Detecting Tax Dodgers With Trade Data
Their dollar claim may be a bit hyperbolic. While their program flags unusual transactions, they are not necessarily fraudulent. Nevertheless, the professors are on to something. Michael Mazerov, research director for the Multistate Tax Commission, a Washington, D.C.-based group representing the interests of 35 states in interstate and multinational business tax issues, stops short of endorsing the $40 billion claim, but agrees that billions of tax dollars are lost due to mispricing of trade goods: "We think it's a very large amount of revenue."
Zdanowicz and Pak buy the U.S. Department of Commerce's database of all imports and exports on CD-ROM every month. Their computer calculates average prices for these products and flags prices that deviate from the average by 50% or more. For example, a company sells a motorcycle to a foreign affiliate at its $4,000 cost; the affiliate then sells the cycle at $8,000, earning a 100% profit margin. By keeping its profits outside the U.S., the company avoids paying U.S. taxes and may take advantage of inefficient tax collections or lower rates in the foreign country. By applying the U.S. corporate tax rate of 34% to suspicious-looking transactions, Zdanowicz and Pak estimate the federal tax loss at $39.5 billion for 1994.
The program has limits. The government deletes company names from the databases it sells to the public, so FIU can't break out suspicious-looking transactions by related companies. Zdanowicz has no doubt, however, that many of the prices are simply fraudulent.
"Our view of the potential use of this program is auditing," he says. "We think we are finding smoking guns." He says the IRS currently can't find these "smoking guns" amid the millions of international business transactions taking place every year. Mazerov agrees. He says flagging transaction prices 50% above or below the norm should be extremely useful to the IRS.
But the IRS hasn't evinced much interest in the program, saying the trade mispricing costs the government only about $3 billion annually. Says Mazerov: "Our feeling about it is that it ought to be a tool that the IRS and Treasury Department look at very seriously. That is not to say that it could not be refined, but the limitations are in the data," not in the program.