euro-philippine network in banking and finance markov-switching market risk carlos bautista daniel...

Euro-Philippine Network in Banking and Finance

Markov-Switching Market Risk

Carlos BautistaDaniel Goyeau

Joel Yu

27 August 2007International Conference on the

Safety and Efficiency of the Financial Sector

Euro-Philippine Network in Banking and Finance

Background

• Market Risk is measured by Beta, cov(r i,rm)/var(rm)

• Traditional Estimation of Beta: Market ModelRi = a + beta (Rm-Rf)

• In the CAPM model, beta explains the cross section of expected returns on securities

Euro-Philippine Network in Banking and Finance

Literature

• Time-varying beta– Empirical evidence– Nature of change/causes of change

• Beta as an explanatory variable and predictor of events

• Difficulty with existing methods

Euro-Philippine Network in Banking and Finance

Model and Data

• Markov-switching market risk– Two states– Fixed transition probabilities– Markov-switching process

• Data– Datastream price data of 22 stocks of 17 firms

in the PSEi– 91-day T-bill rate

Euro-Philippine Network in Banking and Finance

Results

• There is a strong statistical support for non-linearity in the market model

• In most cases, there is a fairly high frequency of changes in the market risk of common stocks in the Philippines from low beta regime to high beta regimes.

• There are few cases where there is a transient change in market risk, albeit significantly longer in duration.

• There are cases where there seems to be a permanent change in the market risk of the firm.

Euro-Philippine Network in Banking and Finance

Results: Case 1

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

Jolibee Food Corp. - smoothed probability of a low beta regime

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

Metro Pacific - smoothed probability of a low beta regime

Jollibee Foods CorpLow Beta: 0.43High Beta: 1.19

Metro PacificLow Beta: 1.23High Beta: 2.35

Euro-Philippine Network in Banking and Finance

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

Meralco B - smoothed probability of a low beta regime

Results: Case 2

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

ICTSI - smoothed probability of a low beta regime

ICTSILow Beta: 0.80High Beta: 2.16

MERALCOLow Beta: 1.01High Beta: 2.28

Euro-Philippine Network in Banking and Finance

Results: Case 3

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

Bank of the Phil. Islands - smoothed probability of a low beta regime

0.0

0.5

1.0

94 95 96 97 98 99 00 01 02 03 04 05

San Miguel B - smoothed probability of a low beta regime

SAN MIGUELLow Beta: 0.54High Beta: 0.89

BPILow Beta: 0.30High Beta: 1.27

Euro-Philippine Network in Banking and Finance

Conclusions

• This study demonstrates that time variation in the CAPM can be adequately modeled through Markov switching techniques.

• Results show that the technique is a productive alternative in evaluating the market risk of firms in the Philippines.

• Shifts in the market risk seem to be related to market developments which can have a permanent or transient change in the volatilities of security returns relative to that of the market.