International Journal of Financial Studies

Article The Relationship between Life Cycle and Korean Stock Market Performance

BokHyun Lee

Korean Intellectual Property Office, Government Complex Daejeon Building 4, 189, Cheongsa-ro, Seo-gu, Daejeon 35208, Korea; [email protected]

 Received: 5 August 2018; Accepted: 25 October 2018; Published: 29 October 2018 

Abstract: Through the three industrial revolutions, technology has enabled rapid changes in society. In a capitalist society, capital is invested where there is utility, for example, economic benefit. We intend to determine that the stock price of a company that uses a particular technology will change with the life cycle of the technology in question. Specifically, we filtered companies that mainly deal with augmented reality and are listed in Korea’s KOSDAQ market. We grouped these companies based on detailed that constitute augmented reality. We used the event study method to calculate the stock returns against a benchmark. As a result, in the “Peak of Inflated Expectations” stage, the portfolios of all companies using augmented reality generally show higher returns than the benchmark. However, it is difficult to ascertain whether a return generated based on one of the detailed technologies that make up augmented reality is higher or lower than that of the benchmark. During the “Trough of Disillusionment” phase, there was neither a consistent trend of cumulative abnormal returns (CAR) nor buy-and-hold abnormal returns (BHAR). However, during this stage, there was a positive correlation of average BHAR and average abnormal returns between the entire sample’s portfolio and each detailed technology firm’s portfolio.

Keywords: ; cumulative abnormal returns; buy-and-hold abnormal returns; augmented reality

JEL Classification: G10; G11; G17; O30

1. Introduction Although there are many factors that affect the economic life span of technology, they can be divided into technical and market factors (Park et al. 2014). These factors affect the value of companies in terms of stock prices. Technology life cycles also analyze technologies based on technical and market factors. A analyzes the current state of a specific technology and considers factors, such as the development stage, interest in the market, market size, and so on. Technology undergoes a process of creation-growth-extinction over time, and its value in society continues to change during this process. Concepts that represent this technology lifecycle include (1) an S-curve, (2) a Hype Cycle, and (3) a Technology Adoption Life Cycle. These life cycle theories assume different shapes when represented graphically, but they have similar interpretations. That is, at first, the speed of generation is slow and user awareness is low. However, the speed of development becomes faster and more recognizable to users as time passes. Ultimately, it reaches a certain limit and a new life cycle begins. In a capitalist society, the value of an object is represented by capital. As of 2008, four of the top 10 global market capitalization stocks were energy companies. However, in 2016, six out of the top 10 of these stocks were technology companies (Jung and Kim 2017). In 2000, Cisco Systems, a network equipment company, was the leading global market capitalization IT stock. However, in 2015, Internet of Things companies, such as Alphabet, accounted

Int. J. Financial Stud. 2018, 6, 88; doi:10.3390/ijfs6040088 www.mdpi.com/journal/ijfs Int. J. Financial Stud. 2018, 6, 88 2 of 22 for a large portion of the global IT market. This suggested changes in society that are based on market capitalization and capital movements depending on which technology is mainly used by IT companies. Based on these ideas, we will explore the relationships that exist between each stage of the technology life cycle and the stock price of each firm related to the technology. If each technology is widely known and it creates a lot of expectations or earnings, its stock price rises more than the benchmark, but if the technology is not known to be useful or profitable, it may likely be less profitable than the benchmark. This contributes to the related literature in several dimensions. First, we use technology life cycles as a means of measuring technological progress. The investigation of the relationship between technology life cycles and stock prices provides new insights into the analysis of a dynamic relationship between technology and stock prices. It is more appropriate and practical to use technology life cycles as background data in order to invest in stocks than to use industry life cycles, because such cycles have a much longer term than technology life cycles. Second, we analyze the stock prices of a group of the total firms dealing with a technology mainly and only a small portion of the group. Portfolio construction can be improved based on the result and it may be possible to imitate the performance of the total group with only small investment. In addition, such a relationship can be utilized when the manager of a company wants to increase the value of the company through the stock market. From the investor’s point of view, this can be used to obtain more earnings. Finally, investment professionals, such as brokerage firms, could potentially make use of these relationships in creating new financial products or investment strategies.

2. Literature Review Recent empirical studies have suggested a dynamic relationship between technology and stock prices. These studies used various means to understand the current state of technology, one being to use information on patents and scientific papers. Park et al.(2011) divided small and medium enterprises (SMEs) that are listed on KOSDAQ by company level and studied the influence of the R&D intensity (R&D expenditure/sales) and patents on the share price of SMEs. They used data from manufacturers listed on KOSDAQ from 2000 to 2009 and data on Korean patent applications. They demonstrated that the relationship between R&D intensity and stock return was statistically significant and R&D intensity had a negative impact on stock return. In addition, they found that the link between the number of patent applications/registrations and stock return was not statistically significant at the customary significance levels. Kim and Jung(1995) demonstrated the relationship between patent application disclosure and stock price using cumulative abnormal returns (CAR). They used the daily stock data listed on the Korea Exchange (KRX) and the data on Korean patent applications from 1989 to 1994. The author analyzed a certain period of time before and after the announcement date and obtained a CAR of about 6%, which was statistically significant. When the expenditure on technology development is large or the research is conducted with other entities, such as firms or laboratories, the effect of patent application disclosure is improved. Similarly, Daizadeh(2007) studied the relationship between the number of United States (US) patents, the number of media and publications related to patents, and US stock indices (Dow, etc.) from 1970 to 2004, and found a strong positive Spearman correlation between them. Nicholas(2008) studied the changes in the market value of patent assets in the 1910s and 1920s and the changes in the market value of patent assets before and after the Great Depression. To carry out this study, the author used the share price of the firms, which had a patentable asset from the 1910s to the 1930s, financial information, the number of patents of the corporation, and the number of times a registered patent between 1975 and 2006 cited the company’s patent. The market value of patent assets in the 1920s was much higher than the value of the patent assets in the 1910s. In particular, high-quality patents had a statistically significant effect on the rise in market value. This effect led to a rise in the stock market in the 1920s. In relation to the Great Depression, there was no evidence that intellectual property helped to gain excess profits from December 1925 through February 1928, but from March Int. J. Financial Stud. 2018, 6, 88 3 of 22

1928 to September 1929, the intellectual property helped to earn excess profits. However, the author showed that, since October 1929, when the Great Depression occurred, patent assets did not bring significant excess returns, even though firms acquired important patents. Thomas(2001) evaluated company stock price using patent and R&D indicators of the specific company, based on the notion that the quality of a company’s technology is reflected in its patent portfolio. The author used the data of all US firms listed on US stock exchanges at the end of 1999 that had been granted at least 50 US patents over the previous five years and the patent data on an annual basis for the period from the end of 1990 to the end of 1997. Patent indicators that evaluate the stock price of companies include the number of patents, the rate of increase of patent registrations, the influence of registered patents, and the degree of reference to the latest scientific papers, and so on. R&D intensity was used as a index. Thomas(2001) predicted market-to-book (MTB) value that is based on patents and R&D data and defined the value as the technology MTB. A company whose actual MTB value was smaller than the technology MTB was defined as an undervalued company. Thomas(2001) found that over 80% of the undervalued companies at the end of 1990 had a higher MTB two years later, and the most undervalued companies showed a higher yield than S&P500 and NASDAQ from December 1990 to December 1999. Baek(2005) selected Samsung Electronics to analyze the correlation between the number of patent applications/registrations and share price. The author used Korean patent application data and the stock data from 1982 to 2001. He found that it is difficult to argue whether there was a correlation between the number of patent applications/registrations and share price related to only one company or not because the stock price of one company is affected by so many factors. Chadha and Oriani(2010) studied the relationship between R&D and stock prices in a country where Intellectual Property Rights (IPR) protection is weak. The authors used the data of manufacturing firms traded on the Bombay Stock Exchange from 1991 to 2005. They demonstrated that R&D in a country with weak IPR protection had more positive influence on stock prices than R&D in a country with strong IPR protection. The results indicated that domestic firms’ R&D investment had a greater impact on stock prices than that of foreign firms’. Mazzucato(2006) reviewed several papers on the relationship between and the stock price and indicated that studies showed that rapid changes in patents and research and development expenditure significantly changed the market value of the company. In particular, the citation weighted patents had more effect on the market value of the company. The author reviewed a study that demonstrated that the relationship between innovativeness (based on R&D expenditures divided by sales and other patent related indicators) and industry level stock return volatility did not have a coherent pattern while using the quarterly return data of 34 industries from 1976–1999. The study also found that firms with higher R&D intensity (R&D expenditures divided by sales) had higher idiosyncratic risk with the monthly firm-level returns and the quarterly firm-level R&D intensity data of five industries from 1974–2003.1 It was assumed that the result of industry level analysis in Mazzucato(2006) was due to the premise that innovation was fixed. In addition, many studies have investigated the relationship between industry life cycle, another means of demonstrating technological progress, and stock price. Pástor and Veronesi(2009) developed a model showing stock prices of innovative firms during technological revolutions. They used monthly prices of steam-powered railroad stock on the NYSE from 1830–1861. They found that stock prices of innovative firms bubbled during technological revolutions due to good news about the new technology’s productivity, but they fell as time went by. They demonstrated that the bubbles in stock prices were not able to be observed ex ante. Mazzucato and Semmler(1999) investigated market share instability and stock price volatility during the industry life-cycle. They found that there is greater market share instability and higher

1 Idiosyncratic risk was defined as the ratio between the volatility of firm-level returns over the volatility of market level return volatility. Int. J. Financial Stud. 2018, 6, 88 4 of 22 stock price volatility in the first phase of each firm’s . They demonstrated that the market share was more stable and the stock price volatility was lower as the industry life cycle matured. Mazzucato (2002) investigated the results of Mazzucato and Semmler(1999) in more detail. Mazzucato(2002) analyzed the market share change and financial volatility in the early phase of the life cycle of an old high-tech industry (the US automotive industry from 1899–1929) and a new high-tech industry (the US PC industry from 1974–2000). The author used firm- and industry-level the annual sales data and stock prices of each firm. The PC industry and the automotive industry showed the same results in terms of stock price volatility but differed in terms of market share instability. Stock prices were most volatile when technology changes its market share the most. and new entry caused market share instability in the US automotive industry. However, this did not occur in the US PC industry until the 1990s. Similarly, Mazzucato(2003) showed that stock price changes in the early phase of the industry life cycle were significantly impacted by turbulence in the market structure, and stock price dynamics in the mature phase were affected more by firm- and market- level fundamentals with a data similar to the data used in Mazzucato(2002). According to Mazzucato(2006), research indicated that by new technologies have a negative impact on capital due to existing technology, resulting in a decline in average industry stock prices. She reviewed a paper that showed that the degree to which stock price was more volatile than the value that was calculated by the efficient market model peaked during the Second and Third Industrial Revolutions due to herd effects, bandwagon effects, and animal spirits in agents’ behavior. Mazzucato(2006) also provided the foundation for a Schumpeterian analysis. Greenwood and Jovanovic(1999) investigated the relationship between the Information Technology (IT) Revolution and the stock market changes. They found that the ratio of market capitalization to GDP tripled from 1985 to 1996 when the companies related to the IT industry, which entered the market after 1968, began to go public on the stock market. They demonstrated that the share of market value of IT startups replaced the share of market value of mainframe computing firms dramatically from 1980. Hobijn and Jovanovic(2001) and Jovanovic and Rousseau(2003) improved on the model mentioned in Greenwood and Jovanovic(1999). Hobijn and Jovanovic(2001) investigated why the stock markets of some OECD countries declined from the early 1970s to the early 1980s. They argued that the IT Revolution was on the horizon in the early 1970s and major technological change ruined the old firms and it took time for the new capital created by the IT Revolution to become part of stock-market capitalization through IPOs. Hobijn and Jovanovic(2001) found that the first oil shock may have caused the productivity slowdown of the 1970s, but it does not seem to clarify the behavior of the stock market. Jovanovic and Rousseau(2003) studied a creative destruction during the Second Industrial Revolution and the IT Revolution using the US stock-market data from 1890–2000. Jovanovic and Rousseau(2003) found that the value that is created by the IPOs of the IT Revolution did not last as long as IPOs of the Second Industrial Revolution because of a dramatic fall in computer prices and the manufacturers’ quick reaction to the microcomputer. Cho and Lee(2003) analyzed South Korean stock prices from 1990 to 2002, which is the period when IT technology was developing gradually. During the development of IT technology, the rate of increase of the stock price of IT technology companies was higher and more stable than that of other industry related companies. Taken together, it can be seen that a company that discloses a patent application or obtains a patent generally shows an increase in stock price, and the higher the quality of the technology, the larger the increase. In the early days of a new industry, share prices and market share change drastically with expectations for the new industry, and as time goes by and the industry grows, the degree of change in both stock prices and market share decreases. In addition, at the beginning of the industry life cycle, the new industry tends to affect the stock price of existing main industries, and the market index tends to decrease. Int. J. Financial Stud. 2018, 6, 88 5 of 22

3. Methodology and Data

3.1. Data The concepts that deal with the technology life cycle include (1) S-curve, (2) Hype Cycle, and (3) Technology Adoption Life Cycle. Among them, S-curve and Technology Adoption Life Cycle are concepts that are generally applicable to any technology. Moreover, as we cannot accurately determine in which phase each technology exists by using the S-curve and Technology Adoption Life Cycle, we use Gartner’s Hype Cycle to identify which stage each technology is in each year. Though the theory of the technology life cycle indicates that the technology phases move forward over time, there is a possibility that some phases reverse with time due to the subjectivity of the participating experts. To solve this problem, among the technologies in the Hype Cycle, we select technologies that meet the following conditions. 1. The technologies should appear at least seven times on the Hype Cycle between 2005 and 2017. 2. They should appear in three or more consecutive stages of the five stages of the Hype Cycle. At least three consecutive phases are required to accurately determine the beginning and the end of a specific phase of the Hype Cycle that is related to a particular technology. 3. If the same technology appears more than once in a Hype Cycle from 2005 to 2017, the following conditions must be met. The stage in the Hype Cycle at point “a” is the same as or later than the stage in the Hype Cycle at point “b” (Here, point “a” and point “b” are time points and point “a” comes after point “b”). Augmented reality was selected and analyzed as one of the technologies satisfying these conditions. To select a company that mainly engages in business related to augmented reality and to analyze the stock price of the company, the information on the official website of each company, the electronic disclosure system of the Financial Supervisory Service (FSS), and the Korea Exchange (KRX) was used. A sample of companies that satisfy the following conditions was extracted. 1. A company that has been listed and is currently operating continuously. 2. More than half of the sales of the company are related to augmented reality-based technology. In the case of a company with many businesses, it is assumed that the stock price is affected by the business that generates more than half of the total sales because the business that affects the stock price is not known exactly. 3. The company has been listed on or before January 2007. In Gartner’s Hype Cycle, the “Peak of Inflated Expectations” phase of augmented reality began in 2010, and therefore, a corresponding evaluation period is required. All the companies that meet the above conditions are listed on the KOSDAQ. Companies such as Samsung Electronics and LG Electronics are large companies that operate various businesses and they are listed on the KOSPI market. Therefore, only a portion of total sales is generated by the augmented reality technology. As the other Korean stock exchange, the KONEX, only opened on 1 July 2013, companies that are listed here do not meet the above conditions. In addition, when they are classified according to the detailed technologies implementing augmented reality, companies that satisfy the above conditions are classified into display companies, camera technology companies, three-dimensional modeling companies, motion recognition technology companies, and so on. However, the display technology company group and the camera technology company group, which include more companies than the minimum number of companies necessary for analysis, were analyzed together with the total augmented reality companies.

3.2. Methodology We generated three groups of companies that met the conditions listed in the previous subsection. Group A includes all the companies, Group B comprises display companies, and Group C consists of camera companies. Groups A, B, and C consist of 10, three, and two companies, respectively. Int. J. Financial Stud. 2018, 6, 88 6 of 22

To compare the stock price of the companies in each group with a benchmark, we calculated the average CAR and buy-and-hold abnormal returns (BHAR) on a monthly basis. We used the KOSDAQ index as the benchmark for calculating CAR and BHAR. We did so by correcting the stock price to eliminate the factors affecting stock prices, such as a decrease in capital and a rights off event. As the “Peak of Inflated Expectations” is the only stage wherein we can accurately determine the beginning and the end, we considered the period before the “Peak of Inflated Expectations” stage as the “Innovation Trigger” stage, and the period from the end of the “Peak of Inflated Expectations” stage to the present stage as the “Trough of Disillusionment” stage. The successive stages of the Hype Cycle were used as the evaluation period and the event period of average CAR. Table1 presents the calculation period.

Table 1. Average cumulative abnormal returns (CAR) analysis period.

Evaluation Period Event Period Jan 2007–Dec 2008 Jan 2010–Dec 2012 (Innovation Trigger) (Peak of Inflated Expectations) Month in which each company was listed-Dec 2008 Jan 2010–Dec 2012 (Innovation Trigger) (Peak of Inflated Expectations) Jan 2010–Dec 2011 Jan 2013–Jul 2018 (Peak of Inflated Expectations) (Trough of Disillusionment)

However, since the listing dates of companies that meet the conditions were different and the evaluation period was the “Innovation Trigger” stage, we considered two evaluation periods. One is the evaluation period from January 2007, the date when the latest listed company was listed on the stock market, to December 2008, and the other is the evaluation period from the date that each company was listed on the stock market to December 2008. Moreover, there was a gap of one year between the evaluation period and event period to eliminate the phase transition effect. We calculated the average BHAR using the data of the “Peak of Inflated Expectations” and “Trough of Disillusionment” stages because the two stages are the only stages wherein the start and the end points are known and we were able to obtain data during the entire period of the stages. Moreover, we used a t-test to check whether the average CAR and the average BHAR are statistically significant (Shah and Arora 2014). We then computed a cross-correlation coefficient between the aforementioned CARs and BHARs of the groups to show the relationship between them. Before calculating a cross-correlation coefficient, we prewhitened the CARs and BHARs with an appropriate ARIMA model and checked whether the prewhitened data were stationary with the Phillips and Perron (PP) test.2

4. Empirical Results

4.1. Average CAR for Evaluation Period: Jan 2007–Dec 2008, Event Period: Jan 2010–Dec 2012. Figure1 shows the average CAR with the evaluation period from 2007 to 2008 and the event period from 2010 to 2012. The average CAR of the entire sample and camera group tended to show an overall upward trend, with an average CAR of 100% and 200% at the end of 2012, respectively. However the average CAR of the display group was just over 0 in 2010 and −40% at the end of 2012. The average CAR of all the samples was generally statistically significant, but the values of the other two groups were not statistically significant. Table2 and Figures2–4 provide detailed cross-correlation coefficients (CCCs) of the average CARs of the groups, the ARIMA model that was used when the average CARs were prewhitened, and the

2 The ARIMA model was chosen based on the Akaike’s information criterion (AIC) and Bayesian information criterion (BIC). Int. J. Financial Stud. 2018, 6, 88 7 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 7 of 23 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 7 of 23 CCCs for the entire sample and display combination, and for the entire sample and camera p-valuesCCCs offor the the prewhitened entire sample time and series display (TS) computedcombination, by theand PPfor test. the Theseentireshow sample that and CCCs camera for the combination, were more than 0.65 when the lag was 0. Table 2 shows that all prewhitened data entirecombination, sample and were display more combination, than 0.65 when and the for thelag entirewas 0. sample Table 2 and shows camera that combination,all prewhitened were data more were stationary. thanwere 0.65 stationary. when the lag was 0. Table2 shows that all prewhitened data were stationary.

Figure 1. Average CAR for Evaluation period: 2007-2008, Event period: 2010–20123. FigureFigure 1. 1.Average Average CAR CAR for for Evaluation Evaluationperiod: period: 2007–2008,2007-2008, Event Event period: period: 2010–2012 2010–20123. 3. Table 2. Cross-correlation of average CARs for Evaluation period: 2007–2008, Event period: Table 2. Cross-correlation of average CARs for Evaluation period: 2007–2008, Event period: Table2010–2012. 2. Cross-correlation of average CARs for Evaluation period: 2007–2008, Event period: 2010–2012. 2010–2012. p-Value of CrossCross p-Valuep-Value of of Cross ARIMA PrewhitenedPrewhitened CorrelationCorrelation CCC ARIMAARIMA Prewhitened Correlation CCC CCC DataData (by (by PP) 4 Group (Lag = 0) 4 (Lag = 0) 5 Data6 (by PP) GroupGroup (Lag = 0) CoefficientCoefficient 5 TS1TS1 6 TS2 Combination ModelModel Coefficient5 TS16 TS2 CombinationCombination Model ma sar drift ma masar sardrift drift ARIMA(0,1,0)(1,0,0)[12] Whole sample & Display 0.663 ARIMA(0,1,0)(1,0,0)[12] 0.3531 3.4809 0.01 0.01 Whole sample & Display 0.663 ARIMA(0,1,0)(1,0,0)[12]with drift 0.3531 3.4809 0.01 0.01 Whole sample & Display 0.663 with drift 0.3531 3.4809 0.01 0.01 Whole sample & Camera 0.682 ARIMA(0,2,1)with drift −0.8814 0.01 0.01 Whole sample & Camera 0.682 ARIMA(0,2,1) −0.8814 0.01 0.01 WholeCamera sample & &Display Camera 0.6820.322 ARIMA(0,2,1)ARIMA(0,1,0) −0.8814 0.010.01 0.010.01 CameraCamera & & Display Display 0.3220.322 ARIMA(0,1,0) ARIMA(0,1,0) 0.01 0.01 0.01 0.01

Figure 2. Cross-correlation coefficient between the average CAR of the whole sample group and the 3 average The average CAR ofCAR the and display the average group forBHAR Evaluation for a specific period: month 2007–2008, in a chart Event are the period: average 2010–2012 of the CAR7. and the 3 The average CAR and the average BHAR for a specific month in a chart are the average of the CAR and the BHAR for each firm from the first month shown in the charts to the specific month. BHAR for each firm from the first month shown in the charts to the specific month. 4 The minimum p-value calculated by PP test in R program is 0.01. The real p-value could be smaller than 4 The minimum p-value calculated by PP test in R program is 0.01. The real p-value could be smaller than 0.01. 0.01. 3 The5 averageThe ma, CAR sar, and and the drift average mean BHAR Θ of for MA a specific model, month Φ inof aseasonal chart are theARaverage model ofand the CARdrift andof ARIMA the BHAR model, for each 5 The ma, sar, and drift mean Θ of MA model, Φ of seasonal AR model and drift of ARIMA model, firmrespectively. from the first month shown in the charts to the specific month. 4 respectively. The6 TS1minimum refersp to-value the calculatedp-value of bythe PP first test time in R programseries of isa 0.01.Group The Combination real p-value could computed be smaller by the than PP 0.01. test and TS2 5 6 The TS1 ma, refers sar, and to driftthe p mean-valueΘ ofof MAthe first model, timeΦ ofseries seasonal of a ARGroup model Combination and drift of ARIMAcomputed model, by the respectively. PP test and TS2 6 refers to the p-value of the second time series of a Group Combination computed by the PP test. For TS1refers refers to to the pp-value-value of of the the first second time seriestime ofseries a Group of a Combination Group Combination computed computed by the PP testby andthe TS2PP referstest. For to the example, the TS1 of the whole sample and display group is the p-value of a whole sample time series p-valueexample, of the the second TS1 timeof the series whole of a sa Groupmple Combinationand display computedgroup is bythe the p-value PP test. of For a whole example, sample the TS1 time of theseries whole samplecomputed and display by the group PP istest, the pand-value the of TS2 a whole of the sample whole time sample series and computed display by group the PP is test, the and p-value the TS2 of ofa thedisplay whole computed by the PP test, and the TS2 of the whole sample and display group is the p-value of a display sampletime and series display computed group is by the thep-value PP test. of a display time series computed by the PP test. 7 Thetime horizontal series computed blue lines are bythe the approximate PP test. 95% confidence interval. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 8 of 23

Int. J. FinancialFigure 2. Stud. Cross-correlation 2018, 6, x FOR PEER coefficient REVIEW between the average CAR of the whole sample group and the 8 of 23 average CAR of the display group for Evaluation period: 2007–2008, Event period: 2010–20127. Figure 2. Cross-correlation coefficient between the average CAR of the whole sample group and the Int. J. Financialaverage Stud. CAR2018 of, the6, 88 display group for Evaluation period: 2007–2008, Event period: 2010–20127. 8 of 22

Figure 3. Cross-correlation coefficient between the average CAR of the whole sample group and the Figureaverage 3. Cross-correlationCAR of the camera coefficient group for between Evaluation the period: average 2007–2008, CAR of the Event whole period: sample 2010–2012. group and the averageFigure CAR3. Cross-correlation of the camera group coefficient for Evaluation between the period: average 2007–2008, CAR of Eventthe whole period: sample 2010–2012. group and the average CAR of the camera group for Evaluation period: 2007–2008, Event period: 2010–2012.

Figure 4. Cross-correlation coefficient between the average CAR of the camera group and the average CARFigure of the4. displayCross-correlation group for Evaluationcoefficient period:between 2007–2008, the average Event CAR period: of the 2010–2012. camera group and the average CAR of the display group for Evaluation period: 2007–2008, Event period: 2010–2012. 4.2. AverageFigure CAR4. Cross-correlation for EvaluationPeriod: coefficien Thet Monthbetween in the which average Each CompanyCAR of the Was camera Listed-Dec group 2008, and the Event4.2. Average Period:average Jan CARCAR 2010–Dec offor the Evaluation display 2012 group Period: for The Eval Monthuation in period: which 2007–2008, Each Company Event Was period: Listed-Dec 2010–2012. 2008, Event Period: Jan 2010–Dec 2012 4.2. FigureAverage5 showsCAR for the Evaluation average Period: CAR withThe Month the evaluation in which Each period Company from Was the monthListed-Dec in which2008, Event each companyPeriod:Figure Jan was 2010–Dec 5 listedshows to 2012 the 2008, average and the CAR event with period the fromevalua 2010tion to period 2012. Thefrom average the month CAR in of which the entire each samplecompany group was remained listed to at2008, around and 30%,the event but then period rose fr fromom 2010 September to 2012. 2011,The average and the CAR average of the CAR entire of Figure 5 shows the average CAR with the evaluation period from the month in which each thesample camera group group remained also rose at moderately around 30%, until but September then rose from 2011, September after which 2011, it rose and sharply. the average The average CAR of company was listed to 2008, and the event period from 2010 to 2012. The average CAR of the entire CARthe camera of the display group groupalso rose showed moderately an overall until downward September trend. 2011, However, after which in allit threerose sharply. groups, theThe sample group remained at around 30%, but then rose from September 2011, and the average CAR of averageaverage CAR CAR were of the generally display not group statistically showed significant. an overall downward trend. However, in all three the camera group also rose moderately until September 2011, after which it rose sharply. The groups,Table the3 and average Figures CAR6–8 wereshow generally that CCCs not for statistically the entire samplesignificant. and display combination, and for average CAR of the display group showed an overall downward trend. However, in all three the entireTable sample 3 and and Figures camera 6–8 combination show that CCCs were morefor the than entire 0.65 sample when the and lag display was 0. combination, Table3 indicates and groups, the average CAR were generally not statistically significant. thatfor allthe prewhitened entire sample data and were camera stationary. combination were more than 0.65 when the lag was 0. Table 3 Table 3 and Figures 6–8 show that CCCs for the entire sample and display combination, and indicates that all prewhitened data were stationary. for the entire sample and camera combination were more than 0.65 when the lag was 0. Table 3 indicates that all prewhitened data were stationary.

7 The horizontal blue lines are the approximate 95% confidence interval.

7 The horizontal blue lines are the approximate 95% confidence interval. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 9 of 23

Int. J. Financial Stud. 2018, 6, 88 9 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 9 of 23

Figure 5. Average CAR for evaluation period: the month in which each company was listed-2008, Event period: 2010–2012. FigureFigure 5. 5.Average AverageCAR CAR forfor evaluationevaluation period:period: the monthmonth in which each company company was was listed-2008, listed-2008, EventEventTable period: period:3. Cross-correlation 2010–2012. 2010–2012. of average CARs for Evaluation period: the month in which each company was listed-2008, Event period: 2010–2012. Table 3. Table 3.Cross-correlation Cross-correlation of averageof average CARs CARs for Evaluation for Evaluation period: period: the month the inmonth which in each which company each p-Value of wascompany listed-2008, wasCross listed-2008, Event period: Event 2010–2012. period: 2010–2012. ARIMA Prewhitened Correlation CCC Cross pData-Valuep-Value (by ofof PP) Group Cross (Lag = 0) ARIMA Prewhitened Correlation ARIMA Coefficient PrewhitenedTS1 TS2 Combination CCC Model Data (by PP) Correlation CCC (Lag = 0) Group ma Coefficientsar drift Data TS1 (by PP) TS2 Group (Lag = 0) Model CombinationWhole sample & ARIMA(0,1,0)(1,0,0)[12] Coefficient TS1 TS2 Combination 0.703 Model ma 0.3516 sar 2.3482 drift 0.01 0.01 Display ARIMA(0,1,0)(1,0,0)[12]with drift ma sar drift Whole sample & Display 0.703 0.3516 2.3482 0.01 0.01 WholeWhole samplesample & & ARIMA(0,1,0)(1,0,0)[12]ARIMA(0,1,0)(1,0,0)[12]with drift 0.703 0.3516 2.3482 0.01 0.01 0.651 ARIMA(0,1,0)(1,0,0)[12] 0.3516 2.3482 0.01 0.01 WholeDisplayCamera sample & Camera 0.651 withwith drift drift 0.3516 2.3482 0.01 0.01 with drift CameraWhole sample & display & 0.328 ARIMA(0,1,0)(1,0,0)[12]ARIMA(0,1,0) 0.01 0.01 0.651 0.3516 2.3482 0.01 0.01 Camera & display 0.328with ARIMA(0,1,0) drift 0.01 0.01 Camera & display 0.328 ARIMA(0,1,0) 0.01 0.01

Figure 6. Cross-correlation coefficient between the average CAR of the whole sample group and the averageFigure CAR6. Cross-correlation of the display coefficient group for Evaluationbetween the period: average the CAR month of the in whole which sample each company group and was the average CAR of the display group for Evaluation period: the month in which each company was listed-2008,Figure 6. Cross-correlation Event period: 2010–2012. coefficient between the average CAR of the whole sample group and the listed-2008, Event period: 2010–2012. average CAR of the display group for Evaluation period: the month in which each company was listed-2008, Event period: 2010–2012. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 10 of 23

Int.Int. J. J. Financial Financial Stud. Stud.2018 2018, 6,, 6 88, x FOR PEER REVIEW 10 10 of of 22 23

Figure 7. Cross-correlation coefficient between the average CAR of the whole sample group and the

Figureaverage 7. Cross-correlationCAR of the camera coefficient group for between Evaluation the average period: CAR the ofmonth the whole in which sample each group company and thewas averagelistedFigure-2008, CAR7. Cross-correlation Event of the period: camera 2010–2012. coefficient group for Evaluationbetween the period: average the CAR month of the in whole which sample each company group and was the listed-2008,average CAR Event of period:the camera 2010–2012. group for Evaluation period: the month in which each company was listed-2008, Event period: 2010–2012.

Figure 8. Cross-correlation coefficient between the average CAR of the camera group and the average Figure 8. Cross-correlation coefficient between the average CAR of the camera group and the CAR of the display group for Evaluation period: the month in which each company was listed-2008, Eventaverage period: CAR 2010–2012. of the display group for Evaluation period: the month in which each company was listedFigure-2008, 8. Cross-correlation Event period: 2010–2012. coefficien t between the average CAR of the camera group and the 4.3. Averageaverage CAR CAR for of Evaluation the display Period: group Jan for 2010–Dec Evaluation 2011, period: Event thePeriod: month Janin which 2013–Jul each 2018 company was 4.3. Averagelisted-2008, CAR Event for Evaluation period: 2010–2012. Period: Jan 2010-Dec 2011, Event Period: Jan 2013–Jul 2018 Figure9 shows the average CAR with the evaluation period from 2010 to 2011 and the event period4.3. AverageFigure from January 9CAR shows for 2013 Evaluationthe toaverage July Period: 2018. CAR TheJan with 2010-Dec average the evaluation CAR2011, ofEvent the period entirePeriod: from sample Jan 20102013–Jul group to 2011 2018 and and the the camera event groupperiod gradually from January dropped 2013 until to July around 2018. AugustThe averag 2015,e CAR then of gradually the entire increased, sample group and thenand the dropped camera sharplygroupFigure fromgradually July9 shows 2017. dropped the The average averageuntil arou CAR CARnd with ofAugust the the display evaluation2015, groupthen periodgradually stabilized from untilincreased, 2010 June to 2015 2011and andthenand then thedropped event rose sharply.sharplyperiod Infrom from this January July case, 2017. the 2013 average The to averageJuly CAR 2018. inCAR allThe threeof averag the groups diesplay CAR were groupof the generally entirestabilized sample not statisticallyuntil group June and 2015 significant. the and camera then rosegroupTable sharply. gradually4 and In Figures droppedthis case,10– 12until the show averagearou thatnd CCCs AugustCAR forin 2015, allall combinations three then groupsgradually were were increased, positive generally when and not then the statistically lag dropped was 0.significant.sharply However, from CCC July for 2017. the cameraThe average and the CAR display of the combination display group was stabilized rather small until when June the2015 lag and was then 0. Tablerose4 Table sharply.shows 4 thatand In allFiguresthis prewhitened case, 10–12 the show average data that were CARCCCs stationary. in for all all three combinations groups werewere positivegenerally when not thestatistically lag was 0.significant. However, CCC for the camera and the display combination was rather small when the lag was 0. TableTable 4 shows 4 and that Figures all prewhitened 10–12 show data that were CCCs stationary. for all combinations were positive when the lag was 0. However, CCC for the camera and the display combination was rather small when the lag was 0. Table 4 shows that all prewhitened data were stationary. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 11 of 23

Int. J. Financial Stud. 2018, 6, 88 11 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 11 of 23

Figure 9. Average CAR for Evaluation period: 2010-2011, Event period: Jan 2013–Jul 2018.

Table 4. Cross-correlation of average CARs for Evaluation period: Jan 2010-Dec 2011, Event period: Jan 2013–Jul 2018.

p-Value of FigureFigure 9.9. AverageAverageCross CARCAR forfor EvaluationEvaluation period:period: 2010–2011,2010-2011, Event period: Jan Jan 2013–Jul 2013–Jul 2018. 2018. ARIMA Prewhitened Correlation CCC Table 4. Cross-correlation of average CARs for Evaluation period: Jan 2010–Dec 2011, EventData period: (by PP) GroupTable 4. Cross-correlation(Lag of average = 0) CARs for Evaluation period: Jan 2010-Dec 2011, Event period: Jan 2013–Jul 2018. Coefficient TS1 TS2 CombinationJan 2013–Jul 2018. Model ma sar drift p-Value of Cross p-Value of Whole sampleCorrelationCross & ARIMA(0,1,0) ARIMA Prewhitened Display 0.749CCC ARIMA −1.3701 DataPrewhitened0.01 (by PP) 0.01 Correlation CCC with drift Group (Lag = 0) Data (by PP) Group (Lag = 0) Coefficient TS1 TS2 Combination Model Coefficient TS1 TS2 CombinationWhole sample & ARIMA(0,1,0)Model ma sar drift Camera 0.477 ma sar drift−1.3701 0.01 0.01 Whole sample & Display 0.749 ARIMA(0,1,0)with drift with drift −1.3701 0.01 0.01 WholeWhole sample & & Camera 0.477 ARIMA(0,1,0)ARIMA(0,1,0) with drift −1.3701 0.01 0.01 Camera & Display 0.273 ARIMA(0,1,0) 0.01 0.01 Display 0.749 −1.3701 0.01 0.01 Camera & Display 0.273with ARIMA(0,1,0)drift 0.01 0.01 Whole sample & ARIMA(0,1,0) Camera 0.477 −1.3701 0.01 0.01 with drift Camera & Display 0.273 ARIMA(0,1,0) 0.01 0.01

Figure 10. Cross-correlation coefficient between the average CAR of the whole sample group and theFigure average 10. CARCross-correlation of the display coefficient group for between Evaluation the period:average JanCAR 2010–Dec of the whole 2011, Eventsample period: group Janand 2013–Julthe average 2018. CAR of the display group for Evaluation period: Jan 2010–Dec 2011, Event period: Jan 2013–Jul 2018.

Figure 10. Cross-correlation coefficient between the average CAR of the whole sample group and the average CAR of the display group for Evaluation period: Jan 2010–Dec 2011, Event period: Jan 2013–Jul 2018. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 12 of 23

Int.Int. J. J. Financial Financial Stud. Stud.2018 2018, 6, ,6 88, x FOR PEER REVIEW 12 12 of of 22 23

Figure 11. Cross-correlation coefficient between the average CAR of the whole sample group and

Figurethe average 11. Cross-correlation CAR of the camera coefficient group betweenfor Evalua thetion average period: CAR Jan of2010–Dec the whole 2011, sample Event group period: and Jan theFigure2013–Jul average 11. 2018. Cross-correlation CAR of the camera coefficient group for between Evaluation the period:average JanCAR 2010–Dec of the whole 2011, Eventsample period: group Janand 2013–Julthe average 2018. CAR of the camera group for Evaluation period: Jan 2010–Dec 2011, Event period: Jan 2013–Jul 2018.

Figure 12. Cross-correlation coefficient between the average CAR of the camera group and the average CARFigure of the12. display Cross-correlation group for Evaluation coefficient period: between Jan the 2010–Dec average 2011, CAR Event of the period: camera Jan 2013–Julgroup and 2018. the average CAR of the display group for Evaluation period: Jan 2010–Dec 2011, Event period: Jan 4.4. AverageFigure2013–Jul Buy-and-Hold12. 2018. Cross-correlation Abnormal coefficient Returns between (BHAR) the for Janaverage 2010–Dec CAR 2012of the camera group and the Figureaverage 13 CAR shows of the the display average group BHAR for Evaluati from 2010on period: to 2012. Jan 2010–Dec The average 2011, BHAREvent period: of the Jan entire 4.4. Average buy-and-hold abnormal returns (BHAR) for Jan 2010-Dec 2012 sample2013–Jul group 2018. continued to decline and to fluctuate until September 2011 and then rose continuously. The averageFigure BHAR13 shows of thethe cameraaverage group BHAR fluctuated from 2010 between to 2012. 0The and average−20% inBHAR the early of the stage, entire but sample the 4.4. Average buy-and-hold abnormal returns (BHAR) for Jan 2010-Dec 2012 averagegroup continued BHAR in January to decline 2012 and changed to fluctuate from negative until September to positive 2011 and and rose then sharply. rose The continuously. display group The achievedaverageFigure anBHAR average 13 shows of BHARthe the camera average of around group BHAR 30% fluctuated infrom May 2010 2010, between to but2012. it 0 declinedThe and average −20% steadily inBHAR the after earlyof that.the stage, entire In this butsample case, the thegroupaverage average continued BHAR BHAR in to of January decline all three 2012and groups tochanged fluctuate were generallyfrom until negative September not statistically to positive 2011 and significant. and then rose rose sharply. continuously. The display The averagegroupTable achieved 5BHAR and Figures anof theaverage 14camera–16 BHAR show group thatof around CCCsfluctuated for30% the betweenin entire May sample2010, 0 and but and− 20%it declined display in the combination, steadilyearly stage, after and butthat. for the In theaveragethis entire case, sampleBHAR the average in and January camera BHAR 2012 combinationof all changed three groups were from more werenegative than generally 0.3to positive when not thestatistically and lag wasrose0. significant.sharply. However, The CCC display for thegroup cameraTable achieved and5 and display an Figures average combination 14–16 BHAR show of was aroundthat small CCCs 30% when for in theth Maye lagentire 2010, was sample 0,but and it declinedand the CCCdisplay steadily for combination, the combination after that. and In wasthisfor the thecase, largestentire the average whensample the BHARand lag camera was of all− 1. threecombination Table groups5 shows werewere that generallymore all prewhitened than not 0.3 statistically when data the were significant.lag stationary. was 0. However, CCCTable for the 5 andcamera Figures and 14–16display show combination that CCCs was for smallthe entire when sample the lag and was display 0, and combination, the CCC for andthe forcombination the entire wassample the andlargest camera when combination the lag was were−1. Table more 5 than shows 0.3 thatwhen all the prewhitened lag was 0. dataHowever, were CCCstationary. for the camera and display combination was small when the lag was 0, and the CCC for the combination was the largest when the lag was −1. Table 5 shows that all prewhitened data were stationary. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 13 of 23

Int. J. Financial Stud. 2018, 6, 88 13 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 13 of 23

Figure 13. Average BHAR for 2010-2012.

Table 5. Cross-correlation of average BHARs for Jan 2010-Dec 2012.

p-Value of Cross FigureFigure 13.13. AverageAverage BHARBHAR forfor 2010–2012.2010-2012. ARIMA Prewhitened CorrelationTable 5. Cross-correlationCCC of average BHARs for Jan 2010–Dec 2012. Table 5. Cross-correlation of average BHARs for Jan 2010-Dec 2012. Data (by PP) Group (Lag = 0) Coefficient pTS1 TS2 Combination Cross Model -Valuep-Value of of CrossCorrelation maARIMA sar drift Prewhitened CCC ARIMA DataPrewhitened (by PP) Whole Correlationsample & CCC Group (Lag = 0) Data (by PP) Group Display (Lag0.368 = 0) ARIMA(0,1,0) Coefficient TS1 0.01 TS20.01 Combination Model Coefficient TS1 TS2 CombinationWhole sample & Model ma sar drift ma sar drift WholeCamera sample & Display 0.594 0.368ARIMA(0,1,0) ARIMA(0,1,0) 0.01 0.01 0.010.01 Whole sample & Whole sample & Camera 0.594 ARIMA(0,1,0) 0.01 0.01 CameraDisplay & Display 0.3680.170 ARIMA(0,1,0)ARIMA(0,1,0) 0.01 0.01 0.010.01 Camera & Display 0.170 ARIMA(0,1,0) 0.01 0.01 Whole sample & Camera 0.594 ARIMA(0,1,0) 0.01 0.01 Camera & Display 0.170 ARIMA(0,1,0) 0.01 0.01

Figure 14. Cross-correlation coefficient between the average BHAR of the whole sample group and the averageFigure BHAR14. Cross-correlation of the display groupcoefficien fort Jan between 2010–Dec the 2012.average BHAR of the whole sample group and the average BHAR of the display group for Jan 2010–Dec 2012.

Figure 14. Cross-correlation coefficient between the average BHAR of the whole sample group and the average BHAR of the display group for Jan 2010–Dec 2012. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 14 of 23

Int.Int. J. J. Financial Financial Stud. Stud.2018 2018, 6,, 6 88, x FOR PEER REVIEW 14 14 of of 22 23

Figure 15. Cross-correlation coefficient between the average BHAR of the whole sample group and

Figurethe average 15. Cross-correlation BHAR of the camera coefficient group between for Jan the 2010–Dec average 2012. BHAR of the whole sample group and the averageFigure BHAR15. Cross-correlation of the camera groupcoefficien for Jant between 2010–Dec the 2012. average BHAR of the whole sample group and the average BHAR of the camera group for Jan 2010–Dec 2012.

Figure 16. Cross-correlation coefficient between the average BHAR of the camera group and the averageFigure BHAR16. Cross-correlation of the displaygroup coefficient for Jan between 2010–Dec the 2012. average BHAR of the camera group and the average BHAR of the display group for Jan 2010–Dec 2012. 4.5. AverageFigure BHAR16. Cross-correlation for Jan 2013–Jul coefficient 2018 between the average BHAR of the camera group and the average BHAR of the display group for Jan 2010–Dec 2012. 4.5. AverageFigure 17BHAR shows for Jan the 2013–Jul average 2018 BHAR from Jan 2013 to Jul 2018. The average BHAR of the whole sampleFigure group 17 fluctuated shows the between average 20% BHAR and from−20%, Jan but 2013 plummeted to Jul 2018. from The July average 2017 untilBHAR January of the whole 2018. 4.5. Average BHAR for Jan 2013–Jul 2018 Thesample average group BHAR fluctuated of the between camera group20% and fell −20%, steadily but toplummeted around − from85% inJuly July 2017 2015, until followed January by2018. a fluctuatingThe averageFigure upward 17 BHAR shows trend of the the for average thecamera next BHAR twogroup years, fromfell andsteadily Jan the2013 average to to around Jul BHAR2018. −85% The of the inaverage displayJuly 2015, BHAR group followed of fluctuated the whole by a betweenfluctuatingsample group 30% upward and fluctuated−40%, trend falling between for the sharply 20% next and fromtwo −20%, Julyyear but 2017.s, and plummeted In the this average case, from the BHAR July average 2017 of BHARtheuntil display January in all threegroup 2018. groupsfluctuatedThe average were between generally BHAR 30% of not the and statistically camera −40%, fallinggroup significant. sharplyfell steadily from to July around 2017. −In85% this in case, July the 2015, average followed BHAR by in a allfluctuating threeTable groups6 and upward Figures were trend generally 18– 20for show thenot statisticallynext that CCCtwo year for significant. thes, and entire the sample average and BHAR display of combinationthe display group was morefluctuated thanTable 0.65 between6 and when Figures the30% lag and 18–20 was −40%, 0.show However, falling that CCCsharply CCCs for for from the other entire July groups 2017. sample wasIn thisand around case, display 0.3the when averagecombination the BHAR lag was was in 0.moreall Table three than6 shows groups 0.65 thatwhen were all thegenerally the lag prewhitened was not 0. statistically However, data were CCCssignificant. stationary. for other groups was around 0.3 when the lag was 0.Table Table 6 6and shows Figures that 18–20all the showprewhitened that CCC data for were the entire stationary. sample and display combination was more than 0.65 when the lag was 0. However, CCCs for other groups was around 0.3 when the lag was 0. Table 6 shows that all the prewhitened data were stationary. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 15 of 23

Int. J. Financial Stud. 2018, 6, 88 15 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 15 of 23

Figure 17. Average BHAR for Jan 2013-Jul 2018.

Table 6. Cross-correlation of average BHARs for Jan 2013-Jul 2018.

p-Value of Cross ARIMA Prewhitened Correlation CCCFigureFigure 17.17. AverageAverage BHARBHAR forfor JanJan 2013–Jul2013-Jul 2018. Data (by PP) Group Table(Lag 6. Cross-correlation = 0) of average BHARs for Jan 2013–Jul 2018. Table 6. Cross-correlation of average BHARs for JanCoefficient 2013-Jul 2018. TS1 TS2 Combination Model Cross ma sar driftp -Valuep-Value of of Cross ARIMA Prewhitened Whole sample & Correlation ARIMA Prewhitened Correlation CCC CCC Data (by PP) Display 0.668 (Lag =ARIMA(0,1,0) 0) Data 0.01(by PP) 0.01 GroupGroup (Lag = 0) Coefficient TS1 TS2 Combination Model Coefficient TS1 TS2 WholeCombination sample & Model ma sar drift 0.374 ARIMA(0,1,0) ma sar drift 0.01 0.01 CameraWhole sample & Display 0.668 ARIMA(0,1,0) 0.01 0.01 Whole sample & CameraWhole & Display sample & Camera 0.374 ARIMA(0,1,0) 0.01 0.01 Display 0.3820.668 ARIMA(0,1,0) 0.01 0.010.01 0.01 Camera & Display 0.382 ARIMA(0,1,0) 0.01 0.01 Whole sample & Camera 0.374 ARIMA(0,1,0) 0.01 0.01 Camera & Display 0.382 ARIMA(0,1,0) 0.01 0.01

Figure 18. Cross-correlation coefficient between the average BHAR of the whole sample group and the Figureaverage 18. Cross-correlation BHAR of the display coefficien group fort Janbetween 2013–Jul the 2018. average BHAR of the whole sample group and

the average BHAR of the display group for Jan 2013–Jul 2018. Figure 18. Cross-correlation coefficient between the average BHAR of the whole sample group and the average BHAR of the display group for Jan 2013–Jul 2018. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 16 of 23

Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 16 of 23 Int. J. Financial Stud. 2018, 6, 88 16 of 22

Figure 19. Cross-correlation coefficient between the average BHAR of the whole sample group and

theFigure average 19. BHARCross-correlation of the camera coefficient group between for Jan the 2013–Jul average 2018. BHAR of the whole sample group and the Figureaverage 19. Cross-correlation BHAR of the camera coefficien group fort Janbetween 2013–Jul the 2018. average BHAR of the whole sample group and the average BHAR of the camera group for Jan 2013–Jul 2018.

Figure 20. Cross-correlation coefficient between the average BHAR of the camera group and the Figure 20. Cross-correlation coefficient between the average BHAR of the camera group and the average BHAR of the display group for Jan 2013–Jul 2018. average BHAR of the display group for Jan 2013–Jul 2018. 5.Figure Main Results20. Cross-correlation coefficient between the average BHAR of the camera group and the 5. MainaverageFor Results the BHAR event of period the display 2010–2012, group there for Jan is only 2013–Jul a difference 2018. in the value of the calculation according to the evaluation period, and the average CAR of the entire sample group and the camera firm group 5. MaintendsFor Results tothe increase event overall, period regardless 2010–2012, of the there evaluation is only period. a difference However, the in averagethe value CAR of thethe display calculation accordinggroup tends to the to stabilizeevaluation after period, declining and until the the averag seconde CAR half of of 2011. the Theentire average sample BHAR group over and the the same camera For the event period 2010–2012, there is only a difference in the value of the calculation firmperiod group shows tends the to same increase results. overall, regardless of the evaluation period. However, the average CAR according to the evaluation period, and the average CAR of the entire sample group and the camera of the displayWhen the group average tends CAR to and stabilize the average after BHAR declining in the until “Peak the ofInflated second Expectations” half of 2011. stage The are average firm group tends to increase overall, regardless of the evaluation period. However, the average CAR BHARconsidered, over the if asame portion period of the shows companies the same whose results. major business is related to the technology that exists of the display group tends to stabilize after declining until the second half of 2011. The average inWhen the “Peak the of average Inflated CAR Expectations” and the stage,average such BHAR as augmented in the “Peak reality, of is setInflated as a portfolio, Expectations” the average stage are BHAR over the same period shows the same results. considered,CAR and BHARif a portion of the portfolio of the companies do not have whose any special major tendency. business is related to the technology that When the average CAR and the average BHAR in the “Peak of Inflated Expectations” stage are exists inHowever, the “Peak it isof evident Inflated that Expectations” the portfolios stage, of all companies such as augmented whose main reality, business is is set the as technology a portfolio, the considered,that exists if in a the portion “Peak of of Inflated the companies Expectations” whose stage major are obtaining business better is returnsrelated thanto the the benchmark.technology that average CAR and BHAR of the portfolio do not have any special tendency. existsInvestors in the “Peak can achieve of Inflated better investmentExpectations” opportunities stage, such by as using augmented this result, reality, and financial is set as companies a portfolio, the However, it is evident that the portfolios of all companies whose main business is the averageare likely CAR to and gain BHAR new financial of the techniques,portfolio do such not as have a new any type special of fund tendency. and a technology-based rather technology that exists in the “Peak of Inflated Expectations” stage are obtaining better returns than thanHowever, an industry-based it is evident sector index.that the Next, portfolios because the of average all companies BHAR of thewhose entire main sample business group and is the thethe benchmark. display company Investors group can periodically achieve better changes investment between positive opportunities and negative by using in the “Troughthis result, of and technology that exists in the “Peak of Inflated Expectations” stage are obtaining better returns than financialDisillusionment”, companies it are is difficult likely toto ascertaingain new a certainfinancia relationshipl techniques, between such the as average a new ratetype of of increase fund and a the benchmark. Investors can achieve better investment opportunities by using this result, and technology-based rather than an industry-based sector index. Next, because the average BHAR of financial companies are likely to gain new financial techniques, such as a new type of fund and a the entire sample group and the display company group periodically changes between positive and technology-based rather than an industry-based sector index. Next, because the average BHAR of negative in the “Trough of Disillusionment”, it is difficult to ascertain a certain relationship between the entire sample group and the display company group periodically changes between positive and negative in the “Trough of Disillusionment”, it is difficult to ascertain a certain relationship between Int.Int. J.J. FinancialFinancial Stud.Stud. 20182018,, 66,, 88x FOR PEER REVIEW 17 17 ofof 2223

the average rate of increase in the share price of each group and the benchmark increase rate, and inonly the the share camera price company of each group group and has the a lower benchmark return increase than the rate, benchmark. and only the camera company group has aAlthough lower return CCCs than between the benchmark. the average CAR and BHAR of each group at lag = 0 are positive, it was Althoughshown that CCCs there between was sometimes the average a large CAR diffe andrence BHAR between of each the group average at lag CARs = 0 areof each positive, group it wasand shownbetween that the there average was sometimesBHARs of aeach large group difference in the between two stages. the averageHowever, CARs a common of each grouptrend andwas betweenassumed. the For average example, BHARs the ofaverage each group BHARs in theof twoeach stages. group However,tended to a commonfall from trendthe second was assumed. half of For2013 example, to the first the averagehalf of BHARs2015. This of eachcan groupbe seen tended more to clearly fall from by thecalculating second half the of cross-correlation 2013 to the first halfcoefficient of 2015. between This can the be monthly seen more rate clearly of return by calculating of each group. the cross-correlation Figures 21 and coefficient 22 show the between monthly the monthlyrate of return rate of of return each of group. each group. Figures Figures 23–28 21 an andd Table 22 show 7 present the monthly the cros rates-correlation of return of eachcoefficient group. Figuresbetween 23 the–28 monthly and Table rate7 present of return. the cross-correlationThe cross-correlation coefficient coefficient between was the positive monthly and rate very of return. large, Theespecially cross-correlation for the combination coefficient with was the positive entire and sample. very large,As the especially average CAR for the and combination the average with BHAR the entirehave cumulative sample. As properties, the average they CAR are and likely the average to have BHARa common have trend, cumulative and the properties, difference they between are likely the toaverage have a CAR common (or BHAR) trend, and of a the group difference and the between average the CAR average (or CARBHAR) (or BHAR)of the other of a group group and is thenot averageexpected CAR to be (or large. BHAR) of the other group is not expected to be large.

40%

30%

20%

10% Whole Sample Display 0% Camera

-10%

-20%

-30%

Figure 21.21. Monthly rate of return forfor JanJan 2010–Dec2010–Dec 2012.2012. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 18 of 23

Int. J. Financial25% Stud. 2018, 6, 88 18 of 22 Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 18 of 23 20%

25%15%

20%10%

15%5% Whole Sample

10%0% Display Camera 5%-5% Whole Sample -10%0% Display

-15%-5% Camera

-10%-20%

-15%-25%

-20% Figure 22. Monthly rate of return for Jan 2013–Jul 2018. -25% Table 7. Cross-correlation coefficient between monthly rate of return (lag = 0). GroupFigure 22. Monthly rate of return for Jan 2013–Jul 2018. Combination Whole Sample & Whole Sample & Camera & Table 7. Cross-correlation coefficient between monthly rate of return (lag = 0). Coefficient Table 7. Cross-correlation coefficientDisplay between monthly rateCamera of return (lag = 0). Display Evaluation Period Group Group Jan 2010–Dec 2012 Combination Whole Sample & Whole Sample & Camera & Coefficient Combination Whole SampleDisplay & Whole SampleCamera & CameraDisplay & CoefficientEvaluation(Peak of Period Inflated Display0.809 Camera0.743 Display0.495 Expectations)Jan 2010–Dec 2012 Evaluation Period 0.809 0.743 0.495 Jan(Peak 2013–Jul of Inflated 2018 Expectations) Jan 2010–Dec 2012 Jan 2013–Jul 2018 0.666 0.624 0.309 (Trough of Disillusionment) 0.666 0.624 0.309 (Peak(Trough of Inflated of Disillusionment) 0.809 0.743 0.495 Expectations) Jan 2013–Jul 2018 0.666 0.624 0.309 (Trough of Disillusionment)

Figure 23. Cross-correlation coefficient between monthly rate of return of the whole sample group and monthlyFigure 23. rate Cross-correlation of return of the display coefficient group between for Jan monthly 2010–Dec rate 2012. of return of the whole sample group and monthly rate of return of the display group for Jan 2010–Dec 2012.

Figure 23. Cross-correlation coefficient between monthly rate of return of the whole sample group and monthly rate of return of the display group for Jan 2010–Dec 2012. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 19 of 23

Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 19 of 23

Int.Int. J. J. Financial Financial Stud. Stud.2018 2018, 6, ,6 88, x FOR PEER REVIEW 19 19 of of 22 23

Figure 24. Cross-correlation coefficient between monthly rate of return of the whole sample group

and monthly rate of return of the camera group for Jan 2010–Dec 2012. Figure 24. Cross-correlation coefficient between monthly rate of return of the whole sample group

Figureand monthly 24. Cross-correlation rate of return coefficient of the camera between group monthly for Jan rate2010–Dec of return 2012. of the whole sample group and monthlyFigure 24. rate Cross-correlation of return of the camera coefficient group between for Jan monthly 2010–Dec rate 2012. of return of the whole sample group and monthly rate of return of the camera group for Jan 2010–Dec 2012.

Figure 25. Cross-correlation coefficient between monthly rate of return of the camera group and

Figuremonthly 25. rateCross-correlation of return of the coefficientdisplay group between for Jan monthly 2010–Dec rate 2012. of return of the camera group and monthlyFigure rate25. Cross-correlation of return of the display coefficient group between for Jan 2010–Decmonthly 2012.rate of return of the camera group and monthly rate of return of the display group for Jan 2010–Dec 2012. Figure 25. Cross-correlation coefficient between monthly rate of return of the camera group and monthly rate of return of the display group for Jan 2010–Dec 2012.

Figure 26. Cross-correlation coefficient between monthly rate of return of the whole sample group and Figure 26. Cross-correlation coefficient between monthly rate of return of the whole sample group monthly rate of return of the display group for Jan 2013–Jul 2018. and monthly rate of return of the display group for Jan 2013–Jul 2018. Figure 26. Cross-correlation coefficient between monthly rate of return of the whole sample group and monthly rate of return of the display group for Jan 2013–Jul 2018. Figure 26. Cross-correlation coefficient between monthly rate of return of the whole sample group and monthly rate of return of the display group for Jan 2013–Jul 2018. Int. J. Financial Stud. 2018, 6, x FOR PEER REVIEW 20 of 23

Int.Int. J. J. Financial Financial Stud. Stud.2018 2018, 6, ,6 88, x FOR PEER REVIEW 20 20 of of 22 23

Figure 27. Cross-correlation coefficient between monthly rate of return of the whole sample group

Figureand monthly 27. Cross-correlation rate of return coefficientof the camera between group monthly for Jan rate2013–Jul of return 2018. of the whole sample group and monthlyFigure 27. rate Cross-correlation of return of the camera coefficient group between for Jan 2013–Julmonthly 2018.rate of return of the whole sample group and monthly rate of return of the camera group for Jan 2013–Jul 2018.

Figure 28. Cross-correlation coefficient between monthly rate of return of the camera group and Figure 28. Cross-correlation coefficient between monthly rate of return of the camera group and monthly rate of return of the display group for Jan 2013–Jul 2018. monthly rate of return of the display group for Jan 2013–Jul 2018. TheseFigure show 28. Cross-correlation that there is a certain coefficient positive between correlation monthly between rate of return the stock of the price camera increase group rate and of the entire sampleThesemonthly show grouprate ofthat andreturn there the of detailed theis a display certain technology group positive for group Jan corre 2013–Jul duringlation 2018. thebetween two phases the stock of the price Hype increase Cycle. Itrate can of bethe assumed entire thatsample there group may beand a certainthe detailed cycle of techno the sharelogy price group of companiesduring the that two has phases a technology of the Hype that isCycle. in theThese It two can show phases be assumed that of thethere Hypethat is therea Cycle,certain may or positive thatbe a therecertain corre is ancyclelation unknown of between the share detailed the price stock phase of pricecompanies in theincrease two that phases rate has of a ofthetechnology the entire Hype sampleCycle. that is It ingroup is the also two and possible phases the detailed that of the the Hype processtechno Cycllogy ofe, shifting orgroup that from thereduring theis anthe “Trough unknown two ofphases Disillusionment” detailed of the phase Hype in phasetheCycle. two to It the phases can “Slope be ofassumed ofthe Enlightenment” Hype that Cycle. there It may phase,is also be wherepossiba certain thele thatcycle technology the of processthe isshare accepted of priceshifting andof companies from utilized, the is“Trough that affected has of a bytechnologyDisillusionment” the process that of is change phasein the intwoto thethe phases users’ “Slope of perception theof Enlighte Hype ofCyclnment” augmentede, or that phase, there reality where is (anLinden unknownthe technology and Fenndetailed 2003is acceptedphase). It is in expectedandthe two utilized, tophases enable is of affected new the typesHype by of Cycle.the stock process It trading is also of skills, possibchange suchle thatinas the the technology-based users’ process perception of shifting pair of tradingfrom augmented the or “Trough statistical reality of arbitrage(LindenDisillusionment” tradingand Fenn that phase 2003). is based to theIt onis “Slope thisexpected kind of Enlighte ofto positive enablenment” correlationnew phase, types ( whereAvellanedaof stock the tradingtechnology and Lee skills, 2008 is accepted).such If the as budgetandtechnology-based utilized, is not available, is affected pair in trading orderby the to or mimicprocess statistical a portfolioof arbitragechange of in the trading the entire users’ that sample perceptionis based group, on aof portfoliothis augmented kind of of detailed positive reality technologycorrelation(Linden and groups (Avellaneda Fenn during 2003). and the It twoLee is phases2008).expected If can the to be budget usedenable instead is newnot available, oftypes a portfolio of instock order of thetrading to entire mimic sampleskills, a portfolio such group. asof technology-basedthe entire sample pairgroup, trading a portfolio or statistical of detailed arbitrage technology trading groups that is duringbased on the this two kind phases of positive can be 6.usedcorrelation Conclusions instead (Avellaneda of and a portfolio Implications and of Lee the 2008).entire Ifsample the budget group. is not available, in order to mimic a portfolio of the entire sample group, a portfolio of detailed technology groups during the two phases can be This paper is primarily aimed at grasping the relationship between a technology at a specific stage used instead of a portfolio of the entire sample group. of6. Gartner’s Conclusions Hype and Cycle, Implications which shows the degree of recognition of a technology over time, and the stock price of a company that mainly deals with the technology. 6. ConclusionsThis paper and is primarily Implications aimed at grasping the relationship between a technology at a specific stageTo of understand Gartner’s Hype this relationship, Cycle, which we shows used the degree event study of recognition technique of to a analyzetechnology the stockover time, price and of companiesthe stockThis pricepaper that mainlyof is a primarilycompany use augmented aimedthat mainly at realitygrasping deals and with the that therelationship satisfy technology. certain between conditions a technology of the Hype at a Cycle. specific stageWe of obtained Gartner’s interesting Hype Cycle, results which through shows the this degree analysis. of recognition In the “Peak of a oftechnology Inflated Expectations”over time, and stage,the stock the portfolios price of a ofcompany all companies that mainly under deals consideration with the technology. were able to obtain higher returns than the Int. J. Financial Stud. 2018, 6, 88 21 of 22 benchmark, but when the companies were classified into groups of detailed technology that forms augmented reality, some of the groups obtained more than the benchmark and others less. For the “Trough of Disillusionment” phase, the value that was computed by the event study method was positive or negative depending on the applied event study technique, evaluation period, and the detailed technology that was used to classify the companies. However, during the “Trough of Disillusionment” phase, there was a positive correlation of the average abnormal return and average BHAR between groups. The results of this study can be used in various fields. First, they can be used by companies to determine how to distribute their R&D budget statistically in order to raise their stock prices. Second, investors can use this study as a basis to judge how to achieve high profits in a stable manner. Third, financial companies can use these results to create new financial products. In particular, this application will be more usable if it is used together with the ideas that are presented in other papers. For example, by analyzing papers and patent data, it is possible to predict the stage of the Hype Cycle of a specific technology and predict the stock price increase ratio of the companies to the benchmark using the technology (Kim et al. 2012; Yoo 2004; Suh and Kim 2016) When considering that considerable efforts have been made to get the technology to be accepted well by the public during the “Trough of Disillusionment” phase, and the affirmation and negativity of news keywords are correlated with the stock price fluctuation (Schumaker and Chen 2009; Kim 2016), we can predict the cycle of average abnormal return and average BHAR of some portfolios within the phase by analyzing the relationship between the positive and negative levels of average abnormal return and average BHAR of each group during the phase and the positive and negative degree of news keywords. However, this study suffered from some limitations. Firstly, this study attempted to identify a relationship between a stock price and a technology in the “Peak of Inflated Expectations” and “Trough of Disillusionment” phases of the Hype Cycle, because (1) there were few technologies that satisfy specific conditions suitable for the research regarding the Hype Cycle; (2) the first phase of the Hype Cycle should be utilized as the evaluation period; and, (3) there were few technologies that were in the “Slope of Enlightenment” stage or the “Plateau of Productivity” stage of the Hype Cycle. If we consider additional technologies besides the technologies published by Gartner Inc. based on additional criteria in the future (Kim et al. 2012), we can apply the event study method to companies that deal mainly with the technologies in the “Slope of Enlightenment” and “Plateau of Productivity” phases. Secondly, the number of companies analyzed in this paper is small, especially the number of members in Group B and Group C. As the history of KOSDAQ is not long and there are only a few companies listed on the KOSDAQ, the number of companies that meet the conditions is small. It would be more reliable if studies on a similar topic to this study were conducted in other stock markets, such as the NASDAQ, where more companies with similar conditions could be listed.

Funding: This research received no external funding. Conflicts of Interest: The author declares no conflict of interest.

References

Avellaneda, Marco, and Jeong-Hyun Lee. 2008. Statistical Arbitrage in the U.S. Equities Market. Available online: https://www.math.nyu.edu/faculty/avellane/AvellanedaLeeStatArb071008.pdf (accessed on 4 April 2018). Baek, Jong-Geun. 2005. Samsung Electronics’ patent trend and correlation with stock price. Patent 21 62: 16–21. Chadha, Alka, and Raffaele Oriani. 2010. R&D market value under weak intellectual property rights protection: The case of India. 82: 59–74. [CrossRef] Cho, Sang Sup, and Jang Woo Lee. 2003. A Relationship on the IT Technology and IT Stock in Korea: A Preliminary Looking. Electronics and Telecommunications Trends 18: 89–98. [CrossRef] Daizadeh, Iraj. 2007. Issued US patents, patent-related global academic and media publications, and the US market indices are inter-correlated, with varying growth patterns. Scientometrics 73: 29–36. [CrossRef] Int. J. Financial Stud. 2018, 6, 88 22 of 22

Greenwood, Jeremy, and Boyan Jovanovic. 1999. The Information-Technology Revolution and the Stock Market. American Economic Review 89: 116–22. [CrossRef] Hobijn, Bart, and Boyan Jovanovic. 2001. The Information-Technology Revolution and the Stock Market: Evidence. American Economic Review 91: 1203–20. [CrossRef] Jovanovic, Boyan, and Peter L. Rousseau. 2003. The Two Technological Revolutions. Journal of the European Economic Association 1: 419–28. [CrossRef] Jung, HyeongJong, and Si-woo Kim. 2017. Investment Ideas from Market Capitalization Changes. Korea Investment & Securities Co., Ltd. Available online: http://file.mk.co.kr/imss/write/20170713102204__00. pdf (accessed on 1 April 2018). Kim, Gi-Beom. 2016. A Study on Stock Price Prediction System for Bio Medicine Sector using Opinion Mining and Learning. Ph.D. thesis, Soongsil University, Seoul, Korea. Kim, MinJo, and HyeongChan Jung. 1995. A Disclosure of Patent Application and Share Value. The Korean Journal of Financial Management 12: 121–142. Kim, Jinhyung, Myunggwon Hwang, Do-Heon Jeong, and Hanmin Jung. 2012. Technology trends analysis and forecasting application based on decision tree and statistical feature analysis. Expert Systems with Applications 39: 12618–25. [CrossRef] Linden, Alexander, and Jackie Fenn. 2003. Understanding Gartner’s Hype Cycles. Gartner. Available online: http: //www.bus.umich.edu/KresgePublic/Journals/Gartner/research/115200/115274/115274.pdf (accessed on 6 April 2018). Mazzucato, Mariana. 2002. The PC Industry: New Economy or Early Life-Cycle? Review of Economic Dynamics 5: 318–45. [CrossRef] Mazzucato, Mariana. 2003. Risk, variety and volatility: Growth, innovation and stock prices in early industry evolution. Journal of Evolutionary 13: 491–512. [CrossRef] Mazzucato, Mariana. 2006. Innovation and Stock Prices: A Review of some Recent Work. Revue de l’OFCE, Presses de Sciences-Po 97: 159–79. [CrossRef] Mazzucato, Mariana, and Willi Semmler. 1999. Market share instability and stock price volatility during the industry life-cycle: The US automobile industry. Journal of Evolutionary Economics 9: 67–96. [CrossRef] Nicholas, Tom. 2008. Does Innovation Cause Stock Market Runups? Evidence from the Great Crash. American Economic Review 98: 1370–96. [CrossRef] Park, Jung-Hee, In-Gook Yeo, and Jong-Beom Moon. 2011. Effect of Small and Medium Businesses’ R&D Intensity and Patents on Their P/E Ratios. Journal of Korea Technology Innovation Society 14: 466–87. Park, HyunWoo, WongHyun Sung, JeomYeol Yun, GeunIm Lee, CheulMin Jung, HanJung Ju, and JeongUn Choi. 2014. Technology Valuation Practice Guide. Ministry of Trade, Industry and Energy. Available online: http://www.valuation.or.kr/assess/%EA%B8%B0%EC%88%A0%EA%B0%80%EC%B9%98%ED%8F% 89%EA%B0%80_%EC%8B%A4%EB%AC%B4%EA%B0%80%EC%9D%B4%EB%93%9C(2014).pdf (accessed on 6 March 2018). Pástor, L’uboš,and Pietro Veronesi. 2009. Technological Revolutions and Stock Prices. American Economic Review 99: 1451–83. [CrossRef] Schumaker, Robert P., and Hsinchun Chen. 2009. Textual analysis of stock market prediction using breaking financial news: The AZFin text system. ACM Transactions on Information Systems (TOIS) 27: 12. [CrossRef] Shah, Priyanka, and Parvinder Arora. 2014. M&A Announcements and Their Effect on Return to Shareholders: An Event Study. Accounting and Finance Research 3: 170. [CrossRef] Suh, Yoonkyo, and SiJeong Kim. 2016. An Exploratory Study of Technology Planning Using Content Analysis & Hype Cycle. Journal of Korea Technology Innovation Society 19: 80–104. Thomas, Patrick. 2001. A relationship between technology indicators and stock market performance. Scientometrics 51: 319–33. [CrossRef] Yoo, Sun-Hi. 2004. A study on the Forecasting Model of Technology Life Cycles by Analysis of US Patent Citation. Journal of Information Management 35: 93–112. [CrossRef]

© 2018 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).