Concern that a buildup in debt around the globe is going to cause some problems for financial and economic stability isn’t controversial. The bursting of the U.S. housing bubble in 2007 made clear the dangers, and research since then on the influence of the growth of credit and debt on recessions has added to that understanding. But when it comes to the destabilizing effects of credit and debt, what’s most concerning: the level of debt in the economy or the pace at which debt is being added?

In other words, should policymakers start worrying when debt-to-gross-domestic-product ratios get high or when credit growth starts consistently outpacing GDP growth?

In a new working paper, economists Jonathan Bridges, Chris Jackson, and Daisy McGregor of the Bank of England look at just this question. Using data from 26 high-income countries since the 1970s, the three economists look at how credit and private debt affect the severity of recessions. In their investigation, the severity of a recession is measured by how much lower GDP-per-person is 3 years after the start of an economic downturn.

Cutting to the chase, their results show that the increase in credit in an economy is a better predictor of the severity of a recession than the level of debt already in the economy. In fact, the predictive ability of their model (the share of variation in the severity of recessions explained by the model) doesn’t change much when the level of debt is added to the pace of credit growth. A 10 percentage point increase in credit growth relative to GDP predicts, in this model, a loss of 2 percent of GDP per person after three years. The level of credit before the recession has no strong predictive power.

But this isn’t to say that the level of debt-to-GDP is never an important consideration. The authors look at how much credit growth affects the severity of recessions when debt-to-GDP is above or below a certain threshold (around 164 percent). When the level of debt is above that threshold, then the impact of a 10 percentage point increase in debt-to-GDP results in a 2.5 percent decline in GDP per person. But when the ratio is below that level, the impact of an increase is only a 1.5 percent decline. The level, then, does matter, but only in increasing the impact of the credit growth.

Other research on the impact of debt and credit on economic output has emphasized the role of household debt. But Bridges, Jackson, and McGregor find that increases in household debt and business debt have an impact on the depth and length of recessions. This is but one paper in the debate on levels versus increases that’s just beginning and is likely to continue for some time. For economists, and especially for policymakers, the more evidence on what kind of credit growth and what kind of debt—as well as whether to look at the level or the change in them—is available, the more likely it is to be able to smooth out economic fluctuations. Unfortunately, we aren’t quite there yet.