Stigler spends a long time developing utility theory. He claims that abstract utility with its questionable foundation is a useful tool, because of its predictive power. Though the foundations are not great, the results justify it. Stigler spends the rest of chapter four showing off the power of utility theory.
In order to test the predictive power, we obviously need to make some predictions from our model of utility maximization. Stigler sticks to mostly graphical predictions that involve reaching the highest indifference curve that is still within the consumer's budget. The graphs allow economists to do comparative statics. If one aspect changes, how does that affect the consumer's bundle?
One type of change is variations of income. On a graph, this corresponds to a shift of the budget, illustrated below. Normally, when people are richer they consumer more of a good, hence normal goods. Other times, when a consumer is richer, he consumes less of this good. The classic examples are ramen and potatoes. As college graduates start making money, ramen consumption declines. These are inferior goods.
While not a formal prediction, the outline of utility theory's predictive power is starting to form. This model of normal and inferior goods suggests which types of goods will increase with wealth. Inferior goods convey the idea of "buying because it is cheap." This is simply a definition, but it clarifies the thinking.
The second comparative static involves changing prices. On the graph, a change in prices represents a change in the slope, since the slope is -dpY/dpx. With most goods1, increases in price lead to decreases in consumption. This is the classic law of demand.
To better understand this mechanism, economists differentiate two effects of changing prices. For example, when the price of gas goes up, I am relatively poorer and buy less gas in terms of quantity. The price change creates a wealth or income effect.
Also, gas becomes relatively more expensive, compared to the train. Now I can buy more train tickets for each gallon of gas, so I switch some consumption to train tickets. I get more bang for my bung. This is the substitution effect. In the real world, these effects are always entangled. However, the distinction is a tool for thinking through changes in prices.
The most obvious prediction of this is that demand curves are downward sloping. Stigler dismisses this prediction, since all known demand curves are downward sloping. This is too general to really be a prediction, although it is always important to keep in mind.
Instead, Stigler focuses on some basic predictions. The first underlies the law of one price. Assume two identical goods are selling for different prices. If we assume the consumer is trying to buy and then consume as much as possible, it follows that he only purchases the cheaper good. Since only the cheaper seller actually sells anything, the market will have only one price. Stigler does not make this jump to the law of one price, but it is there.
Utility theory has also clear predictions about how people react to changes in relative prices. Suppose an entrepreneur wants a nice office making monetary profits. Those are his two goods. A new regulation or tax makes it harder for him to make monetary profits, thus raising the "price" of profits2. It is relatively easier to build a nice office than earn profits.
The model we have constructed predicts that regulation will result in lower profits and nicer offices. This is testable and comes from the substitution effect. We can also say that the entrepreneur is worse off because of the regulation because it raises prices. This comes from the wealth effect, since the entrepreneur can no longer reach his earlier preferred combination of profits and office.
Utility Theory as Bridge
Stigler also believes that utility theory provides a bridge between the objective actions taken by individuals and the subjective preferences. Utility theory says that a consumer will buy the bundle of goods within his budget that gives him the most utility. Therefore, if a bundle of goods, say 4 apples and a TV, was within the consumer's budget and the consumer does not choose it, it would be worse off with the apples and TV than with what he chose.
This idea motivates cost-of-living indexes, which are popular in economics. If a consumer has $1,000 more income in 2014 than 2013, is he better off? Right away, we do not know. There is one way to tell. If he could afford exactly what he chose last year and does not choose it, then he is of course better off. If he could not afford last year's bundle, he is at a lower indifference curve. The index attempts to compare the costs of bundles across time. Comparing utility and saying if 2014 is better requires constant preferences, which is a strong assumption.
The other two applications of welfare analysis are closely related. The first is consumer surplus. If I would have paid $10 for a great burger, but ended up getting it for $5, people say it is a "good deal." Consumer surplus is a way to conceptualize this. Stigler defines it as the gap between what I would have paid and what I did pay. Stigler does not claim that this measures any sort of true psychic utility, but more surplus is clearly better than less.
One application of this is price discrimination. If the seller of that burger knew I was willing to pay $10, he would have preferred to charge me $10. The better he is a discriminating a price, the "better" side of the deal he will get, even though we know that both parties are better off in any voluntary exchange.
For Stigler, economics is prediction. The only way we know that we understand something is if we can predict it. He constantly stresses this. If I claim to understand football games, but do no better than a coin flip in predicting winners, I probably don't know that much about football. This importance of prediction is different from the view of Austrian view, who love to disparage Samuelsonian economics' obsession with prediction and its failure to predict future economic outcomes such as GDP.
However, in this chapter, Stigler is stressing a much more Austrian-friendly approach to prediction. He is not saying that if the price of a good goes up $1, the agent will consume .25 units less. He never does the calculus of a maximization problem to find the exact change in demand. Instead, he makes directional, ceteris paribus, predictions. This is what Austrians do. The model is a tool to understand ideas, instead of a simple input/output algorithm which tells the economist exactly what will happen.
The foundations of utility theory are too ad-hoc and shaky to move Stigler's approach. He is much more hesitant about his claims on utility and somewhat justifies his assumptions. He clearly has some attachment to authors, such as Rothbard, who reject utility theory. For example, Stigler is clear that consumer surplus is how much more the consumer would pay. It is not a measure of utility.
By the time of Mas-Colell, this hesitation is gone. Economics is applied utility theory. One can read this change as finally ignoring questions that were long ago settled or as closing off to outside criticism. I tend towards the latter.
Economists need to be aware of all their assumptions and the baggage that entails. We need to know the limits of our assumptions and their predictions. If we make utility theory and welfare for predictive power, we should not use them to turn economics into a poorly thought through branch of philosophy.
I'm still always worried when even talking about welfare and utility, since so many economists use the tool while forgetting the methodological limitations. In our interest as social scientists we ask, "how happy is she because of this?" Stigler already dismissed any idea of adding, counting, or comparing utilities. But economists love to ask "how big?" and not just which direction. It is a tough field to navigate.
1. Giffen goods decrease in consumption when the price goes down. This is theoretically possible, since negative income effect for inferiors goods could be larger than the positive substitution effect. Giffen goods violate the uncompensated law of demand which we discussed in MWG.
2. The "price" of a good is not just the sticker price, but involves everything required to get that good.