Australian Assistant Minister for Productivity, Competition, Charities and Treasury Andrew Leigh said governments should direct money towards programmes that rigorous evaluation shows work, in an address delivered on 1 September 2026 to Asian Evaluation Week in Manila.

The event was hosted by the Asian Development Bank at its headquarters in the Philippines under the theme "Put Your Money Where Your Evidence Is". An edited version of the address was published on the Devpolicy Blog, run by the Development Policy Centre at the Australian National University.

Leigh said evaluation answers one question: whether a policy or programme improved people's lives. That is hard to answer, he said, because what would have happened without the intervention cannot be observed directly, and rigorous evaluation supplies a credible counterfactual.

Many policies that look promising fail when tested, while some simple and cheap interventions do better than expected, he said. Governments should be learning institutions, and evaluations too often end with a report rather than a decision.

Leigh gave three examples of trials whose findings were used. In the Democratic Republic of the Congo, neighbourhoods in one large city were randomly assigned a progressive or a proportional property tax schedule. The progressive schedule raised revenue by about 56 per cent and was later expanded across the city.

In Zambia, households were guided through a structured exercise to think about future expenses. According to the address, that prompt increased savings by 15 per cent, left households with about one extra month of savings entering the "hungry season", and led to a 9 per cent rise in crop revenue the following year.

In Kenya, minibus drivers were randomly assigned financial incentives for safer driving, measured with vehicle monitoring technology. Speeding fell by 29 per cent among drivers who received the incentive, but the effect faded once the incentive was removed, Leigh said.

He said policy questions on tax capacity, road safety, poverty reduction and service delivery are shared across Asia and the Pacific, and that results do not always transfer between countries. A rigorous study from overseas can still be more useful than poor local evidence, he said.

Leigh said fewer than 25 randomised trials from developing countries were published each year in the 1990s, against 274 in 2012. The Abdul Latif Jameel Poverty Action Lab at MIT has led more than 2,600 randomised trials, and the American Economic Association's registry lists more than 12,000 social science trials.

Australia's Centre for Evaluation, set up in the Treasury portfolio in July 2023, has supported seven randomised trials with four more under way, he said. Its Evaluation Profession network has more than 8,000 members and its public repository holds over 200 evaluations.