Reaganomics essay

20

The breakdown

of the

Keynesian model

PAUL CRAIG ROBERTS

Tru_ is much talk these days

about "the crisis in Keynesian economics." That some such crisis

exists is evident from the bewilderment and impotence our econom-

ic policy makers are displaying in their confrontation with economic

reality. But what exactly is the nature of this crisis? What went

wrong and what can put it right?

The answer, I would suggest, is almost embarrassingly simple.

Today in the United States, public economic policy is formulated in

bland disregard of the human incentives upon which the economy

relies. Instead it is based on the Keynesian assumption that the gross

national product (GNP) and employment are determined only by

the level of aggregate demand or total spending in the economy.

Unemployment and low rates of economic growth are seen as evi- dence of insufficient spending. The standard remedy is for govern-

ment to increase total spending by incurring a deficit in its budget. GNP, it is believed, will then rise by some multiple of the increase

in spending. Keynesian economics focuses on estimating the "spend-

ing gap" and the "multiplier" so that the necessary deficit can be calculated.

This view of economic policy is enshrined in the large-scale

econometric forecasting models upon which both Congress and the

THE BREAKDOWN OF THE KEYNESIAN MODEL 21

Executive Branch rely for simulations of economic policy alterna-

tives. It is a view that is extraordinary in its emphasis on spending.

True, it is obvious that if people did not buy, no one would produce

for market. It also seems obvious that the more people buy, the more will be produced and, therefore, that the use of government

fiscal policy to increase total demand will increase total production or GNP. All this is so obvious to Keynesians that they believe any

fiscal policy that produces an increase in government spending, even

a spending increase matched by a tax increase, will produce an in- crease in GNP.

The concept of the "balanced-budget multiplier" illustrates the

primacy that Keynesians give to spending as the determinant of

production. According to this concept, government can increase total

spending and, thereby, GNP by raising taxes and spending the rev-

enues. The reasoning is as follows. People do not pay the higher

taxes only by reducing their spending (consumption); they also re- duce their savings. Therefore, when taxes are raised, the decrease

in private spending is less than the increase in government spend- ing. Conversely, a cut in tax rates, matched by a decrease in govern-

ment spending, would result in a reduction in total spending (i.e.,

saving would increase), a fall in GNP, and a rise in unemployment. For years after the 1964 Presidential election, college students

were asked a standard question on economic exams: What would

happen if Barry Goldwater's prescription for a tax cut, matched by

a spending cut, were implemented? They missed the answer if they

did not reply that there would be a reduction in aggregate demand

and, therefore, a fall in GNP and employment. Alas, for too many policy makers that is still the answer.

Since the "balanced-budget multiplier" implies that the greater

the increase in taxes and in government spending, the greater the

increase in GNP, it is a wonder no one ever asked what happens to

production as tax rates rise. This question confronts economic policy with the incentive effects it has disregarded. It should be obvious

even to Keynesians that when marginal tax rates are high, people will prefer additional leisure to additional current income, and ad-

ditional current consumption to additional future income. As work

effort and investment decline, production will fall, regardless of how

great an increase there might be in aggregate demand. Such a rec-

ognition of disincentives implies a recognition of incentives, and

Keynesians are gradually having to rethink the answer to their

standard question about Barry Goldwater. Once one recognizes that

people produce and invest for income, and that income depends on

22 THE PUBLIC INTEREST

tax rates, one has reached the realization that fiscal policy causes

changes not iust in demand but also in supply.

The economics of supply

The economics of spending has thoroughly neglected the eco-

nornics of supply. On the supply side there are two important rela-

tive prices governing production. One price determines the choice between additional current income and leisure; the other determines

the choice between additional future income (investment) and

current consumption. Both prices are affected by the marginal tax

rates. The higher the tax rates on earnings, the lower the cost of

leisure and current consumption, in terms of foregone after-tax income.

As an illustration, consider the decision to produce. There are two

uses of time-work and leisure. Each use has a price relative to the

other. The price of additional leisure is the amount of income fore- gone by not working, and it is influenced by the tax rates. The higher

the tax rates, the smaller the amount of after-tax income foregone

by enjoying additional leisure. In other words, the higher the tax

rates, the lower the relative price of leisure. When the marginal tax

rate reaches 100 percent, the relative price of additional leisure be-

comes zero. At that point, additional leisure becomes a free good, because nothing has to be sacrificed in order to acquire it.

We often hear that a person works the first five months of the year for the government, and then starts working for himself. But that is

not the way it goes. The first part of the year, he works for himself;

he only begins working for the government when his income reaches taxable levels. The more he earns, the more he works for the govern-

ment, until rising marginal rates discourage him from further work.

Take the case of a physician who encounters the 50-percent rate

after six, eight, or 10 months of work. He is faced with working

another six, four, or two months for only 50 percent of his earnings. Such a low after-tax return on their efforts encourages doctors to

share practices, to reduce their working hours, and to take longer

vacations. The high tax rates thus shrink the tax base by discourag-

ing them from earning additional amounts of taxable income. They

also drive up the cost of medical care by reducing the supply of

medical services. A tax-rate reduction would raise the relative price of leisure and result in more taxable income earned and also in a

greater supply of medical services. The effect of tax rates on the decision to earn additional taxable

THE BREAKDOWN OF THE KEYNESIAN MODEL 23

income is not limited to physicians or to the top tax bracket; it

operates across the spectrum of tax brackets. Studies by Martin

Feldstein show that the tax rates on the average worker practically

eliminate the gap between his after-tax take-home pay and the level

of untaxed unemployment compensation he could be receiving if he

did not work. In this case, a marginal tax rate of 30 percent (includ-

ing state and Federal income taxes and Social Security taxes) re-

duces the relative price of leisure so much that, by making unem-

ployment competitive with work, it has raised the measured rate of

unemployment by 1.25 percent and shrunk GNP and the tax base

by the lost production of one million workers.

It is useful to give another example to illustrate that it is not just

the top marginal rate that causes losses to GNP, employment, and

tax revenues by discouraging people from earning additional tax-

able income. Blue-collar workers do not yet encounter the top mar-

ginal tax rate (although if inflation continues to push up money incomes, and the tax-rate structure remains unadjusted for inflation,

it will not be many years before they do). Nevertheless, the marginal

tax rates that many blue-collar workers already face are high enough

to discourage them from earning additional taxable income. Take

the case of a carpenter facing only a 25-percent marginal tax rate. For every additional $100 he earns before income tax, he gets to

keep $75. Suppose that his house needs painting and that he can

hire a painter for $80 a day and hire himself out for $100 a day.

However, since his after-tax earnings are only $75, he saves $5 by

painting his own house, so it pays him to choose not to earn the additional $100. In this case, the tax base shrinks by $180-of which

$100 is the foregone earnings of the carpenter, and $80 is the lost

earnings of the painter who is not hired. (Also, the productive ef- ficiency associated with the division of labor vanishes. )

Suppose, instead, that the marginal tax rate on additional earnings

by the carpenter were reduced to 15 percent. In this case, his after-

tax earnings would be $85, and it would pay him to hire the painter.

The reduction in the marginal tax rate would thus expand the tax

base upon which revenues are collected by $180.

Studies by Gary Becker have made it clear that capital and labor

are employed by households to produce utility through non-market

activities (e.g., a carpenter painting his own house). Utility pro-

duced in this way is not purchased with income subject to taxation.

Therefore, the amount of household-owned capital and labor sup-

plied in the market will be influenced by marginal tax rates. The

lower the after-tax income earned by supplying additional labor and

24 THE pUBLIC ENTEREST

capital in the market, the less the utility that the additional income

can provide, and the more likely it is that households can increase their utility by allocating their productive resources to non-market

activities. A clear implication of the new household economics is

that the amount of labor and capital supplied in the market is in-

fluenced by the marginal tax rates.

Now consider how relative prices affect the choice concerning the use of income. There are two uses of income, consumption and

saving (investment), and each has a price in terms of the other.

The price of additional current consumption is the amount of future

income foregone by enjoying additional current consumption. The

higher the tax rates, the smaller the amount of after-tax future in-

come foregone by enjoying additional current consumption. In other

words, the higher the tax rates, the lower the relative price of cur-

rent consumption. Take the case of an Englishman facing the 98-percent marginal

tax rate on investment income. He has the choice of saving $50,000

at a 17-percent rate of return, which would bring him $8,500 per year before taxes, or purchasing a Rolls Royce. Since the after-tax value of that $8,500 additional income is only $170 per year, the

price of additional consumption is very low: He can enjoy having a fine motor car by giving up only $170 per year of additional in-

come. This is why so many Rolls Royces are seen in England today.

They are mistaken for signs of prosperity, whereas in fact they are

signs of high tax rates on investment income.

A tax-rate reduction would raise the price of current consumption relative to future income, and thus result in more savings, making

possible a growth in real investment. A rate reduction not only in-

creases disposable income and total spending, it also changes the

composition of total spending toward more investment. Thus, labor

productivity, employment, and real GNP are raised above the levels that would result from the same amount of total spending more

heavily weighted toward current consumption.

Tax cuts and rebates

The econometric models upon which the government relies for

simulations of policy alternatives do not take into account these sup-

ply-side effects on GNP of these relative price changes. Consider the alternatives faced by the Keynesian policy maker who wants

"to get the economy moving again. _ His goal is to inerease aggregate

demand or total spending. How can he do this? He has the choice

THE BREAKDOWN OF THE KEYNESIAN MODEL 25

between the balanced-budget multiplier (i.e., increasing both taxes

and government spending) or a deficit. He will discard the balanced-

budget multiplier, because it is relatively weak and deficits are more

politically acceptable than legislating higher tax rates. Having set-

tled on a deficit, he has to choose how to produce it. He can hold

tax revenues constant and increase government spending, or he can

hold government spending constant and cut tax revenues. In the

latter case, he has a choice between rebates and permanent reduc- tions in tax rates. Wanting the most stimulus for his deficit dollar,

he will ask for econometric simulations of his three policy alterna- tives: a tax rebate, a tax rate reduction, or an increase in government

spending programs.

The simulations, all based on Keynesian assumptions, will show that a revenue reduction of a given amount, whether in the form

of a rebate of personal income taxes or a reduction in personal-

income-tax rates, will raise disposable income--and thereby spending

and GNP-by the same amount. The policy maker may prefer the

rebate for reasons of "'flexibility.'" The spending stimulus may not

be required in the following year, and, if it is, he has the option

of providing it either by another rebate or by an increase in govern-

ment spending programs. But on the basis of the econometric simula- tion, he will be indifferent as to the choice between rebates or rate

reductions. As for his third option, an increase in government spend-

ing programs, the simulation may report that, dollar for dollar, au

increase in government purchases (as contrasted with transfers) will

have a more powerful impact on GNP because the government spends all of the money, whereas if it is returned to consumers they

will save part of it. Based on the econometric simulation of his al-

ternatives, he will conclude that there is no compelling economic reason in favor of any of the three, and he will make his choice on

a political basis.

But the econometric models have misled the policy maker. Unlike a reduction in personal-income-tax rates, a rebate affects no indi-

vidual choice at the margin. It does not change the relative prices governing the choices between additional current income and leisure

or between additional future income and current consumption. It

does not raise the relative prices of leisure and current consumption. Therefore, a rebate directly stimulates neither work nor investment.

For any given revenue reduction, a rebate cannot cause as great an increase in GNP as a rate reduction, because it does not affect the

choices that would cause people to allocate more time and more

income to increasing production for the market.

26 THE PUBLIC INTEREST

An increase in government spending fares no better by comparison, and may fare even worse. It too fails to raise the after-tax rewards for

work and investment. Furthermore, it increases the percentage of

total resources used in the government sector. If the government

sector uses resources less efficiently than the private sector, as seems to be the case, the result is a decline in the efficiency with which resources are used-which means GNP would be less than it other-

wise would be. Yet the econometric simulations of the policy maker's

alternatives will pick up none of the incentive and disincentive ef-

fects of these relative price changes. Instead, they focus on the

effects of these alternatives on disposable income and on spending.

There are a number of adverse consequences of this extraordinary

preoccupation with spending. One is that the models exaggerate the

net tax-revenue losses that result from cutting tax rates. The only

"feedback effect" on the tax base and tax revenues that they provide

for is the expansion of GNP in response to an increase in demand.

They do not provide for the expansion in GNP that results from

higher after-tax rewards for work and investment. The supply-side

"feedback effects" are ignored. Similarly, revenue gains from tax- rate increases will be overestimated, because the disincentive effects are left out.

A second consequence follows from the popular misidentification

of a tax rebate as a tax cut, and from a similar tendency on the part

of most policy makers to see rebates and rate cuts as variations of

the same policy instrument. If Milton Friedman is correct that per-

sonal consumption is a function of permanent income, a temporary

rebate has little impact even on spending. Thus, on the basis of ex-

perience with rebates, tax cuts per se might come to be seen as

relatively ineffectual, leaving the field open to proponents of govern-

ment spending programs.

A third consequence is that the true effects of large tax increases

(such as the proposed energy taxes, or the $227-billion increase in

the Social Security tax over the next decade) will not be accurately calculated. Policy makers see these tax increases as withdrawals

from disposable income and spending, and their only concern is "to

put money back" into spending so that aggregate demand does not fall. However, these tax increases change the relative prices and in-

centives of leisure and work, consumption and investment. They

produce resource reallocations that have adverse implications for ,,

employment and the rate of economic growth. Yet the econometric

models, as now constructed, flash no warning lights. Consider what Arthur Laffer, in the Wall Street 1ournal, has called

THE BREAKDO_/VN OF THE KEYNESIAN MODEL 27

the "tax wedge." The Social Security tax increase provides a good

example of this phenomenon. It is a tax on employment, and, as

economists should know, a tax on employment will reduce employ-

ment. The employer's decision to hire is based on the gross cost to

him of an employee. The employee's decision to work is based on

his after-tax pay. We know that the higher the price, the less the

quantity demanded, and the lower the price, the less the quantity

supplied. The Social Security tax both raises the price to the de-

mander and lowers it to the supplier. By increasing the Social Se-

curity tax, policy makers reduced both job opportunities and the

inclination to work? They raised the cost of labor relative to capital

for the employer, and they narrowed the gap between unemploy-

ment compensation and after-tax take-home pay for a wider range

of workers. Since the revenues available for paying Social Security

benefits depend on both the tax rates and the number of people paying into the system, the increase in rates will be offset to some

degree by a decrease in the number of people paying into the sys-

tem. It is hard to see how the Social Security system can be saved

by decreasing employment, or how increasing the demand for un-

employment compensation is likely to free general revenues for So- cial Security benefits.

"Crowding out" investment

There are at least two other important points on which economic policy is misinformed by the neglect of incentives and of choices

made at the margin. One is the impact on GNP of reductions in the

corporate-income-tax rate, and the other is the controversy over

whether government fiscal policy "crowds out" private investment. Simulations run by the Congressional Budget Office and the House

Budget Committee on two of the three large-scale commercial econ- ometric models show declines in GNP as a result of reductions in

corporate-tax rates. In one of the models, corporate investment did

not depend on after-tax profits in a very strong way, but was very sensitive to changes in interest rates. Since interest rates rise as the

Treasury increases its borrowing to finance the deficit resulting from

a Theoretically, the effect on work effort depends on the present value of the Social Security benefits and taxes. If the increased tax means increased future benefits, the employee's work decision will take into account his increased future income, as well as his reduced current income. However, the recent changes in the Social Security law raised taxes and reduced benefits as a proportion of pay before retirement. As the Wall Street/ournal put it, "the extra money will go to pay people now or soon to be on the retirement rolls, not to finance your own high living in the 21st century" (February 6, 1978).

28 THE PUBLIC INTIERIgST

the tax cut, investment falls, and the model predicted a _cli_ in

GNP as the result of a tax-rate reduction that increased the profit-

ability of investment. 2

The other model predicted that a corporate-tax-rate reduction

would slightly raise real GNP after a lag of a couple of quarters, but

it predicted a lower nominal GNP for two years. Nominal GNP de-

clined because the corporate-tax-rate reduction reduced the user

cost of capital, the price mark-up, and thereby the inflation rate,

thus lowering the nominal price level.

To the extent that Keynesians think about the "crowding out" of

private investment by fiscal policy, it is in terms of upward pressure on interest rates as a result of government borrowing to finance

budget deficits. They do not realize that investment is crowded out

by taxation, regardless of whether the budget is in balance. To un-

derstand how, consider the following example. Suppose that a 10-

percent rate of return must be earned ff an investment is to be un- dertaken. In the event that government imposes a 50-percent tax

rate on investment income, investments earning 10 percent will no

longer be undertaken. Only investments earning 20 percent before

tax will return 10 percent after tax. Taxation crowds out investment

by reducing the number of profitable investments. When tax rates are

reduced, after-tax rates of return rise, and the number of profitable investments increases.

So "crowding out" cannot be correctly analyzed merely in terms

of events in the financial markets: "Crowding out" occurs in terms

of real output. It is the preempting of production capacity by gov-

ernment outlays, regardless of whether these outlays are financed

by taxing, borrowing, or money creation.

Responding to incentives

A concern with the supply-side effects of fiscal policy is incom-

patible with the concept of eeonomie policy that currently reigns in the Congress and in the Executive Branch. Members of the House

Budget Committee asked Alice Rivlin, Director of the Congressional

Budget Office, and Bert Lance, then Director of the Office of Man-

agement and Budget, about the neglect of the incentive effects of

tax-rate changes on supply and also about the econometric predic-

2According to staff in the Office of Management and Budget, there have recent- ly been changes in the model, but one can still get the perverse result because a reduction in the tax rate directly and substantially reduces multi-unit housing starts.

THE BREAKDOWN OF THE KEYNESIAN MODEL 29

tions that GNP would fall in response to a reduction in corporate tax rates.

Dr. Rivlin said that she and her staff had been "particularly

troubled" by model findings that GNP declines if corporate tax rates

are reduced. However, she went on to say:

Studies have generally found that tax-rate changes are less important than changes in the cost of capital and changes in levels of national out- put in influencing the level of investment. It follows that an investment tax credit or liberalized depreciation will increase investment more than a corporate-tax-rate reduction of equivalent revenue loss. While we do not believe that corporate-tax-rate cuts reduce investment, it would not be surprising to find that tax cuts had only a minor expansionary effect.

The OMB staff reply to this question was ambiguous.

Both CBO and OMB realized that the question about incentive

effects most fundamentally challenged their concept of economic

policy. The comments of Rivlin, Lance, and the OMB staff all un-

equivocally acknowledged that the econometric models upon which

they rely for guidance in the choice of economic policy alternatives

do not include any relative price effects of changes in personal-in-

come-tax rates. However, since they believe that the performance

of the economy is a function of spending levels, not of production

incentives, they expressed no concern over this neglect. They said

that economic theory and empirical studies leave it unclear whether

the neglected supply-side effects are important; regardless of how

the issue is resolved, they questioned the practical importance of

supply incentives for short-run policy analysis.

There are two parts to this argument. One is that it is unclear

whether lowering personal-income-tax rates will increase or reduce work effort. The other is that it is unclear whether any incentive

effects on work effort and investment would show up as quantita-

tively important in a short-run policy framework. The first proposi- tion questions the existence of the incentive effects; the second ques-

tions whether they would be effective in time to deal with an im-

mediate problem of economic stabilization.

It is easy to dispose of the latter point. The long-run consists of

a series of short-runs. If policies that are effective over a longer

period are neglected because they do not have an immediate impact,

and if policies that are damaging over the longer period are adopted

because they initially have beneficial results, then policy makers

will inevitably come to experience, sometime in the future, a period

when they will have no solution for the crisis they have provoked In the United States, that future might be now.

30 THE PUBLIC INTEREST

As for the first point, Rivlin acknowledged that a personal-income-

tax-rate reduction raises the relative price of leisure, and that work

effort will increase as people substitute income for leisure. This is known in economics as the "substitution effect," and it works to in-

crease supply. However, Rivlin also said:

It is also theoretically arguable that when a tax cut provides people with more after-tax income, many of them will reduce effort through what is called the income effect. For most people, leisure has some positive value, and it may even be a "luxury" good; these people could respond to a tax reduction by reducing their working hours, benefiting from more leisure time and still maintaining their after-tax income. For other people who like their work, there may be little or no labor supply response to the income or the substitution effect. In much of the United States economy, work weeks are fixed, leaving little pos- sibility for individuals to make marginal adjustments in hours of work.

In other words, CBO believes that the "income effect" works to de-

crease supply.

Rivlin then went on to say that it was an empirical question whether the "income effect" offset the "substitution effect," referred

to a narrow range of studies that left the question unresolved, and

concluded: "In the range of policy options that we have been deal-

ing with, I think the assumption that changes in marginal tax rates

have no quantitatively significant effect on labor supply is quite

plausible."

But the concept of a targeted or desired level of income unaf-

fected by the cost of acquiring such income is foreign to the price-

theoretical perspective of economic science. Rivlin's idea that people

respond to a cut in income-tax rates by maintaining their existing income levels while enjoying more leisure implies that, if their tax

rates went up, they would work harder in order to maintain their desired income level. Lester Thurow has actually employed this

reasoning to argue for a wealth tax. According to Thurow, a wealth

tax is a costless way to raise revenues because the "income effect" runs counter to and dominates the "substitution effect." He assumes

that people have a targeted level of wealth, irrespective of the cost

of acquiring it. Therefore, he says, a tax on wealth will cause people to work harder in order to maintain, after tax, their desired wealth level.

Note the perverse ways in which people respond to incentives

and disincentives according to the Rivlin-Thurow line of argument:

When tax rates go down and the relative price of leisure rises, peo-

ple demand more leisure; when tax rates go up and the relative price

THE BREAKDOWN OF THE KEYNESIAN MODEL 31

of leisure falls, people demand less leisure. In economics, any time the "income effect" works counter to the "substitution effect," we

have the relatively rare case of what is called an "inferior good"

(i.e., people purchase less of it as their income rises). Since income

is command over all goods, Rivlin's argument implies that all goods

are inferior goods: A tax cut will cause people to purchase only

more leisure, not more income (i.e., goods). What kind of people are these? Well, the only kind of people who fit this kind of eco-

nomic analysis are people who respond to a monetary incentive in perverse ways.

Perhaps Rivlin merely meant to say that lower tax rates would

allow people to have a little more income for a little less work. Even so, as long as she maintains that the "income effect" works counter

to the "substitution effect," her argument carries the implication that goods in general are inferior.

A perverse logic

Whatever the weight one assigns this point, there is a more fun-

damental defect in her argument. Notice the stunning inconsistency:

People respond to a tax-rate reduction "by reducing their working hours.., and still maintaining their after-tax income." But it is im-

possible for people in the aggregate to reduce their work effort and

maintain the same level of aggregate real income! If people respond to tax cuts by working less, real GNP would fall, and it would be

impossible to increase real disposable income, spending, and demand

in the aggregate. Rivlin's argument is directed against the effec-

tiveness of incentives in raising aggregate output, but if she were

correct, it would mean that Keynesian fiscal policy also is ineffective!

The fatal error in the Rivlin-Thurow argument can be put this way: It derives from trying to aggregate a series of partial equilib-

rium analyses (individual responses to a change in relative prices) and, in the aggregate, ignoring the general equilibrium effects.

There are various ways a non-economist can grasp this point. As- sume that the government cuts taxes and maintains a balanced

budget by reducing spending. In this case, the higher income ac-

corded the taxpayers whose rates are reduced must be matched by

a negative impact on the incomes of recipients of government spend- ing. Some or all of these may be the same people. Assume, for

example, that both the tax burden and government spending are evenly distributed. In this case the "income effect" (the substitution

of leisure for work) "nets out" for each individual. Since the ag-

32 _ PUBLIC ][I_rEREST

gregate income effect is zero, it cannot offset the "substitution effect"

(the substitution of work for leisure).

If taxes are cut and government spending is unchanged (resulting

in a budget deficit), the nominal disposable income of taxpayers as

a group will rise relative to the nominal disposable income of the

recipients of government spending as a group. The former will be able to bid real resources away from the latter. The real income gains

of the former will be matched by the real income losses of the lat-

ter. Since the bidding will raise prices, the real income loss might

be suffered by individuals who hold money. Regardless of who loses

and who gains, the individual income effects "net out," leaving only

the "substitution effects," which unambiguously increase work ef- fort.

There can be no aggregate "income effect" unless the impact of

incentives is to raise real aggregate income. Economic theory makes

it perfectly clear that a tax-rate reduction will increase work effort

and total output. In the final analysis, Rivlin's argument is not that the supply-side

incentive effects are unimportant, but the equally false argument

that their impact is perverse-that is, only a tax-rate increase can

produce a rise in real national income! She may not actually believe any such thing, of course-but that is where her reasoning leads her.

From economics to politics

An economist might see the flaw in the Rivlin-Thurow argument,

but it is not obvious to politicians. Take something simple, like Rivlin's assertion that a fixed work-week precludes adjustment of the

labor supply to tax-rate changes. To an economist her assertion is obviously false, but to the politician it sounds reasonable enough. He will not realize that the "adjustments" will be reflected in absen-

teeism rates, turnover rates, the average duration of unemployment,

labor negotiations for shorter work-weeks and more paid vacation

rather than higher wages, and in the quality and intensity of work.

Nor will he think of the entrepreneur who, because of high tax rates, loses his incentive to innovate-to make the economy itself (all of

us) more productive. Besides, one has to have an idealistic view of government to be-

lieve that politicians even want to know. The Keynesian concept of

the economy is that of an unstable private sector that must be

stabilized by fiscal and monetary policies of the government. This

view has served as a ramp for the expansion of the interests of gov-

THE BREAKDOWN OF THE KEYNESIAN MODEL 33

ernment. It has also served the interests of economists by transform-

ing them from ivory-tower denizens to public-spirited social activists,

a transformation which has much increased their power and en-

livened their life styles. Unemployment can always be said to be

too high. And the rate of economic growth can always be found to

be below "potential." This means that there is always a "scientific"

economic reason for expanding government spending programs that

enlarge the constituencies of the Congress and of the Federal bu-

reaucracy. From the standpoint of the private interests of policy makers, Keynesian economic policy will always be iudged a success.

To write about all of the problems of econometrics and economic

policy would require a book, not an article, but one other important problem must be mentioned in closing. Professor Robert Lucas has demonstrated that the standard econometric models assume that the

structure of the economy remains invariant under wide variations in

policy paths. What this means is that the models assume that people

do not learn. But people do learn, and their expectations change as

they experience various policies: They may not repeat the same be-

havior in response to the same policy at different times. Therefore,

the policy simulation may always misinform the policy makers. This

is not an optimistic note on which to end an article about public policy in a country that believes we need a great deal of it. But our

faith in public policy has exceeded our knowledge, and we will find

out that, in this area, there is no such thing as free faith.