Does "Fans Economy" Work for Chinese Pop Music Industry?
DDoes “Fans Economy” Work for Chinese Pop MusicIndustry ?
Hao Wang
Beijing, [email protected]
Abstract — China has become one of the largestentertainment markets in the world in recent years. Due to thesuccess of Xiaomi , many Chinese pop music industryentrepreneurs believe “Fans Economy” works in the pop musicindustry. “Fans Economy” is based on the assumption that popmusic consumer market could be segmented based on artists.Each music artist has its own exclusive loyal fans. In this paper,we provide an insightful study of the pop music artists and fanssocial network . Particularly, we segment the pop musicconsumer market and pop music artists respectively. Our resultsshow that due to the Matthew Effect and limited diversity ofconsumer market, “Fans Economy” does not work for theChinese pop music industry.
Keywords—Fans Economy; Popular Music; Social NetworkAnalysis; Marketing; Market Segmentation; Community Detection
I. I
NTRODUCTION
With the continuous rapid growth of the Chinese economy,China has become one of the largest entertainment market inthe world. [show some statistics] The term “Fans Economy”has been popular (especially in 2015) in China due to thesuccess of the Chinese mobile phone brand Xiaomi. Xiaomiintroduces high-end smart phones with low prices to createloyal fans of their products. Some popular music companiesand websites in China also wants to create fans of the artiststhey newly produce. Meantime, many firms wish to sellproducts related to pop music artists to earn easy money.Following the latest trend in China , many of them havebecome firm believers of the “Fans Economy” assumptionthat for a specific music artist, there exists a group of fans thatare loyal to the artist only. Some entertainment companieswant to “sell” their music artists like Xiaomi sells its smartphones.From marketing theory, we know that market segmentation isan inseparable part of marketing process. Unless the firm is ina perfect competitive market serving the mass population, likeagriculture, the firm needs to know to which people they wantto sell their products. If “Fans Economy” really works in theChinese popular music industry, the consumer market mustfirst be segmentable based on artists. In other words, if anentertainment company considers its newly introduced artistas a product and if there really could be a fans group loyal exclusively to that artist, then the fans should be separableinto different groups according to different music artists.Similarly, if firms want to sell products related to a specificartist based on the assumption there exists loyal fans for thatartist only, the buyers of such products should be differentfrom buyers of another artist. On the other hand, if the musicmarket is segmentable, music artists should also besegmentable based on their particular music styles based onfans.In this paper we would like to debunk the theory that “FansEconomy” work for the Chinese popular music industry. Wepoint out that with pop music artists taken as products , theconsumer market exhibit serious Matthew Effect and verylimited diversity, which invalidates meaningful marketsegmentation. We also point out that based on music fans, theonly determining factor that separates music artists is theirpopularity rather than their music styles etc. , which fromanother perspective invalidates the “Fans Economy”assumption.We collect our data from the B company (For commercialprivacy issues, we would like to keep the anonymity). Bcompany is one of the largest Chinese online music sites withmillions of unique visitors per month. On the website, theuser could listen to music, download music , search for musicor join a discussion board of a specific music artist to interactwith other fans online.The artist discussion board forms a large artist-fan socialnetwork worthy of research study . In the following sectionswe provide social network analysis of this artist-fan socialnetwork in the hope to provide insight to the Chinese musicmarket. In the meantime, we point out the “Fans Economy”assumption of the Chinese popular music industry is invalid.II. R
ELATED W ORK
Social Network Analysis has been attracting researchers’ andindustrial engineers’ attention since the emergence of SNSwebsites such as Facebook and Twitter. SNS websites haveaccumulated large amounts of users’ social interactionnformation , which made social network analysis based onbig data possible. Facebook even created a data departmentwith devoted effort on social network analysis. They havedone a series of insightful research on Facebook data. Forinstance, they discovered Facebook network has a fourdegrees of separation [1] . They have also used social networkinformation for other applications such as predicting user’sgeographical location [2]. Other companies like Renren.com[3] have also done extensive study on social networks.In this paper, to justify “Fans Economy” does not work for theChinese pop music industry, we utilize the communitydetection technique. Community detection is a well studiedtopic with extensive research literature [4] [5][6]. In this paper,we resort to method proposed in [4] for market segmentation.We prefer the method because it has been well tested for yearsand it is well integrated with Gephi, which provides easy-to-use and manipulatable visualization of social networks.“Fans Economy” has been popular in China in recent years . L.Tie and F. Yong [7] points out consumers have a sense ofbeing the producer when they become loyal fans of a productand interact with the real producer of the product. K.Ye [8]claims “Fans Economy” is crucial for a conventionalcompany to successfully transform itself into an internetcompany. IT companies like Xiaomi, Lenovo and musicentertainment website like iQiYi are all followers of theconcept of Fans Economy.III. S
OCIAL N ETWORK A NALYSIS
A. Degree Distribution
We introduce the following Artist-Fan network G(V, E) . Inthe network, there are two types of vertices - the first type isartist vertices, each of which represents an artist; the second isfan vertices, each of which represents a fan. If a fan joins thediscussion board of an artist , then a directed edge from thefan to the artist is formed. There are no edges between artists.The Artist-Fan network from out data set contains data fromthe year 2016 with 13054 artist and 660054 fans.Figure 1 shows the log-log plot of the out-degree distributionof fan vertices and Figure 2 shows the log-log plot of the in- degree distribution of the artist vertices. Both distributions areapproximately power-lawdistributions.Figure 1. Log-log plot of the number of fans following amusic artistFigure 2. Log-log plot of the number of artists that a fanfollowsThe power law distribution effect indicates that only the fewmost popular music artists gain massive popularity. The restof them are much lesser known to the public. Similarly, themajority of popular music consumers follow a small-to-moderate number of artists. . Community Detection
We create the following Fan-Fan social network G(V, E)where each vertex represents a music fan . A weighted edge isformed between two vertices if two music fans join thediscussion board of the same artist in the year of 2016. Theedge weight represents how many music artists they share. Tosimplify our computation, we omitted edges whose weightsare smaller than 3. In the end, we obtain an undirected graphof 152598 vertices and 38282018 edges.We computed the modularities of Fan-Fan network using themulti-level Louvain method [4] shipped with the “igraph”package of R. The community detection algorithm generates83 communities with 6 dominating communities and 77smaller communities of negligible sizes [Fig 3].Figure 3. Community Sizes of Fan-Fan Network produced byLouvain MethodWe compute the most popular artists in each of the 6 largestcommunities and show the result in Table 1. From thestatistics, in 2016 the most popular music artist is Koreanbands EXO, with a couple of Chinese singers coming next. Itis obvious the popular artists from Hong Kong and Taiwanare no longer popular as they used to be.The Matthew effect of each community is very obvious. Mostfans in each community follows one or two of the mostpopular artists with the exception of the smallest community.For example, in the first community of Table 1, the top 10most popular artists are all from the Korean band EXO. Thenumber of fans following other artists are much fewer.This observation could be very frustrating to “Fans Economy”believers because in the music industry, the Matthew effect isso strong even in segmented communities. You could notcreate many artists with self-exclusive fan bases.
The gamerule is you either introduce an artist that is the mostpopular or you get almost no fans at all.
In addition, there aren’t so many different music styles amusic marketer could pick for a newly introduced music artist.Out of the 6 largest communities in Fan-Fan Network , 3 areKorean, 1 is Hong Kong and Taiwan that has no greatprospects . For the other 2 communities, the young Chineseboy band TFBoys dominates 1 community, the Chinese artistsJianyu Feng and Qing Wang dominate the other community.For a Chinese music marketer , you could either import apopular Korean band into the Chinese market, or youintroduce a local artist similar to TFBoys or Jianyu Feng /Qing Wang.
The Chinese pop music market does not haveenough sub-markets for “Fans Economy” to work.
Now let’s take a look at the popular music artists : Weconsider the following Artist-Artist social network G(V, E)where each vertex represents a music artist . A weighted edgeis formed between two vertices if two music artists share amusic fan on both of the artist discussion board in the year of2016. The edge weight represents how many music fans theyshare. We delete edges whose weights are smaller than 2 andget a social network of 923 vertices and 425503 edges.
Figure 4 Community Detection of the Artist-Artist network withresolution 1.0 (A) Community Detection (B) PageRank Values
We compute the modularities of Artist-Artist network usingalgorithm introduced in [4] with Gephi. On appearance, thesocial network has well formed modularity structures. Whenresolution is 1.0, we obtain 3 communities. The communitiescomprise of 87.65% , 9.97% and 2.38% of the vertices. Wealso compute the weighted PageRank values of the vertices.After comparison, we find out the community detection resultcoincides with segmentation based on weighted PageRankvalues. In other words, vertices of the largest 2.38%PageRank values form the first community, 9.97% of the nextlargest PageRank values form the second community, the restof the vertices form the third community. We also try thecommunity detection algorithm when resolution is smaller.However, even when resolution is smaller, we get similarresults to when resolution is 1.0, i.e. , one dominatingcommunity and coincidence with PageRank Values.The community detection result of the Artist-Artist networkdemonstrates that based on music fans , the pop musicrtists could only be segmented by popularity.
This in turndebunks the myth that “Fans Economy” works for the Chinesepopular music industry because based on fans, the onlydetermining factor that segments artists is not their musicstyles or genres but their popularity.IV CONCLUSIONIn this paper, we apply social network analysis to debunk thepopular myth in the Chinese popular music industry that“Fans Economy” is an ideal model for new artist introduction.We demonstrate that popularity and Matthew effectinvalidates the attempt to create exclusive fans for a specificmusic artist based on factors other than popularity. We alsodemonstrate that the diversity of the sub-markets of Chinesepopular music industry is highly limited and unsuitable formarket segmentation. R
EFERENCES[1] L. Backstrom, P. Boldi, M. Rosa J. Ugander and S. Vigna, “FourDegrees of Separation” , WebSci '12 Proceedings of the 4th AnnualACM Web Science Conference , June 2012[2] L. Backstrom, E. Sun, C. Marlow, “Find Me If You Can : ImprovingGeographical Prediction with Social and Spatial Proximity ”[3] Z. Yang, T. Gao, C. Wilson, B. Zhao, X. Wang, Y. Dai, “UncoveringSocial Network Sybils in the Wild”, Proceedings of the 11 th ACM/USENIX Internet Measurement Conference, 2011[4] V. Blondel, J. Guillaume, R. Lambiotte, E. Lefebvre, “Fast Unfolding ofCommunities in Large Networks”, Journal of Statistical Mechanics:Theory and Experiment, October 2008[5] S. Fortunato, “Community Detection in Graphs”, Physics Reports, 2009[6] J. Leskovec, K. Lang, M. Mahoney, “Empirical Comparison ofAlgorithms for Network Community Detection”, Proceedings of the19th International Conference on World Wide Web, 2010[7] L. Tie, F. Yong, “Fans Economy, Pleasure Generation, and Nomads -Cultural Interpretation of the Popularity of MIUI” , Journal of JishouUniversity , June 2013[8] K.Ye , “ 粉丝经济 ”, 中国华侨出版社中国华侨出版社