MACROPRUDENTIAL TOOLS: CALIBRATION ISSUES IN CENTRAL, EASTERN AND SOUTHEASTERN EUROPE

Transcription

1 MACROPRUDENTIAL TOOLS: CALIBRATION ISSUES IN CENTRAL, EASTERN AND SOUTHEASTERN EUROPE Adam Gersl Joint Vienna Institute World Bank Workshop on Macroprudential Policymaking in Emerging Europe Vienna, June 2, 2016 The views expressed here are those of the author and do not necessarily reflect those of the JVI, its management, or its board. ALBANIA ARMENIA AZERBAIJAN BELARUS BOSNIA AND HERZEGOVINA BULGARIA CROATIA CZECH REPUBLIC ESTONIA GEORGIA HUNGARY KAZAKHSTAN KOSOVO KYRGYZ REPUBLIC LATVIA LITHUANIA FYR MACEDONIA MOLDOVA MONTENEGRO POLAND ROMANIA RUSSIAN FEDERATION SERBIA SLOVAK REPUBLIC SLOVENIA TAJIKISTAN TURKEY TURKMENISTAN UKRAINE UZBEKISTAN

2 Focus of the presentation Caps on LTV on mortgage/housing loans in Central, Eastern and Southeastern Europe (CESEE) A pragmatic way to calibrate and/or assess the level of LTV caps (not only in CESEE countries) agersl@jvi.org 2

3 LTV caps in CESEE 3

4 Caps on LTV became popular in CESEE Number of countries in CESEE with LTV caps Pre-crisis wave Post-crisis wave agersl@jvi.org 4

5 Pre-crisis wave of LTV caps (I) Adopted as a part of a package of measures to tame credit growth and strong FX lending in Country LTV cap Time period Poland 75% Croatia 75% Latvia 90% Often combined with limits/recommendations on DSTI, risk weights, reserve requirements etc. Some countries conditioned other instruments (risk weights) on LTVs (Bulgaria) agersl@jvi.org 5

6 Pre-crisis wave of LTV caps (II) Communication not always very clear due to the missing macroprudential framework objectives countercyclical use calibration Despite that, they seem to have contributed to decrease credit growth (especially when combined with or substituted by limits on DSTI) 6

7 The effects of pre-crisis LTV caps Gersl and Jasova (2014, ES): survey of 11 CEE countries during , total of 82 measures analyzed, out of which 5 cases of LTV/DSTI measures Provisioning rules and limits on LTV/DSTI found effective in curbing credit growth in a panel estimation Source: Gersl, A., Jasova, M., 2014, Measures to tame credit growth: are they effective? Economic Systems, Volume 38, Issue 1, March 2014, Pages

8 Post-crisis wave of LTV caps (I) 1. Continuing the fight against FX lending Country LTV cap FX LTV cap Time period Hungary (1) 75% 60% euro % other Hungary (2) 80% 50% euro % other Romania 85% 80% hedged 75% unhedged euro 60% unhedged other Serbia... 80%

9 Post-crisis wave of LTV caps (II) 2. Limiting the unintended side effects of extremely loose post-crisis monetary policy on the housing loans/real estate market Country LTV cap Time period Lithuania 85% Poland 80% (by 2017) Slovakia 90% Estonia 85% Czech Republic 90%

10 Post-crisis wave of LTV caps (III) Much more elaborated measures compared to the first wave, especially for the subwave existing macroprudential framework intensive external communication provisions to limit arbitrage Modalities to allow banks to provide a part of the housing loans above the LTV limit additional collateral or guarantee (PL, LV, LT) fixed percentage (10-25% of loans in CZ and SK) agersl@jvi.org 10

11 Calibration issues 11

12 1. The objective of the tool Macroprudential instruments and frameworks: a stocktaking of issues and experiences. CGFS Paper No 38, agersl@jvi.org 12

13 The ESRB Handbook on Operationalising Macro prudential Policy in the Banking Sector,

14 2. The volatility of house prices more volatile house prices => lower LTV safer as it mutes the PD/LGD channel 60% 50% 40% 30% 20% 10% 0% Peek to trough correction of house prices during the crisis (%) Standard deviation of yoy growth of house prices (%) 14

15 Pragmatic calibration: preventing the under-water situation Assumption: in stress, the ex-post LTV should not increase above a pre-determined stress LTV level (for example 100%): 1 % 1 % The resulting cap depends on the position in the financial cycle within the upturn phase: more time to amortize and to benefit from increases in house prices before a correction if at the peak: no amortization and no house price gains, only correction agersl@jvi.org 15

16 Simulation for the peak Assumptions: stress LTV = 100% 100% 80% 60% 40% 20% 0% correction = max (observed correction; 2*stdev) Desirable LTV cap at the peak Current LTV cap agersl@jvi.org 16

17 Simulation for the upturn phase Assumptions: 140% 120% 100% 80% 60% 40% 20% 0% effect of amortization (10%) and house price gains (20%) decreases the LTV from the level at origination by about 1/4 Desirable LTV cap in the upturn Current LTV cap 17

18 Ideal calibration: conditional forecasts Ideally, in each point of financial cycle, one would like to know (at a minimum): when the cycle turns from upturn to downturn by how much do house prices increase before the cycle turns what is the expected correction of house prices during the downturn Not easy, as things are interrelated, but conditional forecasts could be done using typical patterns of house price cycles (research in progress) agersl@jvi.org 18

19 3. Sensitivity of credit losses from mortgage borrowers that are under water The strength of the PD/LGD channels stress LTV can be higher than 100% if low sensitivity The role of: legal system (stigma of personal bancruptcy) financial assets and wealth (Netherlands) remittances and family help-outs the liquidity of the secondary real estate market 19

20 4. Other existing measures increasing resilience Less strict LTV caps if borrowers resilient to income shocks (strict DTI/DSTI regulation), or banks can withstand high credit losses from mortgages due to large countercyclical capital buffer or other capital buffers higher risk weights on housing loans However, beware different transmission channels of borrower-based versus capital-based tools 20

21 5. FX housing loans 1 % 1 % 1 % 70% 60% 50% 40% 30% 20% 10% 0% Hungary Romania Serbia Peek to trough change (2Y) in exchange rate vis a vis EUR (%) Standard deviation of yoy change of exchange rate vis a vis EUR (%) 70% 60% 50% 40% 30% 20% 10% 0% Hungary Romania Serbia Peek to trough change (2Y) in exchange rate vis a vis CHF (%) Standard deviation of yoy change of exchange rate vis a vis CHF (%) agersl@jvi.org 21

22 Simulation for the peak, FX loans stress LTV = 100%, correction in house prices = max (observed correction; 2*stdev), depreciation of the exchange rate = 3 standard deviations (20% 30%) 100% 100% 80% 60% 40% 20% 80% 60% 40% 20% 0% 0% Hungary Romania Serbia Hungary Romania Serbia Desirable LTV cap at the peak (EUR loans) Desirable LTV cap at the peak (CHF loans) Current LTV cap (EUR loans) Current LTV cap (CHF loans) agersl@jvi.org 22

23 Conclusions LTV caps a popular tool in CESEE, but differences in the design If LTV caps understood as a measure to increase banks resilience: either need to be adjusted countercyclically to react to the position in the financial cycle (change in the risk of house price corrections/exchange rate corrections) or, if kept fixed, other instruments need to be tightened along the cycle to increase banks capacity to bear additional credit losses. agersl@jvi.org 23

24 Thank you! Adam Gersl Senior Economist Joint Vienna Institute 24