A screen on the floor of the New York Stock Exchange shows the closing number for the Dow Jones industrial average.

When central banks can no longer push down short-term interest rates, what are macroeconomic policymakers to do? Policymakers and researchers had a wide-ranging debate about the kinds of unconventional monetary policy that could have been tried to boost growth in the years after U.S. interest rates hit zero back in December 2008. At the same time, fiscal policy—once regarded as unnecessary in an era of omnipotent central banks—made a comeback as the federal government passed a roughly $800 billion stimulus package.

The U.S. economy today is in its seventh year of recovery, and interest rates are on the rise. Yet another recession is inevitable, perhaps another powerful one. Among policymakers and researchers who favor fiscal stimulus, the relative merits of different kinds of stimulus when monetary policy hits the zero lower bound is still up for debate. Should a stimulus program focus more on direct government purchases, or should it favor transfers to households?

New research tries to unpack what kind of fiscal policy is most effective at the zero lower bound. In a recently released Equitable Growth working paper, Neil Mehrotra of Brown University and the Federal Reserve Bank of Minneapolis looks at the relative effectiveness of government transfers—say, a tax rebate—versus government purchases such as the construction of a new road. Mehrotra doesn’t directly answer the question at hand, but he does build a model of the economy and then figure out the conditions under which fiscal policy featuring these two kinds of stimulus are most effective.

What he finds (akin to what other models show) is that fiscal policy in general is most powerful when prices are “sticky,” meaning they don’t change quickly and conventional monetary policy is constrained. Mehrotra’s model shows that the effectiveness of transfer-stimulus programs—think tax breaks—depends on the relationship between household debt and access to credit. The more a decrease in household debt decreases the difference between the rates at which banks borrow and lend to households, the more effective transfer programs are at boosting growth. In other words, if allowing households to pay down debt by giving them government transfers ends up reducing the cost of borrowing, then the result is a bigger boost to economic growth.

This strong relationship between household debt and access to credit means that government transfers boost growth in two ways. The first is that households will spend a larger share of temporary income, which increases the boost in demand in the economy from transfers. The second is that a decline in borrowing costs increases household incomes since households will be spending less to get credit.

While both kinds of fiscal stimulus are effective when monetary policy is at the zero lower bound in Mehrotra’s model, the feedback between household debt and the cost of borrowing is key to understanding the choice between transfers and purchases. Mehrotra points to some research finding a weak relationship, but fully understanding this relationship is an important research question for the future.

We don’t know when the next recession will hit or if it’ll be anywhere near as debilitating as the last one. But we should try to better understand the most effective tools policymakers possess for the next time. Just in case.