Why Do People Contribute Flowers? --A field study of a public good voluntary provision phenomenon in street show market Yunfeng Lu CCER

Xing Zhang JNULab

1

Abstract: In our research, we investigated a street art show market in Zhuhai city in China. In this market, people contribute flowers to singers voluntarily, and the flower selling is the main income of the market. Standard microeconomics predicts that there will be no contribution at all, but this market has become a large industry in this city. Equipped with behavioural economics research and econometric tools, in our paper we attempt to solve this puzzle. Key words: public good provision; Heckman Choice Model; field experiment; gift exchange; showing fondness

I. Introduction Zhuhai is a city with booming industry. On the outskirts of the city, there are many factories with workers from the countryside. In order to provide some entertainment to the workers who have low salaries and earn a good reputation, the government set up the “people’ stage”. Actually, the government does nothing more than providing the place for performance. There are 5 to 10 singers take performance every night. Some of them belong to one group which can take performances legally in the place provided by the government. The group has the necessary equipment for the performance. When the show begins, there will be a flower seller walking around to get attentions of the audiences. If one audience wants to contribute flower to the singer, he just waves his hand, and the seller will come to him and sell the flower. The flowers are very cheap (1CNY each) and are made of plastic which can be used again and again. After buying the flower, the audience waves the flower, and the singer comes down from the stage and goes straight to the one who wants to contribute flower. When the singer receives it, she will shake hand with that audience. Then the flower will be given to the seller for reselling. After investigation, we found the mechanism is very surprising: the income of the performance totally depends on the flowers the singers obtain. Half of the revenue from flower selling is divided to singers as 1

This research is financially supported by Economics Lab in Jinan University. The authors thank Colin Camerer, Yan Chen, Yaohui Zhang, Huojun Sun for their helpful comments and discussion.

their personal income. There are almost 400-600 people who watch the show every night and the group can earn more than 1000 flowers a night! It seems like that the flower had the magic power. Because of the flower contribution mechanism, the “people’s stage” has becomes a large industry in this city. As far as we know, only in our community, there are more than 4 shows every night as the “people’s stage” within 1 square kilometres i.e. there are considerable shows using the same mechanism in our city! We think it is a public goods provision phenomenon in the field. The microeconomics theory predicts that there will be no contribution at all! So we are puzzled:” Why do people contribute flowers?” In this research, we want to find out what is the motivation of the audience to contribute flowers. There are five parts in our paper, the first part is introduction; the second is the hypothesis about the motivation of contribution; the third part is the econometric analysis of the question; the last part is the conclusion and further research prospect. II. Hypothesis of the motivations After interview with the singers, organizers and the government officers, we implemented a survey using questionnaires in June 2007. Based on literature of behavioral economics and experimental economics, we have the following hypothesis about the motivation: Hypothesis 1. Pure altruism; Audiences know that all the income of the singers is from the contribution of flowers, and they contribute flowers just because they want the show to be continued. In many existing public good experiment literature, this social preference is apparent (see Kagel & Roth，1995 and Andreoni, 1990). If this preference is the motivation, people will not care which singer get the flowers, and people will give flowers to singers randomly. Hypothesis 2. Gift exchange; Flower is a kind of reward for the music provided by the singers. Gift exchange phenomenon was first studied by anthropology research. In Mauss (1954), the author described the social pressure surrounding the gift-giving process in human culture. He said that there is an obligation to give, an obligation to receive, and an obligation to repay. In Regan (1971), the gift exchange study indicated that a small initial favor can produce a sense of obligation to agree to a substantially larger return favor. Akerlof（1982）adopted this conception into economics to explain the anomalies in labor market wages. This hypothesis was confirmed both in laboratory (Fehr et al., 1993) and field experiment (Gneezy et al., 2006). If the flower is a kind of gift to singers,

the amount of flowers is correlated to the quality of the performance of the singers. Hypothesis 3. Herd behavior; In The Theory of the Leisure Class, Thorstein Veblen explained economic behavior in terms of social influences such as "emulation," where some members of a group mimic other members of higher status. Behavioral finance study identified the herd behavior in collective irrationality of investors (see Shiller, 2000 and Banerjee, 1992). In this art show market, people imitate others and feel bad when others contribute but they do not. When most of the people who sit there contribute, it is a shame that one never contribute or contribute very few. To some extent, they want to pretend to be more rich and generous than others. Interestingly, in some place, the organizer hires some “fake audiences” to lead the herds, and they believe this trick can increase contributions. If this hypothesis is true, the mount of people’s contribution will be affected positively by others’. Hypothesis 4. Showing fondness; We believe this is a preference which needs more attention. In order to show fondness to one’s favorite writers, singers, actors, people often spend a lot of money on it! In “Super Girl Contest” in China, people can use short message to vote for their favorite singers. By doing so, two telecommunication companies of China earned millions of money. III. Models and Results 1. Assumptions of our models Heckman Choice Model Our questionnaire is designed to investigate the motivations such as the gift exchange, pure altruism, showing fondness, and herd behavior while these questions are effective when people have contributed flowers. Therefore, our questionnaire is divided into two part (refer to the appendix), and only if people said that they had contributed flowers, they could answer the questions in the second part about their motivations of flower contribution. In order to avoid the sample selection bias, we adopt the Heckman Choice Model (In Amemiya(1985), it is called II Tobit Model). y1 = X1β1 + u1 L L L (1)

y2 = 1[ X 2 β2 + v2 > 0]L L L L (2)

(u1, v2 ) ~ bivariate..normal (0,0,σ ,1, ρ )L L L (3)

In the model, X 2 is the variables which affect whether the audience contribute flowers; X 1 includes variables in X 2 and the group of variables in second part of our questionnaire about the motivations of flower contribution; the distribution of u1 and v2 satisfy the assumptions in standard Heckman Choice Model. In this model, both MLE and Heckman Two steps Estimation can be adopted. In our analysis, we use MLE. OLS Model If we ignore the sample selection bias problem, we can do the regression using OLS estimation. Because OLS need not to satisfy the normal distribution assumption in large sample, and it is more robust. So we also adopt an OLS model 2. Data In order to avoid other factors to interfere our analysis, we choose only one market to implement our investigation. This market is located in west part of Zhuhai City which has the most audiences. The performance group includes one organizer, seven singers one DJ and one flower seller. In general cases, there are about 100 audience sitting around the stage while about 500 standing around (see picture 1). We chose 5 days randomly in one month. And at one night, the investigations were done at 19:30, 20:00, 20:30, 21:00, 21:30 due to the unstable population in the market.

Figure 1

There are 122 observations in our original investigation. Because people can refuse to answer the questions which are related to their privacy, we omitted the incomplete questionnaires. So there are 103 final observations in the analysis. The main descriptives are presented below in Table 1: Table 1 Descriptives from the investigation Variables

Mean

Stdev

Max.

Min.

Average contribution per night (Yuan)

21.13208

46.66723

0

400

35.32039 2575.472

10.09 1110.11

17

60

500

4500

Walking time(Minutes)

1.88283

1.061138

．16

5

resident （resident_d）

.1792453

.3853797

0

1

Whether married

.6603774

.475831

0

1

.7735849

.420499

0

1

.1320755

.3401812

0

1

(negative_inf)

.6509434 .1037736 .1037736 .1132075 .1132075 .0283019

（fixed）

.5377358

.4789364 .3401812 .3064154 .3931123 .3183515 .1666217 .8113208

mode3

.1132075

.3183515

mode4

.1603774

.3686989

0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 1 1

Age Income

Whether contribute (contri) Contri for reason 1(number1) (number2) (number3) (number4) (quality) (positive_inf)

Explanation for the dummy variables: 1. marry_d: If people get married, marry=1; if not, marry= 0. 2. Residen_dt: if people are resident, Resident=1; if not, Resident= 0. 3. quail_d: if people will contribute more when the show becomes better, quail_d = 1; if not, quail_d = 0. 4. fixed: if the audience contribute flowers to particular singers, fixed= 1; if not, fixed= 0. 5. contri: if the audiences have ever contributed flowers to the singers, contri=1; if not, contri= 0. 6. positive_influ: if one’s amount of contribution is affected by others’

positively, positive_influ=1; if not, positive_influ= 0. negative_ influ: if one’s amount of contribution is affected by others’negatively, negative_ influ= 1; if not, negative_ influ= 0. 7. Perception of the profit mode of the market. Mode 1: if interviewee thinks that the income of the singers is from sponsor of the government, Mode 1= 1; if not, Mode 1= 0. Mode 2: if interviewee thinks that the income of the singers is from audiences’ flower contribution, Mode 2=1; if not, Mode 2= 0. Mode 3: if interviewee think that the singers are volunteers who do not get money by singing songs here, Mode 3= 1; if not, Mode 3= 0. Mode 4: if interviewee think that the income of the singers is from deduct from the beer and beverage consumption, Mode 4= 1; if not, Mode 4= 0. 8. The factors which affect the amount of contribution. number 1: if the amount of contribution is dependent on whether the singers are beautiful (handsome), number 1=1, if not, number 1=0; number 2: if the amount of contribution is dependent on whether the singers sings well, number 2=1, if not, number 2=0; number 3: if the amount of contribution is dependent on whether the singers know the audience well, number 3=1, if not, number 3=0; number 4: if the amount of contribution is dependent on the recent income of the audience, number 4=1, if not, number 4=0. 9. The reasons for contribution. contri1: if people’ contribution is due to the music, contri1=1, if not, contri1=0; contri2: if people’ contribution is due to some particular singers, contri2=1, if not, contri2=0; contri3: if people’ contribution is due to the opportunity to communicate with the singers, contri3=1, if not, contri3=0; contri4: if people think that contribution is way to show their generosity, contri4=1, if not, contri4=0; contri5: if people’ contribution is due to peer pressure, contri5=1, if not, contri5=0; contri6: if people’ contribution is due to their pure kindness, contri6=1, if not, contri6=0. 10. The reasons for watching the show. reason1: if people come here to listen the music, reason1=1, if not, reason1=0; reason2: if people come here because there is no other amusement around here, reason2=1, if not, reason2=0; reason3: if people come here to meet friends, reason3=1, if not, reason3=0; reason4: if people come here because other kinds of amusement around here are too expensive, reason4=1, if not, reason4=0; reason5: if people come here to communicate with the singers, reason5=1, if not, reason5=0; reason6: if people just walk around here, reason6=1, if not, reason6=0. 2. Results In our model, we focus on which factors will affect the average flower contribution( ave_m ) i.e. which variables compose X i in

equation (1)and (2). According to our hypothesis, we think X i include the following variables: Table 2 Variables in X i Variables groups

X1

X2

Variables

Demographics: marry_d, age, income Other ther preferences: number1，number2，number3，number4 fixed, positive_inf, negative_inf quali_d, reasons4, contri3, Perceptions: mode4 Demographics: marry_d, age, income Preference： Preference：reason4, reasons6, Perception： Perception： mode3

Corresponding to Model (1), we have the main Heckman regression equation: ave_m= β0 + β1*age+β2 *income+β3 *marry_d +β4 *reason4+β5 *contri3+β6 *mode4

+ β7 *number1+β8 * number2 +β9 * number3 + β10 * number4

+ β11*positive_influ+β12 *negative_influ+β13 *fixed+β14 *quali_d+u1

(4)

Corresponding to Model (2), the choice equation is: contri = β0 + β1*income+β2 * reasons4+β3 * reasons6 + β4 *marry_d +β5 *mode3+v 2

(5)

Table 3 results Dependent Variables

(1) Heckman MLE With robust variance

Age

income

marry_d

positive_i~u

negative_i~u

.0019587 （ .0040672) -15.20555 (10.32394) 12.0377 (14.11962) .688851 (24.32677)

(3) OLS with robust variance

(4) OLS with bootstrap variance

-.0334723 ( .6892158) .0043014 （.0043555） -20.28442 （14.12032） 13.08878 ( 17.42511) -4.204743 (11.50134)

-.0334723 .8500927 .0043014 ( .0052454) -20.28442 (16.55168) 13.08878 (18.80806) -4.204743 (18.91085)

fixed

quali_d

number1

number2

number3

number4

reasons4

contri3

mode4

-cons

22.94252 ** (10.46703) 10.46703) 3.285569 (12.05315) -11.08074 (14.44503) 36.0515 *** (14.60583) 14.60583) 63.49051 (15.15887) 15.15887) *** 21.38087 (14.79375) -44.1748 *** (14.48712) 14.48712) 37.86851 ** (17.59694) 17.59694) -11.73644 (11.72855) -11.90735 （19.06757）

25.30505 (12.36529) 12.36529) 5.46145 (10.97788) -16.25471 (14.20845) 33.26014 (25.98305) 66.33529 (22.53146) 22.53146) 17.55819 (20.12959) -40.63734 (21.11863) 21.11863) 37.3003 (23.16036) -14.39242 (10.75059) -16.42409 (31.88713)

**

25.30505 (12.80851) 12.80851) 5.46145 (15.04234) -16.25471 (22.45985) 33.26014 (29.31784) 66.33529 *** (32.91616) 32.91616) 17.55819 (20.55774) * -40.63734 (28.62522) 37.3003 (29.77673) -14.39242 (12.27132) -16.42409 (37.97272)

**

**

Choice Equation

income2

reasons6

reasons4

marry_d

mode3

-cons

.0001832 （ .0001438） -1.767802 *** （ .5175488） .5175488） 1.329281 ** ( .670871 ) -.3744399 （ .3420961） .5700923 （ .5453846） .5298304 （ .4178028）

0.4642

Adj. R-square

Number of obs

ρ

106 -.3990096 (.2578225)

79

0.3470

79

0.2955

Note：* represents 10% significant level，** represents 5% ，*** represents 1% significant level

From results of the model, we may find: 1) The dummy variable number3 is significant (p<0.05), which indicates that holding other factors constant, if the singers acquaint the audiences, the amount of contribution will increase by 66.335 yuan. 2) The dummy variable fixed is significant (p<0.05), which indicates that holding other factors constant, if the audiences’ contribution is fixed to one or several singers, the amount of contribution will increase by 25.3 yuan. 3) The dummy variables postive_inf,negative_inf,quail_d which indicate social preference are not significant. In order to test our hypothesis about the motivation, we joint correlative variables and use Wald test to see whether the joint variables are significant. The Wald test is based on the OLS model which adopts bootstrap. Table 4 Joint Test Hypothesis

Variables

Gift exchange

quali_d number2

2.7

refute

Pure altruism

mode4 negativeinf

1.16

refute

7.09 *

support

0.51

refute

Showing fondness Herd behavior

fixed

number3

Wald test

marry_d

positiveinf

Results

We find that in the joint test above, the hypothesis of showing fondness is significant under 10% significance level (p=6.9%) while other three hypothesis are refuted.

IV Discussion and Prospect As we studied above, the main motivation of contribution is people’s preference of showing fondness which coincides with our intuition. The main contributions of our research are: first, we studied the motivation of contribution of public good using field data; second, we included the social interaction effect on the people’s contribution behavior, which improved the existing research about charity. However, the internal validity of our research is flawed because we did

not control the problems of omitted variable bias and simultaneous causality. These problems threaten the conclusion of our research. Harrison and List （ 2004 ） pointed out that a well controlled field experiment can overcome the problems above. It is free of internal and external validity threats. Therefore, field experimental method is adopted by more and more researchers. We believe that ranking mechanism implemented in the field in next step may help us. In the experiment, we announce how many flowers the singer totally gets accumulated to her current song. The ranking mechanism makes the stage become a singing championship in which audiences vote for their favorite by the purchase of flowers. In a month, the treatment will be randomly assigned to 15 days. And then we can compare the average amount of flower contribution between the controlled days and treatment days. If the hypothesis “showing fondness” is correct, audiences will contribute more flowers to singers.

References Akerlof, Gorege. 1982, Labor Contracts as Partial Gift-Exchange, Quarterly Journal of Economics, 97, pp.543-569. Amemiya, Takeshi 1985, Advanced Econometrics, Harvard University Press Andreoni, James, Impure Altruism and Donations to Public Goods: A Theory of Warm-Glow Giving. Economic Journal, v.100, June 1990, 464-477.

Banerjee, Abhijit A simple Model of Herd Behavior. Quarterly Journal of Economics, 1992, 107(3): pp. 797-817. Gneezy, Uri and John List. 2006, Putting Behavioral Economics to Work: Field Evidence on Gift Exchange, Econometrica, Vol. 74, 5, pp.1365-1384. Harrison, Glenn and John A. List. 2004 ，Field Experiments, Journal of Economic Literature, XLII (December): 1013-1059 Kagel, John & Alvin Roth, 1995, The Handbook of Experimental Economics, Volume 1, Princeton University Press Mauss, Marce.l 1954. The Gift. London: Conhen and West Press Regan, Dennis. 1971. Effects of a favor and liking on compliance. Journal of Experimental Social Psychology, 7, pp.627-639 Robert J. Shiller. 2000, Irrational Exuberance, Princeton University Press, pp. 149-153. Thorstein Veblen. 1899，The Theory of the Leisure Class. An Economic Study in the Evolution of Institutions. New York, London: Macmillan Wooldridge, Jeffrey. 2000, Econometric Analysis of Cross Section and Panel Data, MIT Press

Appendix Questionnairs Excuse me! We are college students who are interested in the mechanism of the street art show market! Could you spent 2 minutes to fulfill this questionnaire? Your answer will be confidential. And if you are willing to do it, you will receive one tin of Coca Cola. Your answer is vital for our research! And we hope you can do us this favor. Thank you! - Your age_____ - Your Occupation is__________________ -How long it will take you to walk here from your home________ minutes -Are you resident of Zhuhai City A. Yes B. No 1. Your income level is A. above 4000 yuan B. 3000-4000 yuan C.2000-3000 yuan C. 1000-2000 yuan D. below 1000 yuan 2. Marital status: A. single B. married, 3. In this month, how many times you have been here?____________ 4. Why do you come here? A. Enjoy the music B. There is no other entertainment around here C. Meet with friends D. For its price. It is cheaper than the other alternatives (because I can attend the show for free). E. Because the audience is involved in the show (we can communicate with the singers) F. Just walk around here 5. Which source does the singers’ income come from? A. From sponsor of the government B. Audiences’ flower contribution C. They are voluntary singers who do not get money by singing songs here D. Deduct from the beer and beverage consumption. - Have you ever contributed to the show? (Yes or No) If yes, please answer the following questions: 6. Why do you decide to buy a flower? A. Because I liked the music. B. Because I liked one or several particular singers. C. Because due to flower contribution, I can communicate with the singers. D. To appear as a generous person. E. Because other audiences contribute flowers, if I do not, I feel bad. F. If we do not contribute, the market cannot exist any more. G. Other reasons____________________________________ 7. Do you always contribute flowers to one or several particular singers?

A. Yes B. No 8. How many flowers do you usually contribute on average every night?_______ How many flowers have you ever contributed at most once a time?_________ 9. How long do you stay here to watch the show?________ 10. Which factors affect your decision of flower contribution? A. Whether the singers are beautiful (handsome) B. Whether the singers sing well C. Whether the singers know you D. Recent income of myself 11. If you see that other people from the audience contribute, do you feel motivated to contribute more or less flowers? A. More B. Less C. Indifferent 12. Will you contribute more flowers when the show become more excellent? A, Yes B. No Thank you again for your assistance!