KTKLOS, Vol. 40 -1987 -Fasc. 4,496-514

On Thoughtless Rationality (Rules-of -Thumb)

AMITAI ETZIO~I*

I.. THE ROLE OF RULES

In strenuous efforts to shore up the beleaguered neoclassic~l paradigm,its followers argue that individuals may render rational decisions with-out processing information or deliberations, by using 'rules of thumb'.

'Cognitive capacity is a scarce resource like any other... To gather the information anddo the calculations implicit in naive descriptions of the rational choice model wouldconsume more time and energy than anyone has... Anyone who tried to makefully-informed, rational choices would make only a handful of decisions each week,leaving hundreds of important matters unattended. With this difficulty in mind, mostof us rely on habits and rules of thumb for routine decisions' [FRANK, 1987, pp. 3-4].

These rules are provided to individuals by their culture, organiza-tions, or are products of their previous experience. The 'discovery' of therules of thumb is tied to admission of information costs into the neoclas-sical paradigm. Early neoclassical economic models assumed that allindividuals have perfect information, instantaneously and freely. Onceinformation costs were included, it became evident that many optimalcalculations are too costly to complete. As BAUMOL and QUANDT [1964,p. 23] put it:'the more refined the decision making process, the more expensive it is likely to be, andtherefore, especially where a decision is not of crucial importance, no more than anapproximate solution may be justified. Since all real decisions are made underconditions of imperfect information, calculation down to the last decimal place ispointless in any event. One can easily formulate the appropriate (though not veryhelpful) marginal condition for what one may call an optimally imperfect decision.which requires that the marginal cost of additional information gathering or morerefined calculation be equal to its marginal (expected) gross yield.'

* Professor at the George Washington University, Washington.

496

ON THOUGHTLESS RATIONALITY

HAYAKAWA and VENIERIS [1977, p. 601], who otherwise depart from theneoclass'ical paradigm, and who recognize the role of social groups inshaping preferences and in processing information about choices, simplyassume that the resulting rules (or heuristics) are 'a useful and orderly way

of defining range of options involved in a variety of choice problems'.The rules themselves are said to be rational because they reflect an

accumulative experience or are the beneficiaries of evolutionary devel-opments: those who follow non-rational rules are driven out by thosewho adhere to the rational ones. Hence, non-rational rules will be swept

aside by the rational ones.The following discussion focuses on the rules of thumb (and other

devices for which the same neoclassic arguments have been advanced) toshow that (a) the empirical evidence about the rationality of these rules isdubious, and that (b) they logically cannot serve as a basis for rationalconduct. He who lives by rules of thumb may be somewhat less orsomewhat more non-rational than those not so guided, but not veryrational by the criteria applied here and generally accepted.

Before turning to an evaluation of the thesis of rationality one stepremoved, or macro-rationality, it must be noted that this position hasbeen advanced for a large variety of collectively provided decisionguides. Some neoclassicists refer literally to rules of thumb, common inthe informal culture, as exemplified in such popular sayings as 'lookbefore you leap'. Others refer to formal rules, for example those insti-tuted in corporate and other organization policies. Still others focus oncognitive patterns such as heuristics or 'routines', for example the use ofabbreviations as memory aids. The same point is also applied to societalinstitutions, value systems or norms of conduct ranging from those thatdefine one's table manners to those that guide behavior at a funeral. Allthese rules 'inform' the individual how best to behave, without anynecessary calculations or forethought [MARCH and OLSEN, 1984; HEINER,1983, p. 573]. In the following discussion the term 'rules' is used to refer

to all these phenomena.

II. AN EMPIRICAL TEST: PRICE SElTING IN FIRMS

The most detailed empirical testing of the rationality of decisions byprovided rules has taken place with regard to price setting in firms. It has

497

AMITAI ETZIONI

been often observed that business executives do not set prices at theplace a simple profit maximization theory states they ought to, wheremarginal costs equal marginal revenue. Instead, prices are set where firmpolicies tell executives to set prices. But do these rules (corporatepolicies), guide executives to the same maximal point?

One of the earliest and most influential studies on the use of rules isthat of HALL and HITCH [1936]. They surveyed 38 British firms and foundthat in 30 firms the executives said they used full cost pricing rules. Fullcost pricing rules basically add a 'normal' or target profit margin (or, apercent return on invested capital) to an estimate of unit costs toestablish a product's price. When the executives were asked why theyused such rules they provided three responses: (1) to deal with theuncertainties in the estimates of demand functions; (2)to avoid chargingtoo high a price (and driving away customers) or too Iowa price (in anoligopolistic market); (3) for businesses that sell thousands of differentitems the marginal cost pricing calculations required would impose agreat burden on them.

MACHLUP [1952] takes issue with HALL and HITCH'S conclusion, sug-gesting that full cost pricing leads to profit maximization. Using HALLand HITCH'S data, MACHLUP argues that executives observed full costpricing only as long as it was consistent with profit maximization.Further, MACHLUP notes that looking closely at the answers the execu-tives provided shows that demand was in fact considered in pricing.Firstly, some businessmen admitted that they would change price inperiods of exceptionally high or low demand. Additionally, MACHLUPpoints out that the answers executives provided to the question of whythey used full cost pricing -fear of competitors and demand beingunresponsive to price -showed that they were taking into accountdemand elasticities 'which to the economist is equivalent to marginalrevenue considerations' [1952, p. 71]. Further, he points out that respond-ents were using demand signals, though the term elasticity might nothave been used. MACHLUP essentially bases his response to HALL andHITCH on the idea that they had expected executives to act like econ-omists, drawing marginal revenue and marginal cost curves andspeaking of elasticity in precise numbers. When this was not the case,HALL and HITCH jumped to the conclusion that profit maximization wasnot taking place. In effect, MACHLUP says, the rules used proved to beguidelines to maximization in all but the terminology used.

498

ON, THOUGHTLESS RATIONALITY

In their influential study, CYERT and MARCH [1963] in effect sup-ported HALL and .HITCH. They were able to predict prices at a largedepartment store with great accuracy by discerning three rules forpricing in different situations: normal pricing, sales, and markdown.While the periods in which normal prices and sales prevail take place atregularly scheduled times, markdown is a contingency employed in thefailure or the perceived failure of these two other rules. For eachsituation a common procedure is used which determines price by theapplication of a predetermined, rule-based markup to a cost.

Using these rules CYERT and MARCH were able to predict prices in anormal pricing situation down to the penny 188 out of 197 times (a95 percent success rate). In regard to sales pricing, 'pricing is a directfunction of either the normal price (i.e., there is a standard sales reduc-tion in price) or the cost (i.e., there is a sales mark up rule)' [CYERT andMARCH, 1963, p.139]. They were able to predict the prices correctly downto the penny 56 of 58 times (a 96 percent success rate).

Markdown is a situatiQn employed when feedback indicates anunsatisfactory sales or inventory position. Even though markdown is acontingent situation, still rules were found to govern the ways it istreated. First, prices were not to be set below cost save as a last resort.Second, prices were to be reduced by one third and the result carrieddown to the nearest 85 cents. With these rules and few others CYERT andMARCH were able to correctly predict prices down to a penny in 140 ofl59markdown situations (an 88 percent success rate). It should be noted thatsuch a level of prediction is practically unknown in economics or inother social sciences, and this study should be regarded as very com-

pelling empirically.It would seem that pricing policies like the aforementioned only

take into account the cost side, not studying the demand side at all. It,hence, would seem that profit maximization is not taking place. How-ever, NICHOLSON [1978, p. 275] argues that despite the fact that the rulesappear to consider cost alone, demand is ultimately to be considered aswell:

'R.M. CYERT and J.G. MARCH spend considerable effort in analyzing the feedbackthat the market provides for the pricing of a product. Even though prices and profitmargins may initially be set without adequate reference to demand, the reaction of themarket provides information on the true demand situation, and prices are adjusted

accordingly.'

499

AMITAI ETZIONI

In short, NICHOLSON is trying to do to CYERT and MARCH what

MACHLUP did to HALL and HITCH: Use their own data to argue for therationality of rules. Other studies provide additional ammunition -to

both sides of the debate.In a study that examined a group of companies over nine years,

LANZILLOTTI [1958] found that companies followed various pricing rules.

These include pricing to achieve a target return on investment; stabiliza-tion of price and mark-up margin; pricing to realize a target marketshare; and pricing to prevent or meet competition. About one half of the

companies indicated that their pricing policies were based mainly uponthe objective of realizing a particular rate of return on investment in agiven year, over the long haul, or both. The question has been raisedwhether a firm is maximizing profits when it sets a target rate of return

or is it setting a satisfying goal it wishes to achieve? LANZILLOTTI pointsout that actual returns over a nine year period (1947-1955) were greaterthan the target returns, at which the prices were set. He casts consider-able doubt on whether the difference can be attributed to profit

maximization.Insofar as how the profit target was set LANZILLOTTI [1958, p.931]

notes:

'The most frequently mentioned rationalizations included: (a) fair or reasonablereturn, (b) the traditional industry concept of fair return in relation to risk factors,(c) desire to equal or better the corporation average return over a recent period,(d) what the company felt it could get as a long-run matter, and (e) use of a specificprofit target as a means of stabilizing industry prices.'

In regard to pricing for stabilization LANZILLOTTI [1958, pp. 931-932]develops the concept of corporate noblesse oblige which scarcely coin-cides with profit maximization but is more like the Marquis of Queens-

bury rules:

'The drive for stabilized prices by companies like U.S. Steel, Alcoa, InternationalHarvester, Johns Manville, du Pont and Union Carbide involves both expectation ofproper reward for duty done, i.e., proper prices, and a sense of noblesse oblige. Havingearned what is necessary during poor times to provide an adequate return, they willrefrain from upping the price as high as the traffic will bear in prosperity. Likewise, inpricing different items in the product line, there will be an effort (sustained inindividual cases by the pricing executive's conscience) to refrain from exploiting anyitem beyond the limit set by cost-plus.'

500

ON THOUGHTLESS RATIONALITY

However, KAHN [1959] argues thatLANZILLOTTI'S data can be used tostrong1y'support t~e thesis that even large corporations try to maximizeprofits.

KAHN [1959, p.671] notes that the widespread use of targetpricing,

'which is really an aspect of full cost pricing, provides their [LANZILwnl, et al.]principal evidence against profit maximization in general and the marginality descrip-tion thereof in particular. This misconstruction is in confusing procedures with"goals". Actually the target return seems above all to reflect what the executives thinkthe company can g~t; and as to the extent actual earnings diverge from the target it isbecause the market turns out to allow more or less' [KAHN, p. 671).

In a response to KAHN's criticism LANZILLOTTI [1959] states that theconcepts of target return and target market share seem far more helpfulin predicting and understanding the price behavior of large corporationsthan any notion of profit maximization. LANZILLOTTI suggests that targetrates of return are useful quantifiable measures corporations strive for,while profit maximization, given its ambiguity, is likely to be an ex postrationalization. LANZILLOTTI also asks '... whether by stating that thetarget return is really another name for profit maximization, KAHN hasexplained anything about the pricing policies of large corporations'[LANZILLOTTI, 1959, p. 682].

KATONA conducted one of the few studies of the use of rules fordecisions involving investment rather than setting of prices. KATONA[1975, p.323] notes that decisions on plant building and locale andacquisition of new equipment are enduring and nonreversible. They ingeneral involve large sums of money; hence, one would especiallyexpect rational behavior in these decisions. 'In fact, even casual observa-tion reveals numerous instances in which rules of thumb prevail andbusiness investment appears to be the habitual or even automatic resultof certain conditions rather than the outcome of careful deliberation.'One example is the 'follow the leader' principle: In various industriesQne firm is often looked upon as a leader whose decisions are imitated byother firms, whether or not it was justified in view of intra-industry'differences of 'first mover' advantages. Decisions to introduce newtechnology have a fadlike quality.

Without attempting to explore here further the ins and outs of thesestudies, their respective merits, let alone review still others (a taskundertaken by SCHERER [1980, pp.187 -189», it is clear that the same data

501

AMITAI ETZIONI

are interpreted by some as indicating that rules can be used to maximize,i.e., are fully rational, and by others to show that rules cannot so serve andin effect perpetuate poor practices. The obvious conclusion is that therules are not sufficiently specified to allow either their rational applicationor -study. Luckily, we shall see shortly, the empirical question need not beanswered in order to reach a conclusion on the merit of rules.

[II.

ORGANIZATIONAL CASE STUDIES

Organizations of all kinds from armies to churches, from public schoolsto prisons, promulgate large numbers of rules that direct and coordinatethe activities of their members. While these rules change over time, asindicated by the term 'bureaucracy' often applied to organizations, theiradjustment to changed realities is frequently slow and far from sufficientto ensure rational adaptation. An overview of the relevant literatureconcludes: 'Rules are broadly adaptive, but there is no guarantee thatthey reflect optimal, or even acceptable, solutions when new problemsarise' [STERN, 1984, p.lIO]. Adaptation of rules are also reported to belimited to the 'local' problem area and not organization-wide. And whenexecutives are faced with unfamiliar, new, complex situations, they tendto break them into components and deal with those elements that arefamiliar or look familiar.

PERROW'S [1981, p.17] study of the nuclear accident at Three MileIsland near Harrisburg, Penn., concluded that 'it is not feasible to train,design, or build in such a way to anticipate all eventualities in complexsystems where the parts are tightly coupled. They are incomprehensiblewhen they occur. That is why operators usually assume something else ishappening, something that they understand and act accordingly'.During the accident significant readings were not communicated to keypersonnel because operators did not believe them. Thus, rules (in thesense of 'routines') delayed proper response to the novel situation becausepeople misinterpreted the situation and did not recognize it as novel.PERROW [1981, p.25] notes: 'This is the significance of the widelyreported comment by the NRC commissioners that if only they had asimple, understandable thing like a pipe break they would know what todo.' Other studies show that organizational rules for collecting informa-tion tend to result in information over-load in which important items gounnoted. It is a problem common to most intelligence services. This is one

502

ON THOUGHTLESS RATIONALITY

main reason the warning about imminent attack on Pearl Harbor was'lost' on 'its way to Washington [WOHLSTETTER, 1962]. When new auto-matic gates closing training crossings were introduced in Britain, a majortrain disaster, known as the Hixon disaster, followed. It was establishedlater that slow moving vehicles could not clear the crossing ,in theallotted 24 seconds. This fact was duly noted in a long technical manualprovided to every policy station as the new gates were introduced butwas obviously 'lost' in the avalanche of information the manual con-tained [TURNER, 1976; see also NEUSTADT and FINEBERG, 1978].

NuTT [1984] studied 78 decision-making profiles of predominantlyservice or voluntary organizations. He finds that managers violate all therules advocated by academics for 'good' (rational) decision-making. Themanagers assume away uncertainty and treat causation and desiredresults as clear and specific thereby creating a false sense of security;NUTT reports, they have a predisposition to focus their rule search verynarrowly as they have a 'low tolerance for ambiguity and a high need forstructure' [1984, p. 446].

CYERT and MARCH [1963, p.121] note that the search process is basedinitially on two simple rules, hardly rational ones: '(1) search in theneighborhood of the problem symptom and (2) search in the neighbor-hood of the current alternative. These two rules reflect different dimen-sions of the basic causal notions that a cause will be found "near" itseffect and that a new solution will be found "near" an old one'.

An example of how such a limited search leads to a nonrationalresponse is provided by HALL [1976, p. 201] in his study on the decline ofthe old Saturday Evening Post. HALL points out that when the managersof the Post faced rising production costs due to excessive expansion thatresulted in depressed profits, they responded by substantially increasingthe subscription rate. This proved to be the Posts undoing for theirreadership growth leveled off and their profits were further depressed asa significant amount of funds had to be spent on promotion merely tohold readership steady. The Post thus staggered into the circulation warsof the 1950s in poor financial shape and never recovered. HALL argues,that this is a typical result of a narrow rule search strategy that leads oneto select an increasing of the subscription rate rather than attacking the'underlying causal structure of the problem (namely, the loss of controlof the annual volume and the consequent increase in production costs)'[HALL, 1976, p. 201].

503

AMITAI ETZIONI

Sometimes rules rather than change inadequately, fail to change atall. Historically developed rules continue on long after their usefulnessis lost and when more efficient alternatives are readily available. Forexample, despite the obvious fact that constant prices are the bestmeasure of economic activity, nominal prices are still very widely used infinancial reports, contracts and government studies [BARAN, LAKONISNKand OFER, 1980J. CYERT and MARCH [1963, p.138J report that frequentlythe mark-ups within an industry remain the same for 40 or 50 years.

Also, rules may be rational but may be implemented incorrectlybecause their guidance was insufficiently specified, promulgated incor-rectly, or inconsistently followed [SPROULL, 1981; BRITAN, 1979J. Forexample, managers often convey rules orally, and then maybe only inresponse to questions, which does not ensure all will know the rules

[SPROULL, 1981, p.119J.Finally, the study of rules as rational assumes there are clear goals, at

least -some set goals even if somewhat unclear. However, this is oftennot the case. A study of OMB under President Kennedy depicts thedifficulties the agency had in formulating specific budgets because thePresident failed to set the total outlay targets [MOWERY, et al., 1980J.Indeed, for many organizations, not just the situation is in flux but alsothe goals [SILLS, 1957J, rendering the use of most rules nonrational. In theUS Navy sailors were trained to step back whenever cannons were firedon deck long after the piece of artillery ceased to jump back when fired.

Many of the points made in defense of the use of rules as rational, and-the criticism -can also be made about the use of agents. For example,neoclassicists argue that individuals may not be able to read an insur-ance policy but they can pay a lawyer to read it for them and thus stillchoose rationally among policies. When it is argued that people cannotevaluate the service of agents rationally, it is suggested that a 'market'will drive out the agents that are serving the client vs. those that do not.This would be true if the individuals could evaluate the advice and ifthere was a competitive market among agents. Often, neither is true.Told, for example, that life insurance X is better than Y, how many timesdo people die and come back to evaluate their death benefits? The sameholds for many other such products from insurance for catastrophicillness or nursing homes to legal advice in the case of a divorce. A studyof patients' evaluation of their physician shows (a) they are unable toevaluate them rationally and (b) are selecting them on the basis of

504

ON THOUGHTLESS RATIONALITY

irrelevant cues (such as the opulence of their reception rooms). The bestpredictor of the agent to sell one insurance is not hi~ or her competitive-ness but how much the agent is able to evoke a sense of affability in theclient [CHESTNUT, 1977]. To further discuss the pros and cons of therationality of the use of agents, and the parallelism and divergences inthe

rules sub-theory and the agents sub-theory, will cause a majordigression and hence is avoided. Agents are mentioned only to add themto

a socio-economic research agenda charted here but not implemented.

IV. RULES CANNOT BE USED RATIONALLY

The empirical evaluation of the use of rules seems inconclusive. Despiteseveral sizable studies, conducted over decades, it is not possible to drawa firm conclusion whether or not rules enable their followers to actrationally, because, it seems the theory is insufficiently specified. As aresult, the same evidence is interpreted by some as supporting, and byothers as negating, the same hypothesis. While several case studiesprovide considerable plausibility to the thesis that the application ofrules often leads to non-rational behavior, such studies, by their verynature of being 'cases', are not fully compelling, especially to those whoare used to quantitative data. There is, though, a way to resolve the issuebecause one can show that rules in many situations cannot providerational guides. It is a matter of sheer logic, not of empirical evidence.

The discussion turns next to provide several reasons in support of theposition that the use of rules is a priori non-rational in many circum-stances.

1. Rules are typically advanced in isolation from one another ratherthan part of an overarching system. As a result they tend to ignore thefact that an efficient response to most challenges in the real, complexworld require taking into account multiple considerations.

At the same time, there is no reason to deny that in some instancessome simple rules prove very useful, at least until their use becomeswidely known. The IRS is using numerous guidance rules, collated intoClassification Handbook IRM 41(12)0. For example, the IRS pays specialattention to children claimed as exemptions by separated parentsbecause they are often claimed by both while only by one is proper.Hence, while one may expect, because of the a priori reasons provided

505

AMITAI ETZIONI

above, that rule guided behavior will often be quite non-rational; thisgeneralization cannot be applied to every instance. As to the question,under what condkions the application of rules is relatively morerational. This seems to be a matter about which relatively little is known.

2. When individuals are provided with more than one rule, thoserules often conflict with one another. For example, investors are told to'buy low and sell high' (an ambigous rule by itself, because no definitionof either 'high' or 'low' is included and hence the rule invites investors toproject their feelings on the situation). Sometimes 'high' or 'low' refers topricelearnings (PIE) ratios. However, what is a high ratio is not agreedand people are advised to invest in stocks of 'high-growth' corporationsdespite the fact that their PIE ratios are often infinite because theyinitially show no profit at all. Investors are also advised 'notto fight thetape'

and to 'let their profits run but cut their losses short'. These rulessuggest that if prices of stock move up, investors ought to expect them tomove still higher, while if the prices fall- and hence are relatively low-investors ought to bailout. These rules are incompatible with oneanother.

3. Rules are typically formulated as if they were universal truths,applying to all circumstances, to all times, and to all people. Threeexamples illustrate the fallacy of universality: Rule number 810 from theonly book exclusively dedicated to 'rules of thumb' [PARKER, 1983] reads:'Don't pay more than twice your average annual income for a house.' Itneglects to mention if gross or net income is at issue and the effects ofone's tax status. For example, given current tax laws it makes much moresense for a person with a high income but few other deductions to violatethe

rule than one with lower income or many deductions. Also, differenttax laws apply to those who are 55 years or older than to younger persons-for instance, the one-time exclusion on capital gain from the sale ofone's house.

Rule number 96 warns those who launch a political campaign that'a certain percentage of voters will be for or against you based purely onparty.affiliation.

Only a small percentage of voters are truly independentand will consider you without bias'. First, like the buy-low, sell-high rule,this

one contains vague terms such as 'certain percentages' and 'truly'.To the extent that its meaning is that political affiliation is much moreimportant than the person, it holds much less now than, say, twenty

years ago.

506

ON. THOUGHTLESS RATIONALITY

When the value of the dollar on foreign exchanges dropped 20 per-cent early in 1985 many economists feared that this would exacerbateinflation in the United States because a 10 percent drop was expected toadd 1.5 percent to 2 percent to inflation. But JAMES E. ANNABLE, chiefdomestic economist at the First National Bank of Chicago, stated thatthis will not be the case because 'rules of thumb that applied in the 1970sdon't apply in the 1980s' [New York Times, March 31, 1983]. He did notelaborate if the rules must be adapted every decade, or more often orhow one tells before the fact if they still apply.

Ample examples of the application of rules out of context is found inthe frequent use of neoclassical economics in policy making. Much ofneoclassical economics is formulated on the assumption that it takesplace in a fully competitive market. However, such markets are often notavailable, not even approximated. Yet rules formulated for full competi-tion are applied, often without even an attempt at adaptation, to thefundamentally different situation or context.

4. When rules conflict people will follow the rule that coincideswith their subjective estimates and own values. For instance, a less riskaverse person may follow rule number 102: 'Don't enter a poker gameunless you have forty times the betting limit in your pocket'. The morerisk averse may pick rule number 103: 'Don't enter a poker game unlessyou have sixty times the betting limit in your pocket' [PARKER, 1983,

p.17].Another pair of conflicting rules is discussed by KINDLEBERGER

[1985, p.l] in his essay on the social responsibility of corporations. Hewrites: 'The least of the issue is the clash between "When in Rome, do asthe Romans do", and "To thine own self be true"'. Many a corporationchooses between those not according to some cost benefit study but inline with its established culture.

5. Another way to see that rules that culture offers are not rational isto see that they are often not sufficiently specific to provide guidance.Barron's publishes a weekly polling the 'sentiments' of traders in Trea-sury bonds and bills, expressed in percentage 'bullish'. But it alwaysadds the note that 'high readings usually are signs of market tops, lowones, market bottoms'. Neither high nor low (or used) are defined.Hence, in effect, the weekly table provides a typical two-faced, highly ifnot completely, ambiguous rule. Ifbonds go up, the poll is 'validated'; ifnot -the note is. Together they provide at best a very vague guidance.

507

AMITAI ETZIONI,

'No price is too low for a bear or too high for a bull' notes Barron's[11-10-86, p.16]. Many rules are such.

Editing and writing are taught by rules but those who have taken sucha course can attest to how bewildering such rules often are. NORMANMAILER regularly violates most rules taught in courses on the writing ofproper English but is considered a very crafty author (reference is tostyle, not content). KENNER [1985, p.I] writes: 'The plain style has beenhard to talk about, except in circles.. .' SWIFT, KENNER reports, refers to itas 'proper words in proper places' but KENNER points out that SWIFTdoes not explain 'how to find the proper words or to identify the properplaces to put them into' [KENNER, 1985, p.I].

6. Psychologists have shown that people often evaluate the validityof messages using rules such as 'length implies strength', 'consensusimplies correctness', 'expert's statements can be trusted' [CHAIKEN andSTANGOR, 1987]. The knowledge that people are vulnerable in this way isused to fashion persuasive advertising. For example, actors who impers-onate doctors are used to 'recommend' over the counter drugs to buyerswho hardly need them. That is, rules people follow in their decision-making are used to deflect them from rational decision-making ratherthan to help.

V.

EXPERIENCE AND EVOLUTION

People are said to drop non-rational rules and retain rational ones astheir personal or collective (shared) and accumulative experienceteaches them which rules work. Actually the users of rules cannot judgeor evaluate them any better than the options they faced to begin with andwhich led them to use rules, for basically the same reasons they cannotmake a rational decision to begin with. (WINTER [1975J made the samepoint about information costs.) The fact that heuristics are systemati-cally biased and hence constitute a poor guide to decision-making is wellestablished. Evidence in support of non-learning is seen in that peoplekeep making the same numerous systematic mistakes in evaluatingprobabilities using the same biased heuristic. Over the course of alifetime their experience does not suffice to teach them to modify therules [TVESKY and KAHNEMAN, 1974, p.1130J. In effect SKINNER provides arather compelling model for how rules may be reinforced even whenthey are quite incompatible with the evidence.

508

O~ THOUGHTLESS RATIONALITY

Too much information to process, limited intellectual capabilities,and values and em,otions all curb the individual's ability to assess rules asit limits their ability to evaluate options. It might be said that there aremany fewer rules than options and hence they are easier to 'process'.However, the evaluation 'of most rules is much more difficult than that ofmost options. Compare, for example, the dilemma of a chess player whomust decide how to proceed in a particular game and situation vs.judging the rule, say, that one ought to develop one's forces (how long?)before one attacks. Compare an executive who must determine at whatlevel to set the price of a specific product in a specific market, vs. todetermine which pricing policy is the most effective. We saw already thedifficulties professional economists have in coping with the latter ques-tion after many years of study and arguments.

Collectively provided rules by society, carried in the culture, are saidto be rational. As TODA [1980] puts it, much learning takes place on the'species' level rather than that of the individual. One reason given is thatindividuals -who are rational- created these rules. 'Norms and Institu-tions were presumably created by men... ...and there is no a priorireason to suppose these men were any the less rational when they wereconstructing institutions than they were in their daily lives' [HEATH, 1976,p. 63]. For the viewpoint presented here, the same statement is to be readin the opposite way the author intended: People are sub-rational in theirdaily life; there is no a-priori reason to believe the rules they promulgatewill be rational. As CROSS [1983, p.17] writes: 'It is one thing to observethat choice making in practice seems to be governed by simple andinexpensive decision rules; it is quite another to show how those rulesthat are in use have come to be selected in preference to aU of the othersimple rules that could have come to be used instead.' Also, howeverrules are originally made, as we have seen, they tend to ossify and lagbehind changes in the environment.

Rules that organizations promulgate have been studied above. Butorganizations (and institutions) themselves are viewed as a mechanismthat limits the need to choose by providing a rational context for'decisions. Traditionally, organizations, especially firms, have beenviewed by sociologists as social units that evolve in part in ways theparticipants are unaware of or do not desire. Originally, an organizationmay have been selectively designed rationally as an efficient means toimplement its particular goal or set of goals. But, organizational studies

509

AMITAI ETZIONI

show, as rules get entrenched and vested interests in them evolve, poiiticsgrows, informal cultures arise, and the organization ceases to be an

efficient tool, if it ever was.Evidence cited above seems to amply support this view. However, in

recent years, neoclassicists have formulated a 'new economics of organi-zations' drawing on a COASE [1937] article and WILLIAMSON'S [1975, 1985]work. Accordingly, firms arise when it is efficient for them to arise; when

organizing production through the price mechanism (or market)becomes too costly. Most of the writing along these lines is highlydeductive, ('Firms must have arisen, because.. .'). Empirical evidence is

very scant.Some neoclassicists have extended the view of rules as efficient

guides from organizations to whole polities and even societies, and tothe march of history. NORTH [1981] for example argues that states are akinto firms in a market: They seek to maximize revenue growth subject. tovarious constraints (for example, technical knowledge). The rise and fallof states is next studied in terms of changes in constraints. For example,states will increase taxes up to the point the costs of the transactioninvolved become too high or citizens' allegiance too low. And, statesdevelop various mechanisms (for example, tax farming) that reflectsefficient adaptation to the constraints they face. (For further discussionand critical review of this approach, see MARCH and OLSEN [1984] and

DOUGLAS and WILDAVSKY [1982].)Finally, it has been argued that rules are rational due to evolutionary

selection. For example, HEINER [1983, p.586] writes: '... evolutionaryprocesses have long been interpreted as one of the key mechanisms

tending to produce optimizing behavior, or conversely, optimizingbehavior will predict the behavior patterns (or rules, AE) that willsurvive in an evolutionary process... [HEINER, 1983, p.560]. And'appropriately structured behavioral rules... will evolve to the extentthat selection processes quickly eliminate poorly administeredbehavior'. FIELD [1984] argues that there is a limited universe of possiblenorms (or rules). Members of social groups, organizations, or politieswork out together -using game theory strategies -which rule to shareand which are the most efficient ones. See also ULLMANN-MARGALIT

[1977].For the viewpoint advanced here, the notion of a state of nature, in

which there are individuals -fully formed and able to render decisions -

510

ON THOUGHTLESS RATIONALITY

before norms existed, is rather flawed, not only historically, but also as aheuristic~ .Individuals are not well formed unless they are members ofcollectivities, which in turn contain nom1s. While there is little to begained by pursuing the question who was first at the creation of thehuman realm, an I or a We, if the question must be broached, theanthropological literature leaves little doubt that collectivities, withinwhich there was little or no individual identity awareness and autono-mous action, preceded the I's. Collectivities can hardly be the product ofrational individuals who preceded them.

Moreover, the conditions for evolutionary selections of rules often donot exist. Among most organizations there often is no competitivemarket; good schools do not shove poor schools out of business; poorlyrun nursing homes are found next to well run ones and so on. Even theexistence of such competition among firms is historically rare and itsscope even in the West is much more limited than it is often acclaimed.And, the environment to which firms and other organizations mustadapt, is largely composed of other firms like themselves, not a setnature. We thus see followers of 'good' and 'bad' rules functioning rightnext to one another in the same environment and reliance on rules ishence no assurance that the rules are rational. Also, as SIMON argued,evolution leads to local or sub-optimal selection; there is no reason tosuppose evolution leads to optimal rules [SIMON, 1982, p. 53ft].

What holds for organizations, holds even more true for societies.There obviously is no market for societies in which the efficient onesdrive out the others. Societies with extremely inefficient institutions orregimes, such as the Soviet mega-bureaucracy, survive for decades. Veryfew societies ever collapse; instead, they change -not necessarily for thebetter!

-very gradually and slowly, and continue to be quite inefficienteven if they give up some of their most inefficient features, withoutacquiring worse ones. Clearly evolution is not working on this leveleither.

In short, the argument in favor of automatic or systemically providedrationality, one in which individuals may not think or deliberate yet actrationally, is poorly supported by facts and is logically untenable. Thefact that it is continuously to be advanced is itself an indicatiqn thatrules, at least in the form of theorems, survive for other reasons than thatthey are logical or empirically supported.

511

AMITAI ETZIONI

REFERENCES

BARAN, ARIE; LAKONISHOK, JOSEF and OFER, MARON R.: 'The Informational Contentof General Price Level Adjusted Earnings: Some Empirical Evidence', TheAccounting Review, Vol. 55 (1980), pp. 22-35.

Ba"on's, Nov. 10, 1986, p.16.

BAUMOL, WILLIAM J. and QUANDT, RICHARD E.: 'Rules of Thumb and OptimallyImperfect Decisions', American Economic Review, Vol. 54 (1964), pp.23-46.

BRITAN, GERALD M.: 'Evaluating a Federal Experiment in Bureaucratic Reform',Human Organinizations, Vol. 38 (1979), pp. 319-324.

CHAIKEN, SHELLY and STANGOR, CHARLES: 'Attitudes and Attitudinal Change', Toappear in the Annual Review of Psychology, Vol. 38 (1987).

CHESTNUT, R. W.: Information Acquisition in Life Insurance Policy Selection: Monitoringthe Impact of Product Beliefs, Affect Toward Agent, and External Memory, Unpub-lished doctoral dissertation, Psychology Department, Purdue University, 1977.

COASE, RONALD: 'The Nature of the Firm', Economica, Vol. N.S. 4(1937), pp. 386-405.CROSS, JOHN G.: A Theory of Adaptive Economic Behavior, Cambridge: Cambridge

Cambridge Univ. Press, 1983.CYERT, RICHARD M. and MARCH, JAMES G.: Behavioral Theory of the Firm, Englewood

Cliffs, New Jersey: Prentice-Hall, 1963.DouGLAs, MARY and WILDARSKY, AARON: Risk and Culture: An Essay on the Selection

of Technological Environmental Dangers, Berkeley: Univ. of California Press, 1982.FIELD, ALEXANDER JAMES: 'Microeconomics, Norms and Rationality', Economic

Development and Cultural Change, Vol. 32 (1984), pp.683-711.FRANK, ROBERT H.: 'Shrewdly Irrational', To be published in Sociology Forum, 1987.HALL, ROGER I.: 'A System Pathology of an Organization: The Rise and Fall of the Old

Saturday Evening Post', Administrative Science Quarterly, Vol.21 (1976), pp.185-211.

HALL, R.L. and HITCH, C.J.: 'Price Theory and Business Behavior', Oxford EconomicPapers, Vol. 2 (1939), pp.12-45.

HAYAKAWA, HIROAKI and VENIERIS, YIANNIS: 'Consumer Interdependence via Refer-ence Groups', Journal of Political Economy, Vol. 85 (1977), pp. 599--615.

JIEATH, ANTHONY: Rational Choice and Social Exchange: A Critique of ExchangeTheory, Cambridge: Cambridge Univ. Press, 1976.

HEINER, RONALD A.: 'The Origin of Predictable Behavior', American Economic Review,Vol. 73 (1983), pp. 560-595.

KAHN, ALFRED: 'Pricing Objectives in Large Companies: Comment', American Eco-nomic Review, Vol. 49 (1959), pp. 670--678.

KATONA, GEORGE: Psychological Economics. New York: Elsevier, 1975.KENNER, HUGH: 'The Politics of Plain Style', New York Times Book Review, Sept 13,

1985, pp.I, 39.KINDLEBERGER, CHARLES P.: Social Responsibility of the Multinational Corporation,

Remarks for a Panel at the Bentley College Conference in Business Ethics, Oct. II,1985.

512

ON THOUGHTLESS RATIONALITY

LANZILLOTTI, ROBERT F.: 'Pricing Objectives in Large Companies', American EconomicReview, Vol. 48 (1958), pp.921-940.

LANZILLOTTl, ROBERT-F.: 'Pricing Objectives in Large Companies: Reply', AmericanEconomic Review, Vol. 49 (1959), pp. 679-686.

MACHLUP, FRITZ: The Economics of Sellers' Competition, Baltimore: The Johns Hop-kins Press, 1952.

MARCH, JAMES G. and OLSEN, JOHAN P.: 'What Administrative Reorganization Tells UsAbout Governing', American Political Science Review, Vol. 77 (1984), pp. 281-296.

MOWERY, DAVID C.; KAMLET, MARK S. and CRECINE, JOHN P.: 'Presidential Manage-ment of Budgetary and Fiscal Policymaking, Political Science Quarterly, Vol. 95(1980), pp- 395-425.

NEUSTADT, R.E. and FINEBERG, H.V.: 'The Swine Flu Affair: Decision Making on aSlippery Disease', Washington, D.C.: US Dept. of Health, Education, and Wel-fare, 1978.

New York Times, March 31, 1983.NICHOLSON, WALTER: Microeconomic Theory, 2nd ed., Hinsdale, Illinois: Dryden

Press, 1978.NORTH, DoUGLAS C.: Structure and Change in Economic History, New York: W.W.

Norton & Co., 1981.NuTT, PAUL C.: 'Types of Organizational Decision Processes', Administrative Science

Quarterly, Vol. 29 (1984), pp. 414-450.PARKER, TOM: Rules of Thumb, Boston: Houghton/Mifflin, 1983.PERRow, CHARLES: 'Normal Accident at Three Mile Island', Society, Vol. 18 (1981),

pp.17-26.SCHERER, F. M.: Industrial Market Structure and Economic Performance, 2nd ed.,

Boston: Houghton/Mifflin, 1980.SILLS, DAVID L.: The Volunteers: Means and Ends in a National Organization, G1encoe,

IL: The Free Press, 1957.SIMON, HERBERT: The Sciences of the Artificial, 2nd ed., Cambridge, MA: M.I.T. Press,

1982.SPROULL, LEE S.: 'Managing Education Programs: A Microbehavioral Analysis',

Human Organization, Vol. 40 (1981), pp.113-122.STERN, PAUL C. and ARONSON, ELLIOT (eds.): Energy Use: The Human Dimension, New

York: W.H. Freeman and Co., 1984.TODA, MASANAO: 'Emotion in Decision-Making', Acta Psychologica, Vol.45 (1980),

pp.133-155.ThRNER, BARRY A.: 'The Organizational and Interorganizational Development of

Disasters', Administrative Science Quarterly, Vol. 21 (1976), pp. 378-397.'J\IERSKY, AMos and KAHNEMAN, DANIEL: 'Judgement Under Uncertainty: Heuristics

and Biases', Science, Vol. 184 (1974), pp.1124-1131.UL~MAN-MARGALIT, EDNA: The Emergence of Norms, Oxford: The Clarendon Press/

Oxford Univ. Press, 1977.WILLIAMSON, OLIVER E.: Economic Institutions of Capitalism, New York: Free Press,

1985.

513

AMITAI ETZIONI

WILLIAMSON, OLIVER E.: Markets and Hierarchies: Analysis and Antitrust Implications,New York: The Free Press, 1975.

WINTER, SIDNEY G.: 'Optimization and Evolution in the Theory of the Firm', in: DAY,RICHARD H. and GROVES, ThEODORE (eds.), Adaptive Economic Models, New York:Academic Press, 1975, pp. 73-118.

WOHLSTE1TER, ROBERTA: Pearl Harbor: Warning and Decision, Stanford, CA: StanfordUniv. Press, 1962.

SUMMARY

The recognition of the existence of imperfect information in decision-making and thecognitive limitations of the human mind put various assumptions of the neoclassicaleconomic paradigm in doubt. Neoclassical economists responded by arguing thatdecision rules, or rules of thumb, can be used to render decision without processinginformation. Further, neoclassicists have suggested that rational rules competeagainst and drive out irrational ones. This paper focuses on the use of rules of thumband posits that the empirical evidence about the rationality of these rules is dubiousand that they logically cannot serve as a basis for rational conduct.

ZUSAMMENFASSUNG

Die Erkenntnis, dass bei Entscheiden nicht vollstandige Infonnation vorhanden ist,sowie das Wissen um die kognitiven Beschrankungen des menschlichen Verstandeslassen an verschiedenen Voraussetzungen des neoklassischen Paradigmas zweifeln.Neoklassische 6konomen entgegnen, dass sich Entscheidungs- und Faustregeln auchohne vollstandige Infonnation anwenden lassen. Weiter gehen sie davon aus, dassrationale Regeln mit irrationalen Regeln konkurrieren und letztere verdrangen. 1mArtikel geht es um das Anwenden yon Faustregeln. Es wird gezeigt, dass die empiri-sche Evidenz fur die Rationalitat der Faustregeln zweifelhaft ist und dass diese Regelnnicht als Grundlage fur rationales Benehmen dienen konnen.

RESUME

La reconnaissance qu'il existe dans Ie processus de prendre une decision des infonna-tions imparfaites et des limites de l'esprit humain, a mis en doute plusieurs supposi-tions du schema economique neo-classique. Les economistes neo-classiques ontrepondu en soutenant que par regle$ de decision, ou procedes mecaniques, il estpossible, meme sans operer l'infonnation, d'arriver a une decision. En outre, ils ontsuggere que les regles rationnelles sont en competition ayec et repoussent les reglesirrationnelles. Cette etude se porte sur l'emploi des procedes mecaniques, et constateque l'evidence empirique de larationalitedeces regles estdouteuse et qu'ils nepeuventlogiquement pas servir comme base de conduite rationnelle.

514