Repeated Bank Robbery: Theme and Variations

This study examines all bank robberies, completed and attempted, reported to the U.K.'s Metropolitan Police in the years 19921994. It shows the rate of repetition against the same branches to be high, with each robbed branch suffering an average of 1.54 robberies, and the most robbed branch suffering six. Repeat robberies follow the success of earlier robberies, with the probability of repetition being roughly predictable from average sum taken (with attempts counting zero) at prior robberies of the same branch. Repeat robberies are less successful than first robberies, presumably because of security enhancements or staff training following the earlier event(s). Repeat robberies tend to happen soon after first robberies, and indirect evidence suggests — consistent with more direct evidence from other studies — that repeat robberies are substantially the work of the original robbers. A surprising and potentially important conclusion of the study is that banks differ greatly in their liability to repeat victimisation. Steps should be taken to supplement the data available, so as to confirm this. However, it is suggested that a meeting of senior bank security staff called Crime Prevention Studies, volume 12, pp. 153-164 154 — Roger Matthews et al. by the Home Office Crime Prevention Agency to discuss the data would not be premature. Commercial robbery, and especially armed commercial robbery, has attracted research interest in recent years (Morrison and O'Donnell, 1994; Hibberd and Shapland, 1993; Hunter and Jeffery, 1992; Challenger 1989; Walsh, 1986). A recent British study of armed robbery (Matthews, 1997) examined police responses in two force areas, the Metropolitan Police District (MPD) and South Yorkshire. In the course of this work, the extent of repeat victimisation by armed robbery seemed to emerge, but within the context of that study, the matter was not dealt with in-depth. Other studies of armed robbery involving interviews with armed robbers (Gill and Pease, 1998; Gill and Matthews, 1994) suggest that some robbers do acknowledge their repeated victimisation of the same target, and that these repeaters differ interestingly from other robbers, primarily in being more professional. This supports the findings of other studies of offenders (e.g., Ashton et al., 1998) and suggests that those who rob the same target repeatedly are more established in criminal careers, thus opening interesting possibilities of offender targeting through the detection of repeat crimes (Pease, 1997). In the present brief study, we take further the analysis of bank robberies in the MPD in the period 1992-94, in order to examine data about the characteristics of such crimes and, in particular, the nature and extent of repeat bank robberies. In 1992, armed robberies with firearms in the MPD constituted some 47% of the total of such offences in England and Wales, although the proportion had declined from 69% in 1982. Nonetheless, the MPD still accounts for a large proportion of the national problem as it relates to firearms. DATA AND PREPARATION The data comprised MPD records of all bank robberies that came to the force's notice and that took place between the beginning of 1992 and the end of 1994. These data were originally gathered for the Matthews (1996) study. They included address and name of the bank, time and date of the robbery, weapon(s) used, number and ethnicity of perpetrator(s), and whether the response to the robbery formed part of a police operation. Not included among the data was the identity of perpetrators, and hence whether perpetrators of repeats had also carried out the first offence. Also missing was the Repeated Bank Robbery— 155 number of branches of each bank in the area, so we could not calculate the probability of a branch being robbed during the three-year period. This also precludes the determination of whether the probability of a repeat among robbed branches was higher than the probability of a first victimisation among branches in general, as work in repeat victimisation generally would suggest. Faced with a set of police data, it should be easy to establish which robberies are repeats; often, it is not (Anderson et al., 1995). It was not easy with this data set, and a full week was taken editing the data in an attempt to get it as near to perfect as possible. For example, if two branches of the same bank on the same street appeared in the data, one with and one without a street number, was this a repeat? In this case, the branch with a known address was telephoned, and the question posed whether it was and had in recent years been the only branch of that bank on that street. Where the answer came that it was, the missing street number was filled in as that of the known bank. Where it was said that it was not the only branch, or where it was said that there used to be another branch, the missing street number was left missing. This did not exhaust the problem of missing values. For example, when there was one branch of bank X that had been robbed at 100 Y Street, another at 300 Y street, and another with a missing number, it was impossible to assign the missing value to one or another of the known branches. The other major problem came with branches on the same street with street numbers that were implausibly close (why would a bank have two branches of the same bank within a few doors?). In such cases, the Yellow Pages were consulted. When a number was given for one or another of the branches, that branch was phoned and asked whether there was another branch of the same bank on the same road. Of course, if both branches were featured in the Yellow Pages, the problem would have been solved without recourse to a call. In practice, this never happened. This quite lengthy process allowed clarification of some ambiguities. This method would have been better had copies of the Yellow Pages been available for each year of the period studied. Obviously, the up-to-date issue of the Yellow Pages did not include details of recently closed branches. Wherever data were inadequate, separate branches were assumed. The third, more trivial, problem came with the misspelling of roads, or the confusions between a branch described as having the address 40, High Street, Anytown, and another having the address 40, Anytown High Street. These were generally easy to edit appropriately, but the process was more troublesome when different areas were recorded for what was clearly the same bank, as in 1 Smith Street, Ealham SW1 156 — Roger Matthews et a/. and 1 Smith Street Seaham SW1, where Ealham and Seaham are contiguous areas of London. In such cases, clarification was sought from the bank concerned. Sometimes, the area was not recorded, making checks impossible. Cleaning the data set substantially increases the number of identified repeats, and incidentally illustrates one of the common reasons for overlooking the extent of repeat victimisation. Even so (as is evident in the process described above), the level of identified repeats will always remain below the true level. A convention in what follows should be stated. A "bank" refers to the institution, e.g., Natwest. A "branch" refers to the premises of a bank at a particular address. Thus, the Natwest has a branch at 1 University Precinct, Manchester M139PL.