Big homeless progress

Apparently there's been a big yet unheralded decline in homelessness in the US, despite the economy, and both Bush and Obama can take credit: the former for a housing-first policy that aimed at providing permanent housing before treatment, the latter for the stimulus. Also, multiple studies show that between arrests and ER visits, homeless people cost $30-45,000 a year, when it'd be $10-16,000 to house them with case worker support.

http://www.theatlantic.com/business/archive/2013/08/the-astonishing-decline-of-homelessness-in-america/279050/
http://www.nytimes.com/2008/07/30/us/30homelessweb.html?_r=0
http://thinkprogress.org/economy/2014/03/24/3418140/charlotte-homeless-study/
Residents of Moore Place collectively visited the emergency room, an
expensive but not uncommon way homeless people access health care, 447
fewer times in the year after getting housing, the study discovered.
Similarly, they spent far less time running afoul of the law, with the
number of arrests dropping 78 percent.
http://thinkprogress.org/economy/2014/05/27/3441772/florida-homeless-financial-study/
http://thinkprogress.org/economy/2013/09/05/2579451/colorado-homeless-shelter/
http://thinkprogress.org/economy/2014/02/05/3228801/osceola-county-homelessness-criminalization/
An average permanent supportive housing unit in Osceola County costs
$9,602 per year, which includes $8,244 for rent and utility subsidies
and $1,358 for a case manager (with a case load of 30 clients). In other
words, each supported housing unit costs the county 40 percent less than
what they’re currently paying to put homeless residents in jail.

Power law problems when instead of a bell curve of normality, you have a
few outliers who are most of your problem -- a few really bad LAPD cops,
a few chronically homeless people who cost tens if not hundreds of
thousands of dollars, a few highly smoggy cars. managing the middle
doesn't help: most of the cops don't need mild training, and it doesn't
help the hard cases; you need to just get rid of them. It's cheaper to
simply house the hardcase homeless people and give them case workers.
Annual smog tests are mostly not needed or cheatable, vs. on-road
testing.
http://archives.newyorker.com/?i=2006-02-13#folio=096

Chicago police corruption and homicide; prison reform

It's usually said that homicides are the trustworthy crime statistic, since it's hard to hide dead bodies. But the Chicago police department has apparently been cooking the books under Garry McCarthy and Mayor Rahm Emmanuel, reclassifying homicides as "death investigations" and other tricks.
http://www.vox.com/2014/6/29/5850100/chicago-brought-its-murder-rate-down-by-not-counting-murders
http://www.chicagomag.com/Chicago-Magazine/May-2014/Chicago-crime-rates/

The vox entry links to a longish (not as long as the Chicago magazine one) article by Yglesias in 2003, on the cost of prison. Link and some quotes:

http://prospect.org/article/research-wars
The trouble with prison isn't that it doesn't work; the trouble is that it doesn't work very well but does cost a fortune compared with other ways of reducing crime. Feeding, clothing, housing and guarding a convict for a year costs more than $20,000.
Post-incarceration, shorter sentences can be combined with properly supervised parole programs that replicate the crime-reduction effects of imprisonment at a fraction of the cost. In Texas (Texas!), a system of graduated sanctions for minor parole violations is credited by officials with an 8,000-person reduction in the state's prison population.
The drug business is a business like any other -- if you eliminate a salesman without eliminating the demand, the salesman's boss is just going to hire someone else. Drug treatment, by contrast, actually works because a reformed drug user isn't automatically replaced with a new addict, and treatment programs aimed at consumption reduction are seven times cheaper than prison.
laws like California's famous "three strikes and you're out" rule go "beyond the point of diminishing returns" because even "career criminals have a period of peak performance." Robbery, he explains, is "a crime of the young," with incidence dropping dramatically in the mid-20s and falling to almost nothing as people move through their 30s.
Similarly, doubling sentencing length doubles (or more) corrections expenditures without doubling the deterrent effect on potential offenders. Simply putting a larger proportion of the people who get arrested behind bars is subject to diminishing returns as well, because as long as prosecutors and judges are minimally competent, they'll have made sure that the worst criminals are already locked up. Whether you look at deterrence or incapacitation, beyond a certain point prison stops being cost-effective.
Zimring went so far as to suggest that even a "prison training program to teach robbers how to burglarize unoccupied dwellings" would work better than more prisons as a method of reducing violent crime. That's far-fetched, of course, but it illustrates a larger point: Giving muggers an alternative to mugging is the best way to get them to stop. Even Heritage's Mulhausen concedes that "there's some research that shows that vocational training helps reduce recidivism."
The notion that jobs rather than jails hold the real key to crime reduction isn't just a bleeding-heart liberal fantasy -- it's supported by sound social-science research.
RAND tried that theory out, conducting a study in which students got money as an incentive for staying in school. This, indeed, caused graduation rates to rise, and Greenwood calculates that 250 serious crimes could be averted for every $1 million spent on such incentives -- far more bang for your buck than the prison system offers. [$4000 per crime prevented]
Ten offenders could be subjected to a tough parole regime for the price of putting one man behind bars, and though testing and sanctions sounds harsh compared with freedom, it looks pretty good compared with prison.