Accounting for Debt Service The Painful Legacy of Credit Booms

BIS Working Papers

No 645


Accounting for debt service: the painful legacy of credit booms

by Mathias Drehmann, Mikael Juselius and Anton Korinek


Monetary and Economic Department

June 2017


JEL classification: E17, E44, G01, D14

Keywords: new borrowing, debt service, financial cycle, real-financial linkages


BIS Working Papers are written by members of the Monetary and Economic Department of the Bank for International Settlements, and from time to time by other economists, and are published by the Bank. The papers are on subjects of topical interest and are technical in character. The views expressed in them are those of their authors and not necessarily the views of the BIS.

This publication is available on the BIS website (


© Bank for International Settlements 2017. All rights reserved. Brief excerpts may be reproduced or translated provided the source is stated.

ISSN 1020-0959 (print)

ISSN 1682-7678 (online)

Accounting for Debt Service The Painful Legacy of Credit Booms

Mathias Drehmann (BIS) Mikael Juselius (Bank of Finland) Anton Korinek (JHU and NBER)



When taking on new debt, borrowers commit to a pre-specified path of future debt service. This implies a predictable lag between credit booms and peaks in debt service which, in a panel of household debt in 17 countries, is four years on average. The lag is driven by two key features of the data: (i) new borrowing is strongly auto-correlated and (ii) debt contracts are long term. The delayed increase in debt service following an impulse to new borrowing largely explains why credit booms are associated with lower future output growth and higher probability of crisis. This provides a systematic transmission channel whereby credit expansions can have long-lasting adverse real effects.


JEL Codes: E17, E44, G01, D14

Keywords: new borrowing, debt service, financial cycle, real-financial linkages

We would like to thank Larry Ball, Claudio Borio, Stjin Claessens, Jon Faust, Andreas Fuster, Hyun Song Shin as well as participants at seminars at the Federal Reserve Board, the International Monetary Fund, the Bank for International Settlements, the Bank of Finland and Johns Hopkins for helpful comments and suggestions. Part of this research was performed when Korinek was a Research Fellow at the BIS. The views presented here are the authors’ and do not necessarily reflect those of the Bank for International Settlements and the Bank of Finland.

Contact information: Wyman 531, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218. Phone: +1-240-575-0995. Email: [email protected]


1 Introduction

Debt service is the inescapable counterpart to borrowing. When taking on new debt, borrowers increase their spending power in the present but commit to a pre-specified future path of debt service, consisting of interest payments and amortizations. In the presence of long-term debt, keeping track of debt service explains why credit-related expansions are systematically followed by downturns several years later. This paper describes the lead-lag relationship between new borrowing and debt service analytically and shows empirically that it provides a systematic transmission channel whereby credit expansions lead to future output losses and higher probability of financial crisis.

We begin by providing a simple accounting framework that describes how new bor­rowing generates a well-specified schedule of debt service. When new borrowing is auto-correlated and debt is long term - features that are present in the real world - we demonstrate two systematic lead-lag relationships: First, debt service peaks at a well-specified interval after the peak in new borrowing. The lag increases both in the maturity of debt and the degree of auto-correlation of new borrowing. The reason is that debt service is a function of the stock of debt outstanding, which continues to grow even after the peak in new borrowing. Second, net cash flows from lenders to borrowers reach their maximum before the peak in new borrowing and turn negative before the end of the credit boom, since the positive cash flow from new borrowing is increasingly offset by the negative cash flows from rising debt service.

Using a panel of 17 countries from 1980 to 2015, we empirically confirm the dynamic patterns identified in the accounting framework. We focus on the household sector since long-term debt contracts are most prevalent in this sector, especially for mortgages. But we also draw comparisons with the corporate sector at several points in the paper. Building on a new BIS database (Drehmann et al (2015)) we obtain time series of new borrowing and debt service. We show that new borrowing is strongly auto-correlated over an interval of six years. It is also positively correlated with future debt service over the following ten years. In the data, peaks in debt service occur on average four years after peaks in new borrowing.

Next, we investigate the implications for the real economy leading to three key findings. First, we document that new household borrowing has a clear positive impact, and its counterpart, debt service, a significantly negative impact on output growth, both of which last for several years. Together with the lead-lag relationship between new borrowing and debt service this implies that credit booms have a significantly positive output effect in the short run, which reverses and turns into a significantly negative output effect in the medium run, at a horizon of five to seven years.

Second, using a novel decomposition method, we demonstrate that most of the neg­ative medium-run output effects of new borrowing in the data are driven by predictable future debt service effects. Our results therefore provide a systematic transmission mechanism for earlier findings in the literature on the real effects of credit booms.1

Third, we also show that debt service is the main channel through which new bor­rowing affects the probability of financial crises. Consistent with a recent literature that has documented that debt growth is an early warning indicator for financial crises, we find that new borrowing increases the likelihood of financial crises in the medium run.2 Debt service, on the other hand, negatively affects the likelihood of crises in the short turn. Taken together and performing a similar decomposition as before, we show that the negative effects of the future debt service generated by an increase in new borrowing nearly fully accounts for the increase in crisis probability.

Our results are robust to the inclusion of range of control variables as well as changes in sample and specification. Our baseline regressions control for interest rates and wealth effects. The results do not change when we control for additional macro factors, including credit spreads, productivity, net worth, lending standards, banking sector provisions and GDP forecasts, nor when we consider sub-samples of the data, e.g. a sample leaving out the Great Recession, or allow for time fixed effects. And despite at most 35 years of data, the relationships even hold at the country level.

We also compare new borrowing and debt service to more traditional measures of credit booms used in the macro-finance literature. Conceptually, new borrowing and debt service are easy to interpret economically as they enter budget constraints directly and therefore capture the economic forces responsible for the delayed adverse real effects of credit booms in a straightforward way. From a statistical point of view, they do not show pronounced trends in contrast to traditional credit-to-GDP aggregates, which have been trending upward for several decades. This is an advantage as it avoids arbitrary de-trending methods like HP-filters or linear trends typically used in the literature. By running a horse race for predicting output or crises, we find empirically that debt service in particular outperforms the traditional measures, both by generating higher R2 statistics and by exhibiting higher t-statistics in regressions in which both variables are included.

1The negative medium-run effect of new borrowing on growth is documented e.g. by Mian and Sufi (2014), Mian et al. (2013, 2017) and Lombardi et al. (2016). Claessens et al. (2012), Jorda et al. (2013), and Krishnamurthy and Muir (2016) document a link between credit booms and deeper recessions.

2See e.g. Borio and Lowe (2002), Reinhart and Rogoff (2009), Schularick and Taylor (2012), and Drehmann and Juselius (2014), among others.

We believe that the transmission mechanism from new borrowing to debt service and real economic activity that we document in this paper is of great relevance for developing realistic models and policies to deal with credit booms and busts.3 For ex­ample, our results highlight a potential trade-off when trying to stimulate the economy by encouraging the expansion of debt. New borrowing has positive effects in the short run, but as it will mechanically increase debt service in the future, these benefits may be offset by the associated drag in growth in the medium run. Equally, this trade-off has potential implications for using monetary policy to lean against the wind as dampening a credit boom with higher policy rates may weaken growth in the short run but avoid higher debt service - and low output and higher crisis risk- in the medium run.4 More broadly, our results show that policy needs to take into account contractual features that affect future debt service and, thus, have a predictable effect on economic activity.

The paper is structured as follows. In the ensuing section, we provide a simple accounting framework to illustrate the main channels at work. In Sections 3 and 4, we discuss the data and document the lead-lag relationship between new borrowing and debt service. In Section 5 we turn to the real implications of debt service and describe the transmission channel from new borrowing to debt service and, in turn, to economic activity and crisis risk. Section 6 concludes. Appendix A contains proofs and further analytic results for the accounting framework and the decomposition of impulse response functions. In AppendixB, we describe the empirical methodology that we develop to decompose what fraction of the future real effects of new borrowing is explained by debt service.

3A recent strand of literature has emphasized the potential aggregate demand effects of flows of resources between borrowers and lenders. See e.g. Farhi and Werning (2016), Korinek and Simsek (2016), and Schmitt-Grohe and Uribe (2016). However, this literature assumes one period debt and is thus silent on the lag structure between credit booms and debt service.

4Juselius et al (2017) develop this theme further by introducing debt service and leverage into a standard reduced form model of the economy. They run counterfactual simulations and conclude that a monetary policy rule that takes financial developments systematically into account during both good and bad times could help dampen the financial cycle, leading to significant output gains and little change in inflation.

2 Accounting for Debt Service

This section lays out a simple accounting framework that clarifies the key mechanism underlying the lead-lag relationship between new borrowing, future debt service, and the net cash flows between borrowers and lenders. The framework highlights the key roles of auto-correlated new borrowing and long-term debt contracts, both of which are present in the data, in generating an interesting lead-lag structure.

Accounting framework Consider a borrower who borrows Bt in period t in long­term debt. Assume, for simplicity, a constant amortization rate 6 and fixed interest rate r. In period t + 1, this contract implies a debt service of (r + 6) Bt, consisting of interest payments and amortization, and a remaining stock of debt outstanding of (1 — i) Bt, which is carried over to the next period. After k periods, a balance of (1 — i)k-1 Bt is left of the original amount borrowed, implying debt service obligations of (r + i)(1 — i)k-1 Bt.

The total stock of debt outstanding at the beginning of period t, Dt, follows the law-of-motion


Hence, the stock of debt can be represented as a moving average of past new borrowing (Bj )j=0.

Total debt service, St, is given by the debt service obligations from all past borrowing (Bj )j=0 that are due in period t, or equivalently, on the stock of debt Dt,


The net cash flow from lenders to the borrowers in a given period t consists of the new borrowing  Bt minus  all the debt service obligations due in period


Observe that the standard case of short-term debt corresponds to i =1. In that case, the above formulas reduce to Dt = Bt-1, St = (1 + r) Bt-1 and Nt = Bt — (1 + r) Bt-1. In other words, with short-term debt, it is unnecessary to distinguish between new borrowing in the previous period and the stock of debt in the current period.

Accounting implications of a credit boom We now use these accounting rela­tionships to trace out the implications of a boom in new borrowing in long-term debt for the lag structure between borrowing and debt service.

Consider an exogenous process of new borrowing {Bt} which involves T > 2 periods of new borrowing Bt > 0 for t G {1, ...T} and that is hump-shaped, i.e. there is a unique interior peak at a time 1 < t* < T such that Bt* = maxte{1,...T} {Bt} and borrowing is increasing up until the peak B1 < B2 < ... < Bt* and decreasing after the peak Bt* > Bf+1 >...> Bt .

For expositional simplicity, we maintain the assumptions of constant interest and amortization rates. Furthermore, we impose a mild condition on timing: the process of new borrowing up until the peak t* cannot be too drawn out over time, captured by the analytic condition (i + r) t* < 1. After T, we assume no further borrowing so Bt = 0 for t > T.

Given these assumptions, we find the following relationships between new borrowing and debt service:

Proposition 1 (Lead-lag structure of new borrowing and debt service). (i)

The peak in debt service t occurs after the peak in new borrowing t*. The lag between the two peaks t* —t is weakly decreasing in the amortization rate i. The net cash flow from lenders to borrowers peaks weakly before the peak in new borrowing and turns negative after the peak in new borrowing but weakly before the end of the credit boom.

The formal proof of the proposition is given in Appendix A.1 but the intuition is straightforward. For part (i) of the proposition, observe that debt service is a function of the stock of debt, or technically speaking, debt service is a moving average of new borrowing. When new borrowing peaks, the stock of debt and thus debt service is still increasing, since new borrowing is still positive and existing debt depreciates at the comparatively low rate of i. After the peak in new borrowing, a lower amortization rate pushes back the time when debt service outweighs the positive (but declining) effects of new borrowing, which moves the peak in debt service further away from the peak in new borrowing.

For part (ii) of the proposition, observe that at the peak of new borrowing, where the growth rate of new borrowing is zero, debt service is still increasing. This implies that the the difference between the two, i.e. the net cash flow from lenders to borrowers, is decreasing and must have already peaked. At some point, the net cash flow turns negative since debt service becomes greater than new borrowing. As long as the credit boom is not too drawn out, this happens after the peak in new borrowing. Furthermore, it happens before the end of the credit boom - once the boom is over and there is no more new borrowing, the net cash flow consists entirely of debt service and must be negative.

Some of the results in the proposition are stated as weak inequalities due to the dis­crete time nature of our framework. Appendix A shows that in an equivalent continuous time framework all of the stated inequalities hold strictly.

Figure 2 illustrates our findings. We assume that new borrowing (light-blue bars) is given by an exogenous bell-shaped process that starts at t0 and lasts for 9 periods, with a peak at t = 3.5 The beige bars depict the resulting debt service obligations, which continue to grow even when new borrowing is already declining. The black line depicts the net cash flow from lenders to borrowers, i.e. the difference between new borrowing and debt service. In line with Proposition 1, the net cash flow peaks before the peak in new borrowing and turns negative before the boom is over.

Analytic results for a unit impulse in new borrowing Although new borrowing in the data is typically a bell-shaped process during credit booms, it is useful to consider the special case of a unit impulse in new borrowing that decays exponentially. This  process allows us to obtain analytic results for the timing of the peak in debt service. It also corresponds to the way that shocks are typically modeled in theoretical models.

5For illustration purposes, we set r = 5% and S = 15% in this simulation.


Assume that there is a unit impulse to new borrowing at time 0 that decays ex­ponentially at rate p G [0,1). As a result, new borrowing at time t is Bt = p*. This process of new borrowing is a limit case of the class of credit boom processes covered by Proposition 1 that is shifted by one period, with t* = 0 and T ^ rc>. The results of the proposition therefore still apply, but they can be sharpened by obtaining analytic expressions for the timing of the peak in debt service.

The debt stock resulting from a unit impulse in new borrowing is a moving average given by the geometric sum


Proposition 2 (Peak in debt service). Following a unit impulse of new borrowing that decays at rate p = 1 — 8 with p, 8 G (0,1), debt service peaks at an integer time index in the interval (t ± 1) where


which satisfies di/dp > 0 and dt/d8 < 0.6

6In the special case p =1 — S, the geometric sum for Dt is given by p (t +1), which is maximized at t = —1/ ln p — 1.

As in the previous proposition, our discrete time setup implies that we can only obtain an interval (t ± 1) for the peak. The appendix provides a proof and shows that an equivalent proposition for a continuous time version of our model delivers a precise value for t.

Intuitively, the proposition captures that a higher amortization rate 8 leads to an earlier peak in debt service since debt is paid off more quickly. Similarly, a higher auto-correlation p leads to a later peak in debt service since borrowers continue to accumulate debt for a longer period.

To showcase that both long-term debt (8 < 1) and auto-correlated new borrowing (p > 0) are necessary to obtain an interesting and non-degenerate lead-lag structure, it is useful to consider the two extremes 8 = 1 and p = 0:

Corollary 3 (Necessity of both auto-correlation and long-term debt). If either 8 = 1 or p = 0, the lag between an impulse to new borrowing and the peak in debt service becomes degenerate and collapses to t = 1.

The case 8 = 1 captures one-period debt contracts as is typically considered in theory models (see the left-hand panel of Figure3 for an illustrative example). New borrowing is still autocorrelated and continues to be given by Bt = p* - after the initial unit impulse at t = 0, it decays slowly. Debt service is given by St = (1 + r)pt-1 for t > 1, and is simply the mirror image of new borrowing lagged by one period. Intuitively, since any new borrowing is immediately paid off in the following period, there is no prolonged lead-lag relationship between new borrowing and debt service. Given that new borrowing peaks at t = 0, debt service peaks at t =1.

The case p = 0 captures a unit impulse to new borrowing without auto-correlation (centre panel, Figure 3). In that case, no new borrowing occurs after the initial impulse. Hence, the stock of debt peaks at t = 1, i.e. in the period right after the impulse to new borrowing, and is declining immediately after. Debt service, given by St = (r + 8)(1 — 8)t-1 for t > 1, follows the same pattern and also peaks at t =1.

The case with auto-correlation (p > 0) and long term debt (8 < 1) is illustrated in the right-hand panel of Figure 3. In this case, we obtain a non-degenerate lag rela­tionship between the peak in new borrowing and the peak in debt service, as described by the corollary. This case is rarely considered in theory papers but is empirically the most relevant .

In summary, our simple accounting framework thus suggests that it is the combined effects of auto-correlated new borrowing and long-term debt that account for the sub­stantial lags between peaks in new borrowing and debt service. The key empirical issues that we address in the remainder of this paper is to document that this relationship holds in the data and to investigate to what extent the lagged response of debt service can account for delayed negative real effects of credit booms.


3 Data and Measurement

Our main variables of interest are new borrowing and debt service. This section dis­cusses how we measure both variables in the aggregate, which variables we use to measure the real effects of credit booms and what controls we employ. We use an unbalanced panel of annual data from 17 countries from 1980 to 2015.7The exact definitions, sources, and availability for all variables are listed in Tables 3 and 4 in Appendix D.

In the main text, we focus on the household sector for three reasons. First, this is the sector in which long debt maturities and auto-correlated new borrowing are most prevalent, giving rise to the most interesting lead-lag relationships. Second, in doing so, we also complement a literature that has demonstrated negative effects of household debt in the medium run (e.g. Jorda et al (2016) or Mian et al(2017)) and show that their results arise from the lead-lag relationship between new borrowing and debt service that our paper identifies. Third, data availability on debt maturities is considerably better in the household sector compared to the corporate sector. For completeness, we also report a summary of results for the corporate sector in Appendix C.

New borrowing and debt service We construct measures of debt service and new borrowing using three main data series as inputs. The first input series is the stock of household debt Di;t in country i at time t from the BIS database compiled by Dembiermont et al (2013). This variable captures credit to the household sector from all sources, including bank credit, cross-border credit and credit from non-banks. The
second series is total interest paid by households, Ri;t, from national accounts.8 The third is debt maturity estimates from the BIS database on debt service by Drehmann et al (2015) that is publicly available from 1999 onward.

7Countries are Australia, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, Korea, the Netherlands, Norway, Portugal, Spain, Sweden, the United Kingdom and the United States.

8As in Drehmann et al (2015), we also include financial intermediation services indirectly measured (FISIM) in our measure of Ri,t. FISIM is an estimate of the value of financial intermediation services provided by financial institutions which consumers pay as part of their borrowing costs. In the be­ginning of our sample, national accounts data on interest paid is not available for all countries. In that case, we proxy interest paid on the stock of debt by using an alternative interest rate such as the average interest rate on bank loans.

From our accounting framework we can obtain expressions for new borrowing and debt service in terms of the raw input series. Equation (1) tells us that new borrowing, Bit, equals the change in the stock of debt plus amortizations


Debt service, Si,t, is similarly given by the sum of interest and amortizations,


The average interest rate on the stock of debt can be calculated by dividing interest expenses by the stock of debt, i.e. ri)t = Ri)t/Di)t.

We estimate the amortization rate by following the methodology of Lucket (1980), Dynan et al (2003) and Drehmann et al (2015), i.e. by approximating it with the amortization rate of an installment loan with a maturity that corresponds to the average maturity m of the stock of debt outstanding.9This implies an amortization rate of



9This methodology is used by both the US Fed and the BIS to construct time series of aggregate debt service. A derivation of the formula is provided in Appendix A.3. See Drehmann et al (2015) for simulation results that show that the approximation error resulting from our formula with average maturity and interest rate does not significantly alter the dynamics of debt service compared to what would be obtained by aggregating individual loan-level data on interest rates and maturities.

With this formula, the amortization rate moves inversely to the interest rate (d8i,t/dri,t < 0) and varies substantially over time, as observed in the data.

Data on maturities is generally limited. Collecting the available evidence, Drehmann et al(2015) find that 18 years is the average maturity for total household debt across countries and time, with a minimum of 16 and a maximum of 19 years.10 As this varia­tion has a limited impact on debt service estimates, the BIS applies a constant maturity for simplicity, which we follow here. For the corporate sector, the average effective ma­turity is 13 years. But as Drehmann et al(2015) acknowledge, this is estimate is more uncertain given the importance of rollovers for corporate sector debt (e.g. Mian and Santos (2017)).

For the ensuing empirical analysis, we normalize both new borrowing and debt service by nominal GDP obtained from national accounts and take logs. We denote the resulting normalized variables by bi>t = Bi>t/Yi>t and si>t = Si,t/Yi,t. We plot these series for the household sector for the individual countries in Figure 19 in AppendixD.

Real variables We study the real implications of new borrowing and debt service by looking at two variables: output growth and the incidence of banking crises. We denote logged real GDP from national accounts by yi,t = ln(^ t/Pi t) so that real output growth is Ayi)t.

We use Laeven and Valencia (2012) for financial crisis dates and extend their dataset using additional information from central banks as discussed in Drehmann and Juselius (2014). Overall, we have 19 crisis observations, 11 of which are related to the Great Financial Crisis. As a robustness check, we also use a broader definition of crises by Reinhart and Rogoff (2009), which adds three further crisis observations prior to 2000.

Controls Since our ultimate interest lies in the real effects of new borrowing and debt service, we control for several variables that could also have real effects through private expenditure. In addition to lagged output growth, our baseline set of controls consist

  •  the real three-month money market rate,
  •  the spread between the short-term money market rate and the prime lending rate,
  •  the change in the average lending rate on household debt,
  •  the growth rate in real residential property prices, and
  •  crisis dummies,

and up to two lags of each of them. The rationale behind these controls is as follows: The (ex-post) real three-month money market rate captures the effect of real interest rates on the expenditure of financially unconstrained consumers. The spread between the short­term money market rate and the prime lending rate captures that the expenditure of constrained consumers is linked to credit spreads, as suggested by conventional models. We control for the change in the average lending rate since outstanding loans are not fully repriced in each period. Together, these controls should ensure that the debt service effects that we capture are not confounded with more conventional effects of interest rates. Finally, we use the growth rate in real residential property prices to capture changes in household wealth that might affect expenditure. In addition, the baseline specification contains a dummy that equals 1 in crisis years as well as a separate dummy that equals 1 in 2009 to control for the global financial crisis, which affected all countries even if they did not have a financial crisis.

We perform additional robustness tests by controlling for further variables that may affect household decisions. We do not include these variables in the baseline specification because data availability is generally limited. Our extended set of controls consist of:

10Drehmann et al (2015) assess the maturity of total household debt including inter alia consumer loans as well as mortgages. As an additional cross check, they confirm this result by collecting matu­rities for mortgages and consumer debt to then derive country-specific weighted averages.

  •  the growth rate in labor productivity11, the change in CPI inflation, the growth rate of unemployment, and the growth rate of the real exchange rate as additional controls for changes in the business cycle environment.12 We also add consensus forecasts for next years output growth as one proxy for future expectations,
  •  net worth calculated as the difference between household assets and household lia­bilities as recorded in the national accounts to capture net worth effects discussed in the literature,
  •  the corporate credit spread and the term spread between short tern and long term government bonds as these variables have been found to have good leading proper­ties for the business cycle (e.g. Gilchrist and Zakrajsek (2012) or Krishnamurthy and Muir (2016)).
  •  lending standards and banking sector provisions to control for potential supply side considerations stemming from the banking sector.

11Gorton and Ordonez (2016) find that shocks to productivity often start booms, and that booms are more likely to end in crisis if productivity is low.

12It is well know that CPI inflation, the unemployment rate and the real exchange rate contain sizable low-frequency components across countries. As this       can  bias their coefficients toward zero when used as regressors for a non-trending variable, such as    real   GDP growth, we use their growth rates rather than levels.

4 New Borrowing and Debt Service

In this section we document that the basic relationships postulated in the accounting framework hold in the aggregate time series in the data: new borrowing is significantly auto-correlated, and there is a robust lead-lag relationship between the peak in new borrowing and the peak in debt service.

4.1 Patterns in the raw data

New borrowing is significantly autocorrelated, with a correlation coefficient p across consecutive periods of 0.8 as illustrated in the autocorrelogram for new borrowing in the left-hand panel of Figure 4. The autocorrelation of new borrowing is positive up to six years ahead.

As the accounting framework suggests, in the presence of high autocorrelation and long-run debt, new borrowing is positively correlated with future debt service for quite some time (Figure4). New borrowing leads debt service by several periods, with the peak correlation occurring in period 4, just before the autocorrelation of new borrowing turns negative. Proposition 2 of our accounting framework implies a similar lead-lag pattern with p = .8 and 8 = 0.05.

Figure 5 showcases the phase-shift between new borrowing and debt service more clearly. It depicts the average evolution of the two variables around peaks in new borrowing (defined as local maxima within a five-year window).13 Such peaks are followed by peaks in debt service three years later. The figure also shows clearly that debt service continues to rise when new borrowing already decreases. Figure 19 in Appendix D documents that the lead-lag relationships are also present at the individual country level.

13Results are quantitatively very similar if typical business cycle dating algorithms as in Harding and Pagan (2002) are used and we impose a minimum cycle length of five years to identify credit booms.


A comparison with the corporate sector illustrates the results of the accounting framework (see Figure 16 in AppendixC). As the autocorrelation across periods is 0.4 in the corporate sector versus 0.8 in the household sector and debt has a shorter matu­rity, the lead-lag relationship between new borrowing and debt service is significantly more compressed. The cross-correlogram is at its maximum after 2 years and peaks in new borrowing are followed by peaks in debt service after 1 year. This suggests that traditional models with one period debt may capture the interactions between debt and the real economy better for the corporate sector than for households.

4.2 Effect of new borrowing on future debt service

To study the relationship between new borrowing and debt service in the data more formally, we use local projections a la Jorda (2005).14 In particular, we estimate for each horizon h projections for new borrowing, b, and debt service, s, with

14We follow the convention of using “prediction” to refer to within-sample impulse responses.

where j is a country fixed effect, controls captures our control variables, and ^hit+h, is the projection residual for j = {b, s}. The h successive [\hb and [Shb coefficients trace out the impulse response of future new borrowing and future debt service, respectively, to a unit increase in new borrowing at time t over h successive years.

The effects of an impulse to new borrowing takes several years to dissipate and is followed by a peak in debt service after four years (Figure6).

The left-hand panel shows the impulse response from a unit shock to new borrowing based on a specification that includes our baseline controls (see Section 3). New bor­rowing remains elevated for five years, with the first four being statistically significant. Immediately after the shock, debt service (right-hand panel) begins to rise significantly; it peaks after four years and remains significantly elevated even after eight years.

The persistence of new borrowing and the lead-lag relationship between new bor­rowing and debt service are very robust in the data. The patterns remains the same if we add time fixed effects (orange line with triangles, Figure 7) to the baseline specifica­tion or exclude the run-up and aftermath of the great financial crisis (green line with diamonds). It also remains the same if we include the additional controls, including credit spreads or changes in net-worth, from Section 3 in the specification (light blue line with circles).


The accounting relationship between lower autocorrelation (and lower maturities) and less time between the peak new borrowing and the peak in debt service is also borne out by the robustness checks. If we add the additional controls, shocks to new borrowing are less autocorrelated, which is largely due to decreased sample size. The impulse of new borrowing only remains statistically significantly positive for three years after which it turns negative. Given decreased persistence of new borrowing, the accounting framework would suggest that debt service should also peak earlier, which is indeed the case. Similarly, the autocorrelation is lower and maturities are shorter for the corporate sector. For this sector, we estimate that the impulse response of debt service following an increase in new borrowing peaks more quickly and the significant impact decays more rapidly compared to the household sector (Figure 17 in AppendixC). As the estimated impulses embed all factors that dynamically respond to the initial debt impulse, including changes in policy rates, these findings underscore the robustness of the accounting framework.

5 New Borrowing, Debt Service and Real Activity

In this section, we investigate the empirical effect of new borrowing on future real economic activity. We show that debt service represents the main transmission channel through which new borrowing affects subsequent output growth and the probability of crisis.

5.1 Effects on future output growth​​​​​​​

To shed light on the link between new borrowing, b, debt service, s, and real output growth, Ay, we estimate local projections of the form


for increasing values of h. The estimates of ^ and [^ for successive values of h trace out the impulse response of GDP growth from unit increases in new borrowing and debt service, respectively.

Figure 8 shows that new household borrowing predicts a slowdown in output growth with a sizable delay, whereas an increase in household debt service has immediate negative effects.15GDP growth significantly increases by around 10 basis points for the first two years following a percentage increase in new debt, after which it declines and becomes 10 basis points lower than normal in years 5 to 7 (Figure 8, left-hand panel). While these may seem like small numbers, peaks in new borrowing are on average between 5 to 10 percentage points higher than normal across countries, implying eventual losses of up to 1 percentage point of GDP growth. The negative effects of new borrowing at medium horizons are in line with the output responses to a unit change in the credit-to-GDP ratio that has recently been documented in the literature (see e.g. Mian et al. (2017)).

15Detailed estimates from several different specification at the 1, 3 and 5 year horizons are reported in Table 5 in Appendix D.


In contrast, the local projection on GDP growth of a unit increase in household debt service (right-hand panel) is large and significantly negative for the first three years and then starts to decline.16 On impact, a unit increase in debt service decreases GDP growth by 25 basis points. This is also large as peaks in debt service are on average between 2 to 10 percentage points above normal across countries. This result is novel and highlights the value added of debt service in the presence of long-term debt contracts for understanding debt dynamics and their impact on the real economy.

The estimated output effects of new household borrowing and household debt ser­vice are robust to alterations in both sample and specification (Figure 9). In particular, the impact of debt service on output is always negative and highly significant at short horizons, whether we use only data up to 2005, add time fixed effects or additional con­trols. The results for new borrowing are somewhat more mixed. While the immediate year impact on output is always estimated to be positive, this effect is not significant for all specifications. In contrast, the significant and negative medium-term effects of new borrowing are a very robust feature of the data.

These results raise the question about long-run effects of household credit booms. Adding up the yearly effects displayed in Figure 8 (left-hand panel) suggests that the cumulative net effects of an impulse to new borrowing on output over the eight year period are negative. A similar picture also emerges from the robustness tests. This suggests that household credit booms and the ensuing predictable busts may have long-run deleterious effects on the level of output.17 mirroring e.g. Cerra and Saxena (2008) who find that crises that follow credit booms are associated with permanent output losses.

16This finding complements micro level evidence in e.g. Olney(1999), Johnson and Li (2010), and Dynan (2012) who document negative effects from debt service burdens on household expenditure.

17For a further elaboration on this point see Juselius et al. (2017).


From a high-level perspective, dynamics are similar for the corporate sector (see Figure 18 in AppendixC). Yet, coefficient estimates are lower and only weakly signifi­cant.18 This is in line with Mian et al (2017) who found little impact of corporate debt on GDP. In light of the accounting framework and our previous results, it is nonetheless interesting to see that the medium-term negative effect of new borrowing on GDP occur earlier for the corporate sector (the minimum is in year 4) than for the household sector (the minimum is in year 6). But given the weak impact of new corporate borrowing on GDP, we concentrate on household debt for the remainder of the paper.

The timing and delay between the impulse to new borrowing and the negative response of GDP growth closely match the timing between peaks in new borrowing and peaks in debt service (see Figure 6). This suggest that the negative effects of debt may be related to rising debt service in line with our accounting framework. We study this question next.

18Significance levels are also somewhat sample dependent. For example, looking at expanding sam­ples from 1995 to the end shows that in most samples new corporate borrowing has a significant negative impact in year 4 but not so if we end in the direct aftermath of the dot-com bubble. Cor­porate debt service, on the other hand, has a significant negative impact on GDP growth in the next year if the sample ends before the Great Financial Crisis. Corporate debt service has also a strong negative impact on next year’s investment growth that is stable across samples.

5.2 A novel method for decomposing local projections​​​​​​​

To assess how much of the negative effects of new borrowing flow through debt service, we decompose the impulse response functions. We provide an intuitive description of our decomposition method in the following and develop a detailed formal description in Appendix B.

The local projections on GDP growth trace out the impulse response function that, after the first round, embodies all factors that dynamically respond to the initial debt impulse and feed into the real economy, including the future debt service obligations that it generates. Hence, they capture the “net effect” of the debt impulse. From equation (7) the net effect at time t + h is


The part of the net effect that goes via debt service can be calculated in two steps. First, for any prediction horizon h > 1, we regress debt service at time t + h — 1 on new borrowing (and controls) at time t as in equation (6). The coefficients on new borrowing, flSh-1, from this regression tell us how debt service in t + h — 1 changes due to a unit increase in new borrowing at t. Second, we know from estimating equation (7) that the direct effect of debt service at t + h — 1 on output growth at t + h is flyjs. Combining these estimates, we calculate the “debt service effect” as:


The other effects will include, for example, any direct effects of new borrowing on output. If these are relevant at the aggregate level, we would expect them to be positive.

5.3 Decomposing the effect of borrowing on future output

The decomposition shows that increasing debt service can, to a large degree, account for the delayed slowdown in GDP growth following an initial increase in new household borrowing. This can be seen from Figure 10 which reports the net effect (black line), the debt service effect (beige bars) and other effects (light-blue bars) given the baseline specification up to 8 years after the debt impulse.

The dynamics of these effects largely follow the predicted patterns from our ac­counting framework (Figure 2). Following an increase in new borrowing at t0, the other effects are generally positive and decline over time. This is broadly in line with what one would expect from new borrowing and the fact that new borrowing is autocorre­lated. The only real exception occurs in years 6 and 7 where the other effects are visibly negative. In contrast, the effects of debt service are always negative and increase until they reach their peak in year 5. The net effect - or the effect of net cash flow in terms of our accounting framework - turns negative in the third year. This is between the peak in new borrowing and the peak in debt service as our accounting framework predicts. The minimum net effect occurs in year 6 or in year 5 if one disregards the additional negative other effects.


The estimated service effect and other effects of the decomposition are robust. Fig­ure 11 compares the baseline decomposition (top left-hand panel) to those obtained from several alternative specifications:19

Results from pre-2005 sub-sample (middle panel, top row) suggest that the results are neither driven by the Great Recession nor the boom that preceded it. As shown in Figure 9 the negative medium-term effects of new borrowing are even stronger in this sample than in the baseline specification. But so is the debt service effect. This may partly reflect the inability of central banks to lower nominal interest rates much below the zero lower bound in the most recent decade (e.g. Korinek and Simsek (2016)). Adding time fixed effects also does not change the picture qualitatively (right-hand panel, top row).

The debt service effect is very robust to the set of control variables, whereas the effect of new borrowing varies a bit more with the specification (second row). For example, if we include the additional controls for net worth, credit spreads, provisions of the banking sector and additional business cycle indicators listed in Section 3, the effects of the new borrowing become slightly stronger in year one, yet turn more quickly downwards and reach lower levels in year 5. Given unchanged service effects, the other effects become now more strongly negative. Yet, the service effect continues to explain around 50% of the negative effects. Controlling for lending standards produces a similar picture.20Equally, controlling only for lags of GDP growth leaves the debt service effect unaffected whilst leading to stronger net-effects than in the baseline specification. These robustness checks are very reassuring as the decomposition could be misleading if the controls, or the lack thereof, bias our coefficient estimates.

20Data in this case are very limited. With the exception of the US, lending standards have only been recorded since 2003 or even 2006 if they are available at all. We therefore only show local projections up to horizon 5.

19We continue to add our baseline controls in all robustness checks, unless otherwise stated.


The results in Figure 11 also suggest that the negative other effects of new borrowing in the medium run are state dependent (third row). For instance, financial crises seem to account for some of the negative effects of credit growth in the medium run. Once we add a country and time specific dummy for the crisis years and the next two years thereafter, the negative other effects in years 6 and 7 almost fully disappear. Equally, if debt service is low at t0, new borrowing has far less negative net effects in the medium run than in the baseline specification (or than if debt service is originally high). But as the initial conditions do not seem to matter for the debt service effect, the other effects are positive in this case.

There also seem to be no other negative effects of new borrowing if we consider the impact on real consumption growth rather than GDP growth (left-hand panel, lower row). In this case, the effects of new debt and debt service are more pronounced as could be expected given our focus on household debt. Interestingly, the estimated decomposition looks virtually identical to our analytical results with long-term debt when shocks to new borrowing are autocorrelated (as simulated in Figure 3).

It is also interesting to relate our results to previous findings in the literature. Mian et al. (2017) find that credit-booms - defined by three-year growth in the household debt-to-GDP ratio - can predict subsequent slowdowns in GDP growth. To compare our results with theirs, we replace new borrowing with the three-year growth rate in debt-to-GDP in our set-up(9). The predicted negative effects on future GDP growth from a unit increase in the three-year growth rate in debt-to-GDP - corresponding to our net effect - documented by Mian et al. (2017) also hold in our sample (middle panel, lower row). And as before debt service can almost fully account for these negative effect.

Using the mean group estimator (Pesaran and Smith (1995)) shows that the panel homogeneity assumption seems restrictive, in particular for the debt service effect (right- hand panel, lower row). In this case, the debt service effect becomes much larger than in other specification (around 2.5 times in comparison to the baseline). The net-effects are also larger, yet the other effects are positive (except in year 6 where they are close to zero). Given the strong results, it is unsurprising that the decomposition also holds at the country level despite the fact that we have at most 35 years of data for an individual country (see Figure 20 in the Appendix D). The marked impact of allowing for country heterogeneity suggests that differences in financing conditions, such as different maturity structures or fixed versus floating rates, have implications for the transmission of new debt to debt service and the economy more broadly.

5.4 Effects on the probability of crises

There is a growing empirical literature on the link between credit booms and financial crises (e.g. Borio and Lowe (2002), Reinhart and Rogoff (2009) or Schularick and Taylor (2012)) that finds strikingly similar lead-lag relationships to the ones that we report for debt and output growth. For instance, a large increase in the debt-to-GDP ratio above a long run trend substantially increases the probability of a banking crisis. In this section, we investigate the extent to which projected future debt service, resulting from a debt impulse, can account for the increase in crisis probability.21This is also natural from a theoretical perspective because it is the debt service obligations, not borrowing in itself, that risk triggering defaults by borrowers, and ultimately a financial crisis.

To study the effects of new borrowing and debt service on the probability of banking crises, we adapt our previous empirical framework to binary response models. We take the crisis indicator at t + h as the outcome variable and use it to model the probability of a crisis conditional on new borrowing, debt service and our baseline control variables in a panel logit framework with country fixed effects. To avoid post-crisis bias, we drop the first two 2 years after a financial crisis (Bussiere and Fratzscher (2006)). As robustness checks, we also use simple OLS with panel fixed effects, additional macro controls, the alternative crisis definition and the 5 year growth rate in the debt-to-GDP ratio in place of new borrowing.22

The local projections show that both new borrowing and debt service increase the probability of a crisis, but the effect of debt service is more direct whereas the effect of new borrowing takes time to materialize. This can be seen from Figure 12 which shows the estimated local projections on the probability of crisis from unit increases to these two variables. New borrowing (left panel) does seem to have some impact on the likelihood in the following year, but the strongest and most significant effects are for years 4 and 5. By contrast, a unit increase in debt service (right panel) has a large and significant effect on the crisis probability in years 1 to 3 after which it declines slowly and becomes insignificant. This is consistent with the lead-lag relationship between new borrowing and debt service that we identified earlier.

Our results are robust across a range of specifications (Figure 13). The results are nearly identical if we use the broader crisis definition of Reinhart and Rogoff (2009). Equally, coefficient estimates for both new borrowing and debt service are very similar for years 1 and 2 if we drop all additional controls. And even in the following years, they are not significantly different from the baseline estimates. Differences seem most stark if we additionally control for corporate credit spreads and banking sector provi­sions. However, this is due to a reduction in sample size, implying also four less crisis observations.23

21The sample size drop from more than 400 to 258 if we include provisions and corporate credit spreads. If we run our baseline specification on this restricted sample, the debt service coefficient in period 1 is for example 2.124 versus 2.166 if we include provisions and spreads as shown in Figure 13. Given the limited number of crisis observations, we could not employ our extended set of controls.


22A few recent studies that look at debt service in this context find that it is an excellent early warning indicator, particularly at shorter forecasting horizons (e.g. Drehmann and Juselius (2013), Detken et al (2014)).

 23Since we only observe household debt service for Korea from 1996 onward, we drop the Korean crisis from our model.



5.5 Decomposing the effect of debt on the crisis probability

Our results are, as before, suggestive of debt service being the main variable through which a credit boom leads to a higher probability of a crisis in the future. This is formally confirmed in Figure 14 which decomposes the local projections from a unit increase in debt on the crisis probability into the service effect and other effects (upper panel) in a similar manner as in Section 5.3: debt service always has a sizable positive effect on the crisis probability. And after year one where the debt service effect is zero by construction, it almost fully explains the net effect of new borrowing.

The estimated service effect and other effects are robust (Figure 15). For instance, relying on the broader crisis definition in Reinhart and Rogoff (2009) (upper left-hand panel), adds 3 further crisis to the previous 19 crisis observations, but has a limited impact on our decomposition.

Corporate credit spreads and banking sector provisions seems to explain most of the positive other effects (upper right hand panel). If we control for both variables in addition to the baseline controls at time t0, the net effect is only marginally affected, whereas the debt service effect is strengthened. Hence, other effects are now zero or even reduce the probability of financial crisis except in year 4.24

24Given the limited number of crisis observations, the logit model does not converge anymore if even more controls are added


Using 5 year changes in the household debt-to-GDP ratio instead of new borrowing, as is more common in the literature (e.g. Schularick and Taylor (2012)), reduces the debt service and net effects. Interestingly, the changes offset each other so that the other effects in particular in years 3 to 5 are virtually unchanged.

5.6 Comparison with alternative measures of credit booms​​​​​​​

It is worthwhile to ask how our measures - new borrowing and debt service normalized by GDP - compare to other measures of credit booms that have traditionally been used in the empirical macro-finance literature, such as credit-to-GDP ratios.

Conceptual differences Conceptually, one difference is that our measures capture flows between borrowers and lenders, which enter budget constraints and thus encapsu­late contemporaneous and future liquidity effects of credit relationships, as emphasized eg by Eberly and Krishnamurthy (2014). Traditional measures, on the other hand, relate to credit stocks. Changes in the stock of credit are a function of the flow of new borrowing and of amortizations but not of the flow of interest payments, as we em­phasized in equation (1) in our accounting framework. Whether stock or flow concepts have more explanatory power for real variables is ultimately an empirical question that we examine below. However, if the described flows matter, the stock of credit or even changes in the stock are imperfect measures of these flows.

A second difference is that credit-to-GDP ratios have been growing for decades in most countries. To econometrically deal with trending variables, researchers have there­fore either detrended (e.g. Drehmann et al(2010)) or differenced credit-to-GDP ratios (e.g. Schularick and Taylor (2012)) in empirical work linking credit booms to real devel­opments, which reduces the informational content inherent in these variables and makes it more difficult to economically interpret the estimated relationships. Calculating debt service, by contrast, can be seen as an economically meaningful way of detrending the credit-to-GDP ratio, since low-frequency changes in the terms of credit over the past four decades, such as declines in nominal interest rates, have roughly offset the upward trends in credit-to-GDP ratios.25

Relevance for real variables Ultimately, however, we are interested in determining whether our measures of new borrowing and debt service outperform traditional credit- to-GDP measures in empirical applications. To assess this, we run a horse race. The alternative measures that we consider are the one-year growth rate of real debt, Ad(t, the three year growth rate in the credit-to-GDP ratio, A3(di,t-yi,t),26 the credit-to-GDP gap that filters out medium-term cycles and that is used under Basel III to calibrate countercyclical capital buffers, gapM, and a credit-to-GDP gap based on the two-sided Hodrick-Prescott filter, gapHP, with A = 6.25, as is standard for annual data. We first assess their predictive performance for one-year ahead GDP growth and then for the probability of banking crises.

25Technically speaking there is a cointegrating relationship between credit-to-GDP ratios and interest rates (Juselius and Drehmann (2014)).

26The five-year growth rate in the credit-to-GDP ratio used for instance by Jorda et al (2013) performs similarly to the three-year growth rate.

27The medium-term credit gap is the deviation of the credit-to-GDP ratio from its HP filtered one-sided trend with A = 1600 for annual data.


Debt service and new borrowing strengthen the empirical relationships for one-year- ahead GDP growth compared to alternatives where these two variables are left out (Ta­ble 1). As can be seen from columns (1)-(5), only new borrowing and debt service are significant predictors of future output one year ahead. On their own, other measures are insignificant. New borrowing and debt service also explain a larger share of the variation in the data as captured by R2. Similar results emerge if new borrowing and debt service are added jointly with each of the alternative measures in turn (columns (6)-(9)). The debt service effect, in particular, remains remarkably stable and signif­icant when including alternative measures. New borrowing is slightly more sensitive, suggesting that there might be some overlap in what is captured by new borrowing and the alternative measures of debt.

Our results in Table 1 also lend support to the view that the liquidity effects of credit booms, captured by new borrowing and debt service, play a significant role in driving real output dynamics as argued, for example, by Eberly and Krishnamurthy (2014).


In predicting financial crises in the next year, similar conclusions on the relevance of new borrowing and debt service versus other measures of credit booms apply (Table 2). As can be seen from column (1), both debt service and new borrowing significantly increase the probability of a financial crises one period ahead, and the within-sample predictive performance of the model is very high (AUC = 0.87). Out of the alternative measures, both the three year growth rate in credit-to-GDP and the medium-term credit-to-GDP gap perform well, although their within sample predictive performance is substantially lower (with AUCs around 0.70). Debt service, in particular, compares favorably to the other measures in joint specifications (columns (6)-(9)). Again, there seems to be some collinearity between new borrowing and other measures of debt. For instance, new borrowing, the three year growth rate in credit-to-GDP and the medium­term credit-to-GDP gap all become insignificant in the joint specification, and the standard HP filtered gap even becomes negative.

When interpreting the results in Table 2 from a policy perspective, it is important to keep in mind that macroprudential policies tend to require a longer lead time than one year. For instance, under the Basel III rules, banks have 12 months to comply with an increase in countercyclical capital buffer requirements. In addition, data are reported with lags, and policy makers generally tend not to act on data developments imme­diately, but observe trends for some time before they change policies (e.g., Bernanke, 2004). As the signalling quality of debt service decreases over time (Figure 12), the medium-term credit-to-GDP gap or medium-term growth rates in credit-to-GDP ra­tios are important inputs in steering macroprudential policies to reduce medium-term financial cycle risks.28

28Comparing the medium-term credit-to-GDP gap and debt service, Drehmann and Juselius (2014) find that the medium-term credit-to-GDP gap performs well, even over horizons of up to five years ahead of crises, but that debt service is a very precise early warning indicator two years ahead of crises. They also show that combining the information from the two indicators is ideal from a policy perspective.

6 Conclusions

This paper shows that debt service accounts for much of the adverse real effects of credit, systematically linking past credit booms to predictable future slumps in eco­nomic activity. We lay out a simple accounting framework that describes how debt service can build up with a sizable lag if debt is long-term and new borrowing is auto­correlated, as it typically is in the data. In a panel of 17 countries from 1980 to 2015, we show that the lag between peaks in new borrowing and debt service is on average four years for the household sector. We also show that predicted future debt service accounts for the majority of the transmission mechanism from an impulse to household borrowing to predicted output losses and increases in crisis probability in the medium run.

Our findings raise several important questions related to the measurement and the­ory of credit cycles. For one, given the important real effects of debt service, it is crucial to improve its measurement. It would be particularly beneficial to obtain more regular and granular information on maturity and amortization schedules. This applies to the household sector, and the more so to the corporate sector, where these data are not too reliable. While our preliminary results for the corporate sector in Annex C confirm the findings for the household sector, better data and more work is needed to enhance our understanding of the transmission from corporate borrowing to the real economy.

Our results also highlight the need for theory models to incorporate the credit market features that account for the lag structure of debt service in the data. In particular, doing justice to the data requires auto-correlated new borrowing and long­term debt. 29Furthermore, the strong and systematic pattern in output and crisis probabilities that is generated by flows from lenders to borrowers and vice versa begs explanation. This pattern is consistent with models in which lenders and borrowers have different marginal propensities to consume and borrowers are financially constrained so the negative demand effects of high debt service cannot be offset by additional borrowing. Monetary policy cannot counter the resulting aggregate demand effects when it is constrained by the zero lower bound.30 However, our paper also finds strong negative output effects of debt service that seem to have not been offset by monetary policy during normal times. This raises the question of why, which we leave for future research.

The systematic transmission channel whereby credit expansions can have long- lasting adverse real effects also raises important questions for policy makers. Our empirical results highlight that new borrowing has positive effects but debt service negative effects on the real economy. But the accounting relationship implies a clear link between new borrowing and future debt service. Hence, this raises question such as how policymakers should optimally respond and how they should trade off current output concerns with future debt service obligations. We hope that our findings will be useful for future efforts to better model financial cycles and optimal policy.

29Garriga et al (2013) and Gelain et al (2015) have recently incorporated long-term debt in quanti­tative models of credit fluctuations.

30For models that explain the real effects of the financial crisis of 2008/09 through this prism, see e.g. Eggertsson and Krugman (2013), Guerrieri and Lorenzoni (2011) and Korinek and Simsek (2016).


Bernanke, B S (2004): "Gradualism". Remarks at an economics luncheon co-sponsored by the Federal Reserve Bank of San Francisco (Seattle Branch) and the University of Washington, Seattle, 20 May 2004.

Borio, C and P Lowe (2002): "Assessing the risk of banking crises", BIS Quarterly Review, December, pp 43-54.

Bussiere, M, and M Fratzscher (2006): "Towards a new early warning system of fi­nancial crises", Journal of International Money and Finance, 25(6), pp 953-973.

Cerra, V, and S C Saxena (2008): "Growth dynamics: The myth of economic re­covery", American Economic Review, 98(1),pp439-457.

Claessens, S, M A Kose and M E Terrones (2012): "How do business and financial cycles interact?", IMF Working Paper, WP/11/88.

Dembiermont, C, M Drehmann, and S Muksakunratana (2013): "How much does the private sector really borrow - a new database for total credit to the private non-financial sector", BIS Quarterly Review, March, pp65-81.

Detken, C, O Weeken, L Alessi, D Bonfim, M M Boucinha, C Castro, . . . P Welz (2014): "Operationalising the countercyclical capital buffer: Indicator selection, thresh­old identification and calibration options", European Systemic Risk Board, Occasional Paper, 5.

Drehmann, M, C Borio, and K Tsatsaronis (2011): "Anchoring countercyclical capital buffers: The role of credit aggregates", International Journal of Central Banking, 7(4), pp 189-240.

Drehmann, M, A Ilnes, M Juselius, and M Santos (2015): "How much income is used for debt payments? A new database for debt service ratios", BIS Quarterly Review, March, pp 89-103.

Drehmann M, M Juselius (2014): “Evaluating early warning indicators of banking crises: Satisfying policy requirements”, International Journal of Forecasting, 30(3), pp 759-780.

Dynan, K (2012): "Is a household debt overhang holding back consumption?", Brook­ings Papers on Economic Activity, pp 299-344.

Dynan, K, K Johnson and K Pence (2003): "Recent changes to a measure of U.S. household debt service", Federal Reserve Bulletin 89(10), pp417-426.

Eggertsson, G B and P Krugman (2012): "Debt, deleveraging, and the liquidity trap: A Fisher-Minsky-Koo approach", Quarterly Journal of Economics 127(3), pp 1469-1513.

Eberly, J and A Krishnamurthy (2014): "Efficient Credit Policies in a Housing Debt Crisis", Brookings Papers on Economic Activity Fall 2014, pp73-146.

Farhi, E and I Werning (2016): "A theory of macroprudential policies in the pres­ence of nominal rigidities", Econometrica 84(5), pp 1645-1704.

Garriga C, F E Kydland, and R Sustek (2013), "Mortgages and Monetary Policy," NBER Working Paper 19744.

Gilchrist, S, and E Zakrajsek (2012): "Credit spreads and business cycle fluctuations", American Economic Review, 102(4), pp 1692-1720.

Gelain, P, K J Lansing and G J Natvik (2015), "Leaning Against the Credit Cycle," Norges Bank Working Paper 4/2015.

Gorton, G, and G Ordonez (2016): "Good booms, bad booms", NBER Working Paper, 22008.

Guerrieri, V, and G Lorenzoni (2011): "Credit crises, precautionary savings, and the liquidity trap", NBER Working Papers, 17583

Johnson, K W and G Li (2010): "The debt-payment-to-income ratio as an indicator of borrowing constraints: Evidence from two household surveys", Journal of Money, Credit, and Banking 42(7), pp 1373-1390.

Jorda, O (2005): "Estimation and inference of impulse responses by local projections", American Economic Review, 95(1), pp 161-182.

Jorda, O, M Schularick and A M Taylor (2013): "When credit bites back", Journal of Money, Credit, and Banking 45(2), pp3-28.

Jorda, O, M Schularick, and A M Taylor (2016): "The great mortgaging: Housing finance, crises and business cycles", Economic Policy (85),pp 107-140.

Juselius, M, C Borio, P Disyatat, and M Drehmann (2016): "Monetary policy, the financial cycle and ultra-low interest rates", International Journal of Central Banking, forthcoming.

Juselius, M, and M Drehmann (2015): "Leverage dynamics and the real burden of debt", BIS Working Paper, no 501.

Korinek, A and A Simsek (2016): "Liquidity trap and excessive leverage", Ameri­can Economic Review 106(3), pp 699-738.

Krishnamurthy A and T Muir (2016): “How Credit Cycles across a Financial Cri­sis,” mimeo.

Laeven, L, and F Valencia (2012): "Systemic banking crises database: An update", IMF Working Paper, WP/12/163.

Lombardi, M, M Mohanty, and I Shim (2016): "The real effects of household debt in the short and long run", BIS Working Paper 607.

Loayza, N V, and R Ranciere (2006): "Financial development, financial fragility, and growth", Journal of Money, Credit and Banking, 38(4),pp 1051-1076.

Luckett, C (1980): "Recent financial behaviour of households", Federal Reserve Bul­letin, 66, pp 473-443.

Mian, A, K Rao, and A Sufi (2013): "Household balance Sheets, consumption, and the economic slump", Quarterly Journal of Economics 128(4), pp 1687-1726.

Mian, A, and J A C Santos (2017): "Liquidity risk and maturity management over the credit cycle", Journal of Financial Economics, forthcoming.

Mian, A, and A Sufi (2014): "House of debt: How they (and you) caused the Great Recession, and how we can prevent it from happening again", University of Chicago Press

Mian, A, A Sufi and E Verner (2017): “Household debt and business cycles world­wide,” Quarterly Journal of Economics, forthcoming.

Olney, M L (1999): "Avoiding default: The role of credit in the consumption col­lapse of 1930", Quarterly Journal of Economics 114(1), pp319-335.

Pesaran, M H, and R Smith (1995): "Estimating long-run relationships from dynamic heterogeneous panels", Journal of Econometrics, 68(1), pp 79-113.

Pesola, J (2011): "Joint effect of financial fragility and macroeconomic shocks on bank loan losses: Evidence from Europe", Journal of Banking and Finance, 35(11), pp3134- 3144.

Reinhart, C M and K S Rogoff (2009). This time is different: Eight centuries of financial folly. Princeton University Press.

Schmitt-Grohe, S and M Uribe (2016). “Downward Nominal Wage Rigidity, Currency Pegs, and Involuntary Unemployment,” Journal of Political Economy, 124, pp 1466­1514.

Schularick, M and A M Taylor (2012): "Credit booms gone bust: Monetary policy, leverage cycles, and financial crises, 1870-2008", American Economic Review 102(2), pp 1029-1061.

A Proofs and Additional Results

A.1 Lag Structure Between New Borrowing, Net Cash Flows, and Debt Service