This paper is devoted to event study analysis of high-tech companies' presentations impact on the stock prices of the companies. First of all, there is a review on a previous research devoted to event study methodology and some innovation announcements research. This research is based on the sample of 49 presentations held by main technological market leaders and on the sample of 12 fully new products presentations. This paper tests two hypothesis the first is whether stock prices of the companies react positively to the presentations and whether new products have greater effect on the stock prices than the presentations related to the update of already existing flagship product of the company. The results show positive cumulative abnormal returns for all presentations and are significant at 99% level for event windows from -1 day to 3 days after event and longer. This study can be useful for private investors, or other market participants, according to the results of this research they can overperform market returns. In the end there are some recommendations for further development of this topic.

Keywords: Product presentation, high-tech, technology, event study, abnormal return, stock market reaction.

TABLE OF CONTENTS

• ABSTRACT
• INTRODUCTION
• LITERATURE REVIEW
• SAMPLE AND METHODOLOGY
• RESULTS OF RESEARCH
• CONCLUSION
• REFERENCES
• APPENDICES
• 

INTRODUCTION

· According to the chosen model of the expected stock price estimation there may be a need of daily market index return estimation;

· During the estimation window evaluations are done to calculate expected returns of the stocks. For this purpose,a model to describe daily returns before the event day should be chosen. The classical approach is based on simple historical price model, which assume constant return in time, according to Brown and Warner (1980, 1985)the returns gained with the help of this method are similar to results of more complex models. Another popular model of estimation is a market return model,in which abnormal return is calculated by subtracting stock daily return of the market index daily return. This model was described by MacKinlay (1997).One more model used for estimation of expected returns is a market model based on the ordinary least squares' method with an assumption that stock and market index daily returns have a linear relation. The model is evaluated in the estimation window (Graph 1). The choice of model and estimation window can play an important role for research because they can influence final results.

· Then according to the idea of research the period T₁ - T₀is chosen where the return of stock evaluates. The event day Ttakes place in event window. Time period including a few days before the event T - T₀ time is used when there is a need to find out whether the market had any information before the event or public announcement of any insider information, in case of airline disasters that period does not play role because it is an unpredictable event, time period T₁ - Tusually takes from few days up to one month depending on the research aim;

Размещено на http://www.allbest.ru/

Размещено на http://www.allbest.ru/

Figure 1. Time diagram for event study (MacKinlay, 1997)

· According to the chosen model daily expected returns for the event window can be calculatedand after that real returns after event compared to expected. The difference between real and expected returns is called abnormal return,ARand calculated for each day of the sample;

· After that,abnormal returns from event day and following are accumulatedin CAR (cumulative abnormal return) to highlight the event effect over random return fluctuations;

· According to the fact that in most cases of event study short event window is used, it is recommended to use not one event but the number of close events and find out the average for the whole sample;

· The last step is to determine whether the abnormal returns are significant using the significance tests.

Some research on the effect of presentations and announcements of a new product or a technology have already been done. They have proved that there is impact on the stock prices.

One of the first researches devoted to innovation and new technology effect on the stock prices was done by Dos Santos et al. (1993). The methodology of the event study was used in this research, they have checked the effect of announcements in the technological industry between 1981 and 1988, they did not get significant results both for the whole industry and for the segments of the industry. high tech stock investor product

According to the Sood and Tellis (2009) appearance of the new products in the market makes a great impact on the whole industry because it moves other companies to create new products and leads to high level of competitiveness sometimes new products can even create fully new markets.

One more research devoted to the stock price reaction to the innovations was done by Guseva and Rogova (2016) this research is based on the investments in the innovation technologies. The sample includes 84 announcements between 2010. And 2013 of Russian companies traded on the stock exchange. They achieved significant results and positive abnormal returns for the sample of 20 research projects and for the sample of 21 projects investments in which was based on the cooperation.

Xin (2010) says that technological development is very important for company growth and can create additional value.But any new product should appear at the right moment and in the right place, when customers really need this good and can create demand for it. She says that investors usually expect positive returns and better financial performance after announcement of the new product. In her study she examined stock market returns and impact on some financial indicators. She got significant average return of 1.71% on the day of event.

In another paper by Lee et al. (2000) they tell us that the company in the market which makes the first move and present new technologyor product gets competitive advantage. This paper is based on the sample which includes product announcements from 1975 to 1990 in brewing, telecommunication and computer industry. As a result of this paper the first movers had negative return on 3 cumulative abnormal return equal to -0.69.

One more paper by Yuan (2012) about product announcements effect on the stock prices in the technological sector. This paper has a sample of 50 events related to phone and smartphone announcements done by 5 companies in the period from 2007 to 2012. This paper tests different time periods to find out any significant returns. The results show that there is no significant effect in the short-term period but for the longer period from 35 to 40 days there is negative change of companies' stock price. In this paper author used his own judgment tochoose 5 companies, not all of this companies have mobile phones as big part of their business and that is the reason why not all of this product in the sample are equally interestingforthe investors.

SAMPLE AND METHODOLOGY

Table 1. Top 5 Smartphone companies' worldwide shipments (in millions) for 2017 and 2018 calendar years. (Source: IDC quarterly mobile phone tracker, January 30, 2019)

Then we checked whether smartphone business is an important part of the company and which part of revenue share does it brings to the company. The reason for this is that if company is not specified in smartphones and the revenue share of smartphone sales is less than 10% we can assume that new announcements in case of such companies do not change stock prices of the companies significantly because investors do not set high expectations for the product which is not the main part of the business. To find out which companies fit this research financial reports of these companies were downloaded.

Samsung revenue distribution between business units are provided in table 2.

Table 2. Samsung's revenue (in trillion KRW) distribution from 2016 to 2018. (Source: Samsung Electronics Announces Fourth Quarter and FY 2018 Results)

From table 2 we can say that Mobile communication unit is one the most important in Samsung's business and brings about 40% of revenue each year so their phone presentations can have significant effect for business.

Apple revenue distribution by products for the last years is presented in the table 3.

Table 3. Apple's revenue (in bln $) distribution between products from 2016 to 2018. (Source: Apple FY 2018 Results)

For Apple company their main product is iPhone with more than 60% of total annual revenue. Apple is the most discussed company nowadays its presentations have millions of viewers all over the world it is one of the most innovative companies nowadays, so consumers and investors are always waiting for the new announcements of the company.

Huawei is private company which is fully owned by employees. The methodology of event study is applicable only for public companies with shares listed at stock exchanges so despite of high level of Huawei sales it is not applicable for this research.

Xiaomi's financial results and business unit share are provided in table 4.

Table 4. Xiaomi's revenue (in bln RMB) distribution between business units from 2017 to 2018. (Source: Xiaomi FY 2018 Results)

From the table 4 we can see that Xiaomi's smartphone business unit is the most important part of its business with more than 65% of annual revenue. During the last years Xiaomi became big player at the high-tech market with a great amount of interesting products but the main part of their business are smartphones. Xiaomi nowadays offers high quality devices with a bit lower price than their competitors have.

Oppo is number 5 company by the amount of smartphones sold all over the world during last 2 years but it is a private company and we can't use it for event analysis because its stocks are not listed at any stock exchange.

After analyzing different companies of technology market, we have found only one more company which suits our requirements for the sample it is Microsoft. This company has diversified business oriented both for users and business (table 5).

Table 5. Microsoft's revenue (in mln $) distribution between products in 2018. (Source: Microsoft FY 2018 Results)

As we can see from the table Microsoft's main parts of business are Office products, Server products and Windows, these products are used in each office and at every home, so updates of them are very important for investors.

After selecting companies, we have to find information about dates of presentations and products presented there.Most of the information was found on official web-sites of the companies which we include in sample, but some old events were found as video recording of official presentations, these videos can provide trustful information about the date of event and products which were presented. We have collected information about product presentations of 4 top high-tech companies. All these presentations were held from 2007 to 2019. The sample includes 25 events for Apple from 2007 to 2019, 20 events for Samsung from 2009 to 2019, 9 events for Xiaomi from 2010 to 2019 and 9 events for Microsoft from 2009 to 2018. There are 66 events all in all.

The next step is to collect historical stock prices for each event. The daily stock prices should include the price on the day when the event took place, daily prices for each day in the estimation window and daily prices for event window. We should also download market index prices for the same periods for local markets of each company. For Microsoft and Apple we have downloaded S&P 500 index prices, for Samsung KOSPI (Korean stock exchange index) and Hang Seng index for Xiaomi (Hong Kong stock index). These indexes will be used to compare our companies' daily returns to market returns. All stock prices and index prices were downloaded from Yahoo! Finance, the website allows to public to download reliable time series of the prices for any period of time for public companies from different markets.

Estimation window is a time period in which model parameters of the model are estimated. The estimation window size differs very much in event study analysis, they can be from 30 up to 350 days everything depends on the effect of which event we want to estimate. In most research devoted to event analysis of innovations estimation window is between 60 and 90 days (Yuan, 2012).

In this study we will use the estimation window of 60 days before the event, because some of the technological companies have presentations once per quarter and to avoid layering of the other presentations effect it reasonable to use rather small estimation window. We have also collected the stock prices for 30 days after the event date for event window calculations further.

While collecting the data for sample it was discovered that Xiaomi was not a public company some time ago and it is possible to collect data only for 1 presentation held by the company. While collecting stock prices information about Samsung it was detected that company had several stock splits during the history of the company. Some of these splits were in the estimation window of events in our sample or in the event window, so we had to delete 3 events from the sample because these company actions had very strong impact on the stock price change. Whenwewereobtaining the information about Microsoft, we found out that the event window of one presentation was in the estimation window of another one, so we deleted one event and have total of 8. SpeakingaboutApple, originally, we had the sample of 28 events but deleted 3 of them because their estimation window was affected by the event windows of previous presentations.

After all these manipulations our final sample decreased from 66 product presentations to 49, which is enough for the event study. The sample is presented in the table 15 in the appendices.

After collecting all the data required to perform event analysis, we have to calculate returns. We need all real daily returns for the event window. First of all,daily returnsRtwere calculated,for each day of each event with the following formula (1), adjusted close prices were used for calculation:

Where P_tis the stock price onday t; - stock price on the day t-1; is a real daily return of the stock.

Then daily market returns Rmt were calculated, for each event based on the index prices of each company market and dates when event took place, with the next formula (2):

Where P_mtis the market index price on day t; is themarket index price on the day t-1; is a real daily return of the local market index. S&P 500 market index was calculated for events associated with Apple and Microsoft, for Samsung KOSPI (Korean stock exchange index) returns were calculated and Hang Seng index for Xiaomi (Hong Kong stock exchange index).

The next step is to find the expected return of each stock E_it. This can be done in different ways, in this research the market model was used to predict expected returns, this model is based on the assumption of a constant linear relation between stock returns and the returns of market index, the next formula is used for calculations (3):

Where is expected return of the icompany on the day t; is a return of the market index portfolio on the day t; and are estimated coefficients of the company i, is a zeromathematical expectation of the regression error.

The estimation of and coefficients of the market model for all presentation events is based on the estimation window of 60 days for each case. The estimation of the expected return for each day in the event window is done by the substitution of daily market index returns in the market model with estimated and coefficients, which are calculated by the ordinary least squares (OLS) method based on the estimation window data.

Then abnormal returns ARcan be calculated for each day before theevent and for the event window. Abnormal return is one of the most important parameters in the event study it shows unexpected changes in the stock prices of the company.It is calculated as a difference between real stock return and expected return.AR has the following formula:

Where is the estimation of excess return of i stock on the day t; is a real daily return of the i stock on the day t; is an expected return of the i stock on t day.

The next step is to calculate theCumulative abnormal return (CAR)for each event. CARis the sum of daily abnormal returns in the event windowfrom the first dayof the event window to the last day and used to evaluate an effect of event or news on the stock price of the company.CAR can be calculated by the following formula (5);

Where is the estimation of sum of the over performed stock returns of the company i; is an estimation of over performance of the stocks of company i on the day.

According to this research our sample consists of the 49 product announcements for the all events and 12 presentations for new products, to test this samples correctly the average abnormal returns (AARs) and corresponding to it cumulative average abnormal returns (CAARs) should be calculated, according to the event study methodology described by Kothari and Warner (2007).

Average abnormal returns (AARs) can be calculated by the following formula (6):

Where is an average abnormal return on day t; - estimation of abnormal return of the stock i on the day t; N - is a number of events in the sample.

Then cumulative average abnormal return should be calculated (CAAR) for both samples, it is calculated as a sum of average abnormal returns (AARs). The formula for CAAR calculation is presented below (7):

Where CAARis an estimation of the accumulated average abnormal returns; is an average abnormal return on the day t.

When all calculations are done it is possible to analyze and estimate the impact of different kinds of events on the stock prices of the companies.

The last stage of the event study analysis is the check of the hypothesis about significant impact of the event on stock prices of the company. All tests for the event study analysis can be divided in two groups parametric and non-parametric. The main difference between this two groups of tests is that parametric tests are applicable only for normal distribution of returns. A lot of research includes tests from different groups. That approach can additionally check the absence of the outlier effect on the research results.

In this research we will also use two different tests to check our hypothesis one parametric test and one nonparametric. Some tests are presented in the table below (table 6).

Table6. Some tests for the hypothesis check. (Source: significance tests for event studies; eventstudytools.com)

Cumulative average abnormal returns will be tested in this research with suitable for the sample testingby Patell (Petell, 1976) and Wilcoxon (Wilcoxon, 1945) tests. To calculate these tests, we will use XLSTAT addon for Microsoft excel.

The aim of this work is to check the two following hypotheses:

· Presentations of high-tech companies lead to higher daily stock returns of the company. We assume that when the new product is presented it is the best at the market or about the same level (according to the sample requirements) so investors are excited about new product and can expect high level of sales and good financial result, which in turn will lead to the increase of stock prices;

The null and alternative hypothesis have the following form:

New product or device presented has greater impact on stock prices than the update of previous one. We assume that new products can have greater impact because new products can significantly increase company's revenueby entering to the new markets and creating demand among new customers or additional product demand among loyal ones.The null and alternative hypothesis have the following form:

For each. of the hypothesis there are two possible outcomes:

· Hypothesis H₀ is rejected and hypothesis H₁ is accepted if testsresults are significant;

· Hypothesis H₀ is not rejected and hypothesis H₁ is rejected if the results of the tests are not significant.

RESULTS OF RESEARCH

Figure 3. Average abnormal returns from day of event to 30 days after, in % for Samsung.

Figure 4. Average abnormal returns from day of event to 30 days after, in % for Microsoft.

Figure 5. Average abnormal returns from day of event to 30 days after, in % for Xiaomi.

The graphs of the average abnormal returns do not provide enough information about the returns of the companies, it is rather difficult to estimate how did stocks perform after the presentation and make conclusion whether presentation of the company affects positively on the stock prices of the companies or not.

Then the average abnormal return for all companies was calculated all AARs for the days from 0 (the day of presentation) to 30 days after public announcement of the product are presented on the figure 6:

Figure 6. Average abnormal returns from day of event to 30 days after, in % for all companies.

From this graph we can see that after presentations companies had mostly positive daily abnormal returns it means that stocks of these companies overperformed themselves on the most days. The highest average abnormal return for all companies equals to 0,5% and is on the 13^th day after presentations the lowest abnormal return is on 23^rd day after the event date and equals to -0,33%

Then the average abnormal return for all companies new products was calculated all AARs for the days from 0 (the day of presentation) to 30 days after public announcement of the product are presented on the figure 7.

Figure 7. Average abnormal returns for new products from day of event to 30 days after it, in % for all companies.

From figure 7 we can find out that new products have positive effect on the stock prices of the companies. Average abnormal returns are positive for 20 out of 31 days with highest average abnormal return of 1,13% on the 30^th day after the event and the lowest abnormal return of -0,69% on the 25^th day after presentation took place.

Then the cumulative average abnormal returns (CAARs) were calculated for different event windows to find out how stock prices behave during different time periods within one month. Cumulative average abnormal returns were calculated for the both samples: all presentations and new products presentations. The next event windows for CAARs calculations were chosen:

· From one day before the event to one day after the date of the event;

· From one day before the event to 3 days after the date of the event;

· From one day before the event to 10 days after the date of the event;

· From one day before the event to 15 days after the date of the event;

· From one day before the event to 30 days after the date of the event.

The results for CAARs are presented in the table 9 below:

Table 9. Cumulative average abnormal returns for different event windows.

According to the cumulative average abnormal results it is possible to make a conclusion that product presentations have positive effect on the stock prices of the companies. For the sample of all products obtained results are positive for all event windows used in this research and these CAARs have a positive dynamics after presentations it growth from 0,57% for the period of 1 day before the event date and 1 day after the event to 2,50% for the event window from 1 day before event to 30 days after the event. The diagram for the whole sample (49 events) cumulative average abnormal return is presented below, figure 8.

Figure 8. Cumulative average abnormal returns for the sample of all product announcements (49).

The situation with new products is partly different. New products have negative cumulative abnormal return for the short event windows (-1 day to +1 day and -1 day to +3 days), but then new products sample greatly overperforms the expected return from 2,26% on the event window from 1 day before the event to 10 days after and up to the 6,82% for the event window from 1 day before the event to 30 days after the event took place.

It can be explained by the following idea: in most cases there are some rumors and information leaks before the event date and investors can set their expectations on the future of the company and its products based on leaks, but when presentations take place real picture can be different from the rumors and leakages and new products cannot meet investors' expectations in this cases stock prices of the company will slightly decrease. After the presentation date in few days the new product sales can begin, preorder can be launched and first reviews on the new product appear, when the first after presentation appears it can affect stock prices significantly. The other possible reason for this kind of stock price change is that it is not always possible to estimate expectations on the totally new product on the day of announcement, people need some time to understand something new and also then, again, after the first information about pre-orders, orders or first reviews appear, the investors can change their expectations.The diagram for the sample of new products presentations (12 events) cumulative average abnormal return is presented below, figure 9.

Figure 9.Cumulative average abnormal returns for the sample of new product announcements (12).

Table 16 in appendices presents calculations for the sample of fully new products.

The last step of the event analysis methodology is significance testing. To check hypothesis and we have checked different event windows with significance test. To accept the null hypothesis the statistics should be less than table value. After calculating all significance tests, the following results were obtained.

Table 10 contains information about testing the event window from 1 day before the event to the 1^st day after the date of the event.

Table 10. Significance testing for event window (-1 to 1 day).

There is no significant results in the tests done for event window of -1 day to day 1, that means that both null hypothesis are accepted and there is no cumulative average abnormal return for presentations of high-tech companies in the period from 1 day before the event to one day after the event takes place.

Table 11 contains information about testing the event window from 1 day before the event to the 3^rd day after the date of the event.

*- means significance at 10% level

Table 11. Significance testing for event window (-1 to 3 day).

For the event window from -1 to 3 days after the event there are significant results at 10% level for the sample of all events, it means that the null hypothesis for the sample of 49 events can be rejected at 10% level of significance and we can conclude that there are cumulative abnormal returns with 90% probability. There are no significant results for new products presentations in the same event window and null hypothesis for the sample of new products is accepted.

Table 12 contains information about testing the event window from 1 day before the event to the 10^th day after the date of the event.

*- means significance at 10% significance level;

** - means significance at 5% significance level;

*** - means significance at 1% significance level.

Table 12. Significance testing for event window (-1 to 10 day).

For the event window from -1 to 10 days after the event there are significant results at 1% level for the sample of all events, it means that the null hypothesis for the sample of 49 events can be rejected at 1% level of significance and we can conclude that there are cumulative abnormal returns with 99% probability. There are also significant results for new products presentations in the same event window and the hypothesis for the sample of new products is rejected at 1% level according to the Patell's test and at 5% according to Wilcoxon signed-rank test.

Table 13 contains information about testing the event window from 1 day before the event to the 15^th day after the date of the event.

*- means significance at 10% significance level;

** - means significance at 5% significance level;

*** - means significance at 1% significance level.

Table 13. Significance testing for event window (-1 to 15 day).

For the event window from -1 to 15 days after the event there are significant results at 1% level for the sample of all events, it means that the null hypothesis for the sample of 49 events can be rejected at 1% level of significance and we can conclude that there are cumulative abnormal returns with 99% probability. There are also significant results for new products presentations in the same event window and the hypothesis for the sample of new products is rejected at 1% level according to the Patell's test and at 1% according to Wilcoxon signed-rank test.

Table 14 contains information about testing the event window from 1 day before the event to 30 day after the date of the event.

*- means significance at 10% significance level;

** - means significance at 5% significance level;

*** - means significance at 1% significance level.

Table 14. Significance testing for event window (-1 to 30 day).

For the event window from -1 to 15 days after the event there are significant results at 1% level for the sample of all events, it means that the null hypothesis for the sample of 49 events can be rejected at 1% level of significance and we can conclude that there are cumulative abnormal returns with 99% probability. There are also significant results for new products presentations in the same event window and the hypothesis for the sample of new products is rejected at 1% level according to the Patell's test and at 1% according to Wilcoxon signed-rank test.

According to the results of the tests completed we can sum up that there is no significance for cumulative average abnormal returns for any presentations in the event window -1 to 1 day. The significant results at 10% level were gained for all presentation (49) in the window -1 day to 3 day. And significant results for both samples at 1% level in the event windows -1 to 10 days, -1 to 15 days and -1 to 30 days. According to these results we can conclude that investors are interested in new product announcements and the market is sensitive to such events.

CONCLUSION