Transcript of a Conference Call on the Staff Discussion Note: "Redistribution, Inequality, and Growth"

February 27, 2014

Washington, D.C.
Tuesday, February 26, 2014

Jonathan Ostry, Deputy Director of the Research Department
Andrew Berg, Assistant Director, Research Department
Ismaila Dieng, Senior Communications Officer

MR. DIENG: Thank you. Good morning everyone and welcome to this conference call on the release of the Staff Discussion Note: Redistribution, Inequality, and Growth by Ostry, Berg, and Tsangarides. You have seen the paper; it's under embargo until 10:30 am.

The main speakers today will be Mr. Ostry and Mr. Berg. They'll start with short opening remarks and then they'll be happy to take your questions. Jonathan?

MR. OSTRY: Good morning. Thank you for joining. So my sense is that pretty much anyone who has dialed into this call needs no reminder as to the extent to which the issue of income inequality has risen on the global policy agenda in recent years. This reflects not only fears about the social and political affects, including questions about the consistency of extreme inequality with democratic governance, but also the economic implications.

While we may all agree that some degree of inequality is needed to provide the incentives for economic growth, excessive inequality seems more likely to detract from sustainable growth. For example, by undermining access to health and education, causing investment reducing political and economic instability, or even thwarting the social consensus required to adjust in the face of major shocks.

High income inequality may also directly contribute to crisis risk. For example, by intensifying leverage and financial cycles, especially if the rich have an interest in allowing financial excesses to balloon.

In some of our earlier work we documented a robust, medium term relationship between inequality and sustainability of growth. We did not, however, have much to say about whether this relationship justifies efforts to redistribute. Indeed, many would argue that redistribution, far from being a friend of growth, is the reason that unequal societies often stumble.

On this view, this vicious cycle runs from inequality to redistribution to low and fragile growth. If this view is right, taxes and transfers are precisely the wrong antidote for inequality, a cure that is worse than the disease itself.

But is this view right? The economics profession hasn't come to a firm conclusion. Some papers point out that some policies that are redistributive, for example, public investments in infrastructure, spending on health and education, and social insurance provision, may be both pro-growth and pro-equality.

Others contend, like Arthur Okun did some 40 years ago, that there is a fundamental trade-off between redistribution and growth, reflecting the disincentive affects of redistributive policies.

In the paper that we're releasing today, we ask what the historical data have to say on these issues. In particular, what is the evidence about the macroeconomic affects of redistributive policies, both directly on growth and indirectly as they reduce inequality, which in turn affects growth? Making use of a new data set on market inequality, that is before taxes and transfers, and net inequality, that is after taxes and transfers, we arrived at two striking results.

First, we continue to find that inequality is a robust and powerful determinant, both of the pace of medium-term growth, and of the duration of growth spells. Even controlling for the size of redistributive transfers. That is it would still be a mistake to focus on growth and let inequality take care of itself, if only because the resulting growth may be low and unsustainable. Inequality and unsustainable growth may be two sides of the same coin.

Second, there is remarkably little evidence of adverse effects of fiscal redistribution on growth. The average redistribution, and the associated reduction in inequality, seems to be robustly associated with higher and more durable growth.

We do find some mixed evidence that very large redistributions may have direct negative effects on growth duration. But even then, taking into account the protective effect that redistribution exerts through a reduction in inequality, the total impact on growth is roughly neutral. Thus, even countries that undertake very large redistributions, presumably with the goal of promoting equality, do not see to pay a clear growth cost.

We are not saying, of course, that countries wishing to enhance the redistributive role of fiscal policy should not pay attention to efficiency of the fiscal measures. If governance and administrative capacity are weak, for example, a massive scaling up of redistributive policies would plausibly be expected to carry a cost.

That being said, the conclusion that emerges from our study is that the things that governments have typically done to redistribute have not led to bad growth outcomes. On average, redistribution seems to have helped support faster and more durable growth.

To put it simply, we find little evidence of the big tradeoff between redistribution and growth that Arthur Okun emphasized nearly half a century ago, and that her permeated the consciousness of economists and policymakers ever since. Inaction in the face of high inequality thus seems unlikely to be warranted in many cases.

Happy to take your questions.

MR. DIENG: Thank you, Jonathan. We'll now open it to questions.

QUESTIONER: Hi, all. Really interesting paper. This may sound cheeky, but it's not, I'm wondering if you could define what a big redistribution is, and does it matter here what type of redistribution we're talking about? In other words, when you talk about the big tradeoff are you pretty much limiting yourself to sort of, you know, sort of classic vanilla, you know, transfer programs and/or investments?

I'm wondering, you know, if we go to the other extreme and you talk about, you know, nationalizing all property and dividing it up into parcels, equal size to all people, I assume that's a big redistribution, right?

MR. OSTRY: On your first question, big means the top quartile of the redistribution distribution. On your question about design, this is obviously a key issue, and one that we don't touch upon in the paper. My sense is that you're right; design is an absolutely critical issue. My colleagues will be coming out in a few weeks' time with a paper that is complementary to this paper in the sense of being about the design issue.

Obviously there are good ways and bad ways of doing redistribution, and we would not want anyone to take away from this paper that design is irrelevant and that it is impossible to hurt your economy if you design redistribution in bad ways.

What we are saying is that on average the things that governments have done have not been destructive on growth as far as the data we have at our disposal are concerned.

QUESTIONER: I assume that data doesn't include things like, you know, land reform in some of the South American countries that have screwed things up or the experience that China or North Korea's going through right now. That type of experience is not factored in here, correct?

MR. OSTRY: Correct. It basically includes the gamut of taxes and transfers that are part of fiscal systems, but doesn't include big ticket items on health and education and social insurance provision. Those to my mind, might actually be be more pro-growth, so I don't see a necessary bias in our results on this score.

MR. BERG: I just want to expand a bit on that. It's an important question. We use a data set that takes some essentially standard definitions of market and net income. But the government does a lot of things that may affect market incomes. For example, it could fix prices for goods, put caps on bonuses, it could do things like that. Those would not show up as redistribution in the data we used because they affect market income.

QUESTIONER: Hi. Thanks for doing this paper. I was wondering if you could just speak a bit about, you know how and when, I guess, this could translate into IMF advice to different countries. I know, you know, the conclusions are still very preliminary. But it does seem that you're putting a much greater focus, you know, in recent papers on inequality than the IMF has traditionally done in its advice. So I'm wondering, you know, does that mean an Article IV that this'll be a greater focus or is it kind of still too early because the research is still ongoing? Thanks.

MR. OSTRY: I want to draw a line between this analytical paper and Fund policy advice and design of programs. You know, the disclaimer on the front of the paper is there for a reason. I mean, that's basically what I can say on the issue. There's no direct implication for Fund policy advice or program design.

QUESTIONER: Thank you for taking my question. I wonder if it's possible what country that has good redistribution policies and one that may use excessive redistribution policies? I don't know if that's a set of 153 countries, if I understood correctly. If you can mention one or two of them in both extremes? Thank you very much.

MR. OSTRY: So this is a cross-country study. We haven't looked at specific country cases. We wouldn't have the data to really examine in detail specific advice for specific countries.

I think what we are saying on the whole is, as I said, the average redistribution had very benign direct affects on growth, and through the impact on equality has had a highly robust and protective affect on both the level of medium run growth and the duration of growth spells. Which, as we know, is key for closing the income gaps between low and middle income countries and rich countries.

QUESTIONER: And I'm right to say that it was 153 countries that you analyzed?

MR. BERG: I think that's the number that we sort of started with, but every time we include another variable, for example if we want to look at say a measure of institutional quality, data may not available for all the countries for all time periods for that data. But that's where we started.

MR. OSTRY: Of course, when we get to discuss hazards and the duration of growth spells we're dealing with a much smaller sample of countries than 153.

QUESTIONER: Hello, good afternoon -- good morning, sorry, you're in Washington.

Your study, did you say that earlier that you don't include transfers that involve Social Security or is it all focused on taxes and other more -- other above-the-line transfers?

MR. OSTRY: Yes. So what's included is the range of taxes and transfers that you're thinking of, but not in-kind transfers for things like Social Security, health and education. Basically our data set follows the Luxembourg Income Survey.

QUESTIONER: So when, I mean, obviously, you're probably aware, and probably to your dismay, that this will be reduced to rather quite crude basis that, you know, that transfers are benign and there are lots of groups who will think that those below-the-line transfers, Social Security in particular, do rob workers of the desire to work, and therefore do work against growth. You're saying that that's not in the report, but it is your belief that that wouldn't be the case. If anything, may actually spur growth, maybe positive?

MR. OSTRY: Let me be clear on this. Our study uses data and it's those data that we can speak about. What I was saying was that I don't see a strong case that what we exclude is, in a way, biasing our results in a particular direction.

You know, going back to Arthur Okun, I think the kinds of taxes and transfers that we include in our study are very much those that he worried about as being antithetical to growth and economic efficiency. You could of course be right though. What we exclude may be the missing evidence to support Okun’s hypothesis. What I said was speculative. But I don't see a strong case ex ante for bias.

MR. BERG: I just want to clarify. I mean, this question of data is obviously very difficult. That's one reason this kind of paper has been hard to write, and so we were happy to see this new data set that does the work of trying to put together comparable data, but it's still hard.

Just one point on what you said is that the taxes that would support, for example, education spending are included as taxes in our data. So insofar as, you know, you have to tax workers and others to provide more education services, the services would not be included as a transfer because it's so hard to allocate them to a particular individual.

QUESTIONER: But the taxes are?

MR. OSTRY: Let me add further to that. So you've got progressive taxation that's financing education. Most people think the progressive taxation may be antithetical to growth. Most people don't think the provision of education of antithetical to growth. So we include the thing that's antithetical to growth in Andy's example, and we exclude that thing that it unlikely to be antithetical to growth. That's where I was coming from.

QUESTIONER: Hi, thanks for the call. I know you say you're drawing a line, this is a research, but it seems to me that Fund research often feeds directly into Fund advice at a later point in time. And you're saying that categorically that is not going to happen?

MR. OSTRY: I didn't say that. I think we're a learning institution, so if we're sitting around writing papers and no one's listening that would be a waste of time. So, of course, we hope it permeates. But the question was: does this have direct implications for Fund policy advice and program design? I said, no. I didn't say that we didn't hope it would influence those things down the road.

QUESTIONER: So you do expect it to influence IMF policy in the future.

MR. OSTRY: I hope it will.

QUESTIONER: You’re just saying that right now it's not?

MR. OSTRY: That's correct.

QUESTIONER: Okay. Secondly, let's see, can you help me to understand when you say unequal societies tend to redistribute more? That seems to be the opposite of what you're trying to say here. That redistribution helps to deal with the inequality in societies and helps to spur growth.

MR. OSTRY: Yes, that's a great question. So the market system produces some degree of inequality. The hypothesis is that as the level of market inequality rises there is greater pressure through the political system for redistribution. Essentially political power is more evenly distributed in democracies than economic power. That may not be the case, of course, if the rich have more political influence then the poor.

Then what we observe as the net, post and tax and transfers inequality, is the outcome of what the market produced and what governments do to partially reverse that.

This is very important because many people interpreted our earlier study on the relationship between inequality and growth as saying there's a relationship between inequality and growth, but the reason you find that is because of redistribution. Unequal societies tend to redistribute more, and that redistribution affects economic incentives so much that growth tumbles.

When we simultaneously control for both redistribution and inequality we can say, well, what is the effect of inequality on growth controlling for the level of redistribution? We find that inequality is bad for growth, even holding constant the level of redistribution.

So we can say that what drives the relation between inequality and growth is not redistribution. It's not because more unequal societies redistribute more that we find that inequality is bad for growth. It's bad for growth in and of itself.

We can also say that redistribution doesn't by itself, controlling for the level of inequality, seem to be that bad for growth unless it is very large.

QUESTIONER: So it seems to me that you're saying that extremes at either extent are bad for growth?

MR. OSTRY: No, I didn't say that.

QUESTIONER: Can I just rephrase that? You're saying that if you redistribution too much from, particularly from people that -- who can't afford the redistribution or if there are overly large, extreme cases, as you phrased, that could weigh on growth?

MR. OSTRY: So there are two effects. When redistribution is large, in the top quartile of the distribution, we find a direct negative effect on growth. But don't forget that redistribution engenders greater equality, and that greater equality itself has a protective affect on growth. So you have to add those two things together to come to an assessment of how redistribution affects growth.

When the redistribution is not extreme, it has a highly protective effect on growth. When it is in the top quartile, it has a roughly neutral effect on growth. So even countries that are in that top quartile don't seem to pay a clear growth cost. The average country achieves greater equality and appears to reap a growth dividend. So that's how we see it.

QUESTIONER: Okay. Finally, you mention in the Executive Summary excessive risk taking in the financial sector. This has been a theme that has cropped up in other IMF research, and actually I think the -- what was it the FSAP -- no, not the FSAP, one of the other big financial papers you do. Talking about or indicating that the IMF is moving towards a financial transaction tax as a beneficial thing. Can you comment on that?

Finally, is there criticism to be found in here of advanced economies, for example the U.S. and Europe, or is this more for developing economies?

MR. OSTRY: So our panel of data encompasses advanced, emerging, and low-income countries. The reference that I think you're talking about is really referring to a line of thinking, in the literature especially, by Raghu Rajan in his book, Fault Lines, where he draws connections between excessive risk taking in the financial sector, excessive leverage, rising inequality, and the emergence of crisis, and the fragility of the pre-GFC growth model of the United States and other countries. Our paper has no direct relation to the issue of how you should address concerns about excessive financial risk taking. That's really a separate topic and far beyond the scope of this paper.

QUESTIONER: Okay. And the second question about -- I mean, it seems to be a large difference between the income inequality in say India and the income inequality in say the UK or the U.S. So I'm wondering if you're seeing the results of your paper as being applicable in advanced economies as well as the emerging market economies or just one or the other?

MR. OSTRY: I personally see the results of our paper applying to all countries in the sample. We don't have the capacity to run hazard regressions on sub-samples. We were very constrained by data. So absent having done the exercise that would allow us to get at your issue I don't have a prior that says that it applies to one group and not the other.

MR. BERG: Our results are average results for the world. They present tendencies that are statistically fairly strong. Whether they apply to any particular case it really is hard for us to say. You have to look carefully at that particular case. Because obviously a lot's going on in any particular country besides things that we could possibly capture in this sort of framework.

QUESTIONER: Sure. I understand, but obviously it's a rallying cry for the folks say -- I forgot what they're called, the 99 percenters in Wall Street. You know who I'm talking about? Versus, you know, major crises in equality in say, the Ukraine.

MR. OSTRY: Let me, without getting into the Ukraine, India and the United States cases, say that there is one thing that I think makes us, in terms of the policy message, cautious.

It's coming back to something I said at the beginning, which is when governance is weak and when fiscal institutions are weak, we would tend to be more cautious about recommending redistribution because in those circumstances redistribution is less likely to be done efficiently than in countries with very well-developed fiscal institutions and strong governance.

MR. BERG: One more thought on your question, which is an important one. At least for me when I think about the question of what result applies where, I go back to the broader literature that talks about the relationship between inequality or redistribution and growth. Because there is a bigger literature that looks at each one, in fact huge literatures in each case. In our earlier paper where we looked at just at inequality and growth, we did go and look at the narrative for some of the particular countries where our data set suggested that there something going on. If you go back and look at that earlier SDN, we found that the narrative was evocative. The results from the statistics seemed to correspond to what we saw in the narrative. One of the main stories, for example, in the literature that goes back to Dani Rodrik, and I'm sure before, is that unequal countries have a hard time achieving social consensus to adjust to shocks and difficult times. This argument was inspired, really, by events a long time ago, from the shocks in the 80s and the difference between countries that collapsed under the pressures of high global interest rates and so on at that time, and other countries that adjusted smoothly through the period.

The punishment seemed to be so much worse than the crime for those countries that had a hard time making the adjustments. The argument was made, and I think you could see it when you look at countries from that time, was that some countries just could not adjust, could not raise taxes or could not do whatever needed to be done to adjust to the shocks.

You have to ask yourself, I think, whether in a particular case we see echoes of that kind of dynamic. Then can we think about relating that to, for example, inequality.

QUESTIONER: So I was struck by -- I know these are averages and all that, but I was struck by the graph on Page 23 which I guess applies to the duration of growth spells argument. Here you see the clear division between the top quartile and the bottom 75 percent. Okay, with Bangladesh as the outlier you're got all these European countries sort of clustered in the top range there where they presumably are not paying a growth penalty, but are not getting the growth dividend of more equality.

I'm wondering why isn't this translating into a more direct sort of, you know, discussion of what they could be doing better? I wonder if there isn't some inconsistency between, for example, the type of labor market reforms that the Fund is recommending in, say France, that could increase inequality by letting employers lay people off more easily?

MR. OSTRY: I mean, these are great questions. So this is snapshot of the latest data that were available to us. That's what you see in Figure 7. Details of labor market reform design, and related policies I think you cannot infer from our analysis.

My sense, though, is that this work will make people think. It will hopefully make people think twice about “freezing” in the face of extreme inequality and being very cautious.

But, you know, the details of country policy design, you're asking more of the paper than it can answer.

QUESTIONER: Okay. Fair enough. Just one last thing real quick. To be clear on the sort of hypothesis you were testing out here, I mean, it sounds like the hypothesis isn't necessarily that redistribution enhances growth as opposed to saying that it simply does not undermine growth?

MR. OSTRY: When we started this, we had absolutely no prior as to what we were going to find. If we expected anything, it would have been that Okun was probably right, that redistribution is bad for growth.

We then read a lot of papers that argued that some redistributive policies, like providing better access to education or health or social insurance provision, may actually be conducive to growth. And so we became more agnostic about what we might find.

And then we began the empirical phase of the project and got results suggesting that the effect of redistribution seems to be close to zero. Our interpretation is that there are probably both pro and anti-growth elements in the typical redistribution package, and these may cancel each other out on net. It also made sense to us that the more redistribution you engage in the more likely you are to find a cost, given the convexity of costs.

MR. BERG: Let me just clarify something. It's a basic point and I'm sure you get it, but I think it's worth underscoring. We have tried to be as careful as we can about the distinction between the direct effect and the total effect.

That's absolutely critical. The whole point of this paper--or one of the points of this paper--is to distinguish those two things. So when we say, for example, “on average redistribution has no affect”, we mean that it has no direct effect.

MR. OSTRY: Exactly. But it still has a favorable total effect taking account of the impact through greater equality as well.

For the large redistributions we do find some evidence of a negative direct effect, but we find that it seems to be about compensated for by the positive indirect affect acting through inequality.

QUESTIONER: Did you get into any -- I mean, I didn't see this in the paper, but do you have any theories about, sort of interaction between the demand and supply side here? I mean, are the ones that sort of get up in the top quartile range having supply side effects that are accounting for the cancellation there as opposed to the sort of more demand side benefits of great equality?

MR. OSTRY: As a matter of logic, the disincentive affects would tend to be on the supply side. So given this convexity of costs you would tend to think that they're more severe at the top of the observed distribution.

MR. BERG: Also, we only found these indications of a negative direct effect when we looked at the growth spells. In general these have a longer time dimension. A growth spell is anything from five years to forty years. What we're looking at is how likely it is to keep going at any point in time. So we see a hint of this negative direct affect for large distribution for spells.

Now we looked for it and didn't find it when we looked at five year average growth rates, i.e. medium-run growth. We didn't see any sign of a negative direct affect at all in the five year growth context.

That to me argues for the supply side, but we don't have any direct evidence.

MR. DIENG: Thank you very much, and thank you all for taking part of this call. As I said, the paper is in embargo until 10:30 today. We'll post the transcript later on our website. Thank you very much. Thanks to our two speakers.

MR. OSTRY: Thank you, all.

MR. BERG: Thank you.


Public Affairs    Media Relations
E-mail: E-mail:
Fax: 202-623-6220 Phone: 202-623-7100