Thursday, May 15, 2014

The Science of Demand (39) - Unofficial Translation of Steven Cheung's 经济解释 - 科学说需求


There are similarities among “other things”, “test conditions” and “constraints”, though there are significant differences in perspective. We might as well use the law of demand to further explain.

A downward-sloping demand curve from left to right (sloping downward from left to right is a law) restrains the relationship between price and quantity of a good, both of which are variables. “Other things” refers to all the variables other than these two, of which we allow some to change and disallow others. We have analyzed the choice between variables and invariables in the previous chapter. To be supplemented here is that certain variables bear no relationship to the good under analysis. We should ignore these irrelevant ones.

The term “test conditions”, seldom used in economics, is what I borrowed from scientific methodology in the study of logic. From the perspective of the law of demand, test conditions, only a minor subset of other things, refers to the conditions required to yield an implication that can be empirically tested. As discussed in Chapter I, an implication or a hypothesis, if not refuted by facts, can be said to have explained the facts, and can also be said to have predicted the occurrence of facts. However, such prediction requires certain conditions: according to the law of demand, for a logically-yielded hypothesis, under certain circumstances, the occurrence of A will lead to the occurrence of B. The circumstances described here are the test conditions.

Though the term “test conditions” is seldom used in economics – economists like to use “other things” or “constraints” – this is what I prefer. The perspectives of these three in evaluating the implication of a theory are different, yet I consider the perspective of test conditions the clearest among the three.

There is a critical test in empirical science. To explain facts or behavior, we can have many different hypotheses. Supposing there are two different hypotheses in front of you, you can test one of them first and then the other. Or you can ponder long and deep to formulate one or several test conditions, which when asserted, only one of the two hypotheses can be logically supported by facts or phenomena. That is, if test conditions are smartly chosen, testing can yield the following result: one hypothesis being right means the other hypothesis must be wrong. This, termed critical test, is the most fascinating and admirable in empirical science. To explain human behavior, instead of investigating into constraints, if a critical test is formulated from the perspective of test conditions, more may be achieved from less effort. An illustration will be provided In Section 4.

“Constraints”, referring to all the conditions that restrain behavior, are the most commonly used in economics. With regard to the law of demand, constraints include not only other relevant things, test conditions, but also price. In terms of testing a hypothesis or an implication, the perspective of constraints is not as acute as that of test conditions. But if the issue is magnified, the perspective of constraints is relatively superior. Alchian’s preference in starting from the constraint of property rights has had far-reaching impact on me; his credence that the constraint of property rights is identical to that of competition enlightened me all of a sudden. The prowess of Coase, on the other hand, is to generalize all constraints into costs.

There are not many principles or postulates in economics as a whole. If you follow the guidance of a master, you will realize how simple these principles are, and where their main points lie. The problem is, when these principles are applied to explain behavior, their level of difficulty will increase substantially. In general, their difficulty lies in three areas. Though earlier discussed, I will systematically repeat them here.

  1. Constraints in the world – restraining every individual to maximize self-interest – are very complicated. Constraints cannot be randomly assumed but must exist in the real world. We need to simplify, yet constraints after simplification must be essentially congruent with reality. On the other hand, since constraints are innumerable, whether they are relevant or irrelevant to a phenomenon has to be clearly differentiated – such “differentiation” cannot be arbitrarily done but has to be restrained by theory.

One example can illustrate the difficulty in investigating relevant constraints. By economic reasoning, when a government subsidizes education, adopting a voucher system is a commendable solution. Many people in fact believe that Hong Kong ought to implement such a system. However, the chance of implementing it is often considered slim. Why is that? To say that is due to opposition by pressure groups is certainly correct, but under what constraints can their opposition be so powerful? Naturally, in explaining why a voucher system is not adopted in Hong Kong, one demand curve plus constraints are sufficient. Demand curve is simple yet constraints are difficult – that’s where the difficulty lies.

  1. Testable implication – the occurrence of A will lead to the occurrence of B – the A and B here, or other relevant variables C, D, etc., must be observable in the real world. Quantity demanded in the law of demand is an intended variable that is non-factual. In other words, the law of demand itself is not testable. While applying the law of demand, we have to vary constraints to logically derive an implication that can be empirically tested. That is, we must derive certain implication so as to logically circumvent the perplexity of the abstractive quantity demanded. To achieve this, the assertion of test conditions poses a mounting challenge.

  1. We have already discussed the choice of change or non-change of other things (variables). If you assume certain other variables unchanged, how can you be sure they indeed do not change? You could do extensive research and then use statistics to control variables that are to change or not to change. Alternatively, you could ponder long and deep to formulate some test conditions. As soon as these conditions are confirmed to exist, other things or variables need not be our concern any more. Doing so is another challenging task.



No comments:

Post a Comment