Ekonomický časopis, 68, 2020, č. 10, s.

981 – 1001 981

“Too Much Debt Will Kill You”:

Although Not in Central Europe, Yet1

Katarína BELANOVÁ – Ľubomíra GERTLER – Rudolf SIVÁK*

Abstract

We take an evidence-based approach and confirm that a high level of house-

hold debt is detrimental to consumption, disposable income, and hence economic
growth in the medium term. Using a panel setting with both fixed and time
effects, we isolate the effect of an excessive household debt level and show that
an additional one percentage point could be associated with a 0.1% drag on
GDP growth over the next three years. Although we confirm this relationship
on a panel of advanced economies, we show that emerging economies such as
Central and Eastern European economies, possess structural characteristics
(a shallow capital market, less finance and more social cohesion) that so far
prevents such a demand constraint from materializing.

Keywords: wealth inequality, secular stagnation, household debt

JEL Classification: E51, H31, R21
DOI: https://doi.org/10.31577/ekoncas.2020.10.01

Introduction

We paraphrase the lyrics of the famous song by Queen to enforce our point
that although easy access to finance had a clear positive effect on welfare in the
past decades, if the debt levels become too high, then the ultimate consequences
could endanger the life of an economy. As findings in a recent paper of Burriel
et al. (2020) suggest, highly indebted countries are usually poorly equipped to
withstand future asymmetric shocks and are exposed to higher output losses,

* Katarína BELANOVÁ – Lubomira GERTLER – Rudolf SIVÁK, University of Econo-

mics in Bratislava, Faculty of National Economy, Department of Finance, Dolnozemská cesta 1,
852 35 Bratislava 5, Slovakia; e-mail: katarinabelanova@centrum.sk; Lubka.Gertler@gmail.com;
Rudolf.Sivak@euba.sk
1
This paper is the outcome of the research project Financial Risks and Their Effect on Credit
Cycle and Financial Stability in Slovakia (VEGA 1/0688/20) supported by the Ministry of Educa-
tion, Science, Research and Sport of the Slovak Republic.
982

a longer period at the zero lower bound, greater effects of spillovers, less stable
private debt and consequently less scope for counter-cyclical fiscal policy.
Negative real interest rates are a new normal worldwide, resulting from
excessive financial deepening and abundant liquidity in the economy. On the
one hand, they helped to ignite economic activity, but on the other hand, also
generally led savers to save more and debtors to owe more. Financial liberalisa-
tion therefore led to more income inequality.2
We frame this topic as a follow-up to our previous findings (Gertler, Sivak
and Kyselakova, 2017) that house prices in the medium to long run are fuelled
by credit to households and their share of disposable income. In other words,
if households are becoming excessively indebted, then house prices rise more
rapidly. This result is not surprising. In persistently good times, interest rates
decrease, and demand for housing is higher and more stable. Housing supply is
not flexible, and therefore, prices adapt predominantly to demand.
This outcome, however, has a flipside. In a longer lasting period of economic
growth, lending conditions steadily improve. Mortgages are more accessible and
reach to wider set of households more easily. Some households that were previously
ineligible become the main target of credit supply, especially in the later stage of
good times. However, at this stage, house prices have already reached high levels.
Such cyclicality makes the most vulnerable class of households even more worse
off. Having purchased their housing late in the economic boom, their disposable
income suffers and becomes sensitive to the downturns that lie ahead.
Some remedies to this natural race to more wealth and income inequality set
up with macro-prudential measures that limit less-endowed households’ engage-
ment in mortgage contracts. Although some reduction of borrowing by the most
vulnerable has surely been achieved, this process still exists. Mian, Straub and
Sufi (2020) recently revisited the issue in a study that coins the notion of dimini-
shing disposable income of certain classes of households as ‘indebted demand’.
In summary, sensitive borrowers reduce spending due to the elevated interest
costs and are trapped in a situation in which they can only afford a limited in-
crease in their demand in good times and need to find alternatives to reduce their
spending in bad times. In bad times, alleviating accommodative policies tends to
generate a debt-financed short-run boom at the expense of indebted demand in
the future.
In this study, we aim to gather some evidence and investigate how the story
of indebted demand and the race to the inequality equilibrium differs in Central
Europe. Given the long-run data availability, it is natural that to conduct this

2
This line of argument dates back mainly to the post-Global Financial Crisis period. Stiglitz
(2012) presents a rich description and multiple dimensions of this process in his book.
 983

kind of research on advanced economies. However, we currently have up to 25

years of relevant historical data series in emerging economies to exploit.
Most of the literature on the effects of debt on economic growth is naturally
focused on advanced economies. In this regard, this study is one of the first
to examine this relationship in Central and Eastern European (CEE) countries.
In addition, private (rather than public) sector debt gained more attention only
recently. This study brings the effects of household sector debt on growth to the
frontline. It also considers whether household debt accumulation added a drag on
the growth of CEE economies, or – on the contrary – served as a catalyst for
further development in the medium term.
Our hypothesis is that despite the apparent economic convergence of CEE
countries and the rapid increase of corporate and especially household indebted-
ness due to low interest rate environments and global financial liberalisation,
some important factors that propel this process are unidentified.
In advanced economies, the rising divide between savers and debtors is ac-
celerated by the presence and accessibility of capital markets, which are largely
missing or inefficient in CEE. Additionally, these countries have a higher degree
of solidarity than in the traditional Anglo-Saxon world (perhaps except the Nordic
countries), resulting in a lower Gini coefficient; that is, a more homogenous socie-
ty. Last, in CEE economies, a large majority of households live in owner-occupied
housing, which makes for most of all household assets. We draw upon these
lines to show that the vicious circle of rising inequality, so much present and felt
nowadays in the western hemisphere, is not affecting emerging European econo-
mies to such great extent yet.
Yet, the playing field may change after the second wave of COVID-19 pan-
demic, which hit Central Europe in full strength and again led to multiple weeks
of economic standstill. A second supply shock within 6 months is fertile ground
for heightened uncertainty in the labour market, consumption patterns and the
ever-increasing pile of debt. However, the way out of this pandemic and its con-
sequences could follow widely different trajectories, which are barely possible to
foresee.
This article is structured as follows. In the following section, we take on the
literature and provide a review of related studies on wealth and income inequali-
ty and how this literature feeds into the current discussion of diverging wealth
inequality. In the second section, we provide the economic story that motivates
the hypothesis. In the third section, we introduce the modelling framework and
provide the results for the advanced economies, which broadly align with the
existing literature. In the fourth section, we incorporate CEE economies into the
analysis and through the lenses of structural indicators, we show why the earlier
findings do not yet apply to these countries. The last section concludes.
984

1. Literature Review

This strand of research aims to connect the observed decline in real interest
rates with the concept of ‘secular stagnation’, which Summers (2014) was
among the first to mention. Researchers identified and analysed several drivers
of this development. First, population aging is the source of the changing bal-
ance between the aggregate desire to save and to consume. The point here is that
older agents expect to shoulder only a portion of the financial burden arising
from the mounting debt, and therefore expect the future generation to service it.3
Second, varying patterns in savings can be explained by different levels of in-
come risk and income inequality. Uninsurable risks may thus prevent agents
from adjusting their consumption, thus making households savvier in general
(e.g. Straub, 2019). The global saving glut, as described by Bernanke (2005),
is the third possible explanation. The rising costs of healthcare and social care
motivate agents to plan more for future safety. Whether this is through existing
social security systems (Western European and Nordic societies) or having their
own plan in their absence (Anglo-Saxon societies), the precautionary saving
channel is gaining traction.4 Lastly, the existing supply of investable financial
assets, which agents use as insurance against economic shocks, explains de-
clining real interest rates and secular stagnation. Large quantities of government
debt were purchased in monetary operations worldwide, thus contracting the
volume of available safe assets in the market. What remains must therefore be of
higher risk in general. The changing composition of household asset holdings
may then potentially have a large real effect on saving behaviour (Caballero and
Fahri, 2017).
The listed causes of the changing propensity to save and consume provide
general answers, but do not tell us much about structural aspects, such as which
classes of entities save, and which do not. On this front, Kumhof, Ranciere and
Winant (2015), using a two-agent endowment model of an economy, conclude
that income inequality leads to more debt and a higher chance of financial crisis.
In their world, preferences are heterogenous depending on the activity and in-
come of the household. More specifically, households on labour income have
higher saving rates than borrowers so, but still have lower saving rates than those
saving out of financial income. Here however, the higher debt taken on by bor-
rowers, who save the least, motivates more demand, reduces income inequality,
and reverts the system in equilibrium.
3
Eggertsson, Mehrotra and Robins (2019), among others, elaborate on the ageing channel in
a modelling framework in detail.
4
Caballero and Fahri (2018), Caballero et al. (2016), and others, analyse this channel in
a model framework.
 985

On the other hand, Mian, Straub and Sufi (2020) portray the system as a dual
equilibrium where less wealthy borrowers take on the most debt and have less or
even negative savings. On the contrary, the savings of the wealthy are channelled
into the new funds used for new borrowing offered to the poorer. Less wealthy
debtors, after a reasonable period of this life cycle, form a critical mass of the
population. Thus, the debt service is heavy enough to depresses their demand
and given the magnitude of this mechanism, it has consequences for the aggre-
gate economy.
To complete the review of the relevant research streams, we need to mention
the rich empirical literature on the origins and consequences of high debt. Various
subdisciplines of this vast topic are connected and extensively explained in Abbas,
Pienkowski and Rogoff (2019).
However, several papers by Schularick and Taylor (2012), Jorda, Schularick
and Taylor (2016) are central to this area in more detail, documenting historical
episodes of debt accumulation and its effect on investment, lending, interest rates
and inflation, as well as its link to the depth of financial crises and the speed of
the recovery that follows.5
This article tries to fill the gap by providing an evidence-based approach to
the indebted demand hypothesis. We work out a simple model in a panel setting,
confirm the hypothesis and identify that the hypothesis does not have to be
universally valid due to structural characteristics that tend to the dissemble the
mechanics behind the argument.

2. Economic Background

A declining interest rate in the long run and a low interest rate environment,
coupled with an increase in credit, is a defining feature underlying the build-up
of debt in recent decades. This is also the main driving force of structural changes
that have differing effects on the debtors that add to new credit and the savers
that participate in financing new claims.
Debt financing has occurred in large volume since the financial liberalisation
of the mid-1980s and unconstrained capital flows, predominantly in advanced
economies. Some small breaks are visible as the consequence of the mid- and
late-1990s crises and the Global Financial Crisis in 2007 – 2008. Debt has been
building up in the background of persistently falling real interest rates in the past
three decades.

5
We also use the Schularick and Taylor Macrohistory Database to back cast our existing time
series. This allows us to work with a sample of annual data between 1945 and 2019 (a consecutive
75 years of history).
986

The CEE economies adopted this trend soon after 2000 when they completed
their economic transformation and entered the European Union single market in
May 2004. Despite a quick upswing in the build-up of debt by some 50% of
GDP (a pace similar to that in advanced economies), the total indebtedness of all
sectors is less than half of that in advanced economies. This all occurred against
the background of declining real interest rates on a very similar scale.
The rapid build-up of debt in advanced economies was fuelled by all sectors.
The debt share to GDP roughly doubled in each of the three sectors between
1980 and 2008. However, the quickest path of growth occurred in credit to
households in the late 1990s and the following decade until the Global Financial
Crisis, as well as in government debt, especially since the crisis.

Figure 1
Credit to GDP and Interest Rates

Notes: Cross-country average of private-sector credit over GDP. The 17 countries that represent advanced
economies are: Australia, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands,
Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom and the United States. The countries that
represent Central Europe are the Czech Republic, Slovakia, Poland, Hungary, and Slovenia.
Source: Own calculations.

In Central European economies, debt built up especially in the household

sector, while corporate sector debt remained rather flat and government debt
increased somewhat after the crisis but reverted to lower levels in recent years.
Stable debt levels in the corporate sector is largely tied to domestic investments
to GDP, which also remained rather flat. Fresh credit was not needed because the
 987

corporate sector was largely funded by foreign direct investment or the expan-
sion of foreign companies into the new markets after the single market grew. In
the advanced economies, the flatter pattern of corporate debt is linked to a low
and declining investment to GDP ratio.
Traditionally, in a circular flow of income, which is a main building block of
Keynesian models, businesses and governments lend resources from households,
and this lending then triggers investment and productivity growth. What we see,
however, is that the excessive rise of household debt is used to finance personal
consumption and government expenditures. As we can also see clearly in the
current post-COVID world, credit expansion is used to increase aggregate de-
mand rather than supply.6

Figure 2
Debt to GDP by Sector

Notes: Cross-country average private-sector credit over GDP, weighted by GDP in 2000. For a fuller picture,
see the breakdown of advanced and emerging (V4) countries’ debt to GDP in Annexes A1 and A2.
Source: Own calculations.

We are also beginning to see empirically that demand remains muted because
the debt levels are simply too high. At moderate levels, debt improves welfare
6
Again, as in previous crises, the current crisis will test the flexibility of heavily indebted
countries and companies. As economies respond to the pandemic, their debt will certainly increase.
Global growth is projected at –4.9% in 2020; –8.0% for advanced economies and –3.0% for
emerging markets (IMF, 2020).
988

and supports growth. However, as some, such as Cecchetti, Mohanty and Zam-
polli (2011) are already highlighting, debt may become a drag on growth when
it reaches certain thresholds. Their estimated thresholds come at 85% of GDP
for the government and 90% for the corporate sector. Similarly, Arcand, Berkes
and Panizza (2012) indicate that financial depth derails growth once credit to
the private sector exceeds 100%. As we can also see in Figure 2, we have now
exceeded these thresholds considerably.
Are we now at the point where private (and public) debt levels are too high,
and are dragging our medium to long run growth down? Worse, would we be
able to escape the current crisis only by piling further debt? To answer these
questions, we run a fixed effect model with both country and time effects and
investigate the effect of households’ private debt level on economic growth,
income, and consumption.

3. The Impact of Debt in the Model

We first review some statistics of debt ratios and savings, consumption, and
growth before we turn to empirical investigation of these questions. We use an
annual dataset of 17 advanced economies and 5 Central European economies.
We use credit series from the BIS, financial series from the IFS and national
accounts and real economy series from the OECD. Where needed and available,
we back casted these time series with the historical series of the Macrohistory
Database from Jorda, Schularick and Taylor (2017).
First, we address the main point of our line of argument and verify whether
increases in household debt tend to reduce growth.
We depart from the model specification of the per capita output growth rate
dependent on indebtedness and other country-specific information following
Solow’s neoclassical growth model used by Barro and Sala-i-Martin (2004),
among others. We also add some other financial and non-financial measures and
vary the use of indebtedness in different sectors.
In such a traditional model, per capita income growth depends in the initial
level of output (or capital-both human and physical), and in the medium to long-
-run convergence to the steady state. The steady state depends positively on the
saving rate and negatively on factor inputs, such as labour force or technology.
Since this family of models worked with regional or cross-country differences,
we instead transform this model specification to a panel setting to estimate a panel
regression with country-and time-specific fixed effects. This approach produces
a clean measure of impact in terms of which factor (in our case indebtedness)
affects growth within a country.
 989

We model the growth of per-capita income of a country g in the following

specification:
gi ,t +1,t + k = ϕ yi ,t + β ′ X i ,t + µi + γ t + ε i ,t ,t + k

where the dependent variable is the k-years forward average annual growth rate
of output; yi,t represents logarithm of contemporary real per-capita output; Xi,t
stands for a set of regressors, including debt measures, saving rate, asset prices
or real interest rate; and µi and γ t are the country- and time-specific dummies,
respectively, which capture common effects across countries and time. We also
add a dummy for the Global Financial Crisis in 2008.
We alter this model in several directions as follows. First, we modify the
extent of indebtedness to see how heavy the debt burden in the individual sectors
is. Second, we modify output growth with households’ disposable income or real
consumption to obtain more direct information about the effect of debt on the
behaviour of households. Third, we estimate a mirror model of Central European
countries to see whether the debt level has a similarly negative effect on the wel-
fare of households or there is still enough time to raise awareness.
In the baseline model, we investigate whether household indebtedness affects
output per capita. We can indeed confirm that both public and household debt
have a negative effect on the future medium-term economic growth. More spe-
cifically, an average one percentage point increase in household debt is associated
with 0.05% less economic growth in the three years to come. For public debt, the
effect for the 17 advanced economies in the post-war period is double, at 0.1% of
GDP.
Both the fixed and time effects are in place and all other explanatory varia-
bles have the expected signs (Table 1). The positive sign with the house prices
relates to the wealth effect, where rising house prices tends to encourage con-
sumer spending and leads to higher economic growth. This relationship works
both ways, although not necessarily symmetrically. If house prices drop, they
tend to do so sharply and adversely affect consumer confidence, which leads to
low or negative economic growth.
Mian, Sufi and Verner (2015) also report a negative relationship, finding that
the negative effects of household debt on income to be particularly pronounced
for countries facing monetary policy constraints. Mika and Zumer (2017) present
similar results. They find that increasing indebtedness in the private sector has
a negative effect on growth rates in the near future. This could be explained by
the fact that households and non-financial corporations are perceived to be over-
borrowing. On the other hand, rising private sector indebtedness is associated
with rising income levels over the long run, as debt allows for consumption and
990

investment smoothing. Overall, as they state, accumulating more debt is a way of

financing as well as obtaining funds to finance investments. Households and
non-financial corporations are expecting to be better off in the future, and hence
are not afraid of borrowing more in the present. Efficient use of such borrowing
stimulates the economy. When considering the effects of deleveraging, Chen et al.
(2015) find that the quicker the private sector deleverages, the greater the posi-
tive effects on growth in the medium term.

Table 1
Baseline Estimate
Dependent: Three-year forward average per-capita GDP
coef. st.dev t-val
Log of output –0.1117*** (0.0079) 0.000
Public debt –0.0010*** (0.0001) 0.000
Household debt –0.0005*** (0.0002) 0.008
House price inflation 0.1130*** (0.0319) 0.000
Real interest rates –0.0013* (0.0008) 0.084
D(GFC) 0.3258*** (0.0484) 0.000
Number of GROUPS OBERVATIONS
17 927
R-square R2 within R2 between R2 overall
0.806 0.032 0.195
Statistics F(74,836) Prob > F Corr(ui, Xb)
46.80 0.000 –0.8879
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.

Similarly, a negative effect of the real interest rate on economic growth is

a long-standing premise of economic theory, dating back to the seminal contribu-
tion of McKinnon (1973) and the following literature on the benefits of financial
liberalisation. Much of this literature investigates how the decreasing level of real
interest rates, which is a manifestation of financial liberalisation, supports econom-
ic growth in the long run. In our baseline, estimate this materialises in 0.13% more
economic growth thanks to a real interest rate lower by 1 percentage point. As later
literature also points out, (e.g. De Gregorio and Guidotti, 1995), this relationship is
far from linear, which is already obvious in the current environment of negative
real interest rates, where further declines are likely to weigh on economic activity
in future. However, we are interested in the story of excessive debt levels and the
extent to which they weight future economic growth. To have a better picture, we
estimate the same model for the cases when debt exceeds certain thresholds.
As it stands out from Table 2, the higher the indebtedness level of households,
the more drag to economic growth in general. As we mentioned earlier, the results
convey that the channel works through the compression of available disposable
 991

income, which can be used for consumption. In other words, if both public and
household debt are high (above 60% of GDP), then the effect of further increases
in household debt almost quadruples the drag on output, to close to 0.2% of GDP
in the three years to come, while the effect of public debt remains roughly the same.

Table 2
Higher Household Debt as a Burden
Dependent: Three-year forward average per-capita GDP
>40% >50% >60%
Log of output –0.1117*** –0.1311*** –0.1600*** –0.1310*
(0.0079) (0.0209) (0.3832) (0.0733)
Public debt –0.0010*** –0.0005*** –0.0009*** –0.0008*
(0.0001) (0.0001) (0.0002) (0.0050)
Household debt –0.0005*** –0.0005* –0.0008*** –0.0019**
(0.0002) (0.0003) (0.0003) (0.0008)
House price inflation 0.1130*** 0.1287*** 0.1869*** 0.1830**
(0.0319) (0.0422) (0.0495) (0.0865)
Real interest rates –0.0013* –0.0025* –0.0006 0.00004
(0.0008) (0.0011) (0.0015) (0.0029)
D(GFC) 0.3258*** 0.3908*** 0.5266*** –0.101*
(0.0484) (0.0817) (0.1380) (0.0385)
Number of observations 927 397 241 123
Number of groups 17 17 15 11
R2 within 0.806 0.798 0.726 0.747
R2 between 0.032 0.012 0.015 0.372
R2 overall 0.195 0.072 0.078 0.166
F(.,.) 46.80 18.49 9.55 4.92
Prob > F 0.000 0.000 0.000 0.000
Corr(ui, Xb) –0.8879 –0.9719 –0.9901 –0.9922
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.

The disappearing effect of the real interest rate that we observe in Table 2
also aligns well with the past literature. A handful of studies provide evidence
that higher real interest rates may be a drag on economic output (e.g. Taylor,
1999). Later studies (e.g. Shaukat, Zhu and Khan, 2019) add that this is mainly
the case for transition economies or economies in the early stage of financial
deepening, where the real interest rate, through its multiple channels, may become
detrimental. Recent studies also find non-linearity and provide the thresholds at
which this is no longer detrimental. However, they are united in their conclusion
that in the current low interest rate and low inflation environment, the relation-
ship between the real interest rate and output disappears (e.g. Bosworth, 2014).
On the other hand, asset prices apparently fuel the economic growth in the
current environment of low inflation, low interest rates and high debt levels. The
higher effect of house prices on economic output in our panel estimate confirms
the wealth effect in our data.
992

Corporate debt seems not to have any negative effect on growth (see Annex,
Table A1), as there is no reasonable channel that would compress households’
disponible income or consumption. It is likely due to the global environment and
foreign financing and other forms of funding that does not require corporations
to look for excessive amounts of credit. However, in countries with higher cor-
porate debt levels (France, the Netherlands, Belgium, Sweden and Norway), we
can still observe some positive effects on economic activity.
Cross-checking the same model specification with a possibly more direct
effect of household debt, we replace economic growth with disposable income.
Interestingly, we see very similar results with a similar magnitude (Table 3).

Table 3
Higher Household Debt as a Drag on Disposable Income
Dependent: Three-year forward average per-capita disposable income
>40% >50% >60%
Log of output –0.1550*** –0.1330*** –0.1477*** –0.1402*
(0.0112) (0.0257) (0.0447) (0.0784)
Public debt –0.0007*** –0.0003** –0.0006*** –0.0004
(0.0001) (0.00015) (0.0002) (0.0008)
Household debt –0.0005*** –0.0005* –0.0009*** –0.0016**
(0.0002) (0.0003) (0.0003) (0.0008)
House price inflation 0.0991*** 0.1112** 0.1589*** 0.1664*
(0.0342) (0.0481) (0.0552) (0.0934)
Real interest rates –0.0021* –0.0018 –0.0008 0.0005
(0.0009) (0.0013) (0.0017) (0.0032)
D(GFC) 0.6304*** 0.0663 0.0031 0.0268
(0.0699) (0.0707) (0.0669) (0.1084)
Number of observations 717 397 225 114
Number of groups 17 17 15 10
R2 within 0.832 0.798 0.679 0.731
R2 between 0.072 0.012 0.008 0.163
R2 overall 0.167 0.072 0.071 0.122
F(.,.) 50.93 18.49 9.29 4.99
Prob > F 0.000 0.000 0.000 0.000
Corr(ui, Xb) –0.9316 –0.9719 –0.9876 –0.9903
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.

If we continue further to examine the factor that should constitute the most
direct effect; that is, that of debt on household consumption, we naturally find
the highest coefficients and the most robust results (Table 4). All coefficients are
in the same ballpark, although real interest rates become insignificant at higher
debt levels.
The estimates portrayed a consistent picture of a channel through which the
debt drag operates. In brief, household debt (and to a minor extent, public debt)
weighs on the household budget and affects consumption, and therefore also
 993

economic activity in general, through which we may also see effects on disposable
income and economic growth. In addition, we show that as the levels of house-
hold and public debt increase, their drag on consumption, disposable income or/
and economic output increases.

Table 4
Higher Household Debt as a Drag on Consumption
Dependent: Three-year forward average per-capita consumption
>40% >50% >60%
Log of output –0.1421*** –0.0959*** –0.0695** 0.0095
(0.0076) (0.1715) (0.0447) (0.0487)
Public debt –0.0009*** –0.0003*** –0.0004** –0.0001
(0.0001) (0.00010) (0.0002) (0.0003)
Household debt –0.0010*** –0.0005** –0.0008*** –0.0021***
(0.0001) (0.0002) (0.0003) (0.0006)
House price inflation 0.11761*** 0.1657*** 0.2204*** 0.2067*
(0.0273) (0.0368) (0.0450) (0.0691)
Real interest rates –0.0031*** –0.0032*** –0.0015 0.0017
(0.0007) (0.0010) (0.0014) (0.0230)
D(GFC) 0.4889*** 0.0127 –0.0935** –0.1125*
(0.0527) (0.0485) (0.0456) (0.0657)
Number of observations 731 357 231 118
Number of groups 16 16 15 10
R2 within 0.891 0.792 0.705 0.784
R2 between 0.064 0.093 0.018 0.076
R2 overall 0.180 0.077 0.112 0.165
F(.,.) 84.85 24.36 10.82 6.96
Prob > F 0.000 0.000 0.000 0.000
Corr(ui, Xb) –0.9177 –0.9548 –0.9541 –0.6796
Note: Standard errors in parentheses; * p<0.10, ** p<0.05, *** p<0.01.
Source: Own calculations.

4. Structural Factors behind the Drag on Growth

How universal are these findings for non-advanced economies? Applying

these estimates to Central European economies, we do not find much signifi-
cance, especially for the variables related to asset markets and finance in general.
There is indeed a ratio to this observation. Small, very open economies are mar-
ket takers and depend closely on their trading partner countries. They operate in
environments with certain interest rate levels and are exposed to the liquidity
provided in the partner economies. Hence, these countries are, to a large extent,
integrated and destined to have similar levels of real interest rates and asset mar-
ket developments to those of their trading partner countries.
Some countries, such as Slovakia and Slovenia, are not only a single market,
but also have a common monetary policy that keeps these countries exposed to
common policies. However, a large part of society remains unaffected, since it is
994

not engaged in the financial market. The level of real interest rates is therefore
uninteresting to the vast majority of households across Central Europe, because
their portfolios consist of illiquid self-occupied housing, some investment trade-
able goods, and deposits. Asset or house price inflation is therefore orthogonal to
households or corporate resources. On the other hand, large corporations are
mostly foreign owned and able to fund horizontally within their supranational
network, or through other sources.

Table 5
CEE Countries’ Debt Burden to Consumption, Disposable Income, and Output
Consumption Disposable income Economic growth
Log of CON, HDI, GDP –0.6523*** –0.8678*** –0.6182***
(0.0859) (0.1301) (0.0975)
Public debt 0.0003** 0.0024 –0.0002*
(0.0001) (0.0020) (0.0001)
Household debt –0.0005*** –0.0009*** –0.0006***
(0.0001) (0.0020) (0.0001)
Deflator inflation –1.2315** –0.3464 –0.7617
(0.5208) (0.7562) (0.5858)
Saving rate –0.0146*** –0.1213*** –0.0141***
(0.0022) (0.0032) (0.0025)
Number of observations 51 49 51
Number of groups 3 4 4
R2 within 0.964 0.944 0.954
R2 between 0.999 0.990 0.997
R2 overall 0.165 0.152 0.164
F(.,.) 24.69 14.17 19.15
Prob > F 0.000 0.000 0.000
Corr(ui, Xb) –0.9983 –0.9989 –0.9982
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.

For this reason, we supplement house prices with deflator inflation and add
a measure of the saving rate, which better characterises the investment climate in
emerging markets. Nevertheless, debt levels are rather low in Central Europe,
and it would be difficult to draw a line between the observed household debt
levels of 20% to 40% of GDP. However, as we pointed out, the depth of the
financial market seems to matter for the level of debt and hence for debt depend-
ency. Although to a lesser extent, such characteristics may be attributed to redis-
tribution and/or social justice. The more social justice is in place, the less need
for debt in a society, since more wealth is redistributed systematically. First and
foremost, these characteristics are of a structural nature and indicate an economy
with a low level of debt.
We represent financial market depth with market capitalisation and social
justice with the Gini coefficient. Figure 3 above makes the point.
 995

Figure 3
Structural Factors behind Lower Debt Levels

Note: Both the market capitalisation index and the Gini coefficient are from the World Bank database and are
the average country values between 1975 and 2017.
Source: Own calculations.

All representatives of Central Europe are grouped at the left bottom segment
of both figures, which indicates the structural characteristics behind the lower
household debt levels. Combined with the estimated relationship between the
levels of debt and output/disposable income/consumption, we can conclude that
we have evidence suggesting that the structural characteristics of Central Euro-
pean economies yet provide a shield, which prevents their debt levels from being
a burden on consumption and growth so far.

Conclusion

We revisited the indebted demand hypothesis, which suggests that high levels
of debt and the consequent interest costs become a constraint on household con-
sumption and hence on aggregate demand. In other words, the large and still
increasing debt already affects a critical mass of society and is a sustained drag
on economic growth in the medium term.
996

We used a simple panel data setting with fixed and time effects based on
a Solow-style growth framework to isolate the effect of household debt on
household consumption, disposable income, and economic growth. Using annual
data of 17 advanced economies for the past 50 years and 5 emerging economies
for the past 25 years, we confirmed our hypothesis and showed that 1% house-
hold debt can be associated with 0.1% in GDP growth in the three years to come.
Moreover, we found that this effect almost doubles when household and general
government (public) indebtedness exceeds 60% of GDP each.
Further, we noted some structural differences between the advanced and
emerging economies of Central Europe, namely the shallower financial markets
and more redistribution, which indicates the higher social cohesion (measured by
Gini coefficient) and prevents household debt from rising dramatically. Holding
debt levels relatively low and experiencing worse access to financial markets
paradoxically helps these economies by preventing the negative effect of reduced
consumption and growth.
Mika and Zumer’s (2017) findings support the conclusion on structural dif-
ferences, as they also claim that country-specific thresholds exist, beyond which
the effects of debt on growth become more negative. These thresholds depend on
the country’s indebtedness level and its other characteristics. This has important
policy relevance, as the policy implications and recommendations in this context
are country-specific rather than ‘one size fits all’.
Ideally, our analysis could be supplemented by a model that combines the
short- and long-run specifications, as in the analysis by Eberhardt and Presbitero
(2015). This would allow us to directly determine the long- and short-term im-
pact of indebtedness on output growth. Moreover, the effect of the current global
recession will provide an opportunity to test this assumption in such a model
in the future.
Thus, although we see specific structural characteristics which still protect
Central Europe against being constrained by excessive debt, things may change
to mirror advanced economies if current trends persist. Escaping the debt trap
then is difficult, but not impossible.
One way would be to reduce incentives to finance business with debt and use
more equity financing instead. One eventual remedy is to reshuffle preferences
by modifying tax systems. It is also possible to shift from debt to equity financ-
ing in the area of housing.
However, one opportunity available currently is to replace government lending
to companies in the COVID-19 crisis with equity purchases. Indeed, with the cur-
rent ultra-accommodative interest rate policies, governments could create instan-
taneous sovereign wealth funds at reasonable terms.
 997

References

ABBAS, S. A. – PIENKOWSKI, A. – ROGOFF, K. (2019): Sovereign Debt: A Guide for Econo-

mists and Practitioners. Oxford: Oxford University Press.
ARCAND, J. – BERKES, E. – PANIZZA, U. (2012): Too Much Finance? [IMF Working Paper
12/161.] Washington, DC: International Monetary Fund.
BARRO, R. – SALA-I-MARTINS, X. (2004): Economic Growth. 2nd ed. Cambridge, MA: MIT
Press.
BERNANKE, B. S. (2005): The Global Saving Glut and the U.S. Current Account Deficit.
[Speech, No. 77.] Board of Governors of the Federal Reserve System (U.S.)
BOSWORTH, B. P. (2014): Interest Rates and Economic Growth: Are They Related? Washington,
DC: Brookings Institution.
BURRIEL, P. – CHECHERITA-WESTPHAL, C. – JACQUINOT, P. – SCHÖN, M. – STÄHLER,
N. (2020): Economic Consequences of High Public Debt: Evidence from Three Large Scale
DSGE Models. [ECB Working Paper Series, No.2450.] Frankfurt am Main: ECB.
CABALLERO, R. J. – FARHI, E. – GOURINCHAS, P. O. (2016): Safe Asset Scarcity and
Aggregate Demand. American Economic Review, American Economic Association, 106, No. 5,
pp. 513 – 518.
CABALLERO, R. J. – FARHI, E. (2018): The Safety Trap. The Review of Economic Studies, 85,
No. 1, pp. 223 – 274.
CECCHETTI, S. – MOHANTY, M. – ZAMPOLLI, F. (2011): The Real Effects of Debt. [BIS
Working Papers 352.] Basel: Bank for International Settlements.
CHEN, S. – MINSUK, K. – MARIJN, O. – WISEMAN, K. – ZDZIENICKA, A. (2015): Private
Sector Deleveraging and Growth Following Busts. [IMF Working Paper 15/35.] Washington,
DC: International Monetary Fund.
De GREGORIO, J. – GUIDOTTI, P. E. (1995): Financial Development and Economic Growth.
World Development, 23, No. 3, pp. 433 – 448.
EBERHARDT, M. – PRESBITERO, A. F. (2015): Public Debt and Growth: Heterogeneity and
Non-linearity. Journal of International Economics, 97, No. 1, pp. 45 – 58.
EGGERTSSON, G. B. – MEHROTRA, N. R. – ROBBINS, J. A. (2019): A Model of Secular
Stagnation: Theory and Quantitative Evaluation. American Economic Journal: Macroeconomics,
11, No. 1, pp. 1 – 48.
GERTLER, L. – SIVAK, R. – KYSELAKOVA, D. (2017): Assessing Explanatory Power of House-
hold Debt for House Prices. Ekonomický časopis/Journal of Economics, 65, No. 4, pp. 335 – 369.
IMF (2020): Fiscal Monitor: Policies to Support People during the COVID-19 Pandemic. Washing-
ton, DC: International Monetary Fund.
JORDA, O. – SCHULARICK, M. – TAYLOR, A. M. (2016): The Great Mortgaging: Housing
Finance, Crises and business cycles. Economic Policy, 31, No. 85, pp. 107 – 152.
JORDA, O. – SCHULARICK, M. – TAYLOR, A. M. (2017): Macrofinancial History and the New
Business Cycle Facts. NBER Macroeconomics Annual 2016, 31, pp. 213 – 263.
KUMHOF, M. – RANCIERE, R. – WINANT, P. (2015): Inequality, Leverage, and Crises. American
Economic Review, 105, No. 3, pp. 1217 – 1245.
Mc KINNON, R. I. (1973): Money and Capital in Economic Development, Washington, DC:
Brookings Institution.
MIAN, A. – SUFI, A. – VERNER, E. (2017): Household Debt and Business Cycles Worldwide.
The Quarterly Journal of Economics, Oxford University Press, 132, No. 4, pp. 1755 – 1817.
MIAN, A. R. – STRAUB, L. – SUFI, A. (2020): Indebted Demand. [NBER Working Papers
26940.] Cambridge, MA: National Bureau of Economic Research, Inc.
MIKA, A. – ZUMER, T. (2017): Indebtedness in the EU: A Drag or a Catalyst for Growth? [ECB
Working Paper Series, No. 2118.] Frankfurt am Main: ECB.
998

SHAUKAT, B. – ZHU, Q. – KHAN, M. I. (2019): Real Interest Rate and Economic Growth:
A Statistical Exploration for Transitory Economies. Physica A: Statistical Mechanics and its
Applications, 534, No. C, pp. 1 – 22.
SCHULARICK, M. – TAYLOR, A. M. (2012): Credit Booms Gone Bust: Monetary Policy, Lev-
erage Cycles, and Financial Crises, 1870 – 2008. American Economic Review, 102, No. 2,
pp. 1029 – 1061.
STIGLITZ, J. E. (2012): The Price of Inequality: How Today’s Divided Society Endangers Our
Future. New York: W. W. Norton & Co.
STRAUB, L. (2019): Consumption, Savings, and the Distribution of Permanent Income. Econo-
metrica, mimeo.
SUMMERS, L. H. (2014): Us Economic Prospects: Secular Stagnation, Hysteresis, and the Zero-
lower Bound. Business Economics, 49, No. 2, pp. 65 – 73.
TAYLOR, M. (1999): Real Interest Rates and Macroeconomic Activity. Oxford Review Economic
Policy, 15, No. 2, pp. 95 – 113.
 999

Annex

F i g u r e A1
Debt to GDP in Central Europe
a) Households
60

SK
CZ
HU
50

PL
V4
Household debt to GDP
20 30 10
0 40

1995 2000 2005 2010 2015 2020

year

b) Corporate sector
100

SK
CZ
HU
80

PL
V4
Corporate debt to GDP
40 60
20
0

1995 2000 2005 2010 2015 2020

year

c) Government
100

SK
CZ
HU
80

PL
Governement debt to GDP

V4
40 60
20
0

1995 2000 2005 2010 2015 2020

year

Source: Own calculations.

1000

F i g u r e A2
Debt to GDP in Advanced Economies
a) Households

US
140

UK
BE
100 120

DK
FR
Household debt to GDP

DE
IT
80

NL
NO
SE
60

CH
FI
40

PT
ES
20

ALL
0

1960 1970 1980 1990 2000 2010 2020

year

b) Corporations

US
UK
BE
60 80 100 120 140

DK
FR
Corporate debt to GDP

DE
IT
NL
NO
SE
CH
FI
40

PT
ES
20

ALL
0

1960 1970 1980 1990 2000 2010 2020

year

c) Government

US
140

UK
BE
100 120

DK
Government debt to GDP

FR
DE
IT
80

NL
NO
SE
60

CH
FI
40

PT
ES
20

ALL
0

1960 1970 1980 1990 2000 2010 2020

year

Source: Own calculations.

 1001

T a b l e A1
Alternative Estimate with Corporate Debt
Dependent: Three-year forward average per-capita GDP
coef. st.dev t-val
Log of output –0.1105*** (0.0079) 0.000
Public debt –0.0010*** (0.0001) 0.000
Household debt –0.0006*** (0.0002) 0.005
Corporate debt –0.0001 (0.0001) 0.663
House price inflation 0.1076*** (0.0321) 0.001
Real interest rates –0.0014* (0.0008) 0.072
D(GFC) 0.3235*** (0.0482) 0.000
Number of GROUPS OBSERVATIONS
17 923
R square R2 within R2 between R2 overall
0.808 0.033 0.199
Statistics F(74,836) Prob > F Corr(ui, Xb)
46.60 0.000 –0.8864
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.

T a b l e A2
Alternative Estimate for the CEE
Dependent: Three-year forward average per-capita GDP
coef. st.dev t-val
Log of output –1.367*** (0.2327) 0.000
Public debt –0.0005 (0.0140) 0.726
Household debt –0.0065*** (0.0011) 0.000
House price inflation –0.0939 (0.1873) 0.623
Real interest rates –0.0035 (0.0008) 0.391
D(GFC) 0.5440 (0.0717) 0.460
Number of GROUPS OBSERVATIONS
4 33
R square R2 within R2 between R2 overall
0.952 0.569 0.154
Stats F(74,836) Prob > F Corr(ui, Xb)
21.15 0.000 –0.999
Note: Standard errors in parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.
Source: Own calculations.
Reproduced with permission of copyright owner. Further reproduction
prohibited without permission.