On ‘A behavioral theory of the firm’ by Richard Cyert and James
March
Hong Zhang
Seems that before this book, the prevailing theory about the firm is based on the two rational assumptions:
1.
Firms seek to maximize profits
2.
Firms operate with perfect knowledge.
The authors first raise questions to these two assumptions. These arguments are pretty close to our view of the firms today. First, there are sociological and psychological facts that are obviously omitted by the traditional approach about firms. In most regions today we still assume rationality of agent and maximization profit of a firm or utility of an investor, etc. And people continuously raise sociological and psychological evidences to repute the rational theories. We have seen almost the same story repeating again recently in finance. This is very interesting. In Physics the by far majority models are linear models (or via linear PDEs). And most of the remaining non-linear models are solved using a linear approximation. This doesn’t mean the world is linear. It simply means human being as a whole so far is better at calculating linear systems. I guess a similar story is there for the rational economic theory. By rationality we easily get linear systems which we can solve. There are conflicts between reality and people’s ability.
Now let me first follow the author’s major theory then give out some comments.
The goal of an organization
The authors first consider the goal of the firm. They notice that individual person has goal, but collectivities of people do not. Nevertheless, for a welldefined theory of organizational decision making, we do need a goal, or whatever analogous to the individual goal, for a firm.
The authors answer to this question is that the goals of a business firm are a series of more or less independent constraints imposed on the organization through a process of bargaining among potential coalition members and elaborated over time in response to short-run pressures. Three things are
important here. First, there is no consensus goal for a firm. There are only constrains for the participants. And the participants bargain with each other for these constraints. Second, the internal organizational process of control should go stabilized and elaborated. Third, the process of adjustment to experience should be altered by environmental changes.
The authors adopt two major organizing devices to analysis the goal (as well as Organization expectations and choices we will come into). The first set is for exhaustive variable categories, the second set for the relational concepts we will mention later. For example, as for the goal of the firm, the first set influences the dimensions of the goals, or what type of things are viewed as important by some or all participants. The second set of variables influences the aspiration level on any particular goal dimension. The second set is composed by past performance, past goal, and past performance of other comparable organizations.
The authors correctly state that conflicting goals exist for most organizations at most of the time. In class we have discussed this a lot. Rousseau had given out very impressive analysis on this point. What’s different in this book is that the author view goals as independent constraints. This is a rather interesting point of view. But to me it doesn’t solve the problem for the conflicts of the firm. Because we know that for a maximization problem we can either use constraints or we can use Langrange Multiplier to let these constraints enter the objective function directly. Therefore this is just to phrase the same problem in another way. This doesn’t give out a way to eliminate the conflict problem.
The Organizational expectations and Choices
To the authors, expectations are seen as the result of drawing inferences from available information. Therefore a good theory for the organizational expectations should consider variables that affect either the process of drawing inferences or the process by which information is made available to the organization. The latter process could also be viewed as search the information.
Of course, at the time of this book, game theory is not ready yet. So in the authors’ explanation, even though they realize the importance of the search and format of the information, they do not mention specifically the agency principal problem. Neither do they emphasize the importance of the
incentive in collecting and reporting true information. Instead, they look at the direction of search and the intensity and success of search. Thus they view the information problem as a flow. The key point for them is to first decide what type of information is important and second how to successfully get it. We have seen that this is still an important approach in a modern society. Just that they still view the firm as a sole identity, so the information problem is for the firm to collect the information. While in a modern contract theory, there is no single distinction for the firm and the outside. Each participant searches his interested information and makes decision based on his contract and his information. Nevertheless, the authors’ traditional view is still important for management decision making.
Based on the information and alternatives, the firm makes decision. The variables that affect choice are those that influence the definition of a problem within the organization, those that influence the standard decision rules, and those that affect the order of consideration of alternative.
Especially, the standard decision rules are affected primarily by the past experience of the organization and/or the past record of organizational slack.
We have seen that the authors have used very technical analysis. When the authors write down divisions like variables, they should have something like a mathematical function in their mind. (Which is partly revealed in the second half of the book where programs for the model are given in length.)
For example, they don’t give out a specific goal of the firm, but their views of the goal of the firm look pretty like a Langrange function (or any other function) of goals of all participants, with some cost functions representing the bargaining process and changing to environment. (I have already discussed that viewing the individual goals as constraints is just another way of stating the same mathematical problem.) Besides, the two sets of variables look like the variables enter the goal function ( the first set), and some rational expectation of the level of the goal function determined by the second set. So the second set is basically historical data and comparative role in the industry.
I guess these types of analysis are deep in the mind of modern economic researches. For example, when we study the stock return, we set up a one period model. In this model the returns might be driven by macro (or any other) variables. Then time series analysis is added to compare lagged return with contemperary return. And general says the returns of a single stock will be compared with that of the whole market, like CAMP.
But what is doubtful to me there is that this model of firm is still rational in terms of that it is the collection of rational individuals. When the author mention goals of participants, they might have a utility function in their mind. Because this is the only way that goals could be viewed as constraints. What’s behavioral here is just that after the aggregating process we won’t easily get an aggregated rational goal function. To me it is more like a chaotic system instead of a behavioral one. When we remember the simple fact that even one individual could have role conflicts, it is not that obvious that utility theory applies here directly. And once it is, as the authors assume, the theory of the firm has a solid rational foundation. If we look at the aggregating of rational preference in the utility theory, there is no sure that a representative rational preference exists. So if only goal of conflicts exists, it is not that behavioral.
Major Relational concepts
Now let’s look at the second set more carefully. The authors use 4 major relational concepts for their analysis: quasi resolution of conflict, uncertainty avoidance, problemistic search, and organizational learning.
Since it is inevitable for conflicting goals in any organization, it is natural to see whether it is still useful to study the aggregating behavior of the firm.
The authors suggest to solve out the conflicts ‘quasily’ by the following process. List all the goals as independent constraints, then assume local rationality to get some compromised goal. Attention should be given to the acceptable level of the decision rules and the sequential attentions to the goal. So here the author believes that the conflicts could be smoothed. But in their suggestion, the process is more or less problem driven. That is, in their setup when there are conflicts, they treat the conflicts as constraints to solve out the possible acceptable solution. As I have mentioned, this process doesn’t solve out the problem as a whole. It just solve the specific problem and doesn’t go to the big picture.
Using a modern word, the uncertainty avoidance is to state that the firm is risk averse. Of course, the risk has very general sense here. The authors suggest that the organizations avoid uncertainty in a way that short run, instead of long run uncertainty will get more attention. The authors assume a feedback-react decision procedures such that the firms solve emerging problems and wait for the next one. Besides, firms tend to find a way to
control the environment, rather than treat the environment as exogenous and suffer the uncertainty.
As for the search, the authors suggest that search is motivated, simpleminded and biased. Seems that the authors are pretty close to the cost of information. So it is a pity that they just accept the fact that search could be biased but don’t go one step further to see how to reduce the biasness. As shown in the previous discussion, I feel that the authors view all the problem dynamically but mechanically. Here the biased information in their view might just cost longer computing time for the true solution to come out. And bias-information problem is ignored for most of the time. Thus the author might believe that the bias of information has some definite distribution.
Still, even in this case, a Baysian distribution should replace the prior distribution. But these problems are ignored by the authors.
Organization learning means that organizations exhibit adaptive behavior over time. Especially, the firm adapts the goal, the attention rules, and the search rules. This is pretty close to the dynamic view of the firm from the contract theory. Still, the authors treat organization as individual.
These are the basic model built up in the book.
Now let me have a somehow modern and bigger view on this book.
As we have sees again and again, in this book the authors adopt a problem driven way of analysis. That is, they solve out problems and wait for another. For example, when there are conflicts, the authors let the firm to set these conflicts as constraints and solve out a possible solution. As another example, the firm makes decision on some given problems and waits for other problems to come. In my own view this could be a major weak point for this book. Even we need to adopt dynamic system to analyze problem, it is not necessary that we can only solve problem one by one. For example, we know that for a social contract, people use law to eliminate conflicts. (see my other paper) At a firm level, common knowledge like accounting system also helps to avoid conflicts. There is no doubt that maximization problems at some level help people to make decision. But on one hand the solution might not exist, on the other hand we really want to find out ways to use the maximization problem as few as possible, that is, to solve the problem as a whole.
Another discussion comes from the distinguishing between behavior and chaotic system. My view is that any economic theory, if assuming the rationality of the participants, should belong to rational theory, even some wired properties could come out as a result of aggregation. At the same time, in modern economic theory, there is very few studies about chaotic system, which is formed by rational individuals but results in strange aggregating properties. Nevertheless, chaotic system belongs to rational world, just that no prediction could be made. In this sense, I feel that many of the topics in this book belong not to behavioral area. Anyway, there is no ready definition and even not belonging to behavioral study this book is still interesting.