George Selgin on the Productivity Norm, Deflation, and Monetary History

George Selgin is the director of the Cato Institute for Monetary and Financial Alternatives and is a former professor of economics at the University of Georgia. He is widely published in monetary and banking theory, monetary history, and macroeconomics. George joins David to discuss what the Fed got wrong in the lead-up to the recent financial crisis. He makes the case that central banks, rather than focusing on the price level or inflation rate, should instead allow inflation to reflect changes in productivity growth. Selgin examines the Great Deflation of the late 1800s and dispels some of the popular myths surrounding that period.

Read the full episode transcript

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected].

David Beckworth: George, welcome to the show.

George Selgin: Thanks, David. It's nice to be here.

Beckworth: It's always good to have my former instructor back, just like the good old days of Monetary Economics with George Selgin.

Selgin: I'll try not to give you such a hard time.

Beckworth: I appreciate that. Let's begin by talking about an idea you've developed. That is the productivity norm. Can you tell us what it is and what it implies for monetary policy?

Explaining the ‘Productivity Norm’

Selgin: Sure, David. The productivity norm, the name is somewhat misleading. It actually refers to a norm for how the price level ought to behave. What it means is that contrary to the very conventional and still very popular view that ideally the price level or at least the inflation rate ought to be as stable as possible, a constant, whether it's two percent or some other number.

Zero used to be the favorite. Contrary to that, under productivity norm, the rate of inflation is allowed to reflect the economy's underlying rate of productivity growth, specifically by adjusting opposite with adjustments to the rate of productivity growth.

In a more rapidly growing economy where productivity is growing more than usual, you would allow a lower inflation rate. In some proposals like my own original one, even a modest rate of deflation to reflect the fact that the economy's getting more productive.

That's another way of saying to reflect the fact that the average unit cost of production of many goods and of goods in general is falling. So if goods get cheaper, prices should generally fall.

Conversely, if you have an adverse supply shock, the rate of inflation should be allowed to go up. The reason for advocating this norm is that I believe if inflation is kept constant despite fluctuations in productivity growth that itself can be a source of trouble. The productivity norm has a link to proposals for market monitorists in that they all involve treating stability of nominal spending rather than of the price level is what's really desirable for overall macroeconomic stability.

Beckworth: Many economists would agree with your point that if there's a negative supply shock so there's sudden reduction in supply of oil, some natural disaster, that prices should temporarily be allowed to go up because you wouldn't want to tighten policy in a situation like that.

They have a hard time with the flip side of that argument, that if there's a positive supply shock, then maybe some disinflation, maybe even mild deflation should be allowed. Why is there this asymmetry in probably most economists thinking?

Understanding Inflation

Selgin: Economists are spooked by deflation generally. That spooking seems to have mainly come about as a result of what happened in the 1930s.

It's important to acknowledge, because some people don't, that, yes, deflation can be very spooky, but it's spooky when it is a consequence of a collapse of spending in the economy or NGDP or whatever metric you want to use to measure spending. In that case of course, when spending collapses, it's impossible for the average firm to recoup its historical expenditures and to make profit.

We have the average firm losing money, which is not a good thing because it's not a useful signal to the economy. It just signals that there's a shortage of money in the economy. That's what happened in the '30s, both at the beginning of the decade and later on, in '36, '37, and '38.

There is such a thing as bad deflation. However, there's also such a thing as good deflation, which is the kind of deflation you were just referring to, which is the flip side of adverse productivity shocks or supply shock where you have positive productivity shocks.

Actually, people are perfectly happy to tolerate good deflation if it's limited to a few goods. But what that means in practice is they're only willing to tolerate it if some other goods are getting more expensive so that the price index or inflation rate measured with conventional price index is not changing.

It's OK for the price of computers to fall, but by gosh something else had better go up to keep the overall price index or the inflation rate from changing.

The reality is, more often than not, productivity improvements are happening in many sectors of the economy at once, affecting many goods at once, with the improvements outweighing or outnumbering situations where unit costs of production are rising. In that case, why not let the general rate of inflation and the general price level reflect the overall changing state of productivity?

Because they've been spooked by the '30s and some other episodes, economists are very reluctant. They tend to speak of deflation as if it was always demand driven. Ben Bernanke routinely does that. He seems to be incapable of recognizing this other kind. He does once in a while very grudgingly.

In fact, if I may add, now there have been quite a few studies of historical deflation. Kito and Atkinson, Boreo, and others, Claudia Barre, they all come to the same conclusion, that in fact the good sort of deflation I was just describing has historically been more common than the other sort.

Therefore, we have more examples of that kind of deflation, the sort that there's really no reason for the authorities to prevent than we have of the bad kind.

Of course we have to be wary of the latter. We certainly don't want risk, allowing it when we shouldn't, when we don't want to, but that alertness should not be allowed to translate into an absolute refusal to allow prices generally to ever fall even when that means you have got to artificially expand nominal spending to prevent that fall.

Beckworth: Let's give a concrete example of this benign deflation. The post‑bellum deflation period, right after the Civil War up until 1896, before they had the big gold discoveries, was a period in US history where we saw on average rapid economic growth as well as a consistent one to two percent deflation a year, yet the world didn't end.

Can you speak to that experience and what it says about benign inflation?

Selgin: Sure. I will. It's a very important experience. It has been widely misunderstood. In fact, deflation of the sort you describe was roughly occurring not only in the United States but in all of the gold standard countries at the time, so it was a phenomenon that was linked to the gold standard and obviously was due to the fact that the gold standard was allowing for.

In general, just enough money growth to offset extensive expansion in production but roughly not enough to meet the expansion in production that was related to productivity improvements, I say roughly because we should not pretend that the gold standard exactly got the productivity norm right. It certainly didn't. But it came pretty close.

A lot of people have based on US experience said, "This was terrible. We had all these crashes, etc." Some even have treated the whole period from the 1870s to the 1890s as one big slump because prices were falling as if deflation were identical to depression, which is certainly not true.

As a matter of fact, of course, the US had many severe downturns, mostly relating to financial crises. These were a peculiar result of problems with the US banking system that weren't repeated in the other gold standard nations.

During those episodes, we tended to have some bad deflation mixed in with the good. That has also been a source of confusion.

As for the general background secular deflation, there's no link in the US any more than in any of the other countries between that and any general slump. We know that people were not unemployed generally at high rates. We know that people didn't feel depressed for most of these periods. Only during those financial crises and for some time after did you really have a slump.

Beckworth: People confuse those crises, 1873, 1893. They were really sharp, but, as you said, they were tied to this faulty national banking system which was very conducive to financial crisis. A good counterexample is the panic of 1907. If you 1907, you go to an inflationary regime.

The panic in 1907 was the worst financial crisis of that post‑bellum period. It speaks to the fact that the financial crisis wasn't a product or a function of the deflation. It was a product of the system because it occurred in an inflationary environment as well as the deflationary environment.

Selgin: That's absolutely right. In fact, there are all kinds of evidence you could bring to bear on this point because you could look at different countries. You could look at Canada.

Canada had the same gold dollar. They had the same secular deflation. They didn't have the crises though because they had a different monetary and banking system without a central bank for what that's worth.

If they had all those elements in common, it can't have been one of these elements they had in common with the United States that was responsible for those crises that did occur.

It had to be something peculiar to the United States. Indeed, after the Klondike and other discoveries, South Africa and especially of the 1890s, the deflationary trend ended. It was replaced by an inflationary trend, but this change had essentially no bearing on the incidence of crises in the US or anywhere else.

Beckworth: One thing we probably need to stress is...This is a question that I often encounter is, "How would the central bank know when it's good deflation or bad deflation?" The productivity norm, correct me if I'm wrong, you don't have to worry about that. It will take care of itself. All you have to do is focus on stabilizing demand. Is that right?

Selgin: That's exactly right. In fact, it's a widespread misconception, by which I mean widespread among the relatively small number of people who have ever heard of a productivity norm, that it calls upon the central bank, assuming that you have a central bank, to take account of productivity changes and react appropriately to them. But that's exactly wrong. That's exactly the opposite of the truth.

It's when the central bank is determined to maintain a constant inflation rate that in principle it has to try to anticipate productivity changes because those changes will affect the rate of inflation unless it takes counter monetary policy actions.

Every central bank that's determined to keep a stable price level target, say the ACB trying to keep two percent, is obliged to inform itself to try to forecast productivity innovations in order to offset them.

Under a productivity norm, those innovations are among the things the central banks should not pay any attention to and should not respond to. So it's one piece of information less that they need to anticipate. Indeed, when you look at the factors that determine output...Because a stable price level...

Let's just speak of stable price level. People can adapt what I'm saying for the case of a stable rate of inflation. What it really calls for in famous formulation is having the amount of money that's chasing goods only adjust as the total amount of goods adjusts, which means you're anticipating changes in total output in trying to have a money supply expansion allowing for changes. Velocity keeps up with that.

The component of output change that's hard to predict is the productivity component. All the others are relatively stable, population or labor force growth, the growth of capital, stock, that sort of thing. But intensive changes in productivity, total factor productivity, those are the ones that are the most difficult to forecast.

If central banks are going to make mistakes in attempting to keep prices stable, they're going to make mistakes with respect to their forecasts of total factor productivity growth. The productivity norm says, "Don't bother. You don't need to offset those. Therefore you don't need to take the risk of doing it incorrectly."

Beckworth: Some people might wonder, "Then do we have a nominal anchor?" But the answer is yes. We were still tied to a nominal measure. In the long run, that anchors down trend prices. But in the short run, there can be deviations. Maybe I should step back a minute.

The Fed was if it were the Fed doing a productivity norm, it would keep spending growing at a stable growth path or growth rate. The question would be how that growth rate was broken down into real growth versus changes in prices. One year it might be higher real growth, lower prices. The next year, it might be the opposite. Is that fair?

Selgin: That's right. A lot of people are hung up on that because they say this would mean that if you're stabilizing nominal spending, this means you're allowing output and the price level to vary more than they would if you were perhaps trying to stabilize some weighted average as in a Taylor rule, for example.

That misses the point because the goal of monetary policy in my opinion and that of market monitors as I take it is not to aim for any absolute stability of either P or Y but rather to avoid monetary influences on Y, deviations from the natural rate we might say. That's a question of stabilizing demand, which is to say total spending.

You can think the amount of NGDP as just indicating the position of the aggregate demand schedule and aggregate supply demand space. In that case, to the extent that the natural is changing and shifting, particularly in response to productivity gains and losses, then that should be reflected in changes in the price level or inflation rate.

I make an analogy or rather respond to an analogy in talking about this where some people say, "Having a price level that varies is like having a thermometer that isn't measuring the temperature correctly or a yardstick made of rubber," and so on. But it's not correct.

Under a productivity norm, it's true that the price level varies, but it's also true that in the world, the temperature varies. The thermometer registers different temperatures that are a good thing.

Similarly, let's say we have a ruler, and we're measuring the average height of people. Sometimes the average height of the entire population changes. We wouldn't want to change our rulers so that average is always constant. Five feet in the Victorian era so it must be five feet today.

Similarly, when the price level changes under productivity norm, it's not telling you that you've got a broken yardstick or rubber on or bad thermometer. It's telling you that the goods are getting cheaper to produce or getting more expensive to produce. Very useful, informative signal.

Then there's another side to the argument which is remember when we're stabilizing output prices or rather allowing output prices to change under productivity norm. What we are stabilizing is an index of factor prices in principle.

Conversely, those who would stabilize output prices in the face of productivity innovations are destabilizing factor prices. They think that they're stabilizing the price level and imagining that this is all prices.

They're forgetting what price level stabilization or inflation targeting involve in practice is looking at one set of prices and not looking at the other. You can make all kinds of good arguments. Economists through the ages have made them, that it's more important that factor prices be kept stable than final good prices.

Beckworth: This speaks to a criticism that I often hear or at least recently have heard. That is that the nominal GDP target. That's what the productivity norm is to be very clear. It's a special kind of nominal GDP target.

Selgin: One of the earliest types.

Beckworth: One of the earliest. Now I stand corrected. But it's basically a nominal GDP target. I believe Marvin Keene had a paper that made this point that is we have a nominal GDP target every so often, we have to out and estimate potential GDP and adjust the target.

That kind of defeats the whole purpose of a nominal target, the whole point of that. Along those lines, could you also speak to this idea that when we're measuring the effect of monetary policy or the stance of it, there's this idea that you've got to decompose nominal measures into the real and inflation as opposed to measuring the actual observed nominal measure.

Selgin: On this question of what the central bank, let's remember that if you have a central bank targeting the inflation rate, what that means is that setting fluctuations in velocity aside which all of these norms the central bank would have to offset changes in velocity. So let's set that aside.

Then, under an inflation targeting regime, what the central bank has to do is try to forecast uncertainty to be aware of current changes in total output. That's little y in the equations. That means that it has to know about both intensive and extensive changes in output where extensive changes are changes based on increased factor input.

Intensive are based on changes increased factor productivity. As I said earlier, it's a factor of productivity that's the hard one to anticipate. It bounces around a lot as we know. From recent experience, we know that very well.

The other one, the factor input, doesn't change that much. It's relatively easy to forecast that. If you're a baseline central banker trying to make sure you hit a fixed inflation rate target, you've got a big forecasting challenge.

If you're just trying to anticipate, to fix an AGDP growth target, you have a considerably easier challenge because you're only having to anticipate that input factor. That is relatively less volatile and easier to guesstimate. They'll get it wrong, but they're not as likely to get it wrong as the other component.

Beckworth: Maybe another way of saying that is if you're an inflation targeting central bank like most are, they have to divine what's causing the changes in inflation to get it right. If it's a velocity change or demand change, they need to respond to that. But if there are supply changes, productivity driven changes, they need to ignore those.

Selgin: Under my regime, they need to ignore the productivity changes, but the point is that's just a matter of not bothering to forecast those.

Beckworth: Right, it's far easier.

Selgin: The problem is that under the inflation targeting regime, they need to forecast future productivity as well as future factor input in order to make sure that changes in neither of those things result in changes in the rate of inflation because they have anticipated and offset them with monetary policy. They've adjusted their monetary targets accordingly.

Whereas with the productivity norm, you don't worry about factoring, but you do worry about if lots of people immigrate and your labor force changes. You don't want to not accommodate that.

Simple way to think of that is that that would put downward pressure on factor prices, specifically wage rates, so you want to have some monetary accommodation to avoid it. But you wouldn't worry about the fact that prices fall because of productivity gains. You would just let them. You wouldn't anticipate them.

The fact that you didn't forecast productivity net, it wouldn't concern. You wouldn't try. The price level would change. You would only have to worry about your mistakes in forecasting factor input.

Beckworth: Right, so your job would be a lot easier.

Selgin: A whole lot easier.

Beckworth: What about the history of thought on that? You resurrected this idea. I think you've done some research pointing out that this actually was a fairly common idea pre‑1930s. Is that correct?

History of Thought on a Nominal GDP Target

Selgin: Oh yeah. Of course people in those days didn't speak about NGDP targeting or anything close nor did they use the expression productivity norm.

Until the 1930s, there were more economists who subscribed to some ideal involving a stability of spending, MV or P times Y, than who advocated stability of the general price level. Mind you, most of these discussions were in the context of gold standards and that sort of thing.

Probably more than any other position, the most popular was that favoring retention of one of these standards as approximating some such ideal better than other institutional arrangements could. Economists back then I think were less naive than they are today about how central planners could implement their ideal theory.

Every economist seems to talk about...Most economists talk about central banks today as if they running them, they would do exactly what their theory says is right. In fact, of course, the central banks disappoint all of them.

Before Keynes came along, you had quite a few economists including many famous ones from different schools, offhand, Dennis Robertson, Alan Fisher in New Zealand, John Williams at Harvard, Fairduke Hayek, Coopins, a Dutch economist, Merdal. Vixel was an exception.

We all know thought that keeping interest rates at their natural levels was the same as keeping a stable price level. People didn't speak of inflation targeting back then. If they advocated a stable inflation rate, it tended to be zero rate.

There was a big debate between Vixel and other Swedes, notably a fellow named David Davidson about this.

We could trace this back much further than these neoclassicals I've been talking about to persons who spoke of a labor standard of value for example in the early 19th century which was quite a different thing from Marx's and others labor theory of value.

What they meant was their conception of an ideal standard would stabilize the price of labor, which is a kind of productivity norm. So it's an old idea.

Very interestingly, as I've written in an article for "History of Political Economy," Keynes himself comes very close to conceding the merits of a productivity norm in "The General Theory" or at least the merits of a monetary policy that would stabilize not output prices but wages or nominal wages. Then he waffles.

He's back and forth in the general theory went back and forth and finally ends up embracing a stable price level norm.

By the way, yes, I just said that Keynes favored a stable price level norm. He does in "The General Theory." It's just that in that book, he's assuming pretty much throughout that the economy is in a state where the equilibrium price level is below the actual price level.

In that situation, of course, any kind of monetary expansion doesn't raise the price level. It just gets you back towards that price level once again being in equilibrium.

Beckworth: The productivity norm is actually an old idea. You resurrected it. You've published on it. I was exposed to it.

I've talked to Ramesh Pumuru who's the writer for "National Review." He was influenced by your work on productivity norms as well, which also lend him into nominal GDP targeting. You work on that has been influential.

I want to think of a new application for it. I want you to hear me out and tell me if this makes any sense. Today there are many who argue that technology is rapidly taking off. There's increased digitization of society, more networking, more smart machines, 3D printers, driverless cars. Some are concerned this is going to accelerate. We're going to have more and more rapid technology growth. They're concerned that...

Selgin: Heaven forefend.

Beckworth: It'd be a great future. There are some, particularly on the left, who are concerned that this will be very disruptive in the short run, and capital will benefit more in the short run than labor.

There's concern about massive structural unemployment as such. Let's take that as given that that's the case. Wouldn't a productivity norm be exactly what we would need in terms of monetary policy for a situation like that?

Because labor would more directly share in those real economic gains through a lower price level. Their nominal wage would be stabilized, but the rapid technological gains would be shared with them via lower prices.

Selgin: Yeah, I think logically you could say that. I suppose you could put it more pithily by saying that if you're out of work, you'd rather be out of work with prices falling than with them staying the same. But of course this is little consolation if you're not getting any income at all which I suppose in the more dire cases might happen.

Not to be so facetious about it, anybody on fixed income benefits under a productivity norm from productivity gains along with all consumers. That includes welfare, social security recipients, and so on, presumably people on unemployment.

Whereas the standard regime that seeks and achieves a stable inflation rate tends to leave the same people out of luck when it comes to partaking of a productivity improvements that others are enjoying because they're enjoying them through increased nominal salaries and earnings. So there's a real serious distribution implication.

I think you're right. That all things considered, a productivity norm tends to be more favorable to groups that are not particularly at the high end of the food chain than price level stabilization. But I hesitate to go too far out on a limb on this because of course there are many well‑to‑do people who also live off of fixed incomes. They tend to be retirees.

There are clearly some real distributional implications of having a productivity norm versus a stable price level that ought to be taken into account with all the other consideration.

Beckworth: That approach would make someone who was much more free market friendly, someone who believes in markets over government, they would embrace that approach compared to some of the other proposals should this world come where machines are doing everything. Again, this is more of a transition story.

An alternative proposal that's often been discussed is that the government should start buying shares of S&P 500 and then take that and pay it to individuals, which would be very much an interventionist approach. Whereas a productivity norm could kind of get the same effect.

Selgin: You know, David, people think that I don't know anything about anything except monetary economics and only ask me questions about that.

At the risk of proving them right, I'd like to say something about this argument about technology because it's always struck me as something very weird, this idea that technological innovations occur in such a way as to result in a tremendous serious mismatch of available skills with available technologies.

Seems to me rather that what's more likely to occur is that innovations take place in response to the labor situation that people don't just invent machines that nobody can run and put them on the market apart from training and everything.

What I mean is that if you don't have the basic skills out there where you could at least train people and make these machines work, conversely if you do have large pools of unemployed labor, you'd innovate in a direction which take advantage of that.

Let me make a more simple example. Suppose that somebody came up with an improved machinery that relied on using some very rare element that wasn't available. That just won't fly because it's too expensive to come up with that element.

If a lot of that stuff suddenly became available, somebody finds a vast new source of some rare commodity or mineral, technology would respond. Technology is doing these things all the time.

When people come to talk about labor shortages and skills, it's as if the technology just fell from the sky willy‑nilly whee the people working on it didn't give a thought to what the actual available resources were.

This doesn't make any sense to me. I'm speaking outside of my expertise but still, somebody needs to consider then when people are doing all this R&D and coming up with these innovations, they're not simply trying to replace labor.

They're trying to replace costly methods of production with cheaper methods of production in a way that's informed by the actual scarcity of the different resources in the economy. At least people should try to argue as if they were aware of this tendency.

Beckworth: I understand. Let's talk about the implications of the productivity norm, not necessarily just the rule itself but the idea that there can benign deflation as well as malign deflation. Let's apply it to the housing boom period.

Let's talk about that period. There was a big productivity surge right after 2001, rapid gains in total factor of productivity. If you go back and read newspaper accounts, there's just all this discussion going on at that time. It really threw a curveball for the Federal Reserve.

Can you explain how it did that and why this may have been a reason the Fed blew it according to some during the housing boom period?

Applying the Productivity Norm to the mid-2000’s Housing Boom

Selgin: As I mentioned before, productivity's hard to forecast. It's also hard to measure. That's another reason why you shouldn't try if you don't have to. Policies that don't require it are better than policies that do.

Having said that, you're right, most estimates of a total factor of productivity had it growing very rapidly after the 2001 dot com crash which mud by the way involved a setback to or an interruption of what had been a similar productivity boom before the dot com phenomenon and that in fact a boom that probably contributed to the dot com boom. It was underlying the dot com boom itself. Then you had a crash.

The Federal Reserve was aware of this rapid productivity growth, but because they were committed to an ideal of stable inflation, they perceived it as supplying a grounds for keeping money easy or easing it. Because otherwise, there would be a tendency for the inflation rate to drop below target.

It's interesting that some of the economists or some of the FOMC members involved including some who were economists openly acknowledged that this probably means going in a sense setting rates below their proper equilibrium values.

Here I think they were to some extent thinking in Vixelian terms. But this was OK because it is not going to result in inflationary pressure. The idea is, ah, look, productivity is growing really rapidly. We can ease monetary policy.

We can keep interest rates lower, even lower than depending on which time in this period we're talking about. It won't have inflationary consequences. Isn't that great? Because that allows us to do more to help the economy to get back on its feet after this crash.

Productivity gains instead of being treated as a reason to modify the inflation target are treated as an opportunity to ease monetary policy or keep it easy without having to face the consequences of a higher headline inflation. This was a big mistake because in fact it involved overly easy policy according to a productivity norm standard or an NGDP growth standard.

Now the NGDP growth rates weren't phenomenally high, but they were higher than usual. They were a couple percentage points higher depending on what long run trend you look at. It would be a bad mistake of course to say that because of this you had this terrible subprime boom and consequent bust. I don't think anybody thinks that.

What I think people think who believe that monetary policy was too easy at this time and what I think is that in combination with many other circumstances regulations, policies, what have you, some private market developments too of course, the excess credit that was being created here is the counterpart of the easier money, which all tended to be shunted into the mortgage market and the subprime mortgage market especially.

We had a perfect storm of conditions in which one element of that perfect storm was excessively easy monetary policy, and what was informing that excessively easy monetary policy was the view that when productivity grows more rapidly, that's not only a license for but an invitation to have an easier monetary policy.

Beckworth: The Fed in the early to mid‑2000s, they eased beyond what they should have because they saw low inflation. Is the story that they took advantage of the low inflation, said, "Hey, we can ease"? Is that the story? Or did they simply misread the inflation?

Selgin: They correctly read the productivity figures.

Beckworth: They knew that inflation was low because of productivity.

Selgin: They knew it would be low. They knew it would be lower than usual for any given monetary policy stance.

Therefore, they took this as an opportunity to have an easier monetary policy stance, saying to themselves and to each other, "We have some softening. We have some downward pressure on prices coming, so we can put more upward pressure on them to offset it," upward pressure on prices from easy monetary policy.

In that sense, they're treating the anticipation of continued rapid productivity growth as allowing them to set an easier monetary policy target than they would have otherwise. This is what was informing the decision first of all to lower the target all the way to one percent, but more importantly, to hold it there for as long as it was held there.

Beckworth: Because one of the challenges we referred to earlier is that when you're doing inflation targeting in real time, it's sometimes hard to know why inflation changes, if it's a supply shock, a demand shock. What you're saying is in this period is they did know why it was going down. They just took advantage of it. They wanted a free lunch and monetary policy.

Selgin: The point is they didn't care why it was going down, which is perfectly consistent with having an inflation target. If you believe a stable inflation rate is what you want.

I'm of course here abstracting from the fact that the Fed's mandate is more complicated than that, but in fact if you throw in the actual Fed dual mandate, you have an even more compelling reason why they would treat any opportunity to ease monetary policy during a recovery as one they should take advantage.

Here what the way they read this was we have a problem where we have recovery that's sluggish. We want it to be faster. We're under an obligation to maximize employment and not blow our inflation, not let inflation get too high. We can do that because we've got a productivity surge.

Beckworth: One of the arguments for flexible inflation targeting is that you don't hit two percent this year or next year. It's over the median term. On the average, you hit two percent, which in theory gives the central bank some flexibility in what inflation is one year to the next. It gives them the room to respond if there's some kind of shock that hits the economy.

Some argue that if they did flexible inflation targeting right, it'd be very close to a nominal GDP target. They would respond properly by ignoring supply shocks. As an example, late 2008, the Fed saw high inflation.

They were worried, so they didn't cut rates. They misread the inflation tea leaves. The ECB actually raised rates in 2008. It raised interest rates twice in 2011 because inflation was going up. I would argue those were supply shocks they should have ignored.

Selgin: That's right. Look, flexibility is vastly overrated because it's true that flexible inflation targeting could allow for, does allow for, the kind of response that you and I might like to see to productivity innovations.

It also allows for all kinds of responses that have to other things or responses to productivity innovations different from what we would like to see or opposite them. Flexibility sounds great.

If you're the kind of person who is optimistic that it'll be used the way you think it should be used, then you always favor flexibility. This gets us back to the more fundamental thing about central banks and rules versus discretion.

Beckworth: It takes us back to the knowledge problem.

Selgin: It takes a very fundamental issue. Most advocates of monetary discretion, it's rather sad to say, they favor it because it allows anything to happen. That infinite set always includes what the advocate of discretion thinks is ideal. What the advocate of discretion is saying is if the central bank could only do anything, it'll do what I would want it to do. Or it's assuming that that's the most likely outcome.

The advocates of discretion never assume that the central bank will use the discretion to do something that they think is terrible, whereas the proponents of discretion, of rules, whether it's targeting the price level or inflation rate or a productivity norm rule if you like or nominal GGB rule are assuming there's a high likelihood that if the central bank could do something else, they'd pick something worse. It's as simple as that.

Beckworth: Even now we see this as the Fed heads into 2016. There's uncertainty. Is inflation low because oil prices are low? Or is it low because the economy is weakening? We just don't know in real time. That's one of the advantages of a productivity norm, a nominal GDP rule.

Look, all you do is keep it simple. Focus on stabilizing demand and let the other pieces fall where they may. It's not the job of the central banker to play god and try to figure out what is happening to the real economy.

Let me move on to the Fed's mistakes in 2008 because I know you've been doing a lot of work on that recently. I have been very critical of the Fed during that time. You've taken a couple of interesting pieces on that period and worked through them.

You've looked at the Fed sterilization of its bank lending program in 2007 through 2008 as well as the Fed's excess reserve policies. Could you explain what those are and how they may have contributed to the worsening of the recession in 2008?

Selgin: Throughout 2008, particularly after the crisis really started to pick up with the Bear Stearns event, the Fed was engaged in all kinds of emergency lending. They didn't call it QE anything yet.

Retroactively, they would call some of the more massive expansion after the Lehman failure retroactively, they would call that QE1. But at the time, they weren't thinking of QE. They were simply rescuing or bailing out various financial firms, etc.

Their balance sheet would have grown in the course of this emergency lending. Eventually it did grow, but the Fed was throughout this period concerned that the rescues which it perceived as being aimed solely at helping these particular enterprises would inadvertently result in an easing of general monetary conditions.

They did not want that to happen because they had their idea of where the federal funds rate target should be, and by gosh, they were going to stick to it.

This is all well and good if the targets where it ought to be. In retrospect, it's pretty clear it was higher than it ought to be, that is, that some monetary easing really was desirable at this time.

By the way, a conventional understanding of the rule of a central bank during crisis would lead one to think that providing more liquidity would have been the natural thing to do, not just to particular firms but to the economy in general.

Nevertheless, the Fed was determined not to do that. They prevented the general liquidity or monetary conditions from easing at first by sterilizing their emergency loans.

This is what they did up to Lehman and particularly up to when AIG was bailed out. That meant that they had a lot of treasuries on their books. As their emergency lending increased through various programs or directly, they would sell off treasuries, that is, buy back reserves exactly in an offsetting manner so that their balance sheet in total reserves didn't change.

The only result is a reallocation of liquidity to the firms being aided through the emergency lending, a reduction in liquidity, equal reduction everywhere else. This was sterilized lending.

If you don't think there's a general lack of liquidity if there isn't, monetary policy is just right. I guess that's the right thing to do. Except it wasn't just right, it was too tight. There all kinds of evidence of this including, most obviously from our point of view, the fact that NGDP growth had slowed down and then eventually would even turn negative for a year.

There would be an absolute decline in NGDP, a level decline. This of course by our understanding is bound to cause the adverse developments in the financial market to become a more general crisis.

After Lehman, the Fed's policy didn't change at all. It was determined to keep it. It made grudging downward adjustments to its federal funds target, but they were grudging in the sense that they were both inadequate and frankly meaningless because the equilibrium federal funds target or the natural target of the effective target, any one of those things you want to talk about had already gone down below the...

The equilibrium rates had already fallen below the target which had become therefore rather meaningless.

Nevertheless, after Lehman, the big bailouts of AIG and subsequent ones that involved even bigger loans, the Fed did not have sufficient treasuries left on its book to sterilize. Sterilization became impossible.

Incidentally, there was another program that involved issuing special bonds to the treasury and having the treasury park the money and issuing Fed bonds and parking the money in these special accounts. That was also again designed to reduce the available effective quantity of total reserves.

All of this was contractionary policy because it was felt that the emergency lending by itself would be too expansionary.

Once they were no longer able to sterilize to offset emergency lending, that's when they implemented interest on reserves. That is positive interest rate on reserves including excess reserves. The express idea, this is not me looking at things after the fact and saying, "Here's what they were really up to." This is what they say they were up to.

The express goal was to have another device, an alternative to sterilization that would again make sure that the expansion in the Fed's balance sheet.

Now it's the whole balance sheet that's growing because they aren't sterilizing anymore, that that expansion does not translate into general easing of monetary policy, a general increase in liquidity, and a lowering of the effective or equilibrium federal funds target as it would if banks took the extra liquidity that was being made available to them through these emergency programs and if some of them used it to either make loans in the intra‑bank market or to purchase securities.

When they purchase securities, that's particularly significant because that would generate a multiplier effect, that would mean that the amount of total monetary and credit expansion ultimately based on a given increment of new base money, of new reserves, would be that much greater.

Essentially the Fed has turned to interest on reserves as a way to effectively achieve what sterilization beforehand had achieved.

That means this. The sterilization means you don't really increase the total amount of the monetary base as you rescue firms. Interest on reserves allows the monetary base to expand, but the idea is to keep the multiplier down so that you have the same outcome. This was the express purpose of both.

That's I'm equally critical of both because I believe that these were the instruments by which the Fed pursued what was in fact an overly tight monetary policy that allowed nominal GDP to collapse when what it needed to have was a looser policy. Here I want to be very clear because I've been misunderstood on my writings on this in many ways. I'm only just touching on the one thing.

Beckworth: I'm aware of them.

Selgin: The ultimate problem was overly tight monetary policy and a desire to maintain effectively a federal funds target that was too high and later a target range that was too high. The instruments by which this overly tight policy was implemented included sterilization and then interest on reserves.

In a sense, there's nothing wrong with sterilization. There's nothing wrong with paying interest on reserves. What is wrong is doing those things in the context in which they were done or starting to do them in the way they were done when they amounted to means for keeping monetary policy excessively tight. That's the sense in which I'm criticizing these things.

I have nothing against...Well, I better be careful. In principle, sterilization could be just the right thing to do under the right circumstances, similarly, paying interest on reserves, especially required reserves but perhaps even excess reserves.

Yes, we all know about the efficiency arguments for that. The Friedman rule and so on which were the basis for the original legislation giving the Fed this power, but it had nothing to do with the decision to implement this rule at the time.

There's something else. You're probably going to ask me about this anyway but I've been hammered on 25 basis points, how much difference can that make? I've been hammered by too many very, very knowledgeable people to by any means wish to deny that they have a point.

However, I have to be very careful what I...I want to be very careful I'm not misinterpreted here. I don't believe 25 basis points in equilibrium make much difference, certainly not too risky bank lending and certainly not if capital is restrained.

The reason bank lending is so low is because of regulation and capital requirements and all sorts of things that have nothing to do with those 25 basis points.

The 25 basis points did was to implement an overly tight policy, particularly by restraining the money multiplier that resulted in banks holding excess reserves and accumulating them instead of using them to buy treasuries, which would have given effect to the multiplier.

I think 25 basis points definitely could have this effect even though it was a small amount because what matters isn't the absolute number of basis points you're talking about but whether whatever's being done in the way of paying interest in reserves, whatever change there is in that policy results in having reserves bear a greater yield or return than many of the alternative liquid assets, specifically treasury securities that are out there.

When that happens, and that can happen because of a three basis point change in interest on reserves in principle, that's a real shift in the circumstances that can affect the total equilibrium.

The equilibrium, the real change here is in the nominal total money supply growth that comes out of any Fed expansion being a much lower number and a lower equilibrium price level than would have resulted otherwise...

Beckworth: Well, on that sobering note, we are out of time. But it's been a joy, George, talking to you and always learn something new. Out guest today has been George Selgin of the CATO Institute. George, thank you for being on the show.

Selgin: You're very welcome, David. Thanks for having me.