The demand for capital punishment

The demand for capital punishment

Intmmtional R&m of Law and Economm (1993) 13, 47-59 The Demand for Capital Punishment SAMC.EI. CAMERON I. Introduction Economists have produced ma...

1MB Sizes 37 Downloads 81 Views

Intmmtional

R&m

of Law and Economm (1993) 13, 47-59

The Demand for Capital Punishment SAMC.EI. CAMERON

I. Introduction Economists have produced many empirical and theoretical studies of crime and punishment (see, e.g., Cameron 1988), including many 011 capital punishment. The fijcus of most of this work has been on two issues: establishing the efficacy of sanctions and determining the optimal level of sanctions. Criminal behavior has been treated as arising from a supply function involving a choice between licit and illicit employment. The division between these activities depends, in general, 011 their relative returns and the wealth of the individual. It follows that potential victims of crime can influence the likelihood of- experiencing a crime by altering the relative returns to the prospective criminal. At an individual level this occurs through burglar alarms and property marking, among other things. The likelihood of victimization can also be influenced through collective security in the form of police and prisons. It follows that here is a demand for protection from crime that may take the form of private investment in security devices or public investment in sanctions or sanctionsgenerating activities. The demand for crime prevention has been studied relatively little. There are empirical studies of the general demand for police personnel (Mathis and Zech lYS5), where this is “derived” from an ultimate demand for deterrence that police are thought to “produce.” Strangely, there are no studies of the demand for particular punitive measures. This is despite the fact that the capital punishment literature (see, e.g., Brier and Fienberg 1980) often assumes that there is a demand response in that the execution rate may depend on the level of homicides. In statistical terms this leads to an estimation of the instrumental variables of the equation used to explain the supply of homicides. The existing studies do not report any estimated equations of the demand response. There is thus a vacuum in economists’ analysis of capital punishment. They have failed to give attention to the public demand for capital punishment. There are many studies by sociologists, psychologists, and legal scholars of attitudes toward capital punishment. Information is also available from opinion polls. The aforementioned studies are carried out through the use of questionnaires. This paper contends that these should provide an economist with some notion of the demand schedule for capital punishment. It must bc stressed that the notion of demand used here is somewhat wider than that in a basic economics textbook, going beyond price and income to the underlying factors that condition support for the use of the death penalty. The use of an economic approach may illuminate some of the debates to be found

48

The demand for capital punishment

in the literature on attitudes on capital punishment. The next section sets out the basic economic framework for analyzing the demand for capital punishment. Section III discusses the attitude literature in the light of this framework. Following this, the implications of an economic approach for attitude research, and vice versa, are assessed. II. Economics of the Demand for Capital Punishment To begin with, let us assume that only the guilty are ever punished and that the public is perfectly informed. One corollary to this is that the public knows exactly how many murders would be deterred by an execution ifcapital punishment is a deterrent. We shall proceed on the assumption that capital punishment is a deterrent and that only this instrumental consideration lies behind public demand for capital punishment. In economic terms, capital punishment is treated as a commodity in the sense that any increase in its amount brings about an increase in consumer welfare. Clearly it does not bring direct benefit in the way that food or clothing does. Kather it operates indirectly, generating benefits in the form of deterrence. An additional execution, which deters, will reduce the probability of being a victim. Deterrence is not a private good like food or clothing. Each time an extra “unit” is produced, the benefits are spread widely over a large number of consumers. Suppose that people were asked to pay a fixed fee for extra capital punishment and that some chose not to pay the fee. Those that did not pay the fee would still benefit from the presence of additional punishment. There is thus scope for free riding, unless an exclusion mechanism is devised to deprive nonpayers of the benefit. Capital punishment-induced deterrence woulcl be a difficult benefit from which to exclude consumers. If the mere presence of the death penalty deters, then all potential murderers reduce their expected supply, and hence all potential victims should gain. Capital punishment may thus be regarded as an “intermediate public good.” The public good nature of capital punishment is C‘VL‘II greater for those disposed to the argument that it operates primarily through a “demonstration effect” of showing a vigorous commitment to the upkeep of law and order. See, e.g., Berm 19’N. The conventional way to derive an individual demand curve for a private good is to use the “revealed preference” apparatus. In this framework all factors are held constant except the price of a good, and changes in consmnption that arise are observed. For a normal good we expect to find an inverse relation with price and ;I positive one with income. The demand fi)r a pure public good cannot be observed in this way as it is not possible to charge for its consumption. Economists have tackled this problem by the use of contingent valuation (Hanley 108Y), which uses a questionnaire to reveal preference. ‘I-he contingent valuation method asks how much people would be willing to pay for different quantities of the good all other things being equal. If people expressed their preferences truthfully, then we could add up the willingness-to-pay schedules (assuming they are independent of each other) to find the community demand. This could be compared with the costs to arrive at the optimal level of punishment (see Cameron 1989 for a discussion of the conceptual problems in arriving at an optimum level of capital purlishment). The determination of a demand schedule requires the identification of prices and quantities for the product. It is more difficult to do this when the product is execution. The quantity has two dimensions: whether or not execution is allowed for an offence, and the extent of its use for that offence. Public opinion has only been

S.CAMER~N

49

sought on the former question and not on the latter. However, when we aggregate across individual opinions, we arrive at a continuous measure of quantity dcmandcd in the form of the degree of support. ‘l‘here are two ways of approaching the prohlem of defining price. One is to take the literal approach of looking at the cost of processing and performing executions. ‘This cost may vary due to adaptive behavior of those to be executed. In the classic scenario someone who has already performed a murder that is certain to result in execution has nothing to lose from performing more in attempting to avoid capture. It is probable that the cost of capital punishment is seriously underestimated by the general public. Nakell (1978) details the expense that arises from a more costly trial process, a more costly appeal process, and the stringent iniprisonnient requirements for a prisoner on death row. ‘I‘he survey, evidence on support for the death penalty does not ask about the literal dimension of price. I( does ask about the second approach, the ranking of’ offences by their seriousness. ‘l‘his gives us the classic inverse demand relation, i.e., support for the death penalty increases 21s the seriousness of crime increases. .I%ere are two circumstances in which an optimuni may not be attained: first, when people behave strategically by over-- or under-reporting their true preferences in an attempt to tilt the outcome; second, when demand is revealed through an electoral mechanism. The most common model used in these circumstances is the median voter model [see Ekelund and Hebert (IYX3: 522-23) for a brief exposition]. Let us now sketch a simplified application of’thc model to capital punishment. Assume that use of the death penalty is to be decided in a single issue election and that the demands of individual citizens are independent of.each other. Individual demands will emerge from derived demand curves in which the demand is ultimately for protection against crime. State executions will, it is hoped, contribute to the deterrence of future criminal acts. .I‘hc derived demand is expected co increase when there is an exogenous increase in the amount of crime. If. public sector deterrence is a “normal” good, then it is also expected that the derived demand will be positively related to income. If this relationship is continuous and monotonic, then the mcdlan level of demand will belong to the voter with the median income. In the median voter model, the volume of expenditure will be that \,oted for by the median voter, i.e., the individual who is just decisive in a majority vote. Kcd~stribution of’ income among the other citizen voters on either side of median income will have no effect on the outcome. A simple median voter model thus implies that support for the death penalty will be positively associated with income and the level of crime. It will also tend to be suboptimal. There are two crucial assumptions that we must attack: perfect information and instrumentalism. Instrumentalis~~~ here denotes the position that punishment is only used because it prevents crime, i.e., it is an instrument for the attaimnent of other ends rather than an end in itself. Information failure occurs at a number of levels. At the level of execution MT may find that some innocent people are executed. Instrumentalists usually conclude that this makes no difference (see Cameron 1989). If we suffer a loss of welfare from knowing that there is a probability of some execution of innocents, then our demand for the death penalty will decrease. ‘This will reduce the optimal and median-voted amounts of it. At the level of public support, we may he misinformed of the effects of capital (and other) punishment. In the most extreme case we might believe that it deters without having consulted any evidence. If pure instrumentalists were to experience a convincing demonstration chat there is no deterrent effect, then all support fijr the

death penalty should evaporate. Justice Marshall, in the F~rn~ol’ case “questioned the relevance of‘ opinions based on ignorance or prejudice to any reasonable definition of enlightened community standards” [EllswortIh anti Ross ( 1983: I ‘LO)]. ‘f‘lirls, it may be that a high level of’ public support is merely ;I manif’estation of a strong degree of public ignorance, in which case we may feel reluctant to put the case to a vote. This has been investigated in the empirical work by looking at whether the receipt of’ information changes expressed prcf’erences. .A more subtle point is that the public may express a set 01. pretcrences that is treated as unified when they have ;I multiplicity 01‘ pref’bretice sets. A simple case is one in which ;I mother of ;I brutally assaulted child urges the death of the perpetrator, when the crime is revealed, but later recants. Her-c the “heat-of-the-rnoIll~Iit” preferences are retributivist while the reflective prcferenccs may bc instl-lImentalist. Recent writings on rational choice theory (Hstcr IOH2, Schelling 19H4, and F:t/ioni 1987) have grappled with the pr-ohlcms 0C ;I multiplicity of pr&rencc sets lvithin one individual. ‘l‘he I‘uritlanicntal proldem is one of IliC;IslIi-emerit. Which pl-c’terences do we enter into a so&d &&ion process? ‘l‘he most relevant schism Ii)r this discussion is that between preferences on 2 jury versus prefcre1mzs as mennbcrs of‘ the gmeral public. Views expressed in ;I situatwn where they carry no cost of‘ wspotisibility ma): well (lift& f’roni those fi)und in the jury situation, \\,here they have weighty consequences. hluch of’ the dif’f’erencc may Aem f’rorn voting fin. the tlcatli ol 211 motiynio~~s person vel-sus dec-iding the death of ati actual person. Parallels can be fi~untl in lift saving; huge I-esou~ws have ken expe~~dctl to find known lost individuals when putting these rcsourws illto risk reduction ~wdd have saved the li\w of‘many more persons who are not knowtl KYutr/r. Although the literature on attitude nm~surmient has been excited ahout diverguices in pretercnce when the “servingon-a7iury” scenario is presented. it has ig~Ioretl ai important criticism 01. dual prt’fcl-cncc’ models (Brentian 1989). ‘l‘hcl-c is no obvious indication that one‘ set of’ prel-

erences is a more coI-rect or tl-uthf’ul I-cl,l.cseIIt;ltioI1 of one’s true wishes than the other. hlany authors seek to impost ;I mod ordering, e.g., they suppose that wanting to he Iazy and get t1t is a “lower ol-der” pref’ercnce than to he active and slim. .I‘tiis is ;t prior ,jutlgnietit ot the observer. In the case of‘ capital sentencing. wc art’ dgain fitted with a problem in democracy. Which prefbrenccs should the clcctetl reprcscntatives cnforcc? It is sirrcly the case that they impose their own mol-al preterences in deciding which art’ the “higher” irnpulscs. ‘Illis leads to the issue ot‘~.etl-ih~ltioli. ‘l‘his is 110t the place to enter into the varieties of retributivist thought, so wc sliall conlilie ourselves to the simple idea that certain acts “tleservc” death. Most usually ttwsc acts involve killing other people. 1)erivation of‘ retributivist tlemand is straigtitfi~l~war~l; every act that warrants execution must result iu it. (:ornplic~ttions arise with the mallt&lity of‘ retributivist preterenccs. Instrumcntal preferences should be highly responsive to infimiiatiot~. c.g., a pure illstrumentalist would give up capital punishment if’it coultl he conclusively shown that

S.

CAMERON

51

it is ineffective. A pure retributivist would never change attitudes. Given what we have said above, it might be argued that retributionist thought might be seen, by some, as an expression of lower order preferences. It would then be contingent OII the scenario presented when eliciting preferences about the death penalty. Some comments are in order on the relationship between this paper and the normative debate on capital punishment. As the material discussed is of a social psychological nature, it does not concern itself with arguing the case for or against the use of the death penalty. This cannot be done with empirical analysis. Rather, it is a positivist analysis of normative thinking. For example, it investigates the robustness of beliefs about right and wrong to one’s place in the punishment process. We have now provided sufficient background for the treatment of attitude studies as expressions of a “demand” for capital punishment. There are a number of-questions that the contextual background suggests. At the most basic economic level, we should ask whether support for the death penalty is a function of the seriousness of the crime and the amount of serious crimes (most simply represented by total homicides). As the demand function is derived from the need for protection, we would expect an increase in demand when there are more threats to be protected from. The next section of the paper is presented as a series of’ questions with answers from the available research. ‘I’he first two questions come from the most basic economic model.

III.

Findings from Attitude Surveys

The collection of evidence is sparse outside the United States and Canada. There is one systematic study of attitudes across a number of nations (Kay 1982). For example, in the United Kingdom there are no academic studies, and the first major opinion poll since 1965 (Shaw 1982) on attitudes to punishment makes IIO reference to the death penalty. There have been a number of questions in general polls in the United Kingdom, but these have not been systematically discussed. This is rectified below. One might conclude from this that, as most developed democratic nations have abolished the death penalty quite some time ago, there is only an interest in the systematic study of public opinion where abolition is on the agenda. This is borne out by the fact that much of the interest in the United States is due to the abeyance of capital punishment in the post-Furman period. Likewise, the Canadian activity in this field stems from debates surrounding abolition in 1976. Evidence comes from two sources: regular public opinion polls and special studies of attitudes. There is a marked difference bctwcen these, as the latter are carried out by social psychologists or legal scholars and the former by pollsters. In the former all that is usually asked is the basic question: “Do you support capital punishment?” (or variants of it). Background economic and/or demographic information may be recorded along with this. In the latter a battery of questions to measure attitudes in terms of, for example, authoritarianism and dogmatism are also included. Some also address the problem of the absence of a contingent valuation element in opinion poll studies (e.g., Ellsworth and Koss 3983 or .Jurow 1971). ‘To be more specific respondents are asked to reconsider what their position would be if asked to serve or1 a capital jury rather than merely express support in a vacuum. The one large and obvious advantage of opinion poll evidence is that the same questions have been asked over a long period of time, so we have a barometer of attitudes that covers the

period frotn 19% in the United States (Zeisel and Gallup 1989) and 1937 in the United Kingdom (Gallup 1976), although the comparability of the United Kingdom questions over time is much less. We now consider the evidence available to answer a series of‘ questiotts. I. Is su#m-t cotmeclrd wilh thr crittw concrtxrd? All opinion polls attd special surveys find that the percentage in f’:tvor of capital punishment rises with the perceived seriousness of the oftcnce. For example, the United Kingdotn Gallup poll for 1981 [Webb and Wybrow (198’L: 12 l)] tinds a top level of support (75%)) for terrorist murders or murdering someone ‘$ts’ for the fun of it,” with 64%) for child rape and 40%S for murdering sot~tconc who disturbs a break-in. Thus we see a feeling that the “price” of crime should he proportional to the loss that it inflicts. ‘l‘his can be interpreted equally coltet-ently in utilitariattiittstrutnentalist or retributivist terms.

The time series of poll data for the United States presented by Zeisel and Gallup (1989) shows support at 619, in 1936. It rises to a peak of 6X% in 1953 and falls thereafter, going below 50% after 1960. ‘l‘he all-time low was 42% in 1966. Support then rose steadily to an all-time high of’ 76% in 1988. The authors make no attempt to relate the above swings to events outside their poll data. Yunkcr (1977) provides a table of’ per capita murder rates over time that shows them declining f‘rottt 1950 to the mid- 1960s; from 1965 there is a steady increase. There is thus broad support, at a descriptive level, for an association of the demand for capita1 punishment with tlte amount of murder. The rise in support for capital punishtnent also happens to coincide with the post-Fu?.?tlclrt suspensiott. If there is a connection, it is not possible to disentangle it from that with the volume of’ homicide. Wc have assembled a United Kingdom series in ‘I‘able 1. It can be seen that there has bern a consistently high level

47 fi9

79 70 7(P 70 6I 71 8:s” ti:?l tifi 66

S.

CAMERON

53

of support in the postwar period (the low 1937 figure is something of a puzzle). There has been no sign of the U-shaped trend that is so distinctive over the same period in America. Given the oft-remarked less violent nature of British society, the strength of British support may also be a puzzle. It would seem that in Britain the level of support is consistently high, having no relationship to the volume of murder or the presence or absence of the death penalty. Smith (1976) in his survey of United States polls finds support for capital punishment to respond positively to reports that the crime problem is worsening (see also Rankin 1979). 3. Does the demand forcapital puntihment respond to irformution? The time series poll data suggest that some people are changing their mind about the issue. If this is not to be regarded as random or irrational, then it must be the consequence of some receipt of information or a fundamental change in attitudes (see below). Relevant information can be split into two categories. The first is supposedly objective research about the operation and effect of capital punishment: for example, does it deter? The second is the experience of “salient” events that make an impact on the individual beyond what would be justified in a pure rational choice decision theory. Under this heading would come events, such as a brutal murder, that capture the public imagination or a personal experience, such as the murder 01 a close relative. The 1986 United States Gallup poll fijund that 16%’ of supporters would change their minds if there was convincing evidence against deterrence (Zeisel and Gallup 1989). Some surveys have looked into the impact of the first kind of information. Sarat and Vidmar (1976) presented their subjects with pamphlets reviewing evidence on deterrent effects on capital punishment in a negative light. This resulted in a decline of support from 51% to 38%. Similar results were found in a more detailed replication by Vidmar and Dittenhoffer (1981). It is entirely logical that the above studies could be conducted the other way round, i.e., evidence in favor of the deterrent effect could be given that would push up the percentage of supporters. ‘l-his does not seem to have been countenanced in the literature, perhaps because there is some normative bias against capital punishment on the part of the academic co~llmunity. It is a well-established proposition in social psychology that people become strongly attached to their prior beliefs. We ought then to be suspicious of glib statements about willingness to revise a long-held position. In particular, we ought to probe beneath the surface to the fact that people who say this kind of thing might refuse to accept any evidence against their established belief as convincing. This proposition is supported by Lord, Ross, and Lepper (1979). In their study, groups of supporters and opponents of the death penalty were given reports of fictitious studies (one pro and one anti in their findings). Subjects gave higher ratings to the study that upheld their prior conviction. It was also found that support for prior conviction increased during the process of exposure to the congruent study. A “rebound effect” emerged whereby the study challenging prior belief was briefly favored but then rejected w further reflection. The lower rating of its quality would assist in post hoc rationalization of this process. As regards salient events, Smith (1976) compared the position of victims of crime and nonvictims over the years 1936-74. The assumption was that the former should find capital punishment more salient as it relates more directly to their own experience, but there was no significant difference between the two groups. This rules out the possibility that the trend in polls (discussed in question 2) is due to swings in the

degree of personal exposure to crime. Smith’s findings are supported in a detailed telephone interview study by Tyler and Weber ( 1982). They find that past experience of crime contributes little to the explanation of’ support for the death penalty. As mentioned at point 1, there is a response to gf~/r~c~l information about crime levels. Some dif’ferent evidence on salience (Cox and Tanford 1989) springs from the issue oP jury selection. Under U.S. law people could be excluded f’rom capital juries if’they were considered to be biased against capital punishment. This may be seen as unfair if’ genera1 punitiveness is correlated with death penalty attitudes, as there will be a consequent increase in the probability of being found guilty. Cox and Tanford expose their experimental subjects to descriptions of vivid, brutal crimes, which results in the inclusion of’ many’fijr ,jury service who would have been excluded by a simple “do you approve” question. An earlier study that provided subjects with “vignettes” of capital crimes (Hanliltott and Kotkin 1979) uncovered a worrying problem of’ ignorance. ‘l‘he responses indicated that the public fail to take into account aggravating and mitigating fjtctors that should, in law, determine whether a capital sentence is imposed. Some new light on the question of’ infi~rm;ttion/saliettce is provided by recent experience in England. I’olls using unsolicited telephone calls are conducted by the Oracle teletext set-vice. ‘l‘hese could not finm the basis of’detailed scientific analysis, given the problems of’ self&lection, among other things. However, they should be roughly comparable across years when an important event occurs. In 1CUE)the service asked “shou1d the death penalty be reinstated?” Eighty-nitte pet-cent I-espot~ded “yes.” This poll was taken at the time of’the latest of’ parliamentary votes on the issue. In 3991 the “Uirmingham Six”-six Irishmen who had been convicted of’ att IRA bombing-were finally f‘recd f‘rom prison. ‘l‘heit- I-elease was precipitated by discoveries of unreliable police evidence and ineptitude in forensic practice. ‘l‘his could be expected to rock public support f’ot- the death penaltv as these men would have been dead if it had been in force. ‘I‘he Oracle poll coinciciing with these events showed a fall in support to 7.‘,‘%. ‘I‘hc satmpIc size was about I 1,000. This would thus be a statistically significant decline, but ir still lcaves a very high level of’ support, suggesting that British sociery is rather prtttitive (see the evidence fin I955 at question 7). 4. 1.s.\up~wl wlnld IO drvtwg~-c~phrc dwcufori,~ tic..,? In the literature, t-ace, sex, age, education, wealth, and religion have been identified as demographic characteristics. According to the 1974 survey by Vidtnar attd Ellsworth, it was well established that support is more likely if’ the respondent is white, older, male, Catholic, less educated and less wealthy. ‘Ilie racial differential could be explained in rational choice terms. A black person is CP~P)L\pcr~-ihl~.~more likely to receive the death penalty. Even if’ one never intends to commit a capital crime, the possibility ot‘wrongful arrest and execution has to be countenanced. Being black makes this a more likely event. At the collective level black people may be opposed to the death penalty on the grounds that it is actuarially unfjtir fill- othet black people. More recent studies, such as .I‘ylet- and Webet- 1982 atid Kay 1982, have f’tiled to find a stalistically significant cottttection with the above variables. ‘I‘he difficulty with this research is that hypothesis testing never rises above bivariate correlations in its level of‘ sophistication. It is thus very difficult to pin much hope on the results as the covariance of’ the explanatory variables is not taken into account. 5. Is su@ort n~fi~v~tion of penovdit~ type? Many rescarchcrs take the position that type of’ personality is a more potent cause than demographics. Support For the death penalty has been linked to conservatism,

S. CAMERON

5.5

toughmindedness, dogmatism, and authoritarianism (Kay 1982, Neapolitan 1983, Thomas and Howard 1977, Tyler and Weher 1982, and Singh and .Jayawardene 1978). It is no easy matter to measure these things. The method adopted is to use multi-item scales developed by social psychologists for other purposes. There seems to he a fair degree of support for the connection with personality type, although it is subject to the same reservations given above about the lack of control for covariante. To take a simple example: if people tend to become more authoritarian as they get older, it is difficult to decide which, both, or neither of these causes support fbr capital punishment if all we have to go on are hivariate correlations drawn from different populations. Singh and Jayawardene draw the conclusion that their results suggest that attitude changes would he useful. Leaving aside the political naivete of this remark, there is social psychological evidence that such a step may not he feasible (see Lord, Koss, and Lepper 1979). Like virtually all authors in the field, they have a predisposition against capital punishment ant1 regard those who support it as needing their attitudes changed. One should be cautious about drawing inferences from correlations with personality in this field, given the reservations above about multiple personality models. There is an obvious statistical problem as to whether personality measurement is stable. Measuring at a point in time gives us an index of personality, hut it is not clear whether the same outcome would arise if’ measurement was retaken at a different time or under dif‘ferent circumstances. Given the discussion at 3 above. it might appear that support varies directly with the popularity of an instrumental view of the role of punishment. Many opinion polls and surveys have elicited the deterrence factor as a strong reason for supporting capital punishment. The problem is that in most cases the conditions are not met fog This is corrected in s01iie a proper contingent evaluation. Choices are unconstrained. instances. For example, the 197?1 Harris poll in the United States (Vidmar and Ellsworth 1973) found that people said they would still support the death penalty even if there were no deterrent effect. ‘1% conflicts with the evidence at 3 above to sonic extent. A rare multivariate stucly by Thomas and Howard (1977) includes measures of personality and use of punishment. ‘rhcy find that support for deterrence is the strongest predictor of support. ‘l‘hero is a problem with 2 above if deterrence is to be taken seriously as the driving force in fluctuations in support fi)r capital punishment. The evidence OII type of‘ crime shows strong support for the use of‘the death penalty in cases where we would expect weak deterrent effects (e.g., terrorism) and much weaker support where we might expect much stronger deterrent effects (e.g., being disturbed during a “crime for profit”). In addition, there is the classic problem with attitude studies that preferences may be deliberately misrepresented. One could draw the implication from the literature that deterrence is not the driving force in support for the death penalty, but it is stated to he so by people because they may think this is more socially acceptable (and they are used to saying this to strengthen their case). 7. Is supporl role-spec$r? The British Gallup Poll (Gallup 1976) in 1955 asked how much support respondents thought they would give to a capital verdict if they were serving on a jury. The percentage giving general support is 73%, and this falls to 67% for those who were asked as though they were on a jury. The decrease is not very dramatic, and it still leaves a substantial majority in favor of the death penalty. Unfortunately there is no

subsequent British evidence on this problen~. Similar questions were asked in the 1975 Harris survey in the United States (Vidmar anti Idlsworth 1974), showing a tall from 59%~ approval of the death penalty in the abstract to 39% for those serving on a jury. The academic survey evidence on this question has usually f’ound a dramatic fall in support when the hypothetical question about ,jury service has been asked. In the study by .Jurow (197 lj 45%’ f’avorccl the death penalty initially but only 8% of those ou the hypothetical jurv f’avored it. A significant directional shift was also ti~nd in the survey by El&or’th and Koss ( 1CM). Davitlow and Lowe ( I979) fi~uncl that law students moderated their views when they wc’rc asked the jury questiou, 1’rom 23% to 2.4%. Since a general expression of‘ support conceals a complex relationship in which support is contingent on the position of’ the rcspondcnt, a prominent theme in the sociological literature is that supper-t is more symbolic than ;ictu;il. ‘l‘his is well put by Hedau (1982: 68):

It could be argued that such ;I position stems f’rom “ahsolutc tletcrrencc.” Many individuals who support the death penalty may f’cel that it must be on the statute books bccausc of its gcneml quality of‘ public good in undcrgirtling gcncral confin-mity with law anti order. ‘l‘hey may ltel that it is irief’fdive as ii specific or marginal deterrent. A tlilferent explanation cm be extracted, f’rom the evidence discussed in question 7, in the fill-m of‘s ctivcrplcr of’m CIT/LP and m fm~ prcfcrcnccs. Once known individuals with names and faces come into the frame, a pro death penalty pemon may turn into au anti one. ‘I‘hc symbolic versus actual argument cm be illuminated with an economic appro~h. Assuming the electorate to be utility maximizers we could stylize them as ireceiving positive utility Lroui the al2iolute deterrent elr‘ect of‘capital punishment but Iwgative marginal utility L‘roni individual executions. In other wortls the preseuce of’ execution is a good, but actual executions are ;I bad. If WC‘ti~llow the standard sub ,jecti\:e expected utility inaxirniz;ition motlcl (SW Canicron 19X9), then thcrc must be some probability 01’ execution, however small. otherwise the absolute dctcrrent cf‘f’cct cannot work. If’ WC‘accept this, then the good must be balanced qainst the Id, i.e., the loss of. w&u-e f~rom dislike of. clctrtnl exwutiou is takeu into account wheu wol-king out the necessary probability or rate of‘ cxccution. ‘l‘his will imply an ef‘licieut solution iu which some people are deliber-ately reprieved the death penalty, even though they arc not so cntitlcd, in tcrnis of justice, than those who arc cxecutecl. ‘There is here a clear problem 01‘ horizontal equity, but this is not the appropridtc place to discuss it (see Cmier-oil 1989).

S.CAMER~N

57

9. Is the reluctance r,f‘goz~rmmer~t.cto use the death prnnlty undemocmtic, given it.\popular-ity with the general public? Some people have drawn conclusions about the democratic process from the fact that there is generally majority support, in opinion polls and surveys, for the death penalty. Clemens (1983) argues that capital punishment is a populist cause in the United Kingdom, yet it has not been repealed by the Tory governments of the 1970s. Kay (1982) suggests that South Africa is a democratic nation on the grounds of a comparison of attitudes to the death penalty among whites in Johannesburg, Los Angeles, Glasgow, London (England), Sydney (Australia), and Manila. He says that “it would appear that the South African government acts more in accordance with the popular will than do many of the other supposedly democratic governments. Ketention of the death penalty is a sign of’ democracy not authoritarianism” (ibid., 288). There have been vigorous objections to this type of reasoning. For example, Lacey (1983: 419) argues against capital punishment on the grounds that the “traditional function” of a politician is to,judge on the basis of the “arguments which support the views which people hold, and his own evaluation of these arguments, not on the fact of these views being held” (emphasis in original). The material in points 3, 7, and 8 above lend some weight to this argument. Their import can be summarized neatly in a quotation from Vidmar and Dittenhof-fer (198 1: 54): “Members of Parliament and other decision-makers should, at the very least, treat superficial poll results with great caution. Opinion polls based on one or two items about capital punishment obscure the possibility that the expressed attitudes may reHect lack of knowledge, uncertainty, or the fact that people are only expressing symbolic attitudes to high crime rates.” IV.

Conclusion

Let us now consider the implications of the above for policy and analysis. From the mid- 197Os, economists have striven to influence penal policy through econometric studies of the effect of capital punishment. In the United States econometric evidence has even been considered by the Supreme Court. ‘l-he focus of most of the debate has been on the size of the deterrent effect. In the framework of’ welfare economics, which is commonly used, a large enough deterrent effect would induce the support of the positive economist for the death penalty (see Cameron 1989). Judging by the heat of the exchanges, there seems to be a general feeling that a clear demonstration of a large deterrent effect would sweep the death penalty into popular approval. This obscures much of the complexity of attitudes to capital punishment, which we may render into economic parlance as the demand for capital punishment. It is somewhat peculiar that economists have labored feverishly ovei estimation of the deterrent effect when it seems to be a relatively unimportant facto1 in support for the death penalty. Further, it has been shown that expressed clen~ands are difficult to take at face value. It must be decided what the relevant bounclary conditions are when evaluating support. Support for the death penalty is an elusive entity. It depends on what has been asked or, more importantly, what the subjects think they have been asked. A striking finding of the literature is the dependence of’ answers to questions on the role envisaged by the respondent. In the past, hardheaded economists would simply have dismissed survey evidence as irrelevant because it does not live up to the strictures of revealed preference methodology. It is

58

Thr dmand

for cupitul pwaishmrnt

difficult to maintain such a position because of the increasing importance of valuing nonmarket entities as evidence in contingent valuation studies. The issue of capital punishment highlights the problem of identifyingjust what a valuation is contingent on. This problem is magnified by the presence of normative elements in the decisionmaking problem.

SARAT, D. AND N. VIDMAR. 1976. Public opinion, the death penalty and the eighth amendment: testing the Marshall hypothesis. Wzsconszn. Law Rwiero 17: 17 l-206. SCHELLING, T. 1984. Self command in practice, in poliry, and in a theory of.rational choice. Amerrcan Economz~ Rruim 74(2): l-l I SHAW, S. 1982. 7‘11~Peop/r’,Ju.\lic~: A 11fapr Poll o/ Public A~zludrs on Crimr and Punk/mm/. Prison Reform Trust. SINGH, A., ANI) C:. H. S. JAYAWARUENE. 1978. Conservatism and toughmindcdness as determinants of attitudes towards capital punishment. CrrnudiclnJoul-,lnl of C~-zrnz~~o~o~~ 20: 19 I-93. SMITH,~. W. 1976. A trend analysis of attitudes towards capital punishment, 1936-1974. In .S!z&.\ of Social Changr since 1948, edited by ,J. E. Davis. Chicago: National Opinion Research Center Report 127b. THOMAS, C., AND R. G. HOWARI). 1977. Public atritudes towards capital punishment: a comparative analysis. Jourr~ul of Br/~rzww/ Economic-.t 6( 112): 189-2 16. .~YI.EX. .r. R., ANI) R. WEBER. 1982. Suppol-t for the death penalty: instl-umental response orsymbolic attitude. I.nw rrwt SOCIPI~Rr~~ivu 17: 22 l-45. VIUMAR, N.. AN,) ?I‘. DITTLNHOFFER. I98 I. Informed public opinion and death penalty attitudes. Canadinrt ,/our nul of Cr-i~~~~nolo,~ 23( 1): 4?-56. VIEMAR. N., AND P. EI.I.SWORTH.1974. Public- opinion and the death penalty. Slnr~fowl Lnw Rrz~r~rc’ 26: 1245-70. Wt~n, N., .4&u R. Wvsnow. 19X2. T/w ~kllnf~ RP/IOI/: IYNI. L.ondot~: Sphere Hooks. YIINKER,,I. A. 1’377. An old rontroversy renewed: introduction to the capital punishment synposium. Jowncll oJ Rr/wviotrt/ Ecorrorrltc t 6(1/2): l-32. ZEISFI., 11.. ANU A. hl. G.41.1 (1~. 1989. Death penalty sentiment in the LJnircd States. ,/orrr& of @~unlilnl~7v C~nni~doy~ 5(3): 285-96.