What macro-innovation risks really are priced in Japan?

Abstract

This article examines whether specified macroeconomic and macro-financial market variables innovations carry risks that are rewarded in the Japanese stock market by a restricted nonlinear multivariate regression model. We find that not all macroeconomic variables priced in the United States are priced in Japan. In addition, we also find, for the first time, that two additional macro-factors, namely the innovations in money supply and in gold and foreign exchange reserves, are strongly priced in Japan. Furthermore, neither market portfolio nor oil price variables are priced separately, as with the evidence from the United States. However, differently from previous US results, we find that innovations in aggregate real per capita consumption are weakly priced in Japan.

Acknowledgements

We are particularly grateful to Professor Mark Taylor (the Editor) for providing constructive comments and advice on improving this article. We also thank the Japan Society for the Promotion of Science and the Zengin Foundation for Studies on Economics and Finance for their generous financial assistance for this research project.

Notes

¹ Chen et al . (1986) used the term ‘innovations’ to mean (unanticipated) changes or growth rates of state variables. We also use the term ‘innovations’ in this article to have that meaning.

² There are numerous studies that follow on from Chen et al . (1986). For more recent research, see Avramov (2002, 2004), Chelley-Steeley and Siganos (2004), Conrad et al . (2003), Ewing et al . (2003), Goyal and Santa-Clara (2003), Jagannathan and Wang (2002), Lioui and Poncet (2003), Maringer (2004), Merville et al . (2001), Payne (2003), Priestley and Odegaard (2004), Poitras (2004), Resnick and Shoesmith (2002), Scruggs and Glabadanidis (2003), Siddique (2003), Smith and Wickens (2002) and Vassalou and Xing (2004), amongst others.

³ Studies of the Japanese stock market that focus on macroeconomic factors, including those of Doukas et al . (1999), Hamao (1988), He and Ng (1998) and Kaneko and Lee (1995), have not exhaustively characterized or tested many sets of influential macro-variables.

⁴ Aquino (2005) analysed exchange rate risk in the Philippine stock market. Chan (1997) implemented multivariate testing of the CAPM in the context of the Hong Kong stock market. Cheung and Lai (1999) investigated the effects of macroeconomic factors on co-movements of the French, German and Italian stock markets. Dickinson (2000) investigated the relationship between stock market integration and macroeconomic fundamentals in the context of the United States, UK, French and German stock markets. Groenewold et al . (1997) investigated the relationship between the stock market and inflation in Australia. Hansson and Hordahl (1998) analysed the relationship between expected returns and time-varying risks in the context of the Swedish stock market. Josev et al . (2001) tested two-factor APT in the context of the Australian equity market. Kavussanos et al . (2002) investigated global sources of risk in 38 international industries by using macro-factors. Liljeblom and Stenius (1997) studied the relationship between macroeconomic volatility and stock market volatility in Finland. Morelli (2003) used data on the UK market to test the CAPM by using an autoregressive conditional heteroscedasticity (ARCH) model. Ong and Izan (1999) investigated the relationship between exchange rates and stock-market indices in Australia, the UK and the United States. Shively (2004) focused on the time-varying risk components in the US equity market. Siklos and Kwok (1999) examined the relationship between stock returns and inflation in the United States. In the context of the Norwegian equity market, Solibakke (2002) tested the univariate conditional CAPM. Staikouras (2005) focused on interest rate risk in the UK equity market. Wirjanto (1997) tested the intertemporal asset pricing model by using data on the Canadian stock market.

⁵ Aylward and Glen (2000) examined the predictive power of asset prices on economic activity in 23 countries including Japan. Benzion et al . (2005) investigated the relationship between exchange rates and stock prices in Japan. Chou and Lin (2002) investigated the unconditional mean-variance efficiency of the Morgan Stanley Capital International (MSCI) world index for 16 OECD countries and Hong Kong. Chou et al . (2002) focused on exchange rate risk in the context of Pacific Basin equity markets. Ostermark (1998) and Ostermark and Aaltonen (1999) examined the relation between the Japanese stock market and Finnish financial markets. Few comprehensive studies of the Japanese equity market have appeared in Applied Financial Economics.

⁶ Subsequent variants and interpretations of APT were introduced by Chen and Ingersoll (1983), Conner (1984), Dybvig (1983), Grinblatt and Titman (1983), Grinblatt (1985), Huberman (1982), Ingersoll (1984), Ohlson and Garman (1980), Shanken (1982) and others. Key issues for these asset-pricing models include the suitability of different estimation techniques (Gibbons, 1982; Shanken, 1985) and the economic interpretation of empirical results.

⁷ While after World War II, an active equity market developed in Japan, the bond market has remained almost undeveloped. Corporations heavily depended on bank loans rather than bond issues, thus the corporate bond market, in particular, remained undeveloped. Furthermore, no long-term government bonds were issued since a balanced budget was strictly sustained for preventing recurrence of the post-war hyperinflation. A long-term government bond was first issued in 1966 upon the amendment of the law requiring a balanced budget and in 1975, massive offerings started. Not until 1977 were financial institutions, the main purchasers of the government bonds, permitted to sell bond holdings in a secondary market. As a result, Japanese financial markets are somewhat different from those in the United States.

⁸ The 33 industries are: (1) fishery, agriculture and forestry, (2) mining, (3) construction, (4) foods, (5) textiles and apparels, (6) pulp and article, (7) chemicals, (8) pharmaceuticals, (9) oil and coal products, (10) rubber products, (11) glass and ceramics products, (12) iron and steel, (13) non ferrous metals, (14) metal products, (15) machinery, (16) electric appliances, (17) transportation equipment, (18) precision instruments, (19) other products, (20) electric power and gas, (21) land transportation, (22) marine transportation, (23) air transportation, (24) warehousing and harbour transportation services, (25) information and communication, (26) wholesale trade, (27) retail trade, (28) banks, (29) securities and commodity futures, (30) insurance, (31) other financing business, (32) real estate, (33) services.

⁹ The series for E [I(t) |t − 1] (and E [I(t + 1) |t]) were constructed as follows. First, according to Fisher (1965), GR(t − 1) can be decomposed into two series, one for the expected real return for month t, E [ρ(t) |t − 1] and one for the expected inflation rate, E [I(t) |t − 1], as follows: GR(t − 1) = E [ρ(t) |t − 1] + E [I(t) |t − 1]. Then, as in Hamao (1988), we assume the following relationship, based on a first-order moving average process: [GR(t − 1) − I(t)] − [GR(t − 2) − I(t − 1)] = u(t) − θ · u(t − 1). Given an estimate of θ, obtained by using the standard Box and Jenkins (1976) methodology and given that GR(t − 1) − I(t) = E [ρ(t) |t − 1] + u(t), we obtain . Then, by using this series for E [ρ(t) |t − 1], we can obtain the series E [I(t) |t − 1] (and E [I(t + 1) |t]) from the equation, GR(t − 1) = E [ρ(t) |t − 1] + E [I(t) |t − 1]. (For more details, see Hamao (1988), p. 50.)

¹⁰ Pricing Equation 2 is consistent with the theoretical asset-pricing model of Sharpe (1977), Breeden (1979), Cox et al . (1985), Long (1974) and Merton (1973). For further discussions as to Equation 2, see, for example, Dybvig and Ross (1985), Shanken (1982, 1985) and Grinblatt and Titman (1983).

¹¹ Early studies of APT used factor analysis: see, for example, Roll and Ross (1980), Brown and Weinstein (1983), Chen (1983), Cho et al . (1984), Kryzanowski and To (1983), Dhrymes et al . (1984), Reinganum (1981), Black et al . (1972), Fama and MacBeth (1973), Miller and Scholes (1972). However, the factor scores and the associated risk prices do not have any clear economic meaning. On the statistical difficulties of this factor analysis in the application to asset pricing, see Anderson (1984) and Conway and Reinganum (1988).

¹² WPI is tested by excluding DEI and UI by taking the high correlation among them into consideration; however, it shows no statistically significant price of risk.

¹³ We consider that to display clearly the evidence as to nonpriced variables is also very important to deepen readers’ knowledge of the Japanese equity market.