Jamie’s Italian traded well for its first five years before eventually collapsing into administration in 2019. Jon Knight, the chief executive at the time, lamented a “growing disconnect” between the restaurant chain and the vision of its founder, the celebrity chef Jamie Oliver. For John List, a professor at Chicago University and chief economist at Walmart, the failure highlighted why certain concepts could not scale. The original restaurant’s intrigue was the direct involvement of Oliver, which couldn’t be replicated across a bigger chain.

List’s new book, The Voltage Effect, draws on his experience as a member of President Bush Jr’s Council of Economic Advisers, as well as stints as chief economist at the ride-hailing apps Uber and Lyft, to provide five “vital signs” to ponder before scaling ideas. Both Tory leadership contenders should heed them. List exposes the pitfalls of much “evidence-based policy” while inoculating readers against the unthinking tendency to extrapolate from small trials to full-blown national programmes.

First, beware false positives when scaling ideas. Small experiments often imply something works, but, without replication, noisy data can conceal an absence of impact.

Next, know your audience. Smart thermostats are supposed to slash CO2 emissions, yet List’s study of a randomised introduction of them in California found meagre energy savings. Engineers had assumed households had the rationality of Spock from Star Trek. They instead got energy users who were more akin to Homer Simpson. Businesses and governments must make sure their concepts work with the actual relevant groups.

Third, assessing the risk of spillovers or changes in behaviour after implementing a policy more widely is crucial. A lot of the secondary impacts of crude lockdowns after the spring of 2020 are only now manifesting. Business decisions often suffer from compensating behaviours too. When Uber introduced tipping to boost wages in the US, more drivers hit the roads. This increase in supply led to more time driving without rides, so neutralising rises in tips.

Fourth, one must carefully consider how costs change with scale. List recalls Lyft data that suggested Facebook adverts were more cost-efficient than Google per 1,000 drivers recruited from each platform. Before Lyft just threw more money Facebook’s way, List demanded analysis about the cost of the last 25 hires on each service. This information revealed Google was significantly cheaper for recent hires. When scaling up, average costs can mislead us about expansion costs. All such considerations apply to government scaling efforts too, of course.

Fifth, as with Jamie’s Italian, a common flaw of the state is the failure to recognise input scarcity. For example, across the West there has been a dash to universalise childcare interventions trialled in highly localised studies. List’s own work on a Chicago Heights pre-school programme found a project that seemed to bestow cognitive benefits on children. He was heartened by its potential, given that his team were careful to hire teachers that such an institution could realistically employ. Yet he realised even this result might not be replicated broadly. Though you could easily hire similar teachers in small numbers across disparate cities, the numbers required to serve all children within every city would mean quickly compromising on quality.

List remains optimistic that awareness of these warnings can improve attempts to scale up within business and government. His book, though, is a good economic primer for why lofty ideas often fail, even when seemingly grounded in evidence. Any reader will recognise the need for humility in determining “what works”, as well as the dangers of the one-size-fits-all approach beloved of politicians.