Are Hotels in Destination Competitive or Cooperative?:

An Empirical Application of Social Network Analysis

Masayuki KanaiSenshu University, Japan

mkanai@isc.senshu‐u.ac.jp

1

The Fifth Joint Japan‐North America Mathematical Sociology ConferenceAugust 16, 2012, Denver CO, USA

Organization of the Presentation

• Puzzle: Competitive or cooperative?:Inter‐firm relations in tourist destinations

• Theoretical argument: Centrality indices on inter‐firm network

• Data and methods• Findings: Cooperative rather than competitive• Conclusions

2

Visit Japan!   Visit Hot Springs!

3

1980 1985 1990 1995 2000 2005 2010

2000

6000

Visitor Arrivals in Japan (1980-2011)

Year

Vis

itor A

rriv

als

(Tho

usan

d)

Source: Japan National Tourism Organization,the Ministry of Justice

Source: Japan Travel Bureau Foundations (2007)

2007  The enactment of the Tourism Nation Promotion Basic Law2008  The establishment of Japan Tourism Agency

1317

8611

Hot springs Nature Food History Seashore Theme park

Favorite Type of Tourism (Multiple Answer)

Per

cent

010

2030

4050

1998 2000 2002 2004 2006 2008 2010

500

600

700

800

900

1000

Number of Rooms in Hotels (1997-2010)

Year

Roo

ms

(thou

sand

)

Hotel styleJapaneseWestern

Stay at RYOKAN?

4

A hotel in the style of traditional Japan(Ginzan hot springs, Yamagata)

Source: the Ministry of Health, Labour and Welfare

Decline of Japanese style hotels in hot springs destinations

982,228

764,316

Inter‐firm Relations in Destinations

• Competition– Share visitors to the destination

• Cooperation– Collaborative network to attract more tourists for the destination

– To win the nation‐wide or global competition among destinations

5

A Puzzle: Competition or Cooperation?

Centrality on Inter‐firm Network

6

betweenness centrality degree centrality

competitive cooperative

If inter‐firm networks are…

… will bring more profits to each hotel.Because of… 

power to control information flow within inter‐firm network

deep involvement in collaborative network for promotion

Data

• Questionnaire Surveyfor Hotels in Hot Springs Destinations– Population: Purposively selected 4 prefectures which have various types of hot springs destinations

– Target: ALL Hotels in ALL hot springs destinations in the 4 prefectures

– Reliable responses were 779 (51.4%).– Mail survey conducted in 2007.

7

Variables and Methods• Unit: Hotel (N=779)• Dependent Variable:

– Profit change compared to five years ago• Independent Variables:

– Betweenness centrality (normalized within destination)– Degree centrality (normalized within destination)

• Control Variables:– Room rate (high or low; dummy variable)– Management efforts

• Method– OLS regression

8

Centrality Indices (1)

• Specify the whole inter‐hotel network structure in each destination

• Derived from participations in the events for promotion held during the last year

9

EventHotel E1 E2 E3 E4 E5H1 1 0 1 0 1H2 1 1 1 0 0H3 1 0 0 0 1H4 1 0 0 0 1H5 1 1 1 1 1H6 1 0 1 0 0H7 1 0 1 0 1H8 0 0 0 0 0

HotelHotel H1 H2 H3 H4 H5 H6 H7 H8H1 3 2 2 2 3 2 3 0H2 2 3 1 1 3 2 2 0H3 2 1 2 2 2 1 2 0H4 2 1 2 2 2 1 2 0H5 3 3 2 2 5 2 3 0H6 2 2 1 1 2 2 2 0H7 3 2 2 2 3 2 3 0H8 0 0 0 0 0 0 0 0

Incidence Matrix Adjacency Matrix

Note: 1 = participated in,  0 = not participated in

Centrality Indices (2)

• Calculate betweenness and degree centrality for each hotel

• Normalize indices within each destination

10

Before NormalizationCentrality

Hotel Betweenness DegreeH1 1.33 6H2 0 4H3 0 4H4 0 4H5 1.33 6H6 0 4H7 1.33 6H8 0 0

CentralityHotel Betweenness DegreeH1 1.21 0.88H2 ‐0.72 ‐0.13H3 ‐0.72 ‐0.13H4 ‐0.72 ‐0.13H5 1.21 0.88H6 ‐0.72 ‐0.13H7 1.21 0.88H8 ‐0.72 2.14

After Normalization

Descriptive Statistics

11

-3 -2 -1 0 1 2 3

Profit Change (N = 757)

Prop

ortio

n (%

)

05

1525

0 1 2 3 4 5 6 7 8

Management Effort (N = 765)

Prop

ortio

n (%

)

05

1020

Betweenness Centrality (N=575)

Prop

ortio

n (%

)

-1 0 1 2 3 4

010

2030

Degree Centrality (N=738)

Prop

ortio

n (%

)

-4 -3 -2 -1 0 1 2

010

2030

40

Profit Betweenness Degree EffortsProfit 1.000Betweenness 0.087* 1.000Degree 0.102** 0.647*** 1.000Efforts 0.252*** 0.162*** 0.157*** 1.000

Correlation Coefficient Matrix

• Significantly positive simple correlation– between Profit and Betweenness centrality– between Profit and Degree centrality

• High simple correlation between Betweenness and Degree centrality (‐> multicollinearity)

12

Note: * < .05, ** < .01, *** < .001.

OLS Regression

13

Note: Dependent variable is Profit Change. “Rate” means higher rate dummy.+ < .10, *** < .001.

• Betweenness centrality is NOT significant.• Degree centrality is significant.

HypothesisCompetition Cooperation

Coef. S.E. Coef. S.E.Betweenness .034 .051Degree .079 + .044Rate .502 *** .102 .482 *** .088Efforts .138 *** .035 .127 *** .030Constant ‐1.639 *** .115 ‐1.670 *** .097adj. R2 .095 * .101 *N 538 690

Findings

• Score of betweenness centrality does not have a significant effect on profit change.– Competition hypothesis was rejected.

• Score of degree centrality has a significant effect on profit change– Cooperation hypothesis was accepted.

14

Conclusions

• Inter‐firm relations in tourist destinations are cooperative rather than competitive.

• Active involvement in the collaborative network for destination development is beneficial for the management performance of each hotel.

• An example of empirical application of social network analysis.

15

Thanks!

16

Your Comments are Welcomed!

This work was supported by MEXT/JSPS KAKENHI Grant Number 23614020.

Interaction betweenDegree Centrality and Price

17

Model 1 Model 2Coef. S.E. Coef. S.E.

Degree centrality .079 + .044 .175 ** .059Rate .482 *** .088 .486 *** .088Degree x Rate ‐.212 * .088Efforts .127 *** .030 .132 *** .030Constant ‐1.670 *** .098 ‐1.673 *** .097adj. R2 .101* .108 *AIC 2075.6 2072.1

Degree Centrality

Prif

it C

hang

e

-2

-1.5

-1

-0.5

-3 -2 -1 0 1

: highprice.c low

-3 -2 -1 0 1

: highprice.c highNote: Dependent variable is Profit Change.N = 690. + < .10, * < .05, ** < .01, *** < .001.

Centrality Indices:Weight Transformation of Ties

• In each adjacency matrix of destination, • Ties with weights greater than median were remained but with weight value 1,

• Ties with weights less than median were deleted.

18

Before Transformation After Transformation

Note: Median of weight in this network is 2.

Population Prefectures

19

Yamagata

Niigata

NaganoGunma

Sampling Design

20

PrefectureNumber ofDestinations

Number ofHotels

ReliableResponses

ResponseRate

Nagano 21 597 312 52.3%Yamagata 12 224 117 52.2%Gunma 9 329 169 51.4%Niigata 14 365 181 49.6%Total 56 1,515 779 51.4%Note: Mail Survey conducted on January to February 2007.

Descriptive Statistics

21

Variables Values MeanStandardDeviation N

Profit change comparedto 5 year ago

‐3 ~ 3 ‐1.017 1.151 757

Betweenness centrality ‐1.339 ~ 3.693 0.000 0.970 575

Degree centrality ‐3.615 ~ 1.529 0.000 0.966 738

Room rate 0(low), 1(high) 0.519 0.500 756

Efforts 0 ~ 8 3.030 1.487 765

Profit Change

22

Profit Change

%

0

10

20

30

40

-2 0 2

low price

-2 0 2

high price

Betweenness Centrality

23

Betweenness Centrality

%

0

10

20

30

-1 0 1 2 3 4

low price

-1 0 1 2 3 4

high price

Degree Centrality

24

Degree Centrality

%

0

10

20

30

40

50

-4 -3 -2 -1 0 1

low price

-4 -3 -2 -1 0 1

high price

Management Efforts

25

Management Efforts

%

0

10

20

30

0 2 4 6 8

low price

0 2 4 6 8

high price

Scatterplot Matrix

26

Profit-1 1 3 0 2 4 6 8

-3-1

13

-11

3 Between

Degree

-3-1

1

-3 -1 1 3

02

46

8

-3 -1 1

Effort