Media outlets are reporting that homicides are skyrocketing in the largest American cities, based on data from the Brennan Center for Justice. Whether that is a long-term trend, a blip in the data or the effect of just a few cities having a bad year is unknown.

Is the trend toward a more violent America real? We simply don’t know. The most recent data from the FBI, which collects information from the nation's 18,000 law enforcement agencies, is from June 2015. A complete reporting of 2015 is not expected until the fall. This year's data will not be compiled and reported until the middle of 2017, and not be final until the end of the year.

But the early reports for 2015 show big increases in violence (up 1.7 percent) and even bigger increases in homicide (up 6.2 percent). And, those increases are biggest in the biggest cities as homicides are up 8.2 percent in American cities with more than 1 million residents. These biggest cities are usually harbingers for broader trends.

The question is: Why? Why would an America in the midst of an economic upturn, with stock values and gross domestic product approaching historical peaks, suddenly become more violent?

Here’s what we know. Violence has declined steadily since 1991. Violence peaked in the early 1980s, peaked again at the end of the decade, steadily declined throughout the 1990s, stayed steady for the most part through the early 2000s, and then plummeted during the great recession beginning in 2007.

Now violence seems to be on its way back up.

Criminologists have cited all kinds of reasons for crime to have declined since the early 1990s. They have cited the removal of lead in gasoline as helping to remove poisons from the atmosphere, leading the next generation of Americans to be less volatile and less violent. They have cited the removal of abortion restrictions as a means of producing fewer criminally minded people. As you might expect, these explanations are widely disputed.

Criminologists also have cited the emergence of evidence-based programs that appear to help the most at risk of offending in the future. Alternatives to incarceration, which are ways to keep people out of prison, such as drug court, help those struggling with drugs and alcohol and mental illness deal with their problems and avoid future criminality. Cure Violence and Operation Cease Fire have demonstrated results reducing gang-involved crime. Community policing has gotten sworn law enforcement officers out of their cars and into neighborhoods, disrupting violence before it happens. Social impact bonds, a new kind of public-private partnership, are funding innovative solutions to intractable problems and scaling evidence-based solutions.

What do all of these innovations have in common? They require substantial public investment.

And that is where the problem seems to exist. Although the private sector has added new jobs for 74 consecutive months, public-sector job growth has languished. And public-sector jobs include sworn law enforcement, corrections personnel, pretrial supervisors and parole agents. All of whom are necessary to implement these new evidence-based innovations and to scale them.

Is the number of sworn law enforcement officers declining? Are there fewer parole and probation officers? It’s hard to know for certain, because reporting is terrible.

But we do know that in 2008, there were “about” 765,000 sworn law enforcement officers in the United States. The same author at the same agency noted that there were about 605,000 sworn officers in 2013. That’s 160,000 fewer officers. In five years.

There are about 25,000 fewer law enforcement officers in 2012 today than in 2009, which means the number of sworn law enforcement per citizen has declined to a level not seen since 1996.

We also know that salaries have declined slightly after increasing quite rapidly, and thus it’s reasonable to assume that officer quality may have declined slightly as well:

There’s no way to know if these numbers are consistent and comparable. But public spending is down overall and the body of evidence suggests that our collective investment in law enforcement has declined. And violence is up.

Correlation or causation? It’s worth looking into.

John Roman is a senior fellow in the Policy Advisory Group and the Justice Policy Center at the Urban Institute.