Quick viewing(Text Mode)

How Becoming Generalists Affects the Performance of Low Status Artists in Kpop

How Becoming Generalists Affects the Performance of Low Status Artists in Kpop

國立中山大學企業管理學系 碩士論文

Department of Bussiness Management National Sun Yat-sen University Master Thesis

演藝圈中地位較低者是否能透過成為通才提升成績

How Becoming Generalists affects the Performance of Low Status Artists in Kpop

研究生:張叡哲 Jui-Che Chang 指導教授:韋岱思 博士 Dr. Thijs Velema 中華民國 108 年 8 月 August 2019

i

摘要 演藝圈中眾多演藝人員的競爭中,往往只有少數巨星能取得足夠能見度並在該領域成 功。但對於非巨星或是尚未闖出一定地位之藝人而言,要怎麼做才能成功呢? 在這個娛樂產業蓬勃發展的時代,競爭程度隨之上升,有才華的藝人要如何透過商業 與策略取得成功而非被埋沒在人海中? 尤其以競爭最激烈的韓國娛樂圈為例,又該怎麼做才 能從每年上百組出道藝人中脫穎而出呢? 本研究透過韓國娛樂產業中 Melon.com 每週暢銷歌曲排行榜成績資料作為分析依據。 以每首歌在排行榜中取得之最佳排名與停留在排行榜時間作為成功與否指標,並以韓國藝人 之地位差異和其過去作品經驗偏向專才或通才之交互作用作為主要變數。其中地位將以歌手 與偶像兩類別作為區分,而通才/專才之經驗會以其過往跨領域的程度(niche width)作為指 標,並量測這兩個主要變數對於演藝人員成績影響程度。 韓國偶像的粉絲往往將轉型與否視為一個藝人發展的指標,但藝人卻不一定願意跨入 自己不熟悉的領域,可以看出觀眾與藝術家之間的不同考量。因此我希望透過此研究之結 果,期待能為競爭程度高但成功機會低之文化、娛樂、音樂或藝術產業之藝術家們提供未來 發展之策略建議,並增加具有才華但缺乏關注之藝術家被大家看見的機會。

關鍵字: 地位、通才、專才、娛樂產業、韓國藝人

ii

Abstract

What kind of artists are more likely to reach success in cultural industry? Usually only the higher status artists, such as superstars, are the ones being focused by most of the audience. However, not everyone can be superstars, what should those lower status artists do to reach success?

In cultural industries nowadays, the competition betweeen artists become more and more intensive.

How do those artists with great talent stand out from the crowds? Especially for Kpop artists, how do they survive from competing against in the environment where hundreds of artists debuted.

In my reserach, I use weekly charts of Melon.com, a Kpop music site as my data source to examine the interaction between status and experience of being a specialist or a generalist in each artist’s career. In addition, I use best ranks and duration of each songs of Kpop artists to indicate the success.

I hope to help those artists to have a better strategy to reach success in the market which becomes more and more intensive, especially those who are talented but still struggling eager to reach success with my research.

Keywords: Status, Specialist, Generalist, Cultural Industries, Kpop

iii

Table of Contents

Ch.1 Introduction ...... 1 Ch.2 Theories ...... 4 2.1 The Concept of Status ...... 4 2.2 Generalists vs Specialists ...... 5 2.3 Interaction Between Status and Generalist/Specialist ...... 8 2.4 Interaction Between Status and Generalist/Specialist of Agencies ...... 10 2.5 Interaction Between Status and Generalist/Specialist When Status Shifts ...... 11 Ch3. Selection of Research Object ...... 13 3.1 Why Kpop? ...... 13 3.2 Hierarchy Between Artists in Kpop ...... 15 3.3 Power of Agencies and Trainee System in Korea ...... 17 Ch4. Data and Methods...... 19 4.1 Source of Data ...... 19 Weekly Charts ...... 19 Artists’ Profile ...... 21 Songs ...... 21 4.2 Data Overview ...... 25 Dependent Variables ...... 28 Main Independent Variables ...... 29 Control Variables ...... 30 4.3 Methods ...... 35 Ch5. Results ...... 37 Ch6. Discussions ...... 41 6.1 Discussion ...... 41 6.2 Limitation ...... 41 6.3 Conclusion ...... 43 Ch7. Reference ...... 45

iv

Table of Figures Figure 1. Average annual income of Korean celebrities in 2016 ...... 13 Figure 2. Weeks of songs stay in the charts (9 songs that appeared longer than half of year were removed) ...... 20 Figure 3. Trend of how ranking goes through weeks ...... 21 Figure 4. Distribution of genres of songs that entered weekly charts in our dataset ...... 22 Figure 5. Distribution of genres in the songs dataset of the 457 artists ...... 23 Figure 6. Distribution of how many artists each agency manages, the figure only shows those who manages more than 6 artists that appears in our dataset ...... 25 Figure 7. Distribution of song width ...... 27 Figure 8. Distribution of artist width ...... 27 Figure 9. Distribution of agency width ...... 28 Figure 10. Relations between the 2 dependent variables, with the correlation = -0.6011 (23 songs whose duration longer than 75 weeks were removed from the chart)...... 29 Figure 11. Distribution of how many artists do 3 kinds of agencies manage ...... 31 Figure 12. Distribution of the groups each artist belongs to ...... 32 Figure 13. Distribution of main music awards ...... 33 Figure 14. Distribution of competition intensity in the market ...... 33 Figure 15. Distribution of career songs ...... 34 Figure 16. Distribution of best ranking ...... 36 Figure 17. Distribution of duration ...... 36 Figure 18. Distribution of ln(duration) ...... 36

v

Table of Tables Table 1 ...... 9 Table 2 ...... 11 Table 3 ...... 12 Table 4. Top 15 of “Top 40 Power Celebrities Of 2018” announced by Forbes Korea) ...... 14 Table 5. Example of how artist widths are calculated ...... 24 Table 6. Overview of numeric variables ...... 26 Table 7. Distribution of idol-singers ...... 26 Table 8. Overview of status ...... 26 Table 9. Overview of factor control variables ...... 27 Table 10. Distribution of gender of artists ...... 30 Table 11. Full model with multilevel coefficients ...... 40

vi

Ch.1 Introduction

What kind of artists are more likely to reach success in cultural industry? Especially for music production in cultural industry that I’m curious about. Normally people would have superstars come to mind as the first answer. Superstars do have a better chance to appear and promote their works than other artists (Adler, 1985). However, most of artists can’t be superstars, so what strategy do those artists take to attract audience and survive in the competitive environment?

For pop music industry in Korea, as knows as “Kpop”, the situation is way more difficult on account of hundreds of new artists debut in a year and the hierarchy between different kinds of artists. There are two kinds of artists focus their career mainly in producing music works in Kpop, singers and idols. Audience often consider it’s a way for Kpop artists (especially idols) to maintain their popularity by transformation in periodically, but does transformation really work for everyone? I came up with this doubt because there was seldom discussion about Kpop singers’ transformation.

Does this mean audience have different expectations on artists based on different circumstance? Or the transformation issue is only a strategy for those lower status artists to survive by creating impressions?

There are two aspects in this question. First, there’s a difference of expectation between audience and artists in an environment. There are different hierarchy between different artists in Korea (Sim,

2017), singers are considered higher status artists while idols are considered lower status as a result.

Different status causes a different expectation from the audience (Sauder, Espeland, 2009). For those higher status artists, they are usually considered having a legitimacy of music in their original field

(Jensen, Kim, 2015), therefore audience won’t expect a big change on their works. Second, there are

1 still concerns for those artists who are expected to transform. Usually transformation leads to spanning genre and cross boundaries to another field and becoming more generalists than those who do not transform, it takes risks for artists to try on unfamiliar field and become more generalist than just focus on the same field as a specialist before. So who is being expected to transform by the audience and who gets benefits of having a better chance to success by transformation periodically becomes the question I’m mostly curious about.

In previous research, Podolny discussed about status and claimed that people consider the higher status ones on account of they get a better preference from the audience, making the financial cost lower and the revenue higher for the higher status ones in the environment (Podolny, 1993). On the other hand, Zuckerman claimed specialists who focus on developing their own field are usually considered more appealing to the audience (Zuckerman et al., 2003). Both of the research discussed about the cause of success, but lack of who’s more suitable to apply these theories empirical in music industry.

I argue that the lower status artists would earn more benefits by transformation than the higher status since they are considered less of legitimacy in music. Therefore, an expectation to try on different fields and become a generalist from the audience appears. On the other hand, since the higher status artists enjoys more privileges with legitimacy of being considered as “true artists”, audience would expect them to stay in what they are specialized in rather than spanning genres too far away. In

Korea, Kpop idols are considered lower status than singers by both artists and audience (read 01 bloggers, 2016; kknews bloggers, 2017; kknews bloggers, 2018).

As a result, is the experience of being a generalist in lower artists’ career a strategy for them to success in the competition against higher status artist is the main question I’m curious. I would like

2 to prove the phenomenon with the arguments about status and experience in my research. How I define specialists and generalists in my research is based on how wide does the transformation in different genres in each artist’s career. The more genres an artist involved in his/her career, the wider niche width the artist has, the more generalist the artist is.

3

Ch.2 Theories

2.1 The Concept of Status

What is status? A producer's status is the perceived quality of that producer's products in relation to the perceived quality of that producer's competitors' products (Podolny, 1993). It can be considered as a hierarchy of value to the customers. The value can affect on different aspects, such as structure of cost and price, greater status increases the utility derived from the association with or consumption of a good. On the non-tangible aspects such as how appealing to the audience.

Usually, status is a perception of quality when quality of products unobservable. But for cultural industries, that is not the case since qualities of artists’ works are open for every audience. In addition to quality, identities may be another reason causing the order of status. With the knowledge that status is valued by audience’s own right, it’s not hard to imagine the order of status is associated to the identities itself. I would like to argue the following reason that lead different status for artists in music industry. There are different legitimacies of “true artist” based on the culture in different places. For example, audience consider white and orchestras are the essential of jazz in early 1900s.

The African-American and smaller improvisational bands are on the other hand considered to be lowbrow, even they had greater market success (Phillips, Kim, 2008). For artists in Kpop nowadays, singers also enjoy the benefits of being perceived as “true artists” by the audience on account of they focus their business mainly on music production while idols may participate in TV shows and commercials.

The result of status can be viewed in two aspects. First, higher status artitst are more appealing to the fans in their field. With the more legitimacy in music, higher status artists such as singers in Kpop

4 would be more appealing to their original fans. Second, costs for the higher status artists to promote their products is lower. They can promote their music to audience who was not interested in their field in the first place easier. More customers simply flow to the producer without the producer actively seeking them out, and often the higher-status producer receives "free advertising" that the lower-status producer is unable to obtain (Podolny, 1993). For example, if a Kpop idol release a new album (aka. “comeback”, the term which they represent for the new album or single comes out), they’ll need much support from their agency to promote them on TV commercials and social media

(Oh, , 2012). If an idol group or the members of which don’t get the chance to have exposure, audience are less likely to notice their works.

Therefore, I argue that higher status artists would have better performance in selling numbers and rankings in music charts than the lower status artists based on higher loyalty and lower costs for promotion. I would like to prove the higher status artists do perform better than the lower status artists in the first hypothesis.

hypothesis 1

An artist performs better as a singer (higher status) than as an idol (lower status)

2.2 Generalists vs Specialists

If we compare specialist and generalist artists in the same market or environment, specialists usually have a higher level of expertise in its style than the generalist has in any style (Negro, Hannan,

2010). With the higher expertise, artists could develop deeper in the genre and increase their fans’ loyalty to keep following their works. On the audience point of view, if an artist stays in a certain field and had good performance, the artist would become the most representative in that field. Take

5

Kpop idol group for example, a girl idol group whose well-known for lovely and innocent appearance since their debut. For the past 7 years of their career, they didn’t make a huge transformation but stick to the same genre, which made them the first come to mind for the audience when talking about lovely idol groups.

However, if remaining unattached to any other categories, the artist would be less likely to be labeled in that field, causing limits to try on different stuffs in the market (Zuckerman et al., 2003). From

Zukerman’s (2003) claim, artists who defies prevailing sociocognitive frames risks causing confusion to the audience. Therefore, artists are pressured to be symbolic. One who attempts to assume a broad identity risks losing the opportunity to have any standing in the market.

With the argument that artists in an environment tend to lower the risks of losing opportunities in any other fields by being symbolic, the concept of specialists do better than generalists start sowing. So what’s the advantage of being a specialist for artists? Consider a generalist artist who tries to maximize its audience, their strategy should be trying to diverse their performance to meet each niche’s preference. On the audience’s point of view, though a generalist’s work includes some of their interests, but nothing deep enough (Hsu, Hannan, & Koçak, 2009). By definition of niche width theory, the fundamental niche is wider for generalists than for specialists (Carroll, 1985). So a trade off appears, to be more appealing in a niche or to maximize their market.

What did the artists think of choosing between the tradeoff of being specialist and generalist?

Zuckerman et al. (2003) interviewed several actors, agents and producers trying to understand their response on being typecasting (i.e. being considered as a specialist of a certain character). In short of the result, they feel more comfortable and dominant in the field they are specialized in, even consider getting a better chance surviving in the market with a better recognition. But on the other hand, some

6 think it would take a lot of effort to break out the stereotype. As a result, it’s good to be attached to a specific label at the very beginning if an artist doesn’t get much attention. As time goes on, trying on different types of performance can prevent them from trapped in the same niche. Afterall, while the companies make the decision on whether their artists are doing an album or film in what genre, the audience’s reaction play a big role. Though the research was about artists in film industry, we can still take a peek of how people in cultural industries think.

There comes another question on how specialist they have to be in the first place and for how long?

If one stays in a certain genre for too long, they’ll face the risk for being too similar to their former works (Askin, Mauskapf, 2017). Being too similar is an issue that especially causing a big discussion in Kpop. This phenomenon can be blamed to the process of idols’ trainee system and agencies’ strategy. Agencies train and produce plenty of groups who experienced a very similar training process (Lee, 2013), means there are so many groups that shares the same features each year. If those groups insist staying in the genre they debut in, they should be more and more appealing to the audience they target in the first place, but lose the chance to expand their audience.

Artists often face the following dilemma: whether the constraints that result from attempting to assume a specialist identity (that one will be consigned to an identity that is more limited than the set of roles that one could potentially play) are more or less serious than those faced by a would-be generalist (that one be regarded as too much of a dilettante to be accepted in any particular role)(Zuckerman, 2003).

However, there are exact opposite examples for Kpop idols spanning genre. Since most of them carry a label with certain genre from their debut, it’s very likely to give audience a stereotype that this group should only perform in this way. Gfriend was known for a label of “power innocence”

7 since they debuted in 2015, but they tried to transform by appearance in their 2017 album

“Fingertip” into a cooler style. The result of this try did not go well which cause a worst selling number of all their records and a lower ranking in music charts. On the other hand, Girls Generation had a different result in transformation. Girls Generation debuted in 2007 with lovely and innocent dance songs as most of girl idols do, but they started to transform their appearance into sexy dressings in their 2009 album; meanwhile trying on genres such as hip hop and R&B from their 2010 album. Each of their experiment on transformation won great responses in selling numbers and charts rankings.

The examples of both successful and unsuccessful generalist can be interpreted that it’s still about which circumstance the artist is facing and whether the situation is suitable for a transformation instead of specialists always perform better than generalists. Therefore, I would like to argue the interaction between status and experience of being how generalists or specialists in each artist’s career in the next section to prove the effect of experience should be based on the scenario the artist in.

2.3 Interaction Between Status and Generalist/Specialist

From the previous section, we know artists do have the desire to try on different fields and genres regardless of their status or situations, but may be constrained by their market positioning and whether to take the risk of not being accepted by the audience. For those who were lower status in the first place, Kpop idols for example, since they are in a lower status disadvantage, a transformation would give them a chance to expand their niche instead of keep challenging the higher status artists. Moreover, they weren’t considered the "true artists” who stands for the legitimacy of music, which gives them a better chance to try on new genres that audience would

8 accept. Therefore, I argue that lower status artists (Kpop idols) do have a better chance to increase their performance by transformation than the higher status artists, such as Kpop singers.

My argument can be expressed by the following Table 1. The main point of this argument comes from the different expectations from the audience and the different niche strategy between status,

Kpop singers would perform better when being a specialist than being a generalist; meanwhile, Kpop idols would perform better when being a generalist than being a specialist.

generalist specialists

high status (Kpop singers) punished when losing the performs better concentrate

legitimacy perception by on single genre specialized

audience when not focusing in and keep the legitimacy of

on the field of the artist’s “true artist”

specialty

low status (Kpop idols) performs better cross decrease performance

boundaries between albums because of the similarity

to match the expectations with all other idols

from the audience

Table 1

hypothesis 2 there’s an interaction between status and experience (specialists/generalists) for artists’ performance. the more generalist the lower status artists be, the better performance lower status artists’ songs get.

9

2.4 Interaction Between Status and Generalist/Specialist of

Agencies

Mostly, agencies play a role of producing artists’ works and take care of each of their activities. But in Korea, what agencies manage is way more than that. Kpop agencies adopt a “total management” strategy on their artists to operate a system (the trainee system, which will be introduced later in the section “Development of Kpop”) of “producing” their artists like factories (Sim, Kim, Lee, 2017).

Therefore, how agencies manage and train with their profession influence the development of artists.

Based on how agencies could affect artists’ development and how they can manage artists’ works to be produced, I argue that the experience of the agency would affect the artists’ development. With more experience in each niche, the more understanding and insight of how audience would like the performance be the agencies get. On the other hand, if an agency does not have that much experience in each genre but focus on a certain genre they’re specialized in, this would make advantage for those artists who want be more appealing in their current niche.

Applying the arguments I made in the previous section of the interaction between status and experience for Kpop singers and idols to the relationship between artists and agencies, I argue that the more generalist the agency is, the better performance lower status artists (Kpop idols) will have on account of the agency would try to develop their artists in a more generalize way and the advantages of their experience.

10

generalist (agency) specialists (agency)

high status (Kpop singers) not specialized enough for helps singer to get deeper

the artists to develop deeper into their current niche

in the current niche

low status (Kpop idols) gives them better support less experience in supporting

when making transformation transformation decrease the

performance of artists

Table 2

hypothesis 3 there’s an interaction between artists’ status and thir agency’s experience (specialists/generalists) for artists’ performance. the more generalist the agency of the lower status artists be, the better performance lower status artists’ songs get.

2.5 Interaction Between Status and Generalist/Specialist When

Status Shifts

With the concept of audience consider some artists more artistic, such as Kpop singers, which gives them a higher status, the incentive for them to transform genres for maximize their market or profits is lower than those who being considered less artistic, the lower status artists such as Kpop idols. On the other hand, Kpop idols are more of business-oriented, what matters the most for them is to appeal as much as possible. But what happens when a Kpop idol tries to decouple his/her identity with the label of idol to become a singer? Would the audience carry another expectation for them and start to

11 consider them as a “true artist” who should specialize on the genre people used to? Or it’s still acceptable for them still carries the advantages of idols, who earns benefits when being a generalist?

Therefore, I argue that there’s an interaction between status and experience when idols try to promote himself/herself into a singer (I’ll use the term an idol-singer in the following article). Once an idol successfully become a singer by producing solo album or participate with other singers, the expectation from the audience would shift from someone who’s just a good-looking celebrity that can dance and attend TV show to a “true artist”. With the difference of expectations, how idol- singers manage their works should make an influence on their performance.

hypothesis 4 after an idol decoupled from his/her former identity, they would be punished when spanning genre.

generalist specialists

artists go from low status to audience would not expect audience would expect a

high status the artist be a generalist with more artistic development

no specialization when they focusing on the same field

are in higher status

Table 3

12

Ch3. Selection of Research Object

3.1 Why Kpop?

In my research, I’ll try to find the relationship between status and experience (generalist or specialist) to see how artists can reach success in different conditions in Kpop with the following reasons.

First, there’s a clear hierarchy between artists of different identities in the order of actors-> singers-> idols in cultural industry in Korea (reasons detailed in the next section, Development of Kpop), which make it possible to distinguish and categorize different artists. However, conversely to the claim of better revenue, Figure 1 and Table 4 shows that idols shows even greater power and incomes than actors, models and MCs.

Figure 1. Average annual income of Korean celebrities in 2016

Second, the information on their music sites show the genre and other information we need of each song including dates and artist backgrounds. With these data, we can discover if an artist stick on developing one certain genre or tried to transform in the past. Further, the music sites rank all songs by week. With the ranking data, we can see the performance of each song. We can track the

13 performance of each artist and how they performed with each kind of genre. For idols in Korea, one of the most common ways they try to obtain higher rankings or to maintain their duration (i.e. to maintain their success) of popularity is transformation in their genre once a while. Idols’ transformation is a topic that Kpop audience always talk about. With the information we can get from the music sites, the effect of status, experience and the interaction between can be calculated and proofed with models we build.

1 BTS (idol group) 9 Kim Yuna (athlete)

2 (idol group) 10 Ryu Hyun Jin (athlete)

3 (idol group) 11 (idol group)

4 (idol group) 12 Suzy (idol)

5 IU (solo singer) 13 Heung Min (athlete)

6 Song Hye Kyo (actress) 14 Hyun (athlete)

7 Song Joong Ki (actor) 15 AOA (idol group)

8 Park Bo Gum (actor) 16 (idol)

Table 4. Top 15 of “Top 40 Power Celebrities Of 2018” announced by Forbes Korea)

Last, the history and development of Kpop was an unique experience that consist of teamwork of companies, artists and government. How they manage their cultural industry makes a great example for other countries who’s also eager to export their cultural products. By studying Kpop, we can dig deeper of how they succeeded in the world and what’s a better strategy to manage in the future.

There are many kinds of market existed in cultural industry now in the world, most of the markets

14 are in highly competition causing only few of the artists can appeal the fans and therefore survive for a certain amount of time. However, the high level of competition and lower prices than in any other large developed country with the relatively small of scale in Kpop market is a much interesting point for strategies analysis (Messerlin, Shin, 2013). Moreover, its innovative ways of systems created and the uniqueness value that audience cares make me consider it a market worth discuss.

In Korea, artists and celebrities are classified in different labels, such as actors, singers, idols, models etc. by what they usually do, just like other countries. However, these labels not only represent their expertise, but also a sign of different status. Singers and idols rely their business mainly in music production, which gives us a comparable basis of their status, experience and performance.

3.2 Hierarchy Between Artists in Kpop

The most unique value that audience care in Korea is that there’s a hierarchy between different types of artists. On audience’s point of view, movie and drama actors are considered the most reputable among all kinds of artists in the cultural industry. This phenomenon can be traced back to the 50’s, which Korea produced movie first being seen in the world. Later on, in the mid 90’s, TV drama

“Winter Sonata” made a huge success in Japan. With the series of success by dramas and movies overseas, the highest status for actors were created since then.

The national consciousness is one of the most well-known characteristics of Koreans (kknews blogger, 2017). The tragic history of Korea in the early 1900’s made them especially appreciate those who can reach success overseas. As a result, actors are considered the highest of status in the cultural industry.

After 00’s, pop music from Korea also conquer the eastern Asia. This gives singers and idols a

15 higher status and attention than before. For singers versus idols, there’s no exact definition between those 2 kinds of artists. Mostly, audience divide them by the following reasons. Those characteristics are highly related to our data, and will be how we classify singers and idols in our data later.

1. Idols aim to be on television, whereas typical artists(singers) are not necessarily.

2. Idols tend to train more skills besides music, whereas singers focus on their music talents. For example, typical singer-songwriters do not tend to practice skills other than singing and song writings. For singers, producers focus on only their singing ability. But for the idols, everything that may be seen by audience is part of their training lesson. This is why singers are considered more of a

“true artist” than idols.

3. Idols tend to be under contract to an entertainment company before debut so that they get trained by a company, whereas typical artists tend to sign a contract when they get debuted.

(quote from blogs and articles on .com, ptt.cc, kknews.cc, kpopmap.com)

The different ways singers and idols being scouted by agencies leads to another issue: quantities.

Singers get the chance to debut by sending their singing demo tapes to the producers, if being accepted, they would have a chance to release their albums. Quantities are being controlled in this way since voice being the only thing to be considered, there are not that many people who owns such a great voice to debut with. The reason why quantities of either artists important is mainly because of the perception of how artistic. How Kpop agencies produce and manage artists now make audience consider them as mass-produced products from factories rather than artists with legitimacy and specialty (Sim, Kim, Lee, 2017), lower the perceived status of idols as the result.

16

3.3 Power of Agencies and Trainee System in Korea

The agency companies control the music production of both singers and idols in Korea. Three main companies dominate the resources of the market, they are SM Entertainment, YG Entertainment and

JYP Entertainment, as known as the “mega agencies”, controlling most of the power in the music industry (Jung, 2018; Lee 2013). There are also small companies such as who only produced 1 group in their history. The main difference as a result of resource is that we can discover the amounts of groups they manage, most of the popular groups in Kpop are under those three agencies, meanwhile, the small ones don’t have the ability to manage that much groups.

Regardless of the scale, the trainee system for idols are very similar. For the agencies, idols are like a product that manufactured and produced by them (Sim, Kim, Lee, 2017). Since the agencies offer cheap price in music streaming, they try to expand their markets in order to increase revenue. The most often ways those agencies expand the scale of their business are developing markets overseas and producing more groups of idols (Lee, 2014). The teenagers who’s interested and qualified to be a trainee would first sign a contract to enter the training system, someone called this “the slave contract” since they’ll be controlled by the agencies for 7 ~10 years after signing the contract. After becoming a trainee, they’ll be trained on vocal, dancing, instruments, all kinds of etiquette lessons and even their appearance. After about three years of training, few of the qualified trainees could get a chance to form a 4~10 people group and debut (Lee, 2013). And it’s only the beginning of their idol career.

It seems to be hard enough to stand out in the competition of all trainees. Take JYP Entertainment for example, they have tested approximately 50,000 idol aspirants every year, which represents that even only ten out of 50,000 could successfully made their debut, it’s still the most attracting way for teenagers to become an artist (Shin, Lee, 2016). But since with the amounts of agencies and their

17 produce strategy, there are still plenty of groups debut each year, causing the teenagers consider being a trainee is the best way for them to become an artist. However, for the audience, the stereotype of “there are already too many groups” became stronger and stronger.

Though the quality of Kpop idols isn’t affected by the quantity, they are even considered the better ones compared to Jpop idols, the influence of quantity and product-alike appearance still end up a lowest status among the artists in Korea. In artistic considerations, people thus consider idols an easy-to-become, if anyone wants to be a celebrity in Korea, becoming a trainee is the easiest way for teenagers with big dreams.

18

Ch4. Data and Methods

4.1 Source of Data

The information of my data comes from three different places on Melon.com: weekly charts of 157 weeks (about 3 years), artists’ profile of which have been on the weekly charts and the songs’ features of the whole career of these artists.

Weekly Charts Weekly charts are published by Melon.com, one of a well-known and most popular music websites in Korea. There are lots of charts based on different time period such as real-time charts, daily charts and monthly charts. The reason why I chose weekly charts are based on the following reasons. First, most of the music shows, which gives artists a stage to perform, are broadcasted weekly, therefore who can be invited to perform in the show are based on the rankings of weekly charts of each music sites. Both fans and artists values weekly charts a more important and representative reference to evaluate. Second, the average lifespan of a song lasts for about 11.5 weeks in the charts (Askin,

Mauskapf, 2017), which means if the performance of the song did not go well, it’s hardly to see the song go on to the charts in the future. Therefore, it’s most suitable to use week as a unit to measure rather than month or days.

My dataset contains the weekly charts from Feb. 21st 2016 ~ Feb. 17th 2019, 157 weeks (about three years) of complete charts. There are 1476 different songs been into the weekly charts. In these songs,

1303 of them were produced by single artist (i.e. one solo artist or group artist) and 173 of them were produced by multiple artists (i.e. a participation between different solo/group artists). My research focuses mainly on the status and experience and the interaction in between, if we include those songs

19 produced by multiple artists, it may be difficult to calculate the status and experience of them.

Therefore, I exclude those songs from my dataset.

We can see the life cycle of these songs from Figure 2, y axis represents the count of songs appears in the weekly charts. The mean of weeks each song stays in the charts in our dataset is about 11.3 weeks, which is close to the claim that Askin and Mauskapf (2017) made that lifespan of a song is about 11.5 weeks. In the descriptive statistics, we can also see most of the songs stays shortly for only a week in the charts. For those songs who can stay for more than a week, the rankings decrease through time (Figure 3).

Figure 2. Weeks of songs stay in the charts (9 songs that appeared longer than half of year were removed)

20

Figure 3. Trend of how ranking goes through weeks

Artists’ Profile There are 492 different artists in solo or group types of format that have been into the weekly charts during the 157 weeks. 35 of them were not Kpop artists but international superstars such as Maroon

5 and Queen. 3 of them were group consist of TV show MCs formed to promote their TV show that doesn’t belong to either singer or idol. Thus, the 35 artists were removed from the dataset as they were incompatible with my research.

For the 457 artists remained, most of the control variables such as gender, type (solo/group), numbers of members in group, numbers of awards were collected from artists profile. There is also information of agencies in the artists’ profile. The agency column can help us calculate the agency width of each artist, which will be introduced later.

Songs The songs dataset were the songs produced by the 457 artists in their whole career, 35,606 songs were collected into my dataset. Each of them was record with released dates, genres and the album it belongs to. With the dates and genre, we can then calculate the song width, artist width and agency width by aggregating the genres of the same song, artist and agency.

21

There are 40 genres in total classified by Melon.com. Each song was classified into at least one genre, and some of them belong to multiple genres. As we can see the distribution of genres in both weekly charts and songs, ballad is the most popular genre in artists’ production. The proportion were

29.3% in the weekly charts and 43% in the songs dataset.

Dance and rap/hip hop are also popular among the artists, but rap/hip hop do not perform well in the charts as we can see there are only 4% in the weekly charts while there are 17% in the songs dataset.

In Figure 4 and Figure 5, we can also notice that except the genres that appears in pop music the most such as dance, ballad, rap/hip hop, R&B/soul and rock/metal, most of other genres hardly ever existed. The cause of this phenomenon can be attributed to the data of songs and artists’ profile were from those who has been into the weekly charts during 2016 to 2018, it’s not hard to imagine those artists’ works belong to the field that most appealing to audience. Therefore, other genres such as country and instrumental are in small proportion in our dataset.

Figure 4. Distribution of genres of songs that entered weekly charts in our dataset

22

Figure 5. Distribution of genres in the songs dataset of the 457 artists

⮚ Niche Width for Evaluation of How Generalist

There are lots of ways to measure experience and how generalist an artist is. With the data from

Melon.com, genres of each songs classified into 40 genres in total. I record those genres with 40 dummy variables. For a dance song in single genre produced by an idol, we can record it as

{1,0,0…..,0} in our data with the length of the vector is 40. If the song is an cross category which classed into both ballad and folk genre, we can record it as {0,1,1,0,….,0} in our data with the length also 40 (Kovács, & Hannan, 2015; Hsu, 2006).

How we get the niche width is to sum up every vectors of the songs each artist produced in their career before the song we observe in the weekly charts (t

푎푟푡푖푠푡 푤푖푑푡ℎ {µ(ß,ß,…,ß)}=1− µ (ß,ß,…,ß) Equation 1

23

Table 5 shows the example of how I calculate the widths. For artist A who produced songs a~c before (t

2 1 1 푤푖푑푡ℎ = 1−( + + ) 4 4 4 Equation 2

Table 5. Example of how artist widths are calculated

As the result, the value of artist width for a single artist would be defined in between 0 to 1. The bigger value of the artist width, means the more generalist the artist is. Meanwhile, for a specialist who never crossed categories into other genre, the artist width of which would be 0.

Compared to the songs each artist produced and their performance in weekly charts (best rankings and duration), we can see whether singers or idols can perform better in different artist width.

How I calculate agency width is similar to artist width, the only difference is that in artist width, the experiences of genre are only calculated from a single artist, but for agency width, the experiences of genre are calculated from all artists under the same agency.

푎푔푒푛푐푦 푤푖푑푡ℎ {µ(ß,ß,…,ß)}=1− µ (ß,ß,…,ß) Equation 3

As mentioned in the development of Kpop, there are “mega agencies” controlling most of the resources in the music industry (Jung, 2018). In the dataset, we can see in Figure 6, there are 140

24 agencies in total and 57% (81 out of 140) of them were small companies that have only 1 artist been into the weekly chart, and 90% (127 out of 140) of them were less than 5 artists. On the other hand, the top 5 agencies (including the 3 mega agencies, SM Entertainment, YG Entertainment and JYP

Entertainment, and two other agencies, and , that doesn’t belong to the 3 mega agencies but also big and powerful) contains 20% of the artists.

Figure 6. Distribution of how many artists each agency manages, the figure only shows those who manages more than 6 artists that appears in our dataset

Since there are some agencies who own lots of artists, the agency width of those artists would be the same in the same time period, but it only appears when two artist who belong to the same agency made their comeback in the exact same day, which is to rare see.

4.2 Data Overview

The final data frame was created based on each song that have been into the weekly charts and without missing values in each column. Therefore, 1,204 observations (songs) remained in the final data frame.

With the distribution (Figure 7), median and standard deviation of song width, we can see most of

25 songs belong to only 1 genre, only a small proportion of songs belong to multiple genres. On the other hand, most of artists and agencies tried on different genres in their career. In particular, distribution of agency width (Figure 8) is left skewed, telling that most of the agencies tried to expand their experience in different niche rather than dig deeper in the same field.

Statistic N Mean St. Dev. Min Median Max Dependent Variables best rank 1,204 49.576 33.23 1 51 100 ln(duration) 1,204 1.707 1.220 0 1.609 4.92

Main Independent Variables song width 1,204 0.077 0.181 0 0 1 artist width 1,204 0.480 0.206 0 0.5 1 agency width 1,204 0.612 0.160 0 0.667 0.84 idol-singer 1,204 0.119 0.324 0 0 1

Control Variables group number 1,204 0.681 0.98 0 0 6 main music awards 1,204 16.509 25.154 0 6 90 competition 1,204 73.742 23.864 3 75 153 career songs 1,204 84.938 91.53 0 46 854 song genre-ballad 1,204 0.351 0.478 0 0 1 song genre-dance 1,204 0.218 0.413 0 0 1 song genre-electronical 1,204 0.006 0.076 0 0 1 song genre-folk/blues 1,204 0.036 0.186 0 0 1 song genre-independent 1,204 0.055 0.228 0 0 1 song genre-Korean drama 1,204 0.098 0.297 0 0 1 song genre-pop 1,204 0.142 0.349 0 0 1 song genre-rap/hip hop 1,204 0.047 0.212 0 0 1 Table 6. Overview of numeric variables

main independent variables type N 1 0 idol-singer logi 1204 143 1061 Table 7. Distribution of idol-singers

main independent variables type N singer idols status factor 1204 759 445 Table 8. Overview of status

26

Table 9. Overview of factor control variables

Figure 7. Distribution of song width

Figure 8. Distribution of artist width

27

Figure 9. Distribution of agency width

Dependent Variables Before choosing dependent variables for the model, we should first define what’s success. People use sales revenue or market share as the index to appraise how successful a business is normally, but in another point of view, how celebrities influence publics can also be another indicator.

We can define success by either the highest rank one gets or the duration of surviving in the environment, one of the most common ways people evaluate how albums or movies is to see how they perform in charts. Either the best ranking or the time they stay in the charts is important, and that’s what artists in cultural industries care about.

Weekly charts provide me the two main dependent variables: best ranking of each song (best ranking) and how long does each song stays in the charts (duration). I’ll create two models for each of the dependent variables. While the two dependent variables are highly correlated (-0.6011), it’s expected to see the results of both model in good correlation but opposite direction of coefficient.

28

Figure 10. Relations between the 2 dependent variables, with the correlation = -0.6011 (23 songs whose duration longer than 75 weeks were removed from the chart)

Main Independent Variables ⮚ Status

This variable is for hypothesis 1, testing whether the performance of higher status artists are superior than lower status artists. Status is expressed in 2 level of factors, singer and idol, which singer stands for the higher status.

⮚ Interaction between Status and Experience

The interaction consisted of two main variables testing the result of hypothesis 2. Experience is represented by artist width which explained earlier in the Evaluation of How Generalist chapter. As mentioned, the bigger the width is, means the more generalist the artist was.

⮚ Interaction between Status and Agency Experience

Hypothesis 3 also argues that not only the experience of singer can affect the performance, but also

29 the experience of agency since agencies play a dominant role in training those artists and producing their albums. Therefore, the status column and the agency width column would have an interaction to test whether the hypothesis is significant.

⮚ Interaction between Experience and Idol-Singer

The last main independent variable is for hypothesis 4, which I argued if an artist successfully turned into a singer from an idol, the artist shouldn’t be that generalist before but should show his/her specialty in a specific field to prove him/herself a “true artist”. Therefore, I argue that if an artist is an idol-singer, the bigger the singer width is, the worse performance the artist will get.

Control Variables Besides the basic information of each artist, such as gender, type of formation (group or solo), nationality and heights, there are some control variables that can reflect scenarios in Kpop.

Table 10. Distribution of gender of artists

⮚ Agencies

I classified those companies into 3 labels by their power and size. The mega agencies, SM

Entertainment, JYP Entertainment and YG Entertainment that audience considered most powerful and with most resources were labeled as “Big 3”. There are other nine big companies such as FNC

Entertainment, Cube Entertainment, and Starship Entertainment that doesn’t have as much power as the “Big 3” were labeled as “Big”. While the rest of the companies are labeled as “Small”.

From Figure 11 we can see, about 20% of the artists that have been into the weekly charts are signed

30 by the “Big 3”, and about 12% of the artists belong to the other big companies.

Figure 11. Distribution of how many artists do 3 kinds of agencies manage

⮚ Group Number

This variable is mainly for controlling how much resource does a solo artist get, no matter the artist is a singer or idol-singer. If an artist is considered important by his/her agency, there will be more chance to cooperate with other artists arranged by agencies or commercial companies. Therefore, the more group the artist participate in means the agency values the artist more, and the bigger amounts of fans the artist gets.

For the artists whose type belongs to “group”, group number are record in 0 since a group won’t belong to other groups. On the other hand, for the artists labeled in “solo”, most of the artists who belongs to only 1 group are idol-singer, the 1 represents the original idol group this artist debuted with.

31

Figure 12. Distribution of the groups each artist belongs to

⮚ Awards

With effect of superstardom, superstars would get more attentions and income than those who are not then before (Adler, 1985; Rosen 1981). I collected the text data of the awards each artist got in their career, labeled them into two kinds of awards, main music awards and TV show awards. I chose main music awards for the model since their votes are from both fans and judges, which is more representative. The main music awards in my data frame are listed below:

The main music awards that being voted by fans and professional judges, including

1. (MAMA),

2. (골든디스크 어워즈),

3. Music Awards (하이원 서울가요대상) ,

4.

I calculated the total numbers of awards above each artist got in their career to represent quality of their works.

32

Figure 13. Distribution of main music awards

⮚ Competition in Market

The intensity of competition in the market is another factor need concerned. I control this variable by counting how many songs released in the same week. The reason I chose week as time scale to evaluate the intensity is the same with why I chose weekly charts, since most of artists compare the performance of a comeback by week instead of other time scale.

Figure 14. Distribution of competition intensity in the market

33

⮚ Career Songs

How I calculate artist width and artists’ career experience in each genre would lead to a confusion between true specialists and those who just debuted and do not have any experience as their artist width record 0. Therefore, how many songs an artist produced in his/her career could help distinguish them the length of career. I calculate the career songs by counting the songs they produced between the date of current song and the date of the artist’s debut.

Figure 15. Distribution of career songs

⮚ Genres

The genre column is about the genre of the song in this “comeback”. Since there’s a trend of higher dance intensity and strong addictive tone that includes emit energy (Messerlin, Shin, 2013), making genres such as dance and hip-hop music more popular, we can see if a certain genre can affect the performance with the control variable.

34

⮚ Multilevel Model for Superstardom Effect

From previous research, we know superstardom effect exists in cultural industry and do affect the performance of each artist (Adler, 1985; Rosen 1981). I tried to use number of awards to interpret the variance between different artists, but it’s still not precise enough. The superstardom effect varies between different artists. However, from the data we currently have, it’s not observable, or we can say it’s not enough to estimate the true effect with different stars. To deal with the mediator we can’t observe in the data by multilevel model, it helps us to use individual artist as the dependent variable to predict, we can get unique slopes and intercepts in linear regression model for each artist and analyze the potential superstardom effect (Nihalani, Wilson, Thomas, Robinson, 2010; Filimon,

López-Sintas, Padrós-Reig, 2009).

Therefore, I added an additional variance based on each artist to make our model more accurate by

Multilevel Random Coefficient modeling (MRC model). In unconditional model, there’s a fixed proportion for every variable in the model, but in MRC, the random portion allows intercepts vary among the groups. I use each artist as a group in my model to represent the vary of superstardom.

4.3 Methods

Two models will be created for the two dependent variables. For the first model, using best ranking as Y. Though it’s an ordered variable and should build with ordered logit regression, but since there are 100 labels of ranks, I chose linear regression rather than ordered logit regression for my model.

As we can see the distribution of best ranking (Figure 16) and duration (Figure 17) aren’t normally distributed, especially duration is right skewed. Therefore, I would use ln(duration) as dependent variable for model 2, which is closer to normally distributed.

35

Figure 16. Distribution of best ranking

Figure 17. Distribution of duration

Figure 18. Distribution of ln(duration)

36

Ch5. Results

I first test hypothesis 1 with status in both models to see if the higher status artists have better performance than lower status artists. Though the p value of status in both models weren’t significant, but from coefficients we can see, the performance for higher status (singers) in both best ranking and duration are better than lower status (idols).

We noticed that the higher status artists would have better performance than the lower status from hypothesis 1. And we can see for all of the artists, the performance in both models increase when the artist has a more generalist (artist width, the higher in artist width, the more general experience is in the artist’s career) career. But for hypothesis 2, I would like to test whether a lower status artist can really increase their performance by being a generalist. I use the interaction between status

(singer/idol) and experience to proof the argument. The results of p value are also not significant, but we can see the direction of coefficients do match our expectation that for higher status artists

(singers), the higher artist width, the lower performance in both best rank and durations.

Similar to hypothesis 2, hypothesis 3 tests the interaction between status of artist and the experience the artist’s agency has (agency width, the higher in agency width, the more general experience is in the producing agency’s history). The results for agency width and interaction between status and agency width is similar to the previous hypothesis. Though it’s not significant related, but the direction of coefficients matches the argument that lower artists (idols) will have a better performance when their agency is more generalist.

Finally, for hypothesis 4, I would like to see if idol-singer get punished when being a generalist after they increase their status. Though the results are also statistically insignificant, but from the direction

37 of the coefficients we can see there’s a tendency that an idol-singer being punished for being a generalist.

From the insignificant results of the main independent variables and R squares are only about 0.101 and 0.118, I argue that the low R square of both my models can be interpreted by the following reasons.

First, it might be because of the limitation of the data, which will be explained completely in the next section. The limitation made control variables not powerful enough to control other factors in the market. Thus, the true effect can’t be discovered in the model causing the p value higher than expected.

In addition, the concept of hypothesis that originally comes from the audience of Kpop might not really reflect the entire market. We all knew that most of the audience pay attention on a few of the most famous stars, leading their perception could be biased with the outlier in the environment. As a result, the model shows that there is no significant relation between performance and the interaction of status and experience. The main reason that really affect performance of artists is still a question that we are eager to explore.

38

Dependent variable best rank ln(duration) (1) (2) Main Independent Variables status(singer) -9.957 0.048 (13.197) (0.464) song width 28.203 -0.565 (36.188) (1.311) artist width 7.818 -0.131 (9.193) (0.327) agency width -12.968 0.165 (17.872) (0.625) idol-singer -1.794 0.230 (11.233) (0.399)

Interaction between Main Independent Variables status(singer):artist width 7.911 -0.362 (13.528) (0.480) status(singer):agency width -2.154 0.321 (20.547) (0.721) artist width:idol-singer 10.103 -0.655 (18.555) (0.659)

Control Variables gender-male 5.070* -0.061 (2.787) (0.097) gender-mix 2.509 -0.080 (8.673) (0.298)

Control of Power of Agencies agency-big 5.639 -0.238 (4.264) (0.149) agency-big 3 4.510 -0.234* (3.981) (0.138)

Control of Popularity group number -1.959 0.027 (1.473) (0.052)

Control of Quality main music awards -0.310*** 0.017*** (0.096) (0.003)

Control of Intensity of Market competition 0.092** -0.003** (0.041) (0.001)

39

Control of Career Length career songs 0.023 -0.002*** (0.015) (0.001)

Control of Popularity of Each Genre song genre-ballad -4.672 0.325*** (3.354) (0.119) song genre-dance -8.729** 0.475*** (3.846) (0.137) song genre-electronical -4.256 0.540 (12.847) (0.464) song genre-folk/blues -10.593 0.571** (6.991) (0.250) song genre-independent -16.538 0.407 (18.990) (0.687) song genre-Korean drama -10.530 0.454 (18.364) (0.665) song genre-pop 1.380 0.162 (3.743) (0.134) song genre-rap/hip hop 12.104** -0.172 (5.800) (0.206) Constant 54.067*** 1.545*** (12.737) (0.449) R2 0.101 0.118 Observations 1,204 1,204 Log Likelihood -5,840.575 -1,926.086 Akaike Inf. Crit. 11,735.150 3,906.172 Bayesian Inf. Crit. 11,872.100 4,043.127 Note: *p<0.1; **p<0.05; ***p<0.01 Table 11. Full model with multilevel coefficients

40

Ch6. Discussions

6.1 Discussion

The main question of my research is that I’m curious about who can success in cultural industries and why do those people success, especially in music production. I argue that for those lower status artists, there’s a better chance for them to increase their performance by being a generalist, that is, try on more different genres to attrack more audience in different niche. On the other hand, for those higher status artists, they should maintain their performance by stick in what they were specialized in.

I tried to answer my question above with the data I collect from Melon.com. Moreover, my research also tried to examine the phenomenon that Kpop audience discuss about Kpop idols’ transformation.

Kpop audience consider transformation a necessary process in their career development. Idols should try to transform their genres and appearance to expand their fan base and gain their discussions online.

From my research, I would like to see the significant results that the interaction between status and experience of being a generalist or specialist would affect the results of each artist’s performance.

With the significant results, I could prove that for those artists who just debut or still struggling to reach better performance a strategy for them. However, the results aren’t statistically significant, and

I would like to discuss the reasons according to the limitations of my data in the following section.

6.2 Limitation

I would like to discuss why the models are with high p value with following reasons, the lack of time range, control variables, difficulties to truly define idols and singers, and lack of genre of

41 appearance.

First, the time range of my raw data from Melon.com contains only 3 years from Feb. 2016 to Feb.

2019 on account of the access to the website. The lack of observations restricts us to analyze more data in the past few years, when Kpop is also popular and full of different types of artists.

Second, the lack of control variables is mainly because of lots of artists who don’t have a big-enough fan base do not have a complete personal data such as birthday and height, which is an important issue in cultural industry since appearance and age affect how popular an artist is and his/her performance. In addition, how I control the effect of superstardom is not precise enough. For example, one of the most popular idol group “” only numbered 17 in the number of main music awards while “Twice” was 44. The result shows that the number of awards can’t really reflect how popular the artist is and how many fans does the artist has. Though the random coefficient can help us deal with about 10% of the variance, it’s still not enough to reflect the reality.

Third, there’s no clear definition between idols and singers, therefore how I distinguish them is mostly by the characteristics that most audience approve that I mentioned in Development of Kpop.

As a result, there might be bias when defining the artist’s status. More, there are some artists that share both characteristics idols and singers have. For those artists, I labeled them into singer on account of the higher status they have than regular idols. But it’s hard to test that whether they will be punished when spanning genre.

Last, which is the most important one, the lack of data of appearance. As how dancing and MVs are more and more critical how audience appraise an album and a song, the influence of pure music becomes lower. For producers such as agencies, they pay more attention on the appearance like

42 dancing styles, dressing, hair style, makeups and even plastic surgery nowadays (Lee, 2013). There aren’t any sites or database that collect those data and organize them yet, therefore, those critical factors couldn’t be included into our analysis.

6.3 Conclusion

From the results of my model, we still can’t see the significance relations of our hypothesis.

However, there are still some interesting findings in the models. The coefficients of genre in both models show that dance songs really appeals audience nowadays, the performance of best ranks and duration both increase if an artist produce a new dance song. Besides dance songs, we can see the duration of folk/blues songs are significant longer than others, which is intersting that this is a genre not that popular nowadays compared to dance and ballad, but still performs well in weekly charts.

Compared to dance and ballad songs, rap/hip hop is another blockbuster genre for teenagers in the decade, however, from the results of our model, rap/hip hop songs do not give artists better performance. Especially in model of best rank, it significantly lowers the ranks of the song in the weekly charts, which is a surprising finding that subvert our perception of rap/hip hop songs.

The honor of music awards also helped artists increase their performance. These awards not only represent an honor from judges and audience, but also reflect on their performance in weekly charts.

On the other hand, the intensity of competiton in the market do significant affect the performance in both best rank and duration. For artist who’s about to have another “comeback”, it’s better for them to avoid when other’s release their new albums, especially those superstars.

From the insights of my models, the effect of genres is more obvious than the main independent variables. Maybe it’s a better strategy for agencies and artists to develop in the genres that appeals

43 audience the most such as ballad and dance songs or genres not that popular but still enjoys fine performance such as folk/blues, instead of focusing on transformation that leads to a risk of losing original fans. On the other hand, Kpop artists and agencies also value fans oversea an important market (Messerlin, Shin, 2013). The issue of transformation may be significant in other markets which we couldn’t examine from the audience in Korea. Therefore, for future research, it may be another interesting topic to compare the difference of main independent variables between different region and different strategies suitable for different audience.

44

Ch7. Reference Adler, M. (1985). Stardom and talent. The American economic review, 75(1), 208-212.

Askin, N., & Mauskapf, M. (2017). What makes popular culture popular? Product features and optimal differentiation in music. American Sociological Review, 82(5), 910-944.

Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in populations of organizations. American journal of sociology, 90(6), 1262-1283.

Filimon, N., López-Sintas, J., & Padrós-Reig, C. (2011). A test of Rosen’s and Adler’s theories of superstars. Journal of Cultural Economics, 35(2), 137-161.

Goldberg, A., Hannan, M. T., & Kovács, B. (2016). What does it mean to span cultural boundaries?

Variety and atypicality in cultural consumption. American Sociological Review, 81(2), 215-241.

Hsu, G. (2006). Jacks of all trades and masters of none: Audiences' reactions to spanning genres in feature film production. Administrative science quarterly, 51(3), 420-450.

Hsu, G., Hannan, M. T., & Koçak, Ö. (2009). Multiple category memberships in markets: An integrative theory and two empirical tests. American Sociological Review, 74(1), 150-169.

Jensen, M., & Kim, H. (2015). The real Oscar curse: The negative consequences of positive status shifts. Organization Science, 26(1), 1-21.

Jung, UK (2018). Kpop Secret.

Kovács, B., & Hannan, M. T. (2015). Conceptual spaces and the consequences of category spanning. Sociological science, 2, 252-286.

Lee, M. (2013). Star Management of Talent Agencies and Social Media in Korea. In Handbook of

Social Media Management (pp. 549-564). Springer, Berlin, Heidelberg.

Lee, M. H. (2014). Penetration Strategies of SM Entertainment in Global Market. Journal of

Information Technology Services, 13(3), 77-92.

Messerlin, P. A., & Shin, W. (2013). The K-pop wave: An economic analysis. Available at SSRN

2294712.

45

Negro, G., Hannan, M. T., & Rao, H. (2010). Categorical contrast and audience appeal: Niche width and critical success in winemaking. Industrial and Corporate Change, 19(5), 1397-1425.

Oh, I., & Park, G. S. (2012). From B2C to B2B: Selling Korean pop music in the age of new social media. Korea Observer, 43(3), 365-397.

Phillips, D. J., & Kim, Y. K. (2009). Why pseudonyms? Deception as identity preservation among jazz record companies, 1920–1929. Organization Science, 20(3), 481-499.

Podolny, J. M. (1993). A status-based model of market competition. American journal of sociology, 98(4), 829-872.

Rosen, S. (1981). The economics of superstars. The American economic review, 71(5), 845-858.

Sauder, M., & Espeland, W. N. (2009). The discipline of rankings: Tight coupling and organizational change. American sociological review, 74(1), 63-82.

Shin, H., & Lee, S. A. (Eds.). (2016). Made in Korea: Studies in popular music. Taylor & Francis.

Williams Jolin, J. (2017). The South Korean Music Industry: The Rise and Success of ‘K-Pop’By:

Johan.

Zuckerman, E. W., Kim, T. Y., Ukanwa, K., & Von Rittmann, J. (2003). Robust identities or nonentities? Typecasting in the feature-film labor market. American Journal of Sociology, 108(5),

1018-1074.

Why Do K-Pop Groups Have So Many Members? Rachelle D, 2014

(https://www.kpopstarz.com/articles/114392/20140924/why-do-kpop-groups-have-so-many- members.htm)

Kpop blog (https://kknews.cc/)

46

Appendix The results models from baseline model to full model. Dependent variables are best rank for a models and ln(duration) for b models, the number of models represents:

1. model with only main independent variables

2. model with only main independent variables and control variables

3. model with only main independent variables and control variables and the interaction of status

and artist width

4. model with only main independent variables and control variables and the interaction of status

and agency width

5. model with only main independent variables and control variables and the interaction of artist

width and idol/singer

6. full model without random coefficients

1a~4a Dependent variable best rank (1) (2) (3) (4) (5) (6) Main Independent Variables status(singer) -4.396* -7.118** -11.070* -8.442 -6.958** -9.607 (2.372) (3.240) (6.054) (11.023) (3.245) (11.292) song width 3.482 26.626 27.059 26.642 26.688 26.983 (5.554) (36.850) (36.861) (36.866) (36.854) (36.885) artist width 8.077 9.813* 5.485 9.705* 8.185 5.436 (5.366) (5.803) (8.066) (5.869) (6.095) (8.105) agency width -19.924*** -14.611* -16.613* -16.127 -14.175* -15.452 (7.406) (8.071) (8.478) (14.513) (8.087) (14.542) idol-singer 2.543 4.791 4.592 4.721 -3.175 -1.673 (3.242) (3.895) (3.904) (3.936) (9.909) (10.437)

Interaction between Main Independent Variables status(singer):artist width 8.374 6.000 (10.839) (12.010) status(singer):agency width 2.082 -0.339 (16.567) (17.369) artist width:idol-singer 14.061 11.178 (16.082) (17.116)

Control Variables gender-male 6.928*** 7.095*** 6.943*** 6.841*** 6.975*** (2.238) (2.249) (2.242) (2.241) (2.258) gender-mix 3.374 3.024 3.300 3.519 3.250 (6.417) (6.434) (6.447) (6.420) (6.462)

Control of Power of Agencies agency-big 5.650* 5.626* 5.537 5.851* 5.811* (3.322) (3.323) (3.443) (3.330) (3.455) agency-big 3 5.617* 5.833* 5.640* 5.713* 5.844* (3.041) (3.055) (3.048) (3.044) (3.059)

Control of Popularity group number -2.160* -2.124* -2.162* -2.123* -2.105 (1.284) (1.285) (1.285) (1.285) (1.287)

Control of Quality main music awards -0.196*** -0.194*** -0.195*** -0.195*** -0.193*** (0.058) (0.058) (0.059) (0.058) (0.059)

Control of Intensity of Market competition 0.092** 0.095** 0.092** 0.091** 0.093** (0.041) (0.041) (0.041) (0.041) (0.041)

Control of Career Length career songs 0.023* 0.024* 0.023* 0.024* 0.024* (0.013) (0.013) (0.013) (0.013) (0.013)

Control of Popularity of Each Genre song genre-ballad -3.603 -3.854 -3.623 -3.684 -3.844 (3.021) (3.039) (3.027) (3.023) (3.042) song genre-dance -7.542** -7.874** -7.602** -7.657** -7.862** (3.418) (3.446) (3.453) (3.421) (3.469) song genre-electronical -2.352 -2.113 -2.335 -1.794 -1.741 (12.843) (12.849) (12.849) (12.860) (12.870) song genre-folk/blues -11.231* -11.693* -11.282* -11.210* -11.537* (6.507) (6.535) (6.522) (6.507) (6.548) song genre-independent -13.094 -13.513 -13.137 -12.935 -13.261 (19.148) (19.159) (19.159) (19.151) (19.176) song genre-Korean drama -8.585 -8.904 -8.603 -8.675 -8.883 (18.688) (18.696) (18.697) (18.691) (18.709) song genre-pop 1.414 1.128 1.385 1.220 1.059 (3.451) (3.471) (3.460) (3.458) (3.478) song genre-rap/hip hop 10.979** 10.469** 10.917** 10.990** 10.632** (5.114) (5.158) (5.140) (5.115) (5.175) Constant 60.100*** 49.027*** 52.019*** 50.100*** 49.573*** 51.430*** (4.712) (6.633) (7.681) (10.813) (6.663) (10.930) R2 0.007 0.05 0.051 0.05 0.051 0.051 Adjusted R2 0.003 0.033 0.033 0.033 0.033 0.032 Observations 1,204 1,204 1,204 1,204 1,204 1,204 Residual Std. Error 33.188 (df = 1198) 32.670 (df = 1182) 32.676 (df = 1181) 32.684 (df = 1181) 32.673 (df = 1181) 32.697 (df = 1179) F Statistic 1.612 (df = 5; 1198) 2.983*** (df = 21; 1182)2.873*** (df = 22; 1181) 2.845*** (df = 22; 1181) 2.881*** (df = 22; 1181) 2.648*** (df = 24; 1179) Note: *p<0.1; **p<0.05; ***p<0.01 Results of models with best rank as dependent variable

1b~4b Dependent variable ln(duration) (1) (2) (3) (4) (5) (6) Main Independent Variables status(singer) -0.251*** 0.071 0.242 -0.127 0.062 -0.070 (0.086) (0.117) (0.218) (0.398) (0.117) (0.407) song width 0.337* -0.507 -0.526 -0.505 -0.510 -0.522 (0.202) (1.330) (1.330) (1.330) (1.329) (1.330) artist width -0.109 -0.285 -0.098 -0.301 -0.191 -0.077 (0.195) (0.209) (0.291) (0.212) (0.220) (0.292) agency width 0.587** 0.334 0.420 0.106 0.308 0.071 (0.270) (0.291) (0.306) (0.524) (0.292) (0.524) idol-singer -0.140 -0.143 -0.134 -0.153 0.319 0.228 (0.118) (0.141) (0.141) (0.142) (0.357) (0.376)

Interaction between Main Independent Variables status(singer):artist width -0.362 -0.295 (0.391) (0.433) status(singer):agency width 0.312 0.430 (0.598) (0.626) artist width:idol-singer -0.815 -0.668 (0.580) (0.617)

Control Variables gender-male -0.109 -0.116 -0.107 -0.104 -0.108 (0.081) (0.081) (0.081) (0.081) (0.081) gender-mix -0.115 -0.100 -0.126 -0.123 -0.125 (0.232) (0.232) (0.233) (0.231) (0.233)

Control of Power of Agencies agency-big -0.138 -0.137 -0.155 -0.15 -0.170 (0.120) (0.120) (0.124) (0.120) (0.125) agency-big 3 -0.253** -0.262** -0.249** -0.258** -0.260** (0.110) (0.110) (0.110) (0.110) (0.110)

Control of Popularity group number 0.031 0.029 0.030 0.029 0.027 (0.046) (0.046) (0.046) (0.046) (0.046)

Control of Quality main music awards 0.015*** 0.015*** 0.015*** 0.015*** 0.015*** (0.002) (0.002) (0.002) (0.002) (0.002)

Control of Intensity of Market competition -0.003** -0.003** -0.003** -0.003** -0.003** (0.001) (0.001) (0.001) (0.001) (0.001)

Control of Career Length career songs -0.002*** -0.002*** -0.002*** -0.002*** -0.002*** (0.0005) (0.0005) (0.0005) (0.0005) (0.0005)

Control of Popularity of Each Genre song genre-ballad 0.286*** 0.297*** 0.283*** 0.291*** 0.295*** (0.109) (0.110) (0.109) (0.109) (0.110) song genre-dance 0.431*** 0.445*** 0.422*** 0.438*** 0.436*** (0.123) (0.124) (0.125) (0.123) (0.125) song genre-electronical 0.508 0.497 0.51 0.475 0.476 (0.463) (0.464) (0.464) (0.464) (0.464)

song genre-folk/blues 0.600** 0.620*** 0.592** 0.599** 0.605** (0.235) (0.236) (0.235) (0.235) (0.236) song genre-independent 0.347 0.365 0.341 0.338 0.345 (0.691) (0.691) (0.691) (0.691) (0.691) song genre-Korean drama 0.398 0.412 0.395 0.403 0.409 (0.674) (0.674) (0.674) (0.674) (0.675) song genre-pop 0.142 0.155 0.138 0.153 0.156 (0.124) (0.125) (0.125) (0.125) (0.125) song genre-rap/hip hop -0.213 -0.191 -0.223 -0.214 -0.209 (0.185) (0.186) (0.185) (0.184) (0.187) Constant 1.549*** 1.602*** 1.473*** 1.763*** 1.571*** 1.692*** (0.172) (0.239) (0.277) (0.390) (0.240) (0.394) R2 0.023 0.083 0.083 0.083 0.083 0.085 Adjusted R2 0.018 0.066 0.066 0.066 0.066 0.066 Observations 1,204 1,204 1,204 1,204 1,204 1,204 Residual Std. Error 1.209 (df = 1198) 1.179 (df = 1182) 1.179 (df = 1181) 1.179 (df = 1181) 1.178 (df = 1181) 1.179 (df = 1179) F Statistic 5.523*** (df = 5; 1198) 5.077*** (df = 21; 1182)4.885*** (df = 22; 1181) 4.856*** (df = 22; 1181) 4.940*** (df = 22; 1181) 4.553*** (df = 24; 1179) Note: *p<0.1; **p<0.05; ***p<0.01 Results of models with ln(duration) as dependent variable

status song_width artist_width agency_width idol_singer gender status 1 0.245259779 0.087585649 -0.305124322 0.281105566 0.008820604 song_width 0.245259779 1 0.163037161 0.004613976 -0.021641988 -0.064548213 artist_width 0.087585649 0.163037161 1 0.432987424 0.158982062 -0.086533949 agency_width -0.305124322 0.004613976 0.432987424 1 0.175417434 -0.089713233 idol_singer 0.281105566 -0.021641988 0.158982062 0.175417434 1 -0.143856515 gender 0.008820604 -0.064548213 -0.086533949 -0.089713233 -0.143856515 1 agency -0.398684636 -0.184142391 0.129161337 0.464760519 0.155894009 0.030683824 group_number 0.303984623 -0.020550701 0.10731437 0.066240119 0.56501772 -0.015193136 main_music_awards -0.555375166 -0.169082553 0.144264444 0.297525476 -0.161919836 0.0838168 competition -0.053951498 -0.042850128 0.018400405 0.030779495 0.039491241 0.064322399 career_songs -0.192069501 -0.021135653 0.17917784 -0.007500454 -0.191710491 0.117233601 song_genre_ballad 0.228305982 0.271129763 0.065669965 -0.054864594 0.009464759 -0.016644892 song_genre_dance -0.49470981 -0.213996218 -0.038179494 0.178452852 -0.038751071 -0.171827227 song_genre_electronical -0.054607104 -0.032560127 0.041529414 0.064567772 0.005693433 0.042888249 song_genre_folk_blues 0.138087154 0.301110038 0.134484394 0.08229016 -0.042984478 -0.006943062 song_genre_independent 0.13148438 0.562346666 0.153278729 0.037851939 -0.077133325 0.005385297 song_genre_Korean_drama 0.206099866 0.76724602 0.080397693 -0.028105284 0.03440952 -0.085388672 song_genre_pop 0.129154326 -0.142490782 0.119647127 0.002778019 0.100684434 -0.034889658 song_genre_rap_hip_hop 0.065363208 0.099440316 0.090062924 0.016379022 -0.033485053 0.070971143

agency group_number main_music_awards competition career_songs song_genre_ballad status -0.398684636 0.303984623 -0.555375166 -0.053951498 -0.192069501 0.228305982 song_width -0.184142391 -0.020550701 -0.169082553 -0.042850128 -0.021135653 0.271129763 artist_width 0.129161337 0.10731437 0.144264444 0.018400405 0.17917784 0.065669965 agency_width 0.464760519 0.066240119 0.297525476 0.030779495 -0.007500454 -0.054864594 idol_singer 0.155894009 0.56501772 -0.161919836 0.039491241 -0.191710491 0.009464759 gender 0.030683824 -0.015193136 0.0838168 0.064322399 0.117233601 -0.016644892 agency 1 0.205229832 0.442786177 0.019512297 0.155716307 -0.147334954 group_number 0.205229832 1 -0.074703712 0.053518136 -0.02762261 -0.010814918 main_music_awards 0.442786177 -0.074703712 1 0.048887637 0.481366663 -0.187684302 competition 0.019512297 0.053518136 0.048887637 1 0.112932802 0.036122232 career_songs 0.155716307 -0.02762261 0.481366663 0.112932802 1 0.182675608 song_genre_ballad -0.147334954 -0.010814918 -0.187684302 0.036122232 0.182675608 1 song_genre_dance 0.231393018 -0.078216545 0.171334474 0.01735363 -0.027586869 -0.389069986 song_genre_electronical 0.115231494 0.036047339 0.036686592 0.001744018 0.10118793 -0.056279031 song_genre_folk_blues -0.067391372 -0.042421688 -0.057831626 0.015780352 -0.053287283 -0.141632456 song_genre_independent -0.158215556 -0.059408019 -0.098480047 0.004137414 -0.078443066 -0.054938664 song_genre_Korean_drama -0.111005582 0.021769606 -0.147551378 -0.059667715 0.028619627 0.371829531 song_genre_pop 0.020366843 0.115456273 -0.026218316 -0.024220762 -0.051682508 -0.28946 song_genre_rap_hip_hop -0.074477605 0.000716195 -0.069065271 -0.119565883 -0.117101379 -0.164059123

song_genre_dance song_genre_electronical song_genre_folk_blues song_genre_independent song_genre_Korean_drama song_genre_pop song_genre_rap_hip_hop status -0.49470981 -0.054607104 0.138087154 0.13148438 0.206099866 0.129154326 0.065363208 song_width -0.213996218 -0.032560127 0.301110038 0.562346666 0.76724602 -0.142490782 0.099440316 artist_width -0.038179494 0.041529414 0.134484394 0.153278729 0.080397693 0.119647127 0.090062924 agency_width 0.178452852 0.064567772 0.08229016 0.037851939 -0.028105284 0.002778019 0.016379022 idol_singer -0.038751071 0.005693433 -0.042984478 -0.077133325 0.03440952 0.100684434 -0.033485053 gender -0.171827227 0.042888249 -0.006943062 0.005385297 -0.085388672 -0.034889658 0.070971143 agency 0.231393018 0.115231494 -0.067391372 -0.158215556 -0.111005582 0.020366843 -0.074477605 group_number -0.078216545 0.036047339 -0.042421688 -0.059408019 0.021769606 0.115456273 0.000716195 main_music_awards 0.171334474 0.036686592 -0.057831626 -0.098480047 -0.147551378 -0.026218316 -0.069065271 competition 0.01735363 0.001744018 0.015780352 0.004137414 -0.059667715 -0.024220762 -0.119565883 career_songs -0.027586869 0.10118793 -0.053287283 -0.078443066 0.028619627 -0.051682508 -0.117101379 song_genre_ballad -0.389069986 -0.056279031 -0.141632456 -0.054938664 0.371829531 -0.28946 -0.164059123 song_genre_dance 1 -0.040428256 -0.101742213 -0.127316268 -0.174264592 -0.215095318 -0.117852495 song_genre_electronical -0.040428256 1 -0.014717026 -0.018416317 -0.025207399 -0.03111357 -0.017047381 song_genre_folk_blues -0.101742213 -0.014717026 1 0.504196078 -0.018278524 -0.078300766 -0.042901634 song_genre_independent -0.127316268 -0.018416317 0.504196078 1 -0.079382894 -0.087529688 0.186857314 song_genre_Korean_drama -0.174264592 -0.025207399 -0.018278524 -0.079382894 1 -0.110106644 -0.020866918 song_genre_pop -0.215095318 -0.03111357 -0.078300766 -0.087529688 -0.110106644 1 -0.090699233 song_genre_rap_hip_hop -0.117852495 -0.017047381 -0.042901634 0.186857314 -0.020866918 -0.090699233 1 Correlation table of all variables included in the model