In partnership with

An analysis of mobile phone theft

September 11, 2014 Simon Ruda

Last Sunday, The Behavioural Insights Team and the Home Office published the report “Reducing Mobile Phone Theft and Improving Security”.

The report provides detailed evidence of which mobile phones are most likely to be stolen, where they are likely to be taken, and who the most typical victims are likely to be. In addition to this, we have been able to quantify, for the first time, the positive impact on thefts of security functions introduced by manufacturers.

Using detailed data from the Metropolitan Police of all thefts of mobile phones between August 2012 and January 2014, we constructed a mobile phone risk of theft ratio to show which mobile phones are most likely to be stolen. The ratio works by calculating an average risk of theft across all handsets, and then calculating which handsets suffered from excess risk of theft over and above the average. If a handset is significantly above the average, it suggests that it is being targeted by thieves for theft due to its value on the black market. Black market value can be determined by several factors, including the ease with which the phone can be re-programmed, how easy it is to access the personal data contained within it, or simply its desirability to consumers.

Designing out crime

Across different industries, we have seen that some of the largest reductions in crime have been due to manufacturers’ ability to improve the inbuilt security features of their products, also known as ‘designing out crime’. Vehicle theft is a useful case study in just how well this approach can work. Between 1993 and 2012, the rate of vehicle theft fell by around 87% in the UK. A significant proportion of this fall can be attributed to a set of technological improvements – particularly central-locking and immobilisation systems. Outside vehicle theft, this design-focused approach to crime prevention is generating innovative solutions to areas such as bicycle and bag theft.

We are beginning to see the mobile phone industry take security more seriously. We hope that, by revealing to consumers the degree to which certain phones are at greater risk of theft, and by quantifying the impact of security features as they are introduced, we can encourage industry to give a higher priority to the safety of their customers.

The risk of theft ratio

The findings are clear: models such as the Apple iPhone 5 were most at risk during the period of the analysis.

The report’s findings also show that mobile phone manufacturers can make a difference. We analysed data around the launch of iOS7 using a technique known as regression discontinuity (RD). We found that although phone thefts increase slightly at first, as is common around the launch of a new phone (iOS7 was launched at the same time as the iPhone 5C and 5S), they then begin to decline – a pattern that is not matched in any other model of phone around this time. The fact that we don’t see a similar pattern in the thefts of other phones allows us to attribute this fall in thefts (which is statistically significant) to the improved security measures built into iOS7, which discourage thieves.

How we constructed the index

The ratio compares a particular handset’s share of all “targeted” thefts (i.e. where the handset was plausibly targeted, such as a phone being snatched from someone’s hand) by the estimated “availability” of that phone (i.e. that handset’s share of all those in active use in London during that time).

It’s important to note that availability is not the same as market share, which only tells us which phones come into use (or become available for theft) each year, and not which phones cease to be used (or become unavailable for theft). Only mobile phone providers have access to precise information on which phones are active over a period of time, and they were unwilling to share it with either us or Government. Therefore, we needed to accurately estimate the proportion of each handset in London in use. We did this by measuring the proportion of handsets in the dataset whose theft could not plausibly have been targeted (e.g. a stolen handbag containing a phone, which the thief could not have known). As these datasets are so large, we are confident that they provide an accurate representative sample for London.

Where next

The current version of the Theft Ratio is intended as the first step in an iterative process. New versions will keep consumers updated about new models and the impact of security enhancements as they are developed. We also look forward to additional engagement from industry. Increased industry cooperation will open up sources of more accurate and granular data, which in turn can help consumers stay better informed.

Smartphone technology has improved so much in recent years that mobile phones today are almost unrecognisable from their predecessors of just a few years before. The ingenuity of designers and manufacturers has transformed their appearance, power and usefulness. We hope that transparency about theft risk can encourage that same ingenuity to be used in transforming their security as well.

Subscribe to our mailing list