On the Observability Requirement in Economics as an Axiomatic Science

1 Biological versus Physical Predictions

Man is a biological creature. To say that economics cannot predict the behavior of man with accuracy is no more than saying that while a biologist predicts butterflies will appear in the summer, a few may show up in the winter. Of course, the biologist may exercise control over an environment so that some butterflies do show up in the winter, but the economist cannot afford such a control for man. During the 20th century many attempts were made by some governments to control the behavior of man. The people’s commune in China is just one example. Economics can explain, however, why the governments did so and predict what would be the consequences.

Predicting an event before it occurs and explaining an event after its occurrence are, in principle, the same in economics because it is an axiomatic science. This is the same as physics using the same principle to predict an apple will fall to the ground if disconnected from the branch of a tree or explain why the apple has fallen from a tree.

In economics, predictions are made in terms of identifying a specific change in constraints that will lead to a specific change in human behavior. Admittedly, it is often easier to trace the change in constraints after an observed change in human behavior than noticing a change in constraints before we predict a change in human behavior. This is the same as saying that in physics it is easier to trace the reason of a fallen apple than to predict when the apple will fall.

2 The Disappearance of a $100 Bill and the Law of Demand

Can economics really predict human behavior before the behavior occurs? When I joined the University of Washington in the fall of 1969, in my first lecture in graduate price theory, I offered an example which has become classic in the profession. I held a coin in my hand, and asked the students: “Anyone wants to bet if I let the coin loose it will fall to the ground?” No student responded. “How about I offer $1000 for your $1.00 that this coin will drop if I let go?” Again, no student responded.

I then pulled out a $100 bill from my wallet, and asked, “If I leave this $100 bill on the ground of a pedestrian sidewalk with no wind blowing, it will disappear. Anyone wants to bet?” No response. I raised the odds as I did with the dropping of the coin, and again no student responded.

I then told the students: of all the sciences known in the history of mankind, whether physics or chemistry, or biology, sociology, history, anthropology, or psychology, none of them can predict the disappearance of the $100 bill. But economics can, and it can also predict under what conditions the $100 bill will disappear sooner or later.

A student may ask: “Why, you do not need economics to predict the disappearance of the $100 bill. Any child can make that prediction! What is the use of economics after all?” I would reply, “If you ask five children they may give you five different answers as to why the $100 bill will disappear, but if you ask five different economists they will give you the same answer, that is, the law of demand.” It saddens me that as an old man now I often hear economics departments everywhere are no longer paying much attention to this law anymore.

In my view, the law of demand is the only law that economics cannot do without. During my graduate student days, 60 years ago, I went through the sequence of five weeks of Baldwin on the of law of demand, five weeks of Hirshleifer on the law of demand, and one whole semester of Alchian on the law of demand.

A reader may think using the law of demand to predict the disappearance of the $100 bill is trivial, but as will be shown later, I used this same law and predicted China would go capitalist in 1980, in no uncertain terms. It was because of the objections of Gary Becker, Doug North, Ted Schultz, and Milton Friedman—all Nobel Prize winners—that its publication was delayed until January, 1982.^[1]

What saved the fate of this pamphlet predicting China was Yoram Barzel, then a colleague at the University of Washington. Yoram also disagreed with my prediction that China would go capitalist, but then he also said the theoretical section was beautiful, brilliant, no flaw, and it would be a great waste if the manuscript did not go into print.

3 The Failures of Two Giants in Predicting Events

Paul Samuelson, a genius in 20th century economics, had the habit of making predictions of worldly events. When his life came to an end in 2009, some of his surviving colleagues pointed out, of the worldly events he predicted he achieved a perfect score of exactly zero.

My close friend Milton Friedman, another genius contemporary with Paul, made numerous predictions on monetary matters in his life time, had fared better. But when I took Milton on a tour to China in 1988, I told him that with the advance of digital technology, the day would come when we could no longer measure the quantity of money. He did not believe me. When in 2013 I told Robert Mundell the same, he asked, “When did you know that?” I replied, “In 1988, I told Milton so.”

When Milton passed away in 2006, among the outpouring of eulogies some of his colleagues noted that he had been consistently wrong in his predictions of monetary effects since 1982.

The above is not meant to belittle Paul and Milton: I have learned so much from them so I could develop my own. The point I want to make here is that given the world is such a complicated place, the constraints subject to which human decisions are made are often too costly to investigate and specify. After nearly 60 years of relentless research I have come to the view that failure in interpreting or predicting economic events rarely stems from the failure of theory proper—which is not profound—but from the failure of our ability to investigate and specify the relevant constraints involved.

4 Predictions Involving Multiple Variables

Let us return to where we began, the proposition of “if A then B” and the testing procedure of “not B implies not A.” Two issues crop up. First, when we perform the testing the variables A and B must be observable. What are we going to do if one of the variables is not observable, or simply no such thing in the real world? This is interesting because in economics there is one non-observable variable which we must keep, namely, “quantity demanded.”

A second issue that crops up is when a proposition involves not only A and B but a multiple of variables. The modern treatment of such a situation is, of course, the use of multiple regression, a technique no doubt prompted by the advancement of the computer. However, Dale Jorgenson, one of my heroes in 20th century economics, told me in 1976 his view that the use of multiple regression to handle data has the weakness that it is not robust.

In 1978 I had the opportunity to perform a massive econometric study, and reached the view that Jorgenson was right. This is a consulting job I did for a petroleum company in California. Several noted economists were also involved, but I was in charge doing the investigation in details. In this effort I produced two lengthy reports which the petroleum company called the Bible, and Armen Alchian said they were the best empirical works in economics he ever saw.

One of the reports, entitled “The Characteristics of Crude Oil as Determinants of Posted Prices,” involved the use of posted prices and crude-oil assays data from two independent laboratories. The data were plentiful, totally objective, and my knowledge of the petroleum industry was taught by experts in that industry. I also had the support of half a garage full of documents. Yoram Barzel was by my side whenever I needed help. Yet during this massive research project numerous questions cropped up on the reliability of the multiple regression technique.

It is one matter to produce a paper for publication; it is another matter if the researcher wants to know the truth. What are you going to do if you want to publish a paper, when one regression result supports your argument while another rejects it? Yes, econometrics studies are inclined to commit the blunder of what Alfred Marshall called post hoc ergo propter hoc.

During my long research career, in the formulation of testable hypotheses I like to use two or very few observable variables for each test, while specifying and holding other constraints constant, and then repeat similar tests using the same approach but a different set of variables. Multiple tests using a few observable variables for each test is, in my view, a far more enlightening method than the use of multiple regressions. My predictions or interpretations of observations have earned such high marks over the years that there is no doubt about the predictive power of economics as an axiomatic science.

5 The Observability Requirement and My Disagreement with Gary Becker

I entered UCLA in 1959 and received my Master’s degree in 1962. The only reason I proceeded for a Ph.D. was because Armen Alchian was visiting Stanford in 1962, and I had to wait for his return to audit his lectures.

Armen was a legendary figure at UCLA even at that early time. It was much later that I realized that his giant status stemmed from a paper he published in 1950, on evolution and economic theory, which ignited the methodology debate in economics for nearly 20 years.^[2] In my view, this was the best paper Armen had ever written. He should have won the Nobel Prize with that piece alone.

I had already gone through the theory courses taught by Robert Baldwin, Jack Hirshleifer, and Karl Brunner, before I turned to audit Armen in the fall of 1963. Armen spent the first five weeks lecturing on the meaning of utility measurement.

Then in the spring of 1964 Gary Becker showed up for a semester at UCLA. Gary was already a superstar in the profession at that time, and in retrospect it was his discussion with Armen that prompted him to use the utility function in literally all of his subsequent works. In terms of analytic ability, Gary was the best I had ever known in the profession, but I did not appreciate the use of “utility” in economics from day one, and by the time I wrote my thesis I adopted the view that economics would be in better shape had Jeremy Bentham never existed.

The view of Armen and Gary was that the concept of utility must be kept because economic goods such as friendship and love cannot be measured in pecuniary terms and therefore cannot be traded in the market. When working on my thesis in 1966, I assumed the view that although non-pecuniary goods cannot be traded, they may nevertheless substitute for pecuniary goods, hence in economics the concept of utility we can do without.

For example, in the fall of 1988, I took Rose and Milton Friedman on a tour to China. In the City of Suzhou, a group of high-ranking comrades entertained us in a lavish dinner. They brought along a translator. Delicious dishes were served on the table while Milton was arguing non-stop with the comrades. I saw the food was getting cold, mumbled a few kind words in Chinese, implying they should argue no more. Milton did not understand my Chinese, but he guessed what I said. He looked at me, and said, “Steve, every man’s soul is for sale; don’t sell yours cheap.”

Yes, my soul was for sale, and its price dropped like a rock when I was hungry and good food was served on the table. This is a substitution between pecuniary and non-pecuniary goods.

Armen’s view is that in order for a utility theory to be operative, the requirement is that we must be able to identify what entity has utility and the cost of acquiring that entity. My view, however, is that since any non-pecuniary good may be substituted by a pecuniary good, at least at the margin, the use of a non-observable entity called “utility” is unnecessary and confusing.

I must continue pressing the issue because it is important. In 1972 I published an article on the marriage contracts in the traditional practice of China, involving issues such as blind marriage, foot binding, concubine, infanticide, bride price, and so on, using wealth maximization subject to constraints as a foundation.^[3] A year or so later Becker published his paper on marriage, using his utility function.^[4]

In terms of popularity contest there was no match—Gary beat me by the miles. English readers were not interested in Chinese affairs at that time. Worse, both Ronald Coase and Harry Johnson asked me not to publish that paper because it exposed some ugly aspects of the Chinese culture. There were others, of course, who urged me to let the paper go into print. The editor of the Economic Journal accepted the paper promptly, but he asked me to shorten the piece. So for sheer convenience I took out the last section, where I explained why I used wealth maximization and abandoned utility maximization. It is interesting to note that although my paper on the Chinese marriage has had only several citations a year, it is still being cited after 50 years with no sign of subsiding.

Jimmy Buchannan and Gordon Tullock, who had read the original manuscript, both wrote and complained that I should not have taken out what they considered as the best section of that paper. So they too, did not like the utility function. My “marriage” paper now in print has a few sentences left saying that utility maximization has the pitfall of presenting a tautology in disguise.

Let me turn to another paper of mine which I like and Becker did not.

It was my paper entitled “Why are Better Seats Underpriced?” published in 1977.^[5] This paper was written to honor Alchian at a conference held in Miami in 1976. The paper argues that in movie theatres or concert halls, where different quality seats are graded and priced separately, the higher-priced better seats are “underpriced” in that they are usually sold out first, for the purpose of preventing seat-jumping.

Before that conference, I sent draft copies of the paper to my old colleagues at Chicago, including of course Becker. Then before Miami I met Becker in an elevator at UCLA. He said to me, “We at Chicago read your paper on theatre ticket pricing and we all think it is wrong.” I asked, “Do you think the paper is interesting?” He replied, “Yes, very interesting.” I asked again, “I tested the hypothesis, did I?” He replied,” Yes, you did.” Then I said, “What more do you guys want?” He said, “We just all think it is wrong.”

A month later I attended the Miami conference in honor of Armen and presented the ticket-pricing paper. Gary was not there. During coffee break I alone was drinking coffee at a corner. Bill Meckling, one of Alchian’s closest friends, joined me. He said, “Steve, do not ever change your choice of subject matter and your style of writing. Your works are always so interesting that only Armen could produce a student like you.” I am sure had I used utility theory to explain ticket pricing, Meckling would not have said that.

6 Predicting China

In the summer of 1979 I received a letter from Arthur Sheldon of the Institute of Economic Affairs in London, saying Margret Thatcher’s office wanted an answer to the question: Will China go capitalist? He said a brief answer of 500 words would do. I went to Canton (now called Guangzhou) in the fall of 1979, spent three days of intensive investigation, then returned to Hong Kong and did more inquiries. Instead of a 500-word answer I sent to Arthur a paper long enough to publish as a pamphlet. In that pamphlet my answer to his question was firm: Yes, China will go capitalist! That was in the fall of 1980.

I was in Seattle and asked my assistant to send copies of the manuscript to a dozen or so people, soliciting comments. They all came back disagreeing with my prediction. Among them four were Nobel Prize winners, two had already won the prize and two to win that prize later.

Milton Friedman’s response was that I was the most optimistic man on Earth. Doug North said no one had ever written a correct history in advance. Gary Becker said I could not possibly be right. And Theodore W, Schultz, who had just won the Nobel prize a year earlier, gave me a lecture: Economics cannot be used to make that sort of prediction. I still possess Ted’s original letter. May be someday I should put it up for auction and see how much it would fare!

My prediction that China will go capitalist is as clear and definite as Newton predicting an apple if separated from the tree will fall to the ground. During my three day trip to Canton in the fall of 1979, 22 years since my last visit, I was received by my oldest sister and her husband, both medical doctors of moderately high cadre ranking.

Canton in 1979 was miserably poor; the streets were pitch dark at night. Queues were observed everywhere to buy items, implying the costs of transactions were enormous relative to the values of the goods. Ranking of the cadres or officials could be identified from where they stood or sat, or from the time they arrived for dinners or meetings—the later arrival must be higher rank.

It was obvious that the cost of transactions was enormous relative to the values of the goods or services. This implied the dissipation of rents was enormous. The verdict I rendered to the comrades receiving me was simple: to sharply reduce this rent dissipation, the economic system of China must be reformed and moved from a system of distributing income in term of cadre rights ranking to a system of property rights ranking.

I told the comrades that human beings are born unequal. I said that if they want an equal distribution of income the human rights must be unequal, and conversely if they want equal human rights the distribution of income must be unequal. I also told them if the costs of transactions were reduced by a few percentage points as a portion of national income, their incomes would soar like a rocket.

The comrades had never heard of the term transactions costs, even less was the term rent dissipation. But they were true experts on transaction costs when I explained to them. They used to stand in lines for hours every day or run around seeking connections through the back doors. Yes, the comrades I met in Canton in 1979, including my sister and her husband, were all world-class experts on the subjects of transaction costs and rent dissipation.

When I made the call that China would go capitalist or change to a market economy, I based it on two observable changes in constraints that were sufficiently clear. First, Deng Xiaoping had made a clear statement that China would open her door to the outside world, and far more important was the observable fact that people from Hong Kong were already pouring into the mainland to see their old relatives, including myself. So, to the mainland Chinese the information costs of knowing the outside world were falling sharply.

Second, industrial investors from Hong Kong were also entering the mainland to employ the far cheaper labor there. This was critical, because the Chinese laborers at that time were all state employees, regulated by the government to a set wage rate of 36RMB per month. These state-assigned workers were lazy, taking long afternoon naps, because no matter what they did they earned 36RMB per month. The Hong Kong investors of course objected to using state laborers, and some of them succeeded to employ laborers by contract, called contract laborers. Here, then, is another set of observable changes in constraints. The government was still controlling the wage rate, so Hong Kong investors turned to pay the workers by counting a worker’s output by the pieces. Their outputs exploded. The workers took afternoon naps no more.

The piece-rate contract began to emerge in China around mid-1980, signaling the state could control the workers no more. The term “state laborer” means the worker was controlled by the state, and “contract laborer” means the worker owned himself. By Irving Fisher’s definition of capital, all contract laborers were capitalists!^[6]

Letting information come in from Hong Kong and the government allowing the use of contract laborers were all the changes in constraints I needed to make a definite call that China would go capitalist. There was no guessing, no luck in my prediction, just as a physicist needs no luck to predict an apple, separated from the branch of a tree, will fall to the ground.

The above approach which I used to interpret the economic transformation of China departs sharply from what my old friend Douglass North used to interpret institutional change. Doug employed the notions of “path dependence” and “institutional inertia” in his interpretation of institutional change in history.^[7] I consider these are non-observable concepts and are therefore incapable of deriving testable implications. Falsification tests can never be performed. My approach is to investigate the observable change in constraints so as to derive testable implications, subject to the dictate of the law of demand.

As noted earlier, the unanimous objections to my prediction that China would go capitalist delayed the publication of the pamphlet until the early spring of 1982. Only a few months after its appearance, rumors were flying that Beijing wanted to make an about turn to go back to the old system. Numerous people approached me to clarify my position. Based on economic theory I again made a definite call: No way China would turn around—the apple was already separated from the tree!

My 1980 prediction was China would reform from a system of assigning rights in comrade ranking to a system of assigning rights in property. By mid-1982 how were they going to carry out this transition became clear. At the earlier stage, the rise of contract laborers was only a small portion of the economy. Now they used a method of contractual substitution which I not only predicted but urged my Beijing friends to push faster. As I expounded in details in a long paper I wrote for a conference organized by Ronald Coase in 2008, the method was through the use of a responsibility contract.^[8] This contract was originally designed to savage the disasters brought about by the people’s communes. By mid-1982 it became clear that the responsibility contract had evolved as a contract delineating and protecting property rights.

In other words, China’s transformation from a system of comrade ranking to a system of property ranking was through a gradual change in contractual arrangements. No revolution, no bloodshed. The change in contractual arrangements was easy to explain after its occurrence, and only a bit more difficult to predict in advance. Correctly assessing the change in constraints is critical.

Friends tend to think I make correct predictions on reform matters all the time. They are wrong. I was doing well at the beginning of China’s reforms, yes, but my predictions got less and less accurate that today I prefer to say no more. The reason is that special interest groups have risen in great numbers. I had to deal with only one group—the ranking comrades—in the late 1970s and the early 1980s. Now there are too many to count. I did predict accurately the number of special interest groups would rise as China’s economy grew—this is easy. But with special interests rising in number the cost of investigation soon became very high. The basic principles or axioms in economics remain the same and I see no reason to change them, but the costs of investigation soar by multiples as special interests increase even in small numbers. And one who chooses to investigate can only live so long.

Let me end this section with an encouraging note. A friend who read my pamphlet Will China Go Capitalist?, written in 1979, and then read my pamphlet The Economic System of China, written in 2008—the two works 30 years apart—told me these two pamphlets gave him the feeling that they were written in one flight of thought. He said he felt as if he was standing on the side of a highway, seeing a car coming from his left, passing by him and leaving from his right—in a continuous movement with no interruption. I replied that as an axiomatic science economics is supposed to have the function of explaining the past and predicting the future, although the required specification of constraints is often highly difficult.

7 The Disastrous Development of Economics

Let me emphasize that as an empirical science economics may not only explain what has occurred in the past but may also predict what will occur in the future. To demonstrate the power of prediction, I have used a simple example of predicting the disappearance of the $100 bill to the complex example of predicting the economic transformation of China. I have also discussed that the used of econometric method to handle multiple variables is not robust, and that economists often commit the fallacy of letting the facts speak for themselves.

I now turn to a major disaster in the development of modern economics, stemming from the use and spread of non-observable entities in describing and analyzing events. We cannot test a hypothesis with entities which we cannot observe! But now non-observable entities are everywhere in the paradigm of modern economics. I may be the man who ignited this unfortunate development. The following is my recollection of the sequence of events.

I was at the University of Chicago revising my book on sharecropping when Armen Alchian spent a year at Chicago on sabbatical from UCLA. That was 1968. He was my mentor and we were together for lunch often. When revising what later became chapter 4 of my sharecropping book, I concentrated on different aspects of enforcing contractual performance.

The idea of “shirking” was somehow central in my thinking at that time, because in different forms of contracts—wage contract, fixed-rent contract, share contracts—the participants would tend to “shirk “or cheap at different margins. This shirking thesis was one of my main concerns in Section 4, Chapter 4 of my sharecropping book. Its publication date was 1969, and I double faulted in letting that Chapter 4 published as a separate paper in the Journal of Law and Economics that same year.^[9]

When pondering this Section 4, I asked Alchian during lunch using the following example. Two men carrying rocks down a hill, if they carry separately each could carry 100 pounds per trip, and the two men combined could carry 200 pounds if they do so separately. But if the two men carry rocks jointly, as it was often observed in China, they may carry 250 pounds per trip. There would be a gain of 50 pounds, or 25 pounds each, if each performs as diligently as when he carries the rocks alone. However, in their joint effort each of the two men would tend to shirk, pushing the weight of the rocks to the other side. Suppose, I asked, with the shirking the two men combined carried a total of 220 pounds, with a share of 110 pounds for each, how is this weight determined?

In my view, this verbal example was where the catastrophic development in modern economics began, and it may even have contributed to the resurrection of game theory beginning in the 1980s. It just did not occur to me at that time that “shirking” is not observable and is therefore worthless in the derivation of testable hypothesis.

Based on this “shirking” idea, Alchian and Demsetz jointly produced a paper on economic organization, which later became the most cited paper in the history of the American Economic Review.^[10] They did cite my work in that paper.

Putting more fuel to the fire, John McManus visited me in Seattle in 1969, soon after I joined the University of Washington. He stayed at my home for several days, and I told him an episode which soon, too, became classic in economics. I told McManus that during WWII, when avoiding the Japanese my mother took six of her children to the province of Guangxi. While travelling by boat on a river, a team of trackers was pulling ropes along the river shore to assist the forward movement of the boat. There was also a monitor holding a whip in his hand, monitoring the performance of the boat trackers.

My mother, who participated in the negotiation of the boat-pulling charges, told me when the boat began to move: “Can you believe this? The man who holds the whip is hired by the boat trackers to whip them.” This may be the truth, but my mother, the brightest person I ever knew, liked to make up stories with imagination to entertain her son.

McManus put the tracker-whipper story in his paper on the firm and submitted to Coase. Coase asked for my opinion. I recommended rejection not because John’s paper was not good, but because I was not sure my mother had told me the truth. McManus’ paper was later published in a Canadian journal.^[11] Then Michael Jensen and Bill Meckling repeated the tracker-whipper story in their famous article on the firm,^[12] giving credit to McManus, and then an economist in Australia put my name on the title of his paper involving trackers and the whipper.

The true fact we can say today is that there were boat trackers and a monitoring whipper, but as to the question of who hires who, my present view is only God knows. I am serious about this. When some decades back the University of Hong Kong wanted to hire a new chancellor, I was on the selection committee. We hired the chancellor, but then he hired some of us and determined our salaries.

The use of non-observable entities began to soar in the economic literature, spawning hypotheses emerging all over the places that were not testable. In 1976 I was asked to review Oliver Williamson’s 1975 book on Markets and Hierarchies.^[13] I simply did not know what to say because that work was full of terms foreign to me, and no testable hypothesis whatsoever could be found in that work. I consulted Chris Hall and Yoram Barzel, and both of them said Williamson wrote a new dictionary. It speaks of the fate of economics that based on this work, Williamson later became probably the most cited economist of our time! Of course, we cannot say Williamson’s works are not economics. But with so many non-observable entities and jargons generated from what he called “opportunism”—which is also not observable—no testable hypothesis can possibly be derived from his works.

Then in 1978 Klein, Crawford and Alchian published their famous piece on vertical integration. This paper is about “hold-up”, another non-observable entity.^[14] Before its publication Armen sent me a copy for comments. In it I found two examples they used that I knew were simply not true.

The authors said that, in order to avoid hold-ups, a petroleum company would own pipelines but would rent tankers for crude oil transportation. I was an expert on transporting crude oil at that time, and wrote Armen, saying that petroleum companies rent pipelines as a matter of routine, but all major refiners have their own tanker fleets. What followed was the joint authors simply deleted the pipeline example and the tanker example, and let the paper go into print!

Game theory in economics, which I studied briefly in 1962, disappeared thereafter but returned to enjoy great popularity beginning around 1980, is also full of non-observable entities and therefore not testable. Its practical usefulness in economic predictions therefore is zero—if we abide by the philosophy of an axiomatic empirical science.

Let me repeat, observability is a vital requirement in an axiomatic empirical science because a hypothesis must pass the refutation test, and such a test can only be carried out with the use of observable variables.

8 The Indispensable Law of Demand

Current economics textbooks tell the students there are 10 or more axioms in economics. There are several of those in the analyses of the indifference curves alone, and then more beyond that. I use only one: the law of demand, which states that quantity demanded is a negative function of price. Throwing the indifference curves away is to clean up the mess.

Yes, the demand curve I use is completely devoid of utility content. Instead of utility I employ the notions of use value and exchange value, and consumer’s surplus is the difference between the two. These are observable entities, at least in principle. The Austrian School also uses them. Presumably they originated in Adam Smith’s analysis of the water-and-diamond paradox. In my view, the fallacy Smith committed in that paradox does not negate the simple and useful notions of use value and exchange value.

Cost and price are different angles looking at the same thing, and the law of demand is the same as the law of diminishing returns. Why? Because all of our consumption items may be regarded as factors of production. Why not, we have to eat and sleep in order to work. Therefore, what must be kept in economics are the law of demand, the concept of cost, and the implications of competition.

I have on numerous occasions said that the world is an enormously complicated place, and it is foolish to interpret complicated observations with complicated theories. What is more, in the classical and neo-classical paradigms, what we have inherited from the grand masters of the past were ideas mainly intended to improve society and not to interpret society. Lionel Robbins wrote a book on just that.^[15] Alfred Marshall, my hero, rendered an analytical framework that guided me to develop my own. However, Marshall floundered in his analysis of the indispensable law of demand because he accepted the Giffin Paradox.

Getting rid of the Giffin Paradox is not difficult. I do not deny this paradox is a logical possibility, but we have three ways to throw it away. One is the method of Armen Alchian and the other two are mine. Armen’s method of getting rid of the Giffen Paradox is to assert it does not exist in the real world, thus rendering the law of demand a postulate or an axiom by itself. I like the following two methods of my own more because they are more interesting and allow us to see the real world in more insightful ways.

My first method is by pointing out that a Giffin good cannot be traded in the market! This is not difficult to prove, and I wonder why in Ph.D. qualifying examinations this elementary question was not asked. My second method to get rid of the Giffen Paradox is to interpret the law of demand as the same as the law of diminishing marginal productivity. Why not? A lady putting on lipsticks and beautiful dresses are for what purpose? To attract gentlemen or to get a better job, thus increasing her income. Therefore, her demand curve for dresses is again the same as the curve of the law of diminishing marginal productivity.

9 The Non-Observable Quantity Demanded and a Fallacy in Alchian’s Law

Let me turn to another issue relating to the law of demand, and this is far more difficult. The law states quantity demanded is a negative function of price. Here, the variable of price is generally observable, but “quantity demanded” is not. There is simply no such thing as “quantity demanded” in the real world.

One way to handle this non-observable “quantity demanded” is through the test of implication, involving the use of additional constraints. For example, a professor obtains a grant to do research, and this grant money is subject to more restrictive use than the same amount is given to him as an increase in his salary. Suppose a dollar of grant money is only worth 60 cents of a dollar increase in salary. Making Xerox copies is allowed under the research grant. The law of demand therefore implies the professor will make more Xerox copies under the grant than if the same amount is given to him through a raise in his salary.

Finally, also in the domain of the law of demand, let me turn to save what has become known as Alchian’s Law.^[16] This law was first introduced to me in Alchian’s price theory lecture at UCLA in the fall of 1963, appearing to everyone as beautiful and simple. But in 1969, J. Gould and J. Segall clearly rejected Alchian’s law through the use of indifference curves analysis. The debate got hot at the University of Washington, with Borcherding, Silberberg, and Umbeck engaging in the debate. No definite conclusion was reached, until I wrote my five-volume treatise on Economic Explanation after the turn of the century.^[17]

My verdict: Armen’s conclusion was correct, but his analysis was wrong.

The issue first cropped up in the fall of 1963, when Armen pointed out that Sunkist oranges, the highest-quality brand in California, was singularly selected for shipment to foreign countries. The question was why the lower quality oranges, which were plenty, were not at all shipped abroad. Armen’s answer today becomes Alchian’s law.

Alchian’s law states that price means relative price. This is of course correct. He argued that when a given transportation cost is added to the shipment of oranges, and this cost is the same for high and low-quality oranges, the price of the high-quality oranges will fall relative to that of the low-quality oranges. Therefore, high-quality oranges like the Sunkist were selected to ship to foreign countries, but not the lower-quality ones.

This logic seemed flawless, but in 1969 Gould and Segall of the University of Chicago Business School published a paper in the Journal of Political Economy, that used indifferent curves analysis and flatly rejected Alchian’s logic.^[18] The analysis of Gould and Segall was also flawless in logic. However, the observation of shipping only the good oranges out to a foreign country remained evident. Why?

When the issue reached the University of Washington in the 1970s the example was shifted to apples, perhaps because the State of Washington was one of the largest apple growing areas in the world. Here again the evidence was incontrovertible. The highest quality apple produced in the State of Washington at that time was the red delicious, and clearly only the red delicious apples were shipped to Hong Kong. Why? Negating Alchian’s argument in logic does not negate the fact the only good oranges and apples were shipped to far away countries.

Among my colleagues at the University of Washington who participated in the debate, Tom Borcherding, Eugene Silberberg,^[19] and John Umbeck^[20] wrote two papers to rebut Gould and Seagall. But the issue remained undecided.

When it came to my turn to tackle the Alchian’s Law after the turn of the century, I could not avoid the apple or orange issue because the first volume of my five-volume treatise was on the law of demand, and I reached the view that looking at good and not-so-good apples is not correct. Rather, we should look at the ingredients inside the apple.

For simplicity, let us assume it is the sugar content that measures the quality of the apple. Let us say the sugar content is worth five cents per unit, and the transportation cost is 10 cents per apple. Now in the State of Washington, with no transportation cost, an apple containing four sugar units is (five cents × 4), or 20 cents, and an apple containing two sugar units is (five cents × 2), or 10 cents. The relative price is 20/10 = 2.

Now we add transportation cost of 10 cents per apple. The good apple with four sugar units becomes 20 cents + 10 cents transportation = 30 cents; while the poor apple with two sugar units is five cents × 2 = 10 cents sugar, plus transportation cost of 10 cents becomes 20 cents per apple. Therefore, by adding a transportation cost of 10 cents, the ratio of good versus poor apples becomes 30/20 cents = 1.5.

It is now clear: the relative price of good/poor apples without shipping costs was 20/10 = 2 but adding transportation costs of 10 cents that ratio becomes 30/20 = 1.5. Therefore, only the good apples are shipped out because they are relatively cheap after shipment. That is, using sugar units to measure quality, the relative price of good versus poor apple before shipment is two but after shipment is 1.5.

This implication of the law of demand is universal when applied to other observations. When someone travels all the way to Paris to drink wine he would ask for higher quality wine than he would drink at home. When I get all dressed up and take a classy lady to a high-class restaurant for dinner, I would not buy her hamburger. When a mother sends a suitcase of clothing to her son studying abroad, if the shipping cost is measured by the suitcase unit and not by its weight, the mother would stuff the suitcase so full and tight that it can hardly be lifted. And in housing development, view lots are usually built with higher-quality houses than lots without views. These behaviors are all implied by the law of demand. I submit that in the paradigm of economics, all human actions are.

The law of demand is the only law which economics cannot do without, and yet one of its variable, quantity demand, is a conceptual entity which we not only cannot observe, there is simply no such thing in the real world. As I have just shown, with only one non-observable entity we find ourselves scrambling, and now economics has developed into a discipline involving numerous non-observable entities. In my view, economics as an axiomatic empirical science has turned disastrous.

10 Concluding Remarks

Economics is an empirical science in that it deals with experience and facts. All social sciences are empirical in nature. Economics is the only social science that is both empirical and axiomatic. The use of axioms in an empirical science renders the power of not only able to interpret or explain what has occurred in the past, but also the power of predicting what will occur in the future. All physical sciences have this dual function. As a social science economics also has this dual function because it is the only social science that is axiomatic.

In order to achieve the above dual function, the testing of hypothesis is absolutely essential—there is no way we can get around this! The point to address here is that in performing a test, we must use variables that are observable. It is unfortunate that in economics we have one non-observable variable which we must keep, namely, quantity demanded. With this one single non-observable variable we often have to scramble to get by. In my view, it is our ability in handling the non-observable “quantity demanded” that saves economics as an axiomatic empirical science.

If you are as good as Gary Becker or Armen Alchian you may add another non-observable variable called “utility”. I prefer to play safe, finding a way to handle the so-called non-pecuniary goods. Such goods exist, yes, but there are ways to get around them. But please, no more non-observables.

I agree with Keynes that compared with the physical sciences, economics is an easy subject. I disagree with Keynes in his explanation of why there are so few in the discipline who excel. In my view, the difficulty of excelling in economics lies mainly in the requirement to test a hypothesis or confirm a proposition, because we cannot test a hypothesis with entities or variables which we cannot observe.

I am 85 years old. It must be the biological nature of mankind that I can remember what my mother said to me 83 years ago but not what my wife said to me 83 min ago. We would have to consult Richard Dawkins to find out if other animals are like that.^[21] So, the readers can imagine that I have to struggle writing this paper. Strangely, my imagination and analytical ability seem to remain intact. And my judgment of what aspects of an idea are important is firmer now than in my younger days.

Economics is an old man’s science because we have no man-made laboratories. The only laboratory we have is the real world. It is such a complicated world that we have to dig so deep and pay so much attention to details before we can obtain a picture that is roughly right. But human beings are biological creatures, and like what we have said about the biologist’s prediction of the behaviors of butterflies, roughly right is good enough.

I waited until my retirement from teaching in 2000, at the age of 65, to begin writing my five-volume treatise entitled Economic Explanation, using the Chinese language. After four rounds of revisions, that task took a total of 20 years. Few axiomatic empirical sciences call for such a long stretch of effort. The reason was that I had to investigate and accumulate data and information from a complicated world which I could not control and manipulate. I was driven by curiosity to investigate. And when conflicting observations were encountered I had to wait for more information to get the conflict resolved. For example, I began investigating piece-rate contracts in Hong Kong factories during the summer of 1969. With knowledge gained from this and other investigations, particularly in the practice of sub-contracting, I published my firm paper in the spring of 1983.^[22]

It is gratifying that my five-volume work is now not only read by some university students in China, but its readers are mainly government officials and businessmen. It appears as if I have brought economics back to the era of the science began, when people read the works of Smith, Riccardo, and Mill did so out of curiosity and wanted to understand the world better.