We’ve written from time to time that not all debt is created equal. Prudent business borrowing enables companies to make investments and expand operations. And even though governments like the US that issue their own currency may nevertheless sell bonds, operationally they can simply create more dough to fund spending. The constraint on spending is creating too much inflation, not bankruptcy. And since as we’ve regularly discussed, the business sector chronically underinvests, deficit spending is necessary and desirable most of the time. Economist Mariana Mazzucato has argued that there are certain risks, such as engaging in basic research, where the uncertainty is too great for entrepreneurs. And that’s before getting to the fact that the party that makes the discovery could easily see its technology exploited by free riders.
However, economic studies have regularly found that high levels of household debt is a negative for economic growth. Moreover, some economists have found a strong relationship between high levels of consumer debt and economic crises. Yet if you read the business press, analysts and government officials see rising consumer borrowing as a plus for growth. How does that make sense?
A recent Bank of International Settlements paper (hat tip UserFriendly) helps reconcile this apparent paradox. The immediate impact of household borrowing does indeed spur the economy near-term but creates drag down the road. And the level at which household borrowing becomes a net negative is 60% of GDP, when nearly all advanced economies are at higher levels. Worse, the dampening effect is more pronounced when the household debt to GDP level exceeds 80% The BIS puts the US as above that threshold
From the study by Marco Lombardi, Madhusudan Mohanty and Ilhyock Shim:
Our results suggest that debt boosts consumption and GDP growth in the short run, with the bulk of the impact of increased indebtedness passing through the real economy in the space of one year. However, the long-run negative effects of debt eventually outweigh their short-term positive effects, with household debt accumulation ultimately proving to be a drag on growth. Our estimates suggest that a 1 percentage point increase in the household debt-to-GDP ratio tends to lower output growth in the long run by 0.1 percentage point, suggesting that policy makers face non-trivial, real costs in stimulating the economy through credit expansion. These findings are robust to alternative lag structures and control variables. Our analysis of the threshold effect suggests that the negative long-run impact of household debt on consumption growth intensifies as the household debt-to-GDP ratio exceeds a threshold of 60%. The estimated threshold is somewhat larger for GDP growth, with the negative debt effects intensifying as the household debt-to-GDP ratio exceeds 80%.
And what did they find explained most of the differences among the 54 countries they studied from 1990 to 2015? How much power the lenders had in forcing borrowers to pay them back:
Another interesting aspect of our results is the role of country-specific characteristics in determining debt limits. One key result is that the only institutional factor able to account for cross-country variation is the degree of legal protection of creditors. In particular, we find that countries with stronger creditor protection tend to experience more drag on long-run growth from higher levels of household indebtedness. We interpret this result as implying that in countries with stronger creditor rights, household borrowers are less likely to default on their loans and more likely to service their debt in the long run. This reduces consumption growth and eventually GDP growth to the extent that banks in the countries do not sufficiently reduce ex ante their loan spreads in consideration of higher expected earnings due to stronger creditor rights.
Note that this paper isn’t set up to capture what amounts to changes in the legal regime. In particular, in the US, mortgage borrowers had the belief, which historically was sound, that if they got into trouble, as in had a work interruption or a financial emergency, that the bank would work with them, as they do with business borrowers who have a bad spell but look to be viable in the long run. That went out the window with the rise of mortgage servicing, since the servicers found it more profitable to foreclose than modify loans, even though the investors would also do better with a successful mortgage modification than a foreclosure sale.
And you can see that in this chart. The US has supposedly had the largest post-crisis household deleveraging of any advanced economy, and much of that was involuntary, meaning the result of foreclosures. Banks also became stringent post crisis about issuing new mortgages, more the result of typical behavior and the need to strengthen their balance sheets than the favorite scapegoat of regulations:
And even with all of its deleveraging, the US is still just above the 80% level1:
The researchers tried to decompose long versus short-term effects:
The right-hand panel of Graph 3 makes two points very clear. First, past increases in household debt are not a good predictor of positive growth but appear to be associated with weaker consumption and higher risks of recession. Second, the downward-sloping line suggests that the negative correlation between household debt and consumption actually strengthens over time, following a surge in household borrowing. What is striking is that the negative correlation coefficient nearly doubles between the first and the fifth year following the increase in household debt.
As is well known, simple correlation does not suggest anything about the causal effects. That said, the preliminary evidence in Graph 3 appears to support the view that credit expansions may have very different effects on the short- and medium-run economic prospects of countries. It also confirms the findings of King (1994) that large increases in private debt in the 1980s made many OECD countries vulnerable to problems of weak growth and “debt deflation”. He shows that the most severe recessions since the 1930s have occurred in countries that have seen the largest increases in private debt in the preceding five years.
If you like nerdy papers, you’ll enjoy the data analysis here. In addition to using several different methodologies to examine the data, they also tested extensively for cross-country explanatory variables and robustness.
The authors point to the open policy questions at the end:
An important question, on which this paper is largely silent, is the role of various factors in the accumulation of household debt.20 One key issue in the context of the risk-taking channel of monetary policy (Borio and Zhu (2012)) is the extent to which low short- and long-term rates over the past eight years may have played a role in the recent rapid rise in household debt in many countries and may even have constrained central banks in raising rates. Even though such a question remains beyond the purview of this paper, any assessment must consider the various short- and long-run effects associated with any strategy aimed at stimulating the economy through ever larger debt levels.
Let us offer our own bit of speculation from ECONNED in 2010:
Let’s use a different metaphor to illustrate the problem. Say a biotech firm creates a wonder crop, the most amazing creation in the history of agriculture. It yields far more calories per acre than anything else, is nutritionally extremely complete, and can be planted and harvested with far less machinery and equipment than any other plant. It is tasty and can be prepared in a wide variety of ways. It is sweet too, so it can be used in place of sugar and high fructose corn syrup at lower cost. We’ll call this XCrop.
XCrop is added as a new element in the food pyramid and endorsed by nutritionists and public health officials all over the globe. It turns out that XCrop also is an aphrodisiac and a stimulant (hmm, wonder how they engineered that in) and between enhanced libido and more abundant food supplies, the world population rises at a faster rate.
Sales of XCrop boom, displacing traditional agriculture. A large amount of farmland is turned over from growing other types of produce to XCrop. XCrop is so efficient that agricultural land is taken out of production and turned to other uses, such as housing, malls, and parks. While some old-fashioned farms
still exist, they are on a much smaller scale and a lot of the providers of equipment to traditional farms have gone out of business.
Twenty years into the widespread use of XCrop, doctors discover that diabetes and some peculiar new hormonal ailments are growing at an explosive rate. It turns out they are highly correlated with the level of XCrop consumption in an individual’s diet. Long-term consumption of high levels of XCrop interferes with the pituitary gland, which controls almost all the other endocrine glands in the body and the pancreas.
The public faces a health crisis and no way back. It would be very difficult and costly to put the repurposed farmland back into production. Some of the types of equipment needed for old-fashioned farming are no longer made. And with the population so much larger than before, you’d need even more farm- land than before. The world population has become dependent on the calories produced by XCrop, so going off it quickly means starvation for some. But staying on it is toxic too. And expecting users simply to restrain themselves will likely prove difficult. The aphrodisiac and stimulant effects of XCrop make it addictive.
Advanced economies have become hooked on debt technology, which, like XCrop, is habit forming and hard to wean oneself off of due to its lower cost and the fact that other approaches have fallen into partial disuse (for instance, use of FICO-based credit scoring has displaced evaluations that include an assessment of the borrower’s character and knowledge of the community, such as stability of his employer). In fact, the current debt technology results in information loss, via disincentives to do a thorough job of borrower due diligence (why bother if you are reselling the paper?) and monitoring of the credit over the life of the loan. And the proposed fixes are not workable. The Obama proposal, that the originator retain 5% of the deal and take correspondingly lower fees, is not high enough to change behavior. And a level that would be high enough to make the originator feel the impact of a bad decision would undercut the cost efficiencies that made securitization popular in the first place. You’d have better decisions, but less lending, and higher interest rates. That’s ultimately a desirable outcome, but as in the XCrop situation, no one seems prepared to accept that a move to healthier practices will result in much more costly and less readily available debt.The authorities want to believe they can somehow have their cake and eat it too.
Note this hasn’t proven to be quite correct; the most rapidly growing category of consumer debt post crisis has been student debt. With most loans government guaranteed, investors have no credit concerns…but the high level of delinquencies and defaults attests to the severity of that ticking time bomb. And we have the more immediate effect that generous student loans simply produce more college education cost bloat, while leaving many graduates with debt burdens that in many cases keep them from getting married, buying a house, and starting a family, and can wind up being a millstone.
Yet both parties shy away from addressing this issue. The Democrats are particularly complicit since academia serves as an informal but very large addition to their think-tank apparatus. Where is our William Jennings Bryant, who will talk about a cross of debt?
1 Bear in mind that the researchers made adjustments for consistency, which may explain why their figures for the US, which exclude student debt (they consider only borrowings made via banks, while the Federal government makes the majority of student loans via colleges) come up with a higher debt to GDP ratio than if you take the New York Fed’s quarterly report on household debt for year end 2016, and divide that by the fourth quarter GDP just released.