Streamlining the Check Payment Process

Despite the barrage of bank ads touting online bill paying and the ease of using debit cards, writing checks the old-fashioned way is still a popular method of parting with one’s money. In the United States alone, about 50 billion paper checks are processed each year at a cost of $1 to $5 per check, according to the Federal Reserve Bank. Amar Gupta, for one, hopes to wring some of those processing costs out of the system.

As codirector of the Productivity from Information Technology initiative at MIT’s Sloan School of Management in Cambridge, Mass., Gupta has been researching ways to streamline the check payment process using IT. A few years ago, he and his team of researchers devised a technology to accurately scan and read handwritten characters. The WinBank Optical Character Recognition System relies on character recognition algorithms and neural networks to read handwritten numerals on checks.

While the goal of the WinBank system sounds straightforward, the vagaries of human handwriting complicate accurate recognition. Gupta and his team had to account for different spaces between characters, overlapping characters, distinctions between commas and decimal points, and handwritten numbers like three and eight that often resemble each other, just to name a few of the myriad considerations.

By automating check reading, the WinBank system will facilitate and expedite the electronic processing of checks, thereby eliminating much of the human intervention the check clearing process now requires. As a result, says Gupta, the process will be faster and less expensive, even if paper checks remain a staple of the financial system.

However, Gupta acknowledges that numerous obstacles remain to widespread, international adoption of the WinBank system. For one thing, the speed of electronic processing will eliminate the ability to write floating checks (a check written just before the funds are available to cover it). Also, the worldwide financial system has no regulatory or operational standards, Gupta says, which further complicates widespread adoption.