I.R.S. Criminal Investigations Continue to Decline

The IRS Criminal Investigation division recently released its annual report.

Here’s the Orwellian happy face the Commissioner of the IRS (Charles Rettig) puts on the group:

“CI supports the efforts of compliant taxpayers by visibly demonstrating the risks of noncompliance thereby helping otherwise honest taxpayers stay honest and compliant.”

The report is mostly a glossy rah-rah yearbook featuring group photos of the inevitably “dedicated” workers at its field offices around the country and brief narratives of some of the more celebrated take-downs (e.g. “Members of Fraudulent Jamaican Sweepstakes Ring Sentenced”).

As with most of the rest of the IRS, the Criminal Investigations division has had to do more with less in recent years. Budgets and staff have dropped, while responsibilities (such as Obamacare enforcement, figuring out how bitcoin works, and combatting increasingly organized and international identity theft and refund fraud) have grown.

The report breaks down the numbers of investigations, prosecutions, and convictions resulting from Criminal Investigations work. It’s surprising just how few federal criminal prosecutions for tax matters there are, given the scope of the agency’s bailiwick and the amount of tax law flouting out there. For example, there are probably tens of millions of American taxpayers (individuals and businesses) who simply do not file tax returns at all in any given year. Of those, the criminal investigations division investigated 271, recommended 128 for prosecution, and managed to bring down the hammer of justice on 111 this year.

They don’t even report numbers for people like me who file but refuse to pay. There is such a thing as a failure-to-pay crime, but it’s rarely prosecuted.

Adding up all of the Criminal Investigations activity over the past few years, which includes everything from Bank Secrecy Act violations to Narcotics to Terrorism to Money Laundering to scofflaws like me, you get this story of continuing decline: