MODELS OF OTHER REGARDING PREFERENCES – PART A
Introduction
Economics is usually referred to as a “social science” because as economists, we seek to understand how individuals react, in a particular environment and how this plays a role in the allocation of goods and services. Economic reasoning’s have argued that it is all characteristically based on the self-interest hypothesis which is the assumption that the dominant drive in individuals is a rational striving to maximize self-interest (Kamarck, 2002). What isn’t talked about in much detail is that this assumption is not considerate of any heterogeneity in regards to other-regarding snd social preferences. Our models usually allow some heterogeneous tastes but in the importance of social preferences mainstream economics always comply to the extreme assumption of self-interest only. There have been a few types of experimental research dedicated to “other- regarding preferences” but since, it is difficult to measure, is a topic that isn’t widely spoken about in detail. Given this, it is important, to appreciate and measure the “other-regarding” preferences of individuals precisely.
The purpose of this report is to analyse the models which show the effects social and other preferences have on our decisions. I will assess the methodological nature of the paper and examine the ways in the models used have captured inequity aversion. I will thoroughly discuss which methods used to measure “other-regarding preferences including the innovative work of Fehr and Schmidt (1999 ; FS) and Bolton and Ockenfels (2000 ; BO) and will decide one I think may be the most suitable.
The procedures used by FS and BO are directed toward two main issues: the extent to which choices depends on the outcome and how these outcomes are achieved as well due to the role which reciprocity and observed intentions plays in these games and the scope of players’ concerns for own income relative to others in the group. These methods will be presented in Section 2 and will further discuss their methodological issues in Section 3. This essay will finish with a conclusion discussing which model would be suitable if hypothetical experiment were to be run.
METHODOLOGY OF OTHER-REGARDING PREFERENCES AND PERFORMANCE OF THESE MODELS
A lot of work concerning other-regarding preferences before, 1995 were based on the results gained from an ultimatum game. During this game two players, A and B, would pick how to divide a sum of money, n, between them. The first player, A, makes an offer to player B, which if it is accepted is divided as player A suggests. However, if player B rejects the offer, both players get nothing. There are many Nash equilibria in this game, however, the subgame perfect Nash equilibrium (SPNE) outcome is where player A offers the least amount of money required (or a small positive) under the assumption that players are only care about own their income (Cooper & Kagel, 2013).
FEHR AND SCHMIDT/ BOLTON AND OCKENFELS
The Fehr-Schmidt (FS) and Bolton and Ockenfels (BO) models assumes that the utility gained from an outcome, for a player, is dependent on the player’s payoff as well as how it is compared to the other players’ payoffs. Both models are based on the assumption that being worse off than others is a worse situation than being better off. We see that in both models, the ultimatum and dictator games, there is concern for “own income” only. Both models are fairly tractable in that the players’ preferences are only dependent on the outcomes of the game and not the steps they have taken to achieve this outcome. This makes it easier to apply both models to new games and compare results.
MENU DEPENDENCE – FFF
Falk, Fehr, and Fischbacher (2003; FFF) investigated the role of intentions by using a group of four small ultimatum games in which the Proposer was allowed to choose between x and y.
All four games were played in different orders with no feedback given after each game. This strategy was adopted so that Responders had enough time to indicate their choices for both the reference point allocation and the alternative allocation.
For all four of the games, the reference point allocation x was the same to keep the game even (an 8, 2 split where the Proposer’s response is exposed first). In one game the second possible allocation was a 5, 5 equal split compared to which seems better overall than the 8, 2 choices which was not a fair decision. The second and third games included an 8, 2 split which was paired with a 10, 0 split and a 2, 8 division. In comparison to the 10, 0 option the 8, 2 split seems relatively reasonable, with the 2, 8 split forcing the Proposer to choose between being fair to either himself or the Proposer. Finally, as a control game, they paired the 8, 2 allocations with itself, so that the Proposer had no choice but to offer an 8, 2 option.
It was found that the rejection in the game with the 5,5 alternative turned out to be significantly higher than all other games. Additionally, the differences between the 10,10 and 2,8 games were important to note.
FFF came to conclusion that differences in rejection rates between all four treatments clearly illustrate that intentions were significantly important. Finally, there is an 18% rejection rate under for the (8, 2) allocation when the Proposer did not have a choice, which FFF quotes as “evidence of pure income inequality aversion”
GUTH AND VAN DAMME – ULTIMATUM GAMES
Güth and van Damme (1998; GD) introduced ultimate games, with a productive framework to analyse the testing of the BO and FS models. In this model, the first player, A, proposes to split income between players A, B and C. Player two, B, then accepts or rejects this “split”” with the outcome of dividing if B accepts and a result of zero if B declines the offer. Player three, in this game, C, is a dummy player. They will have the same role as player two, in the sense of the “dictator game”. GD found that all Proposers (player A) took advantage of the dummy status of player 3 meaning they essentially divided the money between themselves and Y. The possibility of rejecting the offer, was very close to zero. (Güth & van Damme, 1998).
These result are extremely coherent with the model of Bolton and Ockenfels (1998) were the utility of other regarding components is evaluated relative to the social reasoning of equal shares for all players. It is believed that the addition of a third player, dummy variable C, to the ultimatum game changes that equal division “social norm” from 1⁄2 to 1/3, leading to a prediction of a higher volume of acceptance rates. However, also in the BO (1998) results, other-regarding preferences seem to only depend on own shares of overall payoffs. The distribution of payoffs in comparison to the players overall had little to no impact on utility. This is in line with the observation of Güth and van Damme that no rejections could be due to the lower share ratio allocated to the Dummy player, C.
This game was later modified by Kagel and Wolfe (2001) to obtain more demanding results.
Firstly, the responding player was randomly selected out of B and C after player A had made their choice. This was designed to maximize the chance of the Responder offering a relatively lower offer as Proposers, not knowing the identity of the player prior to making an offer, could no longer pay off the Responder at the expense of the Dummy player.
BEREBY-MEYER AND NIEDERLE – THREE PERSON GAMES
The two different three person games that Bereby-Meyer and Niederle (2005; BMN) reported were proposed to differentiate the presence of outcome based preferences and mutuality in bargaining games.
The first game was called the third-party rejection payoff games (TRP) and it is similar to the three player game above with Kager and Wollfe(2003) in the sense that a Proposer will make an offer to a Responder, which will have three variable prizes for player C, the “dummy” play. For this example, we can assume that the prizes are – £0, £5, or £10. There are many differences between this model and the KW model as each person plays the game once under each treatment with no feedback on outcomes until the session is over. Then, we see that the Proposer, player one, chooses to split the money, £10, strictly between themselves and the Responder, player two.
In the second game presented by Bereby-Mayer and Niderle, (referred to as proposer rejection payoff games – PRP) the Proposer, player one, is required to split the £10 prize between player two, the Responder, and the “dummy” variable. If the Responder accepts, the division is compulsory. More commonly, it is expected that he Responder, player two, will accept this as it is in their interest to win some of the prize. On the unlikely occasion that they reject this offer, both the dummy variable and Responder leave with nothing and the Proposer, player one, will get the rejection payoff of £0, £5, or £10.
The results of this experiment show that there were significantly higher rejection rates in the TRP game than in the PRP game for all payoff level. This is highly inconsistent with the outcome based models which had predicted that there would be no difference in players’ behaviour according to the positions in the allocation process. So, this brings us to think there may be multiple reasons that are playing a role in “other regarding behaviour” in these games. The data give is suggesting that the intentions of the players, play a larger role than the allocation of outcome but also that the role of pure outcome based preferences is not equal to zero.
Conclusion
The methods presented within this essay describe are relative approaches used to measure other regarding preferences in experimental situations. All theories have been able to validate the number of theories established in Behavioural Economics. Of course, for all situations, we have a method which is suitable but some methods may not be suitable in various circumstances, as they can create bias or results which aren’t correct for the experiment.
Personally, if I were to run an experiment surrounding the effects of social preferences on the individual and which effects were most detrimental to the consumer’s choice, I would choose the model posed by Falk, Fehr, and Fischbacher (2003; FFF). The results are coherent with those provided in the BO and FS model and this model eliminates any presence of bias. The model has specifically captured ideas that the FS model couldn’t. From a practical perspective, the two models, BO and FS, make similar predictions: we observe that the FS model is likely to be sensitive to variations in the allocation of the different payoffs over other individuals, contrasting with the modifications of the average payoffs of all of the other individuals. This doesn’t not have an effect on the predicted behaviour in the BO model.
the capacity to predict a broad range of outcomes as a function of possible parameter values within a given population.