Reflections on Secular Stagnation

February 25th, 2015

Summers gave the keynote address at Princeton University’s Julius-Rabinowitz Center for Public Policy & Finance on February 19, 2015. In his remarks, Summers gave his perspective on the “profound macroeconomic challenge of the next 20 years in the industrial world: secular stagnation.”

Lawrence H. Summers
Speech at Julius-Rabinowitz Center, Princeton University
February 19, 2015

Thank you for those generous words. I am glad to be here, glad to see so many old friends, my former Treasury colleague Jeff Shafer, from whom I learned much of what little I know about the international economic interactions relating to Europe. My former student, and then government colleague, Alan Krueger, whose work has illuminated so much to do with the working, or the non-working, of labor markets. My friend David Wessel, who’s covered my activities in government for many, many years at the Wall Street Journal. I can now tell you that on any occasion when I looked good, it was because he was reporting accurately. On any occasion when I looked bad, it was because he did not have an accurate rendering.

I just want to say, before I launch into my topic, that as someone who has spent his life, in a way, shuttling back and forth between government and university, I think that conferences like this one, and centers like the one that it’s convened in, are really profoundly important. If, for example, the United States has had a more successful response for financial crisis than Europe or Japan, it is importantly because of the kind of close connections between the worlds of thought and the worlds of action, that the American system makes possible. I believe the cultivation and support of worldly academic research in economics is something that is very, very important. I also believe that economic ideas, when either right or wrong, spur change and spur progress. When right, they make an important contribution. When wrong, they provide important clarification that ultimately proves to contribute to public policy.

What I’d like to do today is talk about my perspective, and I’ll try to recognize that there are multiple perspectives, on what seems to me to be the profound macroeconomic challenge of the next 20 years in the industrial world, and that is a problem of what I like to call secular stagnation, following Alvin Hansen.

I’m going to talk about six things. I’m going to talk about why we’re talking about secular stagnation, the dismal performance of the industrial world in recent years. I’m going to talk about the secular stagnation hypothesis, as Hansen framed it. Talk about what’s the central element in that, the low level of real interest rates. Reflect on some of the challenges that have been posed to the hypothesis, and then discuss what is it to be done.

This shows you US economic performance since 2007, measured relative to what we aspired to in 2007. What you see is that the economy went off a small cliff between 2007 and 2009 and that relative to what we aspired to in 2007, there has been no catch-up. The GDP gap is indeed smaller than it was in 2009, but that is entirely because our judgments about potential has been revised downwards, in the face of dismal performance. If anything, the picture is worse. In Europe, where there’s been essentially no progress, and where the gap relative to potential as we had assumed it would be, has steadily increased, and is continuing to increase. Of course, this is all reminiscent of the Japanese experience, and it would be be a rough summary of macroeconomics in this decade to say that Japan is the old Japan, and Europe is the new Japan. Europe today looks very much like Japan did seven or eight years post-bubble. Demographically challenged, incipiently deflating with severe financial strains, with dysfunctional politics, and ineffective decision-making.

I say that this picture is profoundly, in my view, inconsistent with the dominant presentation of the crisis, and way of thinking about the crisis comes through in most of what has so far been written and said about the US financial crisis. Roughly speaking, and I exaggerate a little bit for effect, the dominant descriptions of the crisis fall within the framework of what I would call financial network failure theory. Something went wrong, bubbles, bank runs, insolvency, whatever it was, there was this vast mass panic. It looked like the financial system was going to collapse. Heroes rode to the rescue, liquidity was provided, stimulus was provided, the system held, we didn’t have the depression and life went on.
If you read Davids book, that is the full and rich and nuanced summary of his book, Davids book (In Fed We Trust)- the best of the books that has been written – about the crisis. He’s pretty clear that it could have gone another way, and the system could have collapsed. That captures a very important aspect of what took place, but it’s very incomplete at capturing of what took place.

Think about the economics of power failures. What we would expect we’d have, if the economy had a terrible power failure, and was without electricity, and the electricity network didn’t work. We’d expect there to be a near-collapse while that was happening. You could try and think about just how that collapse would play out, but it’s pretty clear that if you got a big power failure, there’d be a big collapse. What would you expect when the power was turned back on? You’d expect that the economy would start to function well again, and indeed, you’d expect that there be even more production, because all the inventories that ran down during the power failure had to be run up. All the people who couldn’t go to the stores during the power failure would go to the store. You would expect, actually, that if it was just some temporary network failure, that if anything, recovery would be very rapid, and if anything, output afterwards would be even more because of that. Yet, if you look, you look at credit spreads, you look at the repayment of TARP, if you look at anything, it was five years, more than five years ago, that the lights had been turned back on, and we have not caught up at all, relative to what we expected. That suggests that we need to think about something other than the power failure model, and I believe, something other than a traditional business cycle here.

Now, to further motivate why we need to think of something that’s further outside the conventional sandbox, I invite you to think about the period prior to the crisis. Think about the United States to start with between 2003 and 2007. It was characterized by what was regarded and criticized at the time as being overly expansionary monetary policy. A prescription drug benefit was entered into, the Iraq War was fought, the Bush tax cuts were entered into. Widespread concern about excessively expansionary fiscal policy. We had the mother of all housing bubbles. With housing levels 50% or 75% above fundamental values, and a vast erosion of credit standards that enabled people to take wealth out of that.

Did any of that produce spectacular economic performance? No. Did any of that produce overheating, with respect to inflation, on the large scale? No. With the worst erosion of credit standards, and the biggest bubble since the Second World War, we got adequate, maybe even good economic performance, You look at the pre-2007 part, and you ask yourself about the growth of debt to disposable income, and you say, “Did we have financial sustainability when we were having adequate economic performance?” It doesn’t look like we did. One could say the same thing with respect to Europe, and of course could add that it’s in retrospect clear that the credit flows to the European periphery that were sustaining growth were manifesting unsustainable.

I’ve spoken at length about the period after the crisis, more briefly about the period before the crisis. Of course the period before that was the 2001 recession, and the period before that was the internet bubble. So, if one asks the question, “How long has it been since the American economy enjoyed reasonable growth, from a reasonable unemployment rate, in a financially sustainable way?” The answer is that is has been really quite a long time, certainly more than half a generation. That’s why it seems to me that one has to contemplate macroeconomic theories of a very different kind than suggested by the conventional business sector theory.

Now, we’ll come back to this later, but, just as one other observation for motivating the general approach that I’m taking in talking about secular stagnation. The first bit of sort of empirical inquiry, empirical inference, that one learns about in an introductory economics course, I would suggest is this. You take some good, we’ll call it apples. If the price of apples is going up, and the quantity of apples is going up, then probably the demand for apples has gone up. But, if the quantity of apples is going up, and the price of apples is going down, then probably, the quantity of apples supplied went up, and so one learns to distinguish demand shocks from supply shocks.

The rate of inflation, the rate of expected inflation in the United States has never been lower in the post-war period, or at least the post-war period since 1950. Between the view that one should be looking overwhelmingly to the supply side, and the view that one should be looking overwhelmingly to the demand side. The fact that we are seeing deflation break out all over the world, would tend to lead one to suggest that at least a large part is a problem were on the demand side, rather than on the supply side.

Now, unlike in the case of apples, in the case of the macroeconomy, it is much too facile to suggest that one can completely distinguish the demand side and the supply side. After all, a rapidly growing economy is going to be an economy where there’s much more demand for investment than a slowly growing economy, and so, supply-side growth affects demand. As we’ve seen in the context of the last recession, a period of very short demand, leads people to leave the labor force – once out of the labor force, they’re unlikely to return – leads to reductions of the investment, which reduce subsequent potential. All I really want to convince you of, at this point, is that something profound has happened. Given its association with falling inflation, and diminishing expected inflation over the long term, we should probably think of it as at least being substantially related to demand factors.

Just to reinforce the point in another way, here is the yield curve on inflation indexed securities before the crisis, and very recently. What you see is that it has declined quite profoundly. I invite anybody to consider what’s happened in the – roughly speaking – year since I first started talking about secular stagnation. Quite a remarkable thing has actually happened. Even though the economy seems to have moved more strongly and smartly towards recovery, even though quantitative easing has ended, even though the fed has moved towards signalling that lift-off will take place, the expectation was that interest rates would rise over the last year.

What about expected interest rates five years hence? They declined by almost 200 basis points. Probably about 20 or 25 of those basis points have been recovered in the three weeks since this data comes from. I don’t want to insist on it being 200 basis points. Some of that was because, despite the recovering economy, expected inflation has come down substantially. More of it is because of market judgment about the real interest rate has come down substantially. Now recall, this is about what is expected between 2020 and 2025. People are expecting that the real interest rate will only 0.6%. For people who are technically oriented here, to the extent they believe the term structure slopes upwards, and so we’ll subtract the risk premium, that would reinforce the point made by these calculations. That also suggests some change in expectation about how the economy is functioning.

Alvin Hansen prophesied, or addressed, this kind of problem in the late 1930s when he spoke about secular stagnation, “Sick recoveries which die in their infancy and depressions which feed on themselves.” Hansen was manifestly wrong, in the sense that what prevailed after he wrote those words in 1939 was by no stretch secular stagnation. But, it has always seemed to me, and this is a good example, that economists have a tendency to suppose that because the world equilibrates and is stable, that that means they must operate with models that equilibrate and are stable. That’s fine if their model captures every aspect of the world, but their model should not have capture the Second World War, should not have captured the immense financial repression that took place during the Second World War, could not have reasonably been expected to capture the various features – the baby boom, the post-war housing boom, and more – that drove the economy forward. So, whether the experience invalidated Hansen’s model, or reflected the fact that variables that were exogoneus in Hansen’s model changed very substantially seems to me to be a quite open question.

This is not the stuff of modern dynamic macroeconomics. But, if whether you believe anything I say from this point on – I address this to the graduate students in the room – if you believe the general thrust of what I said about the magnitude of our current problem, and the long period of problematic growth, it makes you think that dynamics stochastic general equilibrium is kind of irrelevant. It makes you think, and what you have to understand, is that there is a big, bad, protracted thing. What you need is a theory of a big, bad, protracted thing. A theory of quarterly time series analysis, and the correlation between one blip and another has next to no potential to contribute to such an understanding. I keep coming back to this because I’m frankly more convinced that the question of understanding macroeconomic performance should be framed broadly the way I’m framing it, than I am with the particulars of the explanation I’m offering are exactly right.

Go back to basic Keynesian economics, and imagine that the point where the IS curve coincides with full employment involves a nominal interest rate that is lower than the attainable nominal interest rate. In that case, the creation, the printing of more money will be unavailing in generating economic growth. The declines in the price level through the vaunted mechanisms of wage price flexibility, will be potentially substantially counterproductive. Counterproductive because the extra money balances they create will provide no stimulus. But, the expectation of deflation will induce higher real interest rates, which will move up the IS curve in the wrong direction. The redistributions from debters to creditors will be adverse for spending, pushing the IS curve to the left.

There is no assurance, there is no theorem in general equilibrium theory that says that an economy where the interest rate is constrained, and I say constrained because one can imagine that the interest rate could fall a little bit below zero, as it now is in Switzerland. Or, one could imagine that there were other kinds of constraints. Fears about financial stability, for example, that prevented the interest rate from being allowed to reach zero. The point is that the interest rate is constrained, there is no assurance that the normal operation of the market will lead to the restoration of full employment. One can draw the picture in a different way by envisioning that investment is a function of the interest rate, and savings are a function of the interest rate. The point at which they balance at full employment, is an unattainably negative level of the nominal interest rate, and so the adjustment has to take place through a reduction in income that inhibits savings.

Now, I promise, I’m acutely aware of the lack of a rigorous micro-foundation for the simply relating the level of investment to the interest rate, and the level of savings to the interest rate. But, at the same time, I would suggest that if we’re not trying to understand quarters and quarter fluctuations, but are trying to understand the broad picture, this has to be a closer, more appropriate– Actually, close to appropriate approximation. Indeed, what is it that has happened over the recent period, there has been a pronounced increase in private savings, suggesting an increase in the savings propensity, at the expense of a diminution in the investment, and also a increase in private savings, and a decrease in the level of investments.

The Japanese case stands out. Japan has had essentially zero interest rates for 15 plus years, and chronically, the level of private savings has been in excess of the level of investment. It’s all very well to say Japan has an imprudent the budget deficit. One has to ask the question, “If private savings were equated without that budget deficit, what would happen to the level of demand in Japan?” It’s pretty clear where I’m going. It should be pretty clear where I’m going with this.

Secular stagnation is the phenomenon that the equilibrium level that savings are chronically in excess of investment, at reasonable interest rates. If secular stagnation was emerging, what would one expect to see? One would expect to see that the world real interest rate, or that the real interest rates of those places where secular stagnation was a pressing problem, were chronically declining. Indeed, what you see here is that for some long time period the real interest rate has been in substantial decline. Now Barry Eichengreen others have suggested that this may, in some sense, be misleading, because this may be back to some kind of normal, and the way we understand things is that real interest rates for some period around 1985 were aberrantly high. Perhaps so.

I’m not old enough to remember this, and therefore I assume very few people in this room are old enough to remember this, but it was the view of John Kennedy’s economic advisors that the 1950s were defined by three Eisenhower recessions associated with a chronic shortfall of demand for much of the 1950s as savings were in excess of investment. Yes, interest rates were low in the 1950s, and the 1950s had their kind of secular stagnation. Much of the 1960s and 1970s were characterized by unexpected inflation that enabled real interest rates to be lower than would otherwise have been possible. What’s true for the world is true for US TIPS. In this latest episode, unfortunately, the ten year real interest rate did not quite get to be negative, but on one day at the end of January, it was down to three basis points. Now, many of you will be thinking, “Well, this is all well and good, it’s sort interesting, and kind of clever in its way, but isnt he really missing the point. Hasn’t he noticed the fed is going to be rates sometimes, here, so even if we have been in a liquidity trap this is kind of about yesterday, it’s not about tomorrow.”

Look at this picture for a minute. Every seven and a half years we have to cut rates by four points. Every four and a half years, we have to cut rates by two points. So yes, we may start increasing interest rates, but if you believe there has been a substantial decline in real interest rates, and a substantial decline in the rate of inflation, will there be room the next time we have a downturn for a 4.5% point decline in rates? If the answer to that question is no, will the fact that there’s a risk of the zero interest rate constraint binding in the future affect expectations about future economic activity, and affect spending today?

Those of you in graduate school, or even undergraduate school, who have studied the consumption function a bit, and learned about liquidity constraints, have learned that just because I’m not liquidity constrained today doesn’t mean that liquidity constraints do not affect my spending decisions. The knowledge that if I spend up to the limit, I will be unable to borrow two years from now, causes me to have the need to maintain a buffer. In the same way, the observation that we’re not constrained all the time does not mean that the possible presence of the zero interest rate constraint will not affect behavior at all moments. I have seen various calculations, and the correct calculation depends on at just what moment one does it. I believe it’s an accurate calculation that at some point in late January, Markets were saying that the probability that the Fed Funds rate would be in the neighborhood of zero, in 2018 was approaching one third.

The point that I’m making, I would suggest is not an entirely hypothetical one in the United States. Of course, the case is by far weakest in the United States. To remind, the ten year interest rate in Germany is 35 basis points, and the ten year interest rate in Japan is 25 basis points. You can kind of work your way around thinking about of that 35 basis points, and how much is inflation, and how much is real interest rate, and you could have good debate about that, but you’re not going to get to a notion of real interest rates and inflation that is going to suggest that the market is prophesying anything sound for a long time. There is a global economy, and the question is to how long the United States can carry that global economy.

What I’ve tried to convince you of so far is that the stagnation hypothesis is something to be, questions of stagnation, and secular stagnation, is the right way to think about the macroeconomic problems of the age. That the idea of the lower constraint on interest rates is a crucial part of that, and there’s substantial reason to think that it’s much more important now than it has been in the past. What I want to do now is just review, very quickly, six or seven reasons for thinking that there has been a change in savings and investment propensities that make secular stagnation plausible.

One, demography. This is what Hansen emphasized. The growth in the work force in the industrial world, is trailing off very very substantially. The easiest way to look at it is to look at the change in the working-age population. If you do it more subtly, you probably reinforce the conclusion, because you had a big surge of women entering the labor force, and if anything, that tied is going out, suggesting a larger demographic change than you get just by looking at the crude population figures. It costs much less to buy capital goods than it used to, a canonical example being the computer. That means a given level of savings can buy much more capital than it used to be able to buy, tending to create any balance of savings over investment.

Another way to think about it, or a more sort of practical way to think about it, is to think about canonical leading companies, and their cash position. It used to be that the canonical, leading, fast-growing companies in the country needed to go to the bond market in order to expand, and couldn’t make dividends because they had so many investment opportunities. Think about Apple, as dynamic as any company in the economy. What activists are demanding it do is pay dividends and repurchase stock. Think about Google. Similarly awash in cash. That kind of, you can already think about.

My favorite example for thinking about these dynamics is think about two companies. Sony, the company is a strong company. It has factories, it’s got offices, its got tens of thousands of people working for it. It’s worth $18 billion. Now, think about Snapchat. All of it – the machines, the people, everything – could fit in this room quite comfortably. It will … It’s about to be valued by our nation’s capital markets at $19 billion. What’s that say, suggests that when you can start a company for nothing, and with nothing, that you will have the possibility of wealth creation without substantial investment, again, reinforcing an increase of savings over investment.

The developing world, for whatever reason, accumulating reserves on a very large scale, as I suspect was discussed earlier in the previous presentation, a set of portfolio changes that have led to very substantial increased demand for safe assets pushing down yields. To touch on a theme that’s important in this conference, rising equality and a rising profit share both operate to raise savings by putting more money in the hands of people who have high savings propensity, relative to those who have low savings propensity.

Inflation tax interactions, which I won’t take the time to explain, lead to downward pressure on the level of nominal rates as inflation rates decline. Sludged up financial intermediation also leads to lower rates on safe assets, relative to other ones. All of these factors operate to suggest that whatever the equilibrium real interest rate was in the past, it is likely to be substantially lower today, and suggests the relevance of the zero lower bound and the secular stagnation hypothesis as a challenge going forward for the industrial economies.

Let me address what seemed to me to be the most natural objections to this line of thought, some of which I’ve already touched on, and then say a bit about if it’s right, what should be done, and then throw this over for discussion. First question, can it really make sense for equilibrium real interest rates to be less than zero? As Paul Samuelson famously put it, “If the interest rates are less than zero, then you should shave off every hill.” Because if you shave off a hill that’s going to be better, for the railroad, over the long term it will yield benefits forever, and so it will be a positive NPV investment at a zero interest rate.

My reaction to this is I guess two-fold. It doesn’t mean that economies are dynamically inefficient, and what we make of this idea that real interest rates are low? First thing to say is that if you just calculate average real interest rates over the last century, they haven’t been very high. As an empirical proposition, real interest rates have a very low average, in fact, average substantially below GDP growth rates, real GDP growth rates, or equivalently, just take a measurement problem out, nominal interest rates are less than nominal GDP growth in most countries over most times. Second thing to say is the Samuelson syllogism depends on the perfection of property rights. It will not be worth my while to invest in shaving off a hill if I’m not sure how long I’ll own that hill, and whether I’ll be able to capture all that benefits indefinitely. In a world of taxation, and a world of uncertain property rights, the nexus between the productivity of capital and the real interest rate is much less clear. The third argument would be, “Look, there’s a market, and it’s out there, and that market has forecasts of real interest rates over a long time period.” I have not chosen the most dramatic examples. Index bonds in Britain suggests real interest rates below zero out for 50 years. As an empirical proposition we probably cannot dismiss the idea of negative equilibrium real interest rates.

Is the issue of secular stagnation a demand side issue or a supply side issue? I have focused on the demand side. I’ve done that, in part, because I’m doing what all academics do, but what policy makers must never do, which is focus on what I have to say that has something really novel to it. I have nothing new or original to say about, I may not have nothing new or original to say at all, but I certainly don’t have anything hugely new or original to say about the supply side. I already explained that a dominantly supply side explanation seems hard to relate to a period of declining inflation, and a period where much of the industrial world is actually experiencing something quite close to deflation. I would remind those of you who are very America-focused that if one looks at what I, at least, would regard as a plausible price construct, US inflation, excluding food, energy, and housing – housing because it’s a complex and somewhat arbitrary calculation in which the price of rental housing is used to make judgments about the cost of owner-occupied housing. Look at the last six months the United States of America had inflation exclude energy, and housing, it is to the nearest tenth of a percent zero.

These concerns suggest that something very important is happening on the demand side, as well as on the supply side. I would be the first to agree that it is quite plausible that there have also been aspects of productivity slow-down that have contributed to reduced economic growth that through the accelerator, mean less demand for investment, that have contributed to the stagnation phenomenon that I described.
Past fears of secular stagnation have indeed proven to be unfounded, but as I already tried to explain, that is largely because exogenous variables changed. It is certainly possible that exogenous variables will change in the future. Indeed and I hope that they will. In an important respect, that is the task of economic policy.

Isn’t the United States approaching full employment? Maybe. The wage inflation would not be overwhelmingly corroborative of that view to this state. Even if it is, how long will it stay at full employment, and isn’t there a substantial risk that the zero constraint will bind before monetary policy can address the next recession?

To remind, and just to score a note of humility regarding economic science, if one looks at recessions since the Second World War in the Unites States, of which we had substantial sample, not a single one was predicted one year in advance by the Council of Economic Advisors, the Federal Reserve Board, the Congressional Budget Office, the IMF, or the Survey of Professional Forecasters. Serenity that somehow it’s okay, because we’re doing okay, we’re pushing the interest rates up a little bit, we can get them off the floor, and there’s going to be a little bit of room for them to move down, seems to me to be not entirely justified.

What’s to be done? If you take this thesis at all, there are three broad approaches going forward. I’m going to talk about them more quickly, because I really wanted to focus, at least the students in the room, on what I thought the question was a macroeconomist should be thinking about, as much as on my particular explanation.

Structural reform. Structural reform that opens up private investment, that raises the demand for private investment, has to be a good idea. Just how much scope for it will vary from place to place. Economists are always for structural reform, and I am too. I would highlight, though, that if you have a gap in which supply is in excess of demand, and in which output is constrained by demand, measures which increase supply will A) not raise the level of output, because it’s constrained by demand, and B) increase the amount of deflationary pressure.

It seems to me that in much of what is said and written about structural reform, there is a unwillingness to distinguish between structural reforms that are demand-promoting, and structural reforms that are supply-promoting. A fair amount of what is suggested seems as likely to be supply-promoting as to be demand-promoting. Structural reform has in any event been the dominant policy mantra in Japan for two decades, and in Europe for the better part of a decade. I think it’s fair to say that the fruits have so far been limited. I would also note that some of what is described as structural triumph has a significantly zero sum element. Much of German success, in particular, is attributable to its emergence as highly competitive, and a large scale exporter. The one thing that economics does know unambiguously is the sum of the world’s exports has to equal the sum of the world’s imports, and that for every surplus, there must be a deficit. Insofar as German policy has taken the form of enabling it to benefit from a large excess of exports over imports, it is almost by definition not a universally replicable model for success, despite how frequently it is urged.

Yes, there is a role for structural reform. What about monetary policies? There is both a short run and a long run question here. The long run question is why don’t we make a base and regular rate of inflation be 3%, or 4%, or 5%, and then there will be plenty of room for the real interest rate to be -3%, or -4%, or -5%. If this zero constraint is really an issue, why don’t we just raise the normal inflation rate to a point where we’re never near the zero bound? That argument has been made by Ken Rogoff repeatedly, and by Olivier Blanchard once, because of what went with being a sitting official of the IMF.

I think it has considerable force as a second-best kind of argument. Two of the biggest weaknesses are that if you frame a question as, “What number should be your inflation target,” then two’s a number, two’s an integer, three’s an integer, four’s an integer, five’s an integer, and they’re all integers, and it doesn’t seem like it’s, if here’s a good to have a higher integer, why not have a higher integer rather than a lower integer? The real case to why two seems like a good integer, is that two is kind of blurry near zero, and you have the idea of price stability. You could say we’re going to defend price stability, and you could understand that in kind of a blurry way from one to two, but can’t really define price stability as four. Once four, why not six? Once six, why not eight? I don’t know a real way to evaluate that argument. There are relatively few countries that have had stable high single-digit inflation. It has a tendency to become higher still. This strikes me as a dangerous recommendation. I’m not confident that it’s worse than what we’re living through, but a dangerous recommendation.

The second problem is that there is, I think, a real question about the allocative efficiency of very very low real interest rates. How desirable really is it to spur those investment projects which were not worth doing with a 0% real interest rate, but only become worth doing with a -2% real interest rate. Is that really the right public policy for stimulating investment? I’m not sure.

There is the further concern that very low real interest rates, particularly when also associated with very low nominal interest rates, may encourage all kinds of imprudent risk-taking. Similarly, similar kinds of arguments apply to what I think is a pretty fully played-out suggestion. Forward guidance involves the same kinds of issues. I’ve always been struck that forward guidance runs the very real risk that the market will not believe forward guidance, and so no substantial stimulus will be delivered. But the Central Bank will feel constrained by some forward guidance, and so when the time comes, will not raise rates in a way they otherwise would have, and so you will get the worst of all worlds.

Again, as between do nothing and forward guidance, forward guidance looks okay, but it’s really not great. Then there is of course the question of quantitative easing, where there are real issues once interest rates are very low of just how much extra investment will take place at 35 basis points on the bund, that would not have taken place at 60 basis points on the bund. How valuable and productive will that investment be, and to what extent will financial stability be undermined? Not least, because with very, very low interest rates, loans involve essentially no coupon payments. Therefore, there is very little monitoring applied to how effectively borrowed proceeds are being used. Yes, there is a role for monetary policy, but it’s not a role that leaves me hugely inspired in addressing secular stagnation, though probably better than the alternative.

I believe the case is much stronger for structural measures to promote private investment, and for expansionary fiscal policy. Expansionary fiscal policies, notice operate, it should push the IS further to the right, to increase equilibrium real interest rates. They are a natural market kind of response to low borrowing costs. As I said many times, thinking about the American context, if a moment when they borrow money for 30 years at close to 2%, in a currency we print ourselves, is not the moment to clean up Kennedy airport, when will that moment come?

There are a set of further rationale for increased infrastructure investment. Can it possibly make sense that at this moment, as I speak to you, the share of public investment in GDP, adjusting for depreciation, so that’s net share, is zero. Zero. We’re not net investing at all, nor is Western Europe. Can that possibly make sense, given the demand issues, given the productivity of public investment, and given that if we have a moral concern about my children’s generation, deferring maintenance is just as surely passing the burden onto them as issuing debt. The burden of deferred maintenance compounds at a rate much greater than zero in real terms. There are other measures that I describe here to promote spending.

I’ll just close with what I think is probably the most remarkable IMF document in the 25 years that I’ve followed the IMF closely. This is not some Keynesian economics professor. This isn’t some guy with some model, arguing for some thing. This is the IMF. It’s not a research-working paper of the IMF. It’s the flagship publication of the IMF. It asks the question, for industrialized countries, if they spend 1% of GDP more on infrastructure, what would the consequence be for their debt to GDP ratio after five years? This is their estimate, not mine. They say that it would be 6% of GDP lower. Why? Because increased economic growth means increased cash revenues. Increased growth in the short run means increased potential in the long run.

What I would suggest to you is that we need to think about the fact that we are likely heading to a world where the most important arguably market price in our economy, the real rate of interest, is going to be substantially lower than most of us have been accustomed to. That is on the one hand telling us something very important about the fundamental forces of savings and investment in our economy. It is on the other hand something that should be a strong impulse for bringing about changes in what we conventionally think of as a [inaudible], that will restore stronger economic performance through a more appropriate investment and saving path. This is, I believe, central to understanding the macroeconomic circumstances of the industrialized world going forward. Thank you very much.