Our purpose here is to formulate a new paradigm for economics, not discuss behavior in general, and we therefore must identify where money fits in the conceptual framework. Money appears in Intentional Action in several places:
- Money may be a constraint on the occurrence of a Stage, and therefore on engaging in a practice at all. Since practices are parts of other, larger, behaviors, this represents the common situation of having an insufficient amount of money to complete a small but necessary step in a larger behavior, preventing it from taking place For example, a new university graduate wants to take a job as a programmer, to engage in the larger behavior of deciding on a career, but lacks money to pay for gasoline to get to the job, which is 1000 miles away. Unless remedied, the “small” amount of money prevents the occurrence of the behavior. This is the issue of credit and capital, an issue currently of extreme micro- and macro-economic interest a number of national economies of the West. For impoverished people, this situation is a common fact of life.
- Money may be a constraint that determines which version of the behavior is engaged in. Consider, for example, a couple celebrating their anniversary. This may be done in a number of ways: going out to dinner, taking a cruise, taking an anniversary trip, buying gifts, etc. Which one is done is constrained by the money available to the couple. Further, the particular version of the behavior is similarly constrained: the cost of the particular restaurant, cruise, trip, and so on.
- Personal characteristics including values, traits, attitudes, and styles related to money affect the behaviors a person will choose and the versions of it, as illustrated by the anniversary celebration example. For example, a wealthy but extremely frugal person may go to a cheaper restaurant to celebrate than will his poorer cousin.
- The traditional “medium of exchange”: referring to the structure of the P parameter above, behaviors involve actual existing “things” (Individuals) serving as Elements in the performance of the behavior. If there is no version of a behavior for which the requisite Individuals for the Elements are present, the behavior cannot occur. In the behavior of baking a pie, for example, there are Elements including ingredients, pie, baker, oven, pie plate, and so on, and there must be actual Individuals to fill each role, or the behavior of baking a pie does not occur. In many purchasing behaviors, money is an Element in the specification of the P parameter, and there must be an actual amount of money to serve in that role, or the behavior cannot occur. The specification of the P parameter may involve a note, bond, or other promise to pay, in a particular amount. The specific instrument is then an Individual serving as an Element in the Performance, and as with the simpler case of the direct medium of exchange there must be such an Individual for the behavior to take place.
- The amount of money a person has – their wealth – is a Personal Characteristic of the person. As articulated here, the value of money to an individual is that it provides potential for behavior: a given amount of money affects the potential for engaging in some set of behaviors, and crudely speaking, the larger the amount, the larger the set of behaviors.
The logical relationship between money and behavior is well known. Kahneman and Tversky  state, “Any discussion of the utility function of money must leave room for the effect of special circumstances of preferences. For example, the utility function of an individual who needs $60,000 to purchase a house may reveal an exceptionally steep rise near the critical value,” acknowledging the direct connection between money and behavior. However, they continue in a way that reveals a profound and fundamental flaw in their analysis and more generally in the concept of value in their work and in the traditional paradigm of economics: “…the derived value (utility) function of an individual does not always reflect ‘pure’ attitudes to money, since it could be affected by additional consequences associated with specific amounts.” The flaw is that there is no such thing as the pure value of money, nor a “pure” attitude toward it. The value of money is always the value of a specific amount for the specific person in the specific circumstances of their life at that moment. Attempting to identify the “pure” value of money is analogous to attempting to identify the “pure” shape of a triangle: there is no such thing. There are specific values of specific angles and lengths of sides. At best, the attempt to define the “pure” value of money may be an attempt to find some value that is the average over a particular population with a specific range of circumstances, but this is no more a pure value than an average value of the angles of a set of triangles is a pure angle value.
Buying fire insurance and buying a lottery ticket have the Significance of protecting oneself from catastrophic harm and making possible a drastic improvement in one’s position in life, respectively. We noted that these are recognizable as behaviors – things people do – but we have not yet addressed two important related questions: 1) Recognizable by whom, and 2) What behaviors are available to be the values of the S parameter? How do we know, for example, that by buying a lottery ticket P is not engaging in giving to charity (a different larger practice)? These questions will be addressed in Section IV, Communities, below.
Full specification of an actual behavior requires specification of the eight parameters of Intentional Action, but this does not mean that only complete specifications are used or are correct. Many ordinary language descriptions of behavior are partial specifications, omitting one or more parameters. “P bought fire insurance” is such a description: the locution indentifies only I and A, but nonetheless identifies a recognizable behavior; on hearing the sentence, “Jane bought fire insurance”, we do not ask, “What did she do?”
Ordinary language does not, however, have standard linguistic “markers” for one kind of specification or another. It is having the fully specified parametric form that allows the identification of the kind of description being used, by comparison with the full one. It is sometimes useful to identify various kinds of behavior description such as the achievement description (specifying only A), performance description (specifying only P), and so on [Ossorio 2006]. “P bought fire insurance” and “P bought the security” are achievement descriptions; “buying the fire insurance” and “buying the security” are performance descriptions of the respective behaviors.
While a partial description such as an achievement description is not defective, it is incomplete, and as the case with an incomplete specification of anything, the incompleteness may be seriously misleading if the incomplete description becomes accepted as “what is really going on.” In such a case the aspect of the phenomenon the description refers to may come to be seen as the entire phenomenon. When this happens the partial phenomenon becomes the focus of study, and the omitted aspects become uncontrolled variables and “outside influences.” This is precisely what has happened in economics: the focus of study has become numerical measures applied to some, but not all, aspects of the A parameter. This situation has not been repaired by current behavioral economics: the form of behavior description used virtually universally is a partial achievement description of the form “P buys X at cost C,” thus identifying the achievement of having X but omitting the other aspects of the behavior, particularly including S, as illustrated by “P buys fire insurance” and “P buys a lottery ticket.” The standard presentation of a gamble and the analysis of it in terms of numerical utility is presenting and analyzing behaviors under this kind of description. We previously noted that buying fire insurance as well as lottery tickets appears to be puzzling due solely to having an incomplete specification of the phenomenon. The IA formulation articulates that incompleteness.
Intentional Action and gambles
Let us see how Principles 1 and 2 apply to the gamble, a device used routinely to analyze economic decisions. A gamble defines an experiment that is a choice between two or more options, each of which is a partially specified behavior. In the simplest form, each behavior specifies only the outcome and its probability; W, Kh, P, PC, and S are omitted. For example, the classic Allais paradox consists of the following two experiments:
Allais Experiment 1
|Gamble 1A||Gamble 1B|
Allais Experiment 2
|Gamble 2A||Gamble 2B|
Each choice defines a particular behavior. Behavior 1A, for example, is:
- I = P
- W: having $1M more money
- K: $1M vs. other amounts; winning vs. other things; certainty of getting $1M
- A: P has $1M more money
- Kh, P, PC, S: omitted
Or, using # to indicate “omitted” or “null value,”
- Behavior 1A = <P, W1A, K1A, #, #, A1A, #, #>
While it is straightforward to specify the behavior, i.e. the IA parameter values, for Gamble 1A, the other cases are not, due to an important subtlety of what the parameters represent. The IA parameters represent what the actor wants (W), what distinctions he is acting on (K), the significance (S) for the actor, and so forth. In particular, consider Gambles 2A and 2B. The experimenter presents the 2A as having an 11% probability of winning $1M and 2B as 10% of winning $5M, and therefore is distinguishing between 10% and 11%, but this does not mean that the actor acts on the distinction of 10% vs. 11%. Whether the actor makes the same distinctions as the experimenter presenting the gambles depends on the personal characteristics of the actor. Unless the subjects of the experiment have had specific learning histories, one would not expect them to distinguish between probabilities of 0.10 and 0.11. It is substantially more common to for actors to make distinctions such as, “very unlikely – unlikely – half-and-half – likely – very likely,” and in fact the empirical data on choices by subjects presented with these experiments is consistent with subjects not distinguishing between probabilities 0.10 and 0.11 in these behaviors. Historically, humans used concepts such as “probably” and “far-fetched but possible” for millennia before the development of mathematical probability theory, and because so many events in the real world require acting on such concepts it is reasonable to suppose that an ordinary person, such as an experimental subject in a behavioral economics experiment with no other training, is competent in their use. Imagine, for example, a person in any ordinary environment (social or natural) unable to act on the distinction between “probably” and “unlikely.”
This is a very common methodological error in behavioral economics: failing to verify that the distinctions the experimental subject is acting on when they answer a question are the same as those the experimenter is acting on in asking it. This is discussed in more detail in Jeffrey .
The Allais paradox will be further discussed in our next post, where we examine the concept of choice.