Chris Yiu: We need new ways of solving the familiar problem of the revenue tax gap

Tuesday, 03 July 2012

Chris Yiu: We need new ways of solving the familiar problem of the revenue tax gap

Synopsis

The tax gap – the difference between the tax people owe to HM Revenue & Customs and the amount they actually pay – stands at an astonishing £35 billion a year. Getting back just a quarter of this amount would pay for a 2p cut in the basic rate of income tax. A prize this big is attractive to politicians on all sides. The tax gap is as big now, however, as it was back in 2004, seemingly stuck at around 8 per cent of total receipts.

The government knows something needs to be done. In 2010, Ministers announced that HMRC would re-invest nearly £1 billion of the savings from its spending review settlement in initiatives to tackle non-compliance in the tax system. Part of this is earmarked for investment in CONNECT, the department's tool for cross-referencing data on individuals and businesses. This exists to help officials spot and target the most risky cases of fraud, error and criminal activity.

This is undoubtedly a step in the right direction. But the snowballing fiasco around tax avoidance – moral or otherwise – justifies a significantly more aggressive approach.
The trouble is that spotting the misbehaviour of a small minority in a large population is difficult, and deciding where best to focus scarce compliance resources is harder still. Broad-brush estimates of the tax gap hide a very messy reality: evasion, avoidance, the hidden economy, legal disputes, non-payment, general non-compliance, errors and criminal attacks are all in play. Nor is this just about bankers, CEOs, celebrities and sports stars. About half the tax gap is attributed to small businesses, a quarter to large businesses, and the remainder to criminals and individuals.

Data and analytics will play a big part in finding a solution. The kind of performance we should be demanding goes beyond an incremental improvement in existing government capabilities. To see what's possible we need to look to industry, where businesses are way ahead on using data and analytics to protect revenues.

Take, for example, the world of credit and debit cards – where last year in the UK alone, customers made over 9 billion purchases worth a total of over £450 billion.

Each time a customer uses a card to make a purchase, an algorithm working for the card issuer makes a split-second decision about whether to authorise the transaction. The best of these crunch a massive amount of information – from the obvious (is the customer over their credit limit, are they abroad?) to the incredibly subtle (does a purchase of this size, in this store, on this day, look unusual compared to everything we have seen the customer do before and everything that all of our other customers are doing?). Of course these systems sometimes produce false positives, which can be frustrating for all involved. On the upside, they have driven losses from fraud down to a fraction of one per cent, which is of huge benefit for everyone interested in making or accepting legitimate card payments.

Back at HMRC, the CONNECT system already aggregates a number of internal and external data sources, and is due upgrades to speed up the supply of key data streams and to access unstructured, text-based assets. Progress could be further enhanced and accelerated by additional investment in cutting-edge analytics technology. There will always be a need for seasoned experts and judgment – but people are expensive, and there are only so many cases a human being can look at each day. The better we can get at using machine learning to spot and close down complex avoidance, the less we will have to look for savings elsewhere. And in time, getting to a place where people contemplating artificial tax dodges are virtually certain they will get caught will, finally, provide a powerful deterrent.

Making inroads into the tax gap is just one area of public policy where data and analytics have transformational potential. In the research we’ve published today we explore some of the other ways that data can be applied to increase public sector efficiency and enhance public service delivery. Collecting the right amount of tax may be just the beginning of a journey to make government faster, smarter and more personal.

Related Staff