A home for sale in Alameda, California.

Any observer of the last housing crisis understands the importance of the U.S. housing bubble to the Great Recession. But the exact way that the bursting of the bubble—and the subsequent 20 percent decline in housing prices—triggered the worst recession in the United States since the Second World War is still very much up for debate.

Some researchers and policymakers emphasize the role of mortgage-backed securities combining with a highly leveraged financial system to bring down the economy. Others focus on the direct effect of the housing price decline on consumption. When it comes to that latter story, however, it’s not readily obvious which exact mechanism dominated. But a new paper from Greg Kaplan, Kurt Mitman, and Giovanni Violante investigates how much of the decline in household spending was due to the hit to households’ net worth.

In a series of influential papers and in their book “House of Debt,” economists Atif Mian and Amir Sufi make the argument that the increases in both household leverage and debt loads, combined with the decline in housing prices, sparked the Great Recession. Debt acted as a financial accelerant and increased the response to the housing price decline. In one of their papers with Kamalesh Rao, Mian and Sufi find that zip codes with more levered households were more likely to pull back consumption when housing prices went down. And given that low-wealth households tend to be more highly leveraged, the distribution of wealth has important implications for the functioning of the macroeconomy.

Yet as important as this research has been, it’s been difficult for other researchers to replicate the findings, as Mian, Rao, and Sufi used proprietary data for household debt loads. Fortunately, the new paper by Kaplan, Mitman, and Violante attempts to replicate the prior research using publicly available data.

Using those public datasets, Kaplan, Mitman, and Violante find results broadly in line with what Mian, Rao, and Sufi found: Areas with larger declines in housing prices saw larger declines in expenditures on non-durable goods, such as groceries or cleaning supplies. The original paper by Mian, Rao, and Sufi focused on overall expenditures (though they did do analysis for just non-durable goods).

Where the new paper differs is how much importance it finds for the role of leverage. When Kaplan, Mitman, and Violante control for the direct effect of the housing price decline, they don’t find much of an independent effect of the household balance sheet effect that Mian and Sufi emphasize. The authors take this to mean that some of the decline in expenditures after the housing burst isn’t due to the leverage effects. This would mean that some of the consumption decline wasn’t due to how a household purchased a house (i.e., how much of its purchase was funded via debt) but simply that a household had a house when its price declined. In other words, that debt might matter less than we currently believe.

But quite a bit of research has found a significant importance for leverage in accelerating economic responses. Mian and Sufi have other research showing the importance of household leverage. Additional research shows how more leveraged companies laid off more workers during the Great Recession. And historical analyses show that the fallout from debt-fueled bubbles is much worse than non-debt fueled bubbles. Perhaps the effects of the housing crisis weren’t entirely due to leverage, but something doesn’t have to explain everything for it to be vitally important.