Transcript

Washington, D.C., June 8, 2004

Participants:

Marcelo Selowsky, Independent Evaluation Office, IMF

Allan Meltzer, Carnegie-Mellon University and American Enterprise Institute

Nancy Birsdall, Center for Global Development

LOUNGANI: Let's get started. I overheard comments on how small the lunch plates were and how tiny the sandwiches. That was just to assure you that you are at the IMF, and this is a forum on fiscal austerity.

Our feature speaker, Marcelo Selowsky, is Assistant Director of the IMF's Independent Evaluation Office, set up in 2001. The IEO operates independently of IMF management and at an arm's length from the IMF's Executive Board. Prior to joining the IEO, Marcelo was the Chief Economist of the Europe and Central Asia region at the World Bank. Marcelo?

SELOWSKY: Thank you, to Prakash for organizing this seminar, to Nancy and Allan for joining in, and to all you for attending; a particular welcome to those from outside the Bank and the Fund. As Prakash mentioned, the IEO is a new group. We are very small, and our objective is to undertake evaluations that can provide recommendations for the IMF to improve its operations.

In this evaluation of fiscal adjustment in IMF-supported programs, we looked broadly at two issues. The first, which is sort of a response to outside concerns, is the quantitative dimension: to what extent do fiscal targets follow a one-size-fits-all approach, leading to excessive adjustment? Second, we looked at the qualitative dimension: to what extent are there issues regarding the efficiency, sustainability and equity of fiscal adjustment?

We identified two main sources of excessive adjustment. The first is that IMF projections of foreign aid in the poor countries may be too conservative - there may be too much aid pessimism - and reflect too much concern about Dutch disease, and so on. This may result in a programmed reduction of expenditures which is more than is required.

The second source -- which is a little more subtle, more of a Keynesian type of argument -- is the extent to which fiscal targets may generate an unnecessary shortfall in aggregate demand. So the first source is a problem of aggregate supply; the second one is a deficit of aggregate demand, vis-a-vis supply, relative to the extent of financing possibilities. To what extent may the fiscal targets be generating too much of a shortfall in aggregate demand when you combine public-sector demand and private-sector demand?

The second issue is: for a given amount of quantity, what can we say about the quality of fiscal adjustment? There may be some tradeoffs between quality and quantity of fiscal adjustment. Fund programs usually have short horizons, at least the standby arrangements, and high-quality fiscal adjustment usually requires more time. So this probably creates some tensions that we want to explore.

In order to carry out this assessment, we used data on about 133 programs. We used this cross-section of programs basically to do some of the quantitative work. Then we used some case studies to study the more qualitative aspects of fiscal adjustment. Here we looked not only at a program in a particular year, but also at the dynamics: what was the context in which that program was put together? We looked back to the surveillance activity-the economic monitoring and policy advice the Fund provides every year in each country-to see if there was a relationship between the current program and surveillance in previous years. Was there a link between the program at Year T and surveillance activity at T-3, T-2 and T-1? We looked at the implementation of specific fiscal reform measures under those programs.

We also took some ad hoc samples. We took 20 concessional programs in sub-Saharan to look into whether there is excessive pessimism about aid flows. To look at how social spending fares under programs, we looked at social spending in years with and without programs in 140 countries over the period 1985 to 2000.

Now, so I can finish in half an hour, I'm going to just take you very quickly over the main findings without getting involved with data issues.

First, we didn't find support for the `one-size-fits-all' criticism. There is significant variability in fiscal targets.

Second, there's also no evidence that programs always involve austerity. For example, in 40 percent of the programs, current account deficits were allowed to increase as part of the program and, in about one-third of programs, primary deficits and primary expenditures were allowed to increase as part of the program.

Third, we didn't find evidence in the PRGF countries-these are the programs in the poor countries, in sub-Saharan Africa-of a systematic underestimation of aid flows.

Fourth, we found that the variability in fiscal targets systematically relates to initial conditions and financing prospects. In other words, the higher the country's initial disequilibria, the stronger is adjustment required in the targets, and the adjustment is related to the prospects for external financing. The more external financing that is projected, the more relaxed can be the fiscal target. Specifically, we found a coefficient of 0.2; in other words, that if additional external financing equal to 1 percent of GDP is projected, 20 percent of it is allowed to go into a higher public sector deficit. So, in that sense, the programs only allow only a fraction of the additional external financing to go to the public sector. But the effect is symmetric. When external financing goes down by 1 percent of GDP, only 20 percent of it is borne by the public sector. So programs try to dampen the impact of fluctuations in external financing on public sector deficits.

Fifth, we found that fiscal targets are flexible. In about two-thirds of programs, targets are revised. They are usually associated with revisions in growth prospects. But there is an asymmetry: fiscal targets are usually relaxed when growth is below expectations but are less likely to be tightened when growth exceeds expectations.

But the problem that we found is that programs do not clearly explain how the pace and magnitude of fiscal adjustment is determined. When targets are revised, the rationale for the revisions is many times not clearly brought up in the program documents. What part of that need for adjustment is due to exogenous factors beyond the control of governments? To what extent part of it is due to a weak policy effort? We call this a lack of story line, and we'll come back to that later on as one of the important recommendations we make. Let me continue with the findings.

The findings thus far are based on an analysis of fiscal targets. Now, what happens when we compare targets with reality? We find that, on average, programs achieve half of the second-year program targets. So the record is pretty mixed. Almost all of the fiscal adjustment is achieved in the first year. This is sort of important. It seems we get like diminishing returns by the second year. This is already something that is food for thought.

There is significant variability in achievement of program targets: sixty percent of programs underperform. The features of programs that underperform are that there was almost no improvement in fiscal balance, and the shortfall reflects weak performance rather than ambitious targets.

And a large part of the shortfall is due to optimism about growth. Expenditures are sticky, so optimism about growth means that expenditure ratios end up higher than programmed. However, in episodes of large fiscal adjustment, an adjustment greater than 3 percent of GDP, the underperformance is due to shortfalls on the revenue side.

Now, what happens to economic growth during programs? Now here I want to be very careful. This is simply a picture of what happened to growth in periods in which the Fund programs were being implemented; I'm not trying to say that all this is due to the Fund. We found that, on average, growth rate did not decline during program years. But there was significant variability. The capital account cases-the ones that have received most publicity, such as Indonesia, Korea, et cetera-were the ones that had the worst performance. For example, growth tended to slow down in 40 percent of cases in programs, but in the case of capital account crises output slowed in 80 percent of the cases. Now, as I mentioned, this doesn't mean that it's necessarily due to the Fund. For example, one could say that maybe output would have decreased even more without the Fund. We don't have a counterfactual against which to evaluate the Fund. Nevertheless, in our view, it was important to put on the table the factual information about what happens to growth during program years.

We found very significant optimism in projecting private-sector activity and growth, particularly when the program commences in very adverse situations. The Fund is extremely optimistic in projecting the recovery of growth, particularly when the country is going through a very rough period--when the program, for example, is put in place in the context of negative growth or of stagnant economy, et cetera, which in some sense was the case in the capital account crises

Let me give you some data on this; this is interesting. For example, slowdown in output happened twice as often as projected, and negative growth ten times as often as projected. And in one-quarter of the cases, investment ratios was 5 percentage points below projections. So there is always an optimism about the recovery of growth, particularly stemming from an optimism in projecting private demand or private effective demand. There is an optimism about how the private sector is responding to the problem.

Now, we come to this area which is a little more subtle. Are fiscal targets unnecessarily contractionary and, hence, inappropriate? We believe that the overoptimism regarding the recovery of private demand may, in many situations, prevent putting on the table the need to discuss possible countercyclical fiscal policy or impulse. The fact that you're always optimistic may reduce the incentive to discuss the possibility of a fiscal policy that may be countercyclical because you always expect that private sector is going to pick up very quickly in response to the problem.

Also, we found some evidence that there is an overperformance on the external side-that current account and reserves perform better than projected-signaling that external financing and unused external financing may be available at the margin.

So these are one or two signals that some kind of a contractionary bias might be there. However, we tried to be very careful here because we said that they may signal a potential contractionary bias. We have to recognize that fiscal deficits cannot only be determined from a countercyclical point of view; you also may have what we economists call "neoclassical reasons" to consider a tight fiscal stance, particularly, for example, if the country already has a very high debt, where markets feel very pessimistic about the country and where the possibilities of financing, of external financing could be highly affected by increasing the fiscal deficit.

So we have to recognize that the fiscal stance does not depend only on countercyclical Keynesian dimensions, but also depends on how the markets are going to react to increasing the fiscal deficit, how it's going to feed into interest rates and also into a further debt accumulation. We say that there is a potential contractionary bias, but not necessarily an actual one. To make a judgment of the appropriateness of the overall fiscal stance, we will need to consider how will the neoclassical and the Keynesian factors interact with each other to say, look, there was a bias here because, really, here in this particular case, we didn't see a debt problem or a market reaction problem.

The problem is, however, is that the program documents do not explicitly discuss these factors. This is a major finding. For example, they don't explain clearly the links between the fiscal stance and the speed of the private-sector recovery. Or the links between private sector recovery and the other parts of the program; for example, if interest rates are going up can the private sector really be expected to recover so quickly? So the tradeoffs between aggregate demand or countercyclical objectives and market response objectives are not well-explained in documents.

So while you cannot answer definitively the question of appropriateness or inappropriateness, there is a lingering sense of some potential contractionary bias because of the optimism in projecting aggregate demand. Another way of putting it is that it seems that programs have a premature concern about crowding out because it is expected that, thanks to the program, the private sector is going to expand very quickly, and you have to make room immediately and therefore, you need to immediately tighten the fiscal stance. But it seems that the private sector takes more time to recover, and in the interim there may be a potential for this deficit of aggregate demand. So there is a need to have a much more explicit discussion of these factors in order to reduce excessive optimism and provide a better rationale of the fiscal adjustment.

Now, let me move now to some of the qualitative dimensions. We look at three aspects of fiscal reforms. First, the balance among policy measures: do programs tilt toward specific areas while neglecting other ones? Second, progress in implementation. Third, learning from the past and the role of surveillance-an issue that has come up many times in discussions of the Board. Is the institution learning from the past? Is the Fund incorporating lessons from the past into the way it does things in the future or are the same mistakes repeated over time?

On the first issue of balance among policy measures, we found that programs usually focus more on the revenue than on expenditure side. There is significant emphasis on the value-added tax, particularly on increasing value-added rates. We believe that in many situations we may have diminishing returns, particularly when the base is narrow. In some programs, for example, initial-run efforts are at increasing the value-added tax very quickly. It does not have a very good effect on revenues because evasion takes very rapidly-particularly when you increase tax rates a lot-and because the base is low.

The Fund has provided less attention to other taxes and to efforts aimed at reducing evasion in customs, income and corporate taxes. There has not been enough effort or curiosity to document the evasion that many times takes place massively, and particularly when strong vested interests are behind that situation. In addition, the Fund is not forceful enough in collecting from well-known taxpayers with arrears, particularly the powerful. Actions of this front would increase the social support for programs.

So we believe that the Fund emphasizes too much measures to increase certain tax rates instead of increasing collections from the groups or for the types of taxes where everybody knows that there is underreporting of income or nonpayment of taxes. This is very true with Customs. In Customs, we have many countries where massive amount of evasion. And in our view, the Fund has not been forceful enough, in the short run, in pressing for improvements in that area.

Tax administration has focused more on improving know-how than on politically difficult actions to empower tax agencies to enforce payments. And while the IMF is doing a very good job in technical assistance in improving the know-how of the tax-revenue institutions to collect better taxes, the issue in the end is lack of teeth: when those agencies want to go and collect, there's not enough legislation, the judicial system is not working or basically the executive or the Parliament are interfering. And here is where the IMF should be more involved because otherwise the technical assistance, which is extensive and many times is extremely good, is not having enough of an impact.

On the expenditure side, programs have often focused on short-term quantitative targets to reduce the wage bill rather than a reorientation of public spending that would be more durable. Hence progress has been easily reversed.

Let me move on to the second qualitative issue, problem in implementation. In no reform area was implementation satisfactory in more than 40 percent of the cases, and many times the reason is that some reform may take more time than envisaged, extending beyond the program horizons. So this is part of the dilemma. However, other reforms could have been implemented within the time frame with more decisive action by the executive. I'll come to that later on. That is a problem of the composition of the actions. Some of them cannot be done very quickly because it takes knowledge, and it takes administration. But other reforms are a matter of pure political will, and there sometimes we believe that the Fund should be even more forceful.

It is critical, therefore, to unbundle reforms into several steps and follow progress in their implementation over time. Distinguish which reforms can be implemented in the short run with enough determination of the executive and, hence, can realistically be implemented. Other reforms may require more time, for instance if they require passing some legislation. Others might require even more time, for instance if they require and improvement in the implementation capacity of agencies. It's very important for the IMF to have such a framework in place.

Now, what is a natural instrument of the Bank--of the Fund, sorry. I'm used to going to the Bank. I have to remind myself.

What is the natural instrument for the Fund to use to have such a framework? Surveillance, in theory, is the best instrument to do this -- in noncrisis, nonprogram years, surveillance should provide that road map for reforms. But surveillance is not doing that job. This is an big deficiency, and to address it we have several recommendations, as I'll come to in a minute.

Now, the third qualitative issue: learning and the surveillance process. Program documents often fail to evaluate past fiscal performance and policy failures under previous programs. They don't look back into the past. I don't want to give the numbers here; they are in the report.

Moreover, we don't have enough of a link between surveillance and programs. Surveillance is not forceful enough in flagging the need to accelerate reform where implementation is lagging. Basically, you get the impression that some things are flagged in surveillance at, say, T-3, then they disappear at T-2, and then they reappear. So why did they not disappear in between? Is it because it was done? It's not important any more? The Fund gave up? So we don't have consistent follow-up, we don't have road maps.

So let me summarize the suggestions for the Fund's work that emerge from these findings on the qualitative dimensions of fiscal reforms. Surveillance should set a clear road map of fiscal reforms and unbundle constraints to action. Surveillance and programs should mutually reinforce each other over time, with subsequent programs picking up reforms requiring more time. Cumulative progress should always be assessed, so that you become aware of reversals--we should always be moving toward a plateau, in some sense.

Doing things in this way has a lot of advantages for the countries too. This will alert authorities ahead of time on critical reforms to be addressing in case IMF financing is needed in the future. It would provide the country's authorities flexibility in timing and modalities of reform and enhance early domestic debate and ownership.

This is always a challenge: how can you provide incentives for the countries to advance reform in noncrisis periods, in the good years? Now, if the Fund has a good road map, identifying some reforms that are really critical, then it can say to the country: `Look, here are two or three priorities that you have not looked at in the past. They are critical, and we believe that if there is a Fund program in the future, these are going to be the reforms to ask early on.'

So you alert the governments early on, so that you don't have to do those reforms under crisis, and you allow flexibility to the governments in the timing. The governments then can have more flexibility in the timing. They can decide how do they want to--they can do it at a convenient time in the political cycle, they can mobilize some social safety net, and so on. You give them more flexibility by alerting them and being more explicit early on, and that process, in some sense, is not taking place as much as we at the IEO think it should.

Doing things this way would also be consistent with the new initiative in the Fund called streamlining conditionality, which says the Fund should get out of some areas of structural reforms. That also avoids the last-minute bunching of reforms under a crisis situation, as in Indonesia where you had a lot of structural reforms that should have been implemented but not necessarily in the context of a crisis; that case showed the need for streamlining.

Only steady progress on the structural aspects of fiscal systems will reduce vulnerability to shocks and permit a better quality adjustment. If you have a shock in the future for which you have to adjust, that adjustment could have a higher quality if the structural underpinnings have been strengthened. If you need to adjust your fiscal deficit by 2 percent of GDP because something happens in the future, if you have a high tax base you have to increase taxes little, but if you have a very small base you have to increase taxes much more. You have to cut certain expenditures. So the better fiscal system you have in place, the better you can adjust to the shocks. So the sort of process we are recommending would be conducive to bringing about that type of situation.

Let me turn to social issues. We looked at what happened to public spending on health and education as a fraction of GDP, and we tried to see to what extent the mere presence of IMF adversely affects that spending. We find no reduction in this social spending, once we correct for other factors and the endogeneity of the IMF presence. This correction is very important. What do I mean by that?

When the country has an IMF program, a lot of other things are also happening that affects adversely social spending-shocks of other types, declines in output. So we control for some of those other factors to try to see to what extent the pure influence of IMF may be affecting social spending. And we didn't find there was a link. This was quite an in-depth work and everything is in the report.

However, when we looked at the case studies we found that in some countries-not necessarily during the exact period of the IMF's presence but when there was an adjustment program-specific social programs critical to the most vulnerable tend to suffer. Many of them are not very expensive: vaccines, food for the patients in hospitals, recurring costs in the schools. These types of expenditures get preempted by the wage bill, because usually what is protected are the doctors' salaries; maintenance of X-ray machines, buying of vaccines or penicillin, these are what suffers.

We found that this to be case in many countries, particularly with low-income countries. These things are not usually very costly. They are not linked to the fiscal adjustment, but simply, by default, they get preempted. To do something about this requires steady work beforehand. You need to get budgetary protection mechanisms in place that can prevent this from happening.

Now we believe that this has not been confronted by the IMF, but not because the IMF is not interested. It's because the original guidelines that IMF had for social sector spending for the non-PRGF countries are a little bit out-of-date. The guidelines are from 1997, before the Asian crisis, and they are very broad -- all over the place. The present guidelines are unclear and unfocused, and therefore, in practice, while many programs of the Fund have worried about social protection, many times it depends on the interests of the particular person in charge of the program.

In fairness, it's not obvious that the IMF has the knowledge. You know, to understand what's happening to health and education, you have to follow over time; you have to talk to the government. I believe that the actual monitoring of health and education spending over time should be done by the World Bank, not by the IMF. Also, judgments about what social programs should be protected should be made by the Bank, not by the IMF. But the IMF should help countries try to protect critical social spending in the face of shocks. This is part of the mandate of the Fund, to try to work with countries to reduce unnecessary social hardship under balance-of-payment shocks, if you wish.

But for this the IMF will need new guidelines stressing that, during the surveillance process, the IMF should encourage countries-working with the World Bank, USAID, WHO, and so on-to put together systems and budgetary mechanisms that can avoid those unnecessary social spending cutbacks to happen in the future. That should be the focus.

So let me just finish-I think I'm basically on time-with a summary of our main recommendations.

First, documents should provide a clear justification for the magnitude and pace of the fiscal adjustment.

Second-and I didn't spend very much time on this during my talk today-the Fund's internal review process should focus much more on early brainstorming so that the story line on the need of the proposed adjustment is discussed more thoroughly within the institution.

Third, programs should make stronger efforts to identify structural reforms that ought to be carried out during the program horizon-as part of a medium-term road map of priority reforms.

Fourth, the surveillance process should be used more explicitly to provide a long-term road map for fiscal reforms, unbundle the constraints, and assess progress.

Fifth, the IMF should encourage non-PRGF countries in preparing systems to protect the most vulnerable in case of budgetary retrenchment. This will enhance early ownership of these programs and reduce the burden on the IMF of having to make quick decisions or judgments under crisis situations.

Let me say that the staff of the Fund has come back with many specific operational guidelines to follow up on many of these recommendations of our report.

Thank you very much.

LOUNGANI: Thanks, Marcelo. To comment on the report, we have Allan Meltzer and Nancy Birdsall. Beginning with Allan, he is, as you know, a professor of political economy at Carnegie-Mellon and a visiting scholar at AEI. To most of us in this building, he is best known as the chair of the so-called Meltzer Commission, which advocated far-ranging reforms of the IMF and the multilateral institutions. He has served as a senior policy adviser to many U.S. Presidents, including the one who is in our thoughts this week, President Reagan. Allan?

MELTZER: Thank you. It's a pleasure to comment on Marcelo Selowsky's presentation and the report on fiscal adjustment by the IMF's Independent Evaluation Office.

The report sets a high standard for work of this kind. It is comprehensive, careful, and balanced in its judgment. The reader gets a sense of the interaction between IMF staff and the government of the country in adjusting fiscal adjustment programs to events both foreseen and unforeseen at the time of the initial program. This is a very different picture than one gets from, for example, Joseph Stiglitz's book on globalization, and with much greater verisimilitude and supporting evidence.

Those who do empirical work learn early, and often, that no matter how careful the study, there is always reason to ask for tests of alternative and supplementary hypotheses. I don't think that that's the purpose of today's forum, so I will confine my comments of that kind to one broad point. Let me add that no study of this kind can settle issues permanently, as the report notes at several places, we cannot construct the counterfactual to learn what would have happened without the program.

Nevertheless, there is an obvious alternative that the report did not consider: how did countries fare if they were faced with similar crises but had no IMF program? A commonly cited example is Malaysia, following the Asian crisis of the late 1990s. It rejected IMF assistance and introduced exchange controls. Malaysia is alleged to have recovered as rapidly as countries with IMF programs, for example, Korea or Thailand, and more rapidly than Indonesia. Is it true that exchange controls, temporary exchange controls can substitute for IMF adjustment? If so, this would be an important finding and highly relevant in the design of fiscal adjustment programs. We should pause before jumping to that conclusion.

Unfortunately, Paul Krugman and others who cite Malaysia do not pause. They neglect the role of world demand for the exports of troubled countries. During the Asian crisis, the United States' current account balance fell from about a minus 128 billion in 1997 to minus 411 billion in 2000. Few would deny that this large, growing increase in global demand made a major contribution to the recovery experienced in Malaysia and other Asian countries. Perhaps it had more to do with Malaysia's recovery than the particular policies chosen by Malaysia. Perhaps Malaysia recovered despite exchange controls. Perhaps it would have done better if it had adopted an IMF program. Once again, the counterfactual problem appears, but the comparison is still informative.

The same is likely to be true of other countries. Rising world demand certainly influenced the speed of recovery in Korea and other parts of Asia. The only way this shows up in the report is in the country's growth rate or its current account position. This helps, but it would help more if world demand were held constant.

I want to now discuss the Executive Director's response to the report and the report's implications for the IMF. While reading the report and the response, I asked myself, What would I want to know next? The question that jumped out at me was: What causes the lack of success in explaining the difference between the planned or envisaged and actual fiscal adjustment? The report finds that nearly 80 percent of the variance of this difference is not explained.

Simply put, we don't know why countries miss their planned fiscal adjustment. The report offers some guidance. It tells us, Page 61, that "Insufficient progress in structural reform in the fiscal area is an important factor behind shortfalls in fiscal adjustment." Again, on page 64, "Revenue from shortfalls seem to be associated with weak implementation." Later, discussing why economic growth outcomes might be less than projected, it offers four reasons, of which one is that, "Policies may not be implemented effectively." This message recurs several times; for example, when discussing effective collection of delinquent taxes that Marcelo spoke about.

My favorite analogy about economic reform is that reforming countries has something in common with raising children. It doesn't do much good to want for your children what they don't want for themselves. The same is true of countries.

For example, compare the success of economic reform programs with Carlos Bolona in the early Fujimori government in Peru and Minister Lavagna in the Argentine government currently. Minister Bolona was committed to reform and reforms were made. He was able to marshal the necessary support to succeed in transforming Peru's economy. The quotations I read from the IEO report speak to the importance of political support in successfully implementing reform. I believe that represents the critical difference in whether fiscal targets, growth targets, and structural reform are achieved. Of course, outside events matter also.

The Executive Directors make a few passing remarks along these lines, but they strike me as lacking curiosity of why many programs do not succeed and what might be changed to make more of them succeed. For example, they noted, on Page 165, that fiscal targets "were not met in a large number of cases," and asked for a better understanding of the reasons that "most of the programs and fiscal adjustments took place in the first year of the programs with little progress thereafter." But they did not direct the staff or the IEO to find a reason. Again, the Executive Directors, on Page 166, "noted the successful fiscal reforms that would require that the authorities have strong ownership of the process," and that the "ultimate responsibility to develop fiscal reform agenda resides with the individual country authorities."

If these are to be meaningful statements, they have to be implemented by reforming the IMF. The principal change has to be a shift away from current command-and-control procedures tied to dollops of money conditioned on the promise to make reforms. The report, and much else, tells us that reforms are not made in a majority of cases and that most conditional programs do not survive into the second year. The problem is to decide what should replace these procedures. My answer is that the IMF must shift to an incentive system, where payments are made for performance, not for promises. There was much talk of country ownership, countries choosing the reforms. This is a good step, an appropriate and desirable step if the country actually makes the reforms and keeps them.

As the report recognizes at several places, structural reform takes longer than most crisis programs, but structural reform is what real adjustment is about. Doesn't anyone at the IMF wonder why countries that got IMF support, like Turkey, Pakistan, Argentina, Ecuador and, in the past, Mexico and India, have had frequent problems? Could it be that they never made the reforms? Isn't it instructive that Turkey took implementation far more seriously when it had the incentive of joining the European Union? It would not get invited unless it made the reforms. India got its incentive from comparing the Hindu rate of growth to China's vastly more successful program. Reforms that were previously politically impossible became distinctly possible.

The main lesson in the IEO report, for me, is that the IMF is not very successful. To be more successful in reforming fiscal and other policies in the client countries, the IMF must reform itself.

I have written and spoken enough about how to increase incentives for reform that I will not repeat my proposals. Let me just say that the CCL [Contingent Credit Lines] was a start toward reform, but the IMF let it die. The CCL contained an incentive. It rewarded actual, not promised, reform. The incentive could be strengthened. Its counterpart, no assistance to countries that get into trouble because they do not make reforms, could supplement it. One thinks of the Turkish or Korean financial systems or the rigid Argentine labor and product markets. Even more, one thinks of financial and exchange rate systems that fail dramatically and produce the largest and most severe crises.

Recent research in economic development asks why development occurs in some countries, but not others? The answer in the current literature is that institutions matter. Unless countries adopt supporting institutions, sustained development does not occur. I believe that crisis reform and crisis avoidance also depend on institutional reforms to require local leadership, sustained commitment and the ability to maintain and support reform for reform during the often costly adjustment period. The IMF cannot bring a country to do what the citizens or their representatives do not want to do.

For me, the most striking aspect of the IEO evaluation of fiscal adjustment is the lack of curiosity by the Executive Directors about how the IMF's record could be improved. After 60 years, it should be time to answer that question. Like Alice, the more I think about it, the more it grows "curiouser" and "curiouser" that they don't ask. Shifting more decision-making authority to client countries facilitates their ownership of the programs and reduces the IMF's perceived responsibility, so the IMF has an incentive, too.

The IEO report expresses concern about maintaining the social safety net during periods of adjustment and distress, but says little about the effect of shifting the burden of adjustment up the income distribution. Welfare decisions of this kind should be made locally, not in Washington. The IMF's incentive for reforming itself should come from the fact that it would be a way of telling its many vociferous critics that it provides incentives for reform and adjustment and leaves these welfare and distributional decisions to the countries, but that hasn't happened. It's "curiouser" and "curiouser."

Thank you.

LOUNGANI: Thanks, Allan. Our second commentator is Nancy Birdsall. She's the founding president of the Center for Global Development here in Washington. Prior to launching the Center, she was at the Carnegie Endowment for International Peace. As I'm sure all of you know, before that she had a very long and distinguished career at the IDB and the World Bank. Nancy?

BIRDSALL: Thank you very much, Prakash. It's a pleasure to be here. It's always a pleasure to come to the IMF, a sister institution of my former institutions, the World Bank and the Inter-American Development Bank.

Let me start by congratulating the IEO for this report, especially Marcelo. I think that office is beginning to show how important it can be to have this kind of independent evaluation.

The report, as Allan suggested, is outstanding. It's on an important issue. It uses a careful approach. It's fine-tuning how to reduce the human costs of adjustment. It's clear and transparent in its analysis. It's responding to longstanding, sometimes misguided, and sometimes exaggerated criticism of the way the IMF deals with fiscal reform issues. The recommendations are practical, and they can be made operational.

What I'd like to do is relate to you one big complaint I have about the report and its context. Then, I want to emphasize two points that Marcelo raised, and I want to conclude by putting the report in a larger context.

Let me start with the complaint. It's not unrelated to some of the remarks that Allan made. My complaint is that the report, for all its technical excellence and subtlety, is like so much of the IMF's work -- willfully naive about the difficult politics of distribution and welfare in so many developing countries. This is the case in many countries, but particularly in developing countries, where institutions are weak, where the checks and balances that democracies provide are missing, where sound economic institutions are not yet well-developed. The politics of distribution are really tough.

By willfully naive, I mean that IMF staff know about these issues. Members of the IMF Board know about the difficult politics of distribution within countries. They have seen how technically sound recommendations can so easily be corrupted in their implementation. They have seen that these consequences hurt the politically weak and thus the poor.

So I think, in a general sense, my complaint is that we cannot forgive the IMF just because it is technically sound. We cannot absolve it of the unintended consequences of its recommendations when the consequences of those recommendations are foreseeable in a political economy sense.

We know that in most developing countries there are entrenched powerful groups, be they opposition politicians, leadership of public unions which are not very democratically organized, industries that enjoy protection, industrialists that enjoy privileged access to cheap credits and insider contracts, the rich who evade taxes, so on and so on. There's also the kind of political inertia that makes it very tough to reduce salaries, even though the costs of the programs that Marcelo and the IMF would like to see protected are so low that reducing salaries or just holding them constant might finance double and triple what the PROGRESA program in Mexico represents, what Paul Seflomilia [ph] in Brazil is trying to do.

So, in fiscal work, I think the IMF makes two kinds of errors. The first is the technical errors, which this report demonstrates are much less pervasive, much less costly than Joe Stiglitz and others might have us believe. I think I have to say it is a coincidence that both Allan, more or less crudely from the right, and Nancy, more or less crudely from the left, referred to Stiglitz in this context.

But there is a second kind of error which is that willful, political naivete. It's a complaint, perhaps unfair, about this report, that it's far too subtle on this issue, and it's unfair because we should not put on one report the burden that I think the IEO faces of finding a way to translate these political economy issues about distribution and welfare into something that can be absorbed, in technical terms, inside the IMF and made operational.

There is other work by the IEO on the prolonged use of IMF resources that refers to many of the points Allan made and has in it the kind of suggestions for this sort of political economy analysis, which are perfectly straightforward to do.

Let me go to two points about the report that I raised in a letter to the Managing Director, representing the Center for Global Development, a Washington-based think tank. I sent this letter in January of 2004. I still haven't gotten a response, but maybe someone here will help me inspire from the IMF a response. There has been, in the interim period, of course, confusion--not confusion--but the change in leadership which may have made it more complicated for the Managing Director's Office to reply.

The first point had to do with the recommendation that Marcelo mentioned and emphasized; namely, to quote, "that the IMF could invite the authorities regularly during Article IV consultations to suggest what are the existing critical social programs and social services they would like to see protected in the event of adverse shocks."

Let's note that this recommendation refers to what I would call the most politically vulnerable programs, which are vulnerable because they serve the most politically vulnerable people. Let me refer you later to Page 57 of the report, which lays out the details on conditionalities and benchmarks in an Ecuador program and illustrates chillingly the fact that some things are recorded, and benchmarked, and tracked, and other things aren't.

And the things that so far are mostly recorded, and benchmarked and tracked, although they are politically difficult to do and therefore they are included as conditionalities, are least likely to help and protect the poor. And those things that, on the contrary, are not benchmarked are the most likely at least to protect the poor.

This is a proposal that is admirably subtle and careful, maybe too subtle. But that very care and subtlety, my fear is that it leaves that recommendation hiding in the shadows. Without a strong message from the board and from the leadership management side within the IMF, I have very little faith that IMF staff, in the Country and Regional Departments, will be systematic and thoughtful in implementing this recommendation.

Indeed, the comments of staff on this proposal, which were published in the IEO report, are oddly lukewarm. The comments of staff raise two issues. One is a reference again to the supposed fiscal tradeoff, which Marcelo and this report demonstrate once again is really not there. It's tough to do politically, but the fact is that some of these programs are incredibly cheap to sustain.

The second thing that the staff referred to is the fact that this is World Bank business. And this reminder to me that, from the point of view of the Fund staff, that social programs are World Bank, not IMF business, also elides completely the point of the recommendation, which was very carefully worded, and it is not to judge, not to design, not to evaluate, not to comment on what the authorities might have to say. It is merely to invite the authorities to think about the issue. It is to invite the authorities in a manner that might give them slightly more political oomph later, when the fiscal hit comes, to protect those programs.

In this sense, it is a subtle invitation to strengthen the arms of those who would fight for what is politically difficult to protect when times are tough. It seems to me that inviting authorities to suggest whether and what programs they would want to see protected is consistent with the IMF's longstanding role in encouraging sound, medium-term fiscal programming, including the logic of fiscal restraint today to permit countercyclical spending tomorrow.

The IMF would, in implementing this recommendation, not only be helping countries protect fiscal adjustment from untoward political pressures, it would be rescuing itself from this widespread perception, which both Allan and I have referred to, that's ultimately harming the Fund's effectiveness, that it supports fiscal discipline, mindlessly and callously, and pushes spending reductions onto the poor. This report shows it ain't so. Still, there's more that could be done.

The second proposal I want to refer to has to do with tax reform. And, again, I'm delighted that Marcelo emphasized this point. Chapter 7 of the report talks about the fact that the IMF has focused so heavily for 20 years or more on expanding of that and reducing trade taxes, with less attention to income and property taxes.

There is no argument economists make about all of this, but I think it is politically naive to assume that all the burdens of adjustment can be soaked up on the expenditure side and that in democratic countries trying to manage a social contract, there should be no burden of adjustment and sharing the burden of adjustment on the tax side.

And, again, I would say, in the published report, the staff response is not encouraging. It refers to the excellent work of the IMF on tax administration. Again, it elides the point, you'll see it on Page 57 in the Ecuador program, that for all of the work on tax administration and all of the tax advice that the IMF provides, worthy as it is, it has not translated into policy work, into policy conditions, into reasonable demands on government.

It is stunning to think that in the late 1990s, IMF research staff published a report showing that in Argentina, where in retrospect the lack of fiscal discipline has tremendously hurt middle-income people and the poor, the effective tax rate on the 10-percent richest households in Argentina was -- sit back and think for a minute what your effective tax rate is, not your marginal rate, but your total tax rate? -- eight percent. It's just too low.

So these are two big issues.

Let me end with a concluding comment, using the report to make a more general point, and that is that we all have to look to the IMF in the 21st century to send more effectively the message that social policy, despite the great ideas and the recommendations in this report, is not just about social programs and certainly not just about spending on social programs.

We have got to go from the concept that social policy is spending more on education and health to a much larger vision of what is the social contract in an open economy? And the foundation of a good social contract in an open economy has to be fiscal discipline which will allow for lower interest rates creating jobs--that's good for people--which will minimize the buildup of debt, allowing later for countercyclical spending to protect the poor, which will create the kind of space on both the expenditure and tax side, political space that will ensure, in the long run, that the IMF is playing its worthy role in reducing poverty and inequality in the world.

Thank you very much.

LOUNGANI: Thanks, Nancy.Let me turn it back to Marcelo or Allan if they have any comments.

MELTZER: I'd like to take up an issue that Nancy raised because I think it's a real difference of opinion between us, and it really goes to the heart of what I think is the issue. Nancy and I agree that the political economy aspect of the IMF and the reform programs are very important, but then she goes on to do something which I think is very much different from what I think needs to be done. What she wants to do is impose or have the country authorities impose or have the IMF impose her idea of social justice on the country. But I think that the reform, that if the IMF is going to get out from under the protests, the criticisms that--I agree with Nancy--that very much interfere with the effectiveness of its program, then what it has to do is it has to take the idea of ownership seriously and let the country decide what its social policy ought to be.

It's not our job to decide what people's social policy ought to be. We can recommend what we think social policy might be, but it's up to them to decide. So, if they want low taxes on the rich, well, that fits very nicely with optimal theories of taxation. For example, lots of theories of optimal issues of taxation say you should take the most productive people and charge very low tax rates because they'll work harder. So you'd like to work them to death; well, not to death, but almost to death.

So that be the country's choice. It may not be my choice, it may not be your choice, but that really is the central issue. The IMF should not be a command-and-control operation. It should provide incentives for countries to do things which are in the country's perceived interests.

LOUNGANI: Nancy, did you want to respond?

BIRDSALL: Well, it's complicated to try and explain why I think Allan and I are more in agreement -- but let me go back to what is so careful and subtle about the recommendation in this report that I referred to. It is to invite the authorities to outline what they would see as programs they would protect. It is not to tell them what to do. It is not to make judgments about what to do.

Where I think Allan has an important point-- I didn't refer to it explicitly in my remarks because he had said it very well -- is that it would make a difference to understand the politics behind these tough decisions that governments have to own. It would make a difference. It would make it easier for the IMF to provide the support at the right time in the right way that responds to the right sorts of ownership. It might mean sometimes not doing the next loan because some particular arrangement that was committed to by a government, owned by a government, wasn't made. I think of the example of Pakistan because I was there recently. There has been at least over 15 years, between the World Bank and the IMF, more than 15 adjustment operations. And in every single one there is a reference to the need to tax the land and to do land reform. That is a politically difficult thing to do. Until the Pakistanis find the wherewithal to do it, I think there should be dialogue, there should be engagement, there should be support in some small way for some reform efforts, but I think there should be a lot more emphasis on not moving lots of money because conditions are waived.

So I think the link between this excellent report and the very good report on prolonged lending of the IEO is something that the board of the IMF should take up explicitly and that the leadership and management at the IMF needs to think about very, very carefully. My own view is it goes back to this willful naiveté about politics, but can be managed. The IMF can do better.

And I think in a world in which weak and failed states are an increasing problem not only for their own people, but for the global system, both economically and politically, and in terms of illegal drug trafficking, war on terror, all of that, it is absolutely critical for the international community to think through how to handle these matters.

We can't just throw up our hands, Allan, and say, "Well, they're not behaving properly, they're not responding to incentives, so we'll walk away." So that's why it's interesting and complicated, and thank goodness we have the IMF to address these issues.

MELTZER: Let me just say that it's rare for me to find so much agreement with Nancy. I'm going to stop there.

[Laughter.]

LOUNGANI: We are running a little bit over time. I realize that people who have binding commitments will want to leave, but I hope others will stick around for questions for our panelists. Thank you all for coming.

[Applause.]