The Saving Function

By definition, consumers save that portion of their disposable income which they do not spend on consumption. The relationship between consumer saving and disposable personal income is shown in Fig. 7-2, which gives the percentage of d.p.i. saved each year. The underlying data are the same as those for Figure 7-1. The percentage saved is of course one minus the percentage consumed for each year.

Figure 7-2 indicates vividly the high percentage of d.p.i. saved in the war years, and the negative savings in 1932—33 at the bottom of the Great Depression. But more significantly it stresses the nearly constant percentage saved (about 4—6 per cent) during the entire period of reasonably stable, prosperous times since 1950.

FIG. 7-2 Personal saving has been
a stable percentage of disposable
personal -income in peacetime pros.
perify, but has fluctuated sharply
in depressions and wars. (Source:
U.S. Department of Commerce.)

What Determines

Consumption Expenditures?

The preceding data reinforce oar intuitive presumption that consumption spending depends mainly on disposable personal income. But the data also warn us that other forces may shift the propensity to consume, slightly or dramatically. We need now to take a more detailed look at the forces which have influenced consumption spending over the past half century.

Income—Present, Past, and Future. Empirical studies show a close relationship between present income and current consumer spending. But sophisticated statistical work over the past decade has suggested that adding in past income and income expectations gives a better explanation of consumer spending than does present income alone.

The influence of past income is persistent. Once families have become used to any real consumption level, they are reluctant to slide back down to a lower level, even if their income drops. Rather than reduce their standard of living as income falls, they will (at least temporarily) reduce their saving levels to well below the amount they would have saved at that income on the way up.

Similarly, when incomes rise sharply, consumption spending rises more slowly. That is, with rising incomes the marginal propensity to consume is lower in the short run than in the long run. For example, following the big federal income tax cut in 1964 which directly increased disposable personal incomes, the percentage of d.p.i. saved rose temporarily, and then gradually shifted back to its earlier level, as previous studies of consumption indicated it would do. Consumption spending takes time to adjust.

Recently, some research workers have advanced a more sophisticated hypothesis. They suggest that consumption spending is a substantially constant and similar proportion of disposable income for average families at all income levels —if we exclude major disruptions like war and mass depression, if we include consumer durables (especially houses, autos, and household furnishings) with saving (investment), and if we consider their “permanent” or “life-span” incomes rather than the particular income of any given year, which may be distorted by special factors. It is the last clause which is the crucial one.

One version of this approach says that families, consciously or subconsciously, estimate their long-range income over the years ahead (over their entire life cycle), and adjust their current consumption spending and saving to their rough expectations for the entire period. As a college professor, I have a pretty good idea how my salary will rise over the years ahead to retirement, barring major calamities and other such special factors. The same is true for many other families. Thus, during their early married years most couples spend most of their income and even go into debt to start families and set up households. A little later, as income rises and these special expenses have passed, they begin to save more, for retirement, to send their children to college, and so on. After their houses are well furnished and their children are educated, they commonly save at a much higher rate. Late in life, after they retire, the saving ratio drops again and often becomes negative. Looking over this life span, the average ratio of consumption to irkome is, if the investigations of this group stand up under further scrutiny, surprisingly similar for typical families at different income levels. By the same token, consumption is a quite stable function of income for the population as a whole.

The “permanent income” hypothesis is the other way of putting the analysis. It says that families base their consumption on what they expect their “permanent” income to be. Temporary deviations in income up or down from this permanent expectation will generally not greatly affect consumption spending. Rather, if income falls temporarily, the family will cut back its savings, use up its liquid assets, or go into debt to maintain consumption. If income bulges temporarily, the bulge is likely to go mainly into saving (including consumer durables). Just how families actually decide what their “permanent,” or lifespan, income is, is not clear; past incomes presumably provide the best indicator.

What is the evidence on the life-span and permanent-income models? Good statistical evidence is hard to get, and interpretation of the data in relation to these models is a tricky problem. Most economists believe that the evidence to date generally supports the models, but with enough contradictory data to require a suspended judgment so far as practical use of the model is concerned.

Money, Liquid Assets, and Other Wealth. If a family has an unusually large accumulation of money and other liquid assets (currency, bank deposits, government bonds, and so on), it probably feels freer to spend out of current income than it would otherwise. If its liquid assets are unusually low, the reverse will be true. This seems an intuitively reasonable hypothesis.

And there is some empirical support for it. During World War II, government borrowing placed nearly $20 billion of new liquid assets (money and government securities) in the hands of the public. After the war, this huge accumulation began to burn a hole in the public’s pocket. Between 1945 and 1947, consumption spending rose from 81 to 98 per cent of d.p.i. Then, as incomes and prices caught up to a more normal relationship to liquid assets, the consumptionincome relationship eased back down to previous high-employment peacetime levels.

The facts are clear. But just how much weight should be given to money and other liquid assets is debatable. Skeptics point out that other factors (especially the pent-up demand for consumer durables unavailable during the war) may also help explain the postwar spending burst. Current research suggests that households’ money holdings (of deposits and currency) may exert an important casual influence directly on consumer spending, through channels to be explained in Chapter 8. But this analysis attributes little causal power to other liquid assets. Best evidence to date is that when money balances (and possibly other liquid assets) get far larger relative to current income (or to consumers’ total wealth) than the public is used to, they exert a strong pressure toward higher spending. But it’s less clear that small variations in money or liquid assets play an important causal role. Reserve judgment until Chapter 8, which integrates money and other liquid assets into our basic aggregate demand model.

More broadly, it is clear that, other things equal, a family with large wealth (ownership of economic assets such as money, stocks, bonds, houses and consumer durables) will spend more on consumption than will a low-wealth family with the same income. For example, suppose a family receives an annual income of $10,000 and has accumulated wealth of $25,000. Imagine now that a wealthy relative dies and leaves family another $25,000. We would expect its level of consumption to rise even if the father’s salary stayed at $10,000. Note, incidentally, that this effect is closely related to the “permanent” income hypothesis above. With more wealth, the family’s permanent, or life-time average, income has increased, since the family’s income now includes the interest, dividends, or other yield on the additional $25,000 of wealth.

Summary: Economists agree that wealth exerts an important effect on the level of consumption, though we must be careful not to doublecount the wealth by also counting its effect on permanent income. Some, but not all, econometric evidence suggests~ that money, as one special form of wealth, exerts a special effect on consumption spending, beyond the impact of the total wealth owned by households.

Consumer Credit. One way of getting around the limitation of income is to borrow money or buy on credit. A net increase in consumer credit correspondingly increases consumer spending power, beyond that provided by current income. Until the last couple of decades, consumers did little credit buying except in purchasing homes. But since World War II a huge volume of houses and consumer durables (especially automobiles) has been bought on credit. By 1967, families were nearly $100 billion in debt on consumer goods, compared to $6 billion in 1929 and in 1946. This growth thus represented a net addition of over $90 billion to total consumer spending power over the period. Mortgage debt on houses approached $250 billion in 1967, up over $200 billion since 1946. In total, such new credit to households thus increased family buying power by nearly $300 billion over the 1946—67 period.

But there is a counterforce at work here too. The deeper consumers are in debt, the less they can count on increasing this supplement to current income in future periods, and the greater is the potential inroad on current consumption spending from required payments on the debt. If you are in debt, interest and debt repayments must be met before current income can be devoted to consumption expenditures. By 1967, the proportion of current disposable income committed in this way had crept up to 15 per cent, which correspondingly restricted consumers’ power to spend disposable income on goods and services.

Availability of Goods. During World War II, you just couldn’t buy lots of “hard goods” like refrigerators and automobiles, because their production was cut back or eliminated by the war. Reflecting this and other factors, the economy’s consumption ratio dropped to below 75 per cent of d.p.i. in 1943 and 1944. By 1946, consumers had accumulated an enormous backlog of demand for such consumer durables, which undoubtedly helps explain the big postwar buying surge. Such shortages of goods are rare, but they may exert a powerful force on the consumption function when they occur.

Consumer Stocks of Durable Goods. The converse effect occurs when consumers have built up unusually large stocks of durable goods. Consumer spending on nondurables (food, clothing, and so on) and on services (housing, utilities, etc.) is relatively stable, but spending on durables (automobiles, refrigerators, TV sets, etc.) fluctuates sharply with fluctuations in income. Since such durables last, their purchase can be postponed far more readily than the purchase of food or services. Thus, after consumers have engaged in a big buying spree on durables (for example, 8 million autos in 1955), they are likely to slack off their buying until the new durables are at least a few years old, even though consumer income holds up. Note that this example applies to only one sector of consumer spending, not necessarily to total consumption expenditures.

Price Expectations. If you expect prices to rise, the time to buy is now, before things go up. If you expect deflation, you had better hold off postponable purchases until prices go down. Changing expectations of future prices can thus bring violent shifts in the consumption-income relationship. Immediately after the outbreak of the Korean War in 1950, for example, current consumer saving dropped almost to zero as consumers rushed to stock up before prices skyrocketed and goods vanished from the market. Such drastic shifts in price expectations are rare, but when they do occur, they can dominate the more stable consumption-income relationships that generally prevail.

Long-run Stability and Short-run Instability. \Vhere do all these considerations leave us on the determinants of consumer spending? Some modern statistical efforts to answer this question are presented in the following section. But at the risk of over-simplification, we can say that over the long pull the ratio of consumption to disposable personal income has been quite stable. At highemployment levels, it has seldom varied from the 94—96 per cent range, moving outside that range for long only in response to strong, identifiable special forces, such as war. In spite of short-run fluctuations, consumers as a whole appear to adjust their consumption habits to rising incomes over the long run so as to maintain about the same average ratio of consumption to disposable income as in past prosperity periods. Will this tendency persist over the years ahead? Nobody knows. But the historical relationship has prevailed long enough to make it a reasonably good bet.