Fighting tax fraud with big data

John F. Kennedy famously said "It is a paradoxical truth that tax rates are too high today and tax revenues are too low," and with April 15 right around the corner, chances are your tax returns leave you hoping that everyone else is planning on paying their fair share like you.

The difference between what is legally owed and what is actually collected by a government in taxes each year is called the "tax gap." The IRS estimates that figure to be around $385 billion, but others see it being much higher. The University of Wisconsin-Madison for instance, calculated it as nearly $600 billion.

Gunay Mutlu | E+ | Getty Images

However, finding fraud, or even honest mistakes, is difficult because tax codes are complex and many tax evaders are clever. Tax violations are difficult to uncover, especially among business taxpayers, who have many justifiable expenses and exemptions that are harder to track.

Big Data and analytics provide a method to find anomalies, possibly leading to predicting where these anomalies may happen again.

Read MoreU.S. seizing tax refunds of children over parents' debt?!

Even experienced tax investigators only look at a few hundred tax cases a year. Sometimes they pick the cases based on memories of past frauds in a similar industry. Occasionally they get lucky, and a disgruntled employee or a divorcing spouse turns whistleblower.

Instead of auditing more returns, government agencies can collate the data they already collect from various departments. Then they can let computers pick out the filings that are most likely to be fraudulent and let human agents focus in on that subset. It's like going from looking for a needle in a haystack, to looking for a big needle in a pile of smaller needles.

Fraud used in the past often turn up in new returns that would usually fly under the radar. By analyzing past cases in which tax fraud was already discovered, tax authorities can then find cases with similar characteristics. Numerous tax agencies around the world are already using analytics successfully on a state, local and even federal levels.

Read MoreTurns out, the really rich don't mind paying high taxes

One Middle-Eastern nation used predictive analytics to examine 10 years worth of existing tax-fraud cases. Using that data, investigators built a system that is able spot new cases without doing time-intensive random searches. Among counterintuitive indicators of fraud that were isolated were businesses that pay outstanding taxes in certain months and businesses involved in trade with southern European counterparts specifically. That country expects to boost recoveries from fraud investigations by 25 percent.

Predictive analytics become even more useful if they look at returns when they come in rather than waiting to analyze them months or years later. A northeastern U.S. tax department reduced refund fraud by using predictive analytics to spot dubious filings before the refund checks were sent out. Getting money back after it has been paid is laborious and acrimonious. The system has helped the state avoid more than $1.6 billion in refunds since 2004.

These are real savings and a real payback into the tax system. Predicting, preventing and thus discouraging fraud boosts tax revenue without blindly raising taxes on law-abiding citizens to compensate for uncollected revenue.

Analytics are already revolutionizing business efficiency across industries from health care to education and retail; with a great foundation in place, it's time for government to start capitalizing on, or expand existing use of, these technologies for tax-policy enforcement.

Read MoreFamous American faces slapped by the taxman

While Ben Franklin did define certainty as "only death and taxes," predictive analytics can help boost the certainty with which tax agencies operate and make sure everyone pays their fair share come April 15.

Curtis Clark is the global director for government at IBM. Prior to joining IBM in 1997, Clark spent more than 17 years with the state of North Carolina in several roles, including deputy state controller for information- resource management.

Watch: Big Data Download: Tax friendly...and unfriendly states