Australia, like the United States, makes it hard for people who get benefits to stay on them.

It’s not simply a matter of withdrawing benefits as people get jobs and work more hours – something Australia’s does more severely than most of the rest of the world – it’s also a matter of imposing onerous conditions on people who want to keep them.

Centrelink generally requires evidence of looking for 20 jobs per month in order to keep receiving Newstart, a demanding requirement the Abbott government tried to double to 40 jobs per month in 2014.

In our new book, Food Stamps and the Working Poor, Professors Peter Mueser, Erdal Tekin, and I examine the impact of similar requirements in the United States, taking advantage of decisions by some US counties to abandon them.

Foodstamps are like Newstart

Unlike Australia, which provides most of its assistance in the form of cash, the United States offers it through a bewildering set of in-kind programs for food (SNAP; the Special Supplemental Nutrition Program for Women, Infants, and Children; and different school meal programs), housing (the Housing Choice Voucher program), utilities (the Low-Income Home Energy Assistance Program), health care (Medicaid), and other services.

In 1996 the United States introduced strict work-readiness requirements and time limits into SNAP for able-bodied adults without dependents (ABAWDs).

ABAWDs had to work or participate in work-readiness activities at least 30 hours per week. If they didn’t, they were only eligible for three months of SNAP in any three-year period.

Read more: Are most people on the Newstart unemployment benefit for a short or long time?

But almost immediately the requirements were relaxed. Individual states were allowed to request waivers for people living in economically distressed areas. As well, all states got the option of exempting up to 15 percent of recipients from the ABAWD restrictions.

The US changed requirements country by county

Our book examines SNAP participation in three states that obtained waivers on a county by county basis.

This provides a natural experiment to test the impact of work requirements, enabling us to compare program outcomes across counties and time periods when the requirements were and were not in effect.

Our findings are striking, at first glace appearing to indicate that tough requirements work.

In two states, SNAP spells for non-elderly, childless people who were subject to the requirements were about 30% shorter than spells for similar people who were not subject to them. In the other state, they were 7% shorter.

We did not see similar differences for elderly people or for parents who would not have been subject to the requirements in any event, suggesting that the differences in the time spent on SNAP are due to the requirements rather than to other conditions applying to the counties at those times.

Almost all the changes in program participation occured in the first few months of being on SNAP, before the the three-month time limit kicked in.

Read more: It's Newstart pay rise day. You're in line for 24 cents, which is peanuts

It looked like successs – tough requirements had got people off benefits – until we examined what happened to those people who stopped receiving benefits.

In a separate study we examined whether the people who left SNAP because of the work requirements found jobs. We found that only half did. Half did not, meaning that many people were forced off benefits without having an independent means to provide for themselves.

Implications for Australia

Several Australian assistance programs, including Newstart, Youth Allowance for job seekers and Parenting Payment single for parents of school-age children, have mutual requirements like those in the US.

These include entering into job plans, actively searching for work and interacting with service providers. People who fail to meet the requirements risk losing some or all of their benefits.

Over time, the proportion of people covered by these requirements has grown.

Sometimes it has happened directly, such as the recent widening in the age range of people subject to the rules.

Sometimes it has happened indirectly as assistance programs without requirements have been terminated and replaced with Newstart.

Read more: Robo-debt class action could deliver justice for tens of thousands of Australians instead of mere hundreds

It might be happening by using the prospect of so-called robo-debt audits to frighten people away from claiming benefits, and it might soon be done by imposing indignities such as drug tests on Newstart recipients.

If Australia is like the US, these measures will get far more people off benefits than they get into work.

We don’t know what will happen to them. One of the downsides of studying administrative records is that when people are no longer included in the records they vanish from sight.

But it won’t mean they will actually disappear. Getting people off benefits shouldn’t be an end in itself, unless we really don’t care what will happen to them.

David C. Ribar receives funding from the Australian Research Council, the Department of Social Services, Commonwealth Bank of Australia, and ASB Bank Limited. His book on Food Stamps and the Working Poor was funded by the W.E. Upjohn Institute for Employment Research.