One of the new mantras among policy wonks has been "evidence-based
policymaking" - the notion that policy ideas should stand or fall on
the basis of research and trials, rather than opinion polling and
Overseas, this has led to some startling discoveries. Young driver
education programs, once thought to reduce road deaths, actually turned
out to increase them - by encouraging high school students to drive at a
younger age. Moving to Opportunity, a US program that provided housing
vouchers for poor people to move out of ghettos, dramatically improved the
health of children and their parents. And studies on class sizes have cast
doubt on earlier assertions that across-the-board reductions boost
students' test scores - as a recent report by the Centre
for Independent Studies has shown.
The lesson is that policies, like medical interventions, can be put to
the test - saving millions of taxpayer dollars, and improving the quality
of government. To be effective, evidence-based policymaking relies on
policy trials, which simulate the randomised conditions of a laboratory
experiment, and access to high-quality data. Unfortunately, both are
largely absent in Australia.
As the NSW election campaign has demonstrated, politicians are about as
ready to engage with policy trials as with redheaded fishmongers migrating
in from the north. The parties, it seems, are big on rhetoric but not on
putting their ideas to the test. If the Coalition believes that Parenting
Partnerships will reduce conflict in schools, they should propose a
one-year experiment - randomly implementing them in 100 schools and
reporting on whether schools with partnerships have better test scores and
retention rates. If the Greens believe that dispensing heroin is the way
to go, they should suggest comparing the outcomes of a group of addicts
who are eligible for it with a group who are not. And instead of watching
incarceration rates skyrocket, Labor could trial and test training and
rehabilitation policies in different jails, to see which does best at
helping ex-cons find jobs.
The only example of evidence-based policymaking that we are aware of in
NSW was last year's Drug Court evaluation. Carefully administered, the
research has provided powerful evidence that this court provides a more
cost-effective solution than the traditional judicial system. At least
it's a start.
Why don't we see more randomised trials in Australia? One impediment is
a cultural attitude that government services are an entitlement, and
therefore must not be rationed. Yet it is time that this conventional
wisdom was balanced against the benefits that can flow from careful
pre-testing of government programs.
Even as Australians have started to embrace testing, our institutions
have failed to follow, denying access to data, or imposing hefty fees. By
contrast, the United States statistics bureaus apply a simple rule to
their data: if the public answered the questions, the public have the
right to analyse the data. And these inputs sustain a proliferation of
thinktanks that debate policies based on outcomes, rather than conjecture.
But in Australia, the picture is transformed. The Australian
Bureau of Statistics makes virtually no data of any complexity freely
available. Vast stores of intriguing data are aggregated into bland facts
for publication in the yearbook, rather than released for primary
analysis. When researchers cannot track individual education, health,
crime and labour market experiences, we lose the ability to make subtle
judgements about policy effectiveness.
Charging for statistical data is a policy that is hard to rationalise.
Simple economics tells us the price that should be charged for
"public goods" - such as clean air, street lighting, or national
defence - is zero, otherwise these public resources will be underused. In
the case of data, there is an extra public benefit: good research leads to
better public policy.
As much as it pains two economists to say it, we are talking about a
free lunch. Leading researchers from around the world are naturally drawn
to the best and most detailed data. The ABS currently collect the best
data - the problem is that they will not release it. It is a sure bet that
if the ABS starts to release these data, then top researchers - including
our colleagues at Harvard and Stanford - will start to analyse them,
providing free insights to Australian policymakers.
Hence we offer this twin challenge: First, Australia's federal
politicians - of all political persuasions - should commit to providing
the ABS with the extra $7 million required to abolish data access fees,
and commit to opening up the databanks. Second, new policy proposals
should be subject to random trials before being funded. The cost of policy
mistakes is surely greater than that of small-scale random trials. And NSW
should take the lead.
To those who don't sign on, we say: "chicken!". One can
barely disagree on cost grounds. Rather, the fear must be that with real
evidence, voters might discover that reality does not match political