Artboard 3 Created with Sketch.
How Data Analytics Can Help Modi Government Catch The Demons Of DeMo

The government will have heaps of complex transaction data from the post-demonetisation months.

Here’s how it can leverage data analytics capabilities to track down those who may have tried to game the banking system.

When Prime Minister Narendra Modi made the demonetisation announcement on the night of 8 November, it started an administrative chain reaction focused on squeezing money launderers, promoting less-cash economy and of course remonetising the banking system. As the all important 30 December cutoff for depositing demonetised currency notes ended, the focus has shifted to the next logical steps. The government and the Reserve Bank of India (RBI) will now have to work on computing tax liability for those who declared their cash and deposited their holdings in the banks and on busting money launderers who tried to game the constantly changing and readjusting demonetisation rules in November and December.

The maturing of financial services technology over the last few years has also coincided with increasing global threats of crime and terrorism, which need movement of money. Over the last couple of decades, global regulators have been trying to keep up with money trails which fund global crimes, tax evasion through tax havens, and the routine day to day conversion of unaccounted wealth to legitimate assets which happens in every economy to varying degree.

Money laundering as a concept is hardly alien to the banking systems worldwide. Peter Reuter in his 2004 book Chasing Dirty Money explained the three step process of money laundering: Placement – putting the illicit wealth in the financial system, layering – creating a maze of financial transactions to create an illusion of legitimacy to their placed wealth, and integration – using these transactions to legalise the illicit wealth.

India has its own set of rules in the money laundering space, primarily governed by the 2002 Prevention of Money Laundering Act, which has since been modified a few times. Banks in India have to comply with a host of regulatory provisions, which include know your customer (KYC), filing cash transaction reports (CTRs) for reporting flow of cash in and out of the banking system, and filing suspicious transaction reports (STRs) for any suspect pattern of transaction activity in bank accounts. These provisions try to catch patterns, which correspond to one or more of the three steps Reuter characterised as laundering.

Indian financial system has several players – foreign banks, public sector banks, private banks, cooperative banks, rural banks, credit cooperatives, and the specialised non-banking financial corporations. The technology capability, understanding of operational risk, regulatory appreciation and general understanding of financial fraud varies in a big way across these myriad institutions. Globally, the financial engineers have demonstrated their ability to always outpace financial regulators in the last decade, and India is perhaps no exception. In this operating environment, a sudden jolt with nationwide systemic impact like demonetisation was always going to expose the weaker innards of financial services.

Yet, the government and the RBI are now faced with the task of finding fiscal upside as well as catching potential tax offenders. The key to success may lie not in the hallowed precincts of North Block in Delhi or in the RBI headquarters in Mumbai. The Financial Intelligence Unit (FIU) of the Ministry of Finance (MOF) and the RBI may have to work jointly with the several data analytics companies spread in various information technology (IT) cities to attain their dual objectives through 2017.

The FIU and the RBI will be saddled with heaps of complex transaction data from the post-demonetisation months. The data volumes for November and December 2016 will be obviously much higher than usual. And buried deep inside these datasets will be the answers government is seeking. This is where data analytics capabilities like pattern matching, linking accounts to individuals, creating 360 degree views of customers, analysing long time series data, and linking transactions to tax returns will come in handy. Thankfully for the government, these skills are aplenty in the Indian technology industry, apart from the homegrown capability with the FIU and the RBI.

Data analytics can come in handy in various ways for the government beyond scrutinising the CTRs and STRs filed by the banks. Some of the areas to dive deeper into are as follows.

First, match cash transactions to account profiles: This is a low hanging fruit where the bank data can directly throw red flags. Checking Pradhanmantri Jan Dhan Yojana accounts, their funding dates, and deposit and withdrawal patterns in the last two months should be the first port of call. There is a lot of anecdotal evidence in media around misuse of these accounts by money sharks, either for a fee or just by threatening the account holders.

Second, check for new activity. Analytics can be used to determine the number of new accounts opened against a specific identifier like Permanent Account Number (PAN). There may also be a chance that fraudsters may have opened multiple accounts to deposit small amounts using different government identity cards and not seeded PAN in all of them. Using data with the Income Tax department on passport or Aadhar to link individual IDs to trace such accounts should be explored. Once a linked database is available, the same can then be used to find transactions like depositing small sums of cash in multiple accounts in primary or secondary holder capacity.

Third, analysis of past pre-demonetisation data: It will be critical to interpret cash movement in accounts as well as entire bank branches in November and December against the activity levels in the last few months. The government has already asked for April to October cash movement data from the banks. Bank branches which have suddenly shown manifold rise in cash movement could be liable to be scrutinised in their entirety. This can be achieved by using a combination of algorithms like time series, clustering and regression to pin point fraud.

Fourth, checking remittances: There could be situations where demonetised cash moved out of India through hawala route and corresponding net of fee money transfers made back to Indian accounts against dubious trade papers. Scrutinising remittances, which are complex exercise given the intermediaries involved in any global money movement, will be critical.

Fifth, too many inward electronic payments in an account: Launderers may have enrolled several accomplices trusting them with small sums of money, which can then be transferred from the agents to the principal electronically. If there are bank accounts with unusually high inward RTGS/NEFT payments bunched together in the last few or next few weeks, it should raise a flag.

Sixth, using credit instruments: A potential laundering activity could be to use credit instruments like credit cards with unusually high sums deposited in these accounts to be used later.

Seventh, profiling bank branches: While a prudent banking IT system will have restrictions or clear trails on backdated transactions and adjustments, it is fair to assume that there could be loopholes known to some bank employees or smaller banks may not have good IT systems and controls in place. Scrutinising bank branches with more than usual STR activity or abrupt changes in complexity of STRs will be handy.

Eighth, creating a ‘related account’ view: If multiple accounts (e.g. of relatives or family connections) have made repeated payment transactions (e.g. deposit in one account, then distribute the money across accounts), creating a view of such related accounts will be important. This may be the most common way of structuring illegal money in the banking system. Techniques like geospatial analysis on bank branches in a given PIN code can be employed

Ninth, matching Income Tax data to bank data: This is perhaps the most crucial aspect of demonetisation analytics, to link income patterns to transaction patterns. The bank account data – at least in theory – is already available with the I-T department. The challenge will in unearthing either incorrect income or incorrect bank data and linking the two using PAN information. Any steps taken in this area will also have to ensure that government does not cross the red lines of tax terrorism – a fear which has been routinely expressed by various commentators in media and periodically acknowledged by various government officials themselves.

Tenth, matching trade invoices cash transactions and positions: This is perhaps the most complex and most productive area for tracing illegal wealth. Trade invoices, along with property, gold, art, and cash, form the main source of black money in India. The government will have to balance its promise of not digging too much in the past when it comes to small businesses with the need for busting big offenders. Matching patterns in trade invoices, cash in hand, and related banking transactions especially taking a historical view against the fourth quarter 2016 view can unearth a lot of muck. But it is not clear how much political capital does government want to spend in this area in the immediate future.

This is also a good time for analytics firms to sharpen their market offerings and for the governments to institutionalise technology driven compliance. There can be no better use case for the “Make In India” initiative too. This government is definitely the most technology friendly regime India has ever had. Will it make a push to tap the wide talent base India has to accomplish important political and economic objectives?

Only time will tell, but given the resolve the Prime Minister has shown in pushing the policy boundaries in the last few months, no one should bet against it.