Document View

An edited transcript of a general session at the Western Economic Association International 70th Annual Conference, held in San Diego, California, on July 7, 1995, is presented. Topics addressed at the session included US fiscal policy, US national saving, and US domestic investment, and how these issues are connected to US generational policy.

Demsetz: Welcome to "Welfare Across the Generations." In accord with the topic, the panel members represent different generations. Those who know me know that this topic is not a specialty of mine. I chose it because of its contemporary importance. It is very topical--in the newspapers and on government officials' lips. I also chose it because my children would like me to learn more about intergenerational wealth transfers. I would like them to learn more about wealth creation.

I am enthusiastic about the panel. The age and idea mixture is a good one, and I really look forward to hearing what they have to say.

The first panelist, Larry Kotlikoff, as you know, is from Boston University and is a research associate of the NBER. In 1981 and 1982, he was a senior economist for taxation and Social Security on the Council of Economic Advisers, so he comes well versed in his topic.

Henning Bohn, the second speaker, is from the University of California, Santa Barbara. He has written several fine articles in the last couple of years on this topic.

Earl Thompson, the third speaker, is my UCLA colleague and one of the most innovative economists I know. He always has interesting things to say about government policy and the family, and I am sure he will not disappoint us this time.

Andrew Samwick, who represents the next generation, is from Dartmouth College and is a faculty research fellow at the NBER. He also has written several very good papers on saving, Social Security, and tax policy. He recently has been thinking about the topic of privatizing Social Security.

Larry, please step up to the podium.

Kotlikoff: I would like to talk about U.S. fiscal policy, U.S. national saving, and U.S. domestic investment--specifically, how they are connected to U.S. generational policy.

First, it is fair to say that we are facing two crises, one with respect to national saving and investment and growth and the other with respect to fiscal policy. And these crises are really very much interrelated. On the saving side, we have a national saving rate that since 1990 has been running at 2.7 percent per year. That is dramatically lower than the saving rates observed in the past. For example, in the 1950s and 1960s, we had a national saving rate of 9.1 percent on average. Our domestic investment rate also is quite low. It is running about a third of what it was in the 1950s and 1960s. As a result of low investment, we have slow growth in capital per worker and, therefore, slow growth in labor productivity and, therefore, slow growth in real wages. That is one of the whammies in the double whammy young people are facing.

The other whammy is that young people are facing pretty high rates of taxes, and these tax rates are likely to soar as we move into the next century. The present administration has continued a longstanding policy that dates to Eisenhower-a policy that allows resources from young and future generations to be passed to older generations. Eisenhower initiated this policy when he increased the Social Security program to about a third of its current size. The program continued under Kennedy, and Johnson, of course, introduced Medicare. Nixon double-indexed Social Security and also allowed Medicare and Medicaid to grow at very rapid rates. The rapid growth of those health care programs continued through the Reagan and Bush administrations and continues under President Clinton.

Last year, Medicare, which is a $160 billion program, grew at 9 percent in real terms-three times the growth rate of the economy. That growth was paid for by current middle-age and young people through payroll taxes. Older generations were not asked to pay an extra penny for the additional growth of Medicare benefits.

Now, though, our politicians in Washington tell us that they are going to deal with this process of transferring from the young to the old. Their focus has been on the budget deficit, and they tell us that the real key to fixing our generational problem is dealing with the budget deficit. They say that if we just can get the budget deficit to be zero, we will have dealt with our problem.

I would like to persuade you that budget balance and generational balance have no necessary connection. The Republicans are telling us that they want to achieve budget balance not now, not next year, not the year after, but in seven years. Thus, the Republicans are not going to reduce the size of the debt. Instead, they are going to let the official debt grow for seven years. And President Clinton tells us that he wants to achieve budget balance not in seven years but in 10 years.

The question is: "Even if you achieve a balanced budget 10 years or seven years from now, is that going to achieve generational balance? Does budget balance imply generational balance? The answer is no. The first reason that these things do not have any necessary connection is that you do not have just a single definition of the debt or of its change, the deficit.

As we saw in the debate about the BaIanced Budget Amendment, there are a great variety of different definitions of the budget and the deficit. Which deficit should be set equal to zero is a big political football. Recall the Democratic senator from South Dakota who voted against the Balanced Budget Amendment because he questioned the definition of the deficit. He had a point. There really is no single correct measure of the deficit. Indeed, economic theory tells us that the deficit is not a well-defined economic concept, that the deficit instead is simply a reflection of our fiscal vocabulary--specifically, how we label government receipts and payments.

For example, if I call this year's contributions that workers make to social insurance programs a loan to the government rather than a tax (because those contributions implicitly are government IOUs), then those contributions will be included as part of the official deficit. This change in words would almost triple the reported official deficit. In using the words they do, our politicians are focusing on one of a zillion alternative measures of the deficit. None of the definitions bears any intrinsic relationship to our true fiscal policy. There is no reason to expect that reducing this particular deficit to zero in seven or 10 years will treat future generations fairly relative to current generations.

If we really care about generational equity, why don't we just measure how we are treating different generations through the fiscal system. There is a very easy way to do this. It is called generational accounting, and it has been in existence since 1989. It is based on work that Alan Auerbach, Jagadeesh Gokhale, and I have done. Generational accounting has been included in some of the budget documents of the federal government in recent years.

Last January, we prepared a chapter on generational accounting for inclusion in this year's federal budget. But at the very last minute, the President's advisors decided that his budget, which is over 2,000 pages long, could not afford to have a couple of pages talking about how this administration is treating our children. So the folks in the west wing of the White House threw out this section on generational accounting from the budget two days before the budget was published. I think that is highly regrettable. For one thing, a lot of countries around the world are looking at what the U.S. government does with respect to its budget documents to guide their own thinking about how they should do budget analysis. As a result of past publications of generational accounting, we have seen countries from New Zealand to Norway to Japan to Italy to Sweden to Thailand starting to do generational accounting. But the deletion is not surprising. We have a situation in which the actual generational picture looks much, much worse than the picture one gets from looking at the federal budget deficit and the official federal debt.

Official federal debt as a share of gross domestic product is a smaller fraction of gross domestic product now than it was in 1950, and the economy did very well for 20 or so years after 1950. So if we were concerned only about federal debt, we could relax. It is a small deal compared to the size of the economy. But we have a variety of other programs which are not recorded in this particular arbitrary measure of the debt--these pay-as-you-go Social Security programs that were not in place in 1950. The payroll tax back then was 3 percent. Right now we have a 15.3 percent payroll tax for Social Security and Medicare, which, according to projections by the actuaries of the Social Security and Medicare trustees, could well double by the early part of the next century.

Generational accounting tries to measure directly how we are treating different generations. Its methodology is very similar to that used by the trustees of the Social Security Administration in their annual Trustees' Report. Generational accounting considers the bills that the government must pay--the government's official debt plus the present value of the government's purchases of goods and services as we project them into the future. These bills must be covered by the present value of the net tax payments of current and future generations. Stated differently, the government's bills, measured in present value, must be paid either by current or future generations. It really is just an expression of the government's intertemporal budget constraint. Generational accounting just attaches numbers to the elements of this equation.

We estimate the present value of projected government purchases, and we add official debt--that is, the size of the bills. Then we calculate the present value of the projected taxes net of transfers--the net taxes that currently living generations will have to pay--and we subtract those from the bills. The difference is the bill that future generations will be left with. That is the residual. When we do that, we find that the projected bill that future generations will have to pay is astronomically large. The ratio of this bill to the labor income, also measured in present value, that future generations collectively are projected to earn, is 84 percent.

Now, you might say that those numbers that you put into your computer model probably are overly pessimistic. The fact is that virtually every single number that goes into generational accounting is provided by the government. The demographic projections come from the Social Security administration. The forecast of taxes and transfers come from the Office of Management and Budget, the Social Security Administration, the Health Care Financing Administration, and so forth. Furthermore, we are using the government's intermediate projections, not its pessimistic projections.

Let's compare the 84 percent number with the lifetime tax rates of generations born in the past. The generation born around the turn of the century paid about a quarter of their lifetime labor income in taxes net of transfers. Under current policy, my generation will pay about a third of our lifetime labor income in taxes net of transfers. But generational accounting shows that current policy is not sustainable. We cannot continue taxing my generation and other current generations and at the rates of 33 percent because such a policy will leave a net tax burden on future generations that is so high that it will not be collectable.

We cannot tax 84 cents of every dollar that future generations earn. We need to adopt policies that will lower that 84 percent number and raise the burden on current generations until they meet--until we achieve generational balance. Indeed, what we really need is a generation balance amendment to the Constitution, not a balanced budget amendment.

What if the Republicans do pass their bill and do balance the budget in the year 2002? Will that help? Will that lower the 84 percent net tax rate facing future Americans? It would go some of the way but not very far. The net tax rate falls to 72 percent, whereas the burden on my generation, for example, goes up from about 33 percent to about 36 percent. The Republican bill is far too little too late.

If you look at the actual details of the Republican proposals, you see that what they are proposing involves Medicare and Medicaid continuing to grow through the next 15 or so years at something like twice the rate of the economy. However, they are not really putting a halt to excessive growth of Medicare and Medicaid. These programs now spend $260 billion, so letting these programs grow at twice the rate of the economy is inviting disaster.

If we could stabilize just these two programs so per capita benefits for Medicare/Medicaid grow at the same rate as labor productivity growth, that 84 percent lifetime tax rate facing future generations would come down to 53 percent.

Now let me try to connect our national saving and our generational problems together. The reason that our national saving rate has declined so dramatically is due to the growth in household consumption as a share of output. Indeed, government consumption as a share of output actually is lower now than it was during the 1950s and 1960s.

Who within the household sector is consuming so much? The answer is older generations. Their consumption relative to that of young people has risen dramatically. For example, consider the average consumption of 70-year-olds compared to that of 30-year-olds. That ratio in 1960 was about 0.8. Today it is about 1.2. "Why has the elderly's consumption gone up so much?" Primarily because their relative resources have risen dramatically, thanks in large part, to the transfer of resources from young people to old people through Social Security and Medicare and Medicaid. Another factor involved is that, compared with the past, a lot more of older generations' resources are in the form of annuities--Medicare, Medicaid, Social Security, private pensions--which continue for as long as the recipients live. So the elderly can consume at a higher rate without worrying as much about running out of money. Indeed, the propensity of the elderly to consume out of any dollar they have in present value resources has gone up. In the case of Americans over 75, it has tripled.

To summarize, we have a policy that systematically is taking from the young and giving to the old, leading to an increase in the consumption by the elderly relative to that by young people, an increase in aggregate consumption relative to output, and a decline in national saving, investment, labor productivity growth, and real wage growth. So young people have been hit with a double whammyhigh taxes, which will go sky high unless we get control of things, as well as a sickly economy that is just chugging very slowly along and producing very little real wage growth. This policy is transforming the American dream-that every generation of Americans would be better off than the previous generation-into something it never has been just a dream.

Demsetz: Thank you very much, Larry. Henning, Are you ready?

Bohn: The panel's topic, "Welfare across the generations," is extremely broad. A vast range of issues fall under this title-many more than the obvious policy problems surrounding the government budget deficit and Social Security. Therefore, before I turn to policy problems, let me comment on two private sector issues, bequests and questions of economy growth.

Bequests and gifts create strong intergenerational linkages. They do not directly involve the government, but they are potentially important for evaluating government policy. For example, if children provide less support for their own parents now than they did before Social Security existed, Social Security may have much less of an intergenerational welfare effect than the raw fiscal data would suggest. Also, if the young are tempted to complain about the government debt left to them by their parents generation, they should not forget the education provided by their parents and through the government.

In this context, issues of economy growth are relevant because if productivity is growing rapidly, one may take for granted that future generations will be much better off than we are. But without productivity growth, the current generation must be much more careful about the burdens they can impose on future generations. For example, if real wages grew at an annual rate of 1.5 percent--which now is the Social Security administration's optimistic scenario, but below the growth rates of the 1950s and 1960s-real incomes would rise by more than 50 percent over the next 30 years. So, if we pay 30 percent taxes on a pre-tax per capita income of about $30,000 and if the next generation pays, say, 40 percent taxes on a real income of $45,000, real disposable income still would rise from $21,000 to $27,000. Future generations would be able to carry a rising fiscal burden and still be better off than we are. Intergenerational redistribution to us at the expense of future generations would then amount to taxing the relatively rich to help the relatively poor, something we commonly do within generations.

The question of how much the government should redistribute from rich to poor long has been the subject of political debate. I would argue that intergenerational redistribution is debatable on the same level. Hence, unless one takes the position that all redistribution is unfair--a position that does not seem popular when intragenerational redistribution is concernedthere is no fundamental reason why intergenerational redistribution per se is normatively objectionable. (As a parent, I do not like this conclusion, but I would be intellectually dishonest to deny it.)

The normative conclusions change, however, if productivity growth is near zero. Any policy that implies increasing tax rates over time then has the potential of leaving future generations worse off in absolute terms than the current generation. Such a policy would be very difficult to reconcile with the general principle of distributional fairness that redistribution should go from rich to poor.

Overall, it seems clear that the post1973 slowdown in productivity growth is a key reason why intergenerational redistribution has become a major policy issue. Much less clear is by what standards such redistribution should be judged. Even if current fiscal policy implies rising tax rates in the future, as Larry Kotlikoff's calculations suggest, we do not know whether the implied redistribution is so large that it amounts to redistributing from the poor to the rich.

With this background, let me turn to some policy issues. We know that the U.S. government has run budget deficits without interruption since 1970. Calculations of U.S. government net worth show increasingly negative values--minus $2.882 trillion at end of fiscal year 1994 versus a value near zero in 1960 (according to the 1995 U.S. budget). The Social Security retirement fund is projected to run out of reserves around 2030, and Medicare even earlier. Taken together, these fiscal data raise some serious questions about the generational impact of government policy. To what extent does the government alter the intergenerational distribution of resources? Does the government favor one generation over another?

To address these questions in a systematic way, one must ask about the appropriate benchmark for evaluating government policy.

Question 1: What would be a policy that is neutral with respect to intergenerational distribution, a policy that does not favor one generation over another? I would interpret Larry Kotlikoff's generational accounting as an attempt to answer this question. But to put the answer in perspective, one should ask a follow-up question.

Question 2: If we found a generationally neutral policy, is there any normative reason why the government should pursue it? Here is where the issue of productivity growth matters. If future generations are better off, why should we not impose a burden on them? As long as distributional judgments are disputable, it is unlikely that there is any one policy that one can single out as being optimal under some coherent set of conditions. At best, we may try to eliminate policies that conflict with some widely accepted principles of fairness (e.g., against redistribution from poor to rich). This raises the question of constraints.

Question 3: What is the set of policies from which the government can pick? Logically, this may be considered a prior question, but assuming that the neutral or "optimal" benchmark policy is feasible, the question is how far a government can deviate from neutrality or optimality. This question also is important from a political economy perspective. If we consider a government acting in the interest of some particular generation(s), it is important to know to what extent one generation can use the government to distribute resources in its favor, i.e., what limits intergenerational redistribution?

I doubt that any general answer to the question about optimal intergenerational redistribution exists, primarily because any specific distributional norm would require interpersonal welfare comparisons that inevitably are controversial. But several policy criteria have been used to make normative judgments. Let me comment on two of them, namely generational accounting-motivated by Larry's presence on this panel--and criteria based on government net worth--building on my own work. (Note that I do not mention conventionally measured budget deficits in this context because I share Larry's concerns that the currently used government accounting system is badly flawed.)

Generational Accounting

Larry already explained the concept. I think it is an excellent idea, but it suffers from at least three major problems. First, generational accounting does not account for the benefits of government spending. Government spending is justified only if it has benefits that exceed the cost. For normative analysis, there is no excuse for not weighting the welfare benefits from the government services enjoyed by a generation against the net taxes paid by the same generation.

Second, generational accounting has no comment on what might constrain future government policy. Constraints are important because they provide a reality check on projected future taxes and transfers. The basic assumption of generational accounting is that all currently living cohorts receive all promised government transfers. But this assumption is unrealistic if the promised transfers are so large that they are infeasible. I suspect that this currently is a problem with respect to Social Security and certainly with respect to Medicare.

Third, and perhaps most important, generational accounting is silent about how people should form expectations about future government policy. Generational accounts are used to evaluate alternative long-run policy plans without worrying about how policy choices actually are made. I believe this is missing the point why ordinary people are concerned about government debt and about the budget deficit. In the United States, we have elections every two years. Each new Congress is free to change tax and transfer policies at its discretion. It therefore is unrealistic to take current policy plans at face value, as if the commitment to execute them were immutable.

A game theoretic, political economy approach to fiscal policy and intergenerational welfare issues would be much more consistent with the institutional realities. We can think of successive governments as being controlled by different age cohorts. Each government controls current spending, transfers, and taxes, but it cannot precommit future tax and spending choices-at least not directly. The key questions for understanding government policy in this context are (i) how individuals form expectations about future policy and (ii) what actions the current government may take to constrain future government choices.

Government debt and other government liabilities are economically relevant in this context, because debt becomes a constraint if future governments are bound to honor an outstanding debt. By incurring debt or other liabilities, the generation controlling the current government may be able to obtain additional resources that can be distributed to its members, at the expense of future generations. I agree with Larry that the labels, such as "debt" or "liability", that we put on different government payments are not the main issue. The central economic issue is whether the current government has the ability to commit future governments to make payments to certain individuals. If such commitments can be made, it is economically important to keep track of them and to distinguish government payments to individuals that are made to honor prior commitments from payments that are made at the discretion of the current government.

The practical question then is: What is the set of financial commitments incurred by current and past governments that constrain future governments? I will call such commitments "debts" or "liabilities."(1) Similarly, it is economically important to know the value of government assets that the current government leaves at the disposal of future governments. This leads to

the second policy criterion.

Changes in Government Net Worth

Government net worth is the difference between government assets and liabilities. It measures the net resources left by the current government to future governments. The concept of government net worth draws a line between the economic resources of the public sector and those of the private sector, which is consistent with the notion that the government has limited powers and operates under various constraints. Most important, the government cannot simply transfer resources from the private to the public sector because taxation is distortionary. Individuals increasingly will engage in tax avoidanceor perhaps even tax evasion activities if tax rates are too high.

As explained before, the set of items included under the heading "liabilities" generally should be determined by an economic analysis of which claims against the government impose constraints on future governments. More specifically, I would argue that in developed democracies in which the government is bound by the rule of law, governments are typically obliged to honor the kinds of commitments that are considered liabilities under generally accepted private sector accounting principles. This includes commitments to pay government employee wages and accrued pensions, to pay government contractors, and to honor the national debt.

The government's intertemporal budget constraint implies that the government's net worth can be written as the present value of primary budget surpluses, which are the differences between tax revenues and the cost of government services. The primary surpluses should of course be computed according to generally accepted accounting principles, which means that the cost of government services includes the user cost of government capital and that the surpluses are not distorted by purchases or sales of government assets. Compared to current federal government accounting practices, the use of generally accepted private sector accounting principles and balance sheets would reduce significantly politicians ability to manipulate the government's accounts. Better information about the government's fiscal status should help citizen to hold politicians accountable for their actions. These are additional benefits of moving to a system in which the government is subject to the same accounting conventions as everyone else.

I already have explained the strategic importance of government liabilities and assets in a game theoretic, political economy context. More details are in my recent Carnegie-Rochester conference paper (Bohn 1992). Instead, let me turn to the more difficult task of explaining why primary deficits and changes in net worth have some direct normative implications.

To see the normative relevance of government net worth and the primary balance, note that the government's intertemporal budget constraint would be satisfied if, starting with a zero net worth, each cohort in each period paid for all the government services it consumes. This policy of contemporaneous payments for all government services provides a natural benchmark for generational neutrality. Moreover, it would provide the proper incentives for each generation to demand only government services for which the benefits exceed the cost. If intergenerational redistribution is considered undesirable--a big if--this policy might be normatively appealing. However, the normative use of net worth and the primary balance presents some practical complications.

Complication 1. The government holds risky assets, such as real estate, mineral rights, gold, and mortgage loans. What should we do about capital gains and losses?

The change in net worth generally can be divided into the primary (non-interest) surplus or deficit and a term representing the total return on initial assets and liabilities. If the government balance sheet contains risky assets and liabilities, the expost return often will include large unexpected capital gains or looses. Then the principle that each generation should pay for the government services it consumes still would call for a non-negative primary balance. But it is not clear that each generation should maintain a zero net worth. It is an open question how the risk of unexpected capital gains or looses on government assets should be shared between generations. There is no reason why current taxpayers should bear all the risk, but they probably should take some responsibility for the management of government assets. (The Federal Savings and Loan Insurance system comes to mind as a recent example of losses due to apparent government mismanagement.) If the current cohort of taxpayers does not fully cover capital gains and losses, the government will start out with a non-zero net worth in next period. This raises the broader question of what to do if the initial government net worth is non-zero.

Complication 2. Suppose the initial government net worth is negative, as it is currently in the United States. Who should pay for the shortfall? A specific answer would require interpersonal utility comparisons. But one can make one generalstatement: If the primary balance is negative, current taxpayers pay less than they receive in government services. They act in way that increases the negative net worth problem left to future taxpayers. In this sense, a non-negative primary balance may be considered a minimum standard for fiscal policy. (Unless redistribution from future to current taxpayers is considered desirable or unless there is a positive net worth to be distributed.) A more aggressive standard would be to require a reasonable contribution toward covering the negative net worth. In a recent paper, I have calculated U.S. primary balances and government net worth for 1948-1989 (see Bohn 1992). I find substantial negative primary balances for 1982-1986 and positive balances for 198789. The average over the 1980s is almost exactly zero. Interestingly, net worth declined by almost $1.7 trillion between 1981 and 1989, largely due to capital losses on mineral rights, gold, deposit insurance, and other items. More broadly, the U.S. primary balance has been oscillating around zero for most of the post-war period. Thus, U.S. taxpayers have not provided a meaningful contribution toward covering the government's negative net worth nor have they taken responsibility for the government's capital losses during the 1980s.

If one wanted to debate in more detail which age cohorts should cover the government's negative net worth, the idea of generational accounting may be very useful. For each age cohort, one could try to compute by how much the present value of taxes exceeds the cost of public services enjoyed by that generation. The difference may be called the cohort's "net worth oriented generational account." It measures the cohort's net contribution toward covering the initial negative net worth. From this perspective, generational accounting and net worth calculations are complementary and not at all incompatible as tools for evaluating fiscal policy.

Generational accounting definitely is a worthwhile and insightful computational exercise. The main problem--apart from not deducting the value of government services from net taxes--is that the existing generational accounts do not pay sufficient attention to the question of which current government plans are likely to be executed by future governments. Net worth and primary deficit calculations do not distinguish between different age cohorts but they provide precise measures of how current taxpayers, on aggregate, redistribute resources to or from future taxpayers. In addition, the concept of net worth is directed towards identifying government promises that future governments will be obliged to honor. Data on government assets and liabilities therefore should help us--as citizens and as generational accountants--to form better expectations about the likelihood that the current government policy plans actually will be implemented.

Demsetz: Thank you, Henning. Earl?

Thompson: I have spent over two decades working on the subject of intergenerational distribution, and I am just now preparing to publish some of my findings. But I cannot let the previous presentations go without comment. So, first I am going to react to what I have heard here. This will motivate getting into some of my own work.

My problem with the presentations of both Kotlikoff and Bohn is that they have no theory of what the distribution should be. They provide no criterion determining how much the poor relative to the rich (the young relative to the old; the future generations relative to the current generations, etc.) should have. Maybe the next generation should have nothing. There is no theory that tells us they should have anything at all. Or maybe they should have everything, and we should have nothing.

We are economists. Usually we require a motivating theory before we make a policy statement. Many statements are made, and not just by Kotlikoff and Bohn but by almost all the authors in this area writing on intergenerational distribution and deficits, without a theoretically explicit, acceptable goal.

Since they do not have a defensible goal about what the distribution should be, why do they spend all this time decrying the fact that the future generations are becoming worse and worse off? In fact, it is extremely easy to see why it would happen. It would happen because in the glory years of the 1950s, which are the years that Kotlikoff and implicitly Bohn were praising, we had just won World War II. Voters were extremely grateful to the young, the soldiers, for winning that war, for risking so many of their lives. The general feeling of gratitude, which prevailed everywhere during the late 1940s and the 1950s and the early 1960s generated wonderful benefits to many of us here.

What happened in the late 1960s and early 1970s was the failure of our ground soldiers in Vietnam. What won the Gulf War was high tech professionalism. The ground soldier was not responsible for the victory. So this society has much less collective benevolence toward the young, and that is reflected in the attitude of the politicians. It is reflected in not only our fiscal policy, not only in our intergenerational distributions, but in all economic policy, including labor, international trade, environmental, and monetary policy. We are looking at a very natural set of policy changes. It happens in history time and time again. It is no big surprise. It is very easy to recount the success of young males in terms of their military contributions. When young males really are not very militarily productive and they do not make great net military contributions, typically they are ground down. That is a normal state of affairs.

We have been living this American dream because we have been living with mass civilian armies that have been tremendously valuable, that have made tremendous war sacrifices. Now that age has come to an end, and we are listening to a bunch of dreamers who have no understanding, in my opinion, of what an optimal distribution would look like. There is no formal structure in their arguments. There is only "Gee, wouldn't it be nice if we lived in the good old days." Well, the military technology supporting those days is obsolete in this country. And if the politicians being criticized were here, they would, in effect, say: "We have been working with these numbers for several decades now. We know these numbers and their relatively harsh implications for the welfare of future generations. Now you want to change the numbers so we do not understand them. What are you trying to do? Are you really trying to help us by changing our accounting system to something we do not understand."

What we have here is just another example of the policy superiority of the subconsciously rational, Pascalian, minds of real-world-disciplined politicians over the pseudo-rational, conscious minds of academically disciplined intellectuals.

More basically, the inability of welltrained economists to think straight about intergenerational distribution is a reflection of a complete absence not only of a historically relevant theory of public choice, but also of a cohesive theory of social organization within which we may see the huge potential role for economics and economists in the broader scheme of things.

What differentiates my own policy analysis from the metaphysically "rationalized" world of standard policy analysis is simply that mine is grounded by a cohesive, thoroughly rational, social theory (Thompson-Faith, 1981). Economists and related social thinkers ideally fit into this world as sellers of information, allowing social leaders to achieve their ideal social equilibrium, one with complete information. Although severe intellectual cartelization has prevented existing economists and related thinkers from filling this ideal role, a practical decartelization of the nonexperimental sciences exists that would induce such a social optimum (Thompson, 1996).

Now, in the context of either the actual or the optimal world, the subject of economics differs from the broader subject of political economy in that "economics" proper accepts the political system--and its politics-dependent distributional goals-as given. In contrast, the subject of "political economy" seeks out changes in the political system enabling the society's underlying military leaders--e.g., kings, tribal chiefs, presidents, etc.--to achieve their personal optima. While the above discussion of intergenerational distribution clearly takes the political system as given and thus can be thought of as economics proper, a broader discussion of the topic of intergenerational distribution may also be warranted here in view of the increasingly widespread dissatisfaction with our political system.

Allow me to introduce the political-economic discussion with a simple example. I call it tribal underdevelopment. Economists do not really explain why tribes--including the myriad of social organizations that abound in Australia, Tropical Africa, and Siberia--have stayed underdeveloped for thousands of years while their neigh-bors have blossomed. If you look at these tribes, typically you see a virile, aware leader. The tribes should do very well, but they do not. Sometimes they have well-developed markets; sometimes they do not. They generally have mixed economies. They are very good at collecting taxes. Look at their neighboring societies that develop great civilizations. The ostensible leader there barely can walk. You see a feeble king, or a vain queen, or a small child who needs a self-serving regent to run the country; you seldom see a virile, aware leader. Yet it is the latter societies that have become rich. So there is something paradoxically disastrous about a social organization that is tribal. And it has not yet been identified.

Do you think this paradox might have something to do with welfare across generations? Given your training, you probably would say that intergenerational efficiency has relatively little to do with it. But recognize that tribal chiefs and leaders of large societies share a common problemthat of making fairly sure that their own children are going to live well. In a system that provides for sharing all real property, such as a socialist system, the tribal chief has a very difficult time ensuring the welfare of his children because the next generation's leader and main surplus recipient will be determined by a contest between his children and the children of his tribesmen. Recognizing that relatively poor males have a substantial underincentive to educate their children, the tribal chief's solution is very simple. It is to let the poorer tribesmen educate their children in their own way--laissez faire education.

A lack of parental incentive is a basic economic inefficiency within the family. When the father educates his child he devotes his own resources to generate benefits that go to the child, not to himself. There is a large positive externality here. The father has a corresponding underincentive to educate or grant bequests to his children. The parent's underincentive to benefit his child is a significant laissez faire market failure. It will happen in all families. It is an unavoidable laissez faire inefficiency. That is a theorem.

Because of the severity of the education externality among relatively poor families, the tribal chief will say, "I will take care of my children. I am wealthy, and I can afford to fully educate my children. I will train them to be leaders like myself. And you other tribesmen, you too can take care of your own children." That is laissez faire education. What will it do? It will make the poorer tribesmen do just a comfortable minimum for their children. They will have their children imitate them or helpout in the hunting grounds. They will thereby teach their children their own skills and duplicate the technology generation after generation. The tribe will remain with the same technology that it inherited a few thousand years ago. It will go nowhere.

Now, consider the society that has property rights. Economists love property rights. But why? They have a theorem about Pareto optimality and competitive equilibrium. But the economic surplus gain over socialism under this theory is of small magnitude and ambiguous sign under realistic transaction costs. The reason property rights are historically so important is that they give current social leaders a way to pass on wealth to their children. The leaders will give, say, half of the society's wealth to their children. Maybe their children are not capable at all, but there is a property rights' religion that is very powerful within their civilized, non-tribal society. That is the difference that property rights mythology makes within any existing civilization.

Property rights protect the resources of the current social leaders' children. Once their children are secure, the leaders will worry about the efficiency of the whole society. They will provide efficient levels of education for the other children, possibly forcing the children to compensate their elders for sacrifices incurred in making the education available. By letting the other children improve their knowledge beyond that of their parents, the leaders are allowing the society to progress to superior technologies. These social leaders have totally progressive political instincts. They will come close to doing the efficient thing for their society, not quite the efficient thing because of informational problems, but they definitely will allow their society to progress out of tribal underdevelopment.

If you think I am talking only about some little tribes, take the example of the U.S.S.R. Here you have a very large tribal society. Suppose you are a leader of the U.S.S.R., in the 1950s, 1960s, 1970s, and you are trying to take care of your children. How do you take care of them when you see that they are about to be ground down by the system? How do you protect them? You make sure that the other bureaucrats who are coming into the system know little about running a government. That way, the existing leaders' favored children, who are granted mid-level positions in the succeeding bureaucracies, are not embarrassed by their bureaucratic cohorts. But, soon the country has a grossly inferior leadership. Soon the people who are running the country are a bunch of engineers with little organizational ability. There is no way they will be able to run a big socialist state. So that state fails because of a growing scarcity of talent in its managerial hierarchy. The scarcity occurs because the leaders educate only their own children, who will do well because they are rewarded according to their relative future bureaucratic talent. Such educational underinvestment is a form of tribal underdevelopment. More basically, the system fails because it does not have inheritable property rights. Such property rights allow the leaders to perpetuate their family's wealth without having to miseducate their entire society in order to do so.

Our civilized society, with its private property system, although escaping tribal underdevelopment, does not reach full efficiency. So, in a way, I am sympathetic to what Larry is saying. But his work contains no intellectual basis for reform. The problem needs an intellectual basis because it is a very important problem. Moreover, the solution to the problem requires political, as well as economic, reforms.

Our society is not helped when we go to politicians and say, "Gee, you are screwing those future kids. Please stop." The politicians know what they are doing. They are clever people. We cannot convince them by saying, "You are really stupid and know little of what you are doing. Here are some numbers that show you what you are really doing." Instead we should say, "This political system must be reformed."

The system is not just inefficient with respect to intergenerational distribution. It is developing inefficiencies everywhere. Professional lawyers are taking over the system. Professional economists are joining in the merriment. It is not going to last long. If we use economics productivelyand I am sure we can--we can induce the society to reform in an efficient way.

When the political system is subject to change, what are economists going to say, unless they have developed some basic insight about what destroys civilized societies? What, in particular, are the underlying political inefficiencies that create the exploitation of children and the takeover by professional rent-seekers?

Returning to the work of the above speakers, when they ask you to support them, what are they really saying? They are saying, "We all can get a good job here doing some simple present-value accounting. We can run an accounting game and get a piece of this really nice action." I say this is cartel-based rent-seeking; this is corrupt. What we should do is try to reform the society, not add to its corruption.

This is just an outline. I am not giving you the reforms, but the reforms are there. There are some very specific political and economic reforms that would end the exploitation of children in a first-best, Pareto optimal, welfare-maximizing way. The current generation of economists are so far away from even setting up the problem, that I do not know whether they will listen to me at all, but solutions do exist.

Demsetz: Thank you, Earl. Everybody will have to read his book when he gets it published.

Andrew, you are the suffering one. Come up here.

Samwick: Funny how I did not think that I was suffering until I got here today.

What I can offer to this discussion is a perspective of what seems to be a different generation from those otherwise represented on the stage. They call us "Generation X," so I guess that means we finally learned algebra. I would like to praise generational accounts. I have at other times been critical of Larry's research, but that was in the context of more academic issues. I have not seen a greater improvement to public discourse about fiscal policy than the introduction of present value accounting into the government's discussion, if not, unfortunately, into their documents.

I have taught for one full year at Dartmouth, and I have enjoyed it. I get to teach the theory of finance. Students who fail to understand the positive net present value rule for investment repeat my class until they do learn it. Possibly, I can flunk the entire Congress. Because policymakers so often fail to consider events beyond the five-year budget horizon, the work that Larry and his co-authors have done is extremely important.

I always marvel at the 84 percent number. I think that wakes people up. I will take Larry's generational accounting story as a point of departure, and I will ask whether there is any way we could make some Pareto improvements just by setting things up a little differently in the realm of old age survivors insurance. Is there a better way for the current generations than there was for past generations? Would it make sense to try to tailor the system to the economic realities of the next 50 years instead of the economic realities of the last 50 years?

I have been developing a growing appreciation of intertemporal choice as an academic issue to research, and a lot of the work I have done has focused on that issue--trying to figure out why people save, what factors contribute to a high savings rate, and under what circumstances do high savings not necessarily mean greater welfare for current or future generations.

Today, I will describe a more flexible retirement system for the current generation of workers that can be implemented by privatizing Social Security. The reality of the situation is that in a pay-as-you-go system, everybody's retirement income is pegged to growth in the population and growth in the average wage. I do not think that those are solid foundations on which to build our retirement. It has been a long time since there was a high fertility rate, and we do not seem to welcome immigrants anymore, so the population is not going to grow too rapidly. The government also has been doing its level best to prevent firms from writing contracts that allow for strong wage growth. Firms also have been trying to push compensation into non-wage forms during the past 20 years, so again we will not see substantial future wage growth.

The current generation of workers, especially young workers--we "generation Xers"--is alarmingly pessimistic about the future of Social Security. I do not know whether to believe the survey results that say that people in my generation think that Social Security will not be around for them. I am not one of them. I just do not think I will get much from it. However, I could be in my parents' generation and still say that to some extent.

I get the feeling that what we are doing is throwing good money after bad. We know that the implied rate of return is lower than the opportunity cost of the funds we are putting in. The onslaught of the baby boom generation may break the bank. I cannot think of a better time to try to devise a more flexible system to provide for the retirement of future generations.

So what prevents reform? The trouble is the generational linkages that are inherent in an unfunded pay-as-you-go system. Political constituencies and the lack of political credibility prevent us from tampering with the system where tampering would be so broadly defined as to include phasing in a different system for the current generation of young workers while leaving the promises inherent to the older workers and current retirees unchanged. It is a shame that such a reasonable proposal is beyond the realm of what is possible to discuss. Nevertheless, I suggest two minimum requirements for an approach to reform. The first is that we do not repudiate any existing promise to the current beneficiaries of the system. The second is that we consider proposals that fundamentally reform the system for those who are not yet collecting benefits from it. The pitfall with any such attempt to change the system is that there will be no change in benefits among the current retirees but, by definition, there will be a reduction in contributions under the existing formulas for current workers. In a pay-as-you-go system, the trust funds become depleted, and the government's total debt obligations increase.

Mandating that some cohort or even every cohort in the population bear the burden of this increased indebtedness would kill the reform proposal, because nobody likes to be mandated to pick up the tab for anyone else. The solution, then, is to make the reform voluntary and to structure the reform so that those who "opt out" of the old system by reducing their contributions in exchange for lower promised benefits bear the entire burden of the increased indebtedness. A voluntary system like this really is not that crazy an idea. Let's consider the trade-offs that people are willing to make.

We have a group of young workers that is dissatisfied with its prospects. The dissatisfaction should be linked to the low implied rate of return in the system compared to what people generally believe they could get outside the system. We know that the legitimate reason to have a Social Security program is to force people to save. If a worker does not save, then he will appear at retirement unable to work and without the assets to support himself. The rest of society will support him anyway because we typically cannot commit ourselves to let him starve in old age even if he could have avoided his plight with a little forethought. Given our failure to commit, the appropriate policy is to force workers to save a reasonable amount for their own retirement while they are young. In terms of reform, we must make sure that no opportunities arise for workers to reduce dramatically their total retirement saving while the future obligations of the public system are being pared down. This is an easy condition to meet simply by requiring all of those who opt out to place at least the former amount of their contributions in another retirement account that they can invest as they choose.

The outline of the plan (presented and analyzed in detail in Feldstein and Samwick, 1995) is as follows: Workers currently pay a payroll tax. About 12 percent of that tax goes for something that looks like a retirement and survivor's account. If workers do not want to participate in the alternative system, they do not have to. They can continue to pay their 12 percent and stay under the old system. If they are willing to save at least a specified amount, for every additional dollar they put in an alternative system up to a particular limit, they will buy out of some fraction of a dollar in contributions they otherwise would have to make to Social Security. Their Social Security benefits will be reduced proportionately.

For example, let's say that every dollar a worker puts in above some threshold buys him out of 50 cents worth of payroll taxes. Let's assume that the threshold is 6 percent of his current payroll. Let's also assume that he is allowed to buy himself out of 9 percent of his 12 percent payroll tax. If he puts 24 percent of his income into this dedicated retirement account, then he will forego his entire benefit and nine out of 12 percent of his contributions.

How would I implement such a system? All the worker would have to do that he does not do now is to tell his employer how much to pay in payroll taxes. Under this simple, quasi-linear system, that uniquely would identify how much of the buy-out the worker wants to participate in. For example, the worker would just tell his employer that he wants to take out 24percent instead of 12, and that uniquely would identify him, conditional on his payroll level, as being fully bought out. Then he just would tell the Social Security system where he wants it invested. This sounds a lot like the 401(a) account that I have with my college. If a college professor can figure this out, normal people can too. That's the system.

This proposal is neutral with respect to the size of the bureaucracy that handles the current system. Ambitious reform proposals could privatize the administrative and record keeping functions of the Social Security Administration, possibly by subcontracting them out to financial services companies. But the Social Security Administration also could administer the reform. The only additional information requiring administration is that pertaining to where the investment funds are channeled.

So what makes me think that this type of privatization is possible? Some international comparisons can be made. The tax set-up I have outlined is similar to what has been done in the United Kingdom since the mid-1970s. Across the Atlantic, it is called "contracting out," and it was the subject of the first economic research project I ever did-my undergraduate thesis. The investment arrangements I have outlined are broadly similar to the reforms that have more recently been implemented in Chile, where mutual funds developed almost overnight to invest the new contributions. Given the comparatively well-developed financial markets in the United States, privatization would be that much easier here.

Returning now to the academic aspects of privatization, there are two critical issues. The first is whether there are scenarios in which the parameters of consumer preferences are realistic and a sizable fraction of eligible workers voluntarily would opt out of the existing program. Not surprisingly, it is possible to construct a privatization scheme that is generous enough to get workers to opt out. The second and more contentious issue is whether any of these scenarios exhibit the property that national savings do not decline in the process of getting workers to opt out. If the change in national savings is positive--i.e., the higher savings required of those who opt out exceed the new debt issued--then there is the potential for a true Pareto improvement via the privatization.

Our research thus far has confirmed that there are scenarios compatible with these two criteria. It is possible to implement this broadly defined privatization scheme without incurring a reduction in national savings. As with any policy reform, however, making the transition from one policy environment to the next also is critical, so many questions remain to be answered. One important question is how much scope exists for shifting existing assets into the retirement account. If most households have large stocks of financial wealth that they already are accumulating for retirement, then to constructing a privatization scheme that keeps national saving from decreasing will be difficult, since most of the dollars flowing into the new accounts will not represent new saving. A similar question concerns the current scope of employer-provided pensions: Would employers and employees find it mutually beneficial to terminate existing pension plans in favor of using the funds contributed to the pension to opt out under the privatization scheme? The privatization scheme probably should not be so generous to induce pension fund terminations, which also would serve to deplete national savings.

Both of these questions are pieces of a much larger puzzle that is very relevant for "Welfare Across the Generations." The larger puzzle is: Why do people save? Is the decision to defer consumption over time the result of people optimizing against economic constraints? If so, what are the contributions of different motives such as retirement, downpayments for home purchases, children's education, bequests, and uncertainty of income, lifespan, and health? Or is the saving that we observe just the residual from other aspects of individual behavior that economists have been reluctant to consider formally? If we do not know why people save, how can we predict their response to a new savings opportunity?

Answering these questions is essential if we are to formulate efficient and fair economic policies for Social Security. I do not believe that we can conduct any constructive policy discussions about welfare across the generations until we face up to the looming consequences of the current system on the economic well-being of future generations.

Today, I have discussed one type of reform proposal that uses a voluntary buyout option to privatize the retirement portion of Social Security. It attempts to walk the very fine line between keeping promises to the current generation of retirees and allowing current and future generations of workers to plan for their own retirement in a way that addresses the realities of their economic environment, rather than that of their parents. Even if this proposal turns out not to be politically viable, I hope that Social Security privatization is an issue that remains at the forefront of the policy discussion. If we do not break the generational dependence built into the current pay-as-you-go system, it certainly will break us.

Professor Bohn mentioned something about appropriating the productivity increases of the future generations. I wrote that model down, too, and I solved it, and it looks interesting, but I have not yet met a parent who would do that to his own children. The fact that I sometimes get to appropriate that money from somebody else's children does not make me feel any better about doing it. There are lots of Pareto optimal equilibria. Some of them strike me as a little bit more credible than others, and I am not willing to jettison that from my research agenda or my thinking on the topic.

As far as this goes, the research implications are clear: Tell me why individuals save, develop a tractable model in which to analyze those policies, and, above all, press for accountability-generational or otherwise--in the conduct of public policy.

Demsetz: Thank you, Andrew. Larry? Kotlikoff: I am not sure which comments I should disagree with first.

One of the things that Andrew suggested is that we are one big happy family, and the kids formerly took care of the parents and now we just are using Social Security to take care of the parents. That is an interesting theory, but it does not square with the facts.

Shifts over time in the age-consumption profile show that there has been a dramatic increase in the consumption of older people relative to that of young people. If we were one big Barro-Becker happy, altruistically linked family in the United States, we would not see this increase in the relative consumption of the elderly.

Also, when we look at microdata, the ratio of consumption of parents to that of their adult children turns out to be very strongly related to the ratio of their incomes. If we are one big, happy, altruistic, extended family, we should find that the relative resources-or the relative income of the members of the extended familydo not determine their relative consumption. So I think there is extremely strong evidence against that kind of Barro-Becker altruism model. I cannot believe that anybody would take it seriously at this point, given all that evidence.

Second, as an economist, I am not here to try to say what is the ethical intergenerational distribution of resources. I was taught, when I went to graduate school, that we cannot make those kinds of statements. Rather, our role is to try to identify the consumption possibilities frontier and that is what generational accounting is doing. It is trying to present the implications of current policy for future generations.

Third, Henning is bothered by the particular counter-factual experiment where we ask, "If current generations are treated as current policy suggests, what will that mean for future generations?" But we have run all kinds of counter-factual experiments. We have asked, for example, "What kind of immediate and permanent federal income tax hike would we need to achieve generational balance? What would get that 84 percent number down to around 40 percent and the number for young and middle-aged people up to around 40 percent--so that we would achieve generational balance and never would have to raise taxes again in the future. It turns out we would need about a 52 percent immediate and permanent hike in the federal income tax.

Another thing that Henning suggested is that somehow the debt and the deficit are the right concepts. He is saying thatgenerational accounting is okay to do on the side, but, there really is an economic theory that says that the debt and the deficit are okay to use. I would like to sit down with him for about 10 minutes with the Economic Report of the President and ask him which debt and deficit he is talking about, because I can flip to the back of the report and within the space of a minute, show him about five different deficit series. And if he gives me a bit more time, I will produce another couple hundred. Just because the government accountants have taken particular transactions and labeled them in a certain way, does not preclude us from taking a look at the same numbers and labeling them differently and producing another time series of what we call debt and deficits that is no less economic than the official ones. There is not just one official series in the back of the Economic Report of the President, there are a number of them. If you look at what the Office of Management and Budget calls the debt versus the National Income Accountants definition, they are quite different, trillions of dollars different.

Henning also suggested that certain government liabilities have more or less commitment associated with them. True. But I do not think that their probabilities of payoff are connected with the word debt and deficit. For example, in the 1970s, the federal government used inflation to default on about a half a trillion of debt. If you look at the 1982 Economic Report of the President, you will find an analysis of this default. Liabilities of the government are not safe, and have not been safe. So I strongly disagree with the proposition that the debt and the deficit are well defined because of the nature of their contingent claims.

Let me turn to Andy's comments about Social Security reform and Chile. First, there is a growing discussion in Washington, in which I have been participating, concerning reforming Social Security for younger people. But I think it would be misleading to think that reforming Social Security is a way to get out from under our generational imbalance problem. What happened in Chile represented, in large part, relabeling of the existing fiscal set-up. The Chilean government in effect told workers, "You do not have to make your contributions to us. Instead, you can contribute to private pensions." And then what did the government do? It borrowed this money out of the private pension system and gave it to older people. So in effect, the government merely changed what it called the contribution. It started calling the contribution a loan to the government rather than a tax. The "reform" was, to a large extent, simply a big shell game involving no fundamental change in the distribution of fiscal burdens across generations. By pegging a very high real rate of return--roughly 6 percent-on the debt that the government issued to these pension funds, it may actually have worsened the imbalance in generational policy in Chile. I think there are some very substantial benefits to be had from privatizing Social Security, but it is easy to overstate them. But I do look forward to looking more closely at your proposal, Andy, and I hope you will look at mine.

Demsetz: Henning? Bohn: I am happy to realize that Larry and I actually agree about something. The first thing we agree upon is that there is no general criterion of what is fair, which generation has the right to appropriate which resources. That leads me back to my point that we are dealing basically with a game in which different generations try to use the government as a tool to redistribute resources. We see that within a generation there is a lot of redistribution going on between different people. Why should we be surprised if this is going on between generations, too? We really do not know what the right intergenerational distribution is. Second, I do agree that we are not living in a Barro-Becker world. However, if we want to quantify the amount of intergenerational payments that take place, we must recognize that there is some nonzero degree of substitution between government transfers and private transfers that affects the quantitative magnitudes of redistribution that takes place.

However, I do think, Larry, that in your published writings, you make the implicit normative claim about generational accounts that if the tax rates that the unborn and the newborn pay are equal, that seems fair and right. So I think there is no less of a normative claim in generational accounting than in any of the deficit numbers.

Where I really disagree is about the importance of measuring government liabilities. If we think of the government as an agency that can redistribute resources, it is absolutely vital to figure out what the pecking order is in which people can make claims on government resources. I think the pecking order is fairly clear. Government bonds and Treasury Bills come first. The government will pay off its debt before it will pay off virtually anything else.

On the next level, people seem to make a distinction between items that we call entitlements and other budget items. Much of the academic and political debate is really about that. If different people have different perceptions about the pecking order among these items, having an open and public debate about each one of them is socially valuable to clarify their status. This is important for deciding which items ought to go on a government balance sheet and which ones should not. It is equally important if we want to construct meaningful generational accounts. Otherwise, if Larry complains about the indeterminacy of the budget deficit, then I would say there is just as much indeterminacy in generational accounting, because you can run 5,000 different simulations with 5,000 different assumptions on what the government might do in the future. Overall, I believe that measures of government net worth and government liabilities are essential to give us a method to determine which government transfer payments and which government debt repayments actually will be executed in the future.

Let me make one comment on the arguments of each of the other two speakers. On Earl's point that the politicians do not want to change the accounting system and that it is just pointless to discuss this, I disagree. Clearly the government and the politicians do not want to change the accounting system because we have an accounting system that holds nobody accountable. Our first priority, if we want to measure what the government is doing with our tax money, should be to adopt an accounting system that the politicians cannot manipulate.

With respect to reforming the retirement system, I am somewhat skeptical. You can do a lot of calculations but there is one big problem that I have not seen anybody resolve and that is the problem that we are faced with an old generation that has not saved, that has put funds into the retirement system, and that now is expecting returns from the system. Somebody must take care of them. If you want to privatize the system, you must find some generation that pays the retirement benefits to the old and accumulates its own savings at the same time, instead of relying on the next generation's savings. Social Security is difficult to fix because a system has been put in place that has made promises to the old generation that now are outstanding. This is difficult to undo.

Demsetz: Earl? Thompson: If you used this present value accounting system, which depends on all sorts of projections in the future and employs a large number of economists, I cannot imagine that would be a more uniform, more non-controversial, more politically simple system relative to the present cash flow system. I really do not think that this is a move toward simplicity even though it is conceptually very nice. That probably is where Larry and I agree, conceptually it is very nice. We both have in mind the same capital theoretic model where people are discounting future tax liabilities, and we do not have this Barro condition, this Ricardian condition. We are looking at the same world but in a much different policy framework. I honestly do not think that politicians are so feeble minded that they do not know what they are doing. In fact, if you are right and we do have a negative net worth, aren't we better off surviving the way we are now rather than declaring bankruptcy?

Kotlikoff: Well, having testified four times this spring to these congressmen, I can personally attest to their feeble mindedness.

Thompson: That is your perception. Maybe they are so clever that they have you convinced that they do not know. Demsetz: Andrew? Samwick: Regarding Larry's point about the recent reforms in Chile, I meant only that they had set up, at least notionally, a different way to channel the funds. I think we still are a little bit away from that level of corruption in government. However, I always have marveled at how people could somehow be relieved to know that their Social Security trust fund was invested in Treasuries.

As far as the double retirement issue goes, that always and everywhere is financed out of making better investments.

By moving toward a fully funded system in which Social Security contributions actually increase the productive capital stock, we simultaneously can retire the debt associated with the benefits of the current generation of retirees and allow future generations of retirees to earn higher rates of return on their contributions.

Bohn: A free lunch, in other words. Samwick: No, it is not a free lunch. Is your argument that the implied rate of return on the current public system is higher than what people could get elsewhere? If it is not, then you are going to appropriate some of that surplus to pay for the double retirement. That is the only way you ever will get reform, and that is the only reason you ever would want reform. That is the issue, and it is hard for me to imagine that the current system provides a rate of return that is high enough for that condition not to hold. The explicit requirement of any reform proposal is that you do not have to buy the support of the current generation of retirees because you do not want to change their benefits. The criterion I also was using to evaluate different reform proposals was that national savings, however classified, do not go down. Demsetz: Larry? Kotlikoff: One last response to Henning. If you write down a rational intertemporal dynamic model--whether or not it has a game-theoretical framework-that model will produce a certain generational distribution of resources and utility over time. One can describe the equations of this model in English or French or Japanese or Chinese without changing what the model tells us, i.e., the economics of the model.

Take any such rational model where labels do not matter--i.e., where the language of describing the symbols in those equations does not matter to the people in the model. Now you can relabel the terms in those equations to produce any deficit or debt measure you would like and any associated time series of deficits. That is what economic theory tells us. It tells us that the deficit is not an answer to an economic question.

Generational accounting is trying to answer an economic question. It is not just accounting. It is trying to answer the economic question of how the utility of future will compare with that of current generations. If you change the labels or the fiscal language, you will not change generational accounting's answer to this question.

Demsetz: I want to thank the audience for listening so attentively to all the generations up here.