Skip Navigation

Who Pays and When? An Assessment of Generational Accounting November 1995

To do so, the accounts start from the premise that all government purchases must be paid for--either at the time with taxes, or later by retiring debt or paying interest. Therefore, the net taxes that current generations pay under a given tax and spending policy will determine the net taxes of future generations.

Finding the implications of a policy for people of different ages requires first estimating how that policy relates their taxes and transfers to their age. Given such relationships, the accounts extend official economic, budget, and population projections to estimate the net taxes of all current generations for the rest of their lives. Those net taxes will determine the bill that future generations will have to pay, given what government will ever buy under that policy.

Thus, the accounts pose a hypothetical question: if a given policy applies to all current generations for the rest of their lives, what would that imply for the net taxes of current and future generations? The answer to that question does not predict actual policy, but is an abstract indicator of how today's policy would distribute resources among generations.
 
 

Forming the Basis of  Generational Accounts

Two standard ideas form the foundation of generational accounts: "present value" compares payments at different times on the same economic basis, and the "zero-sum constraint" enforces government solvency in the long run.

Present Value

Present value puts the prospective net taxes of the average person of every age on the same basis--one payment at one time. It is the net amount that an individual is willing to pay at that time, then never again pay taxes or receive transfers.

A discount (interest) rate is used to calculate present value. For instance, if the interest rate is 5 percent, this year's $100 will grow to $105 next year. Hence, $100 is this year's present value of $105 next year; alternatively, $105 is next year's present value of $100 this year. Other things being equal, present value gives less absolute weight to a prospective dollar if:

• The discount rate is higher (because a smaller amount can grow to a dollar in a given time when it compounds at a higher rate), or
•  The payment is later (because a smaller amount can grow to a dollar at a given rate when it compounds for a longer time).

The zero-sum constraint says there is no free lunch; someone, sometime, must pay for all that government ever spends. That is, the present value of prospective net taxes of all current and future generations must match today's net government debt (liabilities less assets), plus the present value of all prospective government purchases. Purchases that past and current generations do not pay for, future generations must, and with interest.

The zero-sum constraint ensures that government debt cannot forever grow faster than output. Without the constraint, mounting interest costs could swell the debt beyond control and bring on default, either direct or by inflation. (Of course, the constraint is satisfied even if government defaults--then the bondholders pay.)

The zero-sum constraint does not specify that government must ever retire any of its debt or can borrow no more; only that it cannot borrow forever to pay interest. If it could, the bill for a deficit would never come due; each generation could pass the bill to its children, who could pass it to its children, and so on. If government cannot borrow forever to pay interest, it must raise taxes or reduce spending at some time, either to retire debt or pay interest forever--choices that are equivalent in present value.

Some conditions may allow the bill to be passed on forever, but they do not prevail now. It may be feasible to pass the bill if the rate at which output grows is forever greater than the rate at which government pays interest on debt. (Even then, the noninterest part of the deficit must stay within a limit in relation to output.) But current and prospective interest rates are too high for that policy to work; and even if they were not too high now, they may become so later. Therefore, trying to evade the constraint and pass the bill is at best a gamble that exposes current or future citizens to the risk of higher net taxes than expected (see Appendix A). Of course, the borrowing crowds out private assets even when the gamble succeeds.
 
 

Estimating Tax and Transfer Payments by Age

The first step in carrying out the ideas behind the accounts is to find how taxes and transfers are now related to age. The average amount of any tax or transfer can vary greatly by age and sex (see Figures 1 and 2 and Box 1). By the estimates in the accounts, those who pay the highest taxes on income from labor are a bit over the age of 40; those who pay the highest taxes on income from capital are about 60 years old. Excise and property taxes fall more evenly on all age groups. Most Social Security and Medicare benefits go to those who are 65 or older, and benefits from Medicaid and other transfers appear more evenly distributed.

The profiles of taxes and transfers by age that are shown in Figures 1 and 2 reflect the judgments used in making them, not necessarily judgments that the Congressional Budget Office would make. Moreover, the profiles shown reflect an outdated version of the accounts, which contains errors that have since been corrected. The profiles shown for Medicare and Medicaid wrongly exclude disabled people who are younger than 65 or in nursing homes. Such people now receive about 16 percent of all Medicare benefits and 28 percent of all Medicaid benefits. For Medicare, the exclusions make the profile of those older than 65 too high in relation to that of younger people; for Medicaid, the profile is too low. As a practical matter, the exclusions have little effect on the main results considered later.

Figure 1.    Taxes Paid by the Average Member of Each Generation in 1991
SOURCE: Congressional Budget Office, using data provided by the authors as described in Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff, "Generational Accounts: A Meaningful Alternative to Deficit Accounting," in David Bradford, ed., Tax Policy and the Economy, vol. 5 (Cambridge, Mass.: MIT Press, 1991), pp. 55-110.  NOTE: N = newborns.

The accounts do not use the profiles in absolute terms, but as "relative-age profiles." For example, compared with the average 40-year-old man, the average 60-year-old man pays 66 percent as much in payroll taxes; the average 40-year-old woman, 46 percent as much; and so on. Such relationships are assumed to remain fixed. Given the relative-age profiles and population by age, the national total for any tax or transfer can be converted into an amount for the average person of any age and sex. The reverse is also true--that is, amounts per person can be converted to a total. Consequently, for example, total payroll taxes would fall if there were fewer 40-year-old men and as many more 60-year-old men, other things being equal.

Figure 2.   Transfers Received by the Average Member of Each Generation in 1991
SOURCE: Congressional Budget Office, using data provided by the authors as described in Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff, "Generational Accounts: A Meaningful Alternative to Deficit Accounting," in David Bradford, ed., Tax Policy and the Economy, vol. 5 (Cambridge, Mass.: MIT Press, 1991), pp. 55-110.
NOTES: The profiles shown reflect judgments made in constructing generational accounts, not necessarily judgments the Congressional Budget Office would make. The profiles have been updated in the latest version of generational accounts using more recent data or new data sources.  N = newborns.

In order to construct the relative-age profiles, the accounts start from official survey data. They also need to decide what to assume about the "incidence" of each tax and transfer; that is, who effectively pays or receives the cash value of a given tax or transfer? For economic or social reasons, that may not be the legal payer or recipient. Special assumptions are also needed to assign taxes on capital income and taxes and transfers within families.

Using Survey Data

The Current Population Survey, conducted by the Bureau of the Census, was used to estimate average labor earnings at any age in 1988. The accounts assume that people pay payroll and labor income taxes in proportion to their income from labor. Labor income includes the implicit labor income of proprietors, as well as the compensation of employees.

Box 1. How Generational Accounts Treat Taxes and Transfers

Generational accounts broadly consider five groups of taxes and three groups of transfers to persons.  Taxes comprise:  Excise taxes, which consist of sales taxes, tariffs, and property taxes paid by all businesses, including farms; Property taxes on owner-occupied homes; Payroll taxes, which consist of both employees' and employers' shares for social insurance and include the contributions of government workers to their pension funds; Taxes on labor income, which consist of income taxes paid on the income from labor of workers and proprietors; and   Taxes on capital income, which consist of corporate income taxes (excluding taxes paid by the Federal Reserve System), estate taxes, and income taxes paid on the capital income of proprietors, investors, and lenders.  The category also includes seignorage--a minor item that represents the revenue obtained from issuing money. Transfers to persons comprise: Social Security, which consists of Old-Age and Survivors Insurance and Disability Insurance (less federal income tax paid on such benefits), Railroad Retirement, and Supplemental Security Income; Health, which separately treats Medicaid and Medicare (less premiums for Part B); and  Other transfers, which treats separately Aid to Families with Dependent Children, Food Stamps, unemployment insurance, and general welfare.  The earned income tax credit is included with Food Stamps. Traditionally, transfers are defined as payments for which the government does not receive a current good or service in return.  People may receive them under entitlement programs, such as Medicare, or under discretionary programs, such as the Special Supplemental Food Program for Women, Infants, and Children.   In some cases, however, the accounts do not define taxes or transfers to persons in the usual way.  For example, they treat personal nontax receipts--such as licenses and user fees or tuition and hospital charges--as returns on government assets rather than taxes.  As a result, such fees are netted from both taxes and government purchases in the accounts.  Similarly, medical, disability, and retirement benefits for civil service and military personnel and veterans are treated in the accounts as purchases (compensation of employees) rather than transfers.  That treatment supposes that the government makes such payments as deferred compensation for past service under previously agreed terms.  Finally, payments that are not conventionally considered either purchases or transfers to persons must be dealt with.  The accounts treat as purchases both government transfers to foreigners (mostly foreign aid) and subsidies less current surpluses of government enterprises.  Payments of net interest on public debt need not be treated explicitly because they are implied in the process of discounting.

The accounts assume that labor's share of proprietors' income is about 80 percent--the same as its share of the rest of national income.

Other taxes and transfers are assigned in a similar way:

• Property taxes are assigned according to home values reported in the Survey of Income and Program Participation (SIPP) conducted by the Census Bureau;
• Capital income taxes (with special adjustments described below) according to assets reported in the Survey of Consumer Finances conducted by the Federal Reserve Board;
• Excise taxes according to household consumption reported in the Consumer Expenditure Survey presented by the Bureau of Labor Statistics;
• Social Security benefits according to payments reported in the Social Security Bulletin published by the Social Security Administration (SSA);
• Medicare and Medicaid according to health benefits reported in the National Medical Care Expenditure Survey conducted by the National Center for Health Services Research; and
• All other transfers according to benefits reported in the SIPP.

The relative-age profiles will change somewhat each time they are updated from the most recent survey or from a new data source. The change may reflect real trends or misleading results of sampling. In particular, the business cycle will affect relative-age profiles. For example, a young worker is more likely to lose a job during a recession; an old stockholder is more likely to suffer a drop in asset value and dividend income. Year-to-year changes in the accounts must be interpreted with those possibilities in mind.

Deciding on the Incidence of Taxes and Transfers

The way that the relative-age profiles are constructed depends on assumptions about the incidence of each type of tax and transfer. The assumptions made in the accounts imply that the supplies of saving and labor do not respond to changes in incentives that taxes and transfers provide.

Incidence is not obvious for two reasons. First, market forces may "shift" a tax or transfer from the legal payer or recipient to others. For example, if workers supply the same amount of labor regardless of pay, employers can shift their share of the payroll tax to workers by reducing wages.⁽¹⁾ That example illustrates a general rule: the less elastically supply responds to price, wage, or interest rate, the more the supplier bears the tax on the good, labor, or capital.

By contrast, higher retirement benefits might induce old workers to leave their jobs. Owners would have to raise the pay of remaining workers and suffer lower profits or pass the cost to consumers (including owners, workers, and retirees) as higher prices. None of that would happen, however, if old workers stayed at their jobs despite the higher retirement benefits (that is, if they supplied labor inelastically). Those who are retired now--or who will retire later--would get higher benefits, and that would be that.

Second, for familial or social reasons, transfers may "slide" from the direct recipient to others. For example, Social Security recipients might need less support from or give larger bequests to their children because of the benefits. If so, at least some of the benefits slide to the children. Any benefits that slide make the children better off and leave the parents as well off as they would have been otherwise. Benefits can slide even when the parties do not know each other. For instance, hospitals may treat uninsured patients and recoup their costs by charging higher prices to other patients. Those patients would therefore benefit if government provided aid to the uninsured.

The accounts assume that some taxes shift fully, some partially, and some not at all, and that no transfers shift or slide. Business excise taxes and the employer's share of the payroll tax are assumed to shift completely. Consumers and workers are assumed to pay those taxes rather than merchants and employers. Taxes on capital income are assumed to shift partially--that is, owners of capital, financial assets, and homes bear the total tax in the same proportion that they own the total net assets. By contrast, the accounts assume that people pay all of the property taxes on their homes, and workers pay all of the income and payroll taxes on their earnings from labor. And recipients of transfers are assumed to enjoy the full benefits without any effect on anyone else.

The assumptions for shifting and sliding imply that both labor and saving are supplied inelastically. Competition in financial markets ensures that capital, financial assets, and homes earn comparable risk- adjusted rates of return after tax (which is why, in the accounts, their owners bear the tax on income from capital according to their share of total net assets).

Assigning Taxes on Capital Income

The accounts treat taxes on capital income in a complex way because of investment incentives (accelerated depreciation or investment tax credits). The incentives make it necessary to adjust the data from the Current Population Survey in order to assign taxes to the right generations. Those adjustments are required, given the incentives under either current law or a change in law.

Investment Incentives Under Current Law. Investment incentives make owners of existing capital pay higher taxes on their prospective income than owners of new capital--that is, investment (see Appendix B). Hence, existing capital commands a lower price than otherwise-equivalent new capital; the difference in price is the present value of the excess taxes--in other words, the difference in taxes on income from existing and new capital. ("Excess" is used to describe the effect of tax law, not to suggest that taxes are too high.) Given its discounted price, buyers of existing capital will earn the same income after taxes as if they had bought new capital. Current owners pay the excess taxes whether they hold or sell their capital; they pay in fact if they hold and in effect if they sell. (For that purpose, owners include those who hold financial assets or homes because in the long run competition makes them bear their proportional share of the tax.)

Therefore, some taxes in a later year will in effect be paid by this year's owners who have sold in the meantime, rather than by that later year's owners. The accounts must allow for that event in order to assign prospective taxes to people in the right generations. To do so, they prorate the present value of the excess taxes to this year's owners.

Without the adjustments, the accounts would understate prospective taxes of the middle-aged and old and overstate those of the young. For example, the adjustments add about $6,000 to the present value of taxes of the average 60-year-old. That amount is about 15 percent of the value of his or her capital.

Investment Incentives Under a Change in Law.  If investment incentives are raised, resources are transferred from old to young, according to the accounts. The higher incentives effectively reduce taxes on new capital. But taxes on existing capital remain as they were, so excess taxes are even higher than before. The accounts prorate the present value of the increase in excess taxes to current owners. At the same time, the lower effective tax rate on new capital reduces the taxes that prospective owners will pay. Thus, according to the accounts, an increase in incentives raises taxes of current owners (mostly old) and reduces those of prospective owners (mostly young).

In principle, the effect on a current owner when an investment incentive rises is the same as that on a bondholder when the interest rate rises. If both hold their assets until the assets expire, they will pay as much tax or earn as much interest as they would have in the absence of the rise. But their assets will fall in value because new capital would pay less tax, or a new bond would earn more interest. If they sell, they have to absorb the difference in higher taxes or lower interest.

Assigning Taxes and Transfers Within Families

Arbitrary judgments must be made in order to assign taxes and transfers within families. There is no clearly right way to split net taxes between husbands and wives, or between parents and children. Should a payroll tax be assigned to the earner, to husband and wife jointly, or to all family members according to their share of consumption?

The accounts use a variety of methods to assign taxes within families. They assign payroll and income taxes to husbands and wives according to which earns the pay or owns the asset. By contrast, property taxes are split 50-50. And excise taxes are assigned to all family members according to their share of consumption. So, for example, the accounts estimate that a current newborn pays about one-fifth as much in excise taxes as a 40-year-old.

Transfers are assigned to the person who directly receives a payment or service. Therefore, the accounts assign to the head of a family the benefits from Aid to Families with Dependent Children (AFDC), Food Stamps, and general welfare. And the direct recipient is assigned the benefits from Social Security, Medicare, Medicaid, and unemployment insurance.

Those treatments can lead to anomalies, although the most severe problem can easily be avoided. Any method of splitting net taxes between husbands and wives must be arbitrary and can produce misleading results. It is possible to avoid such problems by presenting the results as weighted averages for males and females together rather than separately. This study does so.

Problems remain in treating dependent children, however. For instance, according to the version of the accounts used for this study, children would benefit from an increase in their Medicaid or Survivors' Insurance benefits, but not from an increase in AFDC benefits. But they should benefit from AFDC; that is its purpose. Similarly, the accounts indicate that children would lose if an increase in excise taxes paid for higher income tax exemptions for dependents. The net income of their family, however, would rise.

It would seem preferable to choose one consistent method to treat dependent children--that is, to assign the cash value of net taxes either to adults or to all family members according to their share of consumption. Either choice involves arbitrary elements, but consistency would help clarify matters.
 
 

Calculating Generational Accounts

In order to estimate the net taxes of future generations, the zero-sum constraint is rearranged. In other words, the present value (PV) of net taxes of future generations must equal the current net debt of government, plus the present value of all prospective government purchases, minus the present value of prospective net taxes of current generations:

Two elements are required to project net taxes of current generations and government purchases: a definition of "prevailing policy," which relates taxes and spending to population and income, and projections of population and income.

Defining Prevailing Policy

The relative-age profiles serve as a key to defining prevailing policy. The assumption that the profiles remain fixed under prevailing policy relates people's prospective taxes and transfers with their age. It then remains to relate taxes and spending with their income.

Net Taxes of Current Generations. Prevailing policy is defined in two parts: first by current policy, then by a mechanical rule. That is, current policy determines the total of each tax and transfer in an official projection of the economy and budget. Given the relative-age profiles and a projection of population, the taxes and transfers of the average member of each current generation are calculated through the end of the projection period.

Beyond the official projection, prevailing policy applies a rule: each year, the real taxes and transfers of the average person of a given age grow at the same rate as productivity (loosely, real output per worker). For instance, suppose a 30-year-old man paid $4,000 in payroll taxes in a given year and productivity grew at 1 percent a year. The next year, a 30-year-old man (the previous year's 29-year-old) would pay $4,040 in payroll taxes. That rule keeps relative-age profiles fixed, but allows absolute profiles to grow in line with productivity.

This method enables the accounts to project the net taxes of the average members of all current generations through the rest of their lives. A projection of population determines the number of people of a given generation who will survive in each succeeding year. Therefore, the extensions reflect growth in the economy, and mortality and migration in the population. The entire procedure applies only to current generations because the net taxes of future generations are determined from the zero-sum constraint.

Government Purchases. Government purchases are also determined by an official projection, then by a rule that relates them to the growth of productivity and population. The rule for purchases, however, applies to both current and future generations. In other words, the accounts take all purchases as given by prevailing policy, then ask which generations pay for them with their net taxes.

Implications of Prevailing Policy. The mechanical rules used to define prevailing policy imply that current law would not remain as it is for current generations. For example, the rules would require that the Congress adjust tax schedules so that overall growth in real incomes did not push people now living into higher income tax brackets. Similarly, the Congress would have to adjust welfare benefits so that the payment for the average person (now alive) at each age grew at the same rate as wages (which grow at the same rate as productivity).

Such rules are commonly used for long-run projections because they make all sectors of the economy grow at the same rate as income and output. For instance, the taxes and transfers of current generations would remain constant as shares of their incomes. And government purchases would remain constant as a share of output (when the age of the population remains stable). If sectors did not grow at the same rate in the long run, the fastest-growing sector would grow to the size of the whole. Therefore, mechanical rules are typically used in the absence of better information.

Nevertheless, sectors can grow faster or slower than output for long periods. For instance, over the past century consumer services have grown from 23 percent of output to 39 percent, and farm products have shrunk from 23 percent of output to 1 percent. Thus, the definition of prevailing policy contains a subjective element.

Choosing Projections of Population and Income

Given the rules of prevailing policy, the accounts need economic and demographic assumptions in order to extend and discount the components of the zero-sum equation. For their base case, the accounts assume:

• Productivity, as the accounts define it, grows at the rate that the Office of Management and Budget (OMB) projects through 2004, and thereafter at 0.75 percent a year (roughly its rate since the mid-1970s);
• Population follows the Social Security Administration's midgrowth projection through 2080 and a mechanical extension thereafter;
• The structure of the economy remains as it is today; and
• A real discount rate of 6 percent applies to all streams of taxes and transfers for all generations.

Population. Population is extrapolated from 2080 to 2200 by assuming that the rates of fertility, death, and immigration remain at the levels the SSA projects for 2080. After 2200, the size and composition of the population are assumed to remain constant. (By about 2040, the SSA projections already put growth of the population near zero.)

Structure of the Economy. Fixed relative-age profiles define the structure of the economy. For instance, they imply that, by age and sex, there is never any change in rates of participation in the labor force, relative earnings, the average work week, the ratios of assets to income, health needs, and so forth.

Discount Rate. The authors of the accounts reason that the real rate of discount should be higher than the real rate of interest on long-term government debt (which is about 2 percent or 3 percent). They maintain that payment of prospective taxes and transfers is less certain than payment of interest and repayment of debt. If so, people should use a discount rate that includes a premium to account for the risk that their net taxes may vary from those scheduled. A real rate of 6 percent equals the average historical rate of return on equity before tax. A before-tax rate is used because net taxes are drawn from the before-tax income of the nation.

Calculating the Components of the Zero-Sum Constraint

Given specific projections of the economy, it is now possible to calculate the components of the zero-sum constraint, namely:

• Net government debt,
• Present value of all prospective government purchases,
• Present value of prospective net taxes of current generations, and
• Present value of net taxes of future generations.

The sum excludes government debt held by government entities, such as the Social Security trust fund, because issue of that debt does not show up in the unified deficit. Such issue is merely a bookkeeping entry that authorizes the agency to spend money. Moreover, generational accounts omit tangible assets of government--such as land, schools, or highways--that reduce net debt. Omitting such assets is not serious, however, because including them would also require including an offsetting item (see Box 2).

Present Value of All Prospective Government Purchases. Purchases are projected under current policy for 10 years, through 2004 in the version of the accounts that the Congressional Budget Office used. Federal purchases are projected by OMB, and state and local purchases are assumed to grow at the rate OMB projects for gross domestic product. (The latest presentation of the accounts has OMB's numbers for federal taxes and spending through 2030.⁽²⁾ The difference does not affect the main results presented later.)

Total purchases are then extended by assuming that real purchases per person in each subpopulation group and in the total population grow at the rate of productivity. The present value of prospective purchases through 2200 can be calculated by applying a discount rate. It possible to calculate the present value of purchases beyond that date by a simple equation because the size and composition of the population are assumed to remain stable after 2200.

Over the past 30 years, this method of projecting government purchases would have overstated their growth. During that period, state and local purchases rose as a share of gross domestic product (GDP) from 9 percent to 11 percent, but federal purchases fell from 11 percent to 7 percent, mostly because of slow growth in defense spending. All government purchases fell from 20 percent of GDP to 18 percent.

Present Value of Prospective Net Taxes of Current Generations. Generational accounts combine official projections under current policy from a number of sources. OMB's projection of taxes and spending through 2004 serves as a base in the version of the accounts that CBO used.⁽³⁾ Current policy can be defined in two ways, however: after 1998, it might hold discretionary federal spending constant in either real or nominal terms. The definition OMB used implies more spending through 2004 and, hence, higher net taxes for future generations.

Other official sources are used to extend the totals of some taxes and transfers beyond 2004. Through 2030, the accounts use projections by the Health Care Financing Administration (HCFA) for Medicare and Medicaid; through 2070, the accounts use projections by the SSA for payroll taxes and Social Security benefits. Those agencies provide modifications of their official projections that are consistent with the economic assumptions of the accounts. The accounts then make the yearly total for each tax or transfer grow from 2004 to the end of its official horizon at the same rate as it does in its modified projection.

Beyond the official horizons, prevailing policy assumes that the real taxes and transfers of the average person of a given age grow at the rate of productivity. Thus, projections of most taxes and transfers reflect rules that keep their growth in line with incomes after 2004.

But the largest and fastest-growing transfers reflect current law and official projections through 2030 or 2070. For instance, prevailing policy includes the assumption by HCFA that real medical costs per recipient will grow faster than productivity through 2020 and the phase-in of the earliest age--from 65 to 67--at which Social Security recipients may draw full benefits.

This method yields the prospective net taxes of the average members of each current generation for the rest of their lives. The accounts apply a discount rate to those net tax streams to calculate their present values. Adding those present values for all the people of every age now alive gives the present value of net taxes of all current generations.

Present Value of Net Taxes of Future Generations. The present value of net taxes of all future generations is now given from the right-side components of the zero-sum equation. To find the payments of each future generation, it is assumed that they all pay net taxes at the same rate. Then it is possible to calculate their payments knowing the number of people in each generation and their income. The number of people is given by the population projection, and their income by the growth of productivity. For example, the real income of next year's average newborn will be higher than that of this year's by the growth of productivity, and so on. Arithmetic then gives the present value of net taxes of each future generation. That calculation is not intended to be realistic, but to make it possible to speak of a representative future generation.
 
 

Reporting and Interpreting Generational Accounts

Generational accounts must report the results in a way that provides a basis of comparison among generations. Simply reporting the results as the present values of prospective net taxes under a given policy would not do so. For example, under prevailing policy, the present value of prospective net taxes of a 40-year-old is higher than that of a 50-year-old. The
40-year-old has 10 more years of taxes to pay and is 10 years further from receiving Social Security and Medicare.

It is not possible to know from that comparison, however, whether past and prospective policy treat the two in the same way. For example, the 40-year-old has earned higher real income than the 50-year-old did at comparable ages. Furthermore, has the 40-year-old paid net taxes in the past at the same rates that the 50-year-old had at the age of 40? Will the 40-year-old pay net taxes for the next 10 years at the same rates that the 50-year-old did for the past 10 years?

Reporting the Results

Generational accounts can be reported in at least two ways that provide a basis of comparison among all generations: as a net tax rate paid over a lifetime or as a change in the present value of prospective net taxes under a change in policy.

Lifetime Net Tax Rate. A generation's lifetime net tax rate is its lifetime net taxes as a percentage of its lifetime labor income. Specifically, a lifetime net taxrate is the present value at birth of net taxes over a lifetime as a percentage of the present value at birth of labor income over a lifetime. (Lifetime labor income is used as a base because it is closely related to lifetime consumption--a basic measure of well-being. See Box 3.) This calculation compares all generations on the same basis because it includes the effects of all policy, past and prospective, from birth.

To calculate lifetime net tax rates, the accounts must first estimate net taxes already paid by the average member of each current generation. To do so, the accounts use survey data to estimate the relative-age profiles for labor income, taxes, and transfers that prevailed in the past.
 

Dollar Change in Present Value of Net Taxes. The results may also be presented as the change in the present value of net taxes for any given change in policy. For instance, a new policy might reduce the present value of net taxes of the average 20-year-old by $200 and raise that of the average 50-year-old by $100. This presentation also compares people of all ages on the same basis because the entire effect of the change in policy is prospective. But the results must be interpreted with care because they do not reflect the net taxes that people of different ages have paid or will pay.

Interpreting the Results

Generational accounts serve only as an abstract indicator, not a predictor or goal. They do not say how policy will or should evolve; such questions remain beyond analysis.

The accounts set a standard by which prevailing policy may be judged and other policies compared. In that respect, they resemble other conceptual standards that answer "as if" questions, rather than making realistic predictions. For example, the baseline budget establishes a reference point as if current policy were to remain in force for everyone, alive or yet to be born; or the full-employment budget separates the effects of policy on the budget from the effects of the economy on the budget as if the economy were at full employment. Generational accounts indicate how policy would distribute resources among generations as if prevailing policy were to continue without change for those now living.

As one point of reference, the accounts indicate whether a policy is "sustainable." It is if scheduled rates of taxes and spending according to age need not change to satisfy the zero-sum constraint. Thus, a policy is sustainable if it implies no difference in the lifetime net tax rates of future generations and current newborns. In that case, each generation could pay net taxes at every age at the rates that are scheduled now and satisfy the zero-sum constraint. Of course, those rates must also be feasible; for instance, people cannot pay more in net taxes than they earn in a lifetime.

A policy is not sustainable if there is a difference in the lifetime net tax rates of future generations and current newborns. In that case, scheduled rates of taxes or spending according to age would have to change--for either current or future generations--in order to satisfy the zero-sum constraint. The accounts do not predict how taxes or spending would change.

Sustainability need not imply desirability. For example, future generations will typically be much richer than current generations, so it may be fair for them to pay net taxes at higher rates. The accounts can address only sustainability, not fairness.

Generational accounts and long-term deficit projections both address sustainability, but present the information in different ways. Strictly, the accounts and an infinite projection of the deficit would require the same data and assumptions. (The information they convey would be equivalent if the accounts used a discount rate equal to the interest rate on government debt.) Given an infinite horizon, implicit obligations must show up in the deficit at some time. In that case, it is not possible to obscure the direction of policy by undertaking implicit obligations that do not raise the deficit now, but would raise it later. Thus, a deficit projection would show that policy is not sustainable if government debt would continually rise in relation to output.

A deficit projection would not, however, address distribution by age. Therefore, some advocates of generational accounts maintain that debate about fiscal policy should focus on how it affects those accounts, rather than the deficit (see Appendix C).

1. Congressional Budget Office, An Analysis of the Administration's Health Proposal (February 1994), Chapter 4.
2. Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff, "Restoring Generational Balance in U.S. Fiscal Policy: What Will It Take?" Economic Review, Federal Reserve Bank of Cleveland, vol. 31, no. 1 (First Quarter 1995), pp. 2-12.
3. Budget of the United States Government, Fiscal Year 1995 (January 1994).

---