Econometrics [Econ 426] Econometrics Project Do critic’s ratings have significant impact on the sales of video games? Christopher A. Robinson 4/27/2012 Hypothesis: The video game industry is a segment of the economy that is becoming extremely important to the United States. Every year there are hundreds of games that are released on various platforms in various genres. The way these releases are structured, I believe, has a profound effect on the games success or failure in the market. Once a game is developed, it is then given to critic before it hits the marketplace so that they may review the game and give it a numerical score. This numerical score is a signal given to consumers that can alter their economic behavior. The question I seek to answer with this project is: If critics give positive reviews to a video game, is that video game more successful in the market? Motivation: The Video games industry is a significant industry that is often overlooked in everyday business conversation. Statistics show that 67% of households in the United States play video games and on average spend 13 hours per week playing video games. With $18 billion in revenue per year and growing, video games are quickly becoming the fastest growing export in the U.S. according to researchandmarkets.com. A video game, if received well by consumers, can make a company millions of dollars. For example, the game Call of Duty: Modern Warfare 3 grossed over $775 million in its first 5 day on the market. That is a record for any movie, book, or video game in a 5 day period. The industry consists of many large publishers and well over 1000 developers. From 2003 to 2006 the industry’s annual growth rate exceeded 17%, compared to the U.S. economy growth rate of 2% over the same period of years. Literature Review: There is a paper written by Feng Zhu and Xiaoquan (Michael) Zhang called “Impact of Online Consumer Reviews on Sales: The Moderating Role of Product and Consumer Characteristics.” The authors look at the influence that online consumer reviews have on video game sales. Zhu and Zhang concluded that there was a potential positive relationship; however, they felt that data sets with different types of media (books, magazines, movies, etc.) could yield different results than what they concluded with the video game industry. A paper by David Bounie, Marc Bourreau, Michel Gensollen, and Patrick Waelbroeck entitled “The Effect of Online Consumer Reviews on Purchasing Decision: the Case of Video Games.” This paper investigates the relationship between online consumer reviews on the consumers buying behaviors. In their model they included a variable of whether a game had a demo or not. This could contribute to the success or failure of a video game significantly they found. 1|Page With this project I intend to go in a slightly different direction than Bounie, Bourreau, Gensollen, and Wealbroeck did with their paper. I will investigate the effects of online critic reviews on the sales of video game software. Model: The Theory that my model is empirically attempting to investigate is the theory of information cascading. An information cascade is when people observe signals from others and make the same choice, regardless of their own personal information. I hypothesize that when consumers receive a positive signal from video game critics, it inspires those who may have previously had no intentions to purchase a particular video game to purchase the game. It is possible to have a general idea of how many units a game will sell and that is through pre order data. So I can compare the pre order data to the actual sales figures and perhaps be able to tell if the critics rating of the video game affected the games sales. In other words, I am attempting to see if critics rating of games contribute to the demand of games. Sales: Sales are defined as the amount of units a video game sells. In the model, sales are the dependent variable. The sales data that I will use will be from 2011, US sales only. Multiplatform (+ effect on sales): The variable multiplatform is a category that is defined on whether a video game was released on multiple platforms (Xbox360, PlayStation 3, and Wii). I expect that this variable will have a positive effect on sales because if a game is released on multiple platforms, it has a broader consumer base that could potentially purchase that game. Surveys show that 72% of U.S. households own at least one game console [ESRB, 2010]. ESRB Rating (- effect on sales): The Entertainment Software Rating Board (ESRB) rating, which is similar to the way the Motion Picture Association of America (MPAA) ratings are given to movies, is a categorical variable that can have an effect on who may purchase certain games. ESRB ratings should have a negative effect on the sales of games because it should limit the sale of certain games to certain individuals. Multiplayer (+ effect on sales): Multiplayer is a component of a video game that allows other many players to share in a video gaming experience. Multiplayer is designed to both retard the used 2|Page games market and to increase sales of that product. If a game has multiplayer I believe that it will have a positive effect on the sales of that game. Critic Score (+ effect on game sales): There are many magazines, internet websites, and even newspapers that publish reviews of video games. There are plenty of places where you can read an “experts” opinion on if a video game is good or not. I believe that the higher a score that a game receives increases its sales figures and bad reviews cause the inverse effect. Demonstration Available: Some video game developers believe that their game’s sale will be improved if consumers have a chance to try the game for free. This allows consumers to play some games that they may not have otherwise considered purchasing risk free. This does work both ways, however. Consumers that were going to make the purchase could play the demonstration and decide that the product is not worth their money. A demonstration, I believe, does more good than harm because it gives people a chance to interact with a product and form their own opinions. Therefore I believe that having a demonstration increases sales of video games. Price: Prices of video games are very standard throughout the industry, however, there are some games that decide to release with lower price tags perhaps to encourage sales to those who believe that $59.99 is too much for a particular game. The fact that prices are standard for a generation of consoles led me to believe that price will have a very minimal negative effect on games sales. Analysis The first linear model run included all of the variables outlined and defined above. What I find are that most of the variables have low T-values and are insignificant in determining higher video game sales. The only variable that is significant to a 95% confidence interval was critics score. The critical T value is 1.96, and critics score has a T value of 3.5. This is a positive outcome for my hypothesis because it shows that critic’s scores of video games do have an impact on the sales of video games. The second linear model removes all of the insignificant variables. This leaves a model with the intercept term and critics score. With this model I am checking for dramatic changes to the R squared and standard error terms. Significant changes in these could indicate omitted variable bias. With the changes from model one to model 3|Page two, there is only a 2% drop in what the variation that the model explains and the standard error is lower. Both are positive results. The third linear model is testing to see if there is a variable that could help explain more of the variation that is occurring in video game sales. I created a dummy variable called adult. Adult is defined as any games that are rated teen or above; games that are geared to a more mature audience. I found that this variable was almost significant to the 95% level and that it had a negative effect on sales. The negative effect on sales was expected because there are less potential customers for that product. If this variable were significant in the model it would be equal to the loss of approximately a half a million in sales. That is surprising. All three of the models above were linear models. I had high hopes that my variables would be able to explain much more that 9-11% of the variation of sales from the mean. In a n effort to be able to build better models I decided to run nonlinear regressions of the data and see if there is more explanatory power within the regressors of my model. Nonlinear model one was run with a semi log model. The logged variable was the sales variable. This model indicated that critic’s scores are still significant and they better explain what happens to sales when sales are expressed as a percentage. The model also has a lower mean squared error and a higher R squared. Nonlinear model two was a double log model. The variable added was the logarithmic critic score regressor. This model was supposed to build on the success that the first nonlinear model had started. What I found is that the RMSE increased slightly and the R squared decreased slightly leading me to the conclusion that nonlinear model one was a better model for testing my hypothesis. Nonlinear model three returned to the semi log model however I added in the other variables that were regarded as insignificant in linear regressions to see if they could add any explanatory power and be significant in a nonlinear model setting. While I had no more success in having more significant variables at the 95% confidence interval I am pleased with the lower RMSE and higher R squared. There are several variables that would be very promising at the 90% percent confidence interval and that is another positive thing about this model. This is the strongest model that I have available and I will use it to test my hypotheses. Interpretation The question that this research first set out to answer is whether critic’s scores have a significant impact on video games sales in the marketplace. What my model found was that critic’s scores are significant and can alter that games sales in the 4|Page market. The model shows that for every point given by a critic increases video games sales by 0.05%. That may not seem like much but for example the lowest score given by a critic in the data set was 28.33 and the highest was 95.22. The lowest game received a measly 1.42% increase to sales while this highest game enjoyed a 4.76% boost to sales. To put these numbers in perspective of units, the lowest game sold 14,200 more units than without a critic’s score and the highest games sold 47,600 more copies. More interpretive data is available in table c in the appendix section. From and economics perspective, you could say that critics can send signals to consumers and that consumers can take that information and make an economic decision: do I buy a particular game or not? According to the data and my model critics can only affect sales of a game by a maximum on 5% from the mean. The mean in this case is almost one million units; therefore they only sell about 50,000 extra copies of a game if critics rate a game favorable. That is not a significant impact to most games but in terms of money it is as much as 3 million dollars (50,000 copies of a game time the $60 purchase price). That is a significant number. Limitations The limitations of my findings are born because of how little information is available for viewing. I have a strong belief that the marketing budgets of these companies have very significant impacts on the success or failure of their games. There is also the variable of the intangible assets – what a developers name means to the sale of the game. Well know developers games tend to sell better than developers who have just entered the market with their first game. There are studies that show that information cascades from experts are infrequent [Konovalov, 2011] and the presence of highly informed individuals marketing a product increases the probability that there will be a cascade. Conclusion In conclusion, my model attempts to investigate the effects of critic’s scores on video games sales. What it finds is that critic’s ratings on video games are significant however they impact those sales by a maximum of 5% from the mean. Not many would call that information cascade. The signals sent by critics encourages few to run out an buy a game that they have little to no interest in anyway. What may hold the answer to just what video game sales hinge on could be marketing or it could be the merits of the product itself. 5|Page Appendix Tables, Data Sources, and References 6|Page Table A Results of regressions of video games sales in 2011 based on if the game is on multiple platforms, contains multiplayer modes, critics scores, there is a demonstration available, the price and if it is a gear designed for adults. Dependent Variable: Video games sales (in Millions) Regressor Intercept Multiplatform Multiplayer Critic Score Demonstration Available Price Linear Model 1 Linear Model 2 Linear Model 3 -2.52 (5.03) 0.41 (0.37) 0.29 (0.33) 0.042 (0.012) -0.074 (0.33) -0.003 (0.082) -2.07 (0.91) -1.9 (0.91) 0.04 (0.01) 0.04 (0.01) Adult game (Rated Teen or Mature) -0.56 (0.32) Summary Statistics SER R squared n 1.6 0.11 119 1.59 0.09 119 1.57 0.11 119 7|Page Variable Description Sales Sales are defined as the amount of units a video game sells. 2011 Sales data used. Released on multiple platforms (Xbox360, PlayStation 3, and Wii) Entertainment Software Rating Board (ESRB) rating – Everyone, Everyone 10+, Teen, Mature, Adult Only, Rating pending Multiplayer is a component of a video game that allows other many players to share in a video gaming experience An “experts” rating of a games on a scale of 0 – 100. Ability to preview a video game before release Price of a single unit Games rated Teen or above Log(Sales) Log(Critic’s Score) Multiplatform ESRB Rating Multiplayer Critics Score Demonstration Available Price Adult LGsales LGCS Data sources Source http://www.vgchartz.com/gamedb/ Various Developer and Publisher sources, dataset built manually http://www.esrb.org/ratings/ratings_guide.jsp#rating_symbols Various Developer and Publisher sources, dataset built manually http://www.gamerankings.com/ Various Developer and Publisher sources, dataset built manually Standard prices for Video games $60 Dummy variable Logarithmic Sales Logarithmic Critic’s scores 8|Page Table B Dependent Variable: Logarithmic Video games sales Regressor Intercept Nonlinear Model 1 Nonlinear Model 2 Nonlinear Model 3 -4.36 (0.64) -13.77 (2.39) -2.27 (3.44) 0.39 (0.26) 0.22 (.23) 0.05 (0.008) Multiplatform Multiplayer Critic Score 0.05 (0.008) LGCS(Logarithmic Critic Scores) Demonstration Available 3.04 (0.55) -.46 (0.23) 0.34 (.23) -0.04 (0.06) 0.42 (0.22) 1.12 0.23 119 1.09 .29 119 Price Adult game (Rated Teen or Mature) -0.44 (0.23) Summary Statistics SER R squared n 1.1 0.25 119 9|Page Table C Impact of Critics on games sales Scores Lowest - 28.33 Mean – 76.05 Highest - 95.22 Percentage boost Dollar amount from mean (approx. 1 (number of units million) times $60 per unit price) 1.42% - 14,200 units 3.65% - 36,500 units 4.76% - 47,600 units $852,000 $2,190,000 $2,856,000 References: Activision Investor press release Activision Publishing, Inc – November 17, 2011 http://investor.activision.com/releasedetail.cfm?ReleaseID=624766 The Effect of Online Customer Reviews on Purchasing - Decisions: the Case of Video Games David Bounie, Marc Bourrea, Michel Gensollen, Patrick Waelbroeck - 2005 http://perso.univrennes1.fr/eric.darmon/workcommed/papers/bounie_bourreau_gensollen_waldbr oeck_2_nice.pdf Impact of Online Consumer Reviews on Sales: The Moderating Role of Product and Consumer Characteristics Feng Zhu and Xiaoquan (Michael) Zhang – 2009 http://blog.mikezhang.com/files/gamereviews.pdf Video game Database 2012 www.VGchartz.com 10 | P a g e