CHAPTER 4 THE DRIVE-EFFORT HYPOTHESIS It is time to state the Drive-Effort hypothesis more precisely. The payoff to an alternative may be defined as yes no (Wn+l - Wn) - (Wn+l - Wn) where "yes" and "no" mean respectively that the opportunity is or is not undertaken, n+l and n indicate the states before and after the alternative would be carried out, and W stands for wealth (including otherwise-expected lifetime income) divided by expected years of life. It is reasonable that Drive-Effort--or the expected strength of response to an opportunity, if one prefers to view the matter that way--depends upon (a) the payoff per unit of work time, and (b) the individual's or nation's wealth. The effect of expected payoff is obvious without examples; the effect of wealth becomes equally obvious with a few simple examples such as: (l) In a subsistence-agriculture economy, one does not build a house for oneself if one already has a house. (2) You do not prepare a meal if you have one already prepared. (3) Campers only work at setting up camp until it is set up. In these primitive examples, wealth in the widest sense obviously affects work supply. Hopefully, this observation is enough to establish the point, though the desire for money to buy goods seems less specific than the desire for specific goods, and its marginal utility does not so obviously decrease. It does not seem reasonable, however, that the simple difference between Wn+l and Wn, written (Wn+l - Wn), indicates the strength of the desire for money, or of Drive-Effort; a millionaire will not perform the same acts to get an additional dollar as will a pauper, on average; rather, it is reasonable to think that the difference is relative to the individual's wealth, assuming that the net money flow is positive and that the opportunity offers a payoff higher than the value of time in Becker's reckoning. Therefore, I propose the formulation Drive-Effort Measure = expected likelihood of positive response to a given opportunity by individual or nation ( Wn+l - Wn ) ( ) = a ( --------- ) ( Wn ) which takes the exponential form shown in Figure 4-l. The function has the property that a doubling of wealth leads to a halving of Drive-Effort, where wealth refers to the person's situation with and without the opportunity in question, and wealth therefore is a variable as a person does more and more work. That is, Drive-Effort is proportional to (Wn+l - Wn) at any given level of initial wealth, and it is inversely proportional to Wn, as may be seen in Figure 4-l, and at subsequent wealth positions; this will be made clearer in the monopoly-duopoly analysis in Chapter 6. But there is no particular relationship (other than monotonicity) between opportunity and Effort. ------------ Figure 4-1 ------------ If the effort actually _r_e_q_u_i_r_e_d for a given alternative is below the Drive-Effort Measure for that alternative at a given level of wealth, the opportunity will be accepted; if not, not. For example, assume that the level of effort required for a particular opportunity Z with Wn+l - Wn = 1 is shown by the horizontal line Z in Figure 4-l. At all wealth positions less than Z the alternative will be undertaken. This formulation has attractions in addition to its plausibility and simplicity. First, it corresponds to the classic Weber-Fechner Law in psychophysics, according to which the likelihood of response to a given difference in stimuli is proportional to the lower level of the stimuli. Second, it corresponds to the standard logarithmic conception (when viewed cumulatively) of (Benthamite) diminishing marginal utility of money.1 (The two functions are in fact one and the same, but viewed in different ***** ways. The present function may be seen as an operationalization of the Benthamite diminishing-marginal-utility function, whose relationship to the Weber-Fechner Law has long been recognized.) And of course this shape of function is consistent with the conventional assumption of concave indifference curves between pairs of goods. It should also be noted that there is some empirical support for a semi-logarithmic utility function, e.g., in Galanter's study2 that asked questions about the *** respondent's speculations about his/her own happiness in connection with various levels of wealth. The Drive-Effort Measure might be better represented with some other concave-downwards function. The power function is one such candidate (it has been suggested in some psychophysical experiments), and there may be others. But the particular function is not of crucial importance; only its downward-concavity matters, or even just its downward slope. The relationship of the Drive-Effort concept to the concept of diminishing marginal utility of money suggests that we look into the meaning of the latter for our purposes here. That concept suggests that a given increment of money is "worth" less when one is richer. But how does one measure an increment's "worth" except by evaluating what a person will trade for it? And the only exchange that might provide grounds for analysis would be a nonmarket good, because money has no economic meaning except as power to purchase goods. (Conceivably, one might measure the amounts that rich and poor persons would pay for food or water on a lifeboat, but this would be neither an easy test nor a very meaningful one, for a variety of reasons.) It seems reasonable, however, to measure the utility of money according to the time and effort that people will give up to get it, and the goods that it will purchase.3 If so, the utility and the effort *** become operationally indistinguishable, though some persons may wish to continue thinking of them as separate concepts. One may also measure the utility of money (really, the utility of goods) with such questions as how happy an additional sum would make one, or with the probability of suicide connected with various wealth (income) levels.4 But such measurements seem more relevant ****for making income redistributions than for predicting how various persons will behave in the face of various opportunities, which is our subject here (Simon, 1974). The relationship between wealth and effort implies a cyclical effect: lack of wealth leads to effort, effort leads to wealth, wealth leads to lack of effort, lack of effort leads to lack of wealth. This is similar to the cycle described by "overalls to overalls in three generations." But this cycle need not suggest a trap of economic stagnation, for at least two reasons. First, the advance of technology (itself a result of effort) enables successive generations to live better at a given level of effort. And second, it may be that there is a _r_e_l_a_t_i_v_e wealth effort which causes people at a given level of absolute wealth to exert more effort if the wealth of others is higher, ceteris paribus; this could be explained in absolute wealth terms, however, if higher observed wealth of others increases perceived opportunities, and hence induces more effort. The types of decisions discussed here could conceivably be crammed within the framework of the pure logic of choice. But if that were to be done, there would be no gain in understanding the particular forces that affect the choices, which is vital for economic understanding. SPECIFICATION AND MEASUREMENT OF KEY CONCEPTS If this theory is to be useful, one must be able to measure its key variables. Despite the earlier discussion of the possible direct measurement of utility, the Drive-Effort concept never needs to be measured in practice; the discussion of utility measurement was for expository purposes only. The measurement of wealth is straightforward though it may not be easy. The individual's (or family's. or group's) assets, as valued in market prices, constitute an appropriate measure in most cases. In some other cases, the person's expected stream of earnings may be the appropriate concept, and the stream may often be estimated with characteristics such as education that are related to income, as human capital theory teaches. In some cases an appropriate measure of wealth may not be obvious, and will require hard thought and serious justification. But in principle, the measurement of wealth presents no problem. (The measurement of relative wealth is discussed in the section on income distribution in Chapter 000). The actual measurement of opportunity is likely to be difficult in many specific non-trivial cases. Some cases are straightforward. The size of the opportunity clearly is changed by a) increasing the payment offered for an evening's overtime work; b) increasing the reward offered to a diver who hesitates to dive to a dangerous depth to carry out a task; c) offering a reward to a Delhi pedicab driver if he gets you to the hotel before a specified hour (though I confess I could never bring myself to ride in a pedicab even for experimental purposes). The income effect causes difficulty in measuring a change in opportunity and its effects. Moving a physician from the Philippines to the United States assuredly increases the physicians's opportunity, but the income effect may reduce the amount and intensity of work exerted by the physician. The effect of a price reduction, or of the equivalent of a price reduction in the guise of the onset of availability of a product to a group or individual, raises perhaps the most complex issues of specification and measurement. Such a phenomenon has offsetting impacts through the channels of changes in both wealth and opportunity. For example, if the tax on automobiles in the Soviet Union, were to be removed and the price were to be consequently lowered, people would suddenly be more wealthy because their financial assets would now enable them to own a good that was previously unobtainable by them, thereby leading to a reduction in Drive-Effort. On the other hand, the lower price would represent an increase in opportunity, because additional earnings would represent the increased opportunity to own an auto, thereby inducing an increase in DrEf. Another example is the increased availability of consumer goods in the early stage of economic development. The net effect, and hence the outcome of the two impacts, would seem to be indeterminate, though perhaps further analysis of lifetime income (including present wealth) would produce an unambiguous predicted effect. The form in which the Drive-Effort hypothesis is currently stated will lack esthetic appeal for economists, because it is not now framed in terms of utility maximization. But there is no reason to doubt that further work can forge that connection with microeconomic theory, just as has occurred with such ideas as mean-variance analysis that were originally stated in a form uncorrelated with other theories. D/l36A,#3 85-36 Effort 4 1-15-86 FOOTNOTES 1The concept of utility referred to in this essay harks back to the original Benthamite concept. In the writing of Bentham and his followers, "utility" referred to ex post happiness and pleasure, and was frankly psychological. Economists gradually purged the term of its pleasure-pain psychological content, culminating in the work of Slutsky (l95l) and von Neumann-Morgenstern (l944) in which utility became an unobservable ex ante magnitude that was theoretically useful to postulate as a foundation for a consistent, positive choice theory. This book returns to the original Benthamic meaning of the term. Much confusion also has arisen because of the shift over the years in the use of the concept of utility, from the theory of public policy to the theory of choice. When the utility concept was introduced by Bentham, the subject of discussion was legislation and governance. The "moral arithmetic" of utility was intended to promote wise laws and human social structure. But economists have come to use "utility theory to _e_x_p_l_a_i_n economic behavior (particularly demand theory) and only secondarily (when at all) to amend or justify economic policy" (Stigler, 1965, p. 67). The distinction between the two uses has frequently been blurred or overlooked, which has led to great confusion. This footnote is drawn from my l974 article. Nothing said here implies that I advocate a utilitarian framework for evaluating social policies. 2Described by Stevens (l959, pp. 54-5). 3As Marshall put it: If the desires to secure either of two pleasures will induce people in similar circumstances each to do just an hour's extra work, or will induce men in the same rank of life and with the same means each to pay a shilling for it; we then may say that those pleasures are equal for our purposes, because the desires for them are equally strong incentives to action for per- sons under similar conditions. (l920, p. l6)This suggests that whatever two persons in _d_i_f_f_e_r_e_n_t circumstances will each sacrifice a _g_i_v_e_n amount of time to attain has the same value for the two of them. 4For a review of such material, see: Simon (1974).