By Juan Sánchez, Assistant Vice President and Economist, St. Louis Fed; Ryan Mather, Research Associate, St. Louis Fed; Kartik Athreya, Executive Vice President and Director of Research, Richmond Fed; and José Mustre-del-Rio, Senior Economist, Kansas City Fed
The following is the fourth post in a series examining the potential impact of COVID-19 on people living in areas already experiencing above average levels of financial distress.
There will be economic losses from the COVID-19 pandemic, and knowing how those losses are distributed across the population is just as important as the overall amount.See, for example, our recent working paper in which we estimate that the inequitable distribution of housing price losses during the Great Recession magnified the overall drop in consumption by 40%. Our previous posts in this series have uncovered a troubling pattern. Communities with the highest financial distress—which are the least able to weather economic losses—both:
We used these results to construct a simple prediction of how much more income highly financially distressed areas would lose compared with less distressed areas; we then fed this prediction into an economic model to see how consumption would react.The model used in that post and in this one is described in fuller detail in our recent working paper. The results, unsurprisingly, suggested that consumption losses would be significantly higher in areas with greater financial distress.
In this post, we continue to explore this same exercise in the model, this time asking how the income losses will affect financial assets, delinquency and bankruptcy. Importantly, our definition of financial assets in the model does not include risky assets (such as equities) and instead refers to assets such as checking and savings accounts or certificates of deposit. Equities tumbled toward the beginning of the COVID-19 pandemic, but it is difficult to say how these losses will be distributed across the population. Instead, we merely change income within the model and examine how this channel alone affects savings in riskless assets, home equity, and repudiation of unsecured debt via delinquency and bankruptcy.
The table below makes clear that, once again, financial distress matters. We see that the response of each of the items under study differs substantially across quintiles of financial distress, with the first quintile representing the 20% of ZIP codes with the least financially distress and the fifth quintile being the 20% with the most. We allow the size of income losses to differ across the quintile of financial distress, to capture the unevenness of exposure to COVID-driven employment risk that we documented in our first post.
The first row reports the income losses that were fed into the model for each quintile of financial distress.Details of their construction appear in our second post. The remaining rows show the reaction of various items on household balance sheets to the given decline in income. By construction, line a equals 100% since we are dividing all rows by the quantile-specific drop in income.
|Quintiles of Financial Distress|
|Percentage decline in income||-6.36||-6.78||-7.66||-8.33||-9.37|
|Percentage Share of Drop in Income|
|a. Drop in income||100.0||100.0||100.0||100.0||100.0|
|b. Drop in assets||54.5||49.1||45.5||41.8||32.2|
|c. Increase in debt||1.5||2.0||2.2||2.4||2.8|
|d. Decrease in home equity||17.1||18.2||17.5||18.7||20.1|
|e. Drop in consumption||26.8||30.7||34.8||37.1||45.0|
|SOURCES: Authors’ calculations based on the model used in their paper “Consumption in the Great Recession: The Financial Distress Channel.”|
Focusing on individuals in Q1 (the least financially distressed), we see they reduce consumption the least (line e)—only 27 cents per dollar of income lost. They are able to shield most of their consumption from the drop in income by dipping into their savings (line b), as the drop in their assets offsets over 50% of the drop in income. This drop in savings, combined with a decrease in home equity (line d), in total helps offset more than 70% of the income lost. This suggests that, in the model, these individuals have enough wealth in their homes to borrow against them during times of trouble.
In sharp contrast, because of their “preexisting condition” of being in the highest financial distress category and having weaker balance sheets, individuals in Q5 reduce consumption by nearly twice as much as their counterparts with healthy balance sheets. Their spending falls by 45 cents per dollar of income lost. The drop in consumption occurs not because that’s what these consumers prefer, but rather because they cannot effectively dip into assets (savings and home equity) to weather the income shock. For these individuals, dipping into savings and home equity offsets only about half (52.3%) of the drop in income, leaving their old living standard out of reach.
The role of debt in this decomposition is minimal. Debt, however, is the only source of funds for renters without liquid savings. This is likely true for younger households. Additionally, note that debt may vary because of changes in financial distress. Debt increases for individuals who are charged penalty rates due to skipped payments (delinquency), and it decreases for individuals who decide to file for bankruptcy. The next table describes the evolution of financial distress.
The table below shows how financial distress changes in response to income lost, and how these changes are disproportionately carried out by groups who were already in heightened states of distress. For ease of interpretation, we scale the model to the labor force of the U.S. economy, so that it represents roughly 164 million people.
With this figure in mind, the first row of the table shows that more than 2 million additional people are projected to go delinquent on their debts than would otherwise, and about one-third of a million more will go bankrupt.
|Total among all workers||2,186,146||326,045|
|From the Top 20% of Preexisting Financial Distress|
|Share of total||28%||22%|
|From the Top 50% of Preexisting Financial Distress|
|Share of total||54%||55%|
|NOTES: The top 20% are the 20% of individuals with the highest financial distress. Similarly, the top 50% are the 50% of individuals with the highest financial distress.|
|Authors' calculations based on the model used in their paper "Consumption in the Great Recession: The Financial Distress Channel."|
The remainder of the table shows that the increase in delinquencies and bankruptcies is concentrated among already-distressed individuals. For example, people who are in the top 20% of preexisting financial distress (i.e., those in Q5, the worst financial distress) experience 28% of the extra delinquencies that occur because of the simulated income shock. This is surprising, because these extra cases of delinquency and bankruptcy are coming on top of already elevated preexisting conditions. In other words, people with the worst preexisting financial distress also have the strongest propensity to go deeper into distress, compounding the effects of the preexisting conditions.
This being said, the bottom panel of the second table shows that as we move down the financial distress distribution, things become more proportional. Indeed, people among the top 50% of pre-existing financial distress account for 54% of the extra delinquencies that occur because of the shock. In other words, they account for a share of extra delinquencies that is more in line with their population share.
One thing that is not captured in this table is the timing of the response of delinquencies and bankruptcies. In the model, most of the increase in financial distress occurs in the period after the initial shock to income. This reflects a certain delay that we believe also exists in the real world. For example, assuming that a person is current on their debt payments when the income losses occur, it will take at least 30 days before they can be 30 days delinquent on their debt, which is the threshold at which the New York Federal Reserve/Equifax Consumer Credit Panel (CCP) data set records debt as delinquent.This has been our source for calculating financial distress statistics in previous posts and is the main source used on this subject.
Altogether then, our simulated model suggests two main predictions for the economy during the COVID-19 pandemic: First, communities with the greatest financial distress are likely to cut their consumption the most, both because they are less prepared to weather shocks and because the income shocks that they face are projected to be larger. Second, these same communities are also more likely to go into further financial distress as the pandemic continues.
In our view, these findings suggest that special consideration should be given to distressed communities as the government moves to provide relief efforts. In a future post, we will evaluate some of the suggested policies and provide some insight into which might be the most effective.
1 See, for example, our recent working paper in which we estimate that the inequitable distribution of housing price losses during the Great Recession magnified the overall drop in consumption by 40%.
2 See also our daily updates of these data, which show that highly financially distressed communities continue to be more at risk of contracting the illness.
3 The model used in that post and in this one is described in fuller detail in our recent working paper.
4 Details of their construction appear in our second post.
5 This has been our source for calculating financial distress statistics in previous posts and is the main source used on this subject.