Moon Phases,Mood and Stock Market Returns

Abstract

We employ recent data from 59 international emerging and mature stock markets to provide new evidence of a lunar cycle (full and new moon) effect on their stock market returns. Using a threshold generalised autoregressive conditional heteroscedasticity (TGARCH) model, we further examine the linkages between efficient-market theory, calendar-related effects and investors’ mood resulting from moon phases. The empirical results show significant full moon effects in six markets, and significant new moon effects in eight markets. In line with the theory, we report significant positive effect of new moon on stock market returns in five cases (UK, Switzerland, Bangladesh, Chile and Cyprus), while a negative effect of full moon is reported for the case of Jordan only. In addition, we find that lunar effects are strongly influenced by the calendar anomalies (Monday effect and January effect); several markets—mostly emerging markets—show evidence of full/new moon effects as well as Monday/January effects (Bangladesh, Brazil, Chile, Tunisia, Belgium, Cyprus). Further, we prove that the lunar phases are stronger outside America. These findings are recommended to investors, financial managers and analysts dealing with international stock indices.

JEL Classification: G02, G14, G15

Keywords

Moon mood emerging and mature markets TGARCH stock returns

Introduction

Psychological factors (e.g., overconfidence and optimism) always play a role in economics and financial markets (Akerlof and Shiller 2009; Ciccone 2011; Kahneman 2003a, 2003b; Krugman 2009; Laibson and Zeckhauser 1998). Further, theories from behavioural finance, such as, the efficient-market theory (stock prices reflect all available information) and calendar-market anomalies, that is, Monday (low returns) and January (high returns) effects, may be explained by psychological principles of decision making¹ (Alt et al. 2011; Garling et al. 2009).

Numerous studies suggest that mood significantly affects human’s behaviour and psychology² (Frijda 1988; Loewenstein et al. 2001; Slovic et al. 2002), while recent studies on behavioural finance show significant effects of investors’ mood³ on asset returns (Cao and Wei 2005; Hirshleifer and Shumway 2003; Yuan et al. 2006). Further, there is much research which shows the importance of weather and other geophysical phenomena (Chang et al. 2008; Pardo and Valor 2003; Saporoschenko 2011), including the lunar cycles (moon phases), on the pricing of assets. Evidence suggests that investors are subject to various behavioural biases when making financial decisions, such as cognitive errors, loss aversion, overconfidence and mood fluctuation (Harlow and Brown 1990).

Moon has a natural power which plays a significant role in phenomena of nature and also on human behaviour; in particular, moon tends to affect decision-makers (Levy and Yagil 2011). The phases of the moon show when people are generally more optimistic or pessimistic. There are two phases of the moon: the new moon and the full moon. The full moon period is the day of the full moon and the seven days before it and after it, when people are more pessimistic. By contrast, the new moon is when the moon is ‘hidden’ and people have positive energy. This period marks the beginning of new life, development and the conception of hope. According to Dichev and Janes (2003), phenomena of nature, such as, lunar cycles or even the energy of ocean, can be used to explain and possibly predict the pattern of stock market prices and their returns. On a general level, if lunar phases affect investors’ mood, these phases may affect asset prices, and therefore, asset returns during full moon phases may be different from those during new moon phases (Keef and Khaled 2011; Liu and Tseng 2009; Yuan et al. 2006). In particular, empirical evidence shows that stock returns are lower on days around a full moon than on days around a new moon. This negative relationship around full moon periods was reported first by Dichev and Janes (2003) and then by Yuan et al. (2006). Dichev and Janes (2003) show a significant lunar effect on the US returns; they report that moon cycles affect human behaviour, especially abnormal behaviour, around full moons. Yuan et al. (2006) employ a pooled regression to investigate the lunar phases–stock market returns relationship using data from 48 countries. They show that stock returns are lower on the days around a full moon than on the days around a new moon. Dowling and Lucey (2008) used weather and biorhythm (including lunar phases) to study the relationship between mood and UK equity pricing (UK Main index and UK Small capitalisation index) for the period 12 December 1994 to 10 November 2004. They employ an ARMA(1,1) Leveraged-GARCH(1,1) model with a generalised error distribution (GED) assumption (UK Main index) and ARMA(1,1)-EGARCH(1,1) ARCH-in-mean with a student’s t-error distribution assumption (UK Small capitalisation index). Dowling and Lucey (2008: 239) report ‘problems with interpreting the lunar phases findings because the findings are stronger for the Main index compared to the Small capitalisation index’. Recently, Lucey (2010) shows the existence of a lunar cycle on silver and platinum returns (precious metals) for the period January 1998 to September 2007. They also report that full moons are bad news for precious metal returns. Finally, Keef and Khaled (2011) use a simple ordinary least squares (OLS) to confirm that full moon is absent when daily returns of 62 international stock indices is considered; they find that lunar effects are weakly influenced by the calendar anomalies (Monday effect and turn-of-the-month effect).

In this article, we empirically examine the effect of mood on asset returns, by investigating the empirical relation between moon phases (new moon and full moon) and stock returns from major international markets. We extend previous related studies by employing a threshold generalised autoregressive conditional heteroscedasticity (TGARCH) model which captures volatility clustering as well as asymmetric behaviour of returns (most previous papers use a simple OLS model).

We focus on two sets of results. The main objective of our study is to test the effect of moon on stock market returns using recent daily data from 59 markets (mature and emerging),⁴ and see if our results obtained from TGARCH are consistent with the recent studies by Keef and Khaled (2011) for 62 international stock indices, Dowling and Lucey (2008) for UK and Yuan et al. (2006) for 48 international countries.

Furthermore, empirical evidence shows that the psychology and behavioural finance theories are strongly linked together; previous studies report that optimism⁵ associated with the calendar effects may have implications for the stock markets (Ciccone 2011). Hence, another goal of our article is to investigate whether there is evidence of the potentially confounding effects, efficient-market hypothesis or calendar-related anomalies (Monday and January effects), on the stock market returns using the foregoing methodology; Yuan et al. (2006) use a simple linear regression and report that the lunar effect is independent of calendar-related anomalies,⁶ while Keef and Khaled (2011: 62) argue that ‘it would appear to be a nigh impossible to advance a market inefficiency explanation for the full moon effects and new moon effects’. Hence, we give further explanations beyond investors’ mood, which in behavioural finance is the most common explanation of market anomalies (DeBondt 2008; Zaleskiewicz 2008).

The remainder of the article is organised as follows. The following section describes the data, while the third section outlines the methodology and presents the empirical results. The fourth section discusses the main findings and concludes the study.

Data Description

Most of our data covers a long-dated period (from January 1990 to December 2010). We consider daily data from the major indices of 59 stock markets (emerging and mature) from all parts of the world (as given in Table 1). The stock market data were collected from Datastream and the data on moon phases were collected from the website of the Munich Astronomical Archive (www.maa.mhn.de). Descriptive statistics results (not reported here) confirm that all of the returns⁷ series follow the stylised facts of financial time series, such as leptokurtosis, volatility clustering and leverage effects (see Bollerslev et al. 1994); in addition, the log levels of prices are found to be I(1), that is, the series are non-stationary.⁸ Hence, we are able to use time series models to capture volatility clustering and test the effect of investors’ mood on stock market returns.

Table 1.

Data Information

Countries (Indices*/Sample Period)
Netherlands (AEX/1.01.1990–31.12.2010)	Japan (NIKKEI225/1.01.1990–31.12.2010)
Austria (ATX/1.01.1990–31.12.2010)	Pakistan (Karachi SE 100/1.01.1990–31.12.2010)
Belgium (BEL-20/1.01.1990–31.12.2010)	Philippines (PSEi/1.01.1990–31.12.2010)
France (CAC-40/1.01.1990–31.12.2010)	US (S&P500 Composite/1.01.1990–31.12.2010)
Germany (DAX-30/1.01.1990–31.12.2010)	Canada (S&P/TSX Composite/1.01.1990–31.12.2010)
US (DOW JONES/1.01.1990–31.12.2010)	Taiwan (SE Weighted/1.01.1990–31.12.2010)
UK (FTSE-100/1.01.1990–31.12.2010)	Turkey (Istanbul SE National 100/1.01.1990–31.12.2010)
Spain (Ibex35/1.01.1990–31.12.2010)	Italy (FTSEMIB/31.12.1997–31.12.2010)
US (NASDAQ-100/1.01.1990–31.12.2010)	Iceland (OMX ICE/31.12.1992–31.12.2010)
Denmark (OMXC20/1.01.1990–31.12.2010)	Bulgaria (SE SOFIX/20.10.2000–31.12.2010)
Finland (OMXH/1.01.1990–31.12.2010)	Hungary (Budapest BUX/1.02.1991–31.12.2010)
Sweden (OMXS30/1.01.1990–31.12.2010)	Kenya (NAIROBI SE/1.01.1990–31.12.2010)
Norway (OSLO All Share/1.01.1990–31.12.2010)	Kuwait (KIC General/28.12.1994–31.12.2010)
US (S&P100/1.01.1990–31.12.2010)	Luxembourg (SE GENERAL/1.04.1999–31.12.2010)
Switzerland (SWISS/1.01.1990–31.12.2010)	Malta (SE MSE/27.12.1995–31.12.2010)
Greece (ATHEX/1.01.1990–31.12.2010)	Argentina (MERVAL/8.02.1993–31.12.2010)
Australia (MSCI /1.01.1990–31.12.2010)	Morocco (ALL SHARE MASI/1.02.2002–31.12.2010)
Bangladesh (SE All Share/1.01.1990–31.12.2010)	India (S&P CNX NIFTY 50/23.04.1996–31.12.2010)
Brazil (Bovespa/1.01.1990–31.12.2010)	New Zealand (NZX50/29.12.2000–31.12.2010)
Chile (GENERAL IGPA/1.01.1990–31.12.2010)	Oman (MUSCAT SEC MKT/22.10.1996–31.12.2010)
Hong Kong (Hang Seng/1.01.1990–31.12.2010)	Peru (LIMA SE GENERAL IGBL/1.02.1991–31.12.2010)
India (BSE Sensex 30/1.01.1990–31.12.2010)	Poland (Warsaw General/16.04.1991–31.12.2010)
Ireland (SE Overall ISEQ/1.01.1990–31.12.2010)	Portugal (PSI-20/31.12.1992–31.12.2010)
Israel (TA-100/1.01.1990–31.12.2010)	Czech Rep. (SE PX/4.06.1994–31.12.2010)
Jakarta (IDX Composite/1.01.1990–31.12.2010)	Russia (RTS/9.01.1995–31.12.2010)
Jordan (Amman SE Fin/1.01.1990–31.12.2010)	Slovakia (SAX-16/14.09.1993–31.12.2010)
Korea (SE COMPOSITE KOSPI/1.01.1990–31.12.2010)	South Africa (FTSE/JSE ALL SHARE/30.06.1995–31.12.2010)
Malaysia (FTSE BURSA MALAYSIA KLCI/1.01.1990–31.12.2010)	Tunisia (TUNINDEX/31.12.1997–31.12.2010)
Mexico (IPC BOLSA/1.01.1990–31.12.2010)	Venezuela (SE General/4.01.1993–31.12.2010)
Cyprus (Cyprus General Price Index/3.09.2004–31.12.2010)

Source:Datastream.

Note:* Name of indices as provided by Datastream.

Methodology

In formulating the hypotheses, we consider the relationships between (a) new/full moon and good/bad news (mood), (b) new/full moon and calendar-related effects (Monday/January) and (c) mood and stock market returns. We re-examine the empirical linkages between moon phases and stock market returns using a TGARCH model which captures an effect where negative shocks have a greater volatility impact than positive shocks (this study is the first investigation of the possibility of moon–calendar anomalies using a TGARCH time series approach which is able to capture the seasonal dependency, or the annual cycle of moon, as well as the fluctuation of returns or volatility response from market shocks).⁹

We not only test whether the phases of moon affect stock returns but also we investigate whether the returns are higher in winter (January effect—see Ciccone 2011;¹⁰ Floros 2008a, 2011; Keef and Khaled 2011; Yuan et al. 2006) and lower during the first day of the week (Monday effect—see Alt et al. 2011; Bohl et al. 2010; Levy and Yagil 2011). These calendar effects are heavily related with the behavioural finance theories (that is, investors’ mood) in the way that (a) the January effect assumes that financial returns in January are large compared to the returns of the other months (Floros 2011), and (b) the Monday effect¹¹ is where the returns are significantly lower over the first trading day of the week (see Levy and Yagil 2011; Yuan et al. 2006); explanations include the settlement procedure hypothesis, measurement errors, the timing of earnings announcements and the influence of institutional versus individual investor trading (Bohl et al. 2010), spillover effect from other large markets, risk–return trade-off and speculative short sales (Charles 2010).

Most previous papers on the same area (Keef and Khaled 2011; Lucey 2010; Yuan et al. 2006) employ a simple OLS model to test the effect of moon on stock market returns; however, simple OLS fails to capture the stylised facts of financial returns, that is, (a) returns have very little autocorrelation, (b) distribution has fatter tails compared to the normal, (c) distribution has negative skewness and (d) variance has positive autocorrelation. In this study, we follow the recent works of Floros (2008b, 2011) and Charles (2010) on modelling behavioural finance; hence, we employ several TGARCH volatility models with Normal, Student’s-t and GED distributional assumptions for the standardised residuals. Both AIC and SIC information criteria select the parsimonious AR(1)-TGARCH(1,1) model which accounts for temporal dependence in variance and excess kurtosis, while it controls the effect of good/bad news (mood) on conditional variance (Glosten et al. 1993; Zakoian 1994); further, we model returns using an AR(1) mean equation, consistently with the non-synchronous trading effect (Charles 2010; Floros 2011).

The AR(1)-TGARCH(1,1) model for returns ρ is given by the following: R_t(1)c₁R_t_–1c₂MON_tc₃JAN_tc₄D_FULLtc₅D_NEWt_t $σ_{t}^{2} = ω + \sum_{i = 1}^{q} a_{i} ε_{t - i}^{2} + γ ε_{t - 1}^{2} d_{t - 1} + \sum_{j = 1}^{p} β_{j} σ_{t - j}^{2}$ (2)

The specification for the conditional mean is given by Equation (1) where JAN is a dummy variable for testing for the January effect (JAN = 1 for January, and 0 otherwise) and MON is a dummy variable for testing for the Monday effect (MON = 1 for Monday, and 0 otherwise). D_FULLt and D_NEWt test for the full moon and new moon effects, respectively.

The specification for the conditional variance is given by Equation (2) where d_t = 1 if ε_t < 0 and d_t = 0 otherwise. In TGARCH model, good news (ε_t > 0) and bad news (ε_t < 0)—or good mood and bad mood—have differential effects on the conditional variance. In particular, good news/mood has an impact of α, while bad news/mood has an impact of α + γ. If γ > 0 and significant, then the leverage effect exists and bad news increases volatility, while if ≠ 0 the news impact is asymmetric; if γ = 0, then the news impact curve is symmetric, that is, past positive shocks have the same impact on today’s volatility as past negative shocks (Floros 2011; Glosten et al. 1993; Zakoian 1994).

Empirical Results

The results¹² from the Equation (2) of the AR(1)-TGARCH(1,1) volatility model (not reported here) confirm the empirical finance literature: (a) the sum of ARCH and GARCH coefficients is very close to one, indicating that volatility shocks are quite persistent; (b) the coefficients of the lagged squared returns (ARCH) and the lagged variances (GARCH) are positive and statistically significant for almost all cases (that is, strong GARCH effects are apparent for the selected sample, and news has a greater impact on stock market returns); (c) the magnitude of the GARCH coefficient, β, is especially high, indicating a long memory in the variance; and (d) positive and significant γ parameter shows that the leverage effect exists and bad news increases volatility.¹³

Table 2 reports mean parameter estimates of the selected AR(1)-TGARCH(1,1) model,¹⁴ defined in the Equation (1), for all countries. First, we report evidence of January effect in 15 stock market returns (Austria, Denmark, Greece, Brazil, Ireland, Pakistan, Philippines, Hungary, Kuwait, Peru, Portugal, Tunisia, Venezuela, Belgium and Cyprus). Hence, for these markets (most of them are emerging markets) the stock market returns are higher in January (winter) than in any other month of the year (empirical results indicate that investors’ confidence peaks in January). In other words, many investors choose to sell some of their stock before the end of the year in order to claim a capital loss for tax purposes (Floros 2011); according to Ciccone (2011: 160), ‘stock susceptible to having optimism impounded into their prices should earn superior returns in January’. Further, our results show significant Monday effects (low returns) in 15 cases (Greece, Bangladesh, Brazil, Chile, Ireland, Jakarta, Malaysia, Mexico, Pakistan, Turkey, Luxembourg, India – S&P Nifty, Poland, Norway and Cyprus); hence, these returns are lower in the first trading day of the week due to the fact that firms announce their profits during that period (Floros 2008a).¹⁵

Table 2.

AR(1)-TGARCH(1,1) Results (Mean Equation)

Coefficient/Country	Austria	US (DOW JONES)	UK
AR(1)	0.112736*** (0.0000)	0.002668 (0.8450)	−0.001897 (0.8918)
MON	−7.33E-05 (0.8396)	0.000873*** (0.0018)	−0.000201 (0.4634)
JAN	0.001022** (0.0270)	0.000212 (0.5447)	−0.000315 (0.4256)
D_NEW	0.000637 (0.3308)	0.000180 (0.7008)	0.000882* (0.0873)
D_FULL	2.40E-05 (0.9696)	−0.000175 (0.7223)	0.000460 (0.3599)
Coefficient/Country	Denmark	Sweden	Czech Rep.
AR(1)	0.085807*** (0.0000)	0.034928* (0.0135)	0.133155* (0.0000)
MON	−0.000336 (0.3226)	0.000594 (0.1292)	0.000139 (0.7227)
JAN	0.001027** (0.0244)	0.000180 (0.7275)	−0.00034 (0.5031)
D_NEW	−0.000386 (0.5263)	0.000532 (0.4380)	−0.000672 (0.3269)
D_FULL	0.000355 (0.5866)	0.000621 (0.3736)	−0.000632 (0.3731)
Coefficient/Country	Greece	Australia	Bangladesh
AR(1)	0.144444*** (0.0000)	0.020157 (0.1690)	0.072600 (0.2282)
MON	−0.001142** (0.0176)	0.000122 (0.6583)	−0.002650** (0.0183)
JAN	0.002231*** (0.0010)	−0.000173 (0.6229)	−0.000306 (0.5873)
D_NEW	0.000386 (0.5943)	−0.000941* (0.0604)	0.003053** (0.0455)
D_FULL	−0.000595 (0.4803)	−0.000257 (0.6110)	−0.002065 (0.1201)
Coefficient/Country	Brazil	Chile	India (BSE Sensex)
AR(1)	0.015612 (0.4620)	0.320007*** (0.0000)	0.096326*** (0.0000)
MON	−0.001694** (0.0280)	−0.001858*** (0.0000)	0.000208 (0.6966)
JAN	0.029375** (0.0422)	0.000154 (0.5810)	−0.000993 (0.1267)
D_NEW	0.000672 (0.6251)	0.000649* (0.0782)	0.000512 (0.5417)
D_FULL	0.002495* (0.0796)	0.000541 (0.1226)	2.65E-05 (0.9756)
Coefficient/Country	Ireland	Israel	Jakarta
AR(1)	0.114410*** (0.0000)	0.057212*** (0.0002)	0.199664*** (0.0000)
MON	−0.000864*** (0.0048)	0.000569 (0.1224)	−0.001405*** (0.0003)
JAN	0.001338*** (0.0047)	−0.000561 (0.4371)	0.000614 (0.2149)
D_NEW	0.000336 (0.5189)	−0.000478 (0.5373)	−0.000846 (0.1330)
D_FULL	−4.23E-05 (0.9334)	−0.000352 (0.6168)	−0.000965 (0.1936)
Coefficient/Country	Jordan	Malaysia	Mexico
AR(1)	0.076587*** (0.0041)	0.187768*** (0.0000)	0.141711*** (0.0000)
MON	0.000359 (0.3534)	−0.000999*** (0.0009)	−0.000809* (0.0685)
JAN	0.001113 (0.1742)	0.000283 (0.5553)	4.11E-05 (0.9464)
D_NEW	−0.001024 (0.2588)	3.37E-05 (0.9471)	−0.000315 (0.6811)
D_FULL	−0.001402** (0.0370)	−0.000157 (0.7781)	−0.000270 (0.7290)
Coefficient/Country	Pakistan	Philippines	US (S&P500)
AR(1)	0.123361*** (0.0000)	0.175389*** (0.0000)	−0.000820 (0.9509)
MON	−0.001242*** (0.0050)	−0.000679 (0.1490)	0.000579** (0.0461)
JAN	0.001150* (0.0583)	0.001212* (0.0781)	0.000340 (0.3130)
D_NEW	−0.000212 (0.7768)	−0.000142 (0.8578)	0.000159 (0.7242)
D_FULL	6.03E-05 (0.9435)	4.58E-05 (0.9543)	9.31E-05 (0.8482)
Coefficient/Country	Canada	Turkey	Iceland
AR(1)	0.120064*** (0.0000)	0.083721*** (0.0000)	0.098165* (0.0428)
MON	7.80E-05 (0.7657)	−0.001426* (0.0835)	0.000316 (0.1466)
JAN	−6.09E-05 (0.8486)	0.000958 (0.3886)	−0.000257 (0.5316)
D_NEW	0.000185 (0.6566)	−0.001098 (0.4760)	0.000289 (0.5680)
D_FULL	0.000900** (0.0366)	−0.000824 (0.6288)	2.99E-05 (0.9520)
Coefficient/Country	Bulgaria	Hungary	Kenya
AR(1)	0.1254* (0.0000)	0.107994* (0.0000)	0.3740* (0.0000)
MON	0.000652 (0.1660)	−8.05E-05 (0.8524)	2.26E-05 (0.9407)
JAN	−0.001221 (0.1508)	0.001277** (0.0778)	0.000296 (0.2006)
D_NEW	−0.000161 (0.8276)	−0.000363 (0.7272)	−0.000494 (0.3215)
D_FULL	0.001705** (0.0816)	0.000134 (0.8758)	0.000555 (0.2530)
Coefficient/Country	Kuwait	Luxembourg	Argentina
AR(1)	0.087394* (0.0011)	0.025578 (0.2965)	0.063566* (0.0001)
MON	−0.000174 (0.7093)	0.000820* (0.00723)	−0.000599 (0.3815)
JAN	0.029875* (0.0204)	0.000106 (0.8715)	−0.000837 (0.2801)
D_NEW	−0.000781 (0.3939)	0.001309 (0.1338)	−0.003652* (0.0156)
D_FULL	0.000238 (0.7060)	0.000414 (0.6354)	6.46E-05 (0.9579)
Coefficient/Country	India (S&P NIFTY)	Peru	Poland
AR(1)	0.068589* (0.0003)	0.283191* (0.0000)	0.119260* (0.0000)
MON	−0.001266* (0.0219)	−7.28E-05 (0.9837)	−0.001558* (0.0104)
JAN	−0.000852 (0.3906)	0.001024** (0.0631)	−0.000571 (0.3691)
D_NEW	0.000171 (0.8547)	0.000236 (0.6952)	0.000255 (0.8377)
D_FULL	−9.41E-05 (0.9250)	−0.00043 (0.5086)	−0.00108 (0.2152)
Coefficient/Country	Portugal	Russia	Slovakia
AR(1)	0.135552* (0.0000)	0.118053* (0.0000)	−0.006757 (0.7304)
MON	−0.000259 (0.3765)	0.000567 (0.4212)	−0.000469 (0.4089)
JAN	0.002094* (0.0000)	−0.000891 (0.2964)	0.000892 (0.1019)
D_NEW	0.000593 (0.2494)	−0.000193 (0.8840)	−0.000189 (0.8032)
D_FULL	0.000386 (0.4301)	0.002728** (0.0705)	−0.000262 (0.731)
Coefficient/Country	Tunisia	Venezuela	Netherlands
AR(1)	0.228916* (0.0000)	0.196687* (0.0000)	0.011384 (0.4273)
MON	4.64E-05 (0.8109)	0.000306 (0.5506)	0.000321 (0.3075)
JAN	0.000606* (0.02691)	−0.001847* (0.0238)	0.000136 (0.7674)
D_NEW	−0.001018* (0.0013)	−0.000113 (0.9050)	0.000500 (0.4460)
D_FULL	5.00E-05 (0.8840)	0.000838 (0.4084)	0.000122 (0.8148)
Coefficient/Country	Belgium	France	Germany
AR(1)	0.083877*** (0.0000)	0.000611 (0.9646)	−0.001692 (0.9020)
MON	−0.000101 (0.7048)	−0.000490 (0.1960)	−6.21E-05 (0.8783)
JAN	0.000694* (0.0580)	0.000135 (0.7782)	0.000335 (0.5355)
D_NEW	0.000225 (0.7768)	0.000775 (0.2760)	0.000560 (0.4803)
D_FULL	0.000841* (0.0900)	0.000883 (0.2023)	0.000463 (0.5147)
Coefficient/Country	Spain	US (NASDAQ)	South Africa
AR(1)	0.058013*** (0.0001)	−0.000356 (0.9796)	0.091202* (0.0000)
MON	−0.000581 (0.1829)	0.000265 (0.5472)	−0.000324 (0.4219)
JAN	0.000312 (0.5396)	−0.000193 (0.7399)	−0.000295 (0.4793)
D_NEW	0.000916 (0.1567)	0.000511 (0.5030)	0.000492 (0.4547)
D_FULL	0.000241 (0.7048)	0.000574 (0.4561)	0.000301 (0.7073)
Coefficient/Country	Finland	Norway	US (S&P100)
AR(1)	0.105233*** (0.0000)	0.087432*** (0.0000)	−0.030579** (0.0222)
MON	−0.000376 (0.3564)	−0.000685* (0.0614)	0.000692** (0.0152)
JAN	0.001024 (0.1320)	0.000121 (0.8017)	0.000339 (0.3384)
D_NEW	0.000985 (0.1785)	0.000582 (0.4257)	0.000285 (0.5372)
D_FULL	−0.000328 (0.6573)	3.91E-05 (0.9498)	−6.29E-05 (0.9002)
Coefficient/Country	Switzerland	Italy	Hong Kong
AR(1)	0.028918* (0.0592)	−0.022071 (0.2074)	0.068200*** (0.0000)
MON	−2.10E-05 (0.9485)	0.000539 (0.2136)	−0.000377 (0.4372)
JAN	−0.000132 (0.7562)	0.000319 (0.5830)	−0.000203 (0.7347)
D_NEW	0.001417** (0.0340)	−9.34E-05 (0.8925)	−0.000407 (0.5942)
D_FULL	0.000359 (0.5427)	0.000691 (0.3396)	0.000142 (0.8611)
Coefficient/Country	Korea	Japan	Cyprus
AR(1)	0.035606** (0.0149)	−0.013452 (0.3317)	0.082151* (0.0018)
MON	−0.000564 (0.2900)	−0.000432 (0.3680)	−0.002693* (0.0040)
JAN	−0.000850 (0.2308)	−0.000256 (0.6846)	0.004663* (0.0127)
D_NEW	0.000276 (0.7597)	−0.000684 (0.3742)	0.004024* (0.0350)
D_FULL	0.000282 (0.7487)	−0.001088 (0.1427)	−0.000633 (0.7256)
Coefficient/Country	Taiwan	Malta	Morocco
AR(1)	0.037385*** (0.0087)	0.233525* (0.0000)	0.253613* (0.0000)
MON	0.000369 (0.4912)	−0.000232 (0.3983)	0.000137 (0.6787)
JAN	0.000410 (0.5011)	0.000421 (0.1128)	0.000535 (0.2801)
D_NEW	−0.000732 (0.4196)	−0.000238 (0.5389)	−0.000502 (0.3924)
D_FULL	−0.000209 (0.8488)	0.000429 (0.3395)	0.000507 (0.4523)
Coefficient/Country	New Zealand	Oman
AR(1)	0.046756* (0.0318)	0.21512* (0.0000)
MON	−0.000126 (0.6780)	−0.000676 (0.1223)
JAN	6.44E-05 (0.8644)	−0.000512 (0.4423)
D_NEW	−0.000952 (0.1667)	−0.000881 (0.1661)
D_FULL	−0.000314 (0.5548)	0.000507 (0.4395)

Notes:Probability values of t-statistics are in the parentheses.

*, **, *** significant at 10%, 5% and 1% levels, respectively.

Our results show significant effect of new moon in eight cases (UK, Switzerland, Australia, Bangladesh, Chile, Cyprus, Argentina and Tunisia), while there is evidence of full moon effect in six cases (Brazil, Jordan, Canada, Bulgaria, Russia and Belgium). In particular, we find a negative effect of moon (new and full) on four international stock market returns, as follows: Australia, Argentina and Tunisia (new moon) and Jordan (full moon); the rest of the markets show significant positive effects. This finding is partially in line with Yuan et al. (2006) who report that stock returns are lower on the days around a full moon than on the days around a new moon.

Conclusion

It is widely believed that the phases of the moon affect behaviour (Owens and McGowan 2006). Moon is considered as an influential source of energy; hence it is linked with elements in a life’s cycle. In particular, on the one hand, new moon marks the beginning of new life, development and the conception of hope. On the other hand, full moon marks the beginning of growth and production, while financial decision-making may be explained by psychological factors (Garling et al. 2009; Keef and Khaled 2011). It is widely accepted that the relationship between moon phases and mood may have implications to investors’ behaviour (Levy and Yagil 2011). Yuan et al. (2006) argue that lunar cycles have a mood-related economic impact. According to Economist (2001: 1021), ‘there is a growing, heavenly body of evidence that share prices are influenced by forces from outer space—well, the sun and the moon, anyway.’ Past empirical evidence is mixed; it shows that stock returns are lower on days around a full moon than on days around a new moon or that there are strong lunar cycle effects in stock returns (Dichev and Janes 2001). This article provides new evidence of a lunar cycle effect on stock market returns of emerging and mature markets around the world. Although we have some evidence about the effect of moon phases on stock market returns, we know little about the linkages between efficient-market theory, calendar-related effects and investors’ mood resulting from moon phases. Therefore, we investigate the effects of moon phases (new moon and full moon) on international stock returns of 59 international markets using an AR(1)-TGARCH model; we extend the work of previous authors, primarily by considering calendar-related dummy variables for Monday and January effects. For instance, given the fact that January (Monday) is hypothesised to be a month (day) of renewed optimism (pessimism), investors may feel a sense of optimism (pessimism) related to the overall market performance at the beginning of the year (week) (see Ciccone 2011).

The empirical results show significant full moon effects in six markets, and significant new moon effects in eight markets. We find significant positive effect of new moon on stock market returns in five cases (UK, Switzerland, Bangladesh, Chile and Cyprus), while a negative effect of full moon is reported for the case of Jordan only. This is partially consistent with Yuan et al. (2006) and Lucey (2010); Lucey (2010) argues that asset returns should peak at the new moon and reach a trough at the full moon (that is, full moons are bad news for the returns). In our study, we find that full moon is mainly good news (positive effect) for the stock market returns.

In addition, seven markets show evidence of full/new moon effects as well as Monday/January effects. In particular, Bangladesh and Chile have a positive effect of new moon and Monday effects, while Tunisia shows a negative effect¹⁶ of new moon and January effects; so, this is supported by emerging markets. Belgium also provides evidence of positive full moon effect along with January effect on stock market returns. Further, Brazil shows evidence of a positive full moon effect on stock market returns as well as Monday and January effects. In addition, the stock market returns of the Cypriot stock exchange are influenced positively by the new moon as well as there is a strong evidence of Monday and January effects on the returns. In other words, we find that lunar effects are strongly influenced by the calendar anomalies (Monday effect and January effect); this is not in line with Keef and Khaled (2011) for 62 international stock indices. Hence, we prove that (a) investors dealing with the Cypriot and Brazilian emerging stock markets are influenced by calendar anomalies, and (b) lunar effects are strongly affected by Monday and January effects. In addition, we find that the lunar phases are stronger outside America (this is in line with Dichev and Janes 2003).

Finally, our study shows that half of the international stock markets support the theories of behavioural finance (that is, investors’ decisions are influenced by many factors including their current mood). Therefore, we find that the January (Monday) effects are driven by investors’ optimism (pessimism).

Further research should examine the impact of lunar days (moon day 1–30) using other quantitative approaches and methods (for example, event study methodology, APARCH model, Monte Carlo simulations).

Footnotes

Notes

References

Akerlof

G.A.

Shiller

R. J.

(2009), Animal Spirits: How Human Psychology Drives the Economy, and Why it Matters for Global Capitalism. Princeton, NJ: Princeton University Press.

Alt

Fortin

Weinberger

(2011), ‘The Monday Effect Revisited: An Alternative Testing Approach’, Journal of Empirical Finance, 18(3): 447–60.

Bohl

M.T.

Schuppli

Siklos

P. L.

(2010), Stock Return Seasonalities and Investor Structure: Evidence from China’s B-share Markets’, China Economic Review, 21(1): 190–201.

Bollerslev

Wooldridge

J. M.

(1992), ‘Quasi-maximum Likelihood Estimation and Inference in Dynamic Models with Time Varying Covariances’, Econometric Reviews, 11(2): 143–72.

Bollerslev

Engle

R. F.

Nelson

D. B.

(1994), ‘ARCH Models’, in Engle

R. F.

McFadden

(eds), Handbook of Econometrics, Vol. 4, pp. 2959–3038. North-Holland, Amsterdam.

Cao

Wei

(2005), ‘Stock Market Returns: A Note on Temperature Anomaly’, Journal of Banking and Finance, 29(6): 1559–73.

Chang

S.C.

Chen

S.S.

Shou

R.K.

Lin

Y. H.

(2008), ‘Weather and Intraday Patterns in Stock Returns and Trading Activity’, Journal of Banking and Finance, 32(9): 1754–66.

Charles

(2010), ‘The Day-of-the-Week Effects on the Volatility: The Role of the Asymmetry’, European Journal of Operational Research, 202(1): 143–52.

Ciccone

S.J.

(2011), ‘Investor Optimism, False Hopes and the January Effect’, Journal of Behavioral Finance, 12(3): 158–68.

10.

DeBondt

W.F.M.

(2008), ‘Stock Prices: Insights from Behavioural Finance’, in Lewis

(ed.), The Cambridge Handbook of Psychology and Economic Behaviour, pp. 64–104. Cambridge: Cambridge University Press.

11.

Dichev

I.D.

Janes

T. D.

(2001), ‘Lunar Cycle Effect in Stock Returns’. Available at SSRN http://ssrn.com/abstract=281665 or doi:10.2139/ssrn.281665

12.

Dichev

I.D.

Janes

T. D.

(2003), ‘Lunar Cycle Effects in Stock Returns’, Journal of Private Equity, 6(4): 8–29.

13.

Dowling

Lucey

B. M.

(2008), ‘Mood and UK Equity Pricing’, Applied Financial Economic Letters, 4(4): 233–40.

14.

Economist (2001), ‘What a Little Moonlight Can Do: How the Sun and the Moon Move Markets’, The Economist, 20 October, 102–05.

15.

Floros

(2008a), ‘Stock Market Returns and the Temperature Effect: New Evidence from Europe’, Applied Financial Economics Letters, 4(6): 461–67.

16.

Floros

(2008b), ‘The Monthly and Trading Month Effects in Greek Stock Market Returns: 1996–2002’, Managerial Finance, 34(7): 453–64.

17.

Floros

(2011), ‘On the Relationship between Weather and Stock Market Returns’, Studies in Economics and Finance, 28(1): 5–13.

18.

Frijda

(1988), ‘The Laws of Emotion’, American Psychologist, 43(5): 249–358.

19.

Garling

Kirchler

Lewis

Raaij

F. V.

(2009), ‘Psychology, Financial Decision Making and Financial Crises’, Psychological Science in the Public Interest, 10(1): 1–47.

20.

Glosten

L.R.

Jagannathan

Runkle

D. E.

(1993), ‘The Relationship between Expected Value and the Volatility of the Nominal Excess Return on Stocks’, Journal of Finance, 48(5): 1779–801.

21.

Harlow

W.V.

Brown

K. C.

(1990), ‘Understanding and Assessing Financial Risk Tolerance: A Biological Perspective’, Financial Analysts Journal, 46(6): 50–62.

22.

Hirshleifer

Shumway

(2003), ‘Good Day Sunshine: Stock Returns and the Weather’, Journal of Finance, 58(3): 1009–32.

23.

Kahneman

(2003a), ‘A Perspective on Judgment and Choice: Mapping Bounded Rationality (Nobel Prize lecture)’, American Psychologist, 58(9): 697–720.

24.

Kahneman

(2003b), ‘Experiences of Collaborative Research’, American Psychologist, 58(9): 723–30.

25.

Keef

S.P.

Khaled

M. S.

(2011), ‘Are Investors Moonstruck? Further International Evidence on Lunar Phases and Stock Returns’, Journal of Empirical Finance, 18(1): 56–63.

26.

Krivelyova

Robotti

(2003), ‘Playing the Field: Geomagnetic Storms and International Stock Market’, Working Paper 2003-5, Federal Reserve Bank of Atlanta.

27.

Krugman

(2009), The Return of Depression Economics and the Crisis of 2008. London: Nortons.

28.

Laibson

Zeckhauser

(1998), ‘Amos Tversky and the Ascent of Behavioral Economics’, Journal of Risk and Uncertainty, 16(1): 7–47.

29.

Levy

Yagil

(2011), ‘Air Pollution and Stock Return in the US’, Journal of Economic Psychology, 32(3): 374–83.

30.

Liu

Tseng

(2009), ‘A Bayesian Analysis of Lunar Effects on Stock Return’. Available at SSRN: http://ssrn.com/abstract=1366663

31.

Loewenstein

G.F.

Weber

E.U.

Hsee

C.K.

Welch

(2001), ‘Risk as Feelings’, Psychological Bulletin, 127(2): 267–86.

32.

Lucey

B.M.

(2010), ‘Lunar Seasonality in Precious Metal Returns?’, Applied Economics Letters, 17(9): 835–38.

33.

Nofsinger

J.R.

(2003), ‘Social Mood and Financial Economics’. Available at http://zonecours.hec.ca/documents/E2005-1-395070.Texte19-30-253-00-E05-SocialMood...pdf

34.

Owens

McGowan

I. W.

(2006), ‘Madness and the Moon: The Lunar Cycle and Psychopathology’, German Journal of Psychiatry. Available at http://www.gjpsy.uni-goettingen.de (ISSN 1433-1055).

35.

Pardo

Valor

(2003), ‘Spanish Stock Returns: Where is the Weather Effect?’, European Financial Management, 9(1): 117–26.

36.

Saporoschenko

(2011), ‘The Effect of Santa Ana Wind Conditions and Cloudiness on Southern California Stock Returns’, Applied Financial Economics, 21(10): 683–94.

37.

Slovic

Finucane

M.L.

Peters

MacGregor

D. G.

(2002), ‘The Affect Heuristic’, in Gilovich

Griffin

Kahneman

(eds), Heuristics and Biases: The Psychology of Intuitive Judgment, pp. 397–420. New York: Cambridge University Press.

38.

Yuan

Zheng

Zhu

(2006), ‘Are Investors Moonstruck? Lunar Phases and Stock Return’, Journal of Empirical Finance, 13(1): 1–23.

39.

Zakoian

J.M.

(1994), ‘Threshold Heteroscedastic Models’, Journal of Economic Dynamics and Control, 18(5): 931–55.

40.

Zaleskiewicz

(2008), ‘Behavioral Finance’, in Altman

(ed.), Handbook of Contemporary Behavioral Economics, pp. 707–28. Armonk, NY: Sharpe.