Incapacitation

The empirical literature on incapacitation is of recent vintage, going back to the early 1970s. Although arguably the most important rationale for imprisonment these days, incapacitation continues to be the least studied of the prison's functions. For the period 1990–1999, the Criminal Justice Abstracts lists only 85 incapacitation publications, compared to 509 for deterrence and 639 for rehabilitation. This gap in publications was substantially more pronounced in the 1980s, for example one incapacitation article for every fourteen articles on deterrence (Zimring and Hawkins).

The works by Cohen (1977, 1983), Zimring and Hawkins, Nagin, and Spelman (2000b) provide an excellent basis to track the evolution of the incapacitation literature over the past twenty-five years. Together they also give us a good portrait of what we know in the domain of incapacitation and, particularly, how well we know it. One important observation is that from the very beginning, scholars of high standing in the criminal justice academic community have disagreed on the basic point of whether the crime suppression effects of prison-incapacitation are large or small.

The two most important 1970s articles on incapacitation, published in the same issue of the Law and Society Review in 1975, illustrate the point. As reviewed by Cohen (1977), the first of these articles, written by David Greenberg, concluded that the crimes prevented by incapacitation amounted to no more than 8 percent of the total crimes actually committed, and perhaps as little as 1.2 percent. The second article was authored by a father-son team of noncriminologists, the Shinnars, and was one in a series of three articles essentially presenting the same model, which ultimately would have a lasting influence on the incapacitation literature. The Shinnars concluded that in the 1970s crimes prevented through incapacitation amounted to 25 percent of crimes committed, three times higher than the upper-bound effect suggested by Greenberg. Moreover, they argued that just as recently as the 1960s, when the risk of incarceration per crime committed was substantially higher, the number of crimes prevented through incapacitation amounted to 120 percent of crimes committed.

Despite substantial progress in methods and theory, disparities in the estimates of incapacitation put together by the best in the discipline continue to be huge, leading Nagin to conclude that "The evidence [about incapacitation] is of limited value in formulating policy. . . . [Predicting] the timing, duration, and magnitude of the impact of incremental adjustments in enforcement penalties remains largely beyond our reach" (Nagin, p. 367).

To understand why measuring incapacitation is such a difficult enterprise one needs to remember that the essence of this exercise is to count crimes that did not occur. Unavoidably, one needs to engage virtual reality and ask questions of the following two basic types: How many crimes would the 162,000 inmates incarcerated in California during the year 2000 have committed had they been free? Or, between 1996 and 1997 California increased its prison population by 9,200 inmates: What difference did that make, if any, in the actual crimes committed during 1997?

In answering these questions empirical evidence readily available to scholars consists of the crime rate, as reported to the police, and the actual imprisonment rate. At first sight, these two variables should provide all we need to assess incapacitation, for intuitively, the more people imprisoned, the larger the number of crimes prevented, and, all other things being equal, the smaller the crime rate.

Unfortunately, this logic can be seriously flawed for various reasons. To begin with, incapacitation and deterrent effects are commingled, a fact recognized since the first works on incapacitation (Cohen, 1977) and, to this day we lack a proven methodology to separate the two. The two effects are commingled because they both respond in the same direction to increases in imprisonment levels. As the prison population increases more offenders are incapacitated but also other would-be offenders may be deterred from committing additional crimes, in light of the higher incarceration risk. Both effects would lead to a reduction in the crime rate.

Second, all other things are hardly constant when imprisonment policies change. In all likelihood the opposite is the case: criminal sanction policies often change following observable changes in crime rates or in a number of other contextual factors that may have an effect on crime. Consider, for example, a situation in which the robbery rate increases by a factor of two, for reasons having nothing to do with the current level of criminal sanctions. In this case one could see an increase in the observed robbery rate and also very likely an increase in the number of people going to prison for robbery. Arguably, the number of prevented robberies via incapacitation would have increased because more robbers would now be in prison; it would be entirely likely that because of the increase in the robbery rate, the criminal sanctions would be further toughened, but still, one would not see a decline in actual robberies. Incapacitation would be expanding but not fast enough to catch up with the expansion in robbery offending. One can play the opposite scenario and easily identify a situation in which the prison population is going down, the number of crimes prevented by incapacitation is going down, and the observed crime rate is also declining.

The important point to keep in mind is that measures of simple correlation between incarceration rates and crime rates are not sufficient to assess incapacitation effects and may very well be misleading. Figure 1 provides a telling illustration of the ambiguous relationship over time between the incarceration and crime rates using California data for the period 1963–1998.

For robbery and burglary Figure 1 displays the crime and incarceration rates as percents of their respective 1963 values. Thus, Figure 1a shows that between 1963 and 1980 the incarceration rate for robbery remained roughly stable whereas during the same period the robbery rate increased by a factor of four. The correlation of the crime and incarceration rates for robbery over this period would be close to zero, as the first was rising sharply and the second was roughly stable. After 1980 and through 1988 the correlation is negative as the imprisonment rate steadily increased and the crime rate steadily decreased. After 1988 and through 1993 the correlation switches to a positive value with the robbery imprisonment rate continuing its assent at a constant rate relative to the prior period while the robbery rate showed a steep increase. The most striking segment of Figure 1a is the story it reveals after 1992. One observes what can be described as a meltdown of the robbery rate at a time when the imprisonment rate for robbery began to flatten. One can find segments of time in Figure 1a to advance claims that the incapacitation effect is either large or irrelevant or actually counterproductive.

The story for burglary displayed in Figure 1b is very different from that of robbery and almost uniformly supportive of the hypothesis that the incarceration and crime rates move in the opposite direction, as expected under the most simple scenario of how incapacitation works. Together Figures 1a and 1b show that the ambiguity of the relationship between incarceration and crime rates occurs not only over time within crime types but also across crime types. This introduces yet another possibility: that incapacitation works for some crimes but not for others.