The balance sheet adjustment in the household sector has been a prominent feature of the last US recession and subsequent recovery. The beginning of the economic downturn in late 2007 broadly coincided with a sustained reduction in household liabilities relative to income – that is, household deleveraging – which contrasted with the strong build-up of debt before the crisis. From a peak of around 129% in the fourth quarter of 2007, the household debt-to-income ratio fell by 26 percentage points to around 104% in the fourth quarter of 2013, led by sustained declines in mortgage debt. While there is broad-based agreement that household deleveraging has acted as an important drag on the recovery, the lack of an obvious benchmark to which the debt ratio should converge makes the assessment of progress on balance sheet repair quite challenging. History appears to be of little guidance regarding the adjustment needs in the current cycle, as the recent swings in the household debt-to-income ratio are unusual by the standards of previous recessions (Figure 1).
Figure 1. Developments of the household debt-to-income ratio over current and past business cycles
(index, start of recession = 100; x-axis: quarters)
Source: Federal Reserve Board and authors’ calculations.
Notes: Zero marks the start of each recession. According to the NBER, there have been 10 recessions in the United States since 1950, with the last one starting in 2007Q4.
New methodology to model household equilibrium debt
In a new paper we propose a novel approach to examine the question of how far US household indebtedness stands from its sustainable level at any point in time, by estimating a time-varying equilibrium household debt-to-income ratio determined by economic fundamentals (Albuquerque, Baumann, and Krustev 2014). This approach allows us to assess whether household indebtedness moved beyond what was suggested by its fundamentals during the recent credit boom, as well as to track progress in household deleveraging in the current phase of balance sheet adjustment.
We model the US household debt-to-income ratio in a panel error correction framework. We employ the Pooled Mean Group (PMG) estimator developed by Pesaran, Shin and Smith (1999) – adjusted for cross-section dependence – for a panel comprising the 50 US states (plus the District of Columbia) over the period 1999Q1 to 2012Q4. The data come from the Federal Reserve Bank of New York’s (FRBNY) Consumer Credit Panel, a nationally representative sample drawn from anonymised Equifax credit data.2 In line with the related literature, the long-run dynamics of the household debt-to-income ratio are modelled as a function of:
- wealth (proxied by the house price-to-income ratio),
- the cost and availability of credit (proxied by the nominal interest rate on conventional mortgages and the loan-to-value ratio),
- the collateral available for borrowing (proxied by the homeownership rate),
- income expectations and uncertainty (proxied by the unemployment rate), and
- the demographic structure of the population (where we use the share of 35–54 age group in total population).
The model estimates point to a stable long-run relationship between the debt-to-income ratio and the explanatory variables. The difference between the actual and estimated equilibrium debt-to-income ratio (determined by the long-run relationship) is interpreted as deviations from ‘sustainable/equilibrium’ levels, the so-called debt gap.
Our results show that the evolution of the debt gap went through a number of stages (Figure 2). The debt-to-income ratio in the US household sector was broadly in line with what was suggested by equilibrium debt up to around 2002–2003. Since then, a positive debt gap started to emerge as actual debt rose at a faster pace than equilibrium debt. After mid-2007, the widening of the debt gap was reinforced by a decline in equilibrium debt reflecting deteriorating fundamentals, such as lower house prices, higher uncertainty, more pessimistic income expectations, and reduced collateral availability. These factors were partially offset by lower mortgage rates. The debt gap reached its peak in late 2008, broadly coinciding with the peak in the actual debt-to-income ratio. Thereafter, the gap began to shrink due to stronger deleveraging undertaken by households that outweighed the decline in the equilibrium debt ratio.
More recently, in the course of 2012 and 2013 the gap has shrunk not only due to on-going household deleveraging, but also reflecting a gradual stabilisation and subsequent rise in the equilibrium debt-to-income ratio. The latter is due to improving fundamentals – particularly rising house prices – and a declining unemployment rate. By the end of 2013, the gap between actual and equilibrium debt stood at less than three percentage points, suggesting that the deleveraging process may have ended or is about to end (given the uncertainty around the estimate of equilibrium debt).
Figure 2. Actual and equilibrium debt-to-income ratio and implied gap
Source: FRBNY/Equifax Consumer Credit Panel and authors’ calculations.
Note: Last observation refers to 2013Q4.
The aggregate results mask some heterogeneity across US states. At the time when the national debt gap was at its peak (2008Q4), all US states had a debt gap above zero, although with different deleveraging needs (Figure 3). The synchronised balance sheet adjustment across states carried out since then implied that, by the end of 2012, the number of states with severe household debt imbalances diminished markedly, in particular in those states with pronounced boom-bust cycles in their housing markets (Arizona, California, Nevada, and Florida), while several other states appeared to no longer face deleveraging pressures (shown in light blue). Heterogeneity across states, however, continued to be present.3
Figure 3. Debt gaps across US states
Source: Authors’ calculations.
Note: The debt gap measures the difference between the actual debt-to-income ratio and the estimated equilibrium ratio for each state. A positive debt gap indicates a need for balance sheet adjustment due to the actual ratio being above the estimated equilibrium ratio. The states of Alaska and Hawaii are not shown for convenience.
The build-up in indebtedness in the US household sector since early-2000 – and the subsequent balance sheet adjustment that began later in the decade – are unprecedented by the standards of previous business cycles. After roughly 5 years of balance-sheet adjustment, the process of deleveraging in the US household sector appears to be broadly completed as of the end of 2013. Although the household debt ratio has started to increase again in the last two quarters of 2013, as evidenced in the FRBNY’s Household Debt and Credit Report, this will not necessarily lead to a widening in the debt gap. In fact, provided that the increase in the debt ratio continues to be accompanied by a rise in equilibrium debt, supported by the on-going US economic recovery, the debt gap could remain largely unchanged. On the other hand, a normalisation of monetary policy and a return to a higher interest rate environment might pose some challenges to the deleveraging process in the future by pushing down the sustainable debt-to-income ratio.
Disclaimer: The views expressed in this column should not be reported as representing the views of the European Central Bank (ECB) or of the Eurosystem. The views expressed are those of the authors and do not necessarily reflect those of the ECB or of the Eurosystem.
1. See for example Mian and Sufi (2010) and Mian, Rao, and Sufi (2013) on the link between household leverage and activity, and Eggertsson and Krugman (2012), and Koo (2008, 2013) on the notion of the ‘deleveraging crisis’ and ‘balance sheet recessions’.
2. Household debt as reported in the Consumer Credit Panel (CCP) draws on detailed anonymous Equifax credit-report data and as such differs somewhat from the Board of Governors’ Flow of Funds (FoF) Accounts measure (shown in Figure 1). The FRBNY’s CCP measure of household debt to personal disposable income is on average 15 percentage points lower than the Flow of Funds measure (1999–2012), but shares broadly the same time dynamics (the correlation between the two series is 98.2%) and the major components – home mortgage debt and consumer credit debt – are directly comparable. In our empirical analysis, we scale the FRBNY’s debt measure to personal income.
3. The sample analysed in Albuquerque, Baumann, and Krustev (2014) goes from 1999 to 2012. However, for the purpose of this column, we have updated our estimates at the US aggregate level to the end of 2013. Data available to us at the regional level end in 2012Q4.
Albuquerque, B, U Baumann, and G Krustev (2014), “Has US Household Deleveraging Ended? A Model-Based Estimate of Equilibrium Debt”, ECB Working Paper 1643.
Eggertsson, G and P Krugman (2012), “Debt, deleveraging, and the liquidity trap: a Fisher-Minsky-Koo approach”, Quarterly Journal of Economics, 127(3): 1469–1513.
Koo, R (2008), The Holy Grail of Macroeconomics – Lessons from Japan's Great Recession, John Wiley & Sons, New York.
Koo, R (2013), “Balance sheet recession as the ‘other half’ of macroeconomics”, European Journal of Economics and Economic Policies: Intervention, 10(2): 136–157.
Mian, A, K Rao, and A Sufi (2013), “Household balance sheets, consumption, and the economic slump”, Quarterly Journal of Economics (forthcoming).
Mian, A and A Sufi (2010), “Household leverage and the recession of 2007–09”, IMF Economic Review, 58(1).
Pesaran, M, Y Shin, and R Smith (1999), “Pooled mean group estimation of dynamic heterogeneous panels”, Journal of the American Statistical Association, 94(446): 621–634.