Should you splurge on a restaurant meal or save money by eating at home? Should you buy a new car and, if so, how expensive a model? Should you redo that bathroom or live with it for another year? In the real world, households are constantly confronted with such choices—not just about the consumption mix but also about how much to spend in total. These choices, in turn, have a powerful effect on the economy: consumer spending normally accounts for two-thirds of total spending on final goods and services. In particular, as we’ve just seen, the decision about how much of an additional dollar in income to spend—the marginal propensity to consume—determines the size of the multiplier, which determines the ultimate effect on the economy of autonomous changes in spending.

But what determines how much consumers spend?

The most important factor affecting a family’s consumer spending is its current disposable income—income after taxes are paid and government transfers are received. It’s obvious from daily life that people with high disposable incomes on average drive more expensive cars, live in more expensive houses, and spend more on meals and clothing than people with lower disposable incomes. And the relationship between current disposable income and spending is clear in the data.

The Bureau of Labor Statistics (BLS) collects annual data on family income and spending. Families are grouped by levels of before-tax income, and after-tax income for each group is also reported. Since the income figures include transfers from the government, what the BLS calls a household’s after-tax income is equivalent to its current disposable income.

Figure 26-2 is a scatter diagram illustrating the relationship between household current disposable income and household consumer spending for American households by income group in 2013. For example, point A shows that in 2013 the middle fifth of the population had an average current disposable income of $45,826 and average spending of $42,495. The pattern of the dots slopes upward from left to right, making it clear that households with higher current disposable income had higher consumer spending.

It’s very useful to represent the relationship between an individual household’s current disposable income and its consumer spending with an equation. The consumption function is an equation showing how an individual household’s consumer spending varies with the household’s current disposable income. The simplest version of a consumption function is a linear equation:

where lowercase letters indicate variables measured for an individual household.

In this equation, c is individual household consumer spending and yd is individual household current disposable income. Recall that MPC, the marginal propensity to consume, is the amount by which consumer spending rises if current disposable income rises by $1. Finally, a is a constant term—individual household autonomous consumer spending, the amount of spending a household would do if it had zero disposable income. We assume that a is greater than zero because a household with zero disposable income is able to fund some consumption by borrowing or using its savings. Notice, by the way, that we’re using y for income. That’s standard practice in macroeconomics, even though income isn’t actually spelled “yncome.” The reason is that I is reserved for investment spending.

Recall that we expressed MPC as the ratio of a change in consumer spending to the change in current disposable income. We’ve rewritten it for an individual household as Equation 26-6:

Multiplying both sides of Equation 26-6 by Δyd, we get:

Equation 26-7 tells us that when yd goes up by $1, c goes up by MPC × $1.

Figure 26-3 shows what Equation 26-5 looks like graphically, plotting yd on the horizontal axis and c on the vertical axis. Individual household autonomous consumer spending, a, is the value of c when yd is zero—it is the vertical intercept of the consumption function, cf. MPC is the slope of the line, measured by rise over run. If current disposable income rises by Δyd, household consumer spending, c, rises by Δc. Since MPC is defined as Δc/Δyd, the slope of the consumption function is:

In reality, actual data never fit Equation 26-5 perfectly, but the fit can be pretty good. Figure 26-4 shows the data from Figure 26-2 again, together with a line drawn to fit the data as closely as possible. According to the data on households’ consumer spending and current disposable income, the best estimate of a is $19,343 and of MPC is 0.50. So the consumption function fitted to the data is:

c = $19,343 + 0.50 × yd

That is, the data suggest a marginal propensity to consume of approximately 0.50. This implies that the marginal propensity to save (MPS)—the amount of an additional $1 of disposable income that is saved—is approximately 0.50, and the multiplier is approximately 1/0.50 = 2.00.

It’s important to realize that Figure 26-4 shows a microeconomic relationship between the current disposable income of individual households and their spending on goods and services. However, macroeconomists assume that a similar relationship holds for the economy as a whole: that there is a relationship, called the aggregate consumption function, between aggregate current disposable income and aggregate consumer spending. We’ll assume that it has the same form as the household-level consumption function:

Here, C is aggregate consumer spending (called just “consumer spending”); YD is aggregate current disposable income (called, for simplicity, just “disposable income”); and A is aggregate autonomous consumer spending, the amount of consumer spending when YD equals zero. This is the relationship represented in Figure 26-5 by CF, analogous to cf in Figure 26-3.

The aggregate consumption function shows the relationship between disposable income and consumer spending for the economy as a whole, other things equal. When things other than disposable income change, the aggregate consumption function shifts. There are two principal causes of shifts of the aggregate consumption function: changes in expected future disposable income and changes in aggregate wealth.

Changes in Expected Future Disposable Income Suppose you land a really good, well-paying job on graduating from college in May—but the job, and the paychecks, won’t start until September. So your disposable income hasn’t risen yet. Even so, it’s likely that you will start spending more on final goods and services right away—maybe buying nicer work clothes than you originally planned—because you know that higher income is coming.

Conversely, suppose you have a good job but learn that the company is planning to downsize your division, raising the possibility that you may lose your job and have to take a lower-paying one somewhere else. Even though your disposable income hasn’t gone down yet, you might well cut back on spending even while still employed, to save for a rainy day.

Both of these examples show how expectations about future disposable income can affect consumer spending. The two panels of Figure 26-5, which plot disposable income against consumer spending, show how changes in expected future disposable income affect the aggregate consumption function. In both panels, CF₁ is the initial aggregate consumption function. Panel (a) shows the effect of good news: information that leads consumers to expect higher disposable income in the future than they did before.

Consumers will now spend more at any given level of current disposable income, YD, corresponding to an increase in A, aggregate autonomous consumer spending, from A₁ to A₂. The effect is to shift the aggregate consumption function up, from CF₁ to CF₂. Panel (b) shows the effect of bad news: information that leads consumers to expect lower disposable income in the future than they did before. Consumers will now spend less at any given level of current disposable income, YD, corresponding to a fall in A from A₁ to A₂. The effect is to shift the aggregate consumption function down, from CF₁ to CF₂.

In a famous 1956 book, A Theory of the Consumption Function, Milton Friedman showed that taking the effects of expected future income into account explains an otherwise puzzling fact about consumer behavior. If we look at consumer spending during any given year, we find that people with high current income save a larger fraction of their income than those with low current income. (This is obvious from the data in Figure 26-4: people in the highest income group spend considerably less than their income; those in the lowest income group spend more than their income.) You might think this implies that the overall savings rate will rise as the economy grows and average current incomes rise; in fact, however, this hasn’t happened.

Friedman pointed out that when we look at individual incomes in a given year, there are systematic differences between current and expected future income that create a positive relationship between current income and the savings rate. On one side, people with low current incomes are often having an unusually bad year. For example, they may be workers who have been laid off but will probably find new jobs eventually. They are people whose expected future income is higher than their current income, so it makes sense for them to have low or even negative savings. On the other side, people with high current incomes in a given year are often having an unusually good year. For example, they may have investments that happened to do extremely well. They are people whose expected future income is lower than their current income, so it makes sense for them to save most of their windfall.

When the economy grows, by contrast, current and expected future incomes rise together. Higher current income tends to lead to higher savings today, but higher expected future income tends to lead to less savings today. As a result, there’s a weaker relationship between current income and the savings rate.

Friedman argued that consumer spending ultimately depends mainly on the income people expect to have over the long term rather than on their current income. This argument is known as the permanent income hypothesis.

Changes in Aggregate Wealth Imagine two individuals, Maria and Mark, both of whom expect to earn $30,000 this year. Suppose, however, that they have different histories. Maria has been working steadily for the past 10 years, owns her own home, and has $200,000 in the bank. Mark is the same age as Maria, but he has been in and out of work, hasn’t managed to buy a house, and has very little in savings. In this case, Maria has something that Mark doesn’t have: wealth. Even though they have the same disposable income, other things equal, you’d expect Maria to spend more on consumption than Mark. That is, wealth has an effect on consumer spending.

The effect of wealth on spending is emphasized by an influential economic model of how consumers make choices about spending versus saving called the life-cycle hypothesis. According to this hypothesis, consumers plan their spending over a lifetime, not just in response to their current disposable income. As a result, people try to smooth their consumption over their lifetimes—they save some of their current disposable income during their years of peak earnings (typically occurring during a worker’s 40s and 50s) and during their retirement live off the wealth they accumulated while working. We won’t go into the details of this hypothesis but will simply point out that it implies an important role for wealth in determining consumer spending. For example, a middle-aged couple who have accumulated a lot of wealth—who have paid off the mortgage on their house and already own plenty of stocks and bonds—will, other things equal, spend more on goods and services than a couple who have the same current disposable income but still need to save for their retirement.

Because wealth affects household consumer spending, changes in wealth across the economy can shift the aggregate consumption function. A rise in aggregate wealth—say, because of a booming stock market—increases the vertical intercept A, aggregate autonomous consumer spending. This, in turn, shifts the aggregate consumption function up in the same way as does an expected increase in future disposable income. A decline in aggregate wealth—say, because of a fall in housing prices as occurred in 2008—reduces A and shifts the aggregate consumption function down.