3. Data Description and Research Methodology

Size: px

Start display at page:

Download "3. Data Description and Research Methodology"

May Bennett
5 years ago
Views:

1 3. Data Description and Research Methodology In this chapter, we describe the data, hypotheses, and research methodology. In section 3.1, we first present the sources of data and provide description of the variables. Research hypotheses are proposed in section 3.2. We also demonstrate the methodology used in this paper in the last section. 3.1 Data description Sample Construction We obtained the official corporate name changes announcements from YSE, AMEX, asdaq, OTCBB, Toronto Stock Exchange (TSX), TSX Venture Exchange (TSX Venture), and EX Board (EX). Besides, we double check with the Mergent Events Data to make sure there is no missing company in our sample. Because each exchange begins providing its name changes data from different dates, it limits our sample period between January 2000 and December Then we filter their old names or new names with two words, oil and petroleum, and obtain 246 name changes events. Since the information of name changes is likely to be spread in the market before the official announcement made by the exchanges, we also search for the date of name changes news in Factiva ews. Daily stock price, trading volume, market value 19, benchmark index 20, and oil price are all sourced from the Datastream. We delete 69 companies due to price and volume data unavailable or insufficient in the period from day -15 to day +120 relative to the event day; hence 177 samples are included in the final samples. The Appendix reports the data sources utilized in this study. 19 Market value on Datastream is the share price multiplied by the number of ordinary shares in issue. 20 We use the Datastream orth America Oil & Gas price index as market index in the market-adjusted model. 14

2 3.1.2 Variable Construction These 177 sample companies are divided into subcategories by three criteria. Criterion 1 is whether the words, oil or petroleum, are added into or deleted from their corporate names, or even both in their old and new names. Criterion 2 separates them into major or minor name changes subsamples just like Bosch and Hirschey (1989) used. A major name change means the new name is entirely different from the old. A minor name change would not completely affect the company recognition and might reflect the continuity of the evolving corporate identity. Criterion 3 is whether their old and new names are both resembled to be related with resources 21 or only one. Moreover, we define three kinds of event date. First, Effective date is the date which corporate name changes to be announced effectively by exchanges. Second, ews date is the first available date of name changes news in the Factiva. Third, Early date is the early one between Effective date and ews date. In order to measure the market reaction for the announcement of name changes news, we use the Early date as our event date. We also divide samples into two subperiods so as to conduct the analysis of corporate name changes before 2004 versus after The reason we choose 2004-Q1 to be a separation for the period of steady and hot is showed in figure 3-1. It shows that both the crude oil price and the Datastream orth America Oil & Gas price index break the highest point between 2000 and 2003 and never turn back since 2004-Q1. In addition, all samples are grouped by their country; it depends on their exchange in the U.S. or Canada. 21 In this study, mining, gold, energy, resource, exploration and etc. are considered to be related with resources. 15

3 Figure 3-1 The quarterly crude oil price and the Datastream orth America Oil & Gas price index Q Q Q Q Q Q Q Q Q Q Q Q Crude Oil-WTISpot Cushing U$/BBL.AMERICA-DS Oil & Gas - PRICE IDEX 3.2 Hypotheses The main purpose of our study is to testify whether the corporate name changes in oil industry have significant influence on the stock prices. Thus, we can form the first hypothesis: ull Hypothesis 1 : The announcement of oil related name changes does not influence stock prices. We collect the samples in the U.S. and Canada, since previous research shows conflicting results between the U.S. and Australia (e.g. Karpoff and Rankine (1994); Josev, Chan and Faff (2004)), we hence examine whether the difference exists between the U.S. and Canada or not by the second hypothesis: ull Hypothesis 2 : The magnitude of abnormal return to announcement of oil 16

4 related name changes is the same between the U.S. and Canada. We refer to the dotcom study by Cooper, Khorana, Osobov, Patel and Rau (2005) and then divide samples into subcategories depends on whether they add or delete oil words and announce name changes in the hot or steady period. Hypothesis 3a : The magnitude of abnormal return to announcement of company s name changes with addition of oil is more positive than those with deletion changes. Hypothesis 3b : The magnitude of abnormal return to announcement of company s name changes with addition of oil is more positive than those with deletion changes, especially in the hot period. We also employ the classification of major versus minor name change like Bosch and Hirschey (1989) used to see if the major name changes appeals to investors more than minor name changes. Hypothesis 4 : The magnitude of abnormal return to announcement of company s name changes is more positive for major changes than those with minor changes. Because the skyrocketing price of crude oil, as illustrated in Table 3-1, we consider that firms change name with resource unrelated is more attractive to investors. Hypothesis 5 : The magnitude of abnormal return to announcement of company s name changes is more positive for resource unrelated changes. 17

5 Table 3-1 CAR sign prediction for additions/deletions/both and resource related/unrelated subcategories Old name ew name CAR sign prediction Additions Deletions resource related oil/petroleum + resource unrelated oil/petroleum ++ oil/petroleum resource related +/- oil/petroleum resource unrelated +/- Both oil/petroleum oil/petroleum +/- Another purpose of our study is to testify whether the corporate name changes in oil industry have significant influence on the stock trading volume, thus we can form the sixth null hypothesis: ull Hypothesis 6 : The announcement of oil related name changes does not influence the trading volume. According to previous research, we believe that the name changes of small companies will gain more benefits than big companies because they don t cost too much for preparing all the necessary operations. Hypothesis 7 : The magnitude of abnormal return to announcement of company s name changes is more positive for small companies. 3.3 Methodology Abnormal return 18

6 We employ the standard market-adjusted model 22 of event study methodology to examine the abnormal return. The abnormal return (AR) for the stock of firm i on day t, is calculated as follows ARit = Rit Rmt, (1) where R it is the return for firm i on day t and R mt is the market return using the Datastream orth America Oil & Gas price index for that day. We then compute the cumulative abnormal return (CAR i ) for firm i for various event windows from t = j to t = k as CAR i k = AR (2) t= j it And the mean cumulative abnormal return (CAR) of firms is CAR = i= 1 CAR The corresponding t-statistics that measure whether the CAR is significantly different from zero is tcar ( ) = CAR, Var( CAR i ) (4) where Var(CAR i ) is the variance of CAR i among firms. i (3) The significance of abnormal return is further tested using a non-parametric sign test. It examines whether the number of firms with positive CAR i exceeds 50%, under the null hypothesis of no abnormal return. This binomial statistic is more conservative than t-statistic test and does not require the assumption of normality. 22 It is described by Brown and Warner (1985). And also adopted by Lee (2001), Cooper, Dimitrov and Rau (2001), and Cooper, Khorana, Osobov, Patel and Rau (2005). 19

7 Z sign 1 w = 2, (5) where w is the percentage of firms with positive CAR i. We use two-sample t-test to testing whether two CARs in different subcategories are equal to each other or not. The t-statistic of two-sample mean cumulative abnormal return is t = CAR S CAR S1 S1 S S1 S2 S1 S2, (6) where CAR S1 ( CAR S1 ) is the CAR of subcategory 1(2), S ( S 2 S1 2 S 2 ) is the variance of CAR i in subcategory 1(2), and S1 ( S 2 ) is the number of firms in subcategory 1(2) Abnormal volume We follow the methodology in the study by Lee (2001). That is to compare trading volume in the event period to the non-event period 23, just like stock price event study techniques. First, we compute the abnormal volume ratio (AVR it ) for firm i on the single day T (T= -1, 0, +1) AVR it = V it 16, (7) Vit t= where V it is the trading volume of firm i on day t, and then calculate the mean 23 The non-event period in Lee s study is form -100 to -10. However, our non-event period is from -100 to -16 because our event period starts from

8 abnormal volume ratio (AVR T ) for firms on the single day T as AVR T = i= 1 AVR it (8) The corresponding t-statistics that measure whether the AVR T is significantly different from 1 on the single day T is t( AVRT ) = AVRT 1, Var( AVRiT ) (9) where Var(AVR it ) is the variance of AVR it among firms on the single day T. 21

A Note on Event Studies in Finance and Management Research. Ivana Naumovska INSEAD,

Working Paper Series 2015/84/EFE A Note on Event Studies in Finance and Management Research Ivana Naumovska INSEAD, ivana.naumovska@insead.edu Abe de Jong Erasmus University, ajong@rsm.nl This article