The standard theory of the consumption function states that spending by an indi-vidual consumer unit is related to its expected income, the cost and availability of credit, and the stock of wealth held by that consumer. In addition, people generally spend more than 100% of their income at certain stages of their life cycle, notably when they are in college or just starting to work and after they retire, and save a substantial proportion of their income during most of their working years.
The difference between actual and expected income is critical in understand-ing the pattern of consumer spendunderstand-ing. Most consumer units have a reasonably firm expectation of the amount of income they will receive in the following year.
Of course, that expectation is not always correct. If income is decreased because of cyclical weakness and increased layoffs, but wage-earners believe they will soon be reemployed, consumption declines by less than the drop in income. If, on the other hand, wage-earners view the loss of job as a permanent development (such as the termination of relatively highly paid jobs in steel mills or coal mines), consumption might drop by the full amount of the decline in income.
Given this straightforward concept, it logically follows that in many cases, the income of consumers at the upper end of the income scale in any given year is well above their long-term expected income level, whereas the income of consumers at the lower end of the income scale in any given year is well below their long-term expected level. As a result, we would expect to find that the personal saving rate for those with very low incomes is negative, and the personal saving rate for those with very high incomes is quite large. That is precisely what the data show.
It was once thought such data meant that people with permanently low incomes dissaved, while those with permanently high incomes saved a large proportion of their income. However, that turns out not to be the case. People with low incomes in any given year dissave because their current income is well below their permanent income, not because their permanent income is low. In fact, most poor people could not dissave permanently because they cannot borrow very much and do not have very many assets. Most people who dissave are children, students, the elderly, or those who are temporarily unemployed.
Perhaps this seems fairly obvious, but that was not always the case. Following Keynes, most economists once thought the personal saving rate rose as income increased. That was one of the major points stressed in his General Theory. That assumption led to two untenable conclusions. One was that total GDP could be increased by transferring income from the ‘‘rich’’ to the ‘‘poor.’’ The other was that an increase in income was accompanied by an increase in saving that would not be invested, so the only way the economy could remain at full employment was to increase the proportion of government spending to GDP.
Presumably economists know better now. Yet for the past decade, Japanese economists have been recommending ever-increasing public works projects, but the economy remains mired in recession. Some of the rhetoric directed against the Bush tax cut of 2001 claimed that it would eliminate the surplus and ‘‘rob’’ the Social Security ‘‘lockbox,’’ but others claimed that giving the ‘‘rich’’ a tax cut would not stimulate the economy because they would not spend it. Old-style Keynesians also missed the distinction between the effect of temporary and permanent tax cuts. Hence vestiges of the old, outmoded theory are still found in public pol-icy debates. For this reason, this appendix offers a brief review of the historical development of the consumption function.
Before the Great Depression, most economists were not concerned with the con-cept of an aggregate consumption function. They thought consumers would either spend their income or save it, in which case it would be invested. Thus the criti-cal determinant of economic activity was how the interest rate balanced the scriti-cales between saving and investment, which in turn determined the long-run growth
rate of the economy. A few economists, known as the ‘‘underconsumptionists,’’
claimed that declines in economic activity occurred because consumers did not spend enough, but they were generally considered to be cranks.
All this changed when the worldwide economy fell into depression in the 1930s. Out of that maelstrom, Keynes developed his theory that as income rose, consumption also rose, but at a slower rate. Thus, he said, as income increased over time, the percentage gap between income and consumption would also widen. If that increased saving was not translated into investment, the economy would falter.
Only the government, said Keynes, could pick up the slack. As a result, govern-ment spending would have to rise as a proportion of GDP in order to pull the economy out of perpetual stagnation.
Except for the part about consumption being related to income, this simple Keynesian theory is wrong on all accounts, bearing as it does a great resemblance to the claims of the discredited underconsumptionists. For if this hypothesis were true:
• Consumption would not rise as fast as income over time, and the personal sav-ing rate would increase over time. Yet the savsav-ing rate shows no upward trend, and may have declined over the past two decades even when it is measured correctly.
• Consumption would be a function only of income, with financial variables play-ing an unimportant or negligible role in decisions to spend or save. Yet most short-term fluctuations in consumption are related to monetary factors.
• The saving rate would decline in recessions. Yet it invariably rises.
• Rich people would save a larger proportion of their income than poor people, because the saving rate rises as income rises. Yet in the long run, studies show that both rich and poor people save the same proportion of their average income.
The Time-Series Cross-Section Paradox
Figure 4.6 shows that consumption has been very highly correlated with income for the past 50 years; the key finding is there is no sign of a widening gap. Figure 4.7 shows that the gap between consumption and income varies cyclically over time.
Figure 4.8 shows that the saving rate has not only been constant over long periods of time, but has declined on balance since 1980, although we have already explained why that decline is overstated. Figure 4.9 shows that the saving rate generally rises rather than declines in recessions; the only exception occurred in 1982, when interest rates fell by a record amount.
At first, the data shown in figure 4.10 would seem to support the hypothesis that rich people save a larger proportion of their income than poor people. Yet that statement is inconsistent with the long-run constancy of the saving rate. Real per capita disposable income has risen fourfold since 1929, yet the saving rate has stayed about the same.
Disposable income
Consumption
0 2000 4000 6000
0 2000 4000 6000
Figure 4.6 Scatter diagram, consumption and disposable income, 1947–98
–200 –100 0 100 200 300
’50 ’55 ’60 ’65 ’70 ’75 ’80 ’85 ’90 ’95
Figure 4.7 Residuals from simple consumption function (correlated only with disposable income)
This is known as the time-series cross-section paradox: on the one hand, saving does not rise as income increases over time, while on the other hand, saving appears to rise as individual income increases. Time-series data refers to data over many years, such as from 1947 through 1998. Cross-section data refers to a snapshot of different entities taken at the same time, such as the spending and saving patterns of all consumers in July, 1989.
’50 ’55 ’60 ’65 ’70 ’75 ’80 ’85 ’90 ’95 0
2 4 6 8 10
Figure 4.8 Personal saving as a percentage of disposable income
Savrat 91 Savrat 82 Savrat 80
Savrat 74 Savrat 70 Savrat 61
Savrat 58 4
5 6 7 8 9 10
Figure 4.9 Figures are the personal saving rate for the year before the recession started, the year of recession, and the next 2 years; in all cases, the saving rate rose during the recession except for 1982, when interest rates fell by a record amount
The Permanent Income Hypothesis
Even a brief reflection should reveal what is wrong with accepting the data in figure 4.10 at face value. It shows that the lowest income group has a saving rate of about−80%. Think about it for a minute. How could some people go through life spending almost twice as much as they earn? The only possibility would be that such individuals were very wealthy and were gradually depleting their assets, but in fact those with incomes of under $5,000 per year probably have very few assets.
The actual reason the saving rate is negative for low-income consumers is that most people in that income category don’t usually earn less than $ 5,000 per year;
Average income in class ($000s)
Personal saving rate (%)
0
–100 –80 –60 –40 –20 20 40
0 20 40 60 80 100
Figure 4.10 Personal saving rate by income class, 1989
they usually earn much more. In that year, reported income is well below the long-term average or expected income. These people may be students; or are taking a year off from work for vacation, because of illness, or because of temporary unemployment; or are retired. Clearly, individuals who earn an average of less than $5,000 per year over their entire lifetime and have no substantial assets cannot spend more than $10,000 every year. Most people who spend more than they earn in any given year are experiencing a year when their income is well below normal.
Thus one of the key modifications of the simple consumption function theory states that consumption is based on expected or permanent income, not just income earned that year. That concept is known as the permanent income hypothesis (PIH).
When the PIH was first introduced, it was subject to bitter criticism by the economics profession. The idea given in italics above was developed by Milton Friedman and, as has so often been the case, most economists initially disagreed with him. The greatest controversy focused on the top end of the income scale: it was generally thought that people in the top 1% of the income scale (above $350,000 at 2002 levels) save a larger proportion of their income than, say, people making
$25,000 to $50,000 per year. And, it was argued, the constancy of the saving rate certainly is not true for someone making $5 or $10 million per year.
Friedman pointed out that just as most people in the lowest income classifications for any given year are earning far less than their average income, those in the very highest income classifications are earning far more than their average income.
Perhaps they received a large bonus, cashed in stock options after 10 or 20 years of appreciation, sold their business, or received some other type of one-time payment that probably would not be repeated. They didn’t spend anywhere near 95% of their income – but that was because their average income was well below what they received in that year.
What about people who regularly earn $1 million or more per year? How much they save depends on the individual person, but there are many stories of the rich who nonetheless ended up broke. When Donald Trump overextended him-self and had to be monitored by his bankers, he complained bitterly because his
‘‘allowance’’ was cut to $450,000 per month. Sometime major league baseball player Jack Clark filed for bankruptcy because he could not pay all his debts from a salary of $3.8 million per year. These may be extreme examples, but there are many stories of sports and entertainment stars who end up impecunious in their later years. The cliché ‘‘from shirtsleeves to shirtsleeves in three generations’’ bears tes-timony to the patriarch who built up a fortune, only to see it frittered away by his progeny.
Although controversial when it was first released, the permanent income hypoth-esis is now accepted as a cornerstone of the modern consumption function. One of the most important empirical tests was to treat the value of one’s house as a proxy for permanent income. The ratio of consumption/house value remains constant as income increases; there is no sign that those who live in expensive houses save a higher proportion of their permanent income. Other tests showed that the saving rate was invariant by professions: physicians, attorneys, and accountants did not save more of their income than blue-collar workers. These results are adjusted for volatility in income.
The principal concept of Friedman’s theory of the consumption function is that consumers base their spending plans on some measure of expected or average income. Suppose, he said, someone got paid once a week. Does that mean they would eat a lot that day and nothing the rest of the week? Of course not. People base their spending plans on what they expect their average income will be. If their income is larger than usual in some week (or month, or year) they will save a larger proportion; if it is smaller than usual, they will save a smaller proportion.
This basic idea has evolved into the modern theory of the consumption function, which says consumers maximize their utility by taking into account all known information about their current income, its expected value in the future, the amount of assets they have, the rate of return on those assets, the amount they can borrow, and the rate at which they can borrow.
The PIH also states that if wealth rises, consumption rises for any given level of income. If interest rates rise, consumption declines. Monetary factors have always played an important role in the PIH; the issue is how to incorporate them empirically.
Other factors that may affect consumption, according to Friedman, are the volatil-ity of actual income received, age of the consumer unit, and size of that unit. The change in consumption for any given change in income is likely to be smaller for someone with a highly volatile income – such as a farmer, independent business owner, or financial market trader – than for someone who receives steady wages or dividends. Friedman also found that large families spend a higher proportion of their income than smaller family units, although this factor is negligible at the macroeconomic level.
The key hypothesis of the PIH states that the ratio of permanent consumption to permanent income is independent of the level of income. Rich people do not save a larger proportion of their income than poor people.
From the viewpoint of political ramifications, the PIH stands as an effective rebuttal to Keynes. Most economists no longer think that as income increases, consumption does not rise as fast, so the government must fill the gap by spend-ing more. Nor do they believe that aggregate consumption can be increased by taxing the rich more and giving it to the poor.13
Keynes said that a balanced budget generated by high spending and taxes was better for the economy than a balanced budget generated by low spending and taxes. In fact, Trygve Haavelmo, one of Keynes’s followers, claimed that a $1 bil-lion increase in both spending and taxes would boost real GDP by $1 bilbil-lion.
Friedman showed that was also incorrect, and today very few economists believe that statement.14
Some economists have claimed that Keynes, being an essential pragmatist and always interested in ‘‘what works,’’ would certainly have modified his theories in view of the post-WWII full employment era; also, that these comments fail to take into account the important perspective of the time, when Keynes was trying to save capitalism from the twin evils of fascism and communism. Occasionally one reads a comment to the effect that ‘‘Keynes was not a Keynesian,’’ meaning that he would have modified those views appropriately. Perhaps that would have been the case. However, to the extent that vestiges of the original Keynesian doctrine on the consumption function still enter the debate about the proper role of fiscal policy, a brief airing of these views seems appropriate here.
The Life Cycle Hypothesis
One popular extension of the PIH, developed by Franco Modigliani, is known as the life cycle hypothesis (LCH). The precise exposition has been changed over the years, but the general idea states that consumption depends on the total resources available to the consumer over his lifetime. That would include the present value of all current and expected future labor earnings, existing net worth (also known as wealth), the rate of return on capital, and the age of the consumer. Modigliani also places greater emphasis on expected income, as opposed to average past income, besides giving greater weight to the role of wealth relative to income.
According to the LCH, consumers plan their consumption over their lifetime.
They dissave when first starting a family, gradually save more as retirement age approaches, and then dissave during retirement. As Modigliani says, ‘‘The cor-nerstone of the model is the notion that the purpose of saving is to enable the household to redistribute the resources it gets (and expects to get) over its life cycle in order to secure the most desirable pattern of consumption over life.’’
Today, almost all economists incorporate elements of the PIH and the LCH into their theories of the consumption function. Other recent developments have
shown (a) the importance of the liquidity constraint when the monetary authori-ties decrease the amount of credit, and (b) the increasing important of confidence or sentiment variables. We have shown how both of these factors are empirically relevant.
Notes
1. The first economist to advance the theory that short-term fluctuations in consumption are random was Robert Hall, in his pathbreaking article, ‘‘Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence,’’ Journal of Political Economy, 86 (December 1978). Hall argues that the best forecast of a family’s consumption next period is the amount of consumption this period.
2. This concept, known as excess sensitivity, was first developed by Hall’s student, Marjorie Flavin. See her article ‘‘The Adjustment of Consumption to Changing Expectations About Future Income,’’ Journal of Political Economy, 89 (October 1981).
3. Milton Friedman, A Theory of the Consumption Function (Princeton: Princeton University Press for NBER), 1957.
4. Bush Administration economists claim that the sluggish growth in 2002 reflected a mas-sive decline in high-tech investment and a major decline in exports due to the overvalued dollar; if it were not for the tax cut, the recession would have continued and become much more serious. The unanswered question is whether the stock market would have rebounded if the Federal budget had not slipped back into a substantial deficit.
5. Two standard references are John Y. Campbell and N. Gregory Mankiw, ‘‘Consumption, Income and Interest Rates: Reinterpreting the Time Series Evidence,’’ in O. Blanchard and S. Fischer, eds, NBER Macroeconomics Annual (Cambridge, MA: MIT Press, 1989), and Robert E. Hall and Frederic S. Mishkin, ‘‘The Sensitivity of Consumption to Transitory Income Estimates from Panel Data on Households,’’ Econometrica, 1982.
5. Two standard references are John Y. Campbell and N. Gregory Mankiw, ‘‘Consumption, Income and Interest Rates: Reinterpreting the Time Series Evidence,’’ in O. Blanchard and S. Fischer, eds, NBER Macroeconomics Annual (Cambridge, MA: MIT Press, 1989), and Robert E. Hall and Frederic S. Mishkin, ‘‘The Sensitivity of Consumption to Transitory Income Estimates from Panel Data on Households,’’ Econometrica, 1982.