What are the risks to the broader economy associated with high private and household debt?


15/05/20

Charles Shaw – cshaw11@mail.bbk.ac.uk 


This blog entry maps the evolution of private credit in the UK and other developed countries and considers the role of private debt in determining the impact of a financial crisis. By examining both theoretical and empirical aspects of this topic, we can address the following key questions. What are the risks to the broader economy associated with high private and household debt? What are the mechanisms by which private debt and deleveraging may restrain economic activity? Available evidence suggests that high household debt levels aggravate crises in the following ways: recessions are deeper, and recoveries are weaker when they are preceded by larger surges in household debt. Analogous observations are made for booms and busts in the housing market.

What do we know?

It is by now well documented that the Global Financial Crisis came after a large upward swing in the leverage cycle.[1] Although there is no generally accepted notion of how much debt is too much, the recent state of affairs warrants attention, particularly as current global debt is at record levels and continues to rise. At 322 per cent of world GDP, it is 40 percentage points ($87 trillion) higher than at the onset of the 2008 financial crisis. Analogously, in the last ten years, the global private debt market has almost quadrupled in size. Global household debt is at $47 trillion (60 per cent of GDP), over $12 trillion higher than in the run-up to the 2008 global financial crisis.[2]

We know that the global crisis set in motion debt-deflation dynamics a la Fisher-Minsky (1933; 1963, 1977, 1978), which had the effect of rapidly inflating public debt-to-GDP. Following the recession, the initial response of many governments in Europe and elsewhere was to allow their debts to increase. This was achieved through two channels. The first channel consisted of the automatic stabilizers triggered by the recession-induced fall in government revenues. The second channel consisted of governments taking over private, mostly banking, debt that went or was about to go sour. Following such radical departures from past norms, a considerable amount of empirical work has emerged, with several researchers focusing their attention on the economic interplay between the macroeconomy and private debt.

A large amount of research, largely focusing on advanced economies, has identified meaningful connections between financial crises and rapid credit growth (Taylor, 2012), the critical role of expansion in global credit in general (Dell’Ariccia et al, 2012), and the role of expansion of credit to households in particular (Jorda et al, 2014). Evidence on global household leverage, house prices, and consumption by Glick and Lansing (2010) show that credit growth acts as a significant predictor of financial crises. Seminal work by Schularick and Taylor, (2012) shows that conditional on there being a recession stronger credit growth tends to predict deeper recessions. The main findings of Jorda et al (2011,2014) namely that credit growth emerges as a significant predictor of financial instability, and that external imbalances have played an additional role, but more so in the pre-WWII era of low financialisation than today, are illustrated in the figure and table below.

Building on such evidence, Mian et al (2016) identify negative dynamic correlations between changes in global household debt and global growth. They find that a rise in household sector debt by more than 1 standard deviation above trend is associated with downward pressure on GDP growth in the following three years. They further argue that dramatic increases in private debt increase in the likelihood of incidence in financial crises. They also demonstrate that increases in public debt do not affect the probability of a financial crisis, although Greece appears to be a notable degenerate case to this empirical regularity. The intuitive explanation they offer by way of support is as follows. The distribution of debt matters because people in the lower half of the distribution have a much higher propensity to spend. For a given shock to their wealth, they pull back much more on their spending. Therefore, someone who has a significant level of debt can see their entire equity position wiped out should the house prices crash. This leads to a large drop in household spending in the aggregate. It sets off a cycle of higher unemployment, delinquent mortgages, and other deleterious effects that are problematic for the economy.

The evolution of private credit in the UK and other developed countries

Taking inspiration from Schularick and Taylor (2012), it may be useful to outline in more general terms the evolution of credit in the UK and developed countries from a historical viewpoint. Let us start by observing that the share of total bank lending to the private non-financial sector in the British economy was around 16 per cent of GDP in 1880. Examination of the data for the next century, excluding two world wars, reveals that this share increased gradually, rising to 63 per cent. This figure, which includes credit provided by non-bank institutions, plateaued at 63 per cent over the next three decades, remaining at 63 until 1980.

What followed over the next two decades was the liberalisation and globalisation of the financial sector. During this period, credit grew faster than the economy, and the stock of credit marked rapid growth, rising to 120 per cent by the early 1990s. Although in the early 1990s this trend was dampened by a recession, by 2009 it had reached 177 per cent of GDP. Similar upward trends are observed in advanced economies.

Between 1997 and 2007, the annualised growth rate of real credit was 7 per cent, more than double the rate of the real GDP growth, which was 3 per cent. Had we witnessed a continuation of this trend, the stock of credit would have been 200 per cent of GDP by 2012, and 300 per cent of GDP by 2023. The financial crisis and the recession that followed resulted in a sharp correction. Credit growth fell in real terms by around 2.5 per cent per annum in the period 2010-14. By 2014 the stock of credit fell from its peak of 177 per cent of GDP to 140 per cent, similar levels witnessed in 2002, where it continues to remain.

Since the majority of lending during the credit growth between the mid-1990s and the financial crisis was to UK households, we can witness that the debt to income ratios UK households follows a similar path to the stock of credit to GDP. In 1964, it was 54 per cent, remaining at this level until 1980. However, by 1990 it had grown to over 100 per cent. It rose again from the late 1990s, growing to 155 per cent by 2007. It dropped to 135 per cent, its 2004 level, in 2012 and continued to remain at such levels. Based on these data, it is evident that the UK experienced a vigorous financial cycle during the period that preceded the crisis: there was a sustained build-up of credit to the private sector, after which a sharp correction put an end to some, but far from most, of the credit growth that predated this period.

Does private sector debt matter?

Credit, as does debt, facilitates home purchases, investment, risk management, and consumption smoothing. But if credit is such an essential building block of modern economic development, what is the point at which credit intensity (i.e. credit to GDP) becomes problematic? Is it even problematic at all?

In this vein, a high level of private-sector debt relative to the economy has often been interpreted as a sign of financial development, which in turn is beneficial for long-term growth. A large body of literature — see, for example, Rajan et al (1998, 2008); Banerjee and Duflo (2014) etc — deals with this question and finds, by various theoretical arguments and empirical observations, that after an increase in credit intensity beyond certain levels of credit to GDP countries experience lower growth.  For example, several cross-country empirical studies find that once a certain threshold is reached, larger public debt stifles potential growth, which suggests a non-monotonic and concave (“inverted U”-shape) relationship between public debt and economic growth, eg Cecchetti et al (2011), and Kumar and Woo (2010). Some have gone even further and calculated the debt-to-GDP turning point in this concave relationship to be approximately between 90 and 100 per cent, as in Checherita and Rother (2010) and Reinhart and Rogoff (2010a, 2010b).

As the alert reader no doubt will recall, the econometric analysis of Reinhart and Rogoff (2010a, 2010b) seemed to largely gloss over the empirical potential importance of private debt. Checherita and Rother (2010) made attempts to control for other potentially relevant variables, such as the stock of private debt. However, while doing so, they do not find the stock of private debt, defined as total domestic credit to the private sector, to be statistically significant in determining growth in their sample across any of their models. They, therefore, argued that its inclusion does not modify significantly the results for public debt. However, the argument advanced by Checherita and Rother does not hold empirically.[3]

Does gross debt restrain economic activity?

Some well-cited and influential studies have asserted that gross debt may restrain economic activity. By way of example, two influential reports by McKinsey (2010, 2012) argue that having too much gross debt matters a great deal. They argue that adjustment, i.e. reducing net indebtedness, is needed. They argue that advanced economies need to “clear the way” for economic growth by reining in the rising levels of gross debt. This claim is worthy of examination in great detail. If true, what does this imply for economic performance, given that government debt surged in advanced economies as a result of the crisis, and gross debt has reached levels not seen in several decades?

Given the available evidence, it is difficult to accept on face value the findings of the McKinsey studies. For example, Fatas (2012) points out that focusing on gross debt is “misleading” since what is of consequence is net wealth, rather than gross debt. Let us take the example of Japan, the country with the largest debt to GDP ratio in the McKinsey reports. The Japanese economy is a net creditor to the rest of the world, a fact which reflects decades of current account surpluses. Similarly, the high Japanese stock of gross government debt is related to large private sector assets. Overall, it appears that there is no accepted wisdom regarding which mechanism, or if at all, gross debt restrains economic activity.

There are two other critical points that the McKinsey authors miss. First, is not the gross debt per se that policymakers ought to look at when trying to understand broader risks to the economy. What matters is financial stability. The McKinsey authors fail to ask themselves one key question. What are the economic implications if credit consistently keeps rising faster than GDP, and therefore the debt to income ratio continues its upward trajectory? Surely, it is precisely this pattern that is bound to signal broader economic risks: for the UK, for other advanced economies, and for emerging markets with coterminous economic scenarios.

Second, when McKinsey issue policy advice of reducing net indebtedness, their logic is at odds with the reality of where the broader risks to the economy lie. To illustrate this fundamental misunderstanding of modern economic dynamics by the McKinsey authors and their subsequent, arguably spurious, policy prescriptions let us consider two scenarios that roughly bookend the possibilities. Scenario one: debt is concentrated in households with low liquid financial assets and high MPC. Scenario two: debt is evenly distributed across the population. In the first scenario, the economy is more vulnerable than it would be if the debt was more evenly distributed across the population. Under any scenario, for every financial debt there exists a corresponding financial asset, owned either by foreigners or domestic residents. For every borrower there exists a lender. What is the net indebtedness if not, by definition, zero? What is important is how the assets and liabilities are matched, and how the debt is distributed across the economy. 

The role of household debt

The 2007 financial crisis was not set off by the accumulation of debt by UK households. But the debt made our economy significantly vulnerable when the shock hit us. So what is the relationship between household debt and economic downturns? In this regard, there is considerable evidence pointing to the fact that household debt – especially when secured on housing assets – was at the heart of many financial crises in the past.

To this extent, Laeven and Valencia (2012) present arguments supporting the view that that the collapse of property prices features prominently the banking crises during the period 1970-2010, which has seen over 100 major banking crisis incidents throughout the global economy. Further, as a perspicuous study by Jorda et al (2015) demonstrates, the surge in credit-financed housing debt has emerged as the key contributor to banking crises for a large number of advanced economies since 1870. As their study correctly asserts

“…household debt to asset ratios have risen substantially in many countries. Financial stability risks have been increasingly linked to real estate lending booms which are typically followed by deeper recessions and slower recoveries. Housing finance has come to play a central role in the modern macroeconomy.”

Supporting arguments have been advanced by Gourinchas and Obstfeld (2012), and Borio (2014).

Theoretical approaches

The theoretical literature is lagging behind empirical work in this area. In fact, there have been relatively few theoretical attempts to examine in any detail the interaction between public and private debt. Batini et al (2016) make one such attempt[4]. They use a DSGE model (calibrated to OECD data) to examine the role of fiscal policy in supporting private deleveraging. It builds on the Kiyotaki-Moore framework of credit cycles and demonstrates how minor shocks to the economy are made significant by being amplified by credit restrictions, generating meaningful volatility in output. This model also incorporates borrowing constraints in the housing market a la Iacoviello (2005, 2010). Their model suggests that the high private debt poses more severe constraints on economic growth than what would be posed by public indebtedness. Their framework further suggests that public and private deleveraging forces combine in a deleterious way.

An alternative way to look at potential implications related to government debt reduction is to abstract from structural reasons that induce debt reduction and impose it exogenously. From this approach, there is some available, albeit highly preliminary work that looks to examine possible fiscal instruments for reduction of government debt in models of closed economies with nominal frictions. Such papers tend to introduce monetary policy in the form of Taylor-type rules.

More recently, Andres et al (2020) use a model of small open economy to examine exogenous fiscal consolidation by virtue of several fiscal instruments in the context of private deleveraging within a monetary union. The presence of long-term mortgages in their model has important insights relating to the way private spending adjusts to exogenous fiscal and other types of shocks. Non-linearities arise because of long-term debt contracts, in part due to multipliers associated with fiscal retrenchments. This establishes a theoretical link between the timing and intensity of fiscal adjustments, the duration of private deleveraging processes and, therefore, of recessions.

Final thoughts

This blog entry considered the role of private debt in determining the impact of a financial crisis. Evidence suggests that high household debt levels aggravated crises, at least in advanced economies. Recessions were deeper, and recoveries were weaker when they were preceded by larger surges in household debt. Analogous observations were made for booms and busts in the housing market. Housing bust episodes that came before more substantive run-ups in gross household debt correspond to busts that are deeper, recoveries that are weaker, and household deleveraging that is more conspicuous.

The fall in economic activity was too discernible to be merely a symptom of a decline fall in house prices. Neither was it the result of banking crises alone. Instead, the severity of the contraction appears to be better explained by the coalescence of two processes: the decline in house prices and the pre-bust leverage. These observations are consistent with the predictions of recent theoretical models that attempt to explain how financial shocks might generate deep and prolonged recessions. We should take lessons from that, especially if anticipated COVID-19-related job losses result in a substantive upward shift in household debt relative to income and assets.

———————————————

Footnotes

[1] See Christiano et al (2011); Woodford (2011); Geanakoplos (2010, 2012); Geanakoplos et al (2012); etc

[2] All estimates from the Institute for International Finance, latest available data March 2020, www.iif.com

[3] For clarity: the main thrust of the Checherita and Rother (2010) argument remains i.e. that there is a negative correlation between external debt and economic growth and that this correlation becomes particularly strong when debt reaches a certain threshold. But their secondary claim (i.e. that the stock of private debt, defined as total domestic credit to the private sector, fails to be statistically significant in determining growth) does not old any water once panel/cross-sectional time-series data is estimated using a more robust specification.
[4] Paper presented at Birkbeck’s inaugural BCAM Policy Conference, March 2017.