
- •Contents
- •Introduction
- •Context
- •Graph 1 – Situation X da – company a’s demand dc – competitors’ demand
- •Financial Data Mining Literature
- •Picture 1 – Word ratio and stock performance visual comparison (Nagar and Hahsler 2012, pp. 15)
- •Competitors’ Effect Literature
- •Critical Analysis and Gap in Research
- •Proportion of competitors’ influence on the stock return
- •Proportion of the news’ effect on competitor(s)’ return
- •Justification
- •Industry return as control variable
- •Controlling for the market
- •Graph 3 – Agilent Technologies squared returns
- •Ratio Analysis
- •Competitors’ Influence - Regression
- •Industry return as control variable
- •Influence on Competitors – Regression
- •Regression analysis results
- •Industry return as control variable
- •Conclusion
- •Proportion of competitors’ influence on the stock return
- •Industry return as control variable
- •Proportion of the news’ effect on competitor(s)’ return
- •Possibility for practical use
- •Limitations of the research
- •Recommendations for future study
- •References
- •Appendix
Controlling for the market
However,
one may suggest that a high or low return can arise if generally
investors are buying or selling stocks on the day. The reason for
that might be an economy-wide news piece. Therefore, to identify the
abnormal company-related returns we also subtract S&P 500 index
return from the stock return. Then we shall have a return of the
company
less the overall investors’ buying attitude
– the adjusted company return
.
This would make more sense as it removes market-wide news. Moreover, most of the sampled stocks exhibit appreciation during the sample period, which means that a negative return is less likely to happen than positive (or that the positive returns’ scale is larger), therefore even if a piece of news led to significant positive return for company A (say, 7%) it may only make its competitor to grow slower than usual (say, 0.2% instead of 1.2%). S&P 500 index itself grew by 44% during the sample period and if we subtract it from the company returns we may identify the “losers” and “winners” as well as “better than index”, or positive returns and “worse than index”, or negative returns. In the previous case, say the S&P rose by 2.5%, then company A performed (7-2.5)=4.5% better than the index, while its competitor performed (0.2-2.5)=-2.2% worse than the index. After subtraction it is easier to see whether a news article affected both the company A and its competitor.
Date |
INITIAL RETURNS |
|
|
|
ADJUSTED |
||||
AGILENT TECHS. |
TERADYNE |
THERMO FISHER SCIENTIFIC |
S&P 500 |
AGILENT TECHS. |
TERADYNE |
THERMO FISHER SCIENTIFIC |
|||
28/11/2011 |
0.039 |
0.036 |
0.026 |
2.88% |
0.97% |
0.73% |
-0.26% |
||
29/11/2011 |
-0.004 |
-0.002 |
0.011 |
0.22% |
-0.62% |
-0.47% |
0.85% |
||
30/11/2011 |
0.068 |
0.108 |
0.025 |
4.24% |
2.60% |
6.58% |
-1.75% |
||
01/12/2011 |
0.008 |
-0.007 |
-0.004 |
-0.19% |
1.04% |
-0.48% |
-0.25% |
||
02/12/2011 |
-0.028 |
-0.016 |
-0.011 |
-0.02% |
-2.82% |
-1.56% |
-1.04% |
||
05/12/2011 |
0.008 |
0.025 |
0.005 |
1.02% |
-0.21% |
1.45% |
-0.51% |
||
06/12/2011 |
-0.002 |
0.009 |
0.009 |
0.11% |
-0.33% |
0.78% |
0.74% |
||
07/12/2011 |
-0.009 |
0.001 |
-0.004 |
0.20% |
-1.07% |
-0.05% |
-0.63% |
||
08/12/2011 |
-0.063 |
-0.019 |
-0.037 |
-2.14% |
-4.17% |
0.29% |
-1.53% |
||
09/12/2011 |
0.026 |
0.054 |
0.008 |
1.67% |
0.94% |
3.71% |
-0.90% |
||
12/12/2011 |
-0.046 |
-0.029 |
-0.011 |
-1.50% |
-3.13% |
-1.37% |
0.38% |
||
13/12/2011 |
-0.017 |
-0.034 |
-0.01 |
-0.87% |
-0.86% |
-2.54% |
-0.11% |
||
14/12/2011 |
-0.007 |
-0.029 |
-0.021 |
-1.14% |
0.48% |
-1.77% |
-0.98% |
||
15/12/2011 |
0.016 |
-0.001 |
0.009 |
0.32% |
1.24% |
-0.40% |
0.59% |
Table
2 – Adjusting
data for market returns
The “Adjusted” part of the table (table 2) represents companies’ daily returns less S&P return. On 28th of November, 2011 all three companies showed to have exceptional returns. However, the S&P index itself showed a return of 2.88% on that day as well and therefore we may not conclude that all three stocks had positive news on that day. After the adjustments the companies show relative returns ranging from -0.26% to 0.97%, which are not that large. However, not all events are hidden with the adjustments. On 8th of December, 2011 (table 4) we see that even after the subtraction of S&P’s return Agilent Technologies stock has a significant negative return of -4.7% and other outliers still remain.
However, comparing the adjusted returns themselves is not comfortable. Some returns are negative and some are positive. So I first take the square of the returns, and then find the outliers. Squaring the returns will magnify the heteroscedastic properties of the returns as well as make it easier to compare negative and positive returns.