3. Research Design
3.1. Sample Selection
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3. Research Design
3.1 Sample Selection
This study uses financial statement data from the database compiled by the Taiwan Economic Journal (TEJ) Data Bank. The sample period consists of all trading dates spans from 1971 to 2017. In order to examine whether stock volume would be affected by religious superstition, which can also derive individuals’ sentiment, I use daily data from 1971, which is the beginning year of daily trading data in TEJ data bank, to 2017 over a 47-year period. This period contains the 1997–1998 Asian financial crisis, the 2001 oil price spike, the 2003 SARS outbreak and Iraq War, and the 2007–2008 U.S.
subprime crisis and global financial crisis.
3.2 Dependent Variables — Trading Volume and Stock Return
This study focus on all listed company at TWSE due to my main hypothesis assume that individuals are more likely to be affected by religious superstition and sentiment and then reflect on stock market. And a larger proportion of individuals are clustering in trading public companies rather than those in the OTC market. Restrict to database from TEJ Data Bank, the earliest daily trading volume that can be traced back to was January 5th, 1971, therefore, I use the sample period from 1971 to 2017. There are some earlier trading data omitted due to few companies going publish prior to this date, however, since a large proportion of corporations are listed later by the date, also the sample period is still long enough and contains both recession and thriving in Taiwan stock exchange market, the influence could be ignored. In addition, all transactions before the date of the first listed day of the company would be deleted.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Another restriction is that trading volume cannot be clearly divided into institutions and individuals since the fact that there is no any detail about transactions made by institutions, directly minus institutions’ trading volume will result in duplicated-cancellation. Nevertheless, a few literatures have suggested that institutional investors are more rational; moreover, the religious superstition discussed in this paper is based on traditional Chinese culture, which has much strong effect on individuals. On the basis of the two reasons, it can be expected that abnormal trading volume should be mainly caused by individuals.
As for stock return, the data managing process is almost the same as trading volume, which is daily basis. The only difference is the cancel of the data at first trading day since there is no return at the day. It is expected that negative returns could be found since the effects of ghost month, including possibly low sales, low entertainment activities, do exist and surprise the market even though it occurs every year. And the expected result shows that market is less than fully efficient.
3.3 Independent Variables
The principle in this paper is to examine whether religious superstition would affect trading volume and return. Hence, the main variable would be an event—the seventh month in Chinese lunar calendar, particularly, I would also check the first day of this month separately due to its distinction for the well-known common term “Kuei Men Kai”, which means the open of the gate to the Underworld. Table 1 lists the corresponding date of every seventh month in lunar calendar from 1971 to 2017. It is noticeable that dates vary year by year without a steady pattern, yet most locating in August and September. Without a fix standard reduce the possibility that the results may come from other events like ex-dividend date.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Additionally, due to this month accompanying with relative low investor sentiment, I also take into account other two factors related to corporate characteristics—company size and age—which are found that they have connection with sentiment (Baker and Wurgler (2006)). Intuitively, a smaller and older corporation should usually accompany with a slighter level of influence driven by superstition since a smaller one is deficient in trading volume originally as well as less well known among investors and thereby could mitigate the influence, and an older one could have a stable reputation among individuals. Firm size, measured as price times shares outstanding, has processed by log and based on daily frequency. Age, however, is based on months, so that daily trading data of the same month will have the same age independent variable.
In the seventh month, there are usually many taboos such as avoid having weddings, taking trips and so on, and could result in some specific industries being influenced more than other ones. Thus, industry is another independent variable. The classification of industries in my sample is referring to Chen, Chen and Lee (2013), which suggests that industries should be grouped into a relatively small number of broad categories, and use the Industry Classification Benchmark for the classification. The indices include: basic materials (MATS), consumer goods (GDS), consumer services (SVS), financials (FIN), health care (HEA), industrials (INDU), telecommunications (TELE), technology (TECH), oil & gas (OIL) and utilities (UTIL). Table 2 lists the components of each major sector.
3.4 Regression Design
Regressions are briefly show below. Both return and trading volume share the same regressions, therefore, I just list volume ones. The event contains “Kuei Men Kai” and
“Ghost Month”. The industry one to nine correspond with industry classification in table
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2, with health care industry being the benchmark, and test which industry displays a larger negative relationship with the inauspicious event.
𝑉𝑜𝑙𝑢𝑚𝑒',) = 𝛽,𝐸𝑣𝑒𝑛𝑡) (1) 𝑉𝑜𝑙𝑢𝑚𝑒',) = 𝛽,𝐸𝑣𝑒𝑛𝑡)+ 𝛽2𝑆𝑖𝑧𝑒',)+ 𝛽6𝐴𝑔𝑒',)+ 𝛽9𝐸𝑣𝑒𝑛𝑡)𝑆𝑖𝑧𝑒',)
+𝛽;𝐸𝑣𝑒𝑛𝑡)𝐴𝑔𝑒',) (2) 𝑉𝑜𝑙𝑢𝑚𝑒',) = 𝛽,𝐸𝑣𝑒𝑛𝑡)+ 𝛽2𝑆𝑖𝑧𝑒',)+ 𝛽6𝐴𝑔𝑒',)+ 𝛽9𝐸𝑣𝑒𝑛𝑡)𝑆𝑖𝑧𝑒',)
+𝛽;𝐸𝑣𝑒𝑛𝑡)𝐴𝑔𝑒',)+ 𝛽<𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦,,)+ ⋯ + 𝛽,9𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦C,)
+𝛽,;𝐸𝑣𝑒𝑛𝑡)𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦,,)+ ⋯ + 𝛽26𝐸𝑣𝑒𝑛𝑡)𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦C,) (3) The first regression purely demonstrates the relationship between trading volume and alternative variable of religious superstition. Next, firm size and age are added into consideration, together with their interaction, to see whether this superstition is still significant enough and if firm characteristics play a role. As mentioned before, I hypothesize larger and/or younger firms would be affected more by this event. To further discuss whether this superstition brings different degree to different industries due to some industries having more direct connect with individuals such as traveling, in the last regression, I divided all public companies into ten industries by the sorting-standard shown in table 2, and use it as another variable, which is processed by dummy. To deserved to be mentioned, I take the industry “health care” as benchmark since it has the least impact caused by the event among all industries, which is easily to figure out and compare the level of influence in other industries.
‧
Table 1: The first day of the seventh month in lunar calendar and the period of the month
Year Open of the gate to‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 2: Industry classification
Basic materials (MATS) Health care (HEA)
Chemicals Pharmaceuticals
Mining Biotechnology
Industrial materials Health care equipment and services
Forestry and paper General retailers
Consumer goods (GDS) Industrials (INDU) Automobiles and parts General industrials Personal & household goods Construction and materials
Food producers Aerospace
Food & beverage Engineering and machinery
Tobacco Industrial goods and services
Consumer services (SVS) Oil & gas (OIL)
Transport Integrated and exploration and production
Support services Equipment and support services
Retailers Technology (TECH)
Travel and leisure Information tech hardware and equipment Media and entertainment Software and computer services
Financials (FIN) Telecommunication (TELE)
Banks Fixed line telecommunication
Life insurance/assurance Mobile telecommunications Equity investment Utilities (UTIL)
Real estate Electricity
Gas, water and multi-utilities
Sources: Dow Jones US sectors and FTSE classifications in Datastream.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4. Empirical Results
4.1 Descriptive Statistics
Table 3 presents the results for the descriptive statistics of variables in regression models. The number of trading data is 3,828,200 for all listed companies from 1971 to 2017. Due to data processing, some samples were deleted in return, size and age. As can be seen in the table 3, median of trading volume is far less than mean trading volume, which demonstrates a disproportionate volume distribution among large and small companies.
The underneath part of table 3 contains descriptive statistics divided into ten industries. It is noticeable that trading volume varies from industry to industry, mainly resulting from the amount and the age of firm. The classification of ‘INDU’, ‘TECH’
and ‘MATS’ are the industries which have highest trading volume at 1,204,218, 885,638 and 623,939 respectively. A large proportion of companies in Taiwan is included in industrials and technology owing to their predominant position over others, which are also the foundation of Taiwan economic growth.
On the contrary, category of oil & gas and utilities have the least numbers of sample at 44,602 and 9,370 respectively, due to they are oligopoly in Taiwan—a small island without the possibility of oil exploration that is easily saturated in need of energy related resource. Furthermore, the category of ‘MATS’, ‘GDS’, ‘SVS’ and ‘INDU’ are relatively older industries with the means of their ages all beyond 165 months, almost 14 years. Specially, except the financial industry, which is rather large in size, other industries display a similar size in means, clustering around 8 after log process.
‧
N Mean Median Max Min S.D
Full Sample
Volume 3,828,200 3655 865 1,947,673 0 9850.82
Return 3,827,703 0.0009 0 144.0581 -5.7114 0.0842
Size(log) 3,827,269 8.6424 8.5543 15.6603 1.7918 1.4938
Age 3,828,147 152.4749 126 680 0 123.6185
12,122 299,093
N Mean S.D N Mean S.D
Volume 623,939 3,523 7,934.76 254,488 2,095.02 4,401.77
Return 623,863 0.000317 0.030157 254,459 0.000315 0.02403
Size 623,008 8.6735 1.42 254,488 8.6256 1.3846
Age 623,939 188.0722 138.6232 254,488 181.3498 134.8091
N Mean S.D N Mean S.D
Volume 445,302 2,376.88 5,691.73 156,073 10832 18950.8694 Return 445,255 0.005448 0.03583 156,045 0.000207 0.033139
Size 445,302 8.4996 1.4858 156,073 10.2992 1.41
Age 445,279 173.5261 135.76 156,073 139.6528 116.8698
N Mean S.D N Mean S.D
Volume 1,204,218 3,144.15 9,731.04 74,322 990.1258 1,936.32 Return 1,204,077 0.000486 0.138234 74,304 0.000189 0.022983
Size 1,204,218 8.3246 1.3567 74,322 8.0803 1.0493
Age 1,204,188 165.0205 129.6412 74,322 143.3282 150.3162
N Mean S.D N Mean S.D
Volume 885,638 4,692.41 11,946.79 130,248 3,446.74 5,668.59 Return 885,500 0.000066 0.046436 130,235 0.00048 0.072648
Size 885,638 8.8755 1.5124 130,248 8.8666 1.6879
Age 885,638 103.6129 72.2892 130,248 97.0123 67.19
N Mean S.D N Mean S.D
Volume 44,602 799.0394 2,219.16 9,370 1,218.70 2,450.63
Return 44,596 0.000251 0.021998 9,369 0.000556 0.042673
Size 44,602 8.3859 1.7619 9,370 7.5858 1.3933
Age 44,602 140.739 95.3507 9,370 116.4827 74.6381
OIL UTIL
Sample classified into ten industries
Number of trading days in “Kuei Men Kai ” Number of trading days in Ghost Month
MATS GDS
TECH TELE
INDU HEA
SVS FIN
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4.2 Regressions of Trading Volume
Table 4 reports the empirical results of trading volume with several regressions. On the top of the table, I regress firm size and age first in order to confirm that a larger and/or a younger company would have positive relationship with trading volume, with firm age has much less effect on it. The regressions in the table are corresponding with those in the chapter 3.4.
In first regression, I check merely the relationship between trading volume and the unlucky event and find that there is a negative and statically significant relationship between stock trading volume and event “ghost month”, also the first day of the month— “Kuei Men Kai”, at -208 thousand shares and -350 thousand shares respectively, which means the first day always brings a spike although it occurs every year. The second regression takes two dependent variables, firm size which is a log-processed variable, and monthly-based firm age into account. The results show that at the first day of the month, larger companies tend to be affected more by this inauspicious event, with a significant drop of 323 thousand shares in trading volume, yet firm age does not have obvious influence in this day. However, in the ghost month, a significant decline is more likely to be seen in larger and younger firms, with approximate decrease of 54 thousand shares in larger ones, though much less comparing with “Kuei Men Kai”, and drop of 0.288 thousand shares in younger ones, which is much slighter in comparison with firm size.
Due to the fact that every industry has its own characteristics that may have different impact on investors, the results of third regression suggest that at the first day of ghost month, there is still a significant drop in those firms with larger size, and volume of industries like financial, technology and telecommunication drop at that day, yet no
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
significance. On the contrary, in the ghost month, a reversal can be seen that firm age does be affected by this month firm size, though still slight. Meanwhile, firm size has constantly negative relationship with this inauspicious event. It is noticeable that all industries experience dropping during ghost month, and financial industry has significant the largest drop of 1150 thousand shares, also telecommunication has significant decline at about 308 thousand shares.
4.3 Regressions of Stock Return
Table 5 reports the empirical results of returns with several regressions. Since the figures of return are too small, the table shows in percent as unit to make the outcome clearer. All three regressions are the same as used in trading volume, except for changing dependent variable to stock daily return. On the top of the table, I regress firm size and age first in order to confirm that a larger and/or an older company would have significantly negative relationship with stock return, though both the figures are small.
In the first regression, return drops during the ghost month by 0.106%, while there is no significance at the first day. In the second regression with adding variables of logged-size and monthly-based age, the return still shows no impact on logged-size and age both at the first day and the whole month. In the last regression, we still can notice that no significant relationship with size and age. At the same time, there is no single industry shows noteworthy influence caused by ghost month and the first day of it, yet most industries display a negative relationship with it. The results implicate that though the unlucky superstition let investors lower their trading volume, the return almost does not be affected by this abnormal trading activity. In the meantime, investors would take firm characteristics—firm size and age—into consideration when they make trading decisions, however, those two factors affect nothing in return.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 6 presents further check of stock return in next five-day period and the subsequent month after the end of ghost month. The idea comes from Gervais, Kaniel and Mingelgrin (2001), which suggests that individual stocks whose trading volume is unusually large (small) over periods of a day or a week, tend to experience large (small) returns over the subsequent month. To fit this high volume premium theory to my study, since investors sentiment comes from religious superstition should not last too long intuitively, I also run a separate regression using five-day period. However, both the results are still mostly insignificant, except for a slight fall in the first regression, which is 0.0862% for 5-day and 0.0893% for one month rather than 0.108%. What is noteworthy is that there is a statistically significant decrease after considering firm size and age by 0.323% for 5-day, which is more than twice larger than the result for one-month (0.135%). The outcome shows that the investors’ irrational behavior originating from superstition surely exists for a relatively short time.
‧
Table 4:
Regressions of trading volume(1) (2) (3) (1) (2) (3)
‧
Table 5:
Regressions of stock returnUnit : %
‧
Table 6:
Regressions of stock returns over the subsequent 5 days and one monthUnit : %
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
5. Conclusion
Empirical studies are rare regarding the religious beliefs in eastern countries since the difficulties for searching alternative variables. This paper aims at filling the gap. As a result, the paper uses special event—ghost month—originating from Chinese religious beliefs and Chinese lunar calendar in Taiwan as an alternative variable. The results of regressions reject the null hypothesis, which supports the view that security markets are systematically influenced by investor psychology and superstition and sentiment coming from inauspicious day and argues for models of asset-pricing. In addition, rejection of the null hypothesis supports the hypothesis that financial markets are, to some degree, irrational and, thereby tossing doubt on the hypothesis that security markets reflect nothing but economic information.
Despite proving that religious beliefs and superstition do have influence on stock market, the result shows that there are only slight significant changes in return during the period, which means that investors’ irrational behavior does not definitely promise profits through speculation. Nevertheless, it demonstrates that investor should better avoid selling stocks during ghost month since the obvious dropping in trading volume implies it may be harder to sell. Furthermore, the return of five days after the end of ghost month displays a rather larger and significant consequence (-0.323%) and implies the religious beliefs and superstition truly leads to negative effect on stock market.
As for the constraint, there is a difficulty in distinguishing institutional investors and individuals. Though past studies suggest that individuals are more irrational than institutions and it would be a good explanation for that trading volume and stock return
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
anomalies mainly come from individuals. However, since they could reflect both directions with institutions usually have larger ability to affect them, it would be better to discuss this issue through a separate empirical analysis. The results might be more representative.
This paper uses data and the idea of religious superstition in seventh month in Chinese lunar calendar in Taiwan, a representative area in inheritance of Chinese traditional culture in East Asia, and the result is a part in the puzzle of the relationship between eastern religion and financial markets. Since there are a lot of religions in this region, also the diversified customs within a country, how to find other appropriate and reliable alternative variables is always a difficult question to discuss.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
References
Albaity, M. S., & Rahman, M. (2012). Gender, ethnicity, and religion and investment decisions: Malaysian evidence. Journal of Sociological Research, 3(2), 502-519.
Al-Khazali, O. (2014). Revisiting fast profit investor sentiment and stock returns during Ramadan. International Review of Financial Analysis, 33, 158-170.
Anderson, G. M. (1988). Mr. Smith and the preachers: the economics of religion in the wealth of nations. Journal of Political Economy, 96(5), 1066-1088.
Areni, C. S. (2008). (Tell me why) I don't like Mondays: Does an overvaluation of future discretionary time underlie reported weekly mood cycles?. Cognition and Emotion, 22(7), 1228-1252.
Baker, M., & Wurgler, J. (2006). Investor sentiment and the cross- section of stock returns. The Journal of Finance, 61(4), 1645-1680.
Baker, M., Wurgler, J., & Yuan, Y. (2012). Global, local, and contagious investor sentiment. Journal of financial economics, 104(2), 272-287.
Barber, B. M., Lee, Y. T., Liu, Y. J., & Odean, T. (2008). Just how much do individual investors lose by trading?. The Review of Financial Studies, 22(2), 609-632.
Barro, R. J., & McCleary, R. (2003). Religion and economic growth (No. w9682).
National Bureau of Economic Research.
Bergsma, K., & Jiang, D. (2016). Cultural New Year holidays and stock returns around the world. Financial Management, 45(1), 3-35.
Bhattacharya, U., Kuo, W. Y., Lin, T. C., & Zhao, J. (2017). Do superstitious traders lose money?. Management Science.
Block, L., & Kramer, T. (2009). The effect of superstitious beliefs on performance expectations. Journal of the Academy of Marketing Science, 37(2), 161-169.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Brown, P., & Mitchell, J. (2008). Culture and stock price clustering: Evidence from The Peoples' Republic of China. Pacific-Basin Finance Journal, 16(1-2), 95-120.
Chen, M. P., Chen, P. F., & Lee, C. C. (2013). Asymmetric effects of investor sentiment on industry stock returns: Panel data evidence. Emerging Markets Review, 14, 35-54.
Chui, A. C., & Wei, K. J. (1998). Book-to-market, firm size, and the turn-of-the-year effect: Evidence from Pacific-Basin emerging markets. Pacific-Basin finance journal, 6(3-4), 275-293.
Chung, R., Darrat, A. F., & Li, B. (2014). Superstitions and stock trading: some new evidence. Journal of the Asia Pacific Economy, 19(4), 527-538.
Cohen, R. B., Gompers, P. A., & Vuolteenaho, T. (2002). Who underreacts to cash-flow news? Evidence from trading between individuals and institutions. Journal of financial Economics, 66(2-3), 409-462.
Coutts, J. A. (1999). Friday the thirteenth and the Financial Times industrial ordinary shares index 1935-94. Applied Economics Letters, 6(1), 35-37.
Edmans, A., Garcia, D., & Norli, Ø. (2007). Sports sentiment and stock returns. The Journal of Finance, 62(4), 1967-1998.
Gervais, S., Kaniel, R., & Mingelgrin, D. H. (2001). The high- volume return premium. The Journal of Finance, 56(3), 877-919.
Goetzmann, W. N., & Zhu, N. (2005). Rain or shine: where is the weather effect?.
European Financial Management, 11(5), 559-578.
Golombik, J., Kumar, A., & Parwada, J. (2011). Does Religion Affect Stock Markets and Institutional Investor Behavior?.
Guiso, L., Sapienza, P., & Zingales, L. (2003). People's opium? Religion and economic attitudes. Journal of monetary economics, 50(1), 225-282.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Haggard, K. S. (2015). Stock returns in Chinese markets and lucky numbered days.
Managerial Finance, 41(9), 925-939.
Hilary, G., & Hui, K. W. (2009). Does religion matter in corporate decision making in America?. Journal of financial economics, 93(3), 455-473.
Hirshleifer, D., Jian, M., & Zhang, H. (2016). Superstition and financial decision making. Management Science.
Hirshleifer, D., & Shumway, T. (2003). Good day sunshine: Stock returns and the weather. The Journal of Finance, 58(3), 1009-1032.
Kaplanski, G., & Levy, H. (2010). Sentiment and stock prices: The case of aviation disasters. Journal of Financial Economics, 95(2), 174-201.
Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2003). Winter blues: A SAD stock market cycle. American Economic Review, 93(1), 324-343.
Ke, W. C., Chen, H., Lin, H. W. W., & Liu, Y. C. (2017). The impact of numerical superstition on the final digit of stock price. The North American Journal of Economics and Finance, 39, 145-157.
Kolb, R. W., & Rodriguez, R. J. (1987). Friday the Thirteenth:Part VII'-A Note. The Journal of Finance, 42(5), 1385-1387.
Kumar, A., Page, J. K., & Spalt, O. G. (2011). Religious beliefs, gambling attitudes, and financial market outcomes. Journal of Financial Economics, 102(3), 671-708.
Lip, E. (1992). Chinese numbers: significance, symbolism, and traditions. Times Books International.
Liu, W. H. (2013). Lunar calendar effect: evidence of the Chinese Farmer's Calendar on the equity markets in East Asia. Journal of the Asia Pacific Economy, 18(4), 560-593.
Loughran, T., & Schultz, P. (2004). Weather, stock returns, and the impact of