Eur J Law Econ (2012) 33:205–229 DOI 10.1007/s10657-010-9161-3
An experimental investigation on optimal bankruptcy laws Daniela Di Cagno • Marco Spallone
Published online: 23 June 2010 Springer Science+Business Media, LLC 2010
Abstract We performed an experimental investigation to assess whether the ‘‘restricted auction’’ mechanism proposed by Berkovitch, Israel and Zender in 1997 works effectively as an optimal bankruptcy law or not. An optimal bankruptcy law is a commitment device that implements efficient choices both before (ex ante) and after (ex post) financial distress, even if moral hazard is binding. We designed an experiment focused on ex ante efficiency and we found that the restricted auction mechanism was able to direct an optimal amount of effort toward entrepreneurial activities. This result confirms the theoretical predictions. Nonetheless, we found that under a plain unrestricted auction mechanism our experimental subjects chose to allocate into their firms a larger amount of effort than that predicted by theory. Although difficult to justify on theoretical grounds, this experimental evidence is robust. Our behavioral interpretation is that this result is due to ‘‘moral sentiments’’, such as the natural propensity of subjects toward socially desirable behaviors. In fact, we show that it vanishes once these motives are removed. Keywords
Bankruptcy Laws Optimal Incentives Experimental Evidence
JEL Classification
C7 C9 K0
1 Introduction A central tenet in economics is that competitive markets gain long run equilibrium through the elimination of inefficient firms. This process enhances social welfare D. Di Cagno M. Spallone (&) Luiss-Guido Carli University of Rome, Rome,, Italy e-mail:
[email protected] M. Spallone G. D’Annunzio University of Pescara, Pescara,, Italy
123
206
Eur J Law Econ (2012) 33:205–229
since firms that employ obsolete technologies die and only minimum-average-cost producers survive. Corporate bankruptcy laws are the institutional mechanism through which inefficient firms drop out of the market. Initially, they were thought as a remedy for creditors (i.e., investors or other commercial partners); eventually, they evolved into a relief for debtors (i.e., entrepreneurs). In any case, they should provide an environment in which an efficient allocation of resources is achieved both before (i.e., ex ante) and after (i.e., ex post) financial distress. Ex post efficiency implies that the assets of distressed firms are directed toward their highest valued use. So, ex post efficiency is easily achieved through market based mechanisms, such as auctions. Ex ante efficiency implies that entrepreneurs do not under-invest in firm specific human capital. Simply put, it implies that entrepreneurs put an optimal amount of effort into debt financed firms. Obviously, it is very difficult to attain ex ante efficiency if moral hazard is binding, that is if entrepreneurial effort is not observable. However, under very specific and sometimes peculiar assumptions, some scholars have offered bankruptcy laws that purport to optimize both ex ante and ex post efficiency (Berkovitch et al. 1997; White 2005). Current bankruptcy codes differ significantly from optimal mechanisms; hence, it is not possible to retrieve data to test whether these mechanisms are really effective or not. This is the reason why we decided to perform an experimental investigation on optimal bankruptcy laws, focusing on the ex ante incentives they provide to entrepreneurs. The aim of this paper is to test experimentally whether the ‘‘restricted auction’’ mechanism proposed by Berkovitch, Israel and Zender in 1997 is an optimal bankruptcy law, that is a commitment device that implements both ex ante and ex post efficiency in firm-specific investment decisions when moral hazard is binding. In the article of Berkovitch, Israel and Zender, it is assumed that while both the entrepreneur and the investor have the same information about the future cash flow of the financed firm, all outside investors do not have such information. The auction is restricted in the sense that only outside investors are allowed to bid for the distressed firm: this restriction introduces a bias in favor of the incumbent management and provides correct ex ante incentives. To test this prediction we run an experiment consisting of two separate treatments that reproduce either the case in which the restricted auction mechanism is implemented (treatment RE) or the case in which the unrestricted auction mechanism is implemented (treatment UNRE). In treatment UNRE the bank is allowed to bid for the distressed firm, while in treatment RE only an outside investor, played by the computer, is allowed to make a bid. We obtain two important results. First, we show that the restricted auction mechanism is able to induce entrepreneurs to allocate more effort into their firm than plain unrestricted auction mechanisms. Second, we find that under an unrestricted auction mechanism our experimental subjects put a larger amount of effort into their firms than that predicted by our theoretical model. The first result confirms that the incentives embedded into the restricted mechanism attain ex ante efficiency. This result is robust and easy to interpret. As for the second result, it is experimentally robust, but difficult to justify on theoretical grounds; our interpretation of such result is behavioral. More precisely, we argue
123
Eur J Law Econ (2012) 33:205–229
207
that the choices of our experimental subjects were driven not only by strict economic incentives, but also by ‘‘moral sentiments’’, such as the natural propensity of subjects toward socially desirable behaviors (Bowles and Hwang 2008; Camerer 2003; Fehr et al. 2007). The only way to check for this argument was to remove these motives. So, we performed a follow-up experiment in which we compare the two auction mechanisms, implementing them sequentially following a different order: in FOLUP1 the experimental subjects make decisions under the restricted mechanism first (and under the unrestricted one second), while in FOLUP2 they do the opposite. The switch from a debtor friendly code to a creditor friendly one was aimed at making our experimental subjects aware of the fact that self-interested optimization rather than ethical behavior was appropriate (Cardenas et al. 2000; Gneezy and Rustichini 2000), or that the investor would like to profit unfairly at their expenses, thereby compromising the subjects’ preexisting predispositions of reciprocity or obligation toward the investor (Falk and Kosfeld 2006; Fehr and List 2004; Fehr and Rockenbach 2003). In fact, our surprising result definitely disappears once the experimental subjects choose to respond to material interests only. The economic relevance of our investigation is related to the debate over the trade-off between debtor friendly bankruptcy codes (as the restricted auction mechanism) and creditor friendly ones (as the unrestricted mechanism). In fact, while creditor friendly codes are aimed at attenuating the inefficiencies that arise at the financing stage of an investment project (where many projects with a positive net present value are not financed because creditors fear that the liquidation procedure may limit the amount of cash flow that they can recover), debtor friendly codes are aimed at providing the right ex ante incentives to entrepreneurs when moral hazard is binding. From a macroeconomic standpoint, bankruptcy codes must be evaluated within the institutional environment in which they are implemented (Berkovitch and Israel 1999); so, in countries where investor protection is low, a creditor friendly bankruptcy code may be seen as a guarantee for investors, hence as a mean to stimulate investments. Our findings suggest that debtor friendly bankruptcy codes do not fully discourage socially desirable behaviors. The main contribution of our analysis is the application of the experimental methodology to bankruptcy issues. In fact, we think that this is the best way to validate a set of theoretical results that are difficult to test empirically. Many economists focused their attention on bankruptcy laws, both from a personal and a corporate point of view (for a comprehensive survey, see White (2005). The main task of most of these contributions is to design a bankruptcy law that is able to induce an efficient allocation of resources (both ex ante and ex post). In order to achieve ex post efficiency Bebchuk (1988), Jensen (1991) and Aghion et al. (1992) suggest market-based procedures. In particular, they suggest that firms entering bankruptcy procedures should be sold via a market mechanism (i.e., they should be auctioned off). In fact, market mechanisms ensure that: (1) the assets of the firms are directed to their highest valued use; (2) the absolute priority of the existing claims is safeguarded; (3) neither the shareholders nor the incumbent management receive any advantage in the resolution of distress. However, market based mechanisms do not guarantee ex ante efficiency, that is the possibility of financial distress may induce entrepreneurs to save on effort even if their firm is
123
Eur J Law Econ (2012) 33:205–229
208
going to be auctioned off in case of distress. That is why Berkovitch et al. (1997) propose a ‘‘restricted auction’’ mechanism. As for our knowledge, this is the first attempt to test experimentally the effectiveness of a mechanism designed to be an optimal bankruptcy law from both an ex ante and an ex post perspective. In fact, there are few other experimental contributions on bankruptcy, but their focus is on how to cope with bankruptcy after it occurs. In particular, their focus is on how the assets of a distressed firm should be divided among creditors (Gaechter and Riedl 2004, 2005; Herrero et al. 2002). The paper consists of five sections followed by an Appendix: Section 2 describes the theoretical background; Section 3 illustrates the initial experimental design and clarifies the theoretical predictions that we are going to test; Section 4 describes the experimental results. Section 5 contains the description and the results of the followup experiment. Section 6 concludes the paper. The Appendix contains the instructions for the experiment.
2 Theoretical background We describe thereafter the theoretical setup of our experiment that consists of a simplified version of the model of Berkovitch et al. (1997). At date 0, a risk neutral entrepreneur contracts with a risk neutral investor to raise an initial investment (I) to run a project, issues debt with face value F (F [ I), establishes the firm, decides how to allocate her effort between her firm (h) and other outside activities (1-h). At date 1, the entrepreneur and the investor (and no one else) observe a precise, but non-verifiable signal (e) concerning the final cash flow at period 21. The entrepreneur decides whether to liquidate the firm, continue to operate under the original contract, or attempt to renegotiate the debt contract. It is obvious that the choice of the entrepreneurial effort is affected by the expectations over the future cash flow and by the renegotiation procedure, as we will clarify later. If the firm is liquidated at date 1, the claimants receive cash flow equal to L, the liquidation value of the firm which is common knowledge2. The entrepreneur is replaced and gets the outside option wage, that is a linear function of the effort she decides to allocate outside her firm: ~ ¼ a þ bð1 hÞ w
ð1Þ
If the entrepreneur decides to operate the firm until date 2, the firm generates a final cash flow equal to: y~ðhÞ ¼ h þ ~
ð2Þ
The initial debt is partly or completely repaid and the entrepreneur gets the rest of the final cash flow. The productivity shock e has probability distribution G(e), 1 This informational structure is peculiar since it implies that both the investor and the entrepreneur have the same amount of information. On the contrary, outside investors have no information about the profitability of the firm. 2
Also outside investors know L.
123
Eur J Law Econ (2012) 33:205–229
209
density function g(e), and support ½; . The possible date 2 cash flow can be divided into 3 regions, with the defining values being: ~ h; 1 1 ¼ L þ w ~ h; 2 2 ¼ F þ w It follows that in region 1 ( \1 ) the firm is not economically viable and in financial distress, in region 2 (e1 B e \ e2) the firm is economically viable and in financial distress, and in region 3 (2 \) the firm is economically viable and able to pay back the initial debt. It is obvious that the entrepreneur liquidates the firm in region 1. In fact, since in region 1 the cash flow generated by the firm is smaller than the liquidation value plus the outside option wage, there is no incentive to operate the firm until date 2. On the contrary, the entrepreneur continues to operate the firm under the original contract in region 3. In fact, since in region 3 the cash flow generated by the firm is bigger than the initial debt plus the outside option wage, the entrepreneur prefers to operate the firm until date 2. In region 2, however, the entrepreneur and the investor have to negotiate in order to share the surplus. In fact, the cash flow generated by the firm is not enough to repay the initial debt, but it is bigger than the the liquidation value plus the outside option wage. So, in region 2, it is assumed that the payoffs are determined according to the generalized Nash ~ and L for the entrepreneur and the bargaining solution with disagreement points w investor respectively. More precisely, the negotiation payoffs are: ~ þ c½h þ L w ~ Ve ¼ w
ð3Þ
~ Vi ¼ L þ ð1 cÞ½h þ L w
ð4Þ
and
where the suffixes e and i stand for entrepreneur and investor respectively and c represents the bargaining power of the entrepreneur. It is worth noticing that it is the mechanism through which the firm is auctioned off in region 2 that determines the size of c in the initial renegotiation. In fact, a debtor biased bankruptcy law, that is an entrepreneur biased auction procedure, increases c. For experimental purposes we let the entrepreneur choose among a finite number of levels of firm specific effort (h) to maximize the following utility function: Ue ¼
Lþwð1hÞh Z
½wð1 hÞd
þ
Fþwð1hÞh Z
wð1 hÞ þ c½h þ L wð1 hÞd
ð5Þ
Lþwð1hÞh
þ
Z
ðh þ FÞd
Fþwð1hÞh
123
210
Eur J Law Econ (2012) 33:205–229
Obviously, for any c, selling the distressed firm through a market mechanism (an auction) is enough to guarantee that the assets of the firm are directed toward their most valued use (ex post efficiency). However, the size of c, i.e. the choice of the auction procedure, affects ex ante incentives. In fact, since the entrepreneur receives only a fraction c of the difference between the final cash flow and the liquidation value, the size of such fraction is crucial to determine the optimal effort. More precisely, it is possible to show that the level of h chosen by the entrepreneur is increasing in c. Then, in order to achieve both ex ante and ex post efficiency, the following restricted auction mechanism is implemented (as in Berkovitch et al.): The investor has the right to refuse bankruptcy proceedings. If he refuses, the court enforces the original contracts regardless of any private agreement the parties may have reached. If the investor approves the bankruptcy procedure, the entrepreneur and any other bidder, other than the investor, may participate in a second-price sealed-bid auction for the firm. The proceeds from the auction are paid out according to the original contract. It is easy to show that the above mechanism implements c = 13. So, in equilibrium the entrepreneur chooses a level of effort equal to the one she would choose if c was equal to 1 in the Nash bargaining solution described above. Moreover, since she is the only bidder who knows the final cash flow, she bids, ~ wins the auction and pays L for the firm. yw We define the outside wage function as: wð1 hÞ ¼ 0:5 þ ð10; 000 hÞ
ð6Þ
where the entrepreneur’s endowment of effort has been normalized to 10,000 experimental credits (EC). Moreover, we allow the entrepreneur to choose among the following six levels of effort: 0; 1,000; 3,000; 5,000; 7,000; 10,000. Finally, we set the liquidation value of the firm (L) equal to 90,000 EC, the face value of debt (F) equal to 110,000 EC, and the initial investment (I) equal to 100,000 EC. As for the productivity shock (e), we randomly generate it from a Normal distribution with mean equal to 100,000 and standard deviation equal to 7,000. This parameterization reflects both our need of achieving meaningful payoffs in terms of monetary payments and our willingness of making calculations easy for the subjects involved in the experiment.
3 Experimental design The experiment, run in 15 sessions of 10 subjects4 each, is made of 2 treatments: treatment RE (8 sessions) and treatment UNRE (7 sessions). Each treatment is 3 The right of refusal of the investor is a guarantee against the entrepreneur asking for bankruptcy procedures in case the firm is not financially distressed. The restriction imposed on the investor allows the entrepreneur to win the auction (he has more information than outside bidders about the future cash flow) and obtain all the surplus. If the investor was allowed to bid, he would bid more than the entrepreneur given a higher reservation value (due to the initial investment). 4
Subjects were recruited among undergraduate students of LUISS- Guido Carli University of Rome.
123
Eur J Law Econ (2012) 33:205–229
211
divided into two phases: treatment RE in Phase 0 and Phase 1, treatment UNRE in Phase 0 and Phase 2. This structure implies that Phase 0 is played by all participants to the experiment (Table 1). In order to select only risk neutral experimental subjects we initially performed a simple procedure5 allowing us to detect the risk attitude of 300 potential experimental subjects and we appointed for the experiment only those subjects who behaved as risk neutral agents over a set of lotteries whose expected returns resembled those of our experiment, that is between 10 and 50 Euros. The choice of a sample of risk neutral subjects was due to two reasons: first, we wanted to correct for the fact that we are testing students instead of entrepreneurs; second, we did not want to deal with different degrees of risk aversion, that are difficult to measure and make the theoretical predictions hard to test. Each subject plays only one of the two treatments: either RE or UNRE, that is Phase 0 plus either Phase 1 or Phase 2.6 At the end of the experiment, subjects are paid (immediately and in cash) the money they made. At the beginning of each treatment, the instructions are handed in and read aloud. Subjects have time to read the instructions again and they are free to ask questions if they want to. Moreover, they could play Phase 0 twice before the experiment effectively started. This allows the subjects to get acquainted with the software and to learn more about the consequences of their choices. In the following subsections we describe separately each one of the three phases. 3.1 Phase 0: all subjects In Phase 0 subjects face individual choices under risk in 3 different rounds (R1, R2, and R3) that imply different value of the parameter c. In R1 c is equal to zero, in R2 c is equal to one half, and in R3 c is equal to one. Notice that in the theoretical setup c is the bargaining power of the entrepreneur in the initial renegotiation, while in the experiment, since there is no initial renegotiation, the c’s are exogenously chosen: in our experimental setting the market mechanism replaces bilateral bargaining. In each round, subjects face a set of 6 different lotteries displayed sequentially on their computer screen and they are asked to choose the lottery they prefer. Before selecting the most preferred lottery, subjects are obliged (by the software) to see all the lotteries at least once, but they may see them every time they need to, with no time constraint nor cost. Lotteries are graphically represented by 3 areas of different color (blue, red, and yellow) on which a indicator moves (see Figs. 5, 6 in the Appendix). Each lottery differs from the others both in terms of the payoffs associated to each area (reported on the top of the screen) and in terms of the size of each area. 5 We adopted a screening technique called Multiple Price List (MPL) to elicit risk attitudes Miller et al. (1969), Schubert et al. (1999), Barr and Packard (2002), and Holt and Laury (2002, 2005). In particular, we followed the procedure adopted by Holt and Laury. The experimental literature is almost unanimous on the fact that MPL provides a relatively transparent procedure to elicit risk attitudes. In fact, subjects rarely get confused about the incentives to respond truthfully.This method allowed us to select 150 subjects who actually behaved as risk neutral. 6
Samples performing under the two treatments are independent.
123
Eur J Law Econ (2012) 33:205–229
212
After their choice, subjects are allowed to play the following round which works in the same way. Then, the computer selects a round at random and plays the chosen lottery: the indicator starts moving on the screen that represents that lottery and, once it stops on a point, the payoff corresponding to that point is calculated. So, in Phase 0, subjects are not told to be entrepreneurs; hence, they are not asked to choose the amount of firm specific effort, but to choose among lotteries. However, each round represents a different level of c. Moreover, each lottery represents a different level of effort, and the size of each one of the three areas that represent a lottery is the probability of each one of the three regions described in Sect. 2. More precisely, the blue area represents the region in which the payoff is the outside option wage; the red area represents the region in which the payoff is the Nash bargaining outcome; the yellow area represents the region in which the payoff is the realized cash flow (net of debt repayment) (see Fig. 5 in the Appendix). Obviously, the payoffs of both the blue and the yellow areas are the same in each round, while the payoffs of the red area change from round to round because they depend on c. Probabilities and expected payoffs are displayed in Tables 2 and 3 respectively. Table 1 Experimental design No. of subjects
Phase 0
Phase 1
R1
R2
R3
Treatment RE
80
U
U
U
Treatment UNRE
70
U
U
U
Table 2 Probabilities
Table 3 Expected payoffs
123
Phase 2
U U
Lottery
Blue area prob.
Red area prob.
Yellow area prob.
1
0.237525
0.746413
0.016062
2
0.176556
0.796552
0.026892
3
0.087368
0.845825
0.066807
4
0.037073
0.820939
0.141988
5
0.013405
0.726437
0.260158
6
0.002137
0.497862
0.050000
Lottery
Round 1 (c = 0)
Round 2 (c = 0.5)
Round 3 (c = 1)
1
5,040.194
7,847.566
10,654.94
2
4,571.822
7,850.471
11,129.12
3
3,705.147
7,825.444
11,945.74
4
3,008.099
7,635.955
12,263.81
5
2,600.557
7,215.117
11,829.68
6
2,792.596
6,398.493
10,004.39
Eur J Law Econ (2012) 33:205–229
213
As Table 3 shows, the optimal choices are the following: lottery 1 in Round 1, lottery 2 in Round 2 and lottery 4 in Round 3. 3.2 Phase 1: subjects under RE In Phase 1, subjects are told that they are entrepreneurs who have already undertaken an investment project financed by a bank. They have to choose how much effort to put in their firm among 6 alternatives (H = 0; 1,000; ...; 10,000). Graphics are exactly the same as in Phase 0: each level of effort H defines a stochastic environment (i.e., a lottery) represented by 3 areas of different color on which an indicator will move. For each level of effort 3 different situations may occur: 1. 2. 3.
The firm is The firm is The firm is debt issued
not economically viable and in financial distress (blue area); economically viable and in financial distress (red area); economically viable and is able to pay back the face value of the at time 0 (yellow area).
After subjects have made their choices, the computer shows the corresponding screen and moves the indicator. If the indicator stops in the areas where the payoff is either the payoff of the outside option (blue area) or the realized cash flow (yellow area), the payoffs are computed as in Phase 0 and the experiment is over. If the indicator stops in the area where the payoff is the Nash bargaining payoff (red area), the auction procedure begins. The auction is a second-price sealed-bid auction that the subject and an outside investor (played by the computer) participate7. The investor who financed the project (the bank) is not allowed to participate the auction. Each subject bids an amount p. The winner of the auction is the one who bids the highest p; the winner gets the firm and pays a price equal to the second highest bid. The computer works out the payoffs accordingly. Phase 1 is payoff equivalent to the third round of Phase 0 because the restricted auction mechanism implements c = 1. Since the bank is not allowed to bid, the computer plays the role of an outside investor that does not know the income produced by the firm, but only its liquidation value. Therefore, the computer always bids the liquidation value of the firm (90,000 EC). Subjects are aware that their firm is worth more than the liquidation value, and they also know that this additional information is not common knowledge. Therefore, they should always bid more than 90,000. In fact, subjects who bid more than 90,000 EC win the auction and pay 90,000 EC for the firm, their payoff being the cash flow of the firm net of 90,000 EC. On the contrary, subjects who bid less than 90,000 EC lose the auction and the cash flow of the firm, their payoff being the outside option wage. Here, the optimal level of H is 5,000, that is the H that corresponds to lottery 4 in Round 3 of Phase 0. 7
This procedure resembles a BDM mechanism. Noussair et al. (2004) show that a BDM mechanism and a Vickrey auction are equivalent to truthful revelation of individuals’ willingness to pay apart from some behavioral differences related to sufficient practice and appropriate training.
123
214
Eur J Law Econ (2012) 33:205–229
3.3 Phase 2: subjects under UNRE In Phase 2, subjects face the same framework as in Phase 1. The difference is that the bank is allowed to participate the second-price sealed-bid auction. Now, the computer plays the role of the bank in the auction procedure. Phase 2 is payoff equivalent to the first round of Phase 0 because the unrestricted auction implements c = 0. Since the bank is now allowed to bid, the computer plays the role of the bank whose reservation value is higher than the reservation value of the entrepreneur. Since the liquidation value (90,000 EC) is smaller than the initial investment (100,000 EC), the computer should always bid more than the entrepreneur. Now, the optimal level of H is 0, that is the H that corresponds to lottery 1 in Round 1 of Phase 0. 3.4 Theoretical predictions According to the model described in Sect. 2, the optimal level of effort is increasing in c: in Phase 0 the subjects involved in the experiment should choose lotteries that are increasing through Round 1, 2, and 3; as for Phase 1 and Phase 2, subjects should choose an higher level of firm specific effort in Phase 1 (restricted auction) than in Phase 2 (unrestricted auction). For the law and economics purposes of this paper, the main comparison that we wish to carry over is that between Phase 1 and Phase 2. This comparison will allow us to test whether the restricted auction mechanism is able to achieve ex ante efficiency or not: if the average level of effort selected by experimental subjects in Phase 1 is higher than in Phase 2, we can conclude that the restricted auction mechanism works. Phase 0 was useful in making our experimental subjects familiar with the experimental software and aware of the consequences of their choices. Data retrieved in Phase 0 were also very useful to check for the robustness of our experimental results and to interpret them correctly.
4 Experimental results The choices made by our experimental subjects are displayed in Tables 4 and 5. In R1 of Phase 0 under treatment RE the large majority of experimental subjects chose lotteries 1 and 2 (56 and 19% respectively), in R2 most of them preferred lotteries 2 and 3 (54 and 16% respectively), while in R3 a higher percentage of subjects selected lotteries 4 and 5 (20 and 25% respectively). Under treatment UNRE, the behavior of experimental subjects in Phase 0 was not very different: in R1 the large majority of experimental subject chose lotteries 1 and 2 (44 and 31% respectively), in R2 they moved toward lotteries 2 and 3 (39 and 21% respectively), and in R3 many of them selected lottery 5 and 6 (17 and 19% respectively).
123
Eur J Law Econ (2012) 33:205–229
215
Table 4 Treatment RE—frequencies Lottery
Phase 0
Phase 1 R3
%
R1
%
R2
%
%
1
45
56.25
10
12.5
9
11.25
3
2
15
18.75
43
53.75
12
15
7
3
9
11.25
13
16.25
13
16.25
14
17.5
4
6
7.5
5
6.25
16
20
31
38.75
5
2
2.5
5
6.25
20
25
19
23.75
6
3
3.75
4
5
10
12.5
6
7.5
3.75 8.75
Table 5 Treatment UNRE—frequencies Lottery
Phase 0 R1
%
R2
%
R3
%
Phase 2
%
1
31
44.29
13
18.57
10
14.29
27
38.57
2
22
31.43
27
38.57
12
17.14
25
35.71
3
11
15.71
15
21.43
13
18.57
10
14.29
4
3
4.29
7
10
10
14.29
5
7.14
5
3
4.29
5
7.14
12
17.14
1
1.43
6
0
0
3
4.29
13
18.57
2
2.86
However, while experimental subjects under treatment UNRE behaved in the same way in Phase 2 and in the correspondent round of Phase 0 (R1), subjects under RE behaved differently in Phase 1 and in the correspondent round of Phase 0 (R3). In particular, t-tests performed on our data show that the behavior of experimental subjects under RE is different across Phase 1 and the correspondent round of Phase 0 (R3) with a 70% probability, while it is exactly the same under UNRE in Phase 2 and the correspondent round of Phase 0 (R1)8. This evidence is represented in Figs. 1 and 2. In the following subsections we describe our results. In particular, we describe the comparison between the restricted auction mechanism and the unrestricted one in Sect. 4.1, and we provide an interpretation of the outcome of this comparison in Sect. 4.29.
8
It is important to point out that in both cases choices are not statistically different at a 5% significance level. However, the behavior of experimental subjects under RE is important to exclude that they chose lotteries simply by similarity between Phase 0 and either Phase 1 or 2. 9 Since the the experimental design is complex, we asked our experimental subjects to provide comments about the difficulties that they faced during the experiment. The great majority of them (more than 80%) say that they understood perfectly the trade-off implied by their choices. Another way to check whether the experimental subjects understood the experiment is to look at the bids that they made in the auction environment: again, 80% of the subjects made bids that are consistent with an optimal strategic behavior.
123
216
Eur J Law Econ (2012) 33:205–229
Fig. 1 Phase 0 vs. Phase 1—frequencies
Fig. 2 Phase 0 vs. Phase 2—frequencies
4.1 Restricted versus unrestricted auction mechanism Here, we wish to compare the behavior of the subjects involved in the experiment under different auction mechanisms. That is, we wish to compare the choices made by the experimental subjects in Phase 1 and in Phase 2 under treatment RE and UNRE respectively to see whether the restricted auction mechanism works as an optimal bankruptcy law or not. The comparison between Tables 4 and 5 shows that our experimental subjects chose on average a higher level of effort H under the restricted auction mechanism (Phase 1 of treatment RE) than under the unrestricted auction one (Phase 2 of treatment UNRE): in Phase 1, 39% of the experimental subjects chose lottery 4 and 24% of them chose lottery 5; in Phase 2, 39% of the experimental subjects chose lottery 1 and 36% of them chose lottery 2. This result implies that on average the restricted auction mechanism provided better ex ante incentives to direct resources toward entrepreneurial activities. More precisely, the mean of H in Phase 1 is 3.925,
123
Eur J Law Econ (2012) 33:205–229
217
Table 6 MWW test—Phase 1 vs. Phase 2 No. of obs.
S. means
MWW (U)
p value
N1 = 80;N2 = 70
S1 = 3.925;S2 = 2.057
U = 789
p \ 0.0001
Table 7 T test—Phase 1 vs. Phase 2 No. of obs.
S. means
T-test (t)
p value
N1 = 80;N2 = 70
S1 = 3.925;S2 = 2.057
t = 9.600
p \ 0.0001
while the mean of H in Phase 2 is 2.057. The Mann-Whitney-Wilcoxon (MWW) test performed on our data confirms that the two distributions are statistically different at a 0.01 significance level (Table 6). In particular, the MWW test confirms that the subjects who played Phase 1 (under treatment RE) chose on average an higher level of effort than the subjects who played Phase 2 (under treatment UNRE). In order to check for the robustness of this result, we also performed a t-test to see whether the two sample means were statistically different: the result of the t-test is that the two sample means are statistically different at a 0.01 significance level (Table 7). These findings need to be further investigated. In fact, it is worth recalling that the experimental subjects could evaluate their alternative choices only by visual inspection, that is they were not able to calculate the expected payoffs exactly. Nonetheless, under the restricted auction mechanism (Phase 1) 31 subjects (out of 80) chose lottery 4 (i.e., they made the optimal choice) and 33 subjects chose either lottery 3 or lottery 5 (i.e., they made choices very close to the optimal one); under the unrestricted auction mechanism (Phase 2) 27 subjects (out of 70) chose lottery 1 (i.e., the optimal one) and 25 subjects chose lottery 2 (i.e., the closest lottery to the optimal one). Moreover, it is important to notice that in Phase 1 actual choices do not differ from our theoretical predictions. In fact, the optimal level of H in Phase 1 is 5,000, corresponding to lottery 4 in our tables, that is not statistically different from 3.92510 (Table 8); however, in Phase 2 the optimal level of H is 0, corresponding to lottery 1 in our tables, that is statistically different from 2.057 at a 0.01 significance level (Table 9). The latter result implies that subjects over-invested in firm specific effort under the unrestricted mechanism. More precisely, subjects put a larger amount of effort into their firms than that predicted by our theoretical model. In the next section we address this issue.
10 In a related paper, Di Cagno et al. (2004) found different results in terms of correspondence between theoretical predictions and experimental choices. These difference is probably due to the fact that in our experiment subjects have been allowed to train while in Di Cagno et al they were not.
123
Eur J Law Econ (2012) 33:205–229
218
Table 8 T test—Phase 1 vs. theoretical predictions No. of obs.
S. mean
Th. pr.
T-test (t)
p value
N = 80
S = 3.925
S* = 4
t = -0.564
p = 0.574
Table 9 T Test—Phase 2 vs. theoretical predictions No. of obs.
S. mean
Th. pr.
T-test (t)
p value
N = 70
S = 2.057
S* = 1
t = 8.7330
p \ 0.0001
4.2 Interpretation of results As for the first result, the fact that our experimental subjects chose a higher level of firm specific effort under the restricted auction mechanism than under the unrestricted one is consistent with the theoretical predictions. The robustness of this result can also be checked by looking at the choices made in Phase 0. In fact, in Round 1 of Phase 0 (i.e., in a situation that is payoff equivalent to Phase 2) 107 subjects (out of 150) chose either lottery 1 or lottery 2; in Round 3 of Phase 0 (i.e., in a situation that is payoff equivalent to Phase 1) 84 subjects chose one lottery among lotteries 3, 4, and 5. At this point, we wish to investigate further the discrepancy between the theoretical predictions and the behavior of our experimental subjects in both Round 1 of Phase 0 and Phase 2. Actually, there are two possible explanations for our overinvestment result: the first is related to the fact that subjects are usually drawn to the middle lottery when they face an ordered table of lotteries (Andersen et al. 2006; Kirby et al. 1996, 1999); the second is related to the fact that subjects do not respond to strict economic incentives only, but also to moral sentiments, such as the natural propensity of subjects toward socially desirable behaviors (Bowles and Hwang 2008; Camerer 2003; Fehr et al. 2007)11. The first explanation seems to fit the behavior of our experimental subjects in Phase 0, where they actually face an ordered table of lotteries. Since the optimal choice is the lower corner lottery (i.e., lottery 1), many subjects could be induced to make a choice that is closer to the middle lottery. In fact, as we already mentioned, 25 subjects chose lottery 2 instead of lottery 1. However, we do not find the middle bias appropriate to explain the behavior of our experimental subjects in Phase 1. In fact, the story that our experimental subjects are told in Phase 1 provides them with new pieces of information that should alleviate such bias. In other words, the story they are told should convince them that the corner solutions are plausible boundaries.
11
Actually, there is another possible explanation related to possible order effects (Harrison et al. 2005). However, we are not able to control for such effects because it would not make any sense to switch the sequential order between Phase 0 and Phase 2. Moreover, the results of the follow-up experiment described in the next section seem to exclude such effects.
123
Eur J Law Econ (2012) 33:205–229
219
As for the second explanation, the only way to check for it is to make an attempt to remove these motives, that is to convince our experimental subjects of the supremacy of material interests (i.e., strict economic incentives) over their natural propensity toward altruistic behaviors. This is the reason why we designed a followup experiment. In the next section we describe our experimental design and our results.
5 Follow-up experiment Our aim is to check for the behavioral argument made in the previous section. So, we decided to treat our experimental subjects with both the restricted mechanism (Phase 1) and the unrestricted one (Phase 2). In particular, a subsample of our experimental subjects faced a switch from the restricted mechanism to the unrestricted one (FOLUP1), while another subsample faced a switch from the unrestricted mechanism to the restricted one (FOLUP2). The switch from a debtor friendly code to a creditor friendly one was aimed at making our experimental subjects aware of the fact that self-interested optimization rather than ethical behavior was appropriate (Cardenas et al. 2000; Gneezy and Rustichini 2000), or that the investor would like to profit unfairly at their expenses, thereby compromising the subjects’ preexisting predispositions of reciprocity or obligation toward the investor (Falk and Kosfeld 2006; Fehr and List 2004; Fehr and Rockenbach 2003). We appointed for this follow-up experiment those subjects who participated the first experiment. The reasons for selecting the same sample of students as in the first experiment are two: first, we wanted to save on money and time by selecting risk neutral subjects without eliciting their risk attitudes ex novo. However, we could appoint for the experiment only 93 subjects out of the initial 150 because the others were not available anymore12. We divided our experimental subjects into two subsamples. The first subsample was initially treated with Phase 1, and then with Phase 2 (FOLUP1); the second subsample was initially treated with Phase 2, and then with Phase 1 (FOLUP2)13. The design of the follow-up experiment is summarized in Table 10. First, we compare the choices of the two samples under the first treatments they were given (Phase 1 for FOLUP1, Phase 2 for FOLUP2). This is again a comparison between the restricted auction mechanism and the unrestricted one, hence a robustness test to see whether both the results of the first experiment survive or not. 12 One possible criticism to our procedure is that endogenous sample selection may occurr, that is it might be the case that only subjects who performed well in the initial experiment decided to show up again. If this was the case, our results could have been affected by some sort of sample selection bias; in particular, we could have observed our experimental subjects replicating in the follow-up experiment their initial (successful) behavior. However, as it will be shown later on, our experimental results prove that the decisions taken by our experimental subjects in the follow-up experiment differ significantly from those taken in the initial experiment; hence, sample selection bias does not seem to have kicked in. 13 With respect to our first experiment, we deleted the Phase 0 treatment. Again, we decided to save on money because there was no need to make the subjects acquainted with the experimental software that they had already seen in the initial experiment.
123
Eur J Law Econ (2012) 33:205–229
220
Table 10 Follow-up experiment—experimental design No. of subjects
Sequential order First
Second
FOLUP1
46
Phase 1 (RE)
Phase 2 (UNRE)
FOLUP2
47
Phase 2 (UNRE)
Phase 1 (RE)
Second, we investigate whether the subjects who faced a switch from a debtor friendly code (Phase 1) to a creditor friendly one (Phase 2) changed their choices substantially with respect to our previous experiment. The experimental results, described in Tables 11 and 12, are sharper than what we expected. Again it is evident that the restricted auction mechanism was able to direct a higher level of effort toward entrepreneurial activities than the unrestricted mechanism. In particular, the average level of effort chosen in Phase 1 under FOLUP1 is 3.956, while the average level of effort chosen in Phase 2 under FOLUP2 is 2.064. The Mann-Whitney-Wilcoxon (MWW) test performed on our data confirms that the two distributions are statistically different at a 0.01 significance level (see Table 13). In order to check for the robustness of this result, Table 11 FOLUP1— frequencies
Table 12 FOLUP2— frequencies
Lottery
Phase 1
%
Phase 2
%
1
0
0
41
89.1
2
4
8.70
5
10.87
3
11
23.91
0
0
4
14
30.43
0
0
5
16
34.78
0
0
6
1
2.17
0
0
Lottery
Phase 2
%
Phase 1
%
1
15
31.91
1
2
19
40.42
1
2.13 2.13
3
9
19.15
12
25.53
4
3
6.38
19
40.42
5
1
2.13
12
25.53
6
0
0
2
4.25
Table 13 MWW test—Phase 1 (FOLUP1) vs. Phase 2 (FOLUP2) N. of obs.
S. means
MWW (U)
p value
N1 = 46;N2 = 47
S1 = 3.956;S2 = 2.064
U = 234.5
p \ 0.0001
123
Eur J Law Econ (2012) 33:205–229
221
Table 14 T test—Phase 1 (FOLUP1) vs. Phase 2 (FOLUP2) No. of obs.
S. means
T-test (t)
p value
N1 = 46;N2 = 47
S1 = 3.956;S2 = 2.064
t = 8.455
p \ 0.0001
we also performed a t-test to see whether the two sample means are statistically different: the result of the t-test is that the two sample means are statistically different at a 0.01 significance level (Table 14). As for the respondence between the individual behavior and the optimal choices, 14 subjects out of 46 chose the optimal level of effort in Phase 1 (lottery 4), and 27 subjects chose a level of effort very close to the optimal one (lottery 3, or 5). In fact, the average level of effort in Phase 1 is not statistically different from the optimal one (Table 15). However, as in the first experiment, Subsample 2 over-invested in firm specific effort with respect to the theoretical predictions: in fact, the average level of effort in Phase 2 is statistically higher than the optimal one (Table 16). The analysis of the choices in Phase 2 under FOLUP1 is the core contribution of the follow-up experiment. In fact, the choices in Phase 2 (i.e., under the unrestricted auction mechanism) are different from the choices made in Phase 2 under FOLUP2 (Table 17), hence also different from the choices made by the experimental subjects in the first experiment: in particular, they chose on average a level of effort equal to 1.106. This level of effort is not statistically different the optimal level of effort at a 0.01 significance level (Table 18). Table 15 T test—Phase 1 (subsample 1) vs. theoretical predictions No. of obs.
S. mean
Th. pr.
T-test (t)
p value
N = 46
S = 3.956
S* = 4
t = -0.274
p = 0.785
Table 16 T test—Phase 2 (subsample 2) vs. theoretical predictions No. of obs.
S. mean
Th. pr.
T-test (t)
p value
N = 47
S = 2.064
S* = 1
t = 7.389
p \ 0.0001
Table 17 MWW test—Phase 2 (FOLUP1) vs. Phase 2 (FOLUP2) No. of obs.
S. means
MWW (U)
p value
N1 = 46; N2 = 47
S1 = 1.106; S2 = 2.064
U = 1732
p \ 0.0001
Table 18 T test—Phase 2 (FOLUP1) vs. theoretical predictions No. of obs.
S. mean
Th. pr.
T-test (t)
p value
N = 46
S = 1.106
S* = 1
t = 2.340
p \ 0.024
123
222
Eur J Law Econ (2012) 33:205–229
However, a graphical analysis of our experimental data shows that the choices of experimental subjects under the restricted auction are always the same, no matter whether the restricted auction is played first or second. On the contrary, under the unrestricted auction mechanism the behavior of our experimental subjects change when they play it second. This evidence is illustrated in Figs. 3 and 4. We conclude that, when the subjects face a creditor friendly code after they have faced a debtor friendly one, the over-investment result disappears. In other words, when subjects perceive that the creditor wants to appropriate all the surplus, they react by responding to strict economic incentives only. These findings suggest that debtor friendly bankruptcy codes do not fully discourage socially desirable behaviors. Hence, switching from debtor friendly codes toward creditor friendly ones may end up killing the natural propensity of economic agents to fulfill their obligations. A final consideration regards the choices in Phase 1 under FOLUP2. In this case, the average level of effort is 3.979, which is not different from the average effort chosen in Phase 1 under FOLUP1. This result confirms the robustness of the incentives embedded into the restricted auction mechanism.
Fig. 3 Phase 1—comparison across treatments
Fig. 4 Phase 2—comparison across treatments
123
Eur J Law Econ (2012) 33:205–229
223
6 Concluding remarks We designed a computerized experiment to test whether the restricted auction mechanism proposed by Berkovitch, Israel and Zender in 1997 is an optimal bankruptcy law or not. As mentioned above, an optimal bankruptcy law is a mechanism that is able to guarantee both ex ante and ex post efficiency in firmspecific investment decisions when moral hazard is binding. Here, we especially focused our attention on ex ante efficiency. In our experiment we achieved two main results: 1.
2.
The restricted auction mechanism induced entrepreneurs to allocate more effort into their firm than the unrestricted auction mechanism. More precisely, our experimental subjects chose on average a higher level of firm-specific effort under the restricted auction mechanism than under the unrestricted auction one. Our experimental subjects over-invested in firm specific effort under the unrestricted mechanism. More precisely, subjects put a larger amount of effort into their firms than that predicted by our theoretical model.
Our first result is very sharp and confirms the theoretical predictions. From a policy point of view, since subjects properly responded to different ex ante incentives by varying their firm specific effort, our experimental findings suggest that it is very important to provide debt financed entrepreneurs with correct ex ante incentives. Here, it is worth noticing that both the theoretical and the experimental results are driven by the assumption that the bank and the entrepreneur have the same information and that such information is not available to other agents. It would be very interesting to repeat our experiment under different hypothesis about the informational setup. Our second result is experimentally robust, but difficult to justify on theoretical grounds; our interpretation of such result is behavioral. More precisely, we argue that the choices of our experimental subjects were driven not only by strict economic incentives, but also by ‘‘moral sentiments’’, such as the natural propensity of subjects toward socially desirable behaviors. To prove this argument we designed a follow-up experiment in which we removed these motives. The over-investment result vanished once the experimental subjects realized that it was appropriate to respond to material interests only. Our over-investment result suggests that debtor friendly bankruptcy codes do not fully discourage socially desirable behaviors. Hence, switching from debtor friendly codes toward creditor friendly ones may end up killing the natural propensity of economic agents to fulfill their obligations.
Appendix: instructions Introduction The instructions are simple. If you follow them carefully you can earn up to 50,000 Experimental Credits (25 Euros).
123
224
Eur J Law Econ (2012) 33:205–229
The experiment consists of 2 different phases. In each phase the computer will work out your payoff. At the end of the experiment, you will draw one of the 2 phases and you will be paid immediately and in cash the money you made in the drawn phase. Pay attention: your gain in each phase depends on your choices. Thanks for your participation. If you have any doubt, do not hesitate to contact the experimenters for further information. Phase 0 Phase 0 consists of 3 ROUNDS. In each ROUND, the computer shows you 6 lotteries sequentially. Each lottery is represented by three colored areas. The possible outcomes of each lottery are displayed on your screen (and in the payoff tables on your desk) (Fig. 5). You have to choose the lottery you prefer to play. The computer will force you to scroll all lotteries before making your choice. After your choice, you move to the next ROUND. Each ROUND works in the same way. At the end of the third ROUND, one ROUND among the three you played will be randomly selected and the lottery you chose in that ROUND will be played for real: the computer swings a indicator over the three areas representing that lottery and works out your payoff according to the point where the indicator stops (Fig. 6). Take a closer look at the payoffs’ tables. Notice that, if the indicator stops at any point of the area where the payoff is the payoff of the outside option (blue area), you
Fig. 5 Phase 0—lotteries. Translation: ‘‘Per scorrere…’’: To see the lotteries use arrows or select ENTER to move forward and BACK SPACE to move backward. To look at a single lottery press F1,…, F6. To choose a lottery press F12. ‘‘Per maggiori…’’: For further information press I
123
Eur J Law Econ (2012) 33:205–229
225
Fig. 6 Phase 0—indicator. Translation: ‘‘Per scorrere…’’: To see the lotteries use arrows or select ENTER to move forward and BACK SPACE to move backward. To look at a single lottery press F1,…, F6. To choose a lottery press F12. ‘‘L’esperimento…’’: The experiment is over. Your payoff is 3,500 EC. ‘‘Per maggiori…’’: For further information press I
get a fixed payoff which is different from lottery to lottery; if the indicator stops at any point of the area where the payoff is the Nash bargaining payoff (red area), again you get a fixed payoff which is different from lottery to lottery; if the indicator stops in the area where the payoff is the realized cash flow (yellow area), you get a payoff that depends on the exact point where the indicator stops. This payoff is bounded between a minimum and a maximum, as you can see in the payoffs’ table. The minimum and maximum payoffs are different from lottery to lottery. Before the experiment starts, you are required to play with Phase 0 twice in order to get acquainted with the software and to check the consequences of your choices in terms of payoffs. Phase 1 You are an entrepreneur who has already run a project, financed by a bank (the computer). The final outcome of the investment is uncertain and depends partly on your effort and partly on your luck. You have to decide how much effort to put in your firm, considering the three different final situations which may occur (Fig. 7). Final situations 3 final situations may occur: 1.
Your firm can turn out to be NOT ECONOMICALLY VIABLE and IN FINANCIAL DISTRESS (i.e., it does not generate any wealth and it is not able to pay back its debt);
123
Eur J Law Econ (2012) 33:205–229
226
Fig. 7 Phase—choice of effort. Translation: ‘‘Per scorrere…’’: To see the lotteries use arrows or select ENTER to move forward and BACK SPACE to move backward. To look at a single lottery press F1,…, F6. To choose a lottery press F12. ‘‘Per maggiori…’’: For further information press I
2.
3.
Your firm can turn out to be ECONOMICALLY VIABLE, but IN FINANCIAL DISTRESS (i.e., it generates wealth, but it is not able to pay back its whole debt); Your firm can turn out to be both ECONOMICALLY and FINANCIALLY VIABLE.
In the first situation, your firm goes bankrupt and you come back home with a payoff (blue area in the payoffs’ table) that depends on the level of effort H you have chosen. The liquidation value of your firm is 90,000 Experimental Credits. In the third situation, your firm is healthy and you come back home with a payoff which is bounded between a minimum and a maximum (yellow area in the payoffs’ table). The minimum and the maximum depend on the level of effort H you have chosen. In the second situation, your firm is auctioned off and you have to bid for it (red area in the payoffs’ table). The auction It is a SECOND-PRICE SEALED-BID AUCTION that you and an outside investor (played by the computer) participate. The bank that financed your project is not allowed to participate the auction. The auction mechanism is simple and the computer will tell you what to do at any stage (Fig. 8). First of all, the computer collects the bid of the outside investor. After that, it asks you to type your bid. Then, the computer reveals the bids and announces the winner, i.e. the one who made the highest bid. The winner pays for the firm the price offered
123
Eur J Law Econ (2012) 33:205–229
227
Fig. 8 Phase 1—auction. Translation: ‘‘Per scorrere…’’: To see the lotteries use arrows or select ENTER to move forward and BACK SPACE to move backward. To look at a single lottery press F1,…, F6. To choose a lottery press F12. ‘‘Dopo aver…’’: After you made your bid press ENTER. ‘‘Il tuo…’’: Your competitor bids EC…. ‘‘Digita…’’: Input your bid. ‘‘Per maggiori…’’: For further information press I
by the other bidder, which is the SECOND PRICE. For instance, if you bid 100,000 Experimental Credits and the outside investor bids 50,000 Experimental Credits, you win the auction and the price you pay for the firm is 50,000 Experimental Credits. If you win the auction, you earn a payoff which is bounded between a minimum and a maximum depending on the level of effort H you have chosen (‘‘MIN’’ and ‘‘MAX’’ of ‘‘WON’’ columns of the red area in the payoffs’ table). If you do not win the auction, you earn a fixed payoff (‘‘LOST’’ column of the red area in the payoffs’ table). The choice of the entrepreneurial effort The computer shows you 6 screens, one for each level of H (the level of effort you must choose to put into your firm). Each screen consists of 3 colored areas that represent the possible outcomes of your choice of H. These outcomes are displayed in the attached payoffs’ tables. After you have seen all the screens at least once, you can choose the level of H you prefer. After you chose a certain H, the computer shows you the corresponding screen. Then, it swings a indicator over the three colored areas representing your choice of H and computes your payoff according to the point where the indicator stops. Take a closer look at the payoffs’ tables. Notice that, if the indicator stops at any point of the area where the payoff is the outside option payoff (blue area), you get a
123
228
Eur J Law Econ (2012) 33:205–229
fixed payoff which is different for each H. If the indicator stops at any point of the area where the payoff is the Nash bargaining payoff (red area), you participate the auction and get a payoff which depends both on whether you have or you have not won the auction, and the H you have chosen. If the indicator stops in the area where the payoff is the realized cash flow (yellow area), you get a payoff which is bounded between a minimum and a maximum. Notice that the minimum and maximum payoffs are different for each H. For instance, if you have chosen H equal to 5,000 and the indicator stops in the blue area, you get 2,500 Experimental Credits; if the indicator stops in the red area and you have not won the auction, you get 2,500 Experimental Credits; if the indicator stops in the red area and you have won the auction, you get a payoff from a minimum of 2,500 Experimental Credits up to 22,500 Experimental Credits; if the indicator stops in the yellow area, you get from a minimum of 2,500 Experimental Credits up to a maximum of 45,000 Experimental Credits. Phase 2 Instructions for Phase 2 are the same as those for Phase 1, except for the fact that experimental subjects are now advised that the bank is not only aware of the liquidation value of the firm, but also of its output. From this difference the new description of the auction mechanism follows. The following sentence: ‘‘It is a SECOND-PRICE SEALED-BID AUCTION that you and an outside investor (played by the computer) participate.’’ is replaced with: ‘‘It is a SECOND-PRICE SEALED-BID AUCTION that you and the bank that financed your project (played by the computer) participate.’’ Notice that Phase 2 payoffs are the same as those for Phase 1.
References Aghion, P., Hart, O. D., & Moore, J. (1992). The economics of bankruptcy reform. Journal of Law, Economics and Organization, 8, 423–446. Andersen, S., Harrison, G. W., Igel Lau, M., & Rutstrom, E. E. (2006). Elicitation using multiple price list formats. Experimental Economics, 9, 383–405. Barr, A., & Packard, T. (2002). Revealed preference and self insurance: Can we learn from the self employed in Chile? Policy Research Working Paper, World Bank. Bebchuk, L. A. (1988). A new approach to corporate reorganization. Harward Law review, 101, 775–804. Berkovitch, E., & Israel, R. (1999). Optimal bankruptcy law across different economic systems. Review of Financial Studies, 41 487–497. Berkovitch, E., Israel, R., & Zender, J. F. (1997). Optimal bankruptcy law and firm-specific investments. European economic Review, 33, 441–464. Bowles, S., & Hwang, S. H. (2008). Social preferences and public economics: Mechanism design when social preferences depend on incentives. Working Paper.
123
Eur J Law Econ (2012) 33:205–229
229
Camerer, C. (2003). Behavioral game theory: Experimental studies on strategic interaction. Princeton University Press. Cardenas, J. C., Stranlund, J. K., & Willis, C. E. (2000). Local environment control and institutional crowding out. World Development, 28, 1719–1733. Di Cagno, D., Sabato, V., & Spallone, M. (2004). Bankruptcy laws: Optimal incentives to entrepreneurs. Economia, Societa` e Istituzioni, 16, 171–199. Falk, A., & Kosfeld, M. (2006). The hidden costs of control. American Economic Review, 96, 1611–1630. Fehr, E., Gaechter, S., & Schmidt, K. M. (2007). Fairness and contract design. Econometrica, 75, 121– 154. Fehr, E., & List, J. (2004). The hidden costs and returns of incentives: Trust and trustworthiness among ceo’s. Journal of the European Economic Association, 2, 743–771. Fehr, E., & Rockenbach, B. (2003). Detrimental effects of sanctions on human altruism. Nature, 422, 137–140. Gaechter, S., & Riedl, A. (2004). Dividing justly in bargaining problems with claims. Tinbergen Institute Discussion Paper. Gaechter, S., & Riedl, A. (2005). Moral property rights in bargaining with infeasible claims. Management Science, 51, 249–263. Gneezy, U., & Rustichini, A. (2000). A fine is a price. Journal of Legal Studies, 29, 1–17. Harrison, G. W., Johnson, E., McInnes, M. M., & Rutstrom, E. E. (2005). Risk aversion and incentive effects: Comment. American Economic Review, 95, 897–901. Herrero, C., Moreno-Ternero, J. D., & Ponti, G. (2002). An experiment on bankruptcy. Working Paper. Holt, C., & Laury, S. K. (2002). Risk aversion and incentive effects. American Economic Review, 92, 1644–1655. Holt, C., & Laury, S. K. (2005). Risk aversion and incentive effects: New data without order effect. American Economic Review, 95, 902–912. Jensen, M. (1991). Corporate control and the politics of finance. Journal of Applied Corporate Finance, 4, 13–33. Kirby, K. N., & Marakovic, N. N. (1996). Delay-discounting probabilistic rewards: Rates decrease as amounts increase. Journal of Experimental Psychology, 3, 100–104. Kirby, K. N., Petry, N. M., & Bickel, W. K. (1999). Heroin addicts have higher discount rates for delayed rewards than non-drug using controls. Journal of Experimental Psychology, 128, 78–87. Miller, L., Meyer, D. E., & Lanzetta, J. T. (1969). Choice among equal expected value alternatives: Sequential effects of winning probability level on risk preferences. Journal of Experimental Psychology, 79, 419–423. Noussair, C., Robin, S., & Ruffieux, B. (2004). Revealing consumers’ willingness-to-pay: A comparison of the BDM mechanism and the Vickrey auction. Journal of Economic Psychology, 25, 725–741. Schubert, R., Brown, M., Gysler, M., & Brachinger, H. W. (1999). Financial decision-making: Are women really more risk-averse? American Economic Review (Papers and Proceedings), 89, 381– 385. White, M. J. (2005). Economic analysis of corporate and personal bankruptcy law. NBER Working Paper.
123