Foreign Currency Borrowing in Hungary: the Pricing Behavior of Banks

SEM 2nd Conference
OECD Paris
22-24. July 2015
Pricing foreign currency debt in Hungary
Zoltán Schepp – Mónika Pitz
University of Pécs
Foreign currency borrowing in CEE Countries
• On growing contemporary literature on the cross-country
properties of foreign currency borrowing
 Rosenberg-Tirpák (2009): studies lending motives of banks
 Fidrmuc et.al. (2011): examines motives
 Brown et.al. (2011): specialties of small firms borrowing
 Beckmann et.al. (2012): searches for reasons behind late
repayment
 Brown-De Haas (2012): relationship between foreign assets and
foreign currency denominated lendingng
 Yesin (2013): compares systemic risk of countries
 Brown er.al. (2014): role of asymmetric information
• Hungarian case: unique in many terms
 Competing narratives but quite little empirics
The story to understand
• In Hungary, the proportion of the foreign currency loans in the
total loan volume reached 70%, most of which was in Swiss francs.
• Crisis: higher interest rates and depreciation of HUF
 Increase of monthly instalments
 20% non-performing loans
• Who should take the responsibility?
– Households: lack of financial knowledge, irresponsible risk
taking
– Authorities: no prevention
– Banks: shifting the risks to households
• Core policy and political question:
– Had the Hungarian banks priced their power to the retail
market, or passed through the shocks in funding costs faced by
themselves!?
What was/is different in Hungary?
The proportion of foreign currency loans to HH-s and non-financial companies in
the EU, March 2011 (source: European Systemic Risk Board)
What was/is different in Hungary? (cont.)
The proportion of euro and other (mainly CHF) denominations in the foreign
currency loans in the EU, Aprill 2011 (source: European Systemic Risk Board)
Specialties of the the Hungarian case
• Contrast between individual and (non explicit!) systemic risk
 Wide ER band (+-15%, from spring 2008: floating ER)
 In the Baltic countries: narrow peg to the euro
• No significant transfers from guest workers in western Europe
 Contrary to e.g. Romania
• To high swizz franc ratio
 Compared to Baltic countries (with similar indebtness)
• To high indebtness
 Compared to countries in them swizz franc debt played also an
important role (Austria, Poland)
• High international capital and trade openness
 ca. 200 Bn. EUR total foreign liabilities, and 50 Bn. EUR net
foreign debt (2011 autumn), close to 200% trade/GDP ratio
 This is not unique but of course important
HUF/CHF exchange rate movements
The HUF/CHF rate depreciated on average 20-25%, and the CHF/EUR
appreciated another 20% between 2005 and 2013.
280.00
260.00
240.00
220.00
200.00
180.00
160.00
140.00
120.00
100.00
2005.
2006.
2007.
2008.
2009.
2010.
2011.
2012.
2013.
2014.
Common macro-environment for taking/lending foreign
currency denominated credits in Hungary
• Basic motivational factors
 Currency risk premium on Hungarian financial assets (e.g.
government bonds) – expected cost cut
 Bad properties of HUF denominated credits – to high interest
rates, volatility of term premium of longer maturities
• Factors hindering/distorting risk perception
 Interest rate reactions of the NBH between 2003 and 2008
 Political promises about euro zone accession (between 2002
and 2008 always 5 year ahead of…)
 Disrupted („risk-based”) competition on the banking market
 Systematic underestimation/ignoring of lending risk
Misperception and miscalculation of risk
propability
perceived
de facto
cost
E[C]
Sector specific motivations in the case of local
governments (LG) and business sector (BS)
• The retention (‘own-contribution’) needed to obtain grants
from EU Structural and Regional Funds
 (LG) no liquid capital, they raised the necessary funds by
issuing foreign currency bonds with a maturity of 20 to 25 years
 (BS) In addition to EU subsidies, banks provided FX-loans to
fund real estate projects (weak income-generating capacity)
• The effects of the partial fiscal consolidation carried out by the
second Gyurcsány government beginning in the middle of 2006
 (LG) changes in the terms of financial support and task-sharing
to the disadvantage of local authorities
 (BS) missing demand at companies which had tried to
compensate for declining income by the cost-benefits of FXloans or even by carry trade speculation based on “forwardrate-bias”
Sector specific motivations in the case of local
governments (LG) and business sector (BS) (cont.)
• The low interest rates on FX-loans permitted higher leverage
 (LG) a substantial number of local authorities had been dealing
with long-run financial problems, so issuing foreign currency
bonds simply to kept their economic scope for action alive
 (BS) the liquid part of the equity of Hungarian-owned SMEs was
too low compared to the level of the firm pre-crisis economic
activity and was replaced by the relatively cheap and easily
available sources of credit
• Consequences (after crisis has hit)
 Local governments: the central government took over the
cumulative bank liabilities (ca. €4.5 Bn., between 2011-2014)
 Corporates needed liquidity or restructuring, so „forintisation”
and/or IR change has happent often relative fast and in a
cooperative way (although with not equal powered parties…)
Why has become households foreign currency indebtness
a systemic financial problem? The story once again.
Higher interest
rate (hh loans)
Exchange rate
depreciation
?
Reference rate +
risk premium
Increased
instalments
Moral hazard
Late
repayments
Lower
lending,
GDP…
Higher
unemployment
What else could effect interest rates?
• We assume basically four main price shocks:
– Reference rate
External funds of banks
– Risk premium
– Loan portfolio quality
– Fiscal burden
SVAR
– Interest rate pricing (stock and new loans)
• Housing loans
• Non-purpose mortgage loans
Data
• Time span
 Monthly data between 2005M1 and 2013M12
• Reference rate
 3 month CHF (chflibor) in base points
• Risk premium
 5Y sovereign CDS (cds) in base points
• Loan portfolio quality:
 impairment rate (impair)
 used proxy: recognized impairment of assets (shows the
losses caused by non-performing loans) / % of total assets
• Fiscal burden:
 Corporate tax, special bank tax (2010-), early repayment
losses, financial transaction tax/duty / % of total assets
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
-100
2005-01
2005-05
2005-09
2006-01
2006-05
2006-09
2007-01
2007-05
2007-09
2008-01
2008-05
2008-09
2009-01
2009-05
2009-09
2010-01
2010-05
2010-09
2011-01
2011-05
2011-09
2012-01
2012-05
2012-09
2013-01
2013-05
2013-09
Data
bázispont
bázispont
chflibor
cds
housing
mew
burd_hh
impair_hh
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
-100
Structural VAR model (2008M10-2013M12)
Household sector:
B0zt = k+B1zt-1+B2zt-2+…+Bpzt-p+ut
zt=(d(chflibor)t, d(burd_hh)t, d(cds)t, d(impair_hh)t, d(y)t)T
Corporate sector:
B0zt = k+B1zt-1+B2zt-2+…+Bpzt-p+ut
zt=(d(euribor)t, d(burd_corp)t, d(cds)t, d(impair_corp)t,
d(y)t)T
k= c + crisist
Identification
• Constraints on immediate effects are based of theoretical
considerations
0
0
0 0
1
0

1
0
0
0


B0   0 b32 1
0 0
0 b

b
1
0
42
43


b51 b52 b53 b54 1
burd
cds
impair
y
libor
Impulse response functions – housing loans
Accumulated Response to Nonfactorized One S.D. Innovations ± 2 S.E.
Accumulated Response of D(HOUSING) to D(CHFLIBOR)
Accumulated Response of D(HOUSING) to D(BURD_HH)
8
8
4
4
0
0
-4
-4
-8
-8
-12
-12
1
2
3
4
5
6
7
8
9
10
Accumulated Response of D(HOUSING) to D(CDS)
1
3
4
5
6
7
8
9
10
Accumulated Response of D(HOUSING) to D(IMPAIR_HH)
8
8
4
4
0
0
-4
-4
-8
-8
-12
2
-12
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Impulse response functions – mortgage equity
withdrawals
Accumulated Response to Nonfactorized One S.D. Innovations ± 2 S.E.
Accumulated Response of D(MEW) to D(CHFLIBOR)
Accumulated Response of D(MEW) to D(BURD_HH)
8
8
4
4
0
0
-4
-4
-8
-8
1
2
3
4
5
6
7
8
9
10
1
Accumulated Response of D(MEW) to D(CDS)
2
3
4
5
6
7
8
9
10
Accumulated Response of D(MEW) to D(IMPAIR_HH)
8
8
4
4
0
0
-4
-4
-8
-8
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Impulse response functions – corporate loans
Accumulated Response to Nonfactorized One S.D. Innovations ± 2 S.E.
Accumulated Response of D(CORP) to D(EURIBOR)
Accumulated Response of D(CORP) to D(BURD_CORP)
30
30
20
20
10
10
0
0
-10
-10
-20
-20
1
2
3
4
5
6
7
8
9
10
Accumulated Response of D(CORP) to D(CDS)
1
3
4
5
6
7
8
9
10
Accumulated Response of D(CORP) to D(IMPAIR_CORP)
30
30
20
20
10
10
0
0
-10
-10
-20
2
-20
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Statments based on SVAR results
• Results are highly sensitive to the sample
 Between 2005M1 and 2012M3 increased tax burdens,
augmented country risks and a declining readiness to pay
contributed to the rising interest rates of housing loans.
(Pitz-Schepp, 2013)
 The price flexibility of the population may have been
higher in the case of free-purpose mortgage loans, which
restricts pricing opportunities of banks (2005M1-2012M3)
 When analyzing separately the post-crisis time period
2008M10-2013M12 we found that the above effects no
longer appeared and that only the reference rate
(negative sign!) and the CDS were decisive.
 According to the corporate sector, only the effect of
euribor proved to be significant.
VECM for housing loans
• We suppose a long run equilibrium relationship between
cost components and IR on existing credit stock.
 2005M1-2013M12 monthly data, no crisis dummy
 Lags determined by LR test (hh: 2, mew: 1 period)
• Johansen procedure: cointegrating vector exists
 With all four cost factors being significant
• Question in the politics:
 Where IR changes unfair?
• Policy question:
 Was it pricing to the market or cost based pricing?
• The opportunity to change (the institutional setup) was a
failure in itself.
 but 7,5 year long nothing has happen in the politics!
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
-100
2005-01
2005-05
2005-09
2006-01
2006-05
2006-09
2007-01
2007-05
2007-09
2008-01
2008-05
2008-09
2009-01
2009-05
2009-09
2010-01
2010-05
2010-09
2011-01
2011-05
2011-09
2012-01
2012-05
2012-09
2013-01
2013-05
2013-09
Data
bázispont
bázispont
chflibor
cds
housing
mew
burd_hh
impair_hh
1400
1300
1200
1100
1000
900
800
700
600
500
400
300
200
100
0
-100
Vector error correction modelling (2005-2013)
• For housing loans we used a two-lag model and for mortgage
equity withdrawals a single-lag model
• The Coefficients of the Normalised Co-integrating Vector and
Standard Errors
housing
-1
mew
-1
chflibor
0,96
(0,33)
1,15
(0,29)
Burd
-1,73
(0,69)
-3,53
(0,63)
cds
0,66
(0,11)
0,65
(0,10)
impair
0,56
(0,14)
0,91
(0,14)
• With the Johansen procedure we estimated the cointegrating
equations expressing the long-run dynamic, which indicated a
significant relationship for all four cost typess.
The Significant Coefficients of the Estimated
VEC Models and Test Results
Coefficient
Standard errors
R-squared
Error
correction
-0.012
0.005
0.456
0.392
Error
correction
-0.020
0.005
0.197
Adj. R-squared
0.148
housing
Coefficient
Standard errors
R-squared
Adj. R-squared
mew
d(housing(-1)) d(housing(-2))
0.237
0.221
0.099
0.095
C
2.76
0.77
ER coefficients and constant are significant in both cases.
C
1.360
0.542
A possible narrative
Interest rate
S(huf,t0)
MC(chf,t1)
E[MC]=S(chf,t0)
D(huf,t0)
E[MU]=D(chf,t0)
Stock (chf)
debt
Debate on unfair-banking in Hungary
• Wide ranging conceptual confusions is associated with the
notion of fairness and naivety of banks by repricing:
– IR changes where „unfair” (eg. reverse to LIBOR changes)
– Banks where „naive” because higher IR-s leading to worse
quality of loan portfolio
• The VECM show significant long run equilibrium relationship
between existing cost factors and mortgage interest rates.
• Proportional and symmetric: see LIBOR
• Less than proportional: impairmant (losses) and cds
• Inverse (asymmetric) relationship: taxes
– But this is a good news for the debtors!
• Banks were not naive by IR changes, but they had from the
beginning market power and a kind of monopolistic price
Conclusion
• All four types of shock might have been played a role in
determining interest rates for housing loans, i.e. the cost
shocks of banks are more or less accurately reflected in the
interest rates applied by domestic banks.
• In a long run view pricing seemed to be cost-based.
 This cost have been covered by the debtors, but very
recently banks has to take beck it (some 3 Bn. EUR!)
 This is not necessary fair, because of the regulatory failures
also has been made for a very long time.
• All three parties (state, banks, debtors) made mistakes
 everyone tried to act bilateral based on power distribution
 But the best solution of the problem needs (would have
needed…) a cooperation of all three parties
Many thanks for your attention!