Determinants of Nikkei futures mispricing in international markets: Dividend clustering, currency risk, and transaction costs

• DOI: http://doi.org/10.1002/fut.22038
• Author: Christopher J. Green,Kavita Sirichand,Jieye Qin
• Published date: 01 October 2019
• Date: 01 October 2019

J Futures Markets. 2019;39:1269–1300. wileyonlinelibrary.com/journal/fut © 2019 Wiley Periodicals, Inc.

|

1269

Received: 10 January 2018

|

Accepted: 1 June 2019

DOI: 10.1002/fut.22038

RESEARCH ARTICLE

Determinants of Nikkei futures mispricing in

international markets: Dividend clustering, currency

risk, and transaction costs

Jieye Qin

1

|

Christopher J. Green

2

|

Kavita Sirichand

2

1

Department of Economics, School of

Economics and Management, Anhui

Jianzhu University, Hefei, Anhui, China

2

School of Business and Economics,

Loughborough University,

Loughborough, Leicestershire, UK

Correspondence

Jieye Qin, School of Economics and

Management, Anhui Jianzhu University,

Ziyun Road No. 292, 230601 Hefei, Anhui,

China.

Email: j.qin@ahjzu.edu.cn

Funding information

Anhui Jianzhu University, Grant/Award

Number: 2018QD03; Department of

Education of Anhui Province, Grant/

Award Number: KJ2019A0764

Abstract

This paper develops a comprehensive modified cost‐of‐carry model to study the

mispricing of Nikkei 225 index futures contracts traded in Osaka, Singapore,

and Chicago based on a new 19‐year data set. Using this improved model, we

find that dividend clustering, currency risk, and transaction costs all play an

essential role in the estimation of Nikkei mispricing. An exponential smooth

transition autoregressive‐GARCH model is used to describe the international

dynamics of Nikkei mispricing. The results indicate that generally mean

reversion in mispricing and limits to arbitrage are driven more by transaction

costs than by heterogeneous investors.

KEYWORDS

ESTAR‐GARCH model, index arbitrage, mispricing, Nikkei 225 futures

JEL CLASSIFICATION

G13; G14; G15

1

|

INTRODUCTION

Many securities are traded simultaneously on different international markets. Differences among markets due inter alia

to currency of trading, transaction costs, opening hours, supply of liquidity, and regulation, can all create different rates

of price discovery in separate markets for the same security. This implies that mispricing, arbitrage opportunities, and

price adjustments may vary among the different markets. This is a particularly important issue for international

investors who have the option to trade on these different markets.

Nikkei 225 index futures contracts are a vital investment tool in worldwide trading activities. Nikkei shares are

traded in Tokyo, but the corresponding index futures contracts are traded in Osaka, Singapore, and Chicago. The triple‐

listed nature of Nikkei futures and institutional differences among the three markets substantially affect the economic

significance of Nikkei mispricing, which is of particular concern for global investors who seek to exploit temporary

price deviations around the world. As such, a study accurately estimating the profitability and behavior of Nikkei

mispricing is of crucial empirical and practical importance.

Some previous papers examined mispricing and arbitrage between Nikkei spot and futures markets. Brenner,

Subrahmanyam, and Uno (1989a, 1989b, 1990) and Lim (1992) studied Nikkei mispricing in Singapore; Iihara,

Kato, and Tokunaga (1996) estimated lead–lag relations between the Nikkei index and Osaka futures; Frino and

West (2003) and Covrig, Ding, and Low (2004) compared the contributions of Osaka and Singapore to price

discovery in the spot Nikkei; Booth, Lee, and Tse (1996) and Kao, Ho, and Fung (2015) studied information

transmission among Nikkei futures; Tsuji (2007) examined the basis dynamics of Osaka futures; Ubukata (2018)

studied a conditional hedging strategy to minimize downside risks for Osaka futures. However, there have been

no studies which have specifically compared Nikkei mispricing in all three futures markets. Most importantly,

previous studies did not fully consider issues that have a material impact on estimates of Nikkei mispricing,

information transmission, and market efficiency. These include the clustering of Japanese dividends in the year,

the dollar denomination of Chicago Nikkei futures, the impact of transaction costs, and differences in market

time zones. Board and Sutcliffe (1996) explained the principles of pricing and index arbitrage of the three Nikkei

futures, but did not provide empirical evidence. Past empirical papers mostly predate the 2008 global financial

crisis, which may have affected arbitrage possibilities, and they mostly used limited‐time high‐frequency

datasets.

In this paper, we provide a new and comprehensive analysis of the behavior and determinants of mispricing in

the three main Nikkei futures markets. We contribute to the literature in three ways. First, we estimate and

characterize Nikkei futures mispricing in Osaka, Singapore, and Chicago using a new daily data set from 1996

through 2014, providing evidence on the magnitude, sign, persistence, and path dependence of mispricing and

the bilateral relationship of mispricing with time to expiration and stock volatility. Second, in view of the

complexity of the Nikkei futures, we develop a comprehensive modified cost‐of‐carry model, which allows for

essential factors that have not previously been considered as a whole. These are (a) the dividend clustering

specific to Japanese companies; (b) the currency risk of the Chicago Nikkei futures, given that the Chicago

futures contract is a quanto, that is, it is denominated, traded, and settled in a currency (US dollars) that is

different from the underlying stock index (yen); (c) the different trading hours of the Nikkei markets,

particularly in the context of Chicago’s Globex trading system; and (d) the transaction costs of brokers and

institutional investors. Hence the present model significantly improves on earlier models such as that of Board

and Sutcliffe (1996) used to price Nikkei futures. Third, the exponential smooth transition autoregressive

(ESTAR)‐GARCH model is used to examine the effects of transaction costs and heterogeneous investors on the

nonlinear adjustments of Nikkei mispricing. ESTAR‐GARCH is arguably more satisfactory in describing these

nonlinear adjustments than are threshold cointegration models (e.g., Tao & Green, 2013), as the former avoids

the imposition of potentially arbitrary discontinuities on the aggregate arbitrage process. Recent studies using

smooth transition include the Dow Jones (Tse, 2001), FTSE 100 (McMillan & Speight, 2006), S&P 500 (Taylor,

2007), Hang Seng (Fung & Yu, 2007), and emerging index futures (Sila Alan, Karagozoglu, & Korkmaz, 2016).

However, to our knowledge, the ESTAR‐GARCH model has not been applied to any of the three Nikkei markets,

despite the fact that Nikkei futures contracts are one of the most actively traded derivatives in the world.

The rest of this paper is structured as follows. Section 2 describes the key institutional differences among the Nikkei

markets and data. Section 3 modifies the standard cost‐of‐carry model for Nikkei futures, allowing for dividend

clustering, currency risk, different trading hours, and transaction costs. Section 4 reports in detail on our mispricing

measures and their behavior across the markets and over time in relation to time to expiration, stock volatility, and path

dependence. Section 5 studies Nikkei mispricing dynamics and index arbitrage by the ESTAR‐GARCH model. Section 6

concludes the paper.

2

|

INSTITUTIONAL DIFFERENCES AND DATA

2.1

|

The Nikkei 225 index and index futures markets

The Nikkei 225 consists of 225 common stocks listed in the First Section of the Tokyo Stock Exchange (TSE).

1

Nikkei

225 index futures contracts are traded on three different markets: the Osaka Exchange (OSE), the Singapore Exchange

(SGX), and the Chicago Mercantile Exchange (CME). See Table 1 for details. There are three important differences

among these contracts. First, the three futures exchanges are located in different time zones and the futures contracts

are traded within different time periods. Second, trading and settlement on the CME are in US dollars while trading and

settlement on the other Nikkei markets involve Japanese yen. Third, OSE futures have had a computerized trading

system since inception. SGX futures initially used open outcry, but shifted from open outcry to electronic trading on

1

In June 2014, the market capitalization of the constituent stocks in the Nikkei was $4.49 trillion, or 64% of all stocks in the First Section of the TSE (data from Nikkei Inc.).

1270

|

QIN ET AL.

November 01, 2004. Open outcry and electronic trading were both used for CME Nikkei futures during the sample

period of this paper.

2.2

|

Data description

Daily closing prices of the Nikkei 225 index and settlement prices of the corresponding futures are collected from

Datastream, OSE, SGX, and CME, for June 20, 1996–December 31, 2014 (OSE and SGX) and January 01,

1997–December 31, 2014 (CME). The starting dates are the earliest available with consistent dividend data (OSE and

SGX), or sufficient trading volume (CME). We use futures contracts that mature in the quarterly months (March, June,

September, and December) in each market. Continuous, synthetic futures price series are compiled using the prices of

the nearest contract and rolling over to the next nearest contract at the start of the expiration month.

2

Nontrading days

are excluded and futures and spot prices are matched on each day that all markets are trading. Higher‐frequency data

have been used in some studies (e.g., Iihara et al., 1996; Lim, 1992), but we prefer daily data to study mispricing. Daily

marking to market implies that gains or losses on every futures contract that is open overnight must be realized at the

end of each trading day with reference to the daily settlement price. Thus, settlement prices will reflect the outcome of

TABLE 1 Nikkei 225 futures contracts

OSE SGX CME

Underlying asset Nikkei 225 Nikkei 225 Nikkei 225

Launch date September 03, 1988 September 03, 1986 September 25, 1990

Contract size Index× ¥1,000 Index × ¥500 Index× $5

Tick size 10 index points (¥10,000) 5 index points (¥2,500); 5 index points ($25)

1 index point (¥500) for strategy

trades

Contract months Nearest 3 for March and September;

nearest 10 for June and December

Nearest 6 for serial months; nearest

20 for March, June, September,

December

March, June, September, December

Trading hours 9.00–15.15, 16.30–3.00 (JST) 7.45–14.25, 15.15–2.00 (SGT) Open outcry: 8.00–15.15 (CST)

Electronic trading: 17.00–16.15

(CST)

Trading

mechanism

Electronic trading Open outcry (before November 1,

2004)

Open outcry and electronic trading

a

Electronic trading (after November

1, 2004)

Last trading day The business day before the

settlement date

The business day before the

settlement date

The business day before the

settlement date

Final settlement

day

Second Friday of the contract month Second Friday of the contract

month

Second Friday of the contract

month

Final settlement

price

Special quotation based on the total

opening prices of each constituent of

Nikkei 225 index on the settlement

date

Special quotation based on the total

opening prices of each constituent

of Nikkei 225 index on the

settlement date

Special quotation based on the total

opening prices of each constituent

of Nikkei 225 index on the

settlement date

Settlement

procedure

Cash settlement Cash settlement Cash settlement

Note: It presents details of Nikkei 225 futures contracts traded on the OSE, SGX, and CME as of December 31, 2014. JST is Japan Standard Time. SGT is

Singapore Time. CST is Central Standard Time. Data are from the OSE, SGX and CME.

Abbreviation: CME, Chicago Mercantile Exchange; OSE, Osaka Exchange; SGX, Singapore Exchange.

a

The CME closed the open outcry system for Nikkei contracts on June 19, 2015.

2

When SGX changed from open outcry to the electronic trading system (ETS), investors were given time to adapt to the ETS in overnight sessions and there was a period when both systems were

available for trading. The ETS started to rival open outcry in volume from about November 01, 2004, and shortly after, ETS became the only trading mechanism. SGX futures prices are compiled using

settlement prices on the floor before November 01, 2004, and those on the ETS afterwards. A Quandt‐Andrews unknown breakpoint test suggests that there were no discernible breaks in the compiled

price series around the changeover date.

QIN ET AL.

|

1271