Banks And Interest Rate Risk

Banks and interest rate risk Ila Patnaik Ajay Shah

Banks and interest rate risk – p. 1/42

Broad issues

Banks and interest rate risk – p. 2/42

Interest rates dropped dramatically... 14

10 years

12

10

8

6

10-09-1997

25-05-1998

28-01-1999

05-10-1999

Time

17-06-2000

03-03-2001

10-11-2001

19-07-2002

Banks and interest rate risk – p. 3/42

NPV of a bank We convert assets into cashflows ai and liabilities into cashflows li . Then:

A(0) =

N X i=1

L(0) =

N X i=1

ai (1 + z(ti ))ti li (1 + z(ti ))ti

Banks and interest rate risk – p. 4/42

If interest rates went up? If we get a parallel shift of the yield curve of ∆:

A(∆) =

N X i=1

L(∆) =

N X i=1

ai (1 + ∆ + z(ti ))ti li (1 + ∆ + z(ti ))ti

The impact upon equity capital is (A(∆) − A(0)) − (L(∆) − L(0)). Banks and interest rate risk – p. 5/42

Think NPV, think MTM! A great deal of confusion comes out of “the earnings perspective”. In terms of core economics, we need full MTM of both assets and liabilities. In India, banking regulation, and many banking professionals, do not yet think MTM, do not yet think NPV.

Banks and interest rate risk – p. 6/42

Problem 1: Reduce a bank into a set of cashflows

Banks and interest rate risk – p. 7/42

Imputation of cashflows Life would be very easy if: 1. The banking regulator asked banks to report cashflows for assets and liabilities in time buckets, and 2. Made these disclosures public.

Banks and interest rate risk – p. 8/42

Repricing date versus maturity If a loan to a company is PLR-linked, it’s really six month maturity, regardless of the loan duration. Assets and liabilities can be classified by time to repricing or time to maturity.

Banks and interest rate risk – p. 9/42

“Core versus volatile” demand deposits Technically, savings accounts and current accounts are maturity 0. In practice, there is a lot of stability. One can own some long assets, backed by demand deposits, and be safe. How much is core? How much is volatile?

Banks and interest rate risk – p. 10/42

Where we stand in India on disclosure Banks are required to classify assets by time to maturity (but not repricing), and show this in the annual report. Banks are required to classify assets by time to repricing, and submit this to RBI, but this is not publicly released. NYSE has superior disclosure standards, so ICICI Bank and HDFC Bank release good data. There is no attempt at disclosing cashflows.

Banks and interest rate risk – p. 11/42

How to make progress? Complicated algorithms to impute cashflows out of public domain disclosure. 500 lines of perl. See Interest rate risk in the Indian banking system, by Ila Patnaik and Ajay Shah.

Banks and interest rate risk – p. 12/42

Example of (31/3/2002)

cashflows

for

SBI

(Rs. crore) Bucket

Assets Liabilities

Zero 0-1mth 1-3mth 3-6mth 6-12mth 1-3yrs 3-5yrs > 5yrs

12409 41659 18382 21927 87411 43282 31882 80285

34262 8053 5113 7483 15421 174229 55414 9944 Banks and interest rate risk – p. 13/42

Problem 2: What shocks to worry about?

Banks and interest rate risk – p. 14/42

BIS Proposal Look at five years of daily data for the long rate. Compute the time-series of change-over-one-year Take the 1th and 99th percentile of this distribution.

Banks and interest rate risk – p. 15/42

Using Indian data The NSE ZCYC database allows us to do this. We use 1/1/1997 - 31/7/2002. In India, there are 288 days per year. The shock to worry about is 320 bps.

Banks and interest rate risk – p. 16/42

One of the world’s highest IR vol Rank Country 1 2 3 4 5 6 7 8

Turkey Chile India Mexico U.K. Indonesia Poland Philippines

Volatility 32.93 1.74 1.72 1.36 0.91 0.88 0.81 0.77

Source: Baig (2001), IMF Working Paper, out of list of 25 countries. Banks and interest rate risk – p. 17/42

Method 1: Casual perusal of “gaps”

Banks and interest rate risk – p. 18/42

Look at them: Note: these are cashflows. (Rs. crore) Bucket

Assets Liabilities

Zero 0-1mth 1-3mth 3-6mth 6-12mth 1-3yrs 3-5yrs > 5yrs

12409 41659 18382 21927 87411 43282 31882 80285

34262 8053 5113 7483 15421 174229 55414 9944 Banks and interest rate risk – p. 19/42

Does not get the job done Is SBI carrying a significant risk? What would happen if interest rates went up by 320 bps? Does SBI have enough equity capital to absorb this? A casual perusal of gaps does not convey the materiality of mismatches (if any).

Banks and interest rate risk – p. 20/42

Method 2: Measure the NPV impact of a shock

Banks and interest rate risk – p. 21/42

NPV impact of +320 bps rise in the spot yield curve Shock

∆A

∆L

∆E

∆E E

∆E A

(Rs. crore) 200 320

-11,126 -9,833 -1,294 -8.50 -0.37 -17,079 -15,375 -1,704 -11.19 -0.49

By current rules, in a +320 bps shock, SBI would be forced to recognise -17,079 crore of losses on assets, but no MTM happens on liabilities.

Banks and interest rate risk – p. 22/42

Method 3: Duration

Banks and interest rate risk – p. 23/42

Fisher-Weil duration The sensitivity of PV to a parallel shift of λ:

P (λ) =

N X

ci e−(ri +λ)ti

i=0

∂P (λ) = − ∂λ

N X

t i ci e

−ri ti

i=0

1 ∂P (λ) = −DFW P (0) ∂λ where DFW = (

P

ti ci e−ri ti )/PV. Banks and interest rate risk – p. 24/42

Duration is a first order taylor approximation For small shocks λ it gives a reasonable prediction of what will happen to PV.

Banks and interest rate risk – p. 25/42

A bad idea for us 60 Exact Duration-based 50

Impact upon SBI (billion rupees)

40 30 20 10 0 -10 -20 -30 -40 -400

-300

-200

-100

0

100

200

Banks and interest rate400 risk – p. 26/42 300

Method 4: The stock market

Banks and interest rate risk – p. 27/42

The standard ‘market model’ (rj − rf ) = α + β1 (rM − rf ) +  Note: Everything here is returns, not rates.

Banks and interest rate risk – p. 28/42

Re-express the long interest rate as returns On the ZCYC, the long rate goes up from r1 on day 1 to r2 on day 2. The log returns on the bond, where the bond price goes from p1 to p2 is: log(p2 /p1 ) = −T (log(1 + r2 ) − log(1 + r1 ))

Banks and interest rate risk – p. 29/42

Augmented market model (rj − rf ) = α + β1 (rM − rf ) + β2 (rL − rf ) + 

Banks and interest rate risk – p. 30/42

SBI α β1 β2 R2 T

0.108 (0.218) 0.8369 (6.402) 0.8359 (2.316) 0.3732 104

Speculative position on core business of SBI: long SBI futures, short Nifty futures, short 10-year bond futures.

Banks and interest rate risk – p. 31/42

Results

Banks and interest rate risk – p. 32/42

Roughly 10 out of 44 banks have exposure BIS says that we should worry about banks where over 25% of equity capital would be gained/lost in the +320 bps move. Our results use accounting data for year ended 31/3/2002. This holds for 33 of 42 banks in our sample. SBI and ICICI are not in this group. 7 have ’reverse’ exposures, 26 have ’normal’ exposures.

Banks and interest rate risk – p. 33/42

Policy issues

Banks and interest rate risk – p. 34/42

Poor disclosure RBI rules require valuation at min(MTM, purchase price). “Hidden reserves”. Source of fog and confusion. Banks may not hedge a security at book value of Rs.110 and market value of Rs.120.

Banks and interest rate risk – p. 35/42

Disclosure of cashflows Fixed income analytics starts from cashflows. It shouldn’t be a struggle to get to cashflows.

Banks and interest rate risk – p. 36/42

Frequency Annual disclosure is highly unsatisfactory. We should have daily MTM and daily disclosure.

Banks and interest rate risk – p. 37/42

IFR Hierarchy: Natural hedges, Derivatives, Equity capital. IFR ignores the larger context and assumes the securities portfolio is the only thing. IFR thinks all banks are alike. IFR is to be built up in five years.

Banks and interest rate risk – p. 38/42

Better tools for supervision Do such computations to isolate weak banks. Use stock market coefficients to isolate weak banks.

Banks and interest rate risk – p. 39/42

IRD market We need an interest rate derivatives market. Modern architecture: Bank owns the customer, lays off the interest rate risk.

Banks and interest rate risk – p. 40/42

Metadata

Banks and interest rate risk – p. 41/42

Also see Web page on Indian fixed income: http://www.mayin.org/~ajayshah/FIXEDINCOME/index.html

Banks and interest rate risk – p. 42/42