This report has sought to replicate the costing methodology developed in previous AIC publications by Walker (1997, 1992), Mayhew (2003a, 2003b), and Rollings (2008). Reference should be made to these previous publications to understand the development of the methodology over time and where improvements and variations have been made to earlier approaches. Where more recent and authoritative sources of information have been located, these have been used in preference to the sources relied on in the earlier AIC reports. In addition, where a strong theoretical or practical case could be made for diversion from earlier approaches, then a variation in the methodology was undertaken. The magnitude of multipliers was also reassessed and where more accurate information was available, the multiplier was varied as indicated below. Finally, where no updated or better data were available, the original 2005 figure was taken and inflated by the consumer price index (CPI) to 2011 values using the Reserve Bank of Australia’s inflation calculator (RBA 2013).
The complexities associated with attempting to quantify the cost of crime are examined by Mayhew (2003b) and a review of the cost of crime literature was undertaken as part of the Mainstreaming Methodology for Estimating Costs of Crime project as presented by the Centre for Criminal Justice (2008). The material presented in these publications has not been reproduced here.
Data used in this report relate to the 2011 calendar year (1 January to 31 December), the most recent for which official crime statistics were available. However, some data are reported for financial years only (1 July to 30 June) and in this case, the financial year 2011–12 data were used where they were available. In addition, where regular survey data have been used, such as crime victimisation and business surveys, these results were used where the respondent reference period was the 2011–12 year, or where the reference period extended for shorter or longer periods, the results recalculated to approximate the 2011 calendar year.
It is also important to determine whether an ‘incidence-based’ or a ‘prevalence-based’ approach should be used (see Centre for Criminal Justice 2008). An ‘incidence-based’ cost of crime estimate is based on individual crime episodes that may have cost implications for many years following the commission of the crime. Thus, ‘incidence-based’ approaches count both present and future costs in the year in which the crime occurred. A ‘prevalence-based’ approach, however, examines costs actually incurred during the reference period in question including all costs incurred during the reference period, regardless of when the crime was committed. ‘Incidence-based’ estimates indicate how much could be saved by preventing future crime and therefore, are relevant for criminal justice policy analysis and are used in the present report.
Multipliers and their use
To estimate the cost of a particular crime, the frequency with which the crime occurs needs to be established. A major difficulty in attempting to assess the costs of crime is the ‘unknown’ frequency of many types of crimes. There are several reasons why the number of crimes that occur may be unknown, in that they have not been detected or not reported officially. First, some victims of crime might not be aware that a crime has taken place (eg in the case of charity fraud where donations are made to an ostensibly legitimate organisation that is actually non-existent). Not all crimes may be reported to the police. This is especially the case if the nature of the crime is considered too trivial to report (eg the case of an attempted break and enter where nothing was stolen), or in the case of a ‘victimless’ crime, where it is not clear a crime has taken place (eg shoplifting). There are also incidents in which more serious crimes are not reported to the police for fear of the consequences of reporting due to concern about reprisals or because the victim is uncomfortable or scared to report the crime (as with sexual assault or domestic violence-related assault). This ‘gap’ between the number of reported crime incidents and the actual number of incidents that occur makes the costing of crime difficult.
Secondly, not all crimes reported to police are necessarily recorded by police as a crime, although under the national crime recording standard this should be rare. The non-recording of crimes by police may occur for a number of reasons, including complying with victims’ wishes not to proceed, the police may feel the report is mistaken or dishonest, or the police may feel there is insufficient evidence to proceed with a charge (Mayhew 2003b).
For the purposes of costing crime, the ‘gap’ between recorded and actual crimes is addressed through the use of so-called ‘multipliers’ for each crime type. This is an estimate of how much police-recorded crime (as presented in ABS 2012b) should be inflated to estimate the ‘true’ number of crimes that have occurred. For example, this paper assumes that all homicides are known to police; therefore, homicide is assigned a multiplier of ‘1.0’. However, it is well documented that not all instances of sexual assault are reported to police (eg Daly & Bouhours 2009), so a multiplier of 6.2 was used to adjust the recorded crime figures to obtain an estimate of the actual number of sexual assaults. A multiplier of 6.3 means that only 16.2 percent of sexual assaults were recorded by police.
Where possible, the multiplier is calculated using the nationally representative data collected in the ABS’ 2011–12 Multipurpose Household Survey. Respondents were asked whether they had been the victim of selected offences in the previous 12 months and those responses were taken at ‘face value’. It should be noted that the calculation of the estimated number of crimes in this report is not necessarily the figure presented in ABS Crime Victimisation data because this does not include victims under the age of 15 years, whereas recorded police figures include victims of all ages. To address this issue, the relevant crime victimisation survey data have been inflated to reflect an estimated number of crimes that would have been experienced by persons under 15 years of age. Although this results in an estimated total only, in the absence of better data it has been adopted to provide a similar estimation to that previously undertaken by Mayhew (2008b).
There are also the many limitations that self-report victimisations surveys carry. These include the difficulties associated with seeking information from victims of crime about complex crime types such as fraud. Problems of telescoping information (ie including events outside the survey reference period), exaggerating facts or reporting selectively—all common problems with surveys and personal interviewing—can affect the accuracy of information gathered using conventional techniques. There may also be problems of veracity, where a manager may be reluctant to report circumstances that may be personally incriminating or that may attract negative publicity for the organisation. There may also be problems arising from organisational incentives associated with how criminal behaviour is classified. For example, some financial crime losses are characterised by organisations as bad debt rather than fraud, leading to under-reporting of criminal acts of dishonesty. In the case of large corporations or revenue collecting government agencies, considerations of this nature can massively affect the accuracy of the estimates of fraud arrived at.
Finally, ABS crime victimisation survey data cannot fundamentally include undetected crimes of which victims are unaware and conduct that might not fall within technical definitions of criminality (see in relation to fraud Levi 2008 and Levi & Burrows 2007). This is particularly relevant in the case of financial crimes involving revenue fraud, corporate and personal credit-related fraud and charity fraud, which are not always apparent to victims and therefore not reported officially.
The multiplier is calculated by comparing the number of crimes as reported in the relevant crime victimisation survey with those reported in ABS Recorded Crime—Victims. It is important to match the crime victimisation survey data reference period (2011–12) with the same period used for police recorded crime. As noted above, this was not always possible as police data for some jurisdictions related to calendar years rather than financial years. There is also the difficulty that police data often refer to crimes recorded during the period in question, regardless of when they were perpetrated. Crime victimisation surveys, however, ask participants to recall their experiences of incidents that occurred during the 12 months prior to the reference period in question. For present purposes in determining multipliers, the reference period of the calendar year 2011 has been be used, or where this is impossible, then the financial year 2011–12. On occasions when data for 2011–12 were not yet available, the year 2010–11 was used as the best alternative.
For some property crimes, reporting rates differ according to the seriousness of the offences, whether the offence was completed or attempted and the degree of loss suffered. These differences need to be reflected by applying different multipliers for each subcategory of offence. In the case of fraud, for example, extremely large-value commercial frauds are more likely to be reported to police than the many small-value frauds experienced by consumers, such as losses suffered as a result of minor credit card fraud. Often, such incidents are reported to credit card issuers or banks, who will usually refund the losses to the card holder and charge the losses back to merchants. In such cases, it is the merchant that may ultimately bear the burden of the fraud. Reporting of such crimes to police is very low. The AIC’s annual consumer fraud survey for 2012, for example, found a reporting rate to police of 17.3 percent of victims of scams in 2011—which would equate to a multiplier of 5.8 (Jorna & Hutchings 2013). However, in the case of serious fraud, KPMG (2013) found that 46 percent of major fraud incidents reported in the survey were referred to the police, which would equate to a multiplier of 2.2. There are similarly wide variations in reporting rates for individual and commercial robbery and burglary where commercial incidents are more likely to be reported for insurance purposes.
Intangible costs are those costs not usually ‘exchanged private or public markets, such as fear, pain, suffering, and lost quality of life’ (Cohen 2005: 25). The methodology used by Mayhew (2003b) for assessing the intangible costs of crime was used, based on the more recent study conducted for the Home Office for the year 2003–04 in the United Kingdom by Dubourg, Hamed and Thorns (2005). Mayhew (2003b) describes in detail the various approaches that can be used to the estimation of intangible costs and it is unnecessary to repeat these here. These estimates were adjusting for inflating in the United Kingdom between 2004 and 2011, and for pricing and purchasing power parity differences between the United Kingdom and Australia.
Purchasing power parities and inflation figures
Cost estimates from both the United States and the United Kingdom have been used in this report, as Australian data were not available. In both cases, Purchasing Power Parities (PPP) issued by the Organisation for Economic Co-operation and Development (OECD 2013) were used to convert costs given in US$ or UK£ to AU$. This gave conversion rates of A$2.20 to £1.00 and A$1.49 to US$1.00 for 2011, which were applied in this report.
There were several places in the report where costs estimated by Mayhew (2003b) and Rollings (2008) or where US and UK figures needed to be inflated to 2011 figures. This was done using the CPI inflation rates reported by the Reserve Bank of Australia using their online inflation calculator (RBA 2013). For example, the calculator shows that $100 in 2005=$119.64 in 2011. Inflation rates were applied as the last stage of the conversion process.
Where baseline data were published in £stg prior to 2011, these were revised to account for changes in inflation in the United Kingdom using the Bank of England calculator (Bank of England 2013) and then converted to A$ using the OECDs PPP conversion rate for 2011 (OECD 2013).
Exclusions from the current estimates
Some costs have not been included in the estimates presented in this report. These include the social costs of fear of crime, costs of supporting offenders and their families, local government crime prevention activity, community defensive action, ‘second-generation’ costs of offending (including the costs of victims of crime committing crimes in the future and the costs to the families of offenders through disruption, guilt and dysfunction) and damage to the reputation of victims and offenders (in the case of a financial crime; Gilligan 2007) and costs associated with disinvestment in high-crime areas, which can be substantial. The nation-wide lost productivity of those individuals committing crimes has not been costed and included in estimates due to lack of available data.
Notes when reading this report
Table totals may not add to sub-components due to rounding.
Medical costs have not been estimated for the categories of burglary, thefts of motor vehicles, thefts from motor vehicles, shop theft, other theft, criminal damage, arson or fraud due to lack of available data.
Intangible losses have not been estimated for shop theft, arson, or drug abuse due to lack of available data.
All dollar values reported have been adjusted to 31 December 2011 AU$.