Religious practices and market anomalies: an inquiry into the persistence of the religious holiday effect

Religious holidays are observed with considerable interest worldwide, yet they are not always meant to be joyous. For instance, Buddha Poornima is celebrated in the name of Lord Buddha, Krishna-Ashtami as Lord Krishna’s birthday and Good Friday as the crucifixion of Lord Jesus Christ. Religious festivals are commemorated by historical events such as birth, death, conquest, defeat and so on. The role of religion in people’s lives, behaviour and decision-making is significant (Fernando, 2005); moreover, religious practices and beliefs have a big impact on economic growth (Weber, 1930; McCleary and Barro, 2006; Bryan et al., 2021). Over the past decades, studies have been conducted on how these beliefs affect stock market return and volatility (Elliot and Echols, 1976; Hilary and Hui, 2009; Canepa and Ibnrubbian, 2014; Atmaz and Basak, 2018; Tadepalli et al., 2021 and among others). In this sense, cultural and religious beliefs can influence the psychological aspect of investors that nib trading behaviour in stock exchanges. Nofsinger (2005) claimed actions and performances of people are influenced by their religious or cultural beliefs, derived from the psychological aspect of optimism or pessimism in a society at a given time. Bley and Saad (2010) document that the behaviour of the local market returns depends on the belief system of the foreign investor. Investors trading behaviours are relatively higher during cultural holidays as compared to that of non-cultural holidays (Segaran and Chia, 2021). Religious commemoration impacts investor mood depending on the mood of religious events. Al-Ississ (2015) provides evidence that mood as a potential driving effect for holiday anomalies and stock returns increased with changes in investors moods welcoming the blessing and forgiveness of sin and lowered returns with mourning and atonement.

Asset pricing in market stock has been long-standing research, contrary to efficient markets, a large amount of empirical evidence indicates that relevant information is delayed in being incorporated into prices. Calendar anomalies are those that are unexplained by historical pricing models and seasonal anomalies in the stock market imply market inefficiency. A considerable amount of earlier empirical literature suggests that seasonal trends affect security returns (Rozeff and Kinney, 1976; De Bondt and Thaler, 1987; Heston and Sadka, 2008; Bentzen, 2009). Seasonal market anomalies such as the turn of the month effect, holiday effect and day of the week are some of such calendar anomalies that exist in most of the global securities markets. This trend appears to skirmish the notion of efficiency and market anomalies in the securities market. Compared to other seasonal anomalies, cultural and religious events are the least explored (Mitchell and Ong, 2006; Wasiuzzaman, 2017; Hasan et al., 2022). Researchers have turned their attention to anomalies related to religious and cultural events (Sasikirono and Meidiaswati, 2017; McGowan and Jakob (2010); Wasiuzzaman, 2017; Tadepalli et al., 2021), however, few studies have been conducted on emerging market economies and frontier market. Given the theoretical and practical importance of religious anomalies, it is quite interesting to examine whether specific religious holidays affect the security of stock prices. A concern in this study is whether Indian cultures and manifold religions impact religious events in stock market returns and volatility.

Our contribution to this paper is divided into threefold: Firstly, it enriches the existing literature on behavioural finance perspective that the inefficiency of a country’s stock market depends on its beliefs and cultural aspects in the region. Secondly, to the best of our knowledge, no study has comprehensively examined the holistic (multi-faith observances holiday) religious holidays covering all the major religious events with long memory data over a period of 30 years. Kumar (2012) and Srikanth and Ram (2013) conducted only on the Diwali effect; Goyal et al. (2016) studied a religious holiday for a period of six years and Tadepalli et al. (2021) were concerned with Hinduistic-mandated holidays. Finally, this study is the first to evaluate the pre- and post-effects of religious holidays on the Indian stock market returns using a symmetric and asymmetric GARCH effect model. No such study has been conducted on a threshold religious holiday among the diverse religions in this market. This comparative study provides valuable insights for market participants. The findings can help investors anticipate share price movements and obtain crucial inputs for designing and timing their trading strategies due to religious anomalies.

Our main findings indicate that the persistence of returns and volatility during religious holidays was found to be minimal nevertheless cannot be entirely disregarded. The pre-religious holiday trading return tends to show higher returns as compared to post-religious holiday trading days. This phenomenon aligns with evidence supporting the notion that religious holidays affect certain aspects of individual lives (Fernando, 2005) and investors are more likely to spend time practising religion than trading stocks (Al-Khazali et al., 2017), resulting in post-religious holiday with lower returns. The drop in return volatility during the holiday season can be attributed to the change in investor behaviour caused by religious practices. Our study reveals an increase in return and a dip in volatility during the mandated religious holiday in the Indian stock exchanges under study. In addition, pre-religious holidays are less volatile than post-religious holidays. When the period was divided into subsamples, the persistence of religious anomalies was found to be significant over time.

The remainder of this paper is structured as follows. Section 2 provides a literature review and related studies. Section 3 presents the data and methodology used in the analysis. The empirical analysis and study outcomes are covered in Section 4. Finally, Section 5 concludes the study.

The first evidence of a holiday effect on trading days was conducted by Field (1934), using data from Dow Jones and the New York Stock Exchange. According to this, holiday anomalies cause stock values to fluctuate between higher and lower before and after holidays. The most promising evidence of the holiday effect lies in the psychological factor of investors rather than rational factors, as investors tend to buy stocks before holidays because of “high spirit” and “holiday euphoria” (Vergin and McGinnis, 1999; Marrett and Worthington, 2009; Al-Ississ, 2015). Studies have found that a holiday prior to public holiday results in higher stock returns as compared to other trading days (Lakonishok and Smidt, 1988; Cadsby and Ratner, 1992; Plastun et al., 2019; Pinto et al., 2022 and among others). This implies that investors’ moods are affected by the holiday and that investors do not always behave rationally and predictably but are influenced by certain biases or cognitive. For instance, in Muslim-majority countries, the Ramadan effect increased stock return (Lai and Windawati, 2017) and volatility persistence due to the Ramadan effect (Al-Khazali et al., 2017). The Chinese New Year effect significantly increases risk and return (Yen and Shyy, 1993; Yuan and Gupta, 2014). A distraction effect results in delayed response (bad news) to stock prices on Christian calendar’s Good Friday, Easter Sunday and Christmas holidays were observed (Pantzalis and Ucar, 2014). Frieder and Subrahmanyam (2004) examined the impact of St. Patricks Day and the Jewish holidays of Rosh Hashanah and Yom Kippur on US equity market. They found that returns were significantly higher before the holiday and effect negatively after the holiday and trading volumes were low during the holidays. As a result, the market reflects the solemnity of religious occasions as well as “mood” explains the market movement. Dowling and Lucey (2005) found that investor mood impacts stock return, causing people to make irrational investment decisions when in a good mood. In addition, Johnson and Tversky (1983) claimed that judgement is compatible with mood even when the subject matter has no relation to the cause of the mood. A similar result was reported by Robins and Smith (2019). In light of the above literature, any nonconformity from the normal behaviour of stock prices and volatility could be due to the information processing that investors and their behaviours or sentiments undertake during religious holidays. Based on the above discussion, the following hypotheses are proposed:

The extant literature acknowledges that religious holiday anomalies influence stock return. For instance, the Islamic calendar effect (McGowan and Jakob, 2010; Halari et al., 2015; Hassan and Kayser, 2019) reports that Muslim holidays affect investors’ moods and emotions, affecting investors’ behaviour and stock markets in Islamic countries. Christian holidays also have a significant influence on stock return during Good Fridays and Christmas (Sasikirono and Meidiaswati, 2017). Similarly, Frieder and Subrahmanyam (2004) found significantly higher trading volume and stock return on the preceding day of Jewish Holy days and St. Patrick’s Day. Prior studies on holiday anomalies (Ariel, 1987; Cadsby and Ratner, 1992; Lakonishok and Smidt, 1988) document higher significant returns on pre-holiday than during other periods. Chia et al. (2015) found that Chinese New Year holiday returns were significantly higher two days before and one day after an event. In addition, pre-holiday volatility is lower than post-holiday volatility. Moreover, Chaouachi (2021) findings suggest that the Tunisian stock market experiences positive with higher returns than an average return during these religious festival periods. The persistence of this effect could suggest market inefficiency, cultural factors, higher consumer spending and increased optimism. However, Chan et al. (1996) found no significant difference in return due to Islamic New Year but cultural events show a significant influence on stock return in Kuala Lumpur Stock Exchanges. Al-Khazali et al. (2017) documented a significant change in returns and volatility in the Islamic stock market. Tadepalli et al. (2021) observed the persistence of the Hindu festival in Indian capital market and sectoral indices and concluded the persistence of pre-religious holiday effect in the Indian market. From the discussion above, it can be observed that previous studies are conducted mainly on Islamic festivals, Hinduistic religious holidays, New Year and Christian festivals, we postulate to determine whether mandated religious holidays have a significant pre- and post-influence on the stock returns:

Moreover, various studies claim that there are no religious holiday anomalies in certain stock returns. Srikanth and Ram (2013) conducted a study using a non-parametric t-test, Wilcoxon signed rank test and run test and documented that religious holiday had no impact on the Indian stock exchange. Goyal et al. (2016) found no pre and post-festival effects in the Nifty index and among the BRICS nations. This result was against the finding of Kumar (2012) on Mahurat trading (Mahurat trading occurs during Diwali on Indian stock exchanges for an hour in the evening). A study conducted by Devakumar and Anuradha (2017) examined the impact of the Indian festival on the CNX Nifty index using a paired t-test and found no festival effect. Such studies such as Bialkowski et al. (2012), Akbalik and Tunay (2016) and Hassan and Kayser (2019) found no significant relationship between Ramadan effect and stock market return and volatility. No Eid al-Fitr holiday effect was found in Malaysia (McGowan and Jakob, 2010). Ozkan (2019) also found no significant Hijri calendar effect in the Borsa Istanbul gold market. In that case, it would be interesting to further explore this anomaly by dividing the sample into different sub-sample periods. A sub-sample excluding the financial crisis, pandemic and recession assumed to shed light on these anomalies. No such study has been conducted in Indian stock market on the persistence of religious holiday anomalies over time and we formulate the following hypothesis:

The study uses daily closing stock prices of the Indian stock market from 1 January 1990, to 30 April 2022. The Sensex and Nifty 50 indices were chosen to represent the Indian stock market. Our selection of these indices is based on their substantial market capitalisations and considered as indicators of Indian economic performance. Sensex, comprising 30 stocks, and Nifty 50, consisting of 50 stocks, are both large-cap and robust companies. Considering their composition, these indices represent broader market trends. The daily prices for these indices were obtained from the official websites of the Bombay Stock Exchange (BSE) ^[1] and National Stock Exchange (NSE) ^[2], respectively. We controlled the Asian financial crisis (AFC) during the 1997–1998 period, the Global Financial Crisis (GFC) during 2007–2008 and the global pandemic 2020–2021 period to ensure that seasonal anomalies in Indian stock returns are not influenced by these financial and economic crises. To avoid fluctuation of return during the last day of the year and the first day of the year over the controlled period, we took an average return of the last trading day and second trading days as the first trading days return for the continuous daily return.

The logarithmic daily percentage return for the sample period is determined by applying equation (1):

where R_t is the index return, ln is the natural log of the underlying market series, P_t is the closing value of a given index on a specific trading day and P_{t− 1} is the closing value of a given index on the previous day (t − 1).

Religious holidays are celebrated according to the religions’ calendars. For instance, Jewish society follows the Hebrew calendar, Christians follow the Gregorian calendar and Muslims follow the Hijri (moon cycle) calendar. We collected data on religious holidays from drigPanchang ^[3] (an Indian festival and Hindu calendar) and world time date ^[4] for the study. Religious festivals are defined as observed holidays or public holidays in India resulting in the closure of the stock exchange trade. The holidays taken into consideration for empirical analysis do not always fall on the same Gregorian calendar. The lists are as follows ( Appendix 1): Buddha Purnima, Diwali, Holi, Christmas, Good Friday, Eid al-Fitr, Dussehra, Mahavir Jayanti, Ganesh Chaturthi, Maha Shivratri, Guru Nanak Jayanti, and Muharram. We conducted a pre-religious holidays and post-religious holidays during the event. The pre-religious holiday returns are defined as close-to-close stock returns prior to the market trading break. Likewise, post-holiday returns are the next trading day after the closure of the market.

In examining the existence of religious holiday anomalies in the stock return, we conduct a test on mean return to find out whether significant pre- and post-religious holidays are different from other trading days. In the second stage of analysis, the standard ordinary least square (OLS) regression equation with 12 dummy variables is used to study the effect on return due to religious holidays. The following regression is used to capture the effect of religious holidays on stock index return:

where, R_i,t = the return of stocks index on day (t) on the major stock exchange in India (i), where, β₁ChristmasDum to β₁₁MuharamDum are dummy variables for each religious festival holiday (Gregorian calendar), taking the value of 1 for each holiday, otherwise 0. β₁ to β₁₂ are coefficient for corresponding holiday effect and ε_t is the error term. The religious holidays occurrence vary for each day and no dummy variable traps was found to be significant.

The OLS model has a drawback with the assumption that the variance error term is constant over time. This leads to inaccurate confidence intervals, statistical tests and skewed estimates of the standard error and coefficients. Furthermore, the error term has a non-normal distribution and the volatility of stock returns varies over time, neither of which can be captured by the usual OLS model (Connolly, 1989). Therefore, to address non-normal error terms and robustness standard GARCH model of Bollerslev (1986) and the asymmetric GARCH of Ding et al. (1993) have been applied. This model is capable of capturing the unequal weighting structure (time-varying variability) in the variance of the error terms and long memory behaviour of stock return. Moreover, it is a conditional model that estimates mean and also conditional variance. Bollerslev (1986) specification of the model is expressed as follows:

where, $∑n=112βnDn$ represent dummy variables for all the religious holidays. β₁ to β₄ are coefficients for the corresponding holiday effect. The variance equation $σt2$⁠, which in turn represents the conditional variance equation is as follows:

where, $εt−12$ is the innovation, α_i represent the ARCH term that measures the impact of past innovation on current variance. β_i is the GARCH model that measures the impact of past variance on current variance. ε_t follows the generalised error distribution (GED) to reduce the normal distribution with mean zero and conditional variance h_t suggested by Nelson (1991). In addition, the log-likelihood function of the GED was maximised by using the legacy technique of EViews. GARCH model with GED distribution have minimal mean square error and outperformed in GARCH model estimation (Kumar and Patil, 2016).

On the other hand, GARCH model captures the symmetric behaviour of the market reaction (Nelson, 1991; Chia et al., 2015) without conceiving the asymmetric, leverage effect and long memory data. The Power ARCH or PARCH model (Taylor, 1986), a generalised model specification by Ding et al. (1993), incorporates the long memory of the market return and captures six other ARCH model extensions (Charles, 2010) in addition to testing the leverage effect:

where, the parameter γ captures asymmetry in the model (γ < 0, γ ≠ 0 ∼ asymmetric). δ > 0,|γ_i| ≤ 1 for i = 1… …n. the power term δ captures both standard deviation (δ = 1) and conditional variance (δ = 2). Following Charles(2010), Farag (2013) and Al-Khazali et al. (2017), we conducted the study as a special case ^[5].

The summary statistics of daily returns for the Indian stock exchanges are given in Table 1. The presences of positive mean returns are shown. Pakistan stock exchange and Bangladesh stock exchange show the highest and lowest mean return, respectively. Whereas, positive standard deviations in all the indices were found and BSE shows the highest and Nifty 50 shows the lowest standard deviation. The return series show a positive skewness in Nepal, Sri Lanka and India, this implies extreme positive are more likely to occur than extreme negative returns. The indices exhibit extreme kurtosis in the return series and the Indian stock exchange has a higher value of kurtosis. The JB test shows the return series are normally distributed as the null hypothesis is rejected at a 1% level of significance in all indices. Unit root test using augmented Dickey–Fuller (ADF) shows the distribution of data is around the normal value zero, and hence, the samples of the study are stationary. The non-parametric test using Kwiatkowski–Phillips–Schmidt–Shin (KPSS) failed to reject the null hypothesis on stationary or unit root test. The unit root test result shows a reliability estimation for the GARCH model. The presence of conditional heteroscedasticity using the ARCH-LM test and Breusch–Godfrey correlation LM test in the return series justified the model fit for GARCH and PARCH estimation.

We observed a summary statistic of religious holidays returns versus non-religious holidays in Table 2. The religious holiday mean return in the BSE stock index was found positive on five holidays: Buddha Purnima, Good Friday, Ganesh Chaturthi, MahaShivaratri and Gurunanak. The mean returns of religious holidays are higher than those of non-religious holidays in Buddha Purnima and Ganesh Chaturthi; the t-statistics show religious holidays are significantly different from other holidays at 10% and 5%, respectively. A significantly lower mean return was found in Good Friday, Muharram and Mahavir Jayanti at 5% and 10%, respectively. The mean difference between religious holidays and non-religious holidays in the Nifty 50 index shows a positive mean return on religious holidays except in Buddha Purnima, Dussehra, Good Friday and Holi, which are not necessarily significant. Ganesh Chaturthi’s mean holiday difference shows the highest return and Holi shows the least significant mean return. The Ganesh Chaturthi, Mahavir Jayanti and Muharram holidays mean returns are significantly higher than other trading days at 1%, 5% and 10%, respectively. The results indicate that the persistence of these religious anomalies during the religious festive periods shows a higher return and exhibits holiday anomalies in the market. Moreover, nonconformity and deviations from the typical behaviour of stock prices could be observed relative to religious holidays. This implies that the persistence of change in investor disposition during religious holidays in the market was found. The behaviour of investor during these religious holidays significantly influences investor mood and decisions in their trading strategy, as investors tend to exhibit a propensity to acquire stocks before holidays due to elevated spirits and holiday-induced euphoria.

Based on the ARCH-LM test and the Ljung-Box Q2 statistics in Table 1, the OLS model remains inadequate, as untreated ARCH effects remain in the volatility of returns. As a result, such volatility must be modelled. The religious holiday effect on the stock market can be better understood by taking a closer look at the specified model. For this, we examine the effect of religious holidays on the return and conditional volatility of the stock market indices using symmetric (GARCH) and asymmetric (PARCH) models. The estimated model requires appropriate specifications for both the return equation and variance equation, the GARCH (1,1) and PARCH (1,1) models were selected. GARCH (1,1) and PARCH (1,1) models performed the best fit for modelling volatility of stock return under study (Charles, 2010; Farag, 2013; Namugaya et al., 2014; Al-Khazali et al., 2017). The pre-religious and post-religious holiday effects on BSE and NSE are presented in Tables 3 and Table 4 respectively.

Table 3 presents the results of both the return equation and conditional variance equation in Panels A and B, respectively. The estimated coefficient result from GARCH and PARCH related to religious holidays in Dussehra, Ganesh Chaturthi Mahashivaratri and Muharram show a significant mean difference during the religious holiday. These include religious holidays for Buddhist festivals, Hindu holidays and Islam New Year. The estimate using GARCH model shows a stock market return increase during Ganesh Chaturthi and Mahashivaratri holidays. This means that pre-religious holiday returns one day prior and two days prior are significantly higher than that of other daily returns. On the other hand, Eid al-Fitr shows significantly negative mean return one day before the holiday. Moreover, no significant post-holiday effect was found on religious holidays except for Buddha Purnima and Ganesh Chaturthi. The same result was found for a pre-holiday using the PARCH estimated model, except for a negative significance in Dussehra. The post-holiday effect was found to be significant in Ganesh Chaturthi and Muharram at 1% and 5%, respectively.

The estimated conditional variance equation result shown in Panel B of Table 3. The volatility persistence measured by β is highly significant in GARCH model (0.89, 0.90) during the pre-holiday event as compared to that of PARCH (0.20, 0.18). Volatility cluster α measures short-run volatility and also shows high significance for symmetric and asymmetric estimations indicating that there is a significant impact shock on volatility. The leverage effect γ is significant at the 1% level. This provides evidence of asymmetric negative and positive market reactions towards news (holidays) in the Bombay stock market. The Power ARCH parameter δ shows highly significance at 1% level, indicating that the model is well specified and fits the distribution assumption of the market index. The volatility model with GED estimation in both analyses was less than 2 (GED < 2) in all windows. This implies the normal distributions under fat tailed of stock return are considered. The variances in religious holidays show a significant increase in volatility before the religious festivals at Christmas and a drop in volatility was found during the Diwali, Eid al-Fitr, and Mahashivaratri holidays ( Appendix 2). Moreover, we observe a drop in post-holiday volatility in Christmas and Diwali at a 1% level of significance. This finding is in line with Al-Khazali et al. (2017), which implies religious holidays affect certain aspects of individual lives. Investors are more likely to spend time practicing religion than trading stocks. Consequently, the drop in return volatility during the holiday season can be attributed to the change in investor behaviour that is caused by religious practices. This may lead to a change in the stock market trading activities.

The estimated coefficient results for religious holidays in NSE are represented in Table 4. The estimated coefficient result shows a positive significant pre-holiday effect on Hindu holidays: Diwali, Ganesh Chaturthi at 5% level, a Muslim holiday Eid-al Fitr and Muharram at the 5% and 10%, respectively, and Christian holiday Christmas show significantly higher returns two days before the event. The post-holiday effect was found significant on Christmas, Ganesh Chaturthi and Muharram at 1, 5 and 10%, respectively. This implies that the persistence of the religious holiday effect was significant in major religious holidays. The persistence of the religious holiday effect in the stock market appears to generate abnormal profits by trimming the time difference during the mandated religious holidays. However, the effect seems sporadic when considering the overall religious-mandated holidays.

The estimated coefficient variance equations are presented in Table 4. The volatility persistence β, volatility cluster α, leverage effect γ and Power ARCH parameter δ show the persistence of volatility during the religious holidays. The volatility during one day prior and after the event (GARCH = 0.88, 0.89) are slightly higher as compared to that of two-day trading volatility (GARCH = 0.84, 0.85). The Power ARCH effect model is statistically different from zero and higher on the trading day one day preceding and post-holidays (PARCH = 1.171, 1.601). This implies that there is an asymmetric stock market reaction to religious holidays on the NSE. The volatility effect on a religious holiday as shown in  Appendix 3 shows a drop (negative) in all 12 religious’ holidays before the mandated holiday except Christmas and Good Friday. Interestingly, we found the NSE stock exchange to be more volatile than the BSE stock exchange. The Hindu-mandated holidays in Diwali, Dussehra, Ganesh Chaturthi and Mahashivaratri and Muslim-mandated holidays in Eid-al Fitr and Muharram show a significant decrease in volatility on the first trading day. No significant volatility was found on the second trading day except for Christmas, Eid-al Fitr and Gurunanak Jayanti holidays. In a nutshell, we found an increase in return and a dip in volatility during the mandated religious holiday on the Indian stock exchanges. This is possibly due to a change in investor psychology, which affects the action and performance towards their trading behaviour in the stock (H1). Our finding supports earlier findings (Hilary and Hui, 2009; Canepa and Ibnrubbian, 2014; Atmaz and Basak, 2018; Tadepalli et al., 2021) that belief affects stock market return and volatility.

A study examining religious holiday effects on stock indices using long memory data can reveal potential stock anomalies, such as autocorrelation, weekday and month effects, world price movements due to recessions and financial crises. Such controls are considered to avoid bias in return estimation and volatility coefficients shifting the focus from the unpredictable part of returns (Chau et al., 2014; Bouri, 2015). Therefore, the ARMAX structure considers different autoregressive (AR), moving average (MA) and exogenous (X) orders. Incorporating an ARMA into the conditional mean equation accounts for possible linearity, which corresponds to the work of Westerhoff and Reitz (2005), Kyrtsou and Labys (2007), Al-Khazali et al. (2017) and Tadepalli et al. (2021). This suggests that neglecting this characteristic may undermine some of the dynamics underlying the model’s relationships. An ARIMAX structure allows exogenous variables to be incorporated into a model in this situation. Anggraeni et al. (2015) concluded that ARIMAX performed better than ARIMA when compared to the Akaike information criterion, mean absolute percentage error and root-mean-square error. Using ADF and KPSS tests in Table 1, the results of its ability to analyse variations in stock returns using dummy variables for religious holiday anomalies are considered reliable. The persistence of religious holidays anomalies over time is worth studying. For this, we divide the entire period into three sub-periods: before the AFC, between the AFC and the Global Financial Crisis (GFC) and after the GFC.

The estimated result in Panel A of Table 5 represents the pre- and post-holiday effect before AFC. We observed pre-holiday positive effects in returns for Buddha Purnima, Christmas, Guru Nanak Jayanti, Mahashivratri and Muharram, although these effects were not necessarily statistically significant. A significant decrease in mean return was observed in Diwali and Dussehra during the pre-holiday period in the Sensex index and in Buddha Purnima and Mahashivratri in the Nifty 50 Index prior to the holidays. Panel B of Table 5 shows pre- and post-holiday during the period between the AFC and GFC. The estimated result shows a significant drop in the mean return on the pre-holiday of Good Friday, Mahavir Jayanti and Muharram. An increase in mean return was found in Good Friday after the holiday events. One possible reason for this is the possibility of day of the week effect in the stock. Panel C of Table 5 shows the pre and post religious effect after the GFC, we found a significant decline in Gurunanak Jayanti holiday and an increase in the mean return in Ganesh Chaturthi, Mahashivaratri on trading before the holidays. The post-holiday effect shows a significant increase in Ganesh Chaturthi and a decline in Muharram holidays. Panel D in Table 5 represent the whole sample period. Out of the 12 holidays, we found a significant pre-holiday effect in Buddha Purnima and Mahashivaratri and post-holiday effect in Ganesh Chaturthi and Muharram. This implies that the persistence of religious holidays in the indices failed to reject the null hypothesis H3. The result contradicts the finding of Chia et al. (2015), Devakumar and Anuradha (2017) and Tadepalli et al. (2021) on holiday effect. Overall, religious mandated holidays had a negligible effect on the stock market’s return. Furthermore, the result show that a certain degree of religious holiday anomalies is found to be significant over time.

In this study, we investigate the behavioural perspective of religious festivals that influences the behaviour of stock returns and volatility in an emerging market. It examines whether religious practices influence investors decision towards investment during religious festivals. To anticipate these, we concentrate on two trading days prior and immediate trading day that led to the closing of stock trading. To capture the persistence of religious holiday anomalies on stock returns and volatility, we used symmetric and asymmetric GARCH frameworks for daily Indian stock indices. Moreover, the ARIMAX model was adopted to observe the seasonality effect in the trading days over a period from 1 January 1990 to 30 April 2022.

The primary evidence from our findings indicates that the existence and persistence of anomalies in stock return and volatility were found to have dissipated and varied subjected to specific religious holidays. Firstly, the pre-religious holiday effect with higher return was found in Christmas, Diwali, Eid al-Fitr, Ganesh Chaturthi and Muharram. This finding is consistent with earlier evidence (Fernando, 2005; Yuan and Gupta, 2014; Chia et al., 2015). Particularly, Christmas and Ganesh Chaturthi festival has both pre- and post-holiday effects. Secondly, the volatility persistence on religious holidays show higher in Christmas and the drop in volatility was found in Diwali, Eid al-Fitr, Dussehra, Gurunanak Jayanti, Mahashivaratri and Muharam. Overall, pre-religious holidays are less volatile than post-religious holidays. This result is consistent with the findings of Chan et al. (1996) and Bley and Saad (2010). The findings of our study align with those of Agarwalla et al. (2013), Al-Ahmad and Al-Ali (2016) and Harshita et al. (2019), who observed that the anomaly exhibited inconsistent and unstable characteristics. In other words, the return and volatility persistence due to religious holidays were found to be irregular and insignificant in most cases. An alternative model to validate the robustness of the findings was examined by dividing the sample into three sub-periods: pre-AFC, between AFC and GFC and post-GFC. A significant difference in mean return was found in Buddha Purniam, Diwali, Dussehra, Mahashivaratri in pre-AFC; Buddha Purnima, Good Friday, Holi, Mahavir Jayanti, Muharram during the period between AFC and GFC; Ganesh Chaturthi Gurunanak Jayanti, Maha Shivratri, Mahavir Jayanti, Muharram in post-GFC. We found a significant persistence of religious holiday anomalies overtime.

The findings of the study provide implications for retail investors, traders and regulators. Firstly, religious holidays offer investors more significant opportunities for abnormal returns compared to normal trading days. Retail investors seeking to capitalise on religious holiday anomalies should continuously monitor for the latest evidence, as these effects may diminish over time. The prevalence of religious anomalies in the market was observed across various holidays; investors can mitigate abnormal losses and achieve profits during these events. Secondly, traders have the opportunity to earn abnormal profit by strategically timing and understanding specific holiday related to Diwali effect and Muharram holiday to avoid periods of potential market volatility or illiquidity. Finally, regulators might implement adjusted liquidity management strategies around major religious holidays to ensure market stability. Investors may require additional information regarding holiday effects. It may be necessary to mandate disclosure of potential holiday impacts in company reports and require fund managers to report on holiday effect strategies. Central banks and other financial authorities may need to provide supplementary liquidity or implement specific interventions during these periods. Practically, there may be a need for periodic review and adaptation to incorporate new research findings into policy. Moreover, existing trading halt mechanisms may not adequately account for religious holiday effects.

For further research, it would be interesting to conduct a comparative analysis on a seasonal anomaly in emerged, emerging, frontier markets on how cultural, religious and social practices influence stock performance. In our study, pre- and post-holiday effects were evaluated for two days prior to and after holidays. Further research can focus on longer event windows. Further analyses of individual stock portfolios, sectoral stock indices, the gold market and derivative markets not subject to stock exchanges can be conducted across different markets.

Table A1 

Table A2 

Table A3