The five states under study were selected because they had both expansive forfeiture laws and necessary data—specifically, data tying forfeitures under state law to specific agencies over the study period of 2005 to 2013.52 Helpfully, the states’ forfeiture laws are, and were during the study period, similar, albeit with some variation.
Arizona law makes civil forfeiture both procedurally easy and financially rewarding for law enforcement. Though somewhat improved in recent years, the standard of proof for forfeiting property during the study period was a preponderance of the evidence, meaning the government just had to prove property was more likely than not connected to criminal activity.53 This standard is far from the proof beyond a reasonable doubt required in criminal court. When an owner, as a third party to a seizure, wishes to make an innocent owner claim to retrieve her property, she bears the burden of proving her innocence of the alleged criminal activity giving rise to the forfeiture action.54 Arizona law also gives law enforcement a strong incentive to seize property, awarding agencies 100% of the funds generated through forfeiture.55
Like Arizona’s civil forfeiture laws, Hawaii’s are among the nation’s most permissive for law enforcement. To forfeit property, the government must tie property to a crime by the low standard of a preponderance of the evidence.56 Innocent owners also bear the burden of proving they had nothing to do with the alleged crime for which the government is pursuing forfeiture.57 And under Hawaii law, law enforcement receives 100% of forfeiture proceeds: 25% of funds generated through forfeiture go to police, 25% to prosecuting attorneys and 50% to the attorney general.58
Iowa’s civil forfeiture laws are similar to Arizona’s and Hawaii’s. Though it has, like Arizona’s, improved somewhat in recent years, Iowa’s standard of proof required to forfeit property was a preponderance of the evidence during the study period.59 A person bringing an innocent owner claim bears the burden of proving she had no knowledge of, or involvement in, the alleged illegal use of her property.60 Iowa law awards 100% of forfeiture proceeds to law enforcement.61
Michigan’s civil forfeiture laws are also highly permissive for law enforcement despite modest reforms to the standard of proof and the innocent owner burden in recent years. During the study period, the standard of proof to forfeit property was a preponderance of the evidence,62 and for seizures where drug activity was alleged—as is the case with most seizures—innocent third-party owners wishing to recover their property bore the burden of proving their innocence or ignorance of the activity. The government bore the burden in all other cases.63 Michigan law also gives law enforcement a large financial incentive to seize property, allowing agencies to keep and spend 100% of proceeds from drug-related forfeitures and 75% of proceeds from other forfeitures.64
Of the five states studied, Minnesota provides the best protections for property owners, though its laws still make forfeiture relatively easy and very rewarding. To forfeit property during the study period, the government had to show property was subject to forfeiture by clear and convincing evidence, a moderately high standard, though still lower than beyond a reasonable doubt. (The state’s standard has since improved somewhat.)65 In innocent owner cases, third-party owners must prove they had nothing to do with the alleged criminal activity involving their property.66 The innocent owner defense was not allowed at all in DWI cases until a 2017 reform.67 Minnesota law enforcement also enjoys a large stake in forfeiture, receiving 90% of proceeds in most cases.68
As shown in Table 1, state and local agencies in the five states took in nearly $442 million through forfeitures conducted under state law between 2005 and 2013. For two of the states, Iowa and Minnesota, I am not able to report the full amount forfeited because agencies did not always comply with state requests for data. Consequently, the absolute levels for those two states are understated; however, if underreporting is scattered randomly across the study period, the numbers do provide some sense of trends.
Agencies in the five states took in an additional $218 million through their participation in the U.S. Department of Justice’s equitable sharing program during the study period. Table 1 also reports these amounts. I obtained these data from DOJ rather than the states, and they are reasonably complete.
Year |
Arizona |
Hawaii |
Iowa |
|||
State Forfeiture |
Equitable Sharing |
State Forfeiture |
Equitable Sharing |
State Forfeiture |
Equitable Sharing |
|
2005 |
$14,046,980 |
$5,448,493 |
$896,121 |
$2,020,762 |
$1,157,848 |
$2,287,631 |
2006 |
$13,695,877 |
$7,430,036 |
$1,419,610 |
$2,927,245 |
$2,770,816 |
$1,400,976 |
2007 |
$20,521,147 |
$5,918,261 |
$1,316,772 |
$2,276,178 |
$1,302,789 |
$1,246,584 |
2008 |
$15,192,662 |
$6,413,722 |
$1,484,994 |
$2,053,249 |
$1,592,685 |
$3,417,109 |
2009 |
$23,574,929 |
$4,008,650 |
$1,316,772 |
$677,243 |
$978,990 |
$6,382,194 |
2010 |
$33,235,833 |
$11,702,027 |
$1,763,123 |
$648,346 |
$961,103 |
$4,341,782 |
2011 |
$29,687,058 |
$7,822,161 |
$661,619 |
$643,256 |
$4,084,918 |
$3,790,540 |
2012 |
$33,658,124 |
$3,574,173 |
$515,811 |
$656,094 |
$2,158,161 |
$1,650,927 |
2013 |
$32,908,019 |
$6,610,161 |
$868,376 |
$1,337,168 |
$1,778,570 |
$2,415,217 |
Total |
$216,520,629 |
$58,927,684 |
$10,243,198 |
$13,239,542 |
$16,785,879 |
$26,932,958 |
Mean |
$1,704,887 |
$199,755 |
$276,843 |
$171,942 |
$23,249 |
$80,638 |
Median |
$251,949 |
$50,451 |
$189,076 |
$74,979 |
$2,228 |
$14,715 |
Year |
Michigan |
Minnesota |
Five-State Total |
|||
State Forfeiture |
Equitable Sharing |
State Forfeiture |
Equitable Sharing |
State Forfeiture |
Equitable Sharing |
|
2005 |
$20,148,191 |
$13,475,757 |
$2,443,374 |
$1,444,222 |
$38,692,515 |
$24,676,865 |
2006 |
$18,134,425 |
$11,849,502 |
$2,584,035 |
$2,298,870 |
$38,604,763 |
$25,906,628 |
2007 |
$22,069,366 |
$7,165,004 |
$2,944,729 |
$2,367,527 |
$48,154,803 |
$18,973,554 |
2008 |
$19,028,702 |
$13,465,121 |
$2,327,403 |
$2,275,389 |
$39,626,446 |
$27,624,590 |
2009 |
$25,217,827 |
$9,387,650 |
$3,484,259 |
$2,851,812 |
$54,572,778 |
$23,307,548 |
2010 |
$16,616,912 |
$6,960,934 |
$2,145,718 |
$2,660,746 |
$54,722,689 |
$26,313,836 |
2011 |
$18,217,111 |
$13,562,944 |
$5,280,967 |
$1,617,612 |
$57,931,673 |
$27,436,512 |
2012 |
$13,700,789 |
$17,211,295 |
$5,226,920 |
$1,843,216 |
$55,259,806 |
$24,935,704 |
2013 |
$13,566,177 |
$6,828,370 |
$5,167,082 |
$1,895,658 |
$54,288,224 |
$19,086,573 |
Total |
$166,699,502 |
$99,906,576 |
$31,604,488 |
$19,255,052 |
$441,853,696 |
$218,261,811 |
Mean |
$75,669 |
$150,235 |
$19,630 |
$59,065 |
|
|
Median |
$5,562 |
$34,513 |
$4,769 |
$15,979 |
|
|
Note: Means and medians are of total forfeiture proceeds (or, in the case of federal equitable sharing, distributions) by agency, by year. These figures reflect non-zero proceed/distribution amounts only. Arizona’s and Hawaii’s means are much larger than those of the other states in part because Arizona and Hawaii have smaller numbers of agencies and because Arizona agencies in particular seem to engage in a lot of forfeiture activity.
It is important to note that the data include proceeds from both civil and criminal forfeitures. In general, state and federal data do not distinguish between the two, though civil greatly outpaces criminal in jurisdictions that keep track. For example, a recent study found 93% of Arizona’s forfeitures were processed civilly in 2018 and 2019; just 3% were processed criminally. (The proceeding type for the remaining 4% is unknown.)69 However, law enforcement’s financial incentive is generally the same for both civil and criminal forfeiture. Moreover, the decision of whether to pursue forfeiture through civil or criminal procedures is often made by prosecutors well after police have seized an asset. And the effects of forfeiture funds, if any, are likely to be similar whether funds derive from civil or criminal forfeitures.
Forfeiture amounts varied across agencies and states as well as over time. Indeed, state and local forfeiture amounts fluctuated considerably in all five states over the study period. Two states, Arizona and Minnesota, saw large increases in state forfeiture during the analysis period. Iowa and Michigan present a mixed picture, and Hawaii experienced a decline. Equitable sharing amounts also varied considerably, which is not uncommon since a few large joint task force operations can greatly affect the results in any given year.
My purpose here is to compare trends in forfeiture over time with those in crime clearances, illicit drug use and economic conditions. While it is tempting to do this using data aggregated at the state level, this is inadvisable. Most forfeitures are initiated not by state agencies but by county or municipal agencies. This means local agencies, and not state, are the relevant actors and decision-makers. Using data aggregated up to the state level could obscure relationships between forfeiture proceeds, police activity and other variables that are apparent only with disaggregated data. I have therefore developed time series data at the agency level across all five states. Called a “panel dataset,” this construction allows me to follow variables for each individual agency over multiple years. Besides associating my tests with the actual agencies engaging in forfeitures, this also gives my tests greater statistical power since hundreds of agencies are involved. In the next section, I describe my panel dataset in greater detail.
This study provides the first multistate analysis of effects and causes of state and local forfeiture. It focuses on the individual police agency, which may be a municipal police force, a county sheriff’s office or a state agency. For each police agency, I combined data for forfeiture, crime, number of officers and other variables of interest over a nine-year period, from 2005 through 2013. The result is a panel dataset that follows changes in outcomes, over the same period, for several hundred agencies across the five states. The panel approach offers important advantages over the “cross-sectional” approach taken in many previous studies, which compare agencies to one other at some single point in time.70 Many factors may affect forfeiture and crime across jurisdictions, and some of them may be difficult or impossible to quantify or even identify. In following individual police agencies over time, my approach effectively controls for many of those unobservable variables. Similarly, any number of factors may contribute to variations in drug use across the United States. Controlling for state and region reduces that statistical noise, allowing me to focus on the variables of interest and greatly increasing confidence in the results. Appendix B describes my regression methodology.
For each police agency. The data for forfeiture amounts by police agency were gathered from public authorities in the five states: Arizona, Hawaii, Iowa, Michigan and Minnesota. As noted above, these five states were selected for the study because they had both expansive forfeiture laws and necessary data—specifically, data tying forfeitures under state law to specific agencies over the study period of 2005 to 2013. A larger study encompassing more states would obviously be preferable, but many states still fail to publicly report consistent and reasonably complete forfeiture data, leaving researchers—and state policymakers—without the tools to determine whether forfeiture is achieving its stated goals.71
For the five states studied here, the Institute for Justice obtained annual data specific to agencies or, in Arizona’s case, counties. These reflect proceeds from forfeitures conducted under state law and do not include federal equitable sharing amounts.72 A number of agencies had no proceeds data for some years. In many cases, it was possible to interpret the missing proceeds as zeros.73 However, for many Iowa and Minnesota agencies, as well as for a much smaller number of Michigan agencies, with missing proceeds data for some years, it was not possible to determine whether, in those years, they received no forfeiture proceeds or simply failed to report on their forfeiture proceeds.74 This made it impossible to determine the year-over-year changes that are a crucial part of the panel structure for those agencies. Accordingly, I omitted those agencies from my balanced panel.75 Fortunately, the forfeiture data for Arizona and Hawaii were complete, those for Michigan were largely complete, and those for Iowa and Minnesota were complete for many agencies.
To create the necessary datasets for my regression analyses, I combined the forfeiture data with several highly detailed government datasets on policing, crime, economic factors and drug use. Specifically, I developed datasets at the police agency level that track, over several years, the number of sworn police officers, population served, and the number of serious crimes reported and cleared (i.e., considered solved) by arrest. I combined these datasets with geographically more aggregated data on demographic characteristics, economic conditions and illicit drug use. The resulting datasets contain annual data for all of these variables across hundreds of individual local and county police agencies.
I obtained the number of officers, by year and agency, from the Federal Bureau of Investigation’s Uniform Crime Reporting Program, both directly and through the Inter-university Consortium for Political and Social Research. I obtained crime data through the UCR’s Offenses Known and Clearances by Arrest database. I obtained illicit drug use data from successive tranches of the National Survey on Drug Use and Health. Conducted by the Substance Abuse and Mental Health Services Administration within the U.S. Department of Health and Human Services, NSDUH is widely considered the leading source for drug usage data. It provides information at the national, state and sub-state levels; I used data at the sub-state level for this study. I obtained annual economic data, usually on a county level, from several sources and price inflation data from the Bureau of Economic Analysis. Appendix A provides a list of my data sources.
The time frame for my analyses—the years 2005 through 2013—spanned the Great Recession, with its impact on economic factors and budgets. This time frame provided the variation necessary to test the impact of changes in local economic conditions. The data compilations within the datasets I used were also very consistent during the time frame, which is important in implementing a panel regression.
My analysis finds no material evidence that forfeiture increases police effectiveness or reduces illicit drug use. The relationships between forfeiture and these outcomes are weak and statistically insignificant, with one exception that links increased forfeiture to decreased clearance rates for violent crimes. In contrast, my analysis does find evidence that fiscal stress leads to increased reliance on forfeiture, providing support to critics’ claim that forfeiture promotes “policing for profit.” (Full regression results for all analyses can be found in Appendix B.)
Forfeiture proponents argue that increased forfeiture revenues allow police to pursue crime more vigorously. If true, this implies jurisdictions with increasing forfeiture revenues would experience increased police effectiveness, with more reported crimes cleared, or resolved, by arrest. Likewise, one would expect jurisdictions with falling forfeiture revenues to experience the opposite. Using crime and forfeiture data, I empirically tested whether these presumed connections actually exist and find that they do not. In fact, I find no meaningful relationship between forfeiture revenue and police effectiveness.
To measure police effectiveness, I used crime clearance rates for four sets of crime data: all reported crimes, “Index 1” (or simply “index”) crimes, violent crimes and property crimes. Clearance rate refers to the proportion of reported crimes police consider solved, usually by arrest. All reported crimes is a large category consisting of the serious violent crimes of murder, negligent (or involuntary) manslaughter, rape, robbery, aggravated assault and simple assault and the serious property crimes of burglary, larceny, motor vehicle theft and arson. Index 1 crimes refer to the eight serious violent and property crimes the FBI includes in its annual Crime in the United States reports.76 These overlap with all reported crimes but exclude involuntary manslaughter and simple assault.
These crimes, unlike certain other crimes like drug crimes, rarely lead to forfeiture. This means the promise of forfeiture revenue is unlikely to motivate police to clear them. Forfeiture proponents argue increased income from forfeiture, which arises largely from alleged illicit drug activity, allows police to be more effective generally, including when it comes to serious crimes.77 Not only are serious crime clearances an important measure of police effectiveness, but focusing on them, rather than on drug crime clearances, allows me to analyze whether forfeiture funds received—rather than the promise of further funds—affects police effectiveness.
A focus on drug crime clearances, on the other hand, would make the direction of causality difficult to determine. The particular availability of the forfeiture tool with respect to those crimes is intended precisely to encourage enforcement. Higher clearance rates for those crimes could therefore stem from police being more highly motivated to pursue those crimes by the promise of forfeiture revenue instead of, or in addition to, their being more effective due to having greater resources at their disposal. More practically, this hypothesis is not possible to test with data currently available: Drug crimes, even if reported to the police, are not part of the UCR’s Offenses Known reporting; while the number of arrests is reported, the level of arrests relative to illicit drug trafficking cannot be determined.
To measure the impact of forfeiture on an agency’s crime clearances, I built a panel of forfeiture proceeds per officer in a given agency lagged one year before the same agency’s clearance rate for all reported crimes. The lag is to allow for some delay in deploying forfeiture funds, which are usually of uncertain timing and amount, making it difficult to incorporate expected amounts into operations planning. I controlled for the number of sworn police officers per capita for the populations served as well as for the overall population. Finally, I included year indicator variables to reflect trends not otherwise captured.
My results do not support the contention that forfeiture leads to greater police effectiveness (see Table 2). Forfeiture’s effects on clearance rates for all reported crimes were not statistically significant, suggesting additional forfeiture revenue does not translate into more crimes solved. Further, the association between forfeiture proceeds and clearance rates, beyond being statistically insignificant, was tiny: Even if the effects were significant, they would amount to a very small marginal increase in crime clearances. Specifically, an increase in forfeiture revenue of $1,000 per officer from one year to the next implies an increase in the clearance rate of only 1 additional arrest per 1,000 crimes.
|
All Reported Crimes |
Index 1 Crimes |
Violent Crimes |
Property Crimes |
|||||||
State and Local Forfeiture Proceeds Only |
Not statistically significant |
Not statistically significant |
Strong statistically significant decrease**: A $1,000 increase |
Weak statistically significant increase*: A $1,000 increase in forfeiture proceeds per officer is associated with 2 more crimes solved. |
|||||||
State and Local Forfeiture and Equitable Sharing Proceeds |
Not statistically significant |
Not statistically significant |
Statistically significant decrease**: A $1,000 increase |
Not statistically significant |
**5% level, *10% level.
For detailed results, see Tables B2 and B3 in Appendix B.
As explained above, all reported crimes include the large category of simple assault and the smaller category of involuntary manslaughter, which are not index crimes. I therefore repeated the analysis with these two categories removed, finding similar results for index crimes clearances only. Again, the association between forfeiture proceeds and clearance rates is not statistically significant and, again, it is tiny, in this case suggesting an increase of about 2 clearances per 1,000 crimes for an $1,000 per officer increase in forfeiture revenue.
I also explored the effects of forfeiture proceeds on violent and property crime clearances separately. Efforts directed at violent and property crimes could potentially differ as police gain greater resources. To take an extreme example, murder clearance rates might be unresponsive to further resources because police already make considerable efforts to solve murders. But with greater resources, police might be able to pursue other less serious index crimes more vigorously. To test this, I split the index crime clearances into violent crimes and property crimes. The results for property crimes showed a weak statistically significant increase (10% level) in clearances for an increase in forfeiture proceeds. However, the effect’s magnitude is again very small—an increase of about 2 clearances per 1,000 crimes for an increase in forfeiture proceeds of $1,000 per officer.
These results suggest forfeiture does not materially improve police effectiveness. Indeed, it may make police less effective when it comes to solving violent crimes.
The results for violent crimes are more intriguing. There, I found a stronger statistically significant effect (5% level) from forfeiture revenue on crime clearances. However, the effect was the opposite of that predicted by proponents: A $1,000 increase in forfeiture proceeds is associated with a decrease in violent crime clearance rates of 7 per 1,000 incidents. It is possible increasing forfeiture revenues are tied to increasing application of police resources to drug crimes—with less effort correspondingly being put into resolving violent crimes.
The preceding analyses used proceeds from state and local forfeitures only. As described above, earlier work has considered the impact of federal equitable sharing payments nationwide on policing.78 But state and local proceeds and equitable sharing funds may have a combined effect not captured in separate analyses. To address this, I added the equitable sharing proceeds received by each of the agencies for each year to the amount of state and local proceeds. Using this to calculate forfeiture proceeds per officer, I proceeded as above.
The results for combined federal, state and local proceeds are similar to those for the state and local proceeds alone. Combined forfeiture proceeds have no detectable effect on clearances for either all reported crimes or index crimes alone. For property crimes, combined proceeds are positively correlated with clearances, but the relationship is small and not statistically significant. And with respect to violent crimes, clearance rates are again negatively associated with forfeiture proceeds: Greater proceeds imply lower violent crime clearance rates. The association is smaller than with state and local proceeds alone—with a $1,000 increase in forfeiture per officer implying a decrease in violent crime clearance rates of about 4 per 1,000 incidents—but the relationship is strongly statistically significant (5% level).
These results suggest forfeiture does not materially improve police effectiveness. Indeed, it may make police less effective when it comes to solving violent crimes. These findings provide support for critics’ claims that forfeiture is not a crucial crime-fighting tool. They also suggest forfeiture may distort law enforcement priorities by encouraging them to pursue crimes, such as drug crimes, that are more likely to lead to forfeiture, at the expense of crimes, such as violent crimes, that are less likely to do so.
State and local forfeiture activity is heavily oriented toward illicit drugs. Forfeiture proponents assert that removing the instruments of the drug trade hinders drug operations while removing the profits makes trafficking less attractive. They also claim allowing law enforcement to keep and spend forfeiture proceeds furthers the fight against drugs because police can use the funds to fight drug crime both directly through greater enforcement and indirectly through drug education and other anti-drug efforts. While separating these strands would be difficult, it is possible to cut through the complexities by asking a simple question: Does increased forfeiture revenue lead to decreased illicit drug use? After all, if proponents are right, there should be fewer drugs on the street—and less drug use. Looking at no fewer than four different drug use metrics, I find no association between forfeiture revenues and drug use.
To measure drug use, I turned to the National Survey on Drug Use and Health, the most reliable information available concerning drug addiction and drug abuse. Survey data were gathered consistently during my study period, allowing me to incorporate them into the panel structure. In particular, I tested whether increases in forfeiture revenue experienced by agencies within given NSDUH sub-regions were associated with reductions in drug use in those same sub-regions, controlling for factors that might also affect illicit drug use: the number of sworn police officers and demographic and economic factors sometimes linked to drug use. In all, I measured changes in four NSDUH variables: (1) use of any illicit drug in the previous year, (2) marijuana use in the previous year, (3) nonmedical use of prescription pain relievers in the previous year and (4) cocaine use in the previous year.
None of these four drug use measures showed any systematic association with forfeiture revenues, either for state and local forfeiture proceeds alone or for combined federal, state and local forfeiture proceeds. Correlations were small and never approached statistical significance (see Table 3).
|
Illicit Drug Use |
Marijuana Use |
Non-Medical Prescription |
Cocaine |
State and Local |
Not statistically significant |
Not statistically significant |
Not statistically significant |
Not statistically significant |
State and Local Forfeiture and Equitable Sharing Proceeds |
Not statistically significant |
Not statistically significant |
Not statistically significant |
Not statistically significant |
For detailed results, see Tables B4 and B5 in Appendix B.
For decades, forfeiture proponents have cited the goal of fighting the illicit drug trade as the primary purpose of forfeiture. And the ultimate goal of fighting the drug trade is to reduce illicit drug use. But the evidence analyzed here finds no link between forfeiture and drug use. No patterns emerge across the four drug use measures to suggest the four sets of results somehow understate forfeiture’s impact. The sample size is large, so even fairly modest effects would be picked up if they were widespread across the data. The data simply do not support proponents’ assertion that forfeiture furthers the policy goal of reducing drug use.
The most controversial aspect of forfeiture is that law enforcement often receives some or all of the proceeds when property is forfeited. State and local agencies can often spend forfeiture proceeds on a wide variety of purposes and with little oversight. In the case of federal equitable sharing proceeds, recipient agencies are required to use the funds for law enforcement purposes. Distributions are not supposed to replace appropriated agency resources; instead, they are to be used as a budget supplement. Forfeiture thus purportedly provides discretionary funds to agencies that might otherwise have little flexibility in their budgets. Among other criticisms of the financial incentive, opponents argue police will pursue forfeiture more assiduously during times of fiscal stress—not due to increases in crime that can lead to forfeiture but because of forfeiture’s increasing value as a budgetary supplement. That is, forfeiture arises not incidentally to normal policing but rather as a deliberate strategy to obtain funds, a strategy especially important when budgetary times are tough.
To test whether local economic conditions impact forfeiture activity, I applied two widely used surrogates for fiscal stress and health: the unemployment rate and personal income.79 Increased unemployment increases local fiscal stress due to loss of tax income as sales taxes fall as well as to increased demands upon municipal resources, such as responding to homelessness and public health concerns. Meanwhile, increases in personal income lead to improvements in local fiscal health through their impact on people’s purchasing power (sales taxes) and property values (property taxes). If critics are right that police forfeiture activity is responsive to changes in local economic conditions, increased unemployment should correlate with more forfeiture activity and increased personal income with less. For each agency, I included the annual unemployment and personal income levels from county data. As controls, I included the number of sworn police officers, demographic data, the number of reported offenses, the population served and year indicator variables.
My analysis finds that the unemployment rate has a strong impact on state and local forfeiture proceeds: A one percentage point increase in the unemployment rate, such as from 5% to 6%, is associated with a 12% increase in forfeiture proceeds. (See Figure 1.) This result is statistically significant at the 10% level. For personal income, on the other hand, I find no statistically significant effect. This may be because personal income changes far more slowly over time than does unemployment, making causal relationships, if any, harder to discern in the data. These results are consistent with my previous work demonstrating fiscal stress leads to increased equitable sharing activity.
For detailed results, see Table B6 in Appendix B.
I also tested the effects of fiscal stress on state and local forfeiture proceeds and federal equitable sharing proceeds combined. To do this, I first summed the forfeiture proceeds from any source by agency and year. I then analyzed the impact of unemployment and personal income on these proceeds, again finding unemployment has a positive and statistically significant effect at the 10% level. Here, a one percentage point increase in unemployment is associated with nearly an 11% increase in forfeiture proceeds. Interestingly, and contrary to what one might expect, I also find a very small but statistically significant effect from personal income in one of the tests: As incomes rose, so, too, did forfeiture activity—albeit very slightly, by one-tenth of a percent for a one percentage point increase in personal income (see Table B6 in Appendix B). It could be that higher personal incomes mean more valuable assets are available for forfeiture.
Two limitations of these results are worth noting, both stemming from the proxies I used for fiscal stress. First, in addition to being associated with fiscal stress, unemployment could also lead to higher crime rates, including greater illicit drug use, and thus more incidental forfeiture activity. However, although unemployment may have led to increased crime in some jurisdictions, crime rates did not increase nationwide during the Great Recession.80 Furthermore, I controlled for reported crime. A second limitation is unemployment and personal income statistics are county level rather than by agency, and fiscal stress may affect different agencies within the same county differently. Ideally, I would have used law enforcement agency budgets as a measure of fiscal stress, but I do not have access to police budgets over time for most of the agencies in my sample.
Nevertheless, my results for unemployment and forfeiture activity are consistent across my analyses as well as with those from my earlier research. This suggests forfeiture is not simply incidental to law enforcement, lending support to critics’ claims that forfeiture activity may be motivated by a desire for revenue.81
This study finds no material support for the propositions that forfeiture, either state and local alone or combined with federal equitable sharing, leads to greater policing effectiveness or reduces illicit drug use. It does, however, find that economic conditions affect forfeiture activity, with the relationship both materially important and statistically significant. These results are similar to those from earlier studies, in particular those from my 2019 study of equitable sharing alone,82 and they are especially salient now, when local government budgets are suffering due to the COVID-19 pandemic. The data suggest that it is during times like these that police may make particular recourse to forfeiture.
These findings that forfeiture is not meeting its policy goals would be of considerable concern even if forfeiture were harmless. But forfeiture is not harmless. It is a serious intrusion on civil liberties.
Property is often seized and forfeited based only on a police officer’s probable cause determination, as owners fail to contest seizures of their property because they are stymied by a confusing system, cannot afford legal representation or are compelled to sign away their right to their property to avoid possible criminal charges. Even when people do contest forfeiture, the system provides owners with poor protections that disadvantage them every step of the way.
And while forfeiture proponents claim forfeiture targets serious criminals, the size of many local forfeitures suggests ordinary people are often victimized. Where known, currency forfeitures in the states are typically just $1,300 or less. In a number of states, that figure is even lower, at only a few hundred dollars.83 And in 2017, nearly all of the vehicles swept up by the same Michigan forfeiture program that later claimed Stephanie Wilson’s cars were worth less than $1,000.84 Despite the low values, these properties may be anything but unimportant to their owners.
Forfeiture proponents have never mounted a serious empirical defense of the institution, neither in whole nor in its several parts. This author urges proponents to join the debate in a serious manner, with data rather than assertions. Clearly, at a minimum, better and more public record keeping is needed for policymakers and the public to be able to understand forfeiture’s benefits and its costs. This should be neither controversial nor difficult given that agencies must track the property in their custody in any event. Among other information, agencies should track whether forfeitures are processed according to civil or criminal procedures and whether they are tied to criminal charges or convictions.
Federal courts have upheld forfeiture because of the perception that the government’s interests in fighting crime outweigh the civil liberties infringements. And, indeed, many seized and forfeited properties likely are involved in or derived from crime. But in the absence of evidence that forfeiture works and given mounting evidence that it does not, the courts should reconsider whether the costs do not outweigh the purported benefits. Before taking people’s property, governments should have to prove owners are guilty of a crime. And when property is forfeited, governments should send the proceeds to a general fund. This would neutralize much of the criticism against forfeiture, and, as the results presented here illustrate, no public good is served by awarding forfeiture proceeds to the agencies that seize property.
Dataset |
Download Origin |
Uniform Crime Reporting Program Data: Offenses Known. Provided offenses known, offenses cleared by arrest and population served, by agency and year. |
FBI’s Uniform Crime Reports, accessed through Inter-university Consortium for Political and Social Research |
Uniform Crime Reporting Program Data: Police Employee Data. Provided the number of sworn officers, by agency and year. |
Criminal Justice Information Services Division |
Consolidated Assets Tracking System datasets, including DAG71_T, DISPOSAL_T, NCIC_CD_L; ASSET_T. Provided equitable sharing amounts and agency identification, by individual claim. |
Asset Forfeiture Management Staff, Department of Justice |
Covariate data: Unemployment rates, by county and year. |
Bureau of Labor Statistics |
Covariate data: Annual County Resident Population Estimates by Age, Sex, Race, and Hispanic Origin. Provided minority proportion in the population and age distribution, by county and year. |
Census Bureau, Population Division |
Covariate data: Personal income and expenditures, by county and year. |
U.S. Department of Commerce, Bureau of Economic Analysis, Regional Product Division |
Gross Domestic Product: Implicit Price Deflator, quarterly series, BEA Account Code: A191RD3. |
U.S. Department of Commerce, Bureau of Economic Analysis |
National Survey on Drug Use and Health, sub-state series. |
Substance Abuse and Mental Health Services Administration, U.S. Department of Health and Human Services |
Arizona: Racketeer Influenced Corrupt Organizations (RICO) Forfeiture Monies Reports: Forfeiture funds received, identified to agency and time period. |
Downloaded from the Arizona Criminal Justice Commission website |
Hawaii: Annual Reports to the Legislature of Proceedings Under the Hawaii Omnibus Criminal Forfeiture Act: Forfeiture funds received, identified to agency and time period. |
Downloaded from the Hawaii Department of the Attorney General website |
Iowa: State of Iowa Forfeiture Cases; Iowa: Q & A forfeiture database tables: Forfeiture funds received, identified to agency and time period. |
Downloaded from data.iowa.gov. Data provided to the website by the Iowa Department of Justice, Office of the Attorney General; Open Records Law request to the Iowa Department of Justice, Office of the Attorney General |
Michigan: Annual local government forfeiture reports pursuant to MCLS § 333.7524a (repealed): Forfeiture funds received, identified to agency and time period. |
FOIA requests to Michigan State Police |
Minnesota: Property Seized Subject to Forfeiture: Forfeiture funds received, identified to agency and time period. |
Minnesota Government Data Practices Act requests and downloads from the Office of the Minnesota State Auditor website |
All of the regressions used fixed effects panel methods with robust standard errors. All panels were balanced.
These tests explore whether forfeiture, as measured by forfeiture proceeds, has a measurable impact on the rate at which police clear, or solve, crimes by arresting someone. The regression treated the crime clearance rate as the dependent variable. The FBI’s Offenses Known data provide reported crime and crimes cleared by arrest for UCR codes 01 through 09 (see Table B1 for a listing of all UCR crime codes). Summing total crimes and clearances for these codes by agency and year allowed me to calculate the dependent variable CLEAR as 1,000 x (reported incidents cleared by arrest / reported incidents). Multiplication by 1,000, which results in a clearance rate per 1,000 incidents, conforms the measure to the standard reporting units. The mean rate was 265 clearances per 1,000 incidents in the panel data.
I conducted two sets of four regressions. The first set tests the impacts of state and local forfeiture proceeds alone, while the second tests the impacts of state and local forfeiture proceeds plus federal equitable sharing proceeds. The four regressions for each corresponded to all reported crimes in the UCR (codes 01 through 09), Index 1 crimes, Index 1 property crimes and Index 1 violent crimes.
My regressors (independent variables) were forfeiture proceeds per sworn officer (hereafter referred to as forfeiture per sworn officer or just forfeiture), sworn officers per population served, the natural logarithm of the population served, and year dummies for 2006 through 2013, with the year fixed effect measured against year 2005. Forfeiture per sworn officer averaged $1,275 annually. However, I denominated this variable in thousands of dollars to make the regression coefficients easier to interpret. I included quadratic terms for forfeiture to reflect likely diminishing marginal benefits of increased funds. I measured sworn officers per capita as 1,000 x number of sworn officers per population served. The natural log of the population served was just that. The timing was year t for forfeiture proceeds and year t+1 for the other variables. The specification took the form:
Clearances/thousand offenses = β0(1000 x Forfeiture/officerit) + β1(1000 x (Forfeiture/officer)2it) + β2(Number of officers/populationit) + β3(Log of population servedit) + β4(Year 2006 dummy) + β5(Year 2007 dummy) + β6(Year 2008 dummy) + β7(Year 2009 dummy) + β8(Year 2010 dummy) + β9(Year 2011 dummy) + β10(Year 2012 dummy) + β11(Year 2013 dummy) + εit,
Where i indicates the ith agency and t indicates the change in the level of the variable from period t-1 to t.
Table B2 provides the regression of forfeiture and other variables onto CLEAR in the columns headed All Reported Crimes. Forfeiture per officer has a coefficient of 1.113 on its linear term and ‐0.010 on its quadratic term. Neither coefficient is statistically significant. Their joint marginal effect is 1.113 – (0.010 x Forfeiture). To provide a sense of scale, consider the implied impact of forfeiture proceeds on clearances at the overall mean forfeiture of $1,275 per officer (1.275 in the units used in the regression). The cumulative effect at the mean of $1,275 per officer implied by the regression coefficients is just over one additional clearance per 1,000 incidents, against a mean of 265. Besides being insignificant statistically, the effect of forfeiture is very small in a practical sense.
I applied the same methodology to three other dependent variables: Index 1 crime, property crime, and violent crime. As usually defined, Index 1 crime does not include two of the 01 to 09 codes, those for negligent manslaughter (01B) and simple assault (08). This can matter because simple assault has a very large number of reported offenses. Removing these categories to create a variable CLEAR1, I obtained the output shown in the column headed Index 1 Crimes in Table B2. The forfeiture coefficient increases relative to all reported crime but remains statistically insignificant. A further breakout is between violent crime (columns headed Violent Crimes, which includes codes 01A, 02, 03 and 04) and property crime (columns headed Property Crimes, which includes codes 05, 06, 07 and 09). The effects of forfeiture on subsequent policing reach a weak level of statistical significance (at the 10% level) for property crime but remain small in practical terms. For violent crime, the coefficient was statistically significant at the 5% level but was negative, meaning that greater forfeiture was associated with a lower clearance rate. At -7.055, the coefficient is arguably material, but the quadratic gradually reduces it as the level of forfeiture increases.
Table B3 provides the results of a structurally identical regression that uses combined state and local forfeiture proceeds and federal equitable sharing proceeds for the forfeiture variable. For three of the four regressions, the forfeiture variable has no statistically significant impact on clearance rates. The exception is violent crimes, for which the coefficient is negative and statistically significant at the 5% level but only about half the level as for state and local forfeiture proceeds alone.
Finally, I note that statistical significance must be interpreted appropriately in this context. First, the databases include all agencies for which I had data for the panel period. They do not include agencies, which were mostly small, that failed to report for parts of the nine-year panel period. Also, of course, the results are for a collection of five out of 50 states. States were selected for the study because they had both expansive forfeiture laws and necessary data—specifically data tying forfeitures under state law to specific agencies over the study period of 2005 to 2013. And agencies within those states were selected based purely on data availability. Statistical significance is thus a measure of the reliability of the results if applied to states or agencies not included in the data, but better state reporting of forfeiture remains an important part of improving the accuracy of the analysis. Second, I have reported the results of several regressions. The odds of a false positive—of statistical significance indicated when none exists—are thus increased beyond the indicated power of the tests (1%, 5%, 10% levels). This should inform the reader’s interpretation of the statistical significance—at the 5% level—associated with the negative coefficient on violent crimes.
Note: The UCR Offenses Known (“Return A”) data report number of offenses for each of the “Part I” crimes (including code 01B) as well as for simple assaults (code 08), a non-Part I crime. The UCR Arrests by Age, Sex and Race data report arrest data for all crime codes. Codes other than Part I are referred to as “Part II.”
Variables |
All Reported Crimes |
Index 1 Crimes |
Violent Crimes |
Property Crimes |
||||
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
|
Forfeiture |
1.113 |
1.614 |
2.135 |
1.372 |
**-7.055 |
3.449 |
*2.312 |
1.250 |
Forfeiture2 |
-0.010 |
0.011 |
-0.015 |
0.010 |
**0.048 |
0.024 |
**-0.018 |
0.009 |
# of Officers |
-1.491 |
1.785 |
**-4.243 |
1.729 |
-2.727 |
3.027 |
***-4.071 |
1.358 |
Population |
-1.628 |
31.272 |
-4.755 |
19.458 |
-20.631 |
55.404 |
4.014 |
16.862 |
Year 2006 |
**8.373 |
3.685 |
***8.670 |
3.144 |
13.716 |
11.945 |
***9.034 |
2.858 |
Year 2007 |
***25.572 |
4.580 |
***25.995 |
3.891 |
**25.047 |
12.479 |
***25.804 |
3.694 |
Year 2008 |
***34.817 |
4.802 |
***31.342 |
4.568 |
***33.528 |
12.920 |
***30.084 |
4.306 |
Year 2009 |
***40.240 |
5.399 |
***35.676 |
4.750 |
***44.178 |
13.111 |
***32.640 |
4.637 |
Year 2010 |
***45.919 |
6.783 |
***40.331 |
5.630 |
***50.118 |
15.259 |
***39.838 |
5.746 |
Year 2011 |
***54.858 |
7.097 |
***42.947 |
6.198 |
***42.354 |
15.182 |
***42.272 |
5.927 |
Year 2012 |
***62.823 |
7.575 |
***57.432 |
6.526 |
***36.160 |
13.876 |
***53.464 |
6.345 |
Year 2013 |
***65.942 |
7.597 |
***58.097 |
6.845 |
**34.131 |
15.496 |
***51.929 |
6.629 |
R2 |
0.731 |
|
0.741 |
|
0.594 |
|
0.759 |
|
F Test |
21.29** |
|
21.64** |
|
12.38** |
|
23.94** |
|
*** p-value < 0.01, ** p-value < 0.05, * p-value< 0.10.
Definitions – dependent variables. Units are clearance rates per 1,000 reported crimes:
All Reported Crimes: Clearance rates for crime codes 01 through 09, including 01B and 08.
Index 1 Crimes: Clearance rates for crime codes 01 through 09, excluding 01B and 08.
Violent Crimes: Clearance rates for crime codes 01A, 02, 03 and 04.
Property Crimes: Clearance rates for crime codes 05, 06, 07 and 09.
Definitions – regressors, per agency basis:
Forfeiture: Forfeiture proceeds per sworn officer.
# of Officers: 1,000 x number of sworn officers per population served.
Population: Natural logarithm of the population served by the agency.
Year 2006 through Year 2013: Year fixed effects relative to year 2005.
Variables |
All Reported Crimes |
Index 1 Crimes |
Violent Crimes |
Property Crimes |
||||
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
|
Forfeiture |
-0.139 |
0.826 |
0.088 |
0.744 |
**-3.767 |
1.485 |
0.158 |
0.716 |
Forfeiture2 |
-0.001 |
0.006 |
-0.001 |
0.005 |
**0.026 |
0.010 |
-0.003 |
0.005 |
# of Officers |
-0.877 |
1.960 |
-3.096 |
2.391 |
-8.003 |
5.252 |
-2.649 |
2.284 |
Population |
15.265 |
29.513 |
11.860 |
21.046 |
-16.227 |
70.313 |
21.563 |
18.815 |
Year 2006 |
***10.759 |
3.370 |
***9.489 |
3.153 |
18.958 |
13.287 |
9.053 |
***3.162 |
Year 2007 |
***25.313 |
4.236 |
***24.909 |
3.875 |
*24.653 |
13.227 |
25.294 |
***3.694 |
Year 2008 |
***29.877 |
4.531 |
***26.165 |
4.592 |
**33.166 |
13.544 |
25.920 |
***4.342 |
Year 2009 |
***41.303 |
5.435 |
***33.816 |
4.920 |
***55.858 |
14.644 |
30.612 |
***4.917 |
Year 2010 |
***40.981 |
6.365 |
***34.889 |
5.618 |
***46.497 |
15.975 |
33.775 |
***5.665 |
Year 2011 |
***52.509 |
6.778 |
***39.697 |
6.038 |
***50.882 |
15.522 |
38.456 |
***5.844 |
Year 2012 |
***64.028 |
7.793 |
***55.842 |
7.158 |
**35.559 |
15.919 |
50.157 |
***6.920 |
Year 2013 |
***62.200 |
7.760 |
***55.110 |
6.912 |
*28.757 |
15.471 |
49.454 |
***6.808 |
R2 |
0.762 |
|
0.751 |
|
0.538 |
|
0.761 |
|
F Test |
25.10** |
|
22.89** |
|
9.44** |
|
24.42** |
|
*** p-value < 0.01, ** p-value < 0.05, * p-value < 0.10.
Definitions – dependent variables. Units are clearance rates per 1,000 reported crimes:
All Reported Crimes: Clearance rates for crime codes 01 through 09, including 01B and 08.
Index 1 Crimes: Clearance rates for crime codes 01 through 09, excluding 01B and 08.
Violent Crimes: Clearance rates for crime codes 01A, 02, 03 and 04.
Property Crimes: Clearance rates for crime codes 05, 06, 07 and 09.
Definitions – regressors, per agency basis:
Forfeiture: Forfeiture proceeds per sworn officer.
# of Officers: 1,000 x number of sworn officers per population.
Population: Natural logarithm of the population served by the agency.
Year 2006 through Year 2013: Year fixed effects relative to year 2005.
The purpose of these tests was to investigate whether forfeiture has a measurable impact on illicit drug use, as measured by the Substance Abuse and Mental Health Services Administration through the National Survey on Drug Use and Health. Structurally, I used annual data organized as a fixed effects panel, with the drug outcomes included as the three-year overlapping averages that NSDUH reports.
I treated four NSDUH outcomes as dependent variables in separate regressions: all illicit drug use in the past year, marijuana use in the past year, nonmedical use of prescription pain relievers in the past year and cocaine use in the past year. I used NSDUH data at the sub-state level, the most detailed level possible from the surveys. The data have the odd feature of being reported in overlapping three-year tranches: 2004–2006, 2006–2008, 2008–2010, 2010–2012 and 2012–2014. Starting with the 2014–2016 survey, methodologies changed. Consequently, I restricted the analysis to the five survey periods listed above, starting in 2004 and ending in 2014.
To place the other variables on the same basis as the NSDUH outcomes, I averaged them for three-year periods. I averaged forfeiture amounts for the three-year periods ending with the central year of the three-month NSDUH moving averages, creating an overlap with the first two years of those averages. I had two reasons for this. First, any effect on drug use from receipt of forfeiture proceeds would likely be delayed, so allowing a delay between the independent (forfeiture) and dependent (drug usage measures) variables makes sense. Second, there is a possible identification problem: Since many forfeitures result from drug arrests, one would expect increased drug use to be associated with increased forfeiture. Equitable sharing distributions lag property seizures by somewhat over a year on average, although state figures are hard to develop, so the overlap of seizures with the NSDUH periods is minimized by introducing the one-year lag from the average forfeiture amount to the average from the NSDUH surveys.
The NSDUH outcomes are published as proportions of the population; I multiplied these by one hundred, converting them to percentages of the population, to make interpretation of the coefficients easier. I used a log transform of the forfeiture amounts to reflect likely declining marginal product for forfeiture and to allow a more intuitive interpretation of the results. I used log transforms for the number of sworn officers and for the population served for the same reason.
I included three commonly asserted covariates for drug use: the unemployment rate, minority proportion of the population and percentage of the population age 15–24 years. The unemployment rate was used as published, as a percentage. The minority proportions are numbers such as 0.10. The occasionally large coefficients on the minority proportions and the percentages age 15–24 represent the impact of a hypothetical increase of 1 and so should be interpreted with caution; their main purpose in these regressions is as covariates. For each of the covariates, I calculated the average rates for the three years corresponding to the NSDUH years. I included year dummies for the last four of the five periods.
Table B4 provides the regression results for the four dependent variables for state and local forfeiture proceeds. The forfeiture coefficients are uniformly small and not statistically significant. For example, the coefficient for all illicit drug use in the previous year is -0.0008 when taken to four decimal places. This estimate suggests that a 1% increase in forfeiture proceeds is associated with a 0.0008 percentage point decrease in illicit drug use. The estimate is not statistically significant, being dwarfed by the standard error. This was true across the regressions: Forfeiture had no statistically significant impact on drug usage measures. Table B5 provides similar output using the sum of state and local forfeiture proceeds and federal equitable sharing proceeds; again, the coefficients on forfeiture are very small and not statistically significant.
Variable |
Illicit Drug Use |
Marijuana Use |
Nonmed Prescription |
Cocaine |
||||
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
|
Forfeiture |
-0.0008 |
0.0101 |
-0.0050 |
0.0126 |
0.0035 |
0.000045 |
-0.0007 |
0.0024 |
# of Officers |
*-0.5610 |
0.2980 |
-0.2830 |
0.3260 |
***0.3491 |
0.000927 |
***0.1443 |
0.0436 |
Population |
-0.3960 |
0.3020 |
0.0058 |
0.3290 |
0.1900 |
0.001290 |
-0.1040 |
0.0728 |
Unemployment |
***0.1496 |
0.0241 |
**0.0663 |
0.0260 |
***0.0514 |
0.000107 |
***0.0509 |
0.0049 |
Minority |
***-42.9040 |
4.5800 |
***-63.2900 |
5.7400 |
0.9797 |
0.015800 |
**-1.8820 |
0.7760 |
% 15–24 |
10.0801 |
10.9000 |
-4.1080 |
12.8900 |
***-23.5210 |
0.047200 |
***-8.7270 |
2.4200 |
Year 2007 |
***0.1761 |
0.0592 |
-0.1200 |
0.0739 |
***0.1773 |
0.000300 |
***-0.0690 |
0.0163 |
Year 2009 |
-0.0190 |
0.1510 |
***0.5032 |
0.1870 |
-0.1080 |
0.000705 |
***-0.7340 |
0.0357 |
Year 2011 |
***1.0202 |
0.1120 |
***1.5190 |
0.1350 |
***-0.4600 |
0.000495 |
***-1.0620 |
0.0249 |
Year 2013 |
***1.7051 |
0.1010 |
***2.6701 |
0.1230 |
***-1.0680 |
0.000450 |
***-1.0920 |
0.0227 |
R2 |
0.818 |
|
0.825 |
|
0.839 |
|
0.879 |
|
F Test |
7.94** |
|
9.85** |
|
8.37** |
|
9.07** |
|
*** p-value < 0.01, ** p-value < 0.05, * p-value < 0.10.
Definitions – dependent variables. Units are the proportion of the respondents who have engaged in the listed activity in the previous year:
All Illicit Drugs: Use of any illicit drug.
Marijuana: Use of marijuana.
Nonmed Use: Nonmedical use of prescription drugs.
Cocaine: Use of cocaine.
Definitions – regressors, per agency basis:
Forfeiture: Natural logarithm of forfeiture proceeds per sworn officer.
# of Officers: Natural logarithm of the number of sworn officers.
Population: Natural logarithm of the population served by the agency.
Unemployment: Unemployment rate.
Minority: Minority proportion in the population.
% 15–24: Percentage of population age 15–24.
Year 2007, Year 2009, Year 2011 and Year 2013: Year fixed effects relative to year 2005, where
the year is the middle of the three-year NSDUH rolling average period.
Variable |
Illicit Drug Use |
Marijuana Use |
Nonmed Prescription |
Cocaine |
||||
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
Coefficient |
S.E. |
|
Forfeiture |
0.0000 |
0.0163 |
-0.0390 |
0.0246 |
-0.0020 |
0.0046 |
-0.0010 |
0.0034 |
# of Officers |
0.6654 |
0.4430 |
0.2489 |
0.5220 |
0.0380 |
0.1280 |
0.1159 |
0.0750 |
Population |
-0.0060 |
0.2770 |
0.6959 |
0.4730 |
-0.0310 |
0.1120 |
-0.1030 |
0.1080 |
Unemployment |
-0.0400 |
0.0697 |
-0.0530 |
0.1060 |
-0.0001 |
0.0263 |
0.0099 |
0.0107 |
Minority |
***28.3758 |
7.3900 |
11.6029 |
10.7100 |
*4.4684 |
2.3400 |
***7.0517 |
1.5000 |
% 15–24 |
***-40.1220 |
11.3200 |
***-114.4190 |
20.8200 |
***-20.5660 |
3.8300 |
**-6.6070 |
3.1500 |
Year 2007 |
-0.1190 |
0.0891 |
**-0.3190 |
0.1450 |
***0.4345 |
0.0330 |
-0.0220 |
0.0241 |
Year 2009 |
***-0.8600 |
0.2790 |
-0.2900 |
0.4480 |
0.1306 |
0.1050 |
***-0.6330 |
0.0406 |
Year 2011 |
-0.1160 |
0.2140 |
-0.1050 |
0.3390 |
***-0.2240 |
0.0788 |
***-1.1100 |
0.0373 |
Year 2013 |
-0.0670 |
0.1560 |
**0.5089 |
0.2460 |
***-0.9520 |
0.0518 |
***-1.1670 |
0.0343 |
R2 |
0.716 |
0.629 |
0.864 |
0.944 |
||||
F Test |
6.58*** |
3.43*** |
11.07*** |
5.59*** |
*** p-value < 0.01, ** p-value < 0.05, * p-value < 0.10.
Definitions – dependent variables. Units are the proportion of the respondents who have engaged in the listed activity in the previous year:
All Illicit Drugs: Use of any illicit drug.
Marijuana: Use of marijuana.
Nonmed Use: Nonmedical use of prescription drugs.
Cocaine: Use of cocaine.
Definitions – regressors, per agency basis:
Forfeiture: Natural logarithm of forfeiture proceeds per sworn officer.
# of Officers: Natural logarithm of the number of sworn officers.
Population: Natural logarithm of the population served by the agency.
Unemployment: Unemployment rate.
Minority: Minority proportion in the population.
% 15–24: Percentage of population age 15–24.
Year 2007, Year 2009, Year 2011 and Year 2013: Year fixed effects relative to year 2005, where the year is the middle of the three-year NSDUH rolling average period.
These tests address whether increased financial stress on police agencies causes them to pursue forfeiture more actively. In contrast to the preceding regressions, here forfeiture is the dependent variable, and I use covariates of fiscal stress to determine whether such stress has a significan association with forfeiture. The structure was annual data for all variables.
As regressors, I included the number of sworn officers as directly influencing the amount of seized assets, the unemployment and personal income of each county as proxies for fiscal stress, the minority proportion of the population and proportion of the population age 15–24 as widely used correlates of police activity, the number of offenses reported as a measure of demands upon police, and the population served. My regressions were in logarithms on the forfeiture proceeds, number of sworn officers, number of offenses and population served. The model for one example took the form:
Log of forfeiture proceedsit = β0(Log of number of officersit) + β1(unemployment rateit) + β2(personal incomeit) + β3(Minority proportionit) + β4(proportion aged 15–24it) + β5(Log of number of offensesit) + β6(Log of population servedit) + β7(Year 2006 dummy) + β8(Year 2007 dummy) + β9(Year 2008 dummy) + β10(Year 2009 dummy) + β11(Year 2010 dummy) + β12(Year 2011 dummy) + β13(Year 2012 dummy) + β14(Year 2013 dummy) + εit,
Where i indicates the ith agency and t indicates the change in the level of the variable from period t-1 to t. The unemployment rate, personal income and minority proportions are measured at the county level, then applied to the agencies within the respective counties.
The dependent variable was either state and local forfeiture proceeds alone or state and local forfeiture proceeds plus federal equitable sharing proceeds. Results for both are provided in Table B6. The unemployment rate is a statistically significant and material predictor of forfeiture under either definition. The estimate of 0.119 in the first column, for example, implies that a one percentage point increase in the unemployment rate induces an 11.9% increase in state and local forfeiture proceeds. The relative unimportance of the number of sworn officers, personal income, minority proportion of the population and total population holds across all regressions, as does the importance of some of the year dummies. Proportion of the population age 15–24 has a statistically significant negative coefficient, indicating a negative relationship between changes in this proportion and changes in forfeiture. However, the level of the coefficient is very small.
All sets of the regressions showed reasonably strong goodness of fit, with R2 values ranging from 0.6 to 0.8. Additionally, the F test decisively rejected the null of no joint significance of the regressors in all cases.
|
State and Local Forfeiture Proceeds Only |
Combined State and Local Forfeiture and |
||
Coefficient |
S.E. |
Coefficient |
S.E. |
|
Unemployment |
*0.119 |
0.062 |
*0.109 |
0.062 |
Personal Income |
0.000 |
0.000 |
**0.001 |
0.000 |
# of Officers |
0.593 |
0.612 |
-0.097 |
0.416 |
# of Offenses |
-0.323 |
0.234 |
-0.236 |
0.205 |
Minority |
0.007 |
0.124 |
0.045 |
0.104 |
Proportion 15–24 |
***-0.907 |
0.334 |
***-0.940 |
0.336 |
Population |
1.439 |
0.909 |
1.439 |
1.233 |
Year 2006 |
**0.405 |
0.190 |
***0.491 |
0.187 |
Year 2007 |
**0.529 |
0.224 |
***0.536 |
0.207 |
Year 2008 |
-0.083 |
0.255 |
0.140 |
0.226 |
Year 2009 |
-0.650 |
0.455 |
-0.349 |
0.450 |
Year 2010 |
-0.564 |
0.439 |
-0.320 |
0.409 |
Year 2011 |
-0.221 |
0.378 |
-0.076 |
0.344 |
Year 2012 |
-0.402 |
0.329 |
-0.256 |
0.304 |
Year 2013 |
-0.208 |
0.326 |
-0.249 |
0.314 |
R2 |
0.783 |
|
0.774 |
|
F Test |
21.56** |
|
12.79** |
|
*** p-value < 0.01, ** p-value < 0.05, * p-value < 0.10.
Definitions – dependent variables:
Forfeiture Proceeds: Natural logarithm of the dollar value of forfeiture proceeds by agency.
Definitions – regressors, per agency basis:
Unemployment: Unemployment rate, in percentage points.
Personal Income: Per capita personal income.
# of Officers: Natural logarithm of the number of sworn officers per population.
# of Offenses: Natural logarithm of number of offenses reported to police.
Minority: Minority proportion in the population.
Proportion 15–24: Proportion of the population age 15–24.
Population: Natural logarithm of the population served by the agency. Year 2006 through Year 2013: Year fixed effects relative to year 2005.
1 First Amended Complaint at ¶¶ 105–148, Ingram, et al. v. Cnty. of Wayne, No. 2:20-cv-10288-AJT-EAS (E.D. Mich. May 11, 2020), ECF No. 12, https://ij.org/wp-content/uploads/2020/02/Amended-Complaint.pdf
2 Id.
3 Id.
4 Mich. Comp. Laws §§ 333.7524(1)(b)(ii), 600.4708(1)(f).
5 Arnold, T. (2019, Mar. 28). Wayne County doubling down on forfeiture as legislature moves to reform it. Michigan Capitol Confidential. https://www.michigancapitolconfidential.com/wayne-county-doubling-down-on-forfeiture-as-legislature-moves-to-reform-it
6 Knepper, L., McDonald, J., Sanchez, K., & Pohl, E. S. (2020). Policing for profit: The abuse of civil asset forfeiture (3rd ed.). Arlington, VA: Institute for Justice.
7 In a previous study, I tested the same claims using forfeiture data from the federal government’s equitable sharing program and a limited set of state and local data, but detailed data about forfeitures conducted under state law have been notoriously difficult to gather. Kelly, B. D. (2019). Fighting crime or raising revenue? Testing opposing views of forfeiture. Arlington, VA: Institute for Justice. https://ij.org/report/fighting-crime-or-raising-revenue/
8 For a primer on forfeiture, see Carpenter, D. M., Knepper, L., Erickson, A. C., & McDonald, J. (2015). Policing for profit: The abuse of civil asset forfeiture (2nd ed.). Arlington VA: Institute for Justice.
9 Knepper et al., 2020.
10 A recent study found that in the four states that track whether a claim was filed for return of seized property, claims are filed in 22% of cases or fewer. Knepper et al., 2020.
11 Legally, the presence of cash is not sufficient to establish probable cause. See, e.g., United States v. A) $58,920.00 in U.S. Currency, B) $38,670.00 in U.S. Currency, 385 F. Supp. 2d 144, 151 (D.P.R. 2005); United States v. $191,910.00 in U.S. Currency, 16 F.3d 1051, 1072 (9th Cir. 1994); United States v. $186,416.00 in U.S. Currency, 590 F.3d 942, 954–55 (9th Cir. 2010); United States v. Wilson, No. 14-cr-209-1, 2016 WL 11642732, at *7 (E.D. Pa. Oct. 27, 2016), aff’d, 960 F.3d 136 (3d Cir. 2020). However, in practice, police officers frequently take the presence of cash as probable cause, often citing a drug dog alert or the fact that money was bundled like drug money or hidden or claiming a person seemed nervous or evasive, had no good explanation for why they had the cash, or fit the profile of a drug courier. See, e.g., United States v. U.S. Currency, $30,060.00, 39 F.3d 1039, 1042 (9th Cir. 1994); United States v. $215,300 U.S. Currency, 882 F.2d 417, 419 (9th Cir. 1989); United States v. $60,020.00 U.S. Currency, 41 F. Supp. 3d 277, 286–89 (W.D.N.Y. 2011); United States v. Mathurin, 561 F.3d 170, 179 (3d Cir. 2009).
12 Knepper et al., 2020.
13 Boudreaux, D. J., & Pritchard, A. C. (1996). Civil forfeiture and the war on drugs: Lessons from economics and history. San Diego Law Review, 33, 79–135. See also, e.g., Leonard v. Texas, 137 S. Ct. 837 (Thomas, J., respecting denial of cert.), https://www.supremecourt.gov/opinions/16pdf/16-122_1b7d.pdf and Ellis, M. (2019, Feb. 10). From pirates to kingpins, the strange legal history of civil forfeiture. The Greenville News. https://www.greenvilleonline.com/in-depth/news/taken/2019/02/10/civil-forfeiture-history-pirate-privateers-organized-crime-drug-kingpins/2458836002/
14 See H.R. Rep. No. 106-192, at 2–3 (1999).
15 Racketeer Influenced and Corrupt Organizations Act of 1970, Pub. L. 91-452, 84 Stat. 922 (1970). See also, e.g., Pilon, R. (1994). Can American asset forfeiture law be justified? New York Law School Law Review, 39(1–2), 311–333.
16 Comprehensive Crime Control Act of 1984, Pub. L. No. 98-473, 98 Stat. 1976 (1984).
17 See Smith, D. B. (2018). Prosecution and defense of forfeiture cases (Vol. 1). New York, NY: Matthew Bender.
18 Comprehensive Crime Control Act of 1984, Pub. L. No. 98-473, 98 Stat. 1976 (1984). See also Smith, 2018.
19 Knepper et al., 2020.
20 Comprehensive Crime Control Act of 1984, Pub. L. No. 98-473, 98 Stat. 1976 (1984). See also Smith, 2018, and Carpenter et al., 2015.
21 U.S. Department of Justice Criminal Division Money Laundering and Asset Recovery Section. (2019). Asset forfeiture policy manual 2019. https://www.justice.gov/criminal-afmls/file/839521/download
22 U.S. Department of Justice & U.S. Department of the Treasury. (2018). Guide to equitable sharing for state, local, and tribal law enforcement agencies. https://www.justice.gov/criminal-afmls/file/794696/download
23 See, e.g., Pilon, 1994; Boudreaux and Pritchard, 1996; Williams, M. R., Holcomb, J. E., Kovandzic, T. V., & Bullock, S. (2010). Policing for Profit: The abuse of civil asset forfeiture. Arlington, VA: Institute for Justice. https://ij.org/report/policing-for-profit-first-edition/; Carpenter et al., 2015; Knepper et al., 2020.
24 See, e.g., Tuchman, G., & Wojtecki, K. (2009, May 5). Texas police shake down drivers, lawsuit claims. CNN.com. http://www.cnn.com/2009/CRIME/05/05/texas.police.seizures/; National Public Radio. (n.d.). Dirty money: Asset seizures and forfeitures [Series]. https://www.npr.org/series/91856663/dirty-money-asset-seizures-and-forfeitures; NewsChannel 5 Nashville. (2014, July 15). Timeline: Policing for profit. https://www.newschannel5.com/news/newschannel-5-investigates/policing-for-profit/timeline-policing-for-profit; Stillman, S. (2013, Aug. 12). Taken. The New Yorker.
https://www.newyorker.com/magazine/2013/08/12/taken; Sallah, M., O’Harrow, R., Jr., Rich, S., Silverman, G., Chow, E., & Mellnick, T. (2014, Sept. 6). Stop and seize. The Washington Post. https://www.washingtonpost.com/sf/investigative/2014/09/06/stop-and-seize/; The Greenville News. (n.d.). TAKEN investigation [Series]. https://www.greenvilleonline.com/news/taken/; St. Louis Public Radio. (n.d.). Taken: How police profit from seized property [Series]. https://apps.stlpublicradio.org/asset-forfeiture/
25 U.S. Department of Justice. (2016). Asset forfeiture policy manual, p. 21.
26 State of Hawaii Department of the Attorney General. (n.d., a). Asset forfeiture program. https://ag.hawaii.gov/afp/
27 146 Cong. Rec. 3657 (2000), https://www.govinfo.gov/content/pkg/CRECB-2000-pt3/pdf/CRECB-2000-pt3-issue-2000-03-27.pdf
28 Knepper et al., 2020.
29 Hearing on H.B. No. 748 Before the H. Comm. On Finance, 2019 Leg., 30th Sess. (Haw. 2019) (statement of the County of Kauai Office of the Prosecuting Attorney). See also, e.g., Cary, N., Lee, A., & Ellis, M. (2019, Feb. 3). SC cops defend keeping cash they seize: ‘What’s the incentive’ otherwise? The Greenville News. https://www.greenvilleonline.com/story/news/taken/2019/02/03/sc-civil-forfeiture-police-defend-practice-say-funds-essential-law-enforcement/2746412002/; Public Hearing on S.B. 521 Before the Wisconsin Senate Committee on Labor and Government Reform, 2015 Legis. (testimony of Ron Cramer, Eau Claire County Sheriff), https://docs.legis.wisconsin.gov/misc/lc/hearing_testimony_and_materials/2015/sb521/sb0521_2016_01_26.pdf; Wisconsin Department of Administration Division of Executive Budget and Finance. (2017). Fiscal estimate – 2017 Session, SB061. http://docs.legis.wisconsin.gov/2017/related/fe/sb61/sb61_DA.pdf
30 See, e.g., Rosenstein, R. (2017, Nov. 9). Bernie Madoff and the case for civil asset forfeiture. The Wall Street Journal. https://www.wsj.com/articles/bernie-madoff-and-the-case-for-civil-asset-forfeiture-1510237360
31The most important such decision was Caplin & Drysdale, Chartered v. United States, 491 U.S. 617 (1989), which weighed constitutional protections against asserted compelling government interests in helping fund law enforcement, compensate crime victims and attack organized crime. The Supreme Court concluded that the government interests outweighed the Sixth Amendment right to counsel and applied similar reasoning in United States v. Monsanto, 491 U.S. 600 (1989), decided the same day. In a more recent case, the Court stated, referring to Caplin and Monsanto: “On the single day the Court decided both those cases, it cast the die on this one too.” Kaley v. United States, 571 U.S. 320, 326 (2014).
32 State of Hawaii Department of the Attorney General. (n.d., b). History of asset forfeiture. https://ag.hawaii.gov/afp/history-of-asset-forfeiture/
33 U.S. Department of Justice and U.S. Department of the Treasury, 2018; U.S. Marshals Service. (n.d.). Asset forfeiture program. https://www.usmarshals.gov/assets/; Federal Bureau of Investigation. (n.d.). Asset forfeiture. https://www.fbi.gov/investigate/white-collar-crime/asset-forfeiture; Williams, M. R. (2002). Research note: Civil asset forfeiture: Where does the money go? Criminal Justice Review, 27(2), 321–329.
34 U.S. Department of Justice and U.S. Department of the Treasury, 2018.
35 Erickson, A. C., McDonald, J., & Menjou, M. (n.d.). Forfeiture transparency and accountability: State-by-state and federal report cards. Arlington, VA: Institute for Justice. https://ij.org/report/forfeiture-transparency-accountability/
36 See, e.g., U.S. Department of Justice Office of the Inspector General Evaluation and Inspections Division. (2017). Review of the Department’s oversight of cash seizure and forfeiture activities. https://oig.justice.gov/reports/2017/e1702.pdf and U.S. Department of Homeland Security Office of Inspector General. (2020). DHS inconsistently implemented administrative forfeiture authorities under CAFRA. https://www.oig.dhs.gov/sites/default/files/assets/2020-09/OIG-20-66-Aug20.pdf
37 Kelly, 2019. In earlier studies, I used the widely spaced Law Enforcement and Management and Administrative Statistics
survey to test the effects of LEMAS-reported forfeiture receipts, rather than equitable sharing proceeds, on crime-solving rates. There, I found police solved more crimes with increasing forfeiture amounts. The result was statistically significant; however, the magnitude of improvements was very small and got smaller and smaller at greater forfeiture amounts. Kelly, B. D. (2015). Further results concerning the effects of asset forfeiture on policing. SSRN. https://ssrn.com/abstract=2647629; Kelly, B. D., & Kole, M. (2016). The effects of asset forfeiture on policing: A panel approach. Economic Inquiry, 54(1), 558–576.
38 Knepper et al., 2020.
39 Knepper et al., 2020. Arizona’s median forfeiture is $1,000; Iowa’s is $900; Michigan’s is $423; and Minnesota’s is $607. Hawaii’s median forfeiture is unknown. The 21-state average is skewed by Florida’s $4,500 median, which is almost $2,000 higher than the next highest median. Florida’s median is likely explained by provisions requiring agencies to pay a $1,000 fee and post a $1,500 bond when filing a forfeiture action. If a forfeiture action is unsuccessful, the bond is payable to the property owner. These provisions make forfeitures under $2,500 riskier and less profitable for law enforcement. Knepper et al., 2020; 2016 Fla. Laws 179 (reform entitled “Contraband Forfeiture” approved by governor on April 1, 2016; original Florida Senate bill was CS/CS/SB 1044).
40 Knepper et al., 2020.
41 Knepper et al., 2020.
42 A recent study estimates hiring an attorney to contest a relatively simple state forfeiture case costs $3,000 at a minimum. The same study finds owners rarely file claims for return of seized property. It analyzed claim rates in the four states that track this information, including Arizona and Minnesota, finding owners contest forfeiture in 22% of cases or fewer, depending on the state. Knepper et al., 2020.
43 Brief of Law and Economics Scholars as Amici Curiae in Support of Petitioner, Pet. for Writ of Certiorari, Salgado v. United States, No. 19-659, https://www.supremecourt.gov/DocketPDF/19/19-659/125883/20191217164124310_Salgado%20Amicus%20Brief%20TO%20FILE.pdf
44 Ariz. Rev. Stat. § 13-4314(F); Fla. Stat. §932.704(10); N.Y. C.P.L.R. § 1318(4); Tenn. Code Ann. § 40-33-215; Utah Code Ann. § 24-4-110; Wis. Stat. § 961.555(7); Wyo. Stat. Ann §35-7-1049(p).
45 Bailes, M. S. (1994). Attorney fee shifting: The American Rule vs. the English Rule. Ohio Lawyer, 8, 16; Peter v. Nantkwest, Inc., 140 S. Ct. 365, 370 (2019) (“This Court’s basic point of reference when considering the award of attorney’s fees is the bedrock principle known as the ‘American Rule’: Each litigant pays his own attorney’s fees, win or lose, unless a statute or contract provides otherwise.” (internal quotations omitted)).
46 See ¶ 10.08 in Smith, 2018.
47 For example, under the Civil Forfeiture Reform Act of 2000, attorney fees may be paid from the general treasury. See 28 U.S.C. § 2465 (“[I]n any civil proceeding to forfeit property under any provision of Federal law in which the claimant substantially prevails, the United States shall be liable for—(A) reasonable attorney fees and other litigation costs reasonably incurred by the claimant.”). This is in contrast to the Equal Access to Justice Act, another statute under which attorney fees are available in forfeiture and other cases, which specifies that agencies must pay attorney fees out of their own budgets. See 28 U.S.C. § 2412(d)(4) (“Fees and other expenses awarded under this subsection to a party shall be paid by any agency over which the party prevails from any funds made available to the agency by appropriation or otherwise.”).
48 See, e.g., Williams et al., 2010; Carpenter et al., 2015; Knepper et al., 2020.
49 See, e.g., Stillman, 2013; Sallah et al., 2014; Lee, A., Cary, N., & Ellis, M. (2019, Jan. 17). Taken: How police departments make millions by seizing property. The Greenville News. https://www.greenvilleonline.com/in-depth/news/taken/2019/01/27/civil-forfeiture-south-carolina-police-property-seizures-taken-exclusive-investigation/2457838002/
50 Holcomb, J. E., Williams, M. R., Hicks, W.D., Kovandzic, T. V., & Meitl, M. B. (2018). Civil asset forfeiture laws and equitable sharing activity by the police. Criminology and Public Policy, 17(1), 101–127. See also Holcomb, J. E., Kovandzic, T. V., & Williams, M. R. (2011). Civil asset forfeiture, equitable sharing, and policing for profit in the United States. Journal of Criminal Justice, 39(3), 273–285; Preciado, M. P., & Wilson, B. J. (2017). The welfare effects of civil forfeiture. Review of Behavioral Economics, 4(2), 153–179; Wilson, B. J., & Preciado, M. (2014). Bad apples or bad laws? Testing the incentives of civil forfeiture. Arlington, VA: Institute for Justice. https://ij.org/report/%20bad-apples-or-bad-laws
51 Kelly, 2019. See also Kelly, 2015 and Kelly and Kole, 2016.
52 The five states included here received D or D- grades for their civil forfeiture laws in the second edition of Policing for Profit. Carpenter et al., 2015. Ideally, I would have included more states with low grades in this study. However, most lacked reliable revenue data specific to agencies for some or all of the study period. Louisiana and Virginia had limited data for the time period, preventing their inclusion, while Pennsylvania’s data for the time period were unclear.
53 Arizona has since raised its standard of proof to clear and convincing evidence, a moderately high standard, though still far short of the proof beyond a reasonable doubt required in criminal trials. Ariz. Rev. Stat. §§ 13-4311(M), -4312(H)(5)(a).
54 Ariz. Rev. Stat. §§ 13-4304(4)–(5), -4311(M), -4312(H)(5)(b).
55 Ariz. Rev. Stat. §§ 13-2314.01(D), -.03(D), 13-4315.
56 Haw. Rev. Stat. § 712A-12(8).
57 Haw. Rev. Stat. § 712A-12(8).
58 Haw. Rev. Stat. § 712A-16(2)–(4).
59 Iowa has since adopted a provision requiring a conviction, though not necessarily of the property owner, for forfeitures worth less than $5,000 when a claim is filed. After this provision is satisfied, prosecutors must show the property is subject to forfeiture by clear and convincing evidence. Iowa Code §§ 809A.1(4), 809A.12A(1), (1)(a), (1)(d), (8), 809A.13(7).
60 Iowa Code §§ 809A.12(7), .13(7).
61 Iowa Code § 809A.17.
62 Michigan now has a provision requiring the conviction of “a defendant”—not necessarily the property owner—for contested forfeitures of property worth less than $50,000. After securing a conviction, prosecutors must link property to a drug crime by clear and convincing evidence or to any other crime by a preponderance of the evidence. Mich. Comp. Laws §§ 333.7521a(1–2), a(6), (2), 600.4707(6).
63 In innocent owner claims, the government now bears the burden of proof in most cases. Mich. Comp. Laws §§ 333.7523a(2)(b) (burden on government in drug-related forfeitures), 600.4707(6)(b) (burden on government in other forfeitures); see id. §§ 333.7521a(6), .7523a(1) (procedures do not apply in drug-related forfeitures of property valued over $50,000); see also id. §§ 333.7521(1)(d)(ii), (f), 333.7531(1) (burden on owner in drug-related forfeitures under pre-reform procedure).
64 Mich. Comp. Laws §§ 333.7524(1)(b)(ii), 600.4708(1)(f).
65 Minnesota now has a provision requiring a conviction in some forfeiture cases. The provision does not apply to administrative forfeitures, only judicial ones. And for property worth up to $50,000, it applies only if an owner contests the administrative forfeiture of her property by asking for a judicial determination, forcing her to bear the burden of a costly legal battle to try to regain her seized property. The provision does not require the owner to be convicted—only a person—and it does not apply if the person has agreed to help investigators in order to avoid criminal charges. Once the provision is satisfied, prosecutors must link property to the crime by clear and convincing evidence. Minn. Stat. § 609.5311, subd. 2–3, § 609.531, subd. 6a(b), 6a(b)(2), 6a(d).
66 Minn. Stat. § 609.5311, subd. 3(d); Jacobson v. $55,900 in U.S. Currency, 728 N.W.2d 510, 519–20 & n.6 (Minn. 2007); Blanche v. 1995 Pontiac Grand Prix, 599 N.W.2d 161, 167 (Minn. 1999); see also Minn. Stat. §§ 609.5314, subd. 1(c), 169A.63, subds. 7(d), 9(e).
67 S.F. 151, 2017–18 Leg., 90th Sess. (Minn. 2017).
68 Minn. Stat. §§ 609.5315, subds. 5, 5a–5c, 169A.63, subd. 10(b).
69 Knepper et al., 2020. Similarly, 84% of DOJ forfeitures were civil between 2000 and 2019; the other 16% were criminal forfeitures. Knepper et al., 2020.
70 Examples of such studies are Worrall, J. L., & Kovandzic, T. V. (2008). Is policing for profit? Answers from asset forfeiture. Criminology and Public Policy, 7(2), 219–244; Holcomb et al., 2011; and Holcomb et al., 2018.
71 Erickson et al., n.d. See also McDonald, J. (2018). Civil forfeiture, crime fighting and safeguards for the innocent: An analysis of Department of Justice forfeiture data. Arlington, VA: Institute for Justice. https://ij.org/report/civil-forfeiture-crime-fighting-and-safeguards-for-the-innocent/
72 The Arizona data were labeled as state or local forfeiture monies only. IJ’s calculations of Michigan’s forfeitures by agency excluded federal equitable sharing proceeds. The custodians of Hawaii’s, Iowa’s and Minnesota’s records confirmed that those states’ data do not contain federal forfeitures. G. Senaga. (personal communication, October 23, 2019); J. Jernberg. (personal communication, Sept. 12, 2019); T. Ferguson (personal communication, Aug. 14, 2019).
73 For example, Michigan agency data were provided at the agency level and included proceeds as well as other information such as the number of forfeitures and the number of forfeitures still pending. Using this information, IJ was able to deduce that most instances where proceeds were left blank in reports were true zeros.
74 Iowa and Minnesota agency data were provided at the property level. If proceeds relating to a particular property were blank, it was impossible to know whether this was because there were no proceeds (e.g., because the property was retained for agency use, donated or destroyed or because the forfeiture was pending) or the agency failed to report them.
75 I also omitted a large number of mostly small Michigan agencies due to problems with their crime data. Specifically, I omitted agencies that reported incidents but no clearances in one or more years as it was impossible to tell whether the agencies had no clearances in those years or had clearances but failed to report them. Including these agencies would have created large jumps in percentage clearance rates, introducing statistical noise from what are likely poor data. The dropped agencies represent about half of Michigan’s agencies. However, these agencies are overwhelmingly small and thus represent a much smaller proportion of the population served.
76 Police agencies provide data to the FBI concerning the number of reported crimes and clearances for serious violent and property crimes. This information is collated and provided publicly through the Uniform Crime Reporting Program’s Offenses Known datasets.
77 See, e.g., Caplin & Drysdale, Chartered v. United States, 491 U.S. 617, 629 (1989). It should be noted that some states require forfeiture proceeds be spent on drug enforcement efforts. See, e.g., 42 Pa. Cons. Stat. § 5803(g), (i) and S.C. Code Ann. § 44-53-530(g). Money is, of course, fungible.
78 Kelly, 2019.
79 Francis, N. (2012). Revenue estimation. In R. D. Abel & J. E. Petersen (Eds.), The Oxford Handbook of State and Local Government Finance (pp. 497–514). New York, NY: Oxford University Press. For a recent estimate of the impact of changes in unemployment on state and local revenues, see Bivens, J., and Cooper, D. (2020, June 10). Without federal aid to state and local governments, 5.3 million workers will likely lose their jobs by the end of 2021 [Blog post]. https://www.epi.org/blog/without-federal-aid-to-state-and-local-governments-5-3-million-workers-will-likely-lose-their-jobs-by-the-end-of-2021-see-estimated-job-losses-by-state/
80 Uggen, C. (2012). Crime and the Great Recession (A Great Recession Brief). Stanford, CA: Stanford Center on Poverty and Inequality. https://inequality.stanford.edu/sites/default/files/Crime_fact_sheet.pdf
81 A separate question, beyond the scope of this study, would be to examine the negative effects of forfeiture on property owners in periods of fiscal stress versus periods of expansion. When unemployment is high, the greater need for funds by the agencies may coincide with a greater cost to losing assets for property owners.
82 Kelly, 2019.
83 Knepper et al., 2020.
84 Arnold, T. (2018, Oct. 27). Wayne County took cars from 380 people never charged with a crime. Michigan Capitol Confidential. https://www.michigancapitolconfidential.com/wayne-county-took-cars-from-380-people-never-charged-with-a-crime
Brian Kelly is associate professor of economics at Seattle University. He specializes in applied microeconomics, particularly the economics analysis of law, market power and antitrust, and the empirical analysis of the effects of trade policy regimes. Dr. Kelly’s interest in forfeiture grew out of his teaching and research in the economic analysis of law, particularly incentive effects created by different legal regimes. He has conducted extensive research with respect to forfeiture with an emphasis on quantification. His current projects include quantitative assessments of the civil liberties impacts of forfeiture and the success of forfeiture in meeting its stated goals. Dr. Kelly has consulted extensively in the areas of international trade barriers, mergers and acquisitions, and the market price effects of government policies. He has also provided expert testimony in cases involving international licensing, the exercise of market power and the distributive impacts of indirect taxes. Dr. Kelly received his doctorate in economics from Harvard University. He also holds a Master in Public Affairs from the School of Public and International Affairs at Princeton University and a bachelor’s degree from Stanford University.
Dr. Kelly is grateful for support provided by the Institute for Justice in the preparation of this report. He received indispensable help constructing the datasets from Anthony Ward and Kathy Sanchez. He also benefited greatly from the fine editing provided by Mindy Menjou and from comments provided by Lisa Knepper, Jennifer McDonald, Scott Bullock, Dana Berliner, Wesley Hottot, Paul Avelar, Keith Diggs, Lee McGrath and Robert Frommer. And he had many fruitful discussions with Dr. Dick Carpenter, and the report is organizationally and substantively much improved as a result.