D. Candidate at the University of Chicago Law School, Class of 2025, and a Ph.D. Candidate in Economics at the University of Chicago.