Agostino Consolo
Private Sector Indebtedness
European Central Bank
Directorate General Economic Developments
across euro area countries
Supply side, Labour Market and Surveillance Division
This is a joint work with Beatrice Pierluigi and has ADBI – Managing Private and Local
received comments from Wolfgang Modery and Isabel
Vansteenkiste. We also acknowledge excellent research Government Debt
assistance from Federica Malfa, Joan Piqueras and
Giacomo Pongetti. 30 Nov – 1 Dec 2017, Tokyo
The views expressed are those of the speaker and do not necessarily reflect those of the European Central Bank.
The views expressed in this presentation are the views of the author and do not necessarily reflect the views or policies of the Asian
Development Bank Institute (ADBI), the Asian Development Bank (ADB), its Board of Directors, or the governments they represent. ADBI
does not guarantee the accuracy of the data included in this paper and accepts no responsibility for any consequences of their use.
Terminology used may not necessarily be consistent with ADB official terms.
Overview
Rubric of the talk
2 www.ecb.europa.eu ©
A.1 A bird-eye view on private sector indebtedness
Rubric
• Build-up of private sector debt in the euro area has been heterogeneous
and significant corrections are under way across both HH and NFC sector
Chart 1: Private sector debt across US, JP Chart 2: HH debt change, pre- and post- Chart 3: NFC debt change, pre- and post-
and four largest euro area countries crisis time (in pp of GDP) crisis time (in pp of GDP)
(in % of GDP)
Source: BIS database for long-term time series of Source: Eurostat national and sectoral accounts; ECB Source: Eurostat national and sectoral accounts; ECB
private non-financial sector debt. The grey area computation computation. Given the large changes, IE and LU are
shows the min-max interval across Germany, France, not shown in the picture
Italy and Spain. The median is computed across
these four euro area countries
3 www.ecb.europa.eu ©
A.2 The euro area context
Rubric
• The analysis of the private (non-financial) sector debt is very topical for the
euro area as a whole and for the convergence process across countries
4 www.ecb.europa.eu ©
Structure
Rubric of the talk and key takeaways
• The structure of the talk is based on the following three work streams
looking at macro, financial and framework conditions behind indebtedness
5 www.ecb.europa.eu ©
Structure
Rubric of the talk and key takeaways
6 www.ecb.europa.eu ©
Overview
Rubric of the talk
7 www.ecb.europa.eu ©
B.1 Measuring excessive private sector leverage
Rubric
• For each country in the euro area, the amount of deleveraging is computed
by comparing current debt levels with a robust benchmark debt level for
each sector (HH and NFC)
• A robust benchmark debt level is derived by model averaging across a set
of panel cointegration and unobserved components models
– In each model, the benchmark level of debt is defined as the underlying long-
term/permanent component. This is country- and sector-specific
– The main purpose of using this set of models is to identify the long-term drivers of debt
from the cyclical fluctuations, while going beyond purely statistical/filtering methods
– The set of the explanatory variables used in the empirical work is differentiated between
HH and NFC and accounts for macroeconomic, financial and balance-sheet variables
8 www.ecb.europa.eu ©
B.2 Deleveraging estimates and bank NPLs
Rubric
• Chart 6 shows the median deleveraging estimates across all models for HH and NFC
• Some countries display larger deleveraging estimates in certain sectors, although
strong co-movement
• Across countries, deleveraging estimates are related to higher banks’ NPL (Chart 7)
Chart 6: deleveraging estimates (2015) for HH and NFC Chart 7: deleveraging estimates for the private
across EA countries sector (2015) and the Texas ratio (2016) across EA
(in % of GDP) countries (both in terms of z-score)
NFC Deleveraging Estimates in 2015
• Chart 8 and chart 9 show how deleveraging estimates measured in 2009 are
associated with subsequent economic and financial developments
• This aims at corroborating the importance of excessive leverage in the private sector
and the prolonged effects the deleveraging process can have on the economy
Chart 8: deleveraging estimates in 2009 and Chart 9: deleveraging estimates in 2009 and
change in core inflation from 2010 to 2015 change in HH bank lending from 2010 to 2015
Source: Eurostat national and sectoral account data. ECB bank balance-sheet (BSI) database
Note: Deleveraging estimates are displayed in slide 10. Core inflation refers to the HICP excluding energy index.
10 www.ecb.europa.eu ©
Overview
Rubric of the talk
11 www.ecb.europa.eu ©
C.1 Private sector debt vulnerabilities in the euro area
Rubric
• A composite risk indicator is built, both for HH and NFC, based on four
underlying factors to encompass different definitions of risk:
– Debt-to-GDP ratio (overall sustainability risks)
– Debt-to-asset ratio (balance-sheet leverage risks)
– Debt service-to-income ratio (liquidity/roll-over risks)
– Interest-growth differential (long-term dynamics risks)
12 www.ecb.europa.eu ©
C.2 Private sector debt vulnerabilities in the euro area
Rubric
Chart 10: HH indebtedness vulnerability index (HSIX)
• These charts show the HH (HSIX)
and the NFC (FSIX) Sector
Indebtedness index in 2006, 2009
and 2017
Source: Eurostat and ECB, Quarterly national accounts Source: Eurostat and ECB, Quarterly national accounts
Note: Range computed as the interquartile range across euro area Note: Range computed as the interquartile range across euro area
countries. Interest payment does not include FISIM fees. Income is the HH countries. Interest payment does not include FISIM fees. Income is the
gross disposable income NFC gross operating surplus
15 www.ecb.europa.eu ©
Overview
Rubric of the talk
16 www.ecb.europa.eu ©
D.1 Policies supporting the deleveraging process
Rubric
• Chart 6 shows that, across some euro area countries and sectors, the ongoing
deleveraging process is still incomplete; also, debt write-offs have had only a
marginal contribution until now
17 www.ecb.europa.eu ©
D.2 Policies supporting the deleveraging process
Rubric
• In a panel of OECD and EU countries, the efficiency of the insolvency framework (IFI)
is found significant in supporting the reduction of bank NPL and favouring the
deleveraging process (annex)
• This is especially so when countries experience high unemployment gaps (crisis
times). IFI is more effective in reducing the stock of NPLs more than in limiting their
increase during the early phase of the crisis
Table 1: panel estimates across EU and OECD countries
18 www.ecb.europa.eu ©
Summary
Rubric and concluding remarks
• Summing up,
– the deleveraging process has strengthened and balance-sheet conditions are improving.
– some countries have adopted new legislations and procedures to upgrade their insolvency regimes
– The ongoing adjustment in private sector vulnerabilities and the convergence process in the euro area
remain supportive of the economic recovery
• The European sovereign debt crisis has further highlighted the importance of
vulnerabilities stemming from high private sector debt levels
• High private indebtedness has been either the main driver of country-specific
imbalances or the results of subdue growth and delayed balance-sheet repair
• Finally, such a framework should preferably account for institutional features – e.g.
the insolvency framework – which can enhance the resilience of the economy, avoid
a prolonged deleveraging cycle and support the reallocation of resources.
19 www.ecb.europa.eu ©
Rubric
Thank You.
In case of questions:
agostino.consolo@ecb.europa.eu
20 www.ecb.europa.eu ©
Rubric
21 www.ecb.europa.eu ©
Modelling
Rubric HH and NFC "equilibrium" debt levels
Purpose
• Derive an assessment of stock (private debt) imbalances
i. time varying
ii. country- and sector-specific
iii. structural/economic interpretation
• Encompass different information
iv. balance sheet information
v. macro variables
vi. interest rates
Modelling approach
• Based on the existence of a credit/debt cycle (fin. boom-bust cycle)
• Mean-reversion of debt ratios provides support for trend-cycle analysis
• Two approaches:
Panel cointegration models
Trend-Cycle decomposition (state-space models)
22 www.ecb.europa.eu ©
Data
Rubricdescription and caveats
Data description
• Country coverage: all EA countries
• Frequency: quarterly
• Sample period: 1999 - 2016 (unbalanced)
• Dependent variable (𝑫𝑫𝒕𝒕 ):
• HH loans (consumer and mortgage loans)
• NFC consolidated debt (consolidated loans + bonds)
• Explanatory variables (𝑿𝑿𝒕𝒕 ):
Balance-sheet:
HH: financial and non-financial assets
NFC: equity, fin. assets, notional lev., liquidity ratio
Macro:
HH: income, house prices, unemp. rate, demographics, share of construction value added
NFC: GDP, sector-specific investment, profits
Interest rates: lending rates, US mon policy shadow rate, corporate bond premium
Caveats:
1. For some countries, NFC debt includes loans of foreign companies
2. Some explanatory variables include large revaluation effects
3. Sample period is not homogenous for all countries and, for some, it may capture only the most
recent leverage cycle
23 www.ecb.europa.eu ©
Panel
Rubric cointegration models
𝐷𝐷𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑁𝑁 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽1 𝑃𝑃𝑃𝑃𝑃𝑃𝑖𝑖𝑖𝑖 + 𝛽𝛽2 𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 + 𝛽𝛽3 𝑈𝑈𝑈𝑈𝑈𝑈𝑖𝑖𝑖𝑖 + 𝛽𝛽4 𝐿𝐿𝐿𝐿𝐿𝐿𝑡𝑡 + 𝑒𝑒𝑖𝑖𝑖𝑖 2
𝐷𝐷𝑖𝑖𝑖𝑖𝑁𝑁𝑁𝑁𝑁𝑁 = 𝛼𝛼𝑖𝑖 + 𝛽𝛽1 𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑖𝑖 + 𝛽𝛽2 𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖𝑖𝑖 + 𝛽𝛽3 𝐺𝐺𝐺𝐺𝐺𝐺𝑖𝑖𝑖𝑖 + 𝛽𝛽4 𝑀𝑀𝑀𝑀𝑀𝑀𝑡𝑡 + 𝑒𝑒𝑖𝑖𝑖𝑖 2
Where PYN is disposable income, RMP is the middle-total population ratio, URX is the unemp. Rate, LRX is the lending rate; EQN is NFC equity,
GOS is the NFC gross operating surplus and MPR is monetary policy rate.
24 www.ecb.europa.eu ©
State-space
Rubric models
• State equation:
𝑠𝑠𝑦𝑦,𝑡𝑡 = 𝛼𝛼 + 𝑠𝑠𝑦𝑦,𝑡𝑡−1 + 𝜇𝜇𝑦𝑦,𝑡𝑡 (3.5)
𝑖𝑖 𝑖𝑖
Where 𝑋𝑋1,𝑡𝑡−1 and 𝑋𝑋2,𝑡𝑡−1 are a set of explanatory variable which are sector specific as for the panel model. 𝑠𝑠𝑦𝑦,𝑡𝑡 is the permanent component in GDP
which also drives the permanent component in HH or NFC debt. 𝑠𝑠𝑐𝑐,𝑡𝑡 instead is the cycle component in debt fluctuations. Only permanent effect
component will be used to compute the underlying benchmark level of sustainable debt
25 www.ecb.europa.eu ©
Rubric
26 www.ecb.europa.eu ©
2. HH and NFC debt dynamics
Rubric
The equation for Households (HH) and Non-Financial Corporations (NFC) debt dynamics
reads, for i = HH, NFC:
∗
𝐷𝐷𝑖𝑖,𝑡𝑡 = 1 + 𝑟𝑟𝑖𝑖,𝑡𝑡 𝐷𝐷𝑖𝑖,𝑡𝑡−1 + 𝐹𝐹𝑖𝑖,𝑡𝑡 + 𝐼𝐼𝑖𝑖,𝑡𝑡 − 𝛼𝛼𝑖𝑖 𝑆𝑆𝑖𝑖,𝑡𝑡
where, for each sector i:
• 𝜶𝜶𝒊𝒊 = parameter capturing the share of savings used for debt redemption
www.ecb.europa.eu ©
27
3. Key underlying drivers
Rubric
www.ecb.europa.eu ©
28
3.Rubric
Key underlying drivers (cont.)
5. Price deflators
• In the long term, inflation converges to the euro area historical average (99-14)
• Each deflator converges to the euro-area historical mean. Euro area real GDP,
investment and consumption deflator are, over history, different though
6. Lending rates
• Short-term and long-term lending rates are a function of the corresponding bond
yields. Yields come from BMPE capital market assumptions and the Public DSA
• A constant country-specific spread, as the one prevailing at the end of the forecast
horizon between lending rates and bond yields is used beyond the forecast horizon.
• The effective lending rate used in the debt formula is a weighted average accounting
for the share of floating loans and the composition of loan maturity
www.ecb.europa.eu ©
29
4. Output of the Private Sector DSA Tool
Rubric
1 1 1 1 1
𝑃𝑃𝑃𝑃𝑃𝑃𝑋𝑋𝑖𝑖 = ( �𝑺𝑺𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑦𝑦𝑖𝑖 + 𝑳𝑳�𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑦𝑦𝑖𝑖 ) + 𝑳𝑳�𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑖𝑖 + 𝑭𝑭
� 𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑙𝑙𝑖𝑖
3 2 2 3 3
• Similar indices can be computed for HHs (HSIX) and NFCs (FSIX) separately
www.ecb.europa.eu ©
30
Rubric
31 www.ecb.europa.eu ©
Effect
Rubric of IFI on the deleveraging process
Table
Households NFC
32 www.ecb.europa.eu ©
Effect
Rubric of IFI on the deleveraging process
Table
EU+OECD countries EU countries
NPL ratio 3yr NPL ratio 3yr
NPL ratio 3yr neg. change NPL ratio 3yr neg. change
pos. change (absolute value) pos. change (absolute value)
33 www.ecb.europa.eu ©
Rubric
34 www.ecb.europa.eu ©
Annex:
Rubric NFC Debt-to-GDP vs Debt-to-Equity ratio
Source: Eurostat and ECB, Quarterly national accounts Source: Eurostat and ECB, Quarterly national accounts
Note: Peak (2005-2015) refers to the max value over the period Note: Peak (2005-2015) refers to the max value over the period
35 www.ecb.europa.eu ©
ARubric
simple decomposition of the Debt-to-GDP ratio
Chart 5: change in the NFC debt-to-GDP ratio and related contributions over 2009Q3-2016Q4
36 www.ecb.europa.eu ©