An Experiment and Estimation of Market-Incentive Effects on Altruism Appendix

Appendix A Materials for the experiment A.1 Instructions You are taking part in an economic decision-making experiment. Please carefully read the instructions. It is very important that you do not speak with other participants for the duration of the experiment. If you break these rules, you could be excluded from the experiment and not receive any payment. If you do not understand something, please take another look at the instructions. If you still have questions, please raise your hand. We will come to you at your cubicle and answer your questions in private. You can earn money in the course of the experiment. The amount of your earnings depends on your decisions and decisions made by other participants. At no time will you be told the names of the other participants. They will also not at any time be informed about your identity. For showing up you will receive a fee of EUR 2.50. All monetary amounts in this experiment are expressed in Taler, whereby the following applies: Taler 100 = EUR 1. At the end of the experiment, the amount of money you earned will be paid to you in cash. Your decisions are made on the computer screen present in your cubicle. All data and answers will be evaluated anonymously. You were asked to draw your own personal cubicle number in order to maintain anonymity. The experiment will last around 60 minutes and consists of three parts. Before each of the three parts you will receive detailed instructions and be asked to answer control questions pertaining to these instructions. Please note: Neither your decisions in the …rst part nor in the second part of the experiment have an in‡uence on the other parts of the experiment. We will ask you to answer a few questions at the end of the experiment. You will receive an additional payment for answering this questionnaire. First part of the Experiment. In the …rst part of experiment, you will take on the role of a physician and make decisions about the treatment of various patients. In total, you will determine the quality of care that you would like provide for eight di¤erent types of patients. For each of these patients you can choose quality of 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10. The demand for medical care by the various patient types is determined only after you have made your decisions about the quality of care for all eight types. [Duopoly: You are randomly matched with another participant. This participant also decides in the role of a physician. Also this physician determines the quality for the same eight types of patients. The matching with this participants remains throughout the entire second part of the experiment. You and the other physician chose the quality simultaneously and independently from each other.] [Quadropoly: You are randomly matched with three other participants. These participants also decide in the role of a physicians. Also these physicians determine the quality for the same eight types of patients. The matching with these participants remains throughout the entire third part of the experiment. You and the other physicians chose the quality simultaneously and independently from each other.] In total, 100 patients of each type demand medical care. It will only be determined after you have made your decisions about the quality of care for all eight types how many of the 100 patients of each type wish to seek treatment from you. [Duopoly: Only after you and the other physician, you are matched with, decided upon the quality of medical treatment for the eight patients, it is determined how many of the 100 patients seek treatment from you and the other physician.] [Quadrupoly: Only after you and the others physicians, you are matched with, decided upon the quality of medical treatment for the eight patients, it is determined how many of the 100 patients seek treatment from you and the other physicians.] Earnings. For each patient who seeks medical care from you, you receive a lump sum that is independent of the quality of care you have selected. You incur costs with your selection of the quality of care. These costs depend on the quality level you choose and can vary between the di¤erent patient types. Your earnings for each patient type are as follows: Earnings = (Lumpsum-Costs) Number of patients who seek medical care from you (when read: your earnings are equal to the di¤erence between the lump sum and the costs that arise from the quality of care you have chosen, multiplied by the number of patients who seek treatment from you.) With the quality of care you choose, you determine not only your own earnings, but also the utility enjoyed by the patient. The amount of the lump sum, your costs, your earnings, and the patient’s utility will be displayed on your screen (as illustrated in Subsection A.3) for each patient type. Before you choose the quality of care for each patient type, you have the opportunity to click on the “calculator” button and thereby calculate patients’potential demand for treatment (as illustrated in Subsection A.3). You can enter the quality you would like to provide as many times as you want. Clicking on the “calculate” button provides you with information about the number of patients who would seek care given the quality level you entered. In addition, you receive information about the resulting earnings and patient utility. You de…ne the quality of care that you wish to provide by entering that quality in the …eld “your decision” and con…rming this entry with “OK.” Payment. After the conclusion of the experiment, one of the 8 decisions will be randomly chosen to function as the relevant round for determining your payment for this part of the experiment. The earnings from this randomly-chosen round will be converted into Euro at the end of the experiment and paid out to you in cash. There are no participants present in the lab who take on the role of patients. An actual patient will bene…t from the patient utility resulting from the quality of care you selected in the randomly-chosen round: A monetary value equaling the patient utility derived from your decision, multiplied by the number of patients who seek treatment from you, will be transferred to Christo¤el Blindenmission Deutschland e.V., 64625 Bensheim. This organization will use the funds to enable the treatment of patients su¤ering from cataracts, a serious eye condition. Control questions. Before proceeding to the decisions in the experiment, we would like to ask you to answer several control questions. These control questions should make it easier for you become acquainted with the decision-making situation. If you have questions about this, please raise your hand. The …rst part of the experiment will begin after all participants have correctly answered the control questions. Payment Procedure. In order to ensure that payments to the participants and the transfer of the monetary donation to Christo¤el Blindenmission Deutschland e.V. are carried out correctly, an overseer will be randomly chosen after the third part of the experiment. The overseer receives a fee of Euro 5 in addition to his or her regular payment from the experiment. The overseer will a¢ rm that the transfer to Christo¤el Blindenmission is correctly carried out by the …nancial administration of the University of Cologne. For the transfer to Christo¤el Blindenmission, the overseer will …ll out a payment order to Christo¤el Blindenmission with the amount, in Euro, that corresponds to the patient utility realized in the randomly-selected round. The …nancial administration of the University of Cologne will then execute payment of the donation to Christo¤el Blindenmission using funds allocated for this experiment. The form will be placed in a stamped envelope addressed to the …nancial administration of the University of Cologne. The overseer and the experimenter will jointly deposit this envelope in the nearest mailbox. The overseer will con…rm by signing a form that he or she properly carried out the assigned tasks, as described above. A copy of this form, as well as a copy of the con…rmation from Christo¤el Blindenmission that the donation was received, can be requested by all participants from the o¢ ce of the Seminar of Personnel Economics and Human Resource Management. The copies will be sent by e-mail. A.2 Control questions of the experiment Comprehension questions [The comprehension questions are presented for the market order Monopoly-Duopoly-Quadropoly. Question that are the same irrespective of the market setting are marked with an asterisk (*).] Monopoly 1. In the …rst part of the experiment, you decide in the role of a .(*) 2. For how many di¤erent patient types, do you decide on quality of treatment about the treatment of 3. How many patients of each type demand medical services in total? 4. How many physicians decide on the quality of medical services beside you in a market? 5. Is the following statement true or false? “Your quality choice for a patient does not only determine your pro…t but also the patient’s bene…t.” (*) True False To answer the following two questions, please consider the examples on your computer screen. 6. Please consider Example A on your computer screen. Please assume that you would choose a quality of 1 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 7. Again, please consider Example A on your computer screen. Please assume, that you would choose a quality of 4 for the patients of this type (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is your pro…t? c. What is the patient’s bene…t? 8. Now, please consider Example B on your computer screen. Please assume, that you would choose a quality of 7 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 9. Again, please consider Example B on your computer screen. Please assume, that you would choose a quality of 5 for the patients of this type (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is your pro…t? c. What is the patient’s bene…t? 10. Which of the following statements is true? (*) Your quality choice for a patient type determines the number of patients of this type who demand your treatment quality. For those patients, who demand your treatment, the quality choice determines the patient bene…t. In addition, your quality choice determines your pro…t for the patient type. Your quality choice for a patient type determines the number of patients of that type who demand your treatment quality. While your quality choice has no in‡uence on the patient bene…t it determines your pro…t. Your quality choice for a patient type does not determine the number of patients of that type who demand your treatment quality. Your quality choice has no in‡uence on the patient bene…t and only determines your pro…t. None. 11. Please complete the following sentence! After the completion of the experiment, it will be determined , which of your decisions from this part of the experiment is relevant for determining your payment and the patient’s bene…t. (*) Duopoly 1. For how many di¤erent patient types, do you decide on quality of treatment 2. How many patients of each type demand medical services in total? 3. How many physicians decide on the quality of medical services beside you in a market? To answer the following two questions, please consider the examples on your computer screen. 4. Please consider Example A on your computer screen. Please assume that you would choose a quality of 1 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 5. Again, please consider Example A on your computer screen. Please assume, that you would choose a quality of 4 for the patients of this type. The other physician would choose a quality of 3 (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is the patient demand for the other physician’s treatment quality? c. What is your pro…t? d. What is the other physician’s pro…t? e. What is the patient’s bene…t resulting from your quality decision? e. What is the patient’s bene…t resulting from the other physician’s quality decision? 6. Now, please consider Example B on your computer screen. Please assume, that you would choose a quality of 7 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 7. Again, please consider Example B on your computer screen. Please assume, that you would choose a quality of 5 for the patients of this type. The other physician would choose a quality of 6 (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is the patient demand for the other physician’s treatment quality? c. What is your pro…t? d. What is the other physician’s pro…t? e. What is the patient’s bene…t resulting from your quality decision? e. What is the patient’s bene…t resulting from the other physician’s quality decision? Quadropoly 1. For how many di¤erent patient types, do you decide on quality of treatment 2. How many patients of each type demand medical services in total? 3. How many physicians decide on the quality of medical services beside you in a market? To answer the following two questions, please consider the examples on your computer screen. 4. Please consider Example A on your computer screen. Please assume that you would choose a quality of 1 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 5. Again, please consider Example A on your computer screen. Please assume, that you would choose a quality of 4 for the patients of this type. The other physicians would choose a quality of 3 (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is the patient demand for the second physician’s treatment quality? c. What is the patient demand for the third physician’s treatment quality? d. What is the patient demand for the fourth physician’s treatment quality? e. What is your pro…t? f. What is the second physician’s pro…t? g. What is the third physician’s pro…t? h. What is the fourth physician’s pro…t? i. What is the patient’s bene…t resulting from your quality decision? j. What is the patient’s bene…t resulting from the second physician’s quality decision? k. What is the patient’s bene…t resulting from the third physician’s quality decision? l. What is the patient’s bene…t resulting from the fourth physician’s quality decision? 6. Now, please consider Example B on your computer screen. Please assume, that you would choose a quality of 7 for patients of this type. For one patient, what is a. your capitation? b. your costs? c. your pro…t? d. the patient’s bene…t? 7. Again, please consider Example B on your computer screen. Please assume, that you would choose a quality of 5 for the patients of this type. The second and the third physician would choose a quality of 6. The fourth physician would choose a quality of 4. (Hint: To answer the questions below, please use the calculator on your computer screen.). a. What is the patient demand for your treatment quality? b. What is the patient demand for the second physician’s treatment quality? c. What is the patient demand for the third physician’s treatment quality? d. What is the patient demand for the fourth physician’s treatment quality? e. What is your pro…t? f. What is the second physician’s pro…t? g. What is the third physician’s pro…t? h. What is the fourth physician’s pro…t? i. What is the patient’s bene…t resulting from your quality decision? j. What is the patient’s bene…t resulting from the second physician’s quality decision? k. What is the patient’s bene…t resulting from the third physician’s quality decision? l. What is the patient’s bene…t resulting from the fourth physician’s quality decision? A.3 Screen shots and experiment parameters Figure 9: Decision screen shot Figure 10: Duopoly calculator screenshot Figure 11: Duopoly calculator screenshot with qualities inputted Figure 12: Quadropoly calculator screenshot Appendix B Robustness: alternate utility function, and betweensubject subsample B.1 Constant absolute risk aversion utility Instead of the linear utility function, we now assume that utility takes the form U (x) = 1 exp( 0:1x), where the coe¢ cient of absolute risk aversion is set at 0:1. Table 11 reports the means of estimated ’s The relative magnitudes between these means are quite close to those for the linear utility function in Table 3. For example, the mean in incentive con…guration (p = 10, c = 0:075, b = 0:5) is two times of that in con…guration (p = 10, c = 0:075, b = 1). The same is true for the linear utility model; see the …rst two rows in Table 3. Using the same normalization (subtracting the monopoly mean), we report the means and standard deviations of estimated ’s in Duopoly and Quadropoly in Table 12. 0 Table 10: Experiment parameters Quality, q 1 2 3 4 5 6 7 8 9 10 Incentive con…guration 1 (p = 10, c = 0:1, b = 1) Capitation, p 10 10 10 10 Cost, c(q) 0 0.1 0.4 0.9 Pro…t, p c(q) 10 9.9 9.6 9.1 Patient bene…t, q 0 1 2 3 10 1.6 8.4 4 10 2.5 7.5 5 10 3.6 6.4 6 10 4.9 5.1 7 10 6.4 3.6 8 10 8.1 1.9 9 10 10 0 10 Incentive con…guration 2 (p = 10, c = 0:075, b = 1) Capitation, p 10 10 10 10 Cost, c(q) 0 0.075 0.3 0.675 Pro…t, p c(q) 10 9.925 9.7 9.325 Patient bene…t, q 0 1 2 3 10 1.2 8.8 4 10 1.875 8.125 5 10 2.7 7.3 6 10 3.675 6.325 7 10 4.8 5.2 8 10 6.075 3.925 9 10 7.5 2.5 10 Incentive con…guration 3 (p = 15, c = 0:1, b = 0:5) Capitation, p 15 15 15 15 Cost, c(q) 0 0.1 0.4 0.9 Pro…t, p c(q) 15 14.9 14.6 14.1 Patient bene…t, q 0 0.5 1 1.5 15 1.6 13.4 2 15 2.5 12.5 2.5 15 3.6 11.4 3 15 4.9 10.1 3.5 15 6.4 8.6 4 15 8.1 6.9 4.5 15 10 5 5 Incentive con…guration 4 (p = 15, c = 0:1, b = 1) Capitation, p 15 15 15 15 Cost, c(q) 0 0.1 0.4 0.9 Pro…t, p c(q) 15 14.9 14.6 14.1 Patient bene…t, q 0 1 2 3 15 1.6 13.4 4 15 2.5 12.5 5 15 3.6 11.4 6 15 4.9 10.1 7 15 6.4 8.6 8 15 8.1 6.9 9 15 10 5 10 Incentive con…guration 5 (p = 10, c = 0:1, b = 0:5) Capitation, p 10 10 10 10 Cost, c(q) 0 0.1 0.4 0.9 Pro…t, p c(q) 10 9.9 9.6 9.1 Patient bene…t, q 0 0.5 1 1.5 10 1.6 8.4 2 10 2.5 7.5 2.5 10 3.6 6.4 3 10 4.9 5.1 3.5 10 6.4 3.6 4 10 8.1 1.9 4.5 10 10 0 5 Incentive con…guration 6 (p = 10, c = 0:075, b = 0:5) Capitation, p 10 10 10 10 10 Cost, c(q) 0 0.075 0.3 0.675 1.2 Pro…t, p c(q) 10 9.925 9.7 9.325 8.8 Patient bene…t, q 0 0.5 1 1.5 2 10 1.875 8.125 2.5 10 2.7 7.3 3 10 3.675 6.325 3.5 10 4.8 5.2 4 10 6.075 3.925 4.5 10 7.5 2.5 5 Incentive con…guration 7 (p = 15, c = 0:075, b = 1) Capitation, p 15 15 15 15 15 Cost, c(q) 0 0.075 0.3 0.675 1.2 Pro…t, p c(q) 15 14.925 14.7 14.325 13.8 Patient bene…t, q 0 1 2 3 4 15 1.875 13.125 5 15 2.7 12.3 6 15 3.675 11.325 7 15 4.8 10.2 8 15 6.075 8.925 9 15 7.5 7.5 10 Incentive con…guration 8 (p = 15, c = 0:075, b = 0:5) Capitation, p 15 15 15 15 15 Cost, c(q) 0 0.075 0.3 0.675 1.2 Pro…t, p c(q) 15 14.925 14.7 14.325 13.8 Patient bene…t, q 0 0.5 1 1.5 2 15 1.875 13.125 2.5 15 2.7 12.3 3 15 3.675 11.325 3.5 15 4.8 10.2 4 15 6.075 8.925 4.5 15 7.5 7.5 5 Table 11: Estimated means of in monopoly under CARA Incentive con…gurations mean (p = 10, c = 0:075, b = 0:5) 0.066 (p = 10, c = 0:075, b = 1) 0.033 (p = 10, c = 0:1, b = 0:5) 0.084 (p = 10, c = 0:1, b = 1) 0.041 (p = 15, c = 0:075, b = 0:5) 0.051 (p = 15, c = 0:075, b = 1) 0.026 (p = 15, c = 0:1, b = 0:5) 0.071 (p = 15, c = 0:1, b = 1) 0.034 Table 12: Normalized means and standard deviations of distributions under CARA Incentive con…gurations Monopoly Duopoly Quadropoly mean st. dev. mean st. dev. mean st. dev. (p = 10, c = 0:075, b = 0:5) 0 0.059 -0.083 0.067 -0.098 0.054 (p = 10, c = 0:075, b = 1) 0 0.030 -0.050 0.042 -0.062 0.043 (p = 10, c = 0:1, b = 0:5) 0 0.082 -0.086 0.073 -0.143 0.112 (p = 10, c = 0:1, b = 1) 0 0.040 -0.054 0.052 -0.068 0.054 (p = 15, c = 0:075, b = 0:5) 0 0.045 -0.105 0.050 -0.126 0.051 (p = 15, c = 0:075, b = 1) 0 0.023 -0.065 0.040 -0.078 0.058 (p = 15, c = 0:1, b = 0:5) 0 0.069 -0.104 0.078 -0.127 0.065 (p = 15, c = 0:1, b = 1) 0 0.033 -0.062 0.046 -0.078 0.053 Again, the means have all become lower when the market becomes more competitive. The di¤erences between the normalized duopoly and quadropoly means also point in the same direction as those in the linear utility model although the magnitudes have now become smaller (see Table 4). For brevity, we do not present the estimated values. Figures 13, 14 and 15 are the histograms of estimated normalized altruism distributions for the three markets. The comparisons between these with those under linear utility (histograms in Figures 4, 5, and 6) just show the di¤erences in estimated values. Figure 13: Histograms of estimated for CARA in each incentive con…guration in monopoly Figure 14: Histograms of estimated for CARA in each incentive con…guration in duopoly Figure 15: Histograms of estimated for CARA in each incentive con…guration in quadropoly B.2 Between-subject subsample We use subjects’…rst experiences for a between-subject experiment. From Table 1, roughly a third of the 361 subjects played each of the three markets in their …rst round, so we only can use about 1=3 of the entire data. In the experiments, 124 subjects played the monopoly game …rst, 119 played the duopoly …rst, and 118 played the quadropoly …rst. The 8 decisions of these …rst games constitute the subsample. Table 13 presents the …rst-round summary statistics of the 8 incentive-con…guration games in the 3 markets. There are some small di¤erences in the means and standard deviations between the smaller, between-subject subsample and the full sample. Nevertheless, the means and standard deviations follow the same pattern in Table 2. Figures 16 to 18 present the quality choice distributions by incentive con…gurations for the three markets. Table 13: Between-subject subsample summary statistics Incentive con…gurations Monopoly (n = 124) Duopoly (n = 119) Quadropoly (n = 118) mean st. dev. mean st. dev. mean st. dev. (p = 10, c = 0:075, b = 0:5) 4.403 2.659 7.437 1.701 7.958 1.577 (p = 10, c = 0:075, b = 1) 4.460 2.688 7.765 1.598 8.017 1.764 (p = 10, c = 0:1, b = 0:5) 4.065 2.569 6.597 1.463 6.932 1.688 (p = 10, c = 0:1, b = 1) 3.871 2.521 6.622 1.408 6.958 1.538 (p = 15, c = 0:075, b = 0:5) 5.113 3.007 8.420 1.670 8.780 1.675 (p = 15, c = 0:075, b = 1) 5.266 3.021 8.672 1.698 8.898 1.892 (p = 15, c = 0:1, b = 0:5) 4.823 2.891 7.664 1.801 8.102 1.692 (p = 15, c = 0:1, b = 1) 4.734 2.930 8.000 1.616 8.254 1.949 Table 14 presents the means of estimated ’s in Monopoly, and they are similar to those in the full sample in Table 3. Table 14: Estimated means of in monopoly Incentive con…gurations mean (p = 10, c = 0:075, b = 0:5) 1.321 (p = 10, c = 0:075, b = 1) 0.669 (p = 10, c = 0:1, b = 0:5) 1.626 (p = 10, c = 0:1, b = 1) 0.774 (p = 15, c = 0:075, b = 0:5) 1.534 (p = 15, c = 0:075, b = 1) 0.790 (p = 15, c = 0:1, b = 0:5) 1.929 (p = 15, c = 0:1, b = 1) 0.947 In Table 15, we present the means and standard deviations of estimated ’s in Duopoly and Quadropoly (under the same normalization as before). There are some di¤erences from Table 4. In particular, the means tend to be higher in magnitude than those in the full sample. The standard deviations are also bigger, but that can be accounted for by the smaller sample size. Table 15: Normalized means and standard Incentive con…gurations Monopoly mean st. dev. (p = 10, c = 0:075, b = 0:5) 0 0.798 (p = 10, c = 0:075, b = 1) 0 0.403 (p = 10, c = 0:1, b = 0:5) 0 1.027 (p = 10, c = 0:1, b = 1) 0 0.504 (p = 15, c = 0:075, b = 0:5) 0 0.902 (p = 15, c = 0:075, b = 1) 0 0.453 (p = 15, c = 0:1, b = 0:5) 0 1.156 (p = 15, c = 0:1, b = 1) 0 0.586 deviations of distributions Duopoly Quadropoly mean st. dev. mean st. dev. -1.532 1.170 -1.789 1.076 -0.893 0.531 -1.141 1.005 -1.639 1.053 -2.762 2.782 -1.011 0.588 -1.315 1.322 -2.188 1.045 -2.665 1.511 -1.345 0.733 -1.743 1.903 -2.377 1.708 -2.832 1.641 -1.323 0.706 -1.743 1.585 We next present the histograms of the actual qualities in the subsample in Figures 16, 17, and 18, with the frequencies written on top of each quality level. Qualities in monopoly in the full and between-subject subsample show more variations. However, the duopoly and quadropoly quality distributions are remarkably similar. Figure 16: Between-subject quality histograms in monopoly Figure 17: Between-subject quality histograms in duopoly Figure 18: Between-subject quality histograms in quadropoly Figures 19, 20, and 21 plot the histograms of estimated distributions. (Again for brevity, we have omit- ted the actual estimated values.) As with the case of qualities, the estimated distributions in monopoly show more di¤erences between the full sample and the between-subject sample, but the estimated distrib- utions in duopoly and quadropoly are remarkably similar. Overall, we think that our results are robust with respect to between-subject and within-subject designs. Figure 19: Between-subject histograms of estimated monopoly Figure 20: Between-subject histograms of estimated duopoly Figure 21: Between-subject histograms of estimated quadropoly B.2.1 Reduced-form analysis for between-subject subsample Table 16 reports descriptive statistics on subjects’…rst-experience average qualities for low and high parameter levels of price, cost, and patient bene…ts. The entries are written with the same convention as in Table 8. The average qualities in Table 16 exhibit the same pattern as those in Table 8. The average quality is higher in each market at the higher price, but the relative di¤erence declines as the market becomes more competitive. Average qualities are lower at higher cost, but the relative di¤erence hardly varies with competition. Patient bene…t does not seem to a¤ect average qualities much. We conclude that the reduced-form analysis is robust with respect to the between-subject and within-subject designs. Table 16: Descriptives on the variations in price, costs, and patient bene…t in the subjects’ …rst market session Low parameter level High parameter level Relative N Parameter Mean st. dev. Mean st. dev. di¤erence Price (p = 10; p = 15) Monopoly 4.200 2.614 4.984 2.962 0.187 496 Duopoly 7.105 1.736 8.189 1.623 0.153 476 Quadropoly 7.466 1.720 8.509 1.832 0.140 472 Cost (c = 0:075; c = 0:1) Monopoly Duopoly Quadropoly 4.811 8.074 8.413 2.866 1.734 1.778 4.373 7.221 7.561 2.757 1.693 1.826 -0.091 -0.106 -0.101 496 476 472 Patient bene…t (b = 0:5; bH = 1) Monopoly Duopoly Quadropoly 4.601 7.529 7.943 2.807 1.781 1.781 4.583 7.765 8.032 2.834 1.743 1.919 -0.004 0.031 0.011 496 476 472 Regression results for the between-subject analysis are reported in Table 17. The notation here is the same as in Table 9, except of course that there are no market-order dummies. Because of the smaller sample, the R2 ’s are uniformly smaller than regressions in Table 9. Most estimates happen to be a little smaller in their magnitudes than in Table 9, but their signi…cance remains the same. Table 17: Between-subject quality regressions Model: (1) (2) Dependent variable: Quality Quality Duopoly 3.194*** 3.125*** (0.373) (0.371) Quadropoly 3.809*** 3.834*** (0.391) (0.387) High price (= 1 if p = 15) 0.967*** 0.784*** (0.0459) (0.0761) High cost (= 1 if c = 0:1) -0.710*** -0.437*** (0.0437) (0.0811) High bene…t (= 1 if b = 1) 0.100** -0.0181 (0.0423) (0.0660) Duopoly High price 0.300*** (0.107) Quadropoly High price 0.258** (0.114) Duopoly High cost -0.415*** (0.111) Quadropoly High cost -0.414*** (0.102) Duopoly High bene…t 0.253** (0.101) Quadropoly High bene…t 0.107 (0.101) Session dummies Yes Yes Constant Observations Subjects R2 4.051*** (0.334) 2,888 361 0.386 4.066*** (0.331) 2,888 361 0.388