afea 1

1
IAME 2006 CONFERENCE MELBOURNE, 12-14JULY 20006
Market Interactions and Volatility Spillover
Effects Between Shipping, Oil and Stock Markets
Andreas MERIKAS Professor of Finance Shipping
Studies Dpt., University of Piraeus Theodore
SYRIOPOULOS Assistant Professor of
Finance Shipping, Trade and Transport Dpt.,
University of the Aegean Efthimios ROUMPIS Ph.D.
Candidate
2
Contents

• Abstract
• Introduction
• Literature Review
• Data and Descriptive Statistics
• Methodology
• Empirical Results
• Conclusions

3
ABSTRACT
This study analyses information spillover effects
and shock transmission between shipping, oil and
stock markets. A VAR(1)-bivariate BEKK GARCH(1,1)
model is employed to estimate dynamic conditional
volatility and correlations between these
markets. The empirical evidence supports market
interactions and volatility spillover effects
between the shipping, oil and stock markets.
Different shipping market segments exhibit
varying degrees of dynamic volatility response
and lead-lag behavior with the other markets.
This exercise is considered to be useful, as the
empirical findings have implications for
efficient corporate decisions, risk management
and hedging strategies.
4
Introduction

• Major sources of risk in shipping industry
• - Freight rates volatility, bunker prices,
  vessel values, Interest rates
• Any shocks in these risk factors can have a
  profound impact on shipping companies
  operational cost, cash-flow surplus,
  profitability and market value
• At first, we investigate the robustness of any
  dynamic relationships, interdependencies and
  volatility spillover effects between shipping and
  oil markets.
• Also, we include major international stock
  markets in our study, because we expect that
• - Business cycle swings and economic growth
  fluctuations are affected by shocks in the oil
  markets, and appear to exert a similar impact on
  shipping and stock markets
• - The role of stock markets as a
  market-driven corporate valuation mechanism has
  been upgraded in recent years
• This paper contributes to the financial
  literature in a number of ways
• - It investigates the dynamic pattern of
  volatility in the shipping, oil and stock markets
  
• - The attempt to jointly model dynamic
  volatility spillover effects between these
  markets is an innovative contribution that has
  not been undertaken previously
• - The empirical methodology is founded on
  recent developments in the field of multivariate
  generalized conditional heteroscedasticity
  (MGARCH) models (Alexander, 2001).

5
Literature Review (1)

• A body of recent empirical research is based
  on the MGARCH models in order to investigate
  volatility interactions, spillover effects
  and shock transmission between underlying markets
  and/or instruments (e.g. Fleming et al., 1998
  Kearney and Patton, 2000 Ganon and Yeung, 2004)
• Volatility in freight rates, vessel size
  class, bunker oil prices, operational flexibility
  and interest and exchange rates have been
  identified as major risk factors in shipping
  industry (e.g. Kavussanos, 1997, 2003 Glen and
  Martin, 1998 Chen and Wang, 2004 Syriopoulos
  and Roumpis, 2006 inter alia).
• A thin body of research has focused on
  volatility dynamics and spillover effects between
  spot and forward shipping markets as well as
  bunker oil and shipping markets (e.g. Kavussanos
  and Nomikos,2003, Alizadeh and Nomikos, 2004a,
  2004b)
• Futures prices tend to discover new
  information more rapidly than spot prices in the
  freight markets (Kavussanos and Nomikos,2003)
• Alizadeh and Nomikos (2004a) study the dynamic
  relationship between oil futures and spot markets
  and tanker freight rates across two major tanker
  routes. They find no evidence to support the
  existence of a relationship between tanker
  freight rates and physical-futures differentials
  in the crude oil market

6
Literature Review (2)

• A number of studies investigate the dynamic
  linkages and information spillover effects
  between oil price volatility and financial
  markets (Malik and Hammoudeh, 2006, Ciner, 2002,
  Huang et al.,1996)
• Malik and Hammoudeh (2006) used a bivariate
  GARCH model to analyze volatility and shock
  transmission among US equity, global crude oil
  markets and equity markets of Saudi Arabia,
  Kuwait, and Bahrain. The results show significant
  transmission effects among second moments
• Ciner (2002) adopts nonlinear causality tests
  to examine the dynamic linkages between oil
  prices and the stock market. He suggests that oil
  price shocks affect stock index returns, which is
  consistent with the documented influence of oil
  on economic output

7
Data and Descriptive Statistics (1)

• Our dataset consist of weekly frequencies and
  covers the period running from January 5th 1990
  to February 25th 2005
• Worldscale freight rates for the
  transportation of 250.000 tons (VLCC ship) for
  the following tanker shipping markets
• - Middle East Gulf to Japan (MEGJ), from Ras
  Tanura to Chiba
• - West Africa to US Atlantic Coast (WAUS),
  from Off Shore Bonny to Philadelphia
• - North Sea to Continent (NOSC), from Sullom
  Voe to Wilhelmshaven
• We include spot prices on the two primarily
  crude oil markets
• - Brent Crude oil
• - West Texas Intermediate (WTI)
• For the market portfolio, we include the
  Standard Poors-500 (SP500) composite index

8
Data and Descriptive Statistics (2)
9
Methodology (1)
We employ the BEKK model for the parameterization
of the conditional variance-covariance matrix
(Baba et al., 1987 Engle and Kroner, 1995)
Ht C?C A?et-1 e?t-1 A B?Ht-1B
where C is a n x n upper triangular matrix, A and
B are n x n coefficient matrices. the elements
aij of matrix A measure the degree of innovation
from market i to market j and the elements ßij of
matrix B indicate the persistence in conditional
volatility between market i and market j. This
can be expressed for the bivariate BEKK model as
Under the assumption of conditional normality,
the model can be estimated by maximizing of the
following log-likelihood function
10
Methodology (2)
The conditional mean return, rt, is modeled in a
vector autoregressive (VAR) framework
ei,t?It-1 ? N(0,Ht)
where, ?(L), d(L), f(L) and ?(L) denote lag
polynomials of order p the vector of error
terms, eit, represents the unexpected excess
return on market i (i 1, 2) and is assumed to
be normally distributed (denoted as N) with the
conditional variance-covariance matrix Ht and,
It-1 is the information set at time t-1.
11
Empirical Results (1)

• The full BEKK GARCH(1,1) model appears to
  perform reasonably well and most coefficients
  appear to be significant
• The bivariate VAR(1) model of shipping and oil
  market returns indicates some interaction effects
  
• The variance-covariance structure of the
  shipping and oil markets reflects information
  shocks and volatility spillover effects. In the
  conditional tanker variance equation, h11,t, the
  coefficients a11 and a21 are significant. This
  implies that both lagged squared freight rate
  shocks and lagged squared oil shocks affect the
  conditional freight rate variance
• Significant volatility spillover effects are
  predominantly found between the VLCC market and
  the WTI market and past volatility in one market
  appears to have a feedback impact to the
  volatility of the other market
• In the case of smaller vessel classes of the
  Suezmax and Aframax market segments, volatility
  spillover effects are mainly detected with the
  Brent market
• The VLCC, Suezmax and Aframax segments, both,
  lagged squared freight rate shocks and past
  variance affect current tanker freight rate
  variance

12
Empirical Results (2)

• The examination of interactions between
  shipping and stock markets reveals dynamic
  volatility effects mainly between stock market
  shocks and the VLCC and Suezmax segments
• The stock market and oil market are found to
  exert spillover effects in both the mean and
  variance equations
• Increase in Brent oil market returns has an
  adverse impact on stock market returns (SP500
  index), probably due to expectations for a
  potential growth slowdown
• In the variance equation statistically
  significant interactions are found mainly between
  the stock market and the Brent oil market
  cross-effects (ß12, ß21) point to volatility
  increase in one market due to spillover effects
  from the other markets

13
Empirical Results (3)
14
Empirical Results (4)
15
Empirical Results (5)
16
Conclusions

• This study has attempted to offer an
  innovative perspective on explaining dynamic
  market interactions between the shipping, oil and
  stock markets
• The first (mean) and second (variance) moments
  of the underlying markets were jointly modelled
  in a VAR(1) - bivariate full BEKK GARCH(1,1)
  approach
• The empirical results indicate significant
  information spillover effects of varying degrees
  between the markets of interest
• The past innovations exert an impact on
  current volatility levels, and the past values of
  the own conditional variance also matter
• The VLCC segment was found to be sensitive to
  WTI oil price changes, whereas Suezmax and
  Aframax markets to Brent oil price volatility
• The tanker market was found to be relatively
  sensitive to stock market volatility as robust
  economic activity is a key driver for both
  markets
• Dynamic volatility effects were mainly
  revealed between stock market shocks and the VLCC
  and Suezmax segments
• The stock and oil markets were found to
  exhibit spillover effects in both the mean and
  variance equations. An upward movement in (Brent)
  oil market has an adverse impact on stock market
  (SP500) returns, probably due to expectations of
  potential growth slowdown
• The current empirical findings could be
  further expanded by bringing the respective
  forward markets into the discussion.