In addition to Grattan Institute’s annual Prime Minister’s Summer Reading List, our Wonks’ List highlights some of the year’s best technical policy reads, for anyone who wants to take a deeper dive. Here’s our 2023 Wonks’ List.

Zero-Sum Thinking and the Roots of US Political Divides

Sahil Chinoy, Nathan Nunn, Sandra Sequeira, and Stefanie Stantcheva

Zero-sum thinking infects many policy debates. It’s the view that gains for one group must come at another’s expense – such as men fearing that gender equality initiatives will leave them out in the cold, or the belief that unions and businesses bargain over a fixed pot of money. This is often a very limited, short-term way to think about the world: policies that make society more inclusive and productive can ultimately benefit us all. But the belief persists.

This US paper investigates what people who hold zero-sum beliefs have in common. It argues that when and where you grow up has a big impact on your outlook. People who grew up when times were good, or whose family’s lot improved from one generation to the next, are less likely to think in zero-sum terms. In other words, if your beliefs were formed at a time when all boats seemed to be rising, you’re more likely to expect this to continue later in life. Conversely, people whose ancestors were slaves or who suffered other forms of subjugation tend to have a more zero-sum worldview, consistent with the reality of those experiences.

Zero-sum beliefs shape people’s policy preferences today – the paper shows that in the US, this way of thinking is associated with support for more redistribution and stronger strictions on migration. But the relationship also goes the other way: the policies and economic environment of today can shape beliefs and norms for generations. It’s a theme also picked up in our next Wonks’ List pick, Men.

Read the paper

Men. Male-biased sex ratios and masculinity norms: evidence from Australia’s colonial past

Victoria Baranov, Ralph De Haas, and Pauline Grosjean

How are ‘masculinity norms’ created, and how do they manifest themselves in Australia today?

When Britain sent convicts to Australia, it sent many more men than women. By the mid-1800s, there were 28 convict men in NSW and Tasmania for every convict woman. An oversupply of men in some parts of Australia meant more competition for female partners. Research has shown how this competitiveness increased women’s power in marriages, allowing them to work less. This behaviour became normalised, and was passed down through families and among peers, such that women in historically male-dominated parts of the country are less likely to work even today.

While a burgeoning stream of literature has begun to unpack the impact of gendered norms on women, we knew very little about the consequences of masculine norms on men – until now.

This paper uses a wide variety of datasets to show how differences in historic sex ratios across different parts of Australia caused differences in masculine norms today. The results are astounding. In areas that had more male convicts relative to women, men were more likely to voluntarily enlist for World War 1 a century later. And in the present day, these areas have significantly higher rates of violence, male suicide, and death from preventable health conditions, and fewer men working in traditionally female industries. 

This tells us that historical norms perpetuate and continue to harm men today, but also serves as a warning. In some developing countries, skewed sex ratios from sex-selective abortions and other cultural practices could have harmful long-term effects. And within Australia, problems could emerge from male-biased settings, such as gender-segregated workplaces, schools, and prisons.

Read the paper

The rise and fall of peer review

Adam Mastroianni

Calls for evidence-based policy are ubiquitous, including from the Grattan Institute. But evidence-based policy is only as good as the evidence underpinning it. How do we know if the evidence is any good? One safeguard is the peer review process, which ensures that every paper published in reputable journals is first checked by other academics in the field.

But this provocative post from Adam Mastroianni argues that the peer review model is fundamentally flawed. First, it has proven deficient at its main job: catching errors. There is evidence of this directly – Mastroianni cites studies where reviewers detect less than one-third of errors deliberately added to papers – and indirectly, in the ‘replication crisis’ engulfing many fields, including economics.

Second, peer review imposes costs. The cumulative time academics spend reviewing, responding to comments, and resubmitting across journals is substantial. Worse, peer review may set back progress in science. The imprimatur of peer review can give a false sense of authority to shoddy research, making it harder to debunk. And the process may stymie truly revolutionary research, which Mastroianni argues elsewhere matters far more than catching errors.

This post doesn’t have all the answers, but it should spark further reflections and conversations about how we can improve the quality of research underpinning policymaking.

Read the post

Generative AI exists because of the transformer

The Financial Times Visual Storytelling Team and Madhumita Murgia

It’s barely been a year since ChatGPT was released to the general public and brought conversations about AI to the mainstream. But how large language models (LLMs), such as ChatGPT and Google Bard, work is often poorly understood. This can lead to embarrassing slip-ups, like mistaking AI ‘hallucinations’ – reasonable-sounding, but made-up statements – for fact. And it makes it more challenging to have reasoned debates about how policy should respond to AI.

This story from the Financial Times is a great primer on how LLMs work. With helpful visualisations, it explains how they use the proximity between different words in training data to build an ‘understanding’ of word meanings, and predict the next word in a piece of text.

It also outlines some big developments in recent AI history. The ‘transformer’ tool, first published by a group of AI researchers at Google in 2017, has been crucial to the generative AI we see today. The transformer allows models to process language sentence by sentence, rather than word by word. It also allows LLMs to consider context from beyond sentence boundaries to build its ‘understanding’ of word meanings.

The opportunities and risks of AI are enormous. As the world grapples with its policy implications, ensuring debate is grounded in facts – not hysteria, or hallucinations – is more important than ever.

See the data visualisation story

A trio of natural experiments

We often blame the individual for their health problems, but three studies published in the past year show just how much is outside the individual’s control. Each study used a natural experiment – where a change happened to one group but not another group – to look at how things outside our control can affect our health.

A US study compared cities that introduced a tax on sugary drinks with those that didn’t. Cities with the tax recorded substantial declines in gestational diabetes, which will mean healthier mothers and babies. A Norway study compared children who grew up near fast-food outlets with those who didn’t. Growing up near fast-food restaurants is linked to weight gain and obesity and lower cognitive ability. A third study looked at the rollout of Facebook across college campuses in the US. Students exposed to the social network reported that their mental health declined, adding to the evidence that social media may be contributing to troubling trends in youth mental health. In each case, the studies controlled for other factors to isolate the impact of a specific aspect of our environment: sugary drinks, junk food, and social media.

Companies have an incentive to increase consumption of their products, even if it harms our health. These studies show that there can be significant, long-term consequences. And the example of a tax on sugar-sweetened drinks, which Grattan Institute has previously proposed for Australia, shows how governments can step in to tackle the commercial determinants of health, helping keep people healthy.

Read the papers

Sugar-Sweetened Beverage Taxes and Perinatal Health: A Quasi-Experimental Study
Kaitlyn E. Jackson, Rita Hamad, Deborah Karasek, and Justin S. White

Swallow This: Childhood and Adolescent Exposure to Fast Food Restaurants, BMI, and Cognitive Ability
Sara Sofie Abrahamsson, Aline Bütikofer, and Krzysztof Karbownik

Social Media and Mental Health
Luca Braghieri, Ro’ee Levy, and Alexey Makarin

While you’re here…

Grattan Institute is an independent not-for-profit think tank. We don’t take money from political parties or vested interests. Yet we believe in free access to information. All our research is available online, so that more people can benefit from our work.

Which is why we rely on donations from readers like you, so that we can continue our nation-changing research without fear or favour. Your support enables Grattan to improve the lives of all Australians.

Donate now.

Danielle Wood – CEO