Alexey Ponomarenko, Alexey Porshakov, Elena Vasilieva

Bank of RussiaResearch and Information Department

MadridSeptember, 2009

Overnight interest rates and liquidity of the Russian money market: the

impact of financial turmoil

2

Motivation

Model set-up

Description of variables

Results

Conclusion

Structure of presentation

3

Motivation

* MIACR – actual average weighted interest rate on overnight interbank ruble credit on the Moscow market.

BoR’s interest rates corridor and overnight market rate (MIACR*), p.p.

0

5

10

15

20

25

05.0

1.04

05.0

4.04

05.0

7.04

05.1

0.04

05.0

1.05

05.0

4.05

05.0

7.05

05.1

0.05

05.0

1.06

05.0

4.06

05.0

7.06

05.1

0.06

05.0

1.07

05.0

4.07

05.0

7.07

05.1

0.07

05.0

1.08

05.0

4.08

05.0

7.08

05.1

0.08

05.0

1.09

Tom-next deposit rate Minimum REPO auction rate (1 day)

MIACR* Overnight credit rate

4

Motivation

• Analyze the process of interest rate determination on the money market and identify the main drivers behind it

• Estimate the magnitude of the effects stemming from different factors to facilitate the process of interest rate steering

• Assess the impact of the financial turmoil

5

Model set-up: related literature

The theory of money market and liquidity managementBindseil 2004

Valimaki 1998, 2001

Empirical modeling of liquidity effectWurtz 2003

Bindseil, Seitz 2001Jurgilas 2006

Macro-analysis of excess liquidityAgenor et al. 2004Saxegaard 2006

6

Model set-up• Econometric Techniques: non-linear EGARCH model

• Dependent Variable: deviation of actual overnight market interest rate from the middle of the BoR’sinterest rate corridor

• Explanatory Variables:I. LiquidityII. BoR’s interest ratesIII. Foreign interest ratesIV. Exchange rate expectationsV. Adaptive expectationsVI. Calendar effectVII. Time series components

• Sample period: January 2006 – January 2009

7

Liquidity

∑−

=

1

1i

t

i

CASF = + CAt-1* (T-t) - RR*T ) - Ln (AccERN)

СА – banks’ current accounts; RR – reserve requirements; AccERN –neutral level

of accumulated excess reserves (non-linear trend which forms the lower boundary

of the excess reserves fluctuations until August 2008); DEP –BoR’s overnight

deposits, REF –BoR’s overnight credit; T –end of RMP; t – current period

Simple forecast

Perfect forecast

PF = Ln (

Ln (

)(

)(

tT

REFDEPT

tiii

−

−∑= )

8

LiquidityActual and neutral level of accumulated excess reserves

(bn. rubles)

5000

10000

15000

20000

25000

30000

35000

10.0

1.20

06

10.0

3.20

06

10.0

5.20

06

10.0

7.20

06

10.0

9.20

06

10.1

1.20

06

10.0

1.20

07

10.0

3.20

07

10.0

5.20

07

10.0

7.20

07

10.0

9.20

07

10.1

1.20

07

10.0

1.20

08

10.0

3.20

08

10.0

5.20

08

10.0

7.20

08

10.0

9.20

08

10.1

1.20

08

10.0

1.20

09

Actual level Neutral level

9

LiquidityMIACR (deviation from the mid-point of the interest rates band, p.p.) and

accumulated excess reserves (deviation from the neutral level, %)

-10

-8

-6

-4

-2

0

2

4

10.0

1.20

06

10.0

3.20

06

10.0

5.20

06

10.0

7.20

06

10.0

9.20

06

10.1

1.20

06

10.0

1.20

07

10.0

3.20

07

10.0

5.20

07

10.0

7.20

07

10.0

9.20

07

10.1

1.20

07

10.0

1.20

08

10.0

3.20

08

10.0

5.20

08

10.0

7.20

08

10.0

9.20

08

10.1

1.20

08

10.0

1.20

09

MIA

CR

-40

10

60

110

160

210

Acc

umul

ated

exc

ess r

eser

ves

MIACR MIACR (moving average) Accumulated excess reserves Accumulated excess reserves (moving average)

10

Foreign Interest Rates and Exchange Rate Expectations

- Interest rate parity concept (under fixed exchange rate regime)

Composite foreign interest rate indicator:

0.8*USD money market interest rate + 0.2* EURO money market interest rate

- High level of dependence of Russian banks on the international money markets

Foreign interest rates variable:

Exchange rate expectations variable:

- Persistent expectations of ruble’s depreciation (during the period of gradual controlled depreciation conducted by BoR)

Exchange rate expectations indicator:

January 2006 – September 2008: adaptive expectations (the average rate of bi-currency basket appreciation over last 5 days)

October 2008 – November 2008: 0.5*adaptive expectation + 0.5*forward-looking expectations

December 2008 – January 2009: forward-looking expectations (based on 1-month forward rates)

11

Calendar EffectMonthly dynamics of overnight MIACR

Calendar Effect Modeling: dummies for the first day and last 5 days of the RMP

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1 2 3 4 5 5 4 3 2 1

day of RMP

MIA

CR

(rat

io to

mon

thly

ave

rage

)

dynamics during RMP average dynamics in 2006-2009

days before end of RMP

12

Other Variables

Adaptive expectations - the average of the spread on the last weekof the previous RMP

Dummy-variables used for the observations when MIACR wasfluctuating beyond the BoR’s interest rate band

Time series component – 1-day lag of the spread (takingaway those observations, which correspond to the last day of the previousRMP)

13

Results: Equation for the mean

** - significance at 5%-level..* - significance at 10%-level.

0.242ConstantAutonomous factor

0.230**1-day lag of actual spread (corrected for spillover effects)Time series component

0.246**Dummy variable (2 banking days until the end of the RMP)

0.205**Dummy variable (3 banking days until the end of the RMP)

0.303**Dummy variable (4 banking days until the end of the RMP)

0.345**Dummy variable (5 banking days until the end of the RMP)

0.247**Dummy variable (last banking day of the RMP)

-0.478**Dummy variable (first banking day of the RMP)

Calendar effect

0.004**Devaluation expectations (August 2008 – January 2009)Exchange rate expectations

0.025*Spread between short-term foreign interest rate and the middle of BoR’sinterest rate band (September 2008 – January 2009)

0.071**Spread between short-term foreign interest rate and the middle of the BoR’s interest rate band (January 2006 – September 2008)

Foreign interest rates

0.047**Average spread for the last 5 days of the previous RMPAdaptive expectations

-0.051**“Perfect” forecast of net deposits until the end of each RMP

-0.672**“Simple” forecast of excess reserves at the last week of the RMP

-0.401**“Simple” forecast of excess reserves before the last week of the RMP

Liquidity

ML Estimate VariableCategory

14

Results: Equation for the conditional variance

-0.571Constant

0.308**Absolute value of normalized disturbance (lag 1 day)

-0.247**Normalized disturbance (lag 2 days)

0.343**Normalized disturbance (lag 1 day)

0.721**GARCH-component – lag 1 day

0.298**Dummy variable (August 2008 – January 2009)

0.486**Dummy variable (last week of the RMP)

ML EstimateVariable

** - significance at 5%-level..

15

Results

Demand for liquidity during RMP

-4

-3

-2

-1

0

1

2

3

4

-181 -147 -114 -81 -47 -14 19 53 86 119 153

deviation of excess reserves from neutral level, %

effe

ct o

n th

e sp

read

, per

cent

age

poin

ts

last week of RMP before last week of RMP

liquidity shortageL*

excess liquidity

16

Results

Actual dynamics of the MIACR and the dynamic forecast from the EGARCH(percentage points)

0

5

10

15

20

25

30

10.0

1.20

06

10.0

3.20

06

10.0

5.20

06

10.0

7.20

06

10.0

9.20

06

10.1

1.20

06

10.0

1.20

07

10.0

3.20

07

10.0

5.20

07

10.0

7.20

07

10.0

9.20

07

10.1

1.20

07

10.0

1.20

08

10.0

3.20

08

10.0

5.20

08

10.0

7.20

08

10.0

9.20

08

10.1

1.20

08

10.0

1.20

09

actual MIACREGARCH dynamic forecast of MIACRBoR's overnight credit rateBoR's overnight deposit rate

17

Results

Monthly averages of absolute contributions of different factor categories to thespread (percentage points)

-6

-5

-4

-3

-2

-1

0

1

2

Jan-06

Mar-06

May-06

Jul-06

Sep-06

Nov-06

Jan-07

Mar-07

May-07

Jul-07

Sep-07

Nov-07

Jan-08

Mar-08

May-08

Jul-08

Sep-08

Nov-08

Jan-09

Liquidity Foreign interest ratesAdaptive expectations Time Series ComponentCalendar effect Exchange rate expectationsResiduals Actual spread

7,21

18

Results

Monthly averages of relative contributions of different factor categories to thespread (excluding time series component and residuals, %)

0%

20%

40%

60%

80%

100%Ja

n-06

Mar

-06

May

-06

Jul-0

6

Sep-

06

Nov

-06

Jan-

07

Mar

-07

May

-07

Jul-0

7

Sep-

07

Nov

-07

Jan-

08

Mar

-08

May

-08

Jul-0

8

Sep-

08

Nov

-08

Jan-

09

Exchange rate expectations Liquidity Foreign interest rates Calendar effect Adaptive expectations

19

The estimation of the model of the short-term interbank rate in Russiabasically confirmed the relative ineffectiveness of the country’s money market.Nevertheless, our model provides empirical backup for the martingale hypothesis as a crucial tool for analyzing the dynamics of the money market.

Over the last several years, abrupt changes of the liquidity conditions, whichwere to some extent driven by the managed floating exchange rate regime inRussia, provoked substantial fluctuations of interbank rates.

In the second half of 2008 general tightening of liquidity conditions occurred. In addition, tighter conditions of borrowing abroad and persistent depreciation expectations both contributed to sharp growth of the interbank rate.

Conclusion