Restricted

Higher Bank Capital Requirements and Mortgage Pricing:

Evidence from the Counter-Cyclical Capital Buffer*

Christoph Basten (FINMA & ETH Zurich) & Cathérine Koch (BIS) Joint workshop by the Research Task Force of the BCBS, the CEPR and the JFI. 22-23 January 2015

* Any views expressed are those of the authors and do not necessarily reflect those of the BIS, ETH Zurich or the Swiss Financial Market Supervisory Authority FINMA.

Restricted

1 The Counter Cyclical Buffer (CCB)

The Counter Cyclical Buffer (CCB) as macroprudential tool of Basel III. Basel II

8% minimum capital requirements (MCR) of RWA Capital requirements had a pro-cyclical effect Swiss implementation in 2007

Basel III (Swiss law since June 20, 2012)

8% minimum capital requirements (MCR) based on RWAs Capital Conservation Buffer: In Switzerland 2.5-6.4% of RWAs On top, authorities can temporarily activate the Countercyclical Capital

Buffer - Extra equity for up to 2.5% of RWAs - Goals (with equal weights in e.g. Swiss law):

• more equity to bear potential losses • moderate lending growth during booms

CCB: Interesting tool when monetary policy already committed.

2

Restricted

1 Research Question(s)

«How does the CCB affect mortgage pricing in Switzerland?»

We examine the interaction with :

... bank sensitivity measures (capital-constrained, mortgage-specialized)

... risk-weighting schemes tied to LTV ratios

… differences between banks and insurers. «Policy leakage?»

3

Restricted

1 Switzerland: Timing of Policy Changes

Focus on CCB activation (Feb. 13, 2013)

rather than availability (July 1, 2012)

Extra capital worth 1% of RWAs secured by domestic residential property transition period until September 2013

Clean of other changes Basel III law adoption, new mortgage market regulation

4

Restricted

1 Key Findings

CCB’s activation impacts the composition of mortgage supply:

Constrained banks with low capital cushions … Mortgage-specialized banks … … raise prices more. => Shift mortgages from less to more resilient banks

Banks charge extra on high LTVs with high risk weights,

but risk-weighting schemes do not amplify CCB effects.

Insurers (not CCB subjected) raise rates as well. => No Leakage

5

Restricted

1 Our Contributions

Disentangle

mortgage demand (borrower) and mortgage supply (lender)

Composition of mortgage supply. First empirical paper on how CCB affects loan pricing

(offers as opposed to contracted rates)

Link distinct offers to bank balance sheet characteristics as bank sensitivity measures

Assess effectiveness of risk-weighting schemes (LTV Thresholds)

Investigate possible policy leakage.

6

Restricted

1 Data on Swiss Mortgage Market

Online mortgage platform www.comparis.ch

Customer ... provides data on financial situation, real estate property, requested

mortgage amount, maturity… pays CHF 148 (about USD 160)

Lenders (banks and insurers) …

get anonymized customer data, submit offers or rejections, if offer: interest rates on (tranches of) the mortgage, cannot see the responses of competitors.

⇒ Multiple offers per individual request ⇒ Many distinct offers per bank over time

7

Restricted

Outline

1) Introduction

2) Empirical Approach and Results

3) Summing Up

8

Restricted

Which banks do exhibit the highest sensitivity to the

CCB?

9

Restricted

2.1 Bank Sensitivity Measures

The CCB applies to … mortgages on balance sheets new mortgage issuance In response to the CCB: H1a: Constrained banks with low capital cushions … H1b: Specialized banks with a mortgage-intensive business model per

unit of capital … … raise mortgage rates relatively more.

10

Which banks do exhibit the highest sensitivity to the CCB?

Restricted

2.1 Bank Sensitivity Measures

𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼1 + 𝛽𝛽1𝑠𝑠𝑅𝑅𝑠𝑠𝑠𝑠𝑖𝑖, 201𝑥𝑥 +𝛽𝛽2𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖𝑠𝑠𝑅𝑅𝑠𝑠𝑠𝑠𝑖𝑖, 201𝑥𝑥 +𝐹𝐹𝐹𝐹𝑖𝑖 +𝐹𝐹𝐹𝐹𝑗𝑗 +𝜀𝜀𝑖𝑖𝑗𝑗

With Customer i and lender j 𝑠𝑠𝑅𝑅𝑠𝑠𝑠𝑠𝑖𝑖, 201𝑥𝑥 as balance sheet data from the past annual report. 𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖= 1 if i > Feb 13, 2013 and 0 otherwise Using request 𝐹𝐹𝐹𝐹𝑖𝑖 implicitly controls for time, individual customer risk,

property type, location and macroeconomic developments. Robust Standard Errors

11

Which banks exhibit the highest sensitivity to the CCB?

Restricted

2.1 Bank Sensitivity Measures

12

Effects of the CCB Banks that are • capital constrained • mortgage-specialized … charge higher rates.

⇒ Cannot reject H1a &b ⇒ Change in the

composition of supply Placebo: Liquid banks charge lower rates in general, but do not change their behavior after the CCB.

With controls; FE (request, lender, robust SE)

Offered Mortgage Rate

(1) (2) (3) (4)

Sensitivity Measures

Constrained 9.19 2.74(6.01) (6.29)

CCB*Constrained 4.26*** 2.72**(0.90) (1.20)

Specialized 3.93 4.03(2.49) (2.52)

CCB*Specialized 6.11*** 5.57***(0.92) (1.26)

Liquid -2.61 -4.21**(1.72) (1.83)

CCB*Liquid -2.08** 0.49(0.89) (1.33)

Constant 190.32*** 194.81*** 201.26*** 191.07***(6.17) (2.84) (1.42) (6.85)

Observations 4,045 4,045 4,045 4,045R-squared 0.83 0.83 0.83 0.83

Full Sample

Which banks exhibit the highest sensitivity to the CCB?

Restricted

Do risk-weighting schemes amplify the CCB effect?

13

Restricted

2.2 Risk-Weighting Schemes

CCB: extra capital worth 1% of RWAs => Loan-to-value (LTV) ratio 35% risk weight on tranches with LTV≤ 66 75% risk weight on tranches with 66<LTV ≤ 80 100% risk weight on tranches with LTV > 80

14

H2: Risk-weighting schemes linked to LTV ratios amplifiy the CCB effect and lead to a relatively larger increase in mortgage rates • for borrowers with LTV

ratios above 66% • and for yet a larger

increase for LTV ratios above 80%.

Do risk-weighting schemes amplify the CCB effect?

Restricted

2.2 Risk-Weighting Schemes

𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼1 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝑙𝑙𝑖𝑖 +𝛽𝛽2𝑙𝑙𝑅𝑅𝑙𝑙67𝑖𝑖 +𝛽𝛽3𝑙𝑙𝑅𝑅𝑙𝑙80𝑖𝑖 +𝛽𝛽4 𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖𝑙𝑙𝑅𝑅𝑙𝑙𝑙𝑙𝑖𝑖 + 𝛽𝛽5𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖𝑙𝑙𝑅𝑅𝑙𝑙𝑙0𝑖𝑖 +𝛾𝛾1𝑟𝑟𝑅𝑅𝑟𝑟𝑖𝑖𝑠𝑠𝑖𝑖 +𝛾𝛾2′𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖 + 𝐹𝐹𝐹𝐹 +𝜀𝜀𝑖𝑖𝑗𝑗

With 𝑙𝑙𝑅𝑅𝑙𝑙𝑙𝑙𝑖𝑖 =I(𝑙𝑙𝑅𝑅𝑙𝑙𝑖𝑖 ≥ 67) and 𝑙𝑙𝑅𝑅𝑙𝑙𝑙0𝑖𝑖 =I(𝑙𝑙𝑅𝑅𝑙𝑙𝑖𝑖 ≥ 80) 𝑟𝑟𝑅𝑅𝑟𝑟𝑖𝑖𝑠𝑠𝑖𝑖 the Swiss 10y swap rate 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖` =(income; wealth; debt; age) 𝐹𝐹𝐹𝐹 = (FEj; FE_monthi; FE_typei; FE_cantoni)

15

Do risk-weighting schemes amplify the CCB effect?

Restricted

2.2 Risk-Weighting Schemes

16

Find: • LTV per se insignificant, but

banks charge extra for LTV ≥ 66 and LTV ≥ 80.

• LTVs Thresholds do not amplify the CCB effect.

⇒ Reject H2. ⇒ Suggests thresholds flesh out risk, but

risk weights do not amplify the CCB. ⇒ CCB not risk-sensitive

With controls ;FE (time, lender, type, canton);robust SE

Offered Mortgage Rate

(1) (2) (3) (4) (5)

Mortgage Characteristics

LTV 0.03 0.03 0.03 0.03 0.03(0.02) (0.02) (0.02) (0.02) (0.02)

LTV67 (0/1) 2.13*** 2.58*** 2.39*** 2.38*** 2.35***(0.70) (0.69) (0.70) (0.70) (0.70)

LTV80 (0/1) 1.81** 1.85** 1.57** 1.56** 1.54**(0.75) (0.74) (0.75) (0.75) (0.75)

CCB*LTV67 (0/1) -1.50 -1.49 -1.49 -1.49 -1.52*(0.92) (0.91) (0.91) (0.91) (0.91)

CCB*LTV80 (0/1) 0.87 1.34 1.45 1.46 1.48(1.17) (1.15) (1.15) (1.15) (1.15)

Refinancing ControlSwap Rate 10y 73.69*** 75.11*** 74.41*** 74.37*** 74.27***

(4.69) (4.66) (4.66) (4.67) (4.67)Request ControlsIncome -3.91*** -3.14*** -3.15*** -3.20***

(0.47) (0.51) (0.51) (0.52)Wealth -0.84*** -0.84*** -0.81***

(0.22) (0.22) (0.23)Debt (0/1) 0.14 0.18

(0.54) (0.54)Age -0.02

(0.02)Constant 131.33*** 174.16*** 177.18*** 176.75*** 177.47***

(23.03) (23.35) (23.33) (23.35) (23.38)

Observations 4,045 4,045 4,045 4,045 4,045R-squared 0.76 0.76 0.76 0.76 0.76

Do risk-weighting schemes amplify the CCB effect?

Restricted

What about «leakage effects»?

17

Restricted

2.3 Banks versus Insurers

The CCB requires banks to hold extra capital, but not insurers. Insurers might… anticipate banks to raise prices. underbid banks and expand their market shares (“policy leakage”). prefer to keep their market share constant, attempt to boost profits. Maybe insurers tailor their mortgage issuance to an optimized, well-diversified, low-risk portfolio. => portfolio limits => raise their prices in parallel with banks => boost profits H3: CCB-exempt insurers tailor their mortgage rates to self-imposed optimal portfolio limits that in turn induce them to raise their mortgage rates instead of winning market shares from banks.

18

What about «leakage effects»?

Restricted

2.3 Banks versus Insurers

𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑖𝑖𝑖𝑖 = 𝛼𝛼1 + 𝛽𝛽1𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖𝑐𝑐𝑅𝑅𝑠𝑠𝑏𝑏𝑖𝑖 +𝛽𝛽2𝑐𝑐𝑐𝑐𝑐𝑐𝑖𝑖𝑠𝑠𝑛𝑛𝑠𝑠𝑐𝑐𝑖𝑖 +𝐹𝐹𝐹𝐹𝑖𝑖 +𝐹𝐹𝐹𝐹𝑗𝑗 +𝜀𝜀𝑖𝑖𝑗𝑗 With 𝑐𝑐𝑅𝑅𝑠𝑠𝑏𝑏𝑖𝑖 = 1 if lender is a bank and 0 otherwise

𝑠𝑠𝑛𝑛𝑠𝑠𝑐𝑐𝑖𝑖 = 1 if lender is an insurer and 0 otherwise Test for the difference between 𝛽𝛽1 and 𝛽𝛽2 (Puri et al., 2011)

19

What about «leakage effects»?

Restricted

2.3 Banks versus Insurers

20

Find: Both banks and insurers raise their mortgage rates. On average insurers raise their offered rates by 9.05 bp more than banks. ⇒Cannot reject H3. ⇒ Insurers do not opt for higher

market share. ⇒ portfolio limits

FE

Offered Mortgage Rate

(1) (2)

(a) CCB*NONB 67.72*** 69.44***

(3.13) (3.09)(b) CCB*BANK 58.66***

(2.99)(c) CCB*KANTONALBANK 56.06***

(2.95)(d) CCB*FOREIGNBANK 61.87***

(3.05)(e) CCB*OTHERBANK 63.25***

(3.03)Constant 203.23*** 205.96***

(12.85) (13.92)

Observations 5,459 5,459R-squared 0.82 0.82

Wald EstimatesDID estimate (a)-(b) 9.05***Wald test (a)-(b) p-value 0.00DID estimate (a)-(c) 13.38***Wald test (a)-(c) p-value 0.00DID estimate (a)-(d) 7.57***Wald test (a)-(d) p-value 0.00DID estimate (a)-(e) 6.19***Wald test (a)-(e) p-value 0.00

Full Sample

What about «leakage effects»?

Restricted

2.4 Robustness

Subsamples New mortgages vs rollovers requests Best offers only LTV>66 and LTV>79 only Homogenous 10y fixed sample without variable components

Estimation Drop FE-lender Standard Errors: Cluster by request

21

Restricted

Outline

1) Introduction

2) Empirical Approach and Results

3) Summing Up

22

Restricted

3 Summing Up

Sensitivity Measures We find that the CCB changes the composition of mortgage supply . Capital-constrained banks raise prices more after CCB activation. Mortgage Specialized banks raise prices more after CCB activation. => Shift mortgages from less to more resilient banks.

LTV Thresholds/Risk-Weighting Schemes Banks charge extra on high LTVs with high risk weights,

but risk-weighting schemes do not amplify CCB effects.

Banks vs. Insurers Insurers (not CCB subjected) do also raise their rates.

=> No «Leakage»

23

«How does the CCB affect mortgage pricing in Switzerland?»

Restricted

Thank you!

24