Attorney General Eric Holder will announce Monday that the Justice Department will no longer charge nonviolent drug offenders with serious crimes that subject them to long, mandatory minimum sentences in the federal prison system. As my colleague Sari Horwitz explains, Holder "is giving new instructions to federal prosecutors on how they should write their criminal complaints when charging low-level drug offenders, to avoid triggering the mandatory minimum sentences."

He's also expected to call for the expanded use of prison alternatives, such as probation or house arrest, for nonviolent offenders and for lower sentences for elderly inmates. And he'll endorse legislation by  Sens. Dick Durbin (D-Ill.), Pat Leahy (D-Vt.), Mike Lee (R-Utah), and Rand Paul (R-Ky.) that would increase federal judges' flexibility in sentencing nonviolent drug offenders.

The changes Holder wants will likely make a big difference at the federal level. But that won't be enough to solve America's mass incarceration problem.

Go home federal prison system, you're on drugs

Focusing on drug offenses is a smart way to go about reducing the federal incarceration rate. According to data in Why Are So Many Americans in Prison?, a new book by UC - Berkeley's Steven Raphael and UCLA's Michael Stoll, the most serious charge for 51 percent of federal inmates in 2010 was a drug offense. By comparison, homicide was the most serious charge for only 1 percent, and robbery was the most serious charge against 4 percent.

Tougher drug sentencing accounts for much of the increase in the incarceration rate. "If you go back and decompose what caused growth in the federal prison system since 1984, a large chunk can be explained by drug offenses, around 45 percent," Raphael says. The other big category accounting for the federal increase is weapons charges, such as the five-year mandatory minimum faced by drug offenders caught with guns. Raphael estimates that that accounts for 18 to 19 percent of the increase.

There's also been an increase in incarcerations on immigration charges, with the rest of the increase in other areas. But there's no doubt that the biggest category of crime behind the increase in the federal incarceration rate is drugs. Easing up on drug sentencing would make a big dent.

The states are different

But the federal system isn't really where the action is. The most recent Bureau of Justice Statistics (BJS) estimates find that there are 1,353,198 people incarcerated at the state level and 217,815 incarcerated federally. So about 13.9 percent of U.S. prisoners are in federal institutions; the other 86.1 percent are in state facilities. And most prisoners at the state level are not there for drug crimes.

In 2004, about 20 percent of state-level inmates were incarcerated on drug convictions, Raphael and Stoll find. Compared with the federal population, those incarcerated at the state level are much likelier to have committed violent offenses. In 2004, 14 percent were in prison for homicide, 9 percent for rape or sexual assault, 12 percent for robbery and 8 percent for aggravated assault. In 2011, it was much the same, according to BJS stats on state inmates serving sentences of a year or more. Fifty-three percent of inmates were in prison for violent offenses, 18.3 percent for property crimes, 10.6 percent for "public order" offenses such as drunk driving, weapons possession or vice offenses, and 16.8 percent for drug convictions.

Raphael and Stoll's estimates of what's accounting for the higher incarceration rates suggest that violent crimes are a big part of the state-level story. They find that harsher sentencing for violent offenders explains 48 percent of growth in incarceration rates, compared with about 22 percent attributable to increases in drug sentencing, and 15 percent due to increases in property crime and other sentences.

Then again, most people who go through state criminal justice systems do so on drug offenses. If you look at admission rates, rather than incarceration rates, at the state level, drugs become a much bigger part of the picture. For admissions, Raphael and Stoll find "relatively modest increases for violent crimes and property crimes and pronounced increases for drug offenses, parole violations, and other less serious crime." And while higher admissions for less serious crimes with shorter sentences don't affect the incarceration rate as much as increases in sentencing for serious crimes, they do dramatically affect the lives of those admitted, who have to find work as ex-offenders and live with the sundry restrictions states impose upon those who've served time.

It's not hopeless

Holder is taking a fairly plausible approach to reducing the U.S. incarceration rate at the level where he can effect it. But that's not the level that matters most, and if we were to get serious about reducing the state-level incarceration and admissions rates, we need to talk not just about reducing sentences for drug crimes but also about reducing prison admissions for drug offenses, and perhaps also lowering sentences for property crime and even violent offenses, particularly robbery.

There has been growing enthusiasm for reforming state sentencing laws, even backed by many conservatives. The American Legislative Exchange Council has joined the cause, creating model legislation for loosening state mandatory minimum laws. Especially if it's not just limited to drug offenses, that kind of reform could greatly reduce the state incarceration rate.