CEEL Projects

Projects - Experimental Economics

Last update:

Target the two

Massimo Egidi.
The first experiments conducted by M. Cohen and Bacdayan (1994) with Target the Two provide evidence that after the initial learning period players' behavior becomes more and more routinized . As noted in the introduction there are two complementary ways to measure routinization: one may either focus on the repetitive sequences of patterned behaviors, as in Cohen and Bacdayan (1994, page 558), or also include considerations regarding the mental model involved. This second experimental level is suggested by the incompleteness that arises if observation is restricted to the sequences over time. In Cohen and Bacdayan' experiment, routines are revealed as repeated action sequences. As we noted before, to define what it means that two sequences are "the same" depends upon the representation of the problem and the level of observation involved. Cohen and Bacdayan point out, for example, that the sequence of moves Up-Up-anything-Target (UU*T) is often used by subjects (the median pair solves more than one game in four playing UU*T), and that there is a statistically significant positive correlation between the number of times UU*T is played and the number of moves used in the tournament. However, other relevant repetitive sequences exist like Search DownC-Search DownN-Pass-Up (CNPU) which are the typical reactions of over-cooperative players to particular configurations The players realize that their partner is searching for a card (C) which is presumed to be useful. They search in the opposite covered position (N) in order to offer the useful card to the partner (U) after he has passed. The sequence is therefore CNPU. Of course the sequence: search DownN-search DownC-Pass-Up reflects the same procedure in terms of the mental model involved but it corresponds to a different sequence of actions. The point is, therefore, that CNPU should be considered as perfectly equivalent to NCPU in terms of routinized thinking, although they differ as regards the sequence of actions. Therefore, there is a limit to defining routinized behaviors as repetitive behaviors which arise in similar conditions. As the above example shows, two phenomena can be considered as similar in relation to the level of abstraction at which the observers are thinking, and more generally in relation to their mental representation of the phenomena. To overcome this difficulty it is convenient to define routinized behaviors as behaviors based on fixed condition-action rules, following March and Simon (1993). As we have seen, March and Simon refer to routinized behaviors as sequences of choices reduced to a "systematic computing routine", instead of appealing to a set of condition-action rules: yet it is well known in Computation Theory that the two computing devices are equivalent (Cutland 1988), in the sense that it is always possible to build a system of condition-action rules that is equivalent to a "computing routine", i.e. to a Turing machine. In defining "routinized behaviors" we must therefore carefully distinguish between the sequence of actions realized over time, and the set of rules which generate this sequence . If we identify a set of rules governing the behavior of a player, we can resolve the conceptual difficulty raised by defining routinized behaviors as "repetitive" : in fact a set of condition-action rules prescribes the action to be made for every condition of the game, and therefore implicitly defines the abstraction level. As we shall see, in our experiment identification of a set of rules governing players' behaviors is possible and statistically significant. To find the set of condition-action rules which allow players to coordinate their efforts and achieve the common goal, we must find a way to represent the problem (the goal to be achieved) in terms of sub-goals and discover the logic of collective action which is involved.

Tax Evasion Multiple Agent Game (temag)

Luigi Mittone.
La metodologia sperimentale si può considerare un valido strumento empirico per analizzare il comportamento di frode fiscale in un ambiente isolato dall'influenza di fattori esterni. Il disegno base che caratterizza la maggior parte degli esperimenti sul comportamento dei contribuenti è sintetizzabile come segue. Viene selezionato un campione di soggetti, in genere studenti universitari, ai quali è attribuito un determinato compito (che può essere una semplice dichiarazione fiscale del reddito, oppure una serie di decisioni più complesse attraverso cui il vero obiettivo dell'esperimento è mascherato) che deve essere eseguito all'interno di un laboratorio sotto il controllo degli sperimentatori. La simulazione, in genere, si svolge in un contesto dinamico, estendendosi su vari rounds sperimentali: all'inizio di ogni round dell'esperimento, a ciascun individuo è attribuita una somma in denaro ed essi devono decidere quanto dichiarare di tale reddito percepito; sul totale del reddito riportato vengono poi pagate delle imposte in base ad una determinata aliquota fiscale prefissata (t), mentre il reddito non dichiarato può essere scoperto con una qualche probabilità (p) e, in tal caso, al soggetto evasore è imposto il pagamento di una multa (f) sulle imposte non versate. Al termine dell'esperimento, ogni partecipante viene retribuito con un compenso (diciamo, i guadagni accumulati) che dipende dalla sua performance durante la simulazione. All'interno di tale sistema microeconomico possono poi essere introdotte diverse variazioni di politica fiscale (cambiamento dell'aliquota d'imposta, della probabilità di controllo, della dimensione della sanzione), come pure un'ampia gamma di fattori di "contestualizzazione", osservando così la reattività dell'evasione individuale al mutare di certi fattori chiave.
Ad ogni esperimento, realizzato nel Laboratorio di Economia Sperimentale dell'Università di Trento, partecipa, solitamente, un gruppo di 30 studenti ai quali non è comunicato il numero totale di rounds di cui consiste la simulazione, che è predeterminato a 60. All'inizio di ogni round ad ogni partecipante è assegnato un reddito in lire, decrescente da 1000 £ (primi 48 rounds) a 700 £ (ultimi 12 rounds), specificando che tutti i guadagni accumulati al termine dell'esperimento verranno distribuiti ai rispettivi vincitori. Ai soggetti viene, inoltre, precisato che il reddito non dichiarato non è soggetto a imposizione fiscale, mentre sulla quantità riportata grava una aliquota d'imposta variabile durante la simulazione (0.20, primi 10 rounds; 0.30, rounds 11-30; 0.40, rounds 31-48; 0.57, ultimi 12 rounds); ognuno e sottoposto ad una probabilità di audit, comunicatagli sullo schermo del suo p.c. e modificata nel corso dell'esperimento (6% fino al round 20, 10% fino al round 40 e 15% negli ultimi 20 rounds) e, nel caso venga controllato, l'accertamento si estenderà anche alle dichiarazioni fiscali dei precedenti tre rounds, comportando il pagamento di una sanzione pari ad un qualche multiplo (ad esempio, 4.5) dell'ammontare complessivo di imposte evase. All'inizio di ogni round viene anche visualizzato sul terminale di ogni partecipante il reddito assegnato loro in quel round, l'importo dell'imposta da versare, il loro nuovo "bilancio" al netto delle imposte e delle multe pagate, la percentuale di evasori del round precedente, la percentuale di indagati del round precedente e il gettito medio riscosso dallo Stato da tutti i partecipanti. Vengono, infine, introdotte diverse variazioni del contesto sperimentale: alcune simulazioni, ad esempio, hanno considerato l'esistenza di due classi sociali (benestanti e indigenti), assegnando un reddito differente all'una e all'altra, altre invece hanno presentato l'esperimento come un semplice gioco d'azzardo, con l'utilizzo di una terminologia neutrale rispetto al fenomeno indagato (restituzioni di danaro al posto di imposte, obbedienza all'ordine anziché adempimento fiscale, ecc.), mentre in altri casi ancora si è introdotta la presenza di una destinazione da parte dello Stato del gettito introitato (erogazione di trasferimenti individuali di reddito con finalità di equità ridistributiva, oppure fornitura di un bene pubblico con possibilità di scelta dell'impiego delle entrate fiscali da parte dei partecipanti tra diverse alternative presentate).

Coordination, flexibility and the emergence of grammar

Massimo Warglien in cooperation with Reinhard Selten.
Our experiments are motivated by an interest in the emergence of language as a solution to a coordination problem. Thus, our focus is on language as an interactive phenomenon. In particular, we are exploring simple coordination games in which language is not just instrumental to achieve a solution of the game (like in the cheap talk literature), but rather is the very solution of the game.
In particular, we are interested in the emergence of the generative aspect of language - the fact that agents can create "grammatical" languages that will allow them to successfully coordinate "ex-ante" in situations never met before. A language needs not have necessarily such quality - one may think of an arbitrary conventional association of signs (e.g. sounds) and "meanings" as a primitive, grammarless language with no generative structure in it (a protolanguage, in Bickerton's terms) . We investigate under which conditions a grammatical language emerges instead of a protolanguage as a response to coordination problems in novel environments.

Short-term memory and agents' heterogeneity

Massimo Warglien in cooperation with Yaakov Kareev.
Short term memory is a fundamental bottleneck of human cognitive processes. In this project, I explore some implications of individual differences in s-t memory capacity for heterogeneity of behavior in individual decision-making and games.
In particular, two experimental strands are under development:

• s-t memory capacity and the perception of risk (how do s-t memory capcity constraints affect the perception of risk? What are the implications for the trade of riskyt assets?)
• s-t memory capacity, complexity and mental representations of games (how does s-t memory capacity limit individual ability to edit complete/correct representations of strategic interactions (games)?

Laboratory experiments as a tool for organizational design

Alessandro Rossi and Massimo Warglien
This research project uses laboratory experimentation as an empirical research tool for the design of organizations. Main research topics are the division of labor among organization members and managerial incentive systems. The laboratory is used to study the emergence of organizational phenomena and to design organizational policies, both to test theoretical predictions, and to pretest changes in managerial policies in actual organizations. In this sense, experiments are used to gather evidence of effectiveness of organizational changes before actual implementation in the field.
An example of this stream of research is in Rossi and Warglien, 1999

Fairness and reciprocity within and between organizations

Alessandro Rossi and Massimo Warglien
Abstract ricerca: This research project focuses on the impact of fairness and reciprocal behavior among and within organizations. The former approach is currently investigating how bargaining takes place within chains of organizations (such as industry supply chains) and the role of reciprocity and fairness in affective the final outcome of the negotiation process. Experiments results are also compared to field evidence on industry bargaining within supply chains. The latter approach investigates experiments based on games where hierarchical structures are introduced, allowing to simultaneously study a vertical dimension of interaction (between superiors and subordinates) and a horizontal dimension of interaction (such the one between teammates or peers).
An example of this stream of research is in Rossi and Warglien, 2001

The spontaneous insurgence of altruism as incentive to efficiency in organisations

Luigi Mittone
The "significant regularity" is displayed by organisations in which one observes the operation of processes optimising the skills of 'strong' (or highly productive) workers driven by non-contractual and non-monetary incentives due to the presence of 'weak' workers, for whom they feel solidarity. This type of organisation can be imagined as a transmission line which promotes the individual search for productively more efficient solutions by means of non-contractual and non-monetary incentives. These incentives take the form of positive externalities in favour of the firm, and they are sustained by a strong sense of group membership. The actual existence of firms of this kind, which would be a manager's paradise, can be envisaged in the set-up to which Williamson alludes - that is, it could come about by virtue of the relationships between firms' production choices and the special nature of the motivations that influence the behaviours of individuals who work in a group.
These non-contractual incentives, sustained by a "moral" or ethical system owned by some of the workers, produce effects that contribute to generate a public good environment. More precisely the positive effects produced on the efficiency level of the organisation through the exceeded effort "spontaneously offered" by the ethically motivated workers can be conceptually assimilated to a public good. The positive result of the exceeded effort in fact is shared with all the participants of the organisation therefore one could say that the motivated workers are producing a sort of public good.
There are three main differences between the classical treatment of the spontaneous provision of public goods and our approach. The first discrepancy is that our agents act in a "contractual" environment. The second difference is that we consider heterogeneous agents represented by the workers and the firm. Finally the third difference is that we are looking for spontaneous provision of public goods in a repeated choices environment, but without any possibility to implement a reciprocity mechanism. The contractual context here considered is the one defined by a labour contract and the interaction is mainly of two kinds; among the workers and between the workers and firm itself. The absence of a reciprocity mechanism, i.e. the impossibility for the agent to implement a punishment device, allows to isolate the "altruistic" behaviours and to investigate on their effects on the efficiency of the production process. Isolating a dynamic process of altruistic interaction that produces efficiency is equivalent to simulate a context very near to the one just described, that is a virtuous process of production of positive externalities in favour of the firm.
The puzzle experiment consists in proposing the collective composition of a puzzle expressly conceived for the experiment at issue to groups of 4 students. The experimental subjects are randomly selected among economics students of the University of Trento. Each group is made up of two girls and two boys. The rules of the game are imparted to all the players of each group according to a fixed scheme: therefore each player knows the rules and is informed on the fact that the other players know the rules of the game. The game structure is the same for each group as well as the distribution of puzzle cards to the players. Therefore the starting situation is the same for all the experimental groups.
The puzzle has to be composed in the following way: first of all, experimental subjects have to compose the first "sub-puzzle", that is the central octagon which is divided into 8 parts. The first "sub-puzzle" ended, they have to go on with the composition of the second "sub-puzzle", then of the third one and so on. Each "sub-puzzle" has to be composed following the direction indicated by the supervisor/experimenter at the beginning of the game.
Each player faces during the puzzle composition what we define as "critical points" in this experiment, that is those turns in which the player does not have the needed puzzle card. In those cases the following player (if he/she has the puzzle card) can choose between the two following alternatives :
to position the puzzle card in preceding player's place. In this case the player who has positioned the puzzle card in preceding player's place avoids the "reward circle" reduction of £. 1000 and loses £. 300 of his/her own final reward. We count this behaviour as 1 move. If a player intervenes in an "altruistic" way, the game goes on with the player who follows the one who has moved in the "altruistic" way.
Not to position the puzzle card in preceding player's place. In this case, the player who decides not to position the puzzle card induces the "reward circle" reduction of £. 1000. We count this behaviour as 2 moves. The game goes on with the player who follows the one short of the puzzle card. If the player does not move in the altruistic way, the game goes on with the player who has not behaved in the altruistic way. In this case the player who has not intervened in the altruistic way does not loose anything of his/her own individual reward, but s/he causes the increase of the number of moves necessary to complete the puzzle and, as a consequence, the decrease of the "reward circle". The reward circle is an imaginary circle that represents the experimental subjects' monetary rewards and has a starting contents of £. 150.000. The final contents of this circle is divided among the 4 players of each group according to the moves occurred during the experiment. The final contents of the circle is an inverse function of the number of moves done to compose the puzzle: more precisely, each move in excess of the 120 minimum moves necessary to complete the puzzle causes a circle's decrease of £. 1000 :
x = f - (n) ,
where x = circle's final dimension
n = number of moves
The final contents of the circle is equally divided among the 4 players. The individual reward depends also on the number of times the player has added a puzzle card out of his/her turn, in order to intervene in favour of another player. In order to know when the experiment is completed how many times each player has added a puzzle card out of his/her turn in favour of another player and how many times a player has not acted in an "altruistic" way, we count the labels connected with each critical point that the supervisor has collected during the composition of the puzzle: blue labels represent "unsuccessful altruistic moves" (that is the reward circle's decrease of £ 1.000) ; yellow labels marked by the letter of the corresponding player, represent the altruistic moves (that is the "altruistic" player 's reward decrease of £. 300).

Motivations and collusion among agents in the evasion of indirect taxes: an experimental approach

Luigi Mittone
Since the seminal paper by Allingham and Sandmo (1972) the theoretical microeconomic approach to tax evasion has almost exclusively treated only personal income tax and, more recently and with fewer examples, profit taxes (e.g. Kreutzer and Lee, 1986; Lee, 1997, Yaniv, 1995, Panteghini, 2000). The evasion of indirect taxes, and more precisely of value added tax (VAT), is an almost unexplored topic for microeconomic theory, and the few papers that have explicitly treated it from a theoretical perspective (e.g. Marrelli, 1984) have done so within the production theory framework, i.e. once again as part of a problem of profit maximisation. None of these works analyse those interesting aspects of VAT evasion tied to the highly social nature of this kind of tax evasion. The social-psychological dimension of tax evasion is not a new topic, and it has been widely analysed from both the theoretical (e.g. Gordon, 1989) and the empirical-experimental perspectives (e.g. Webley, P.Robben, H., Elffers, H. and Hessing, D., 1991; Bosco, Mittone, 1997), but once again this literature refers only to income tax.
The most distinctive characteristic of the evasion of VAT is that it typically involves three actors - the seller, the buyer and the state - whereas in the evasion of income tax the interaction concerns only the taxpayer and the state. The interaction among these three agents may give rise to the following phenomena:

• the taxpayer, i.e. the buyer of a given good or service, can evade only if s/he is able to collude with the seller, who should behave as tax collector for the state.
• The collusion between the seller and the buyer is facilitated by the mutual advantage accruing to the two agents from the collusion. By colluding, in fact, both agents can reduce their fiscal burdens: the buyer does not pay the VAT and the seller can declare an income lower than the real one because s/he under-reports the amount of his/her business, and consequently must pay less profit tax.
• The seller can decide to confiscate the tax yield that she has collected from his/her buyers. The approach chosen here to analyse VAT evasion is an experimental one. The main advantage offered by the experimental approach is that it enables isolation of each of the aspects just described and empirical investigation into the individual roles played by these factors in influencing VAT evasion.

1. assuming that the sellers operate in some form of imperfect competition market (i.e. assuming that they can fix their selling price) what is the seller's optimal price-collusion-evasion strategy?
2. Which is the optimal collusion-evasion choice for the buyer?
3. Does the traditional tax evasion theory fit with the seller's decisional problem of keeping the indirect tax yield collected from his/her buyers?

The experimental design
The context modelled by the experiments discussed here is that of a market of an homogeneous good with the following features:

1. operating on the market are several sellers and buyers, each characterised by different reservation values. The reservation value for the buyers is depicted by a reservation price, while for the sellers the reservation values are represented by their total production costs;
2. neither the buyers nor the sellers can alter their reservation values;
3. each agent (seller and buyer) can close only one transaction (consisting of only one unit of the good) per each time period (round of the game);
4. the experiment is carried out using computers; the experimental subjects interact via a local net;
5. all relevant items of information are given only via the computer screen;
6. each subject receives a role at the beginning of the experiment - seller or buyer - which does not change throughout the entire experiment;
7. each subject receives an identification number at the beginning of the experiment so that the subjects' real identities are not known to each other;
8. each subject receives (via the computer screen) her/his "personal information" i.e. her/his production cost if s/he is a seller, or her/his reservation price if s/he is a buyer, this values change over time but the subjects are constantly informed;
9. the money reward for the experimental subjects is given by the difference between the actual value of the transaction and its cost of production, or its reservation price, minus the indirect tax;
10. both the sellers and the buyers can make public offers for the good at the price that they believe most advantageous, obviously the sellers ask a price to sell while the buyers offer a price to buy the good;
11. both the sellers and the buyers can choose to close a contract from the list of offers shown on the computer screen by clicking with the left button of the mouse on the identification number of the agent that has offered the price that they believe good for them,
12. the sellers as well as the buyers can try to collude with a potential partner by clicking on a special button on the screen called "collusion"; when this button is clicked two buttons appear on the screen : "yes" and "no"; a subject who receives a proposal for collusion can accept by clicking on the yes button or can refuse by clicking on the no button;
13. collusion is always total, i.e. it regards the entire amount of tax due to the state, and it is a private relationship, so that the other players cannot know if a given seller (or buyer) has already agreed to collude with someone else;
14. during the experiment a given number of transactions are monitored by the fiscal authority, and if the subjects have colluded they must pay a fine that will be deducted from their final rewards;
15. the expected value from collusion (i.e. the values of audit probability and of the fine) is the same for both the sellers and the buyers;
16. in correspondence to the equilibrium point the lottery is fair, i.e. the expected value from evasion is equal to the sure choice value;
17. the subjects are informed about the fiscal audit probability and the fine to pay;
18. if the sellers are allowed to expropriate the VAT collected, a special window opens on the computer screen: the "pay tax yield to the state" window; when the subjects decide to expropriate VAT, they must write only the amount of money that they have decided to pay to the government in the window.

top

Department of Economics and Management via Inama, 5 I-38100 Trento tel. +39 461 282100