Employee surveillance is on the rise, with banks looking for innovative new ways to detect and even predict bad behaviour. But how useful is it really?
As banks move towards building a more integrated surveillance model, employee surveillance is increasingly viewed as a vital part of the solution. Industry anecdotes range from monitoring employee working hours and holiday take-up to surveillance of their social media accounts and even culture scoring based on their perceived buy-in to audits.
Behavioural science could yet open a whole new frontier in employee surveillance – and potentially help to encourage good behaviour rather than simply rooting out bad – although practical applications are scant.
But until banks have the tech in place to triangulate the potentially vast amounts of data that such surveillance produces, and have gathered enough historical patterns to demonstrate correlations between specific behaviours and incidents of market abuse or poor conduct, can employee surveillance really hold the key to identifying critical risk indicators missing from trade and communications coverage?
Behavioural science has attracted the attention of many banks and could yet play a real role in employee surveillance, senior surveillance practitioners believe.
Research has suggested, for example, that failure to meet continuing mandatory training requirements or changes in working hours could be linked with an employee committing a market abuse violation or stealing from the firm. Other potential uses include patterning alerts that get closed out for an individual to see if it creates a heat map, some say. Machine learning in particular holds promise for trader profiling, with the ability to view numerous dimensions at once, helping to spot where changes in a trader’s behaviour over time coincide with potential manipulative practices, one surveillance chief notes.
With 25% of those polled in 1LoD’s Surveillance Benchmarking Survey saying they planned to make significant investments in employee surveillance over the next 12 months – and nearly 70% saying they anticipated some – the activity is clearly becoming higher priority for many banks. Another 25% cited behavioural science as a skillset that their function would have significant need for, while 63% said they would need some requirements, suggesting the ability to understand and anticipate human behaviour is becoming more central to employee surveillance.
There is a still a lot of cross-over between employee surveillance and communications, with banks exploring, for example, metadata around voice and text communications – such as who employees contact, how often, and when – to detect potential instances of sharing confidential information.
Not so fast
Banks however have a big to-do list before they are ready to apply behavioural science to their employee surveillance in any meaningful way, including pulling together fragmented data pools, improving the efficiency of comms and trade surveillance programmes, limiting false positives and ensuring the right skillsets are in place, some practitioners say.
“Behavioural analytics heavily depends on investment in technology and data to be effective and, as with the elusive concept of holistic surveillance, I think there are several hurdles to cross before we begin to see results,” says Steven Schluter, head of compliance monitoring strategy at Morgan Stanley.
“With various types of monitoring being conducted in silos, there typically has been little consideration given to how the output of the given control can be used by different control functions down the line.”
Terri Duhon, chair of risk at a number of financial institutions, questions whether employee surveillance is even an efficient use of a bank’s compliance resources, at least until the science is proven. “What kind of indicators or behaviour patterns outside the organisation are we looking for in individuals, that are going to tell us they’re going to exhibit poor conduct at some point?” she asks. “We have to tread very carefully here. And have we actually made the industry safer? Because remember, that’s the end goal. I wonder if we’ll be generating too many false positives if we’re not thoughtful enough.”
Combining network analysis with semantic modelling – using ontologies to model data in triplet form to enable more effective searches – could be a better way in the short term for banks to gain a broader view of their employees and potential risks, Schluter believes.
Ultimately, where behavioural science may offer more value for banks is in trying to improve employee behaviour, rather than simply detecting bad behaviour. Although Schluter says he hasn’t yet seen firms investing in behavioural science to mitigate regulatory risks, he believes taking a more employee-centric rather than rules-based approach to compliance could help instil a better conduct culture within the banking hierarchy.