We have drawn a line to all the dots to show the so-called fitted consumption function. All the points fall to the right of the 45-degree reference line (where real disposable income = real consumption), such that every year, aggregate or national saving was positive.

Keynes's theory seems to have been borne out by the empirical evidence. Real consumption is greatly dependent on real disposable income, and as real income changes, so too does real consumption (though by a smaller amount). There seems, however, to be a difference between the historical consumption function represented in Figure 12-1 and the hypothetical household consumption function represented in the graphs in Chapter 12 in Economics Today. In Figure 12-1 here, there is no autonomous consumption; that is, the consumption function never crosses the 45-degree, C = Y_d, reference line. In fact, it intersects the vertical axis at zero. How can this be? These two different consumption functions can be reconciled by distinguishing between the short and long run.

SHORT-RUN VERSUS LONG-RUN CONSUMPTION FUNCTIONS

The historical (long-run) consumption function for the United States in Figure 12-1 goes through the origin. This means that because it starts at zero, the long-run consumption function shows no autonomous consumption. Such is not the case with the hypothetical example that we used in Chapter 12 in Economics Today. In that example there was an autonomous component to consumption. Also, we can see that the slope of the consumption function in Figure 12-1 here and the slope of the consumption functions in the graphs in Chapter 12 in Economics Today are quite different. The hypothetical consumption functions in the graphs are similar to actual short-run consumption functions; the empirical evidence suggests that distinct short-run and long-run consumption functions exist. The long-run consumption function in Figure 12-1 here shows a marginal propensity to consume of about 0.9. (Incidentally, the average propensity to consume is also 0.9. Why?) We see, then, a basic inconsistency between the short-run consumption functions used in our examples and the historical long-run consumption function. One way to reconcile this inconsistency involves the permanent-income hypothesis.

THE PERMANENT-INCOME HYPOTHESIS

The permanent-income hypothesis (PIH), as developed by economist Milton Friedman, can be distilled into two basic statements:

1. Currently measured annual real income is a poor clue as to the economic status of an individual, a family, or a nation.
2. Individuals or families base their consumption on their economic status, not on their current annual real income.

Consider the first statement. Suppose all you know about a person is his or her income for one year. How much information concerning that person's economic status do you really have? Probably not much. For example, that individual might be a college student whose current income vastly understates what he or she will earn in the future (one hopes). Current income for most young people generally understates their economic status or permanent income.

Or suppose the worker in question is retired: his or her accumulated wealth and economic status may well exceed the economic status that the individual's current annual income suggests. Or, during the period when a person's income is measured, that person could be temporarily on strike, ill, or injured. Likewise, such groups as the self-employed, farmers, or salespersons often have an exceptionally good or an exceptionally bad year economically.

In short, it is important to know an individual's normal long-term permanent income before you categorize his or her economic status.

Now consider statement 2, which, in effect, says that people base their consumption on their permanent income and largely ignore deviations from that permanent income if they expect the deviations to be only temporary.

College students, anticipating higher income levels in the future, will tend to spend more than will people whose permanent incomes equal the college students' temporarily low income. Workers on strike or who are temporarily ill will not radically reduce their consumption to match what is expected to be a temporary income reduction. Instead, they will sell accumulated assets and spend the proceeds. This point is very important, because the consumption function normally used in textbooks maintains that current income constrains current consumption, but the permanent-income hypothesis maintains that wealth is the true constraint on current consumption. [Permanent income is closely related to wealth for Friedman, who defines permanent income as the amount that can be spent currently and yet leaves wealth unaltered. That amount is r · W, where r is the interest rate and W is total wealth. Total wealth is composed of both nonhuman wealth (real estate, stock, bonds, and so on) and human wealth (the present value of future labor earnings).]

When we get data on families with different income and different levels of spending, we can assume that many high-income people are experiencing those levels of earnings only temporarily, not permanently. For such people, their average propensity to save (APS) is higher than normal because they base their consumption on their normal, or permanent, income. They save much of their temporarily higher income in order to acquire assets on which to draw when their income is temporarily low in the future. (Or they purchase durable goods, which is a form of saving.) Their average propensity to consume (APC), therefore, appears to be relatively low. Conversely, many people with low income may be at earnings levels that are abnormally low for them when compared to a higher level that they consider more permanent. These people will probably be saving very little--or even dissaving--in order to maintain their normal consumption, which depends on their permanent income. Thus we can imagine plotting a consumption relationship like the ones in Chapter 12 in Economics Today. The permanent-income hypothesis would predict such a consumption relationship at any point in time.

This hypothesis would also be useful in explaining the behavior of medical students, for example. While they are in medical school, students' consumption expenditures, in general, greatly exceed their actual income. This is understandable, however, because medical students' permanent (anticipated) income will be much higher than their current income. Consumption is geared to expectations of permanent income rather than to current levels of income.

This hypothesis also predicts the long-run consumption function in Figure 12-1 here. Proponents of the hypothesis contend that there is no difference between marginal and average propensities to consume if one looks at permanent income rather than current income. Accordingly, we would not expect to see the "permanently" rich saving a larger percentage of their income (in the long run) than the "permanently" poor, other things being constant.

HOW DO WE MEASURE PERMANENT INCOME?

Permanent income is difficult to measure because there are difficulties in deciding which part of current measured income is transitory and which part is permanent. Difficulties arise in the determination of what the individual's (or family's, or country's) future income will be; difficulties arise also in the determination of the human and nonhuman wealth of the subjects involved. Even if we could amass an impressive amount of economic information about a family or a country, we still have the problem that people themselves may not know what their permanent income is.

THE ADAPTIVE-EXPECTATIONS HYPOTHESIS

Friedman proposed a simple method for estimating permanent income that may approximate the method by which people themselves estimate their permanent income; this method is now referred to as the adaptive-expectations hypothesis. This theory maintains that permanent income is related systematically to present income and past income.

Assume, as Friedman did, that an estimate of permanent income can be made from what actual measured income has been in the past and what it currently is. In a complex model, a proxy for permanent income can be derived by using a weighted average of measured income over many periods and by assigning relatively larger weights to the more recent years. An even more complicated model would also take into account the growth of permanent income through time; and Friedman has done this, too.

THE PERMANENT-INCOME HYPOTHESIS AND ECONOMIC STABILITY

The permanent-income hypothesis suggests that the private economy is inherently stable for two reasons. First, because permanent income is a wealth concept, then, to the extent that the permanent-income hypothesis is valid, it provides support for the Pigou (wealth) effect. A.C. Pigou of Cambridge University demonstrated that a falling price level increases the real value of currency--paper money and coins--holders without decreasing anyone else's wealth. Thus the Pigou effect indicates that a falling price level due to some real effect such as unemployment (not a reduction in the money supply) will increase the wealth of some persons but not decrease the wealth of others.

Therefore, a falling price level will shift any consumption function upward. This means that price flexibility is capable of restoring the economy to a full-employment equilibrium after the economy has been shocked. Second, when the economy strays from its natural growth path during booms and recessions, the permanent-income hypothesis implies that consumers will respond only in part to such transitory changes in national income. Temporary booms, therefore, will not be fueled by large increases in consumption; and temporary recessions will be cushioned somewhat by consumers who maintain most of their consumption expenditures. Stated differently, the permanent-income hypothesis implies that the average propensity to consume falls in booms and rises in recessions; this countercyclical tendency is inherently stabilizing.

Copyright © 1995-2010, Pearson Education, Inc., publishing as Pearson Addison Wesley Legal and Privacy Terms