Meldrum 1. Chris Meldrum. Econometrics. Nelson. 30 April Econometrics Report

Transcription

1 Meldrum 1 Chris Meldrum Econometrics Nelson 30 April 2010 Econometrics Report 1) What is the significant contributor to the sales of a comic book? That is the question that spurred this research. Is the fact that the title is related to an event a primary contributor for the sale of the comic? 2) The motivation behind this report is there is no economic research done on the comic book industry. I have been a comic book fan for the past few years and want to know what goes into the decision of a new comic book. 3) There is no empirical work done on the comics industry. The only academic work on the subject is about comics as literature and the history of comics. There is no doubt that there is some sales tool on the subject but it is probably entirely done and kept in-house, to figure out sales data and provide a new lasting series. I have found an article entitled A Brief Glossary of Comic Book Terminology by Jane Griffin in Serials Review. The information from this article can be used to describe the various aspects of comics that follows in this report. 4) Comic Sales= (Wizard Rank) + e < 0 This variable follows the basic law of demand. As prices increase, the quantity demanded will decrease. However, the interesting fact of comics is that the higher priced comics are typically event comics. > 0 This variable will be similar to in that is a symbol of the differentiation. The industry, concerning issues not companies, can be viewed as monopolistically competitive. A higher issue number is a symbol of success, since some comics such has Batman is nearing issue 700, since it has been in circulation since This implies that the quality of the title is greater than others. > 0 Event comics are typically huge events that are crossovers (meaning that other comic book titles are involved within the story telling process) that are written by the more distinguished authors and drawn by the most well respected artists. They usually take years to plan, all of these factors add up to the fact that they are generally better products than a nonevent comic, which increases the desirability of reading such a story. > 0 The comic book industry is primarily ran by Marvel and DC comics, with them having over 77% of all units shipped within the month of March. Also, they own the rights to the most well known characters, such as Superman, Spider-Man, and the X-Men. So, it is more likely that a comic published from these two companies will be a better selling comic. > 0 Since the ranking is down from 1 being the best and 200 being the worst, lower numbers will provide higher sales, if the variable is a good indicator of sales. Since, Wizard ranks on the

2 Meldrum 2 quality of the character, it could be assumed that a character closer to the number 1 will be a more desirable character to read about. The primary factor which is causing the sales, thus is product distinction in this above mentioned monopolistic competition of the comics, not to be confused with the oligopolistic market in genereal. 5) The following is a list of sources used to complete the research proposed within this topic. 1) Comichron provided the information about the estimated comic sales, price of the issue, issue # and publisher. 2) DC comics website provided information to whether or not a specific issue was involved in an event. 3) Marvel comics website provided information to whether or not a specific issue was involved in an event. 4) Wizard s 200 th Issue This issue provided the information for how great a character is in the history of American comics. 6) The following are the data tables I was able to construct through the SAS output. I settled upon four models to present in this project. The techniques range from logging the functions to attempting joint tests. The null hypothesis is that one of the following variables is a good indicator of sales. Variables Definition Sample Average Price Price of an individual comic for March 2010 $3.48 IGN Rating system, 1 is worse 10 is best 6.9 Issues Number of issues in comic series 160 Event 1 is event or tie-in, 0 if not.4 Wizard ranking, 1 being best 42

3 Meldrum 3 Regression Model Linear-log Joint Quadratic Dummy Joint Test=1 Test=0 Issues ( ) IGN ( ) lnign (8.63) Dmarv (22.21) DDC (22.39) Ddyn (30.73) DDH (30.78) Wizard (184.92) ( ) Wizard (1.04) Constant (.0506) ( ) (25.81) (25097) Adj *All standard errors are at the 95% significance level and within the ( ) 7) The four models that I reconstructed above gave me the highest Adj- however, it is important to note that none gave me statistically significant independent variables. The first model listed looks like the following: Log(sales/1000)= issues + e where all comics that are linked to an event have been taken out of the regression. The model above means that a raising the issue number by one, within a specific comic book series, would increase the sales by.0245%. However, as I stated before the variable is not statistically significant with a t-value of The next model that I was able to construct was a quadratic model which looks like the following: Sales= wizard wizar + e The t-values for the variable and the variable-squared are 1.51 and -1.99, respectively. Thus neither stand as statistically significant. The next model that was built through SAS involved me constructing dummy variables: Sales/1000= lnign Dmarv DDC Ddyn DDH +e Where Dmarv= marvel, DDC=DC, Ddyn= Dynamite, and DDH= Dark Horse; All of these are various comic book publishing companies that appeared within my data set. Once again, all of the t-values failed to be statistically significant ranging from.44 (DDH) to 1.03 (DDC). I used

4 Meldrum 4 this technique because I had reason to believe that the way the other variables interacted would change depending on which publisher was being discussed. Lastly, where my personal hopes where at the outset of this project, is the following joint test model applying only to comics that were used if they were tie-ins or the event title themselves: Sales= wizard ign + e This model once again provided me with no statistically significant t-values, wizard s value being -.08 and ign s value being.77. When a Proc Corr was run, there were no variables that had a high correlation with sales that could affect the modeling. So there was no chance of omitted variable bias within my regressions. 8) When looking at my various models it becomes clear one of the major limitations that I had was number of observations. I imputed the 65 top selling comics in the month of March as my data set and when I started to use variables that eliminated some of the observations that decreased my sample size. For instance, in my final model, after eliminating all none event comics and the further comics that had no value listed for either variable I was left with only 14 observations to use, which is unusable. One needs a minimum of 30 observations for statistical analysis to provide reasonable numbers. Another limitation of my study is there variables; I had no independent variables that related to the quality of either the writer or the artist, except for the ratings offered by a prominent website. If I could have found a good variable, such as years a published writer or number of comic books series drawn, that would have given a better indicator of the quality of the books being presented. Furthermore, I had no previous work to base my own study off of, so I did not know until I ran the models how my well my data represent the sales. However, I still wish to continue this research to see if there is a way to link these variables into a way to explain sales. I still feel like the most promising lead to creating a workable model is the title relationship to an event. Another comic book sales area that I find interesting to further research is if one can identify cultural and demographic trends and the genre of comic book sold. For instance, Jane Griffin points out in American Comics the most popular genre is super heroes (73). I find the topic very interesting, which is why I pursued this path to begin with.

5 Meldrum 5 Work Cited Griffin, Jane K. A Brief Glossary of Comic Book Terminology. Serials Review 24.1 (1998): Electronic Journal Center. 30 March 2010 <