BETWEEN June 2014 and February 2015 the price Americans paid for petrol fell by a third. Economists predicted this would boost growth by causing consumers, newly flush with cash, to spend more on other goods and services. Instead, the economy seemed to slow, with early estimates putting annualised growth in the first half of the year at a paltry 1.4%. Many claimed Americans were saving the windfall, or using it to pay down debts. Estimates of growth in the first half of the year have since been revised up sharply, to 2.3% annualised. Reinforcing this turnaround, a report released today argues that Americans are spending most of the oil-price windfall after all.

Researchers at the JPMorgan Institute, a think-tank tied to the bank, examined anonymised data from one million of the bank’s credit- and debit-card customers. The number-crunchers divvied up customers according to how much they spent on fuel before prices fell. Gas-guzzlers gain the most when fuel gets cheaper; reluctant-refuelers benefit less. Comparing the two groups’ spending before and after the price collapse can reveal how much of a dollar saved at the pump is spent elsewhere.

To mitigate the problem of mean reversion—high spenders spend less over time by virtue of being outliers to begin with—customers were categorised as gas-guzzlers or otherwise based on average spending on fuel by their zip-code neighbours. The researchers found that for every extra dollar those in gas-guzzling neighbourhoods saved at the pump, their spending elsewhere rose by 73 cents. This increased to 89 cents after adjusting for the fact that fuel is more likely to be bought with a debit or credit card than other expenses.

If this estimate is right, low oil prices are significantly boosting American consumption after all. This should reassure those who fret that low prices have reduced investment in oil and gas extraction without boosting consumer spending by much. The finding also contradicts recent survey evidence: one conducted by Gallup, a pollster, for instance, found that only 24% of Americans say they are spending their savings from cheaper gas.

The analysis, though, is not definitive. In particular, it relies on the similarity of gas-guzzlers and reluctant-refuelers along dimensions other than fondness for petrol (lest some other difference between the group be driving their divergent spending patterns). In support of this assumption, the authors point to the similar demographics of the two groups. For instance, both have a median monthly income of around $5,300. The two groups’ spending also follows a similar patterns before the oil price fall.

Much variation in fuel spending is driven by geography: gas-guzzlers are concentrated in spacious south, whereas almost three-quarters of low spenders are in the more metropolitan north-east. Divergent economic fortunes for different regions could, therefore, distort the results. In addition, there is a wide statistical margin of error around the estimates.

But the findings are still important, for three reasons. First, they provide some reassurance that boosting consumers’ disposable income does help the economy. The report’s authors note that the average household will gain $700 in 2015 from cheaper fuel—more than tax rebates issued in 2008 as a stimulus measure. Second, they suggest that consumers cannot always give pollsters an accurate picture of their budgets, especially when the question is complex. Even low earners spend only about 6% of their income on gas, so working out what they are doing with the savings is difficult. Finally, they show the potential for banks’ large data sets on individual behaviour to help answer big macroeconomic questions.