Harsher prison sentences have meant Australia is jailing more people than at any time in the last two decades — while rates for most violent crimes continue to fall. What’s the connection?

Corrective services data recently released by the Australian Bureau of Statistics shows the number of people in prison on an average day in Australia was over 36,000 in June this year — the highest number since the ABS began collecting imprisonment data in 1998, and the highest imprisonment rate ever as well: 196.5 Australians per 100,000 adult population.

Just four years ago the imprisonment rate was below 170 per 100,000, the level around which it had hovered after a big rise from the late 1990s. In 1998 it was 138, in 2002 it was 144, by 2006 nearly 160.

Imprisonment rates among indigenous people haven’t risen at the same rate, but recent falls have been reversed and the 2015 level of 2257.3 people per 100,000 is the third-highest of recent years. Indigenous Australians are jailed at more than 10 times the rate of non-indigenous Australians.

While imprisonment rates have gone up, recent crime trends have mostly been down. The ABS’ most recent crime data for the whole of Australia — based on victim reports, independent of police, in the national Crime Victimisation Survey — shows that victimisation rates for physical assault have fallen significantly, from 3.1% of the population in 2008-09 to 2.3% in 2013-14, and threatened assaults have fallen from 3.9% to 2.7%. Recorded crime data from the ABS (i.e. based on police reports) shows 2014 recorded a five-year low in homicide rates and attempted murder rates and numerical lows — that is, not adjusted for population growth — for robbery.

This is backed up by data from New South Wales. The NSW Bureau of Crime Statistics and Research data shows homicide numbers stable over the last decade to 2014 despite rising population; non-domestic violence-related assault has fallen by 25% in numerical terms, and robberies of all kinds have more than halved. Victoria is a little different: robbery has fallen dramatically in the last five years; the homicide rate is stable; and assaults have fallen, but only in the last 12 months. In Queensland, assaults have fallen significantly in the last four years, as have robberies and break-and-enter crimes, while homicide has remained level.

However, reports of sexual assault, domestic violence and drug offences have all increased significantly. The incidence of sexual assault in victim reports (that is, not police reports) has remained steady, according to the ABS; NSW, Victoria and Queensland figures all show a big rise in reports to police of sexual assault, which may suggest higher reporting rates.

In Queensland, the reported incidence of sexual assault and attempted sexual assault has increased every year for the last four years and is now 25% higher than in 2011. The reported incidence of sexual offences, adjusted for population, in Victoria has increased every year for the last five years, and in NSW the reported incidence of sexual assault has steadily risen until the last 12 months.

Domestic violence assaults have also increased in NSW, though not by as much, and they, too, fell slightly in 2014-15. In Victoria, the number of violent crimes flagged as “family-related incident” have more than doubled since 2010-11, including of assaults, sexual assaults and abductions. Domestic violence assaults aren’t broken out for Queensland, but the rate at which domestic violence protection orders are breached has risen over 50%. And the rate of drug offences (adjusted for population) in Victoria has increased nearly 75% in the last five years; the comparable rate has nearly doubled in Queensland; in NSW, the number of incidents relating to dealing amphetamines has tripled, and cocaine has doubled.

What’s the link between the fall in some crimes and rising prison populations? The traditional assumption, particularly by politicians, is that the more criminals you lock up, the safer the streets are. But what complicates the issue is that there are different reasons why the prison population might be increasing: more offenders might be refused bail (the rise in the imprisonment rate has been among both unsentenced and sentenced prisoners.); more might be being arrested, more might be being imprisoned if convicted, or they might be serving longer sentences.

A NSW study, for example, found that between 1994 and 2013, the number of defendants refused bail across nearly all categories of crime rose significantly: less than a quarter of defendants were refused bail for assault-related offences in the mid-1990s but in 2013 nearly half were; sexual assault offences saw a rise from 12% of defendants refused bail to around 30%. And the average length of sentences rose for nearly all types of offence. Sentences for murder rose by 25% in the period; assault-related offences by 75%, firearms offences by 30%. Sexual assault sentences rose by 11% (though in 2013 those convicted of sexual assault were serving less than three years); drug offences recorded only a 9% increase, there were 9% fewer robberies. Local courts, which deal with more minor offences attracting shorter sentences, have also been handing down longer sentences for some offences, like sexual assault.

But not merely had sentence lengths risen, the probability an offender would be jailed at all significantly increased. The study found, for example, that offenders were 25% more likely to be imprisoned for assault, 12% more likely to be jailed for sexual assault, 22% more likely to be jailed for theft.

Victorian data shows sentences increasing in that state; while the proportion of prisoners with sentences of 10 years or more has remained around the same from 2004 to 2014, the proportion of prisoners serving less than a year fell from over a third to just over a quarter, while those serving one to five years has increased from 45% to 50%. The number of those serving between five and 10 years has increased to 14%. There have been complaints for a number of years that harsher sentences are filling Victorian prisons.

The relationship between higher incarceration levels and lower crime is a vexed issue in the United States, where drug laws and anti-violent crime measures caused the US incarceration rate for males to soar from just over 200 per 100,000 population for the first three-quarters of the 20th century to over 500 in the 1980s, over 900 in the 1990s and peaking at around 1000 just a few years ago, before declining to around 750 in 2013. The primary targets of this surge in incarceration have been Hispanic and African-American males: throughout the 2000s, the imprisonment rate for African-American men was just under 5000 per 100,000 population, that for Hispanic males, around 1800.

In the same period, violent crime has dramatically reduced in the United States. Studies suggest that a rise in incarceration does reduce violent crime by locking up dangerous offenders for longer — some US criminologists suggest there’s a very strong link between longer sentences and lower crime — but diminishing returns set in at some point, after which locking up people for longer has minimal impact on crime and simply wastes taxpayer money (especially given higher incarceration may not be the lowest-cost means of reducing crime anyway). Other factors like economic growth, more heavy-handed policing and more unusual theories like the legalisation of abortion in the 1970s, and the removal of lead from gasoline have all been identified as other possible factors in the dramatic drop in violence in the US.

But one of the few Australian studies to tackle the issue found that it is the likelihood of arrest, and the likelihood of being jailed if convicted, that has the greatest impact on both violent and non-violent crime rates, not sentence length. The 2012 study, carried out by the NSW Bureau of Crime Statistics and Research, concluded that:

“The arrest rate and imprisonment rate [in that study, the number of offenders receiving prison sentences] each have significant negative relationships with both property and violent crime, over both the short and long run. Variations in the average sentence impose no short-run or long-run effects on property crime or violent crime … In terms of long-run (equilibrium) effects, a 1 per cent increase in the arrest rate produces a .135 per cent reduction in property crime and a .297 per cent reduction in violent crime. By comparison, a 1 per cent increase in imprisonment rate produces a .115 per cent reduction in property crime and a .170 per cent reduction in violent crime.”

This appears to be backed by a study by the Australian Institute of Criminology’s Rick Brown earlier this year on the decline in property offences, in which offenders most commonly identified improved security, greater risk of detection through better policing and greater risk of imprisonment as deterrents to property crime.

The question for Australian policymakers — to the extent that rational policymaking is possible in this area (behold the fate of Greg Smith, the NSW Liberal Attorney-General who tried to apply logic and evidence to the NSW criminal justice system) — is whether the point of diminished returns might have been reached with longer sentences, and the focus should be on better policing, and jailing more offenders who would otherwise receive lesser punishments, particularly in areas like sexual assault and domestic violence where crime rates have remained stubbornly high, while other forms of violent crime have been falling.