The centrality of policy to how long recessions last

For economists

Paul Krugman reminds us that one of the most misguided questions in macroeconomics is ‘are business cycles self-correcting’. This is a particular case of another mistake, which is to say that the duration of the business cycle depends on the speed of price adjustment. That answer is seriously incomplete, because it only holds for a particular set of monetary policy rules (plus assumptions about fiscal policy).

It is very easy to see this. Suppose monetary policy is so astute that it knows perfectly all the shocks that hit the economy, and how interest rates influence that economy. In that case absent the Zero Lower Bound the business cycle would disappear, whatever the speed of price adjustment. Or suppose monetary policy followed a credible rule that related real interest rates to the output gap rather than excess inflation. Once again the speed of price adjustment is not central to how long business cycles last. As Nick Rowe points out, if you had a really bad monetary policy recessions could last forever.

A better answer to both questions (self-correction and how long business cycles last) is it all depends on monetary policy. Actually even that answer makes an implicit assumption, which is that there is no fiscal (de)stabilisation. The correct answer to both questions is that it depends first and foremost on policy. The speed of price adjustment only becomes central for particular policy rules.

So why do many economists (including occasionally some macroeconomists) get this wrong? Why are textbooks often quite unclear on this point? It could be just an unfortunate accident. We are so used to teaching about fixed money supply rules (or in my case Taylor rules), that we can take those rules for granted. But there is also a more interesting answer. To some economists with a particular point of view, the idea that getting policy right might be essential to whether the economy self-corrects from shocks is troubling. They prefer to think of a market economy as being ‘naturally’ self-correcting, and to think that government intervention only has a role to play if there is some serious ‘market imperfection’. The market imperfection in the case of business cycles is price rigidity.

Focusing on this logic alone can lead to big mistakes. I have heard a number of times good economists say that in 2015 we can no longer be in a demand deficient recession, because price adjustment cannot be that slow. This mistake happens because they take good policy for granted. It is almost certainly true that no recession should last this long, because fiscal policy can substitute for monetary policy at the Zero Lower Bound. But with sub-optimal policy the length of recessions has much more to do with that bad policy than it has to do with the speed of price adjustment.

Just how misleading a focus on the speed of price adjustment can be becomes evident at the Zero Lower Bound. With nominal interest rates stuck at zero, rapid price adjustment will make the recession worse, not better. Price rigidity may be a condition for the existence of business cycles, but it can have very little to do with their duration.        

More asymmetries: Is Keynesian economics left wing?

In the textbooks it is suggested that Keynesian economics is what happens when ‘prices are sticky’. Sticky prices sound like prices failing to equate supply and demand, which in turn sounds like markets not working. Hence whether you believe in Keynesian theory depends on whether you think markets work, so it obviously maps to a left/right political perspective.

Reality is rather different. Suppose we start from a position where firms are selling all they wish. Aggregate demand equals aggregate supply. If then aggregate demand for goods falls, perhaps because consumers or firms are trying to rebuild their balance sheets after a financial crisis, producers of these goods will start to reduce output, and lay off workers. The idea that they would ignore the fall in demand and just carry on producing the same amount is ludicrous. So output appears to be influenced by aggregate demand at least in the short run, which is at the heart of what most economists think of as Keynesian theory.

So where do sticky prices come in? Here we have to go back to the textbooks, and to an imaginary world where the monetary authority fixes the money supply. Firms, in an effort to stimulate demand for their goods, cut prices. Lower prices mean people do not need to hold so much money to buy goods. However if the nominal money supply is fixed, interest rates will fall to encourage people to hold more money. The textbooks encourage us to think of a market for money, with interest rates as the price that equates supply and demand. Lower interest rates provide an incentive to consumers and firms to increase demand, which in turn raises output.

Now suppose that firms carry on cutting prices as long as they are selling less than they would like. The process just described will continue, with interest rates getting lower and aggregate demand rising in response. The process stops when firms stop cutting prices, which means aggregate demand has increased back to its original level. Suppose further that prices adjusted very quickly. This mechanism would work very quickly, so we would only observe aggregate demand being below supply for very short periods. If prices were extremely flexible, we could ignore aggregate demand altogether in thinking about output. Hence aggregate demand matters only if ‘prices are sticky’.

Note that this correction mechanism is quite complex, and some way from the simple microeconomic world of the market for a single good. But we need to move back to the real world again. Monetary authorities do not fix the money supply; they fix short term interest rates. So they are directly in charge of the correction mechanism that is at the heart of this story. If central banks had some way of knowing what aggregate supply was, and also had perfect knowledge of aggregate demand and how interest rates influenced it, they could make sure aggregate demand equalled supply without any need for prices to change at all. Equally, if prices were very flexible but the monetary authority always moved nominal rates in such a way as to fail to stimulate aggregate demand, aggregate demand and therefore output would not return back to equal aggregate supply. Demand would still matter, even with flexible prices.

Once you see things as they are in the real world, rather than as they are portrayed in the textbooks, the importance of aggregate demand (and therefore of Keynesian theory) is all about how good monetary policy is, and not about sticky prices. If monetary policy was perfect, then Keynesian theory would only be used by central banks in order to be perfect, and everyone else could ignore it. Of course for many good reasons monetary policy is not perfect, and so Keynesian theory matters.

We could re-establish the link between Keynesian theory and price flexibility by assuming the monetary authority follows a rule which would make policy perfect if and only if prices moved very fast, but the key point remains. The importance or otherwise of Keynesian theory depends on monetary policy. It is not about market failure. Keynesian economics is not left wing, but it is about how the economy actually works, which is why all monetary policymakers use it.

It is also common sense, which is why I’m often perplexed by those who dispute Keynesian ideas. Now maybe they are confused by the strange world portrayed in textbooks, but even if they think it is all about ‘sticky prices’, the evidence that prices are slow to adjust is overwhelming, so it is hard to dispute Keynesian theory on those grounds. Yet a whole revolution in macroeconomic theory was based around a movement that wanted to overthrow Keynesian ideas, and build models where this correction mechanism I described happened automatically. The people who built these models did not describe them as assuming monetary policy worked perfectly: instead they said it was all about assuming markets worked. As a description this was at best opaque and at worst a deliberate deception.

So why is there this desire to deny the importance of Keynesian theory coming from the political right? Perhaps it is precisely because monetary policy is necessary to ensure aggregate demand is neither excessive nor deficient. Monetary policy is state intervention: by setting a market price, an arm of the state ensures the macroeconomy works. When this particular procedure fails to work, in a liquidity trap for example, state intervention of another kind is required (fiscal policy). While these statements are self-evident to many mainstream economists, to someone of a neoliberal or ordoliberal persuasion they are discomforting. At the macroeconomic level, things only work well because of state intervention. This was so discomforting that New Classical economists attempted to create an alternative theory of business cycles where booms and recessions were nothing to be concerned about, but just the optimal response of agents to exogenous shocks.

So my argument is that Keynesian theory is not left wing, because it is not about market failure - it is just about how the macroeconomy works. On the other hand anti-Keynesian views are often politically motivated, because the pivotal role the state plays in managing the macroeconomy does not fit the ideology. Is this asymmetry odd? I do not think so - just think about the debate over climate change. Now of course it is true that there are a small minority of scientists who do not believe in manmade climate change and who are not politically motivated to do so, and I’m sure the same is true for Keynesian theory. But to claim that the majority of anti-Keynesian views were innocent of ideological preference would be like – well like trying to pretend that monetary policy has no role in stabilising the business cycle.

There are of course many differences between climate change denial and anti-Keynesian positions. One is the extent to which the antagonism has infiltrated the subject itself. Another is the extent to which the mainstream wants to deny this influence. I do wonder if the unreal view of monetary policy that remains in the textbooks does so in part so as to not offend a particular ideological position. I do know that macroeconomics is often taught as if this ideological influence was non-existent, or at least not important to the development of the discipline. I think doing good social science involves recognising ideological influence, rather than pretending it does not exist.

  

Sticky prices: how we confuse students, and sometimes ourselves

For teachers and students of macroeconomics.

I’m about to teach a small number of first year undergraduate students Keynesian macroeconomics, and my aim will be not to tell them that this is the macroeconomics of sticky prices. Yet I realise I’ve already gone wrong. In week one I talked about time periods in macro, and how the ‘short run’ was the length of time ‘it takes prices to fully adjust’. I must have been saying this for years. But it is at best highly misleading.

In both the New Keynesian closed economy model, and the IS/LM model, the short run is the length of time it takes the central bank to stabilise inflation (output goes to its natural rate), or less precisely to achieve full employment. For students we could equally say it is the period it takes monetary policy to achieve the real rate of interest implied by the RBC, or Classical, model.  Calling this the time period it takes prices to fully adjust only makes sense when monetary policy involves some kind of nominal anchor, like a fixed money target in IS/LM. It makes no sense when monetary policy involves a central bank trying to choose the best nominal interest rate. The impact of an unexpected but subsequently known preference/demand shock, for example, would be very short lived when such a central bank knew what it was doing. (See this excellent post from Nick Rowe.)  

The big danger in equating Keynesian economics with sticky prices is that students forget about the crucial role monetary policy is playing. Too many think that after an increase in aggregate demand, if contracts and menu costs were absent, higher prices would in themselves choke off the increase in aggregate demand.  As they have just learnt micro, it is a natural mistake to make. They then get very confused when price flexibility does (at best) nothing at the zero lower bound.

Yet the linking of the short run with sticky prices is ubiquitous. In the edition of Mankiw I have to hand it says

“In the long run, prices are flexible and can respond to changes in supply or demand. In the short run, many prices are sticky at some predetermined level. Because prices behave differently in the short run than the long run, economic policies have different effects over different time horizons.”

This kind of statement makes sense in a fixed money supply world, but it makes much less sense in the real world. (Mankiw uses the term ‘long run’ where others would use ‘medium run’, but let us not worry about that.) Compare it with this alternative statement:

“In the long run, monetary policy adjusts to achieve steady inflation, which means output goes to its ‘natural’ or Classical level. In the short run, monetary policy fails to achieve this, so we need to look at movements in aggregate demand to explain output.”

This works for any sensible monetary policy.

In my second year lectures, I ask my students to think about a monetary policy that involved moving real interest rates in response to the output gap, but not to excess inflation.  If that policy stabilised a closed economy, then what impact would the speed of price adjustment have on anything except inflation? Inflation aside, a world where price adjustment was quick would look much like a world where prices were much stickier. The ‘short run’ would have the same length, irrespective of how quickly prices adjusted.

All this is about how Keynesian economics is taught, rather than about how it is done. Yet how it is taught can also influence how it is eventually understood. One of the problems some people have with understanding that we are still in a situation of deficient demand is that it is five years after the recession ‘and surely prices should have adjusted by now’. There is also a rather more profound point. Many anti-Keynesians use this misunderstanding about price adjustment to dismiss Keynesian economics. When they say ‘I ignore Keynesian economics, because I think prices adjust rapidly’ they are really saying ‘I ignore Keynesian economics because I think monetary policy is very successful’. And in the real world, monetary policy can only be very successful by understanding Keynesian economics! 

Two Anti-Keynesian myths

This is mainly of interest to economists, but given the importance of these issues, I have tried to write it in an accessible manner.

Stephen Williamson, in commenting on this post, remarks acidly that

“Part of what defines a Keynesian (new or old), is that a Keynesian thinks that his or her views are "mainstream," and that the rest of macroeconomic thought is defined relative to what Keynesians think - Keynesians reside at the center of the universe, and everything else revolves around them.”

In that post I was careful to distinguish between academic research and policy. Of course macroeconomists research many things, and only a minority are using New Keynesian models, and probably even some of those do not really need the New Keynesian bit. That is the great thing about abstraction. Working with what can be called ‘flex price’ models does not imply that you think price rigidity is unimportant, but instead that it can often be ignored if you want to focus on other processes. So in terms of research I talked about the significant divide being between mainstream and heterodox. In terms of research, mainstream macroeconomists talk the same language.

I used the term anti-Keynesian in the context of macroeconomic policy, and in this context I was not talking about academics, but the set of economists involved with macro policy. They could be academics, but they could be working for central banks, a finance ministry, or an international organisation like the IMF or OECD. For this group, I think we have good reason to believe that the large majority are not anti-Keynesian.

Just look, for example, at any central bank publication discussing recent movements in output. This will typically focus on movements in components of aggregate demand: consumption, investment etc. The reason is a belief that output in the short run is demand determined. That, for me, is the defining feature of Keynesian analysis. If you look at the core models used in central banks (which, unlike models used by academics, need to be ‘horses for all courses’) the same will be true.

Now Nick Rowe and David Glasner suggest that this view is not uniquely Keynesian - I could equally call it monetarist, for example. But what then are the criteria you will use to restrict the Keynesian set today? Different views about unconventional monetary policy? Different views about the efficacy of fiscal policy at the zero lower bound? This seems way too narrow. People have views about the relative merits of fiscal policy for all kinds of reasons, which may be very context specific, so this does not look like a good method of defining a label for economists today.  

Yet not everyone is a Keynesian using my deliberately broad definition. The only logical way to make sense of statements like those discussed here, for example, is to imagine we are in a world where output is determined from the supply side, so if one particular component of demand, like government spending on goods and services, goes down its impact on output will be offset by some means. That is an anti-Keynesian view. My first anti-Keynesian myth is that among economists involved with policy this is not a minority view.

The second myth is that all you need to justify this anti-Keynesian view is to observe that wages and prices move in response to booms and recessions. For example Roger Farmer points to negative inflation following the Great Depression. Language can be confusing here. As an analogy, suppose someone has been ill, and you ask them whether they are now better. Do you mean better than they were (but still ill), or completely recovered? For us to take an anti-Keynesian view, we do not just require prices to move, we require them to move by just the amount needed so that we can ignore demand. So, for example, in an open economy under fixed exchange rates, for a devaluation to have no demand impact requires prices to immediately rise by the same amount. Prices will begin to rise for sure, but ‘flexible prices’ means more than that.

If we take a simple closed economy, then ‘flexible prices’ is short for the real interest rate (nominal rates less expected inflation) always being at its ‘natural level’, which is the level that ensures demand matches supply. This immediately tells you that we are not just talking about price flexibility, but also monetary policy. Imagine in this economy there is a negative demand shock, caused by a fall in government spending. In this economy, the immediate impact is that firms will reduce output as well as prices. To offset this, the natural interest rate will fall to increase private consumption. Will the actual real interest rate do the same? This could happen without prices changing if the central bank cut nominal rates by exactly the required amount. It could happen without the central bank doing anything to nominal rates if expected inflation rose by exactly the required amount. Or it could be some combination of the two.

One justification for assuming that the real interest rate is always at its natural level is that monetary policy is super efficient, moving nominal rates to always offset the impact of demand shocks. For some reason this is not an argument anti-Keynesians usually make. The alternative justification is that, conditional on whatever monetary policy does, prices move by just the amount required to give you the expected inflation rate necessary to generate the natural real interest rate, and therefore offset the demand shock.

This becomes clear when nominal interest rates are stuck at the zero lower bound. In that case, the natural real interest rate is large and negative, but monetary policy cannot get there because nominal interest rates cannot be negative. For flexible prices to get you to that real rate you would need expected inflation to be significantly positive. (We now believe there is no independent Pigou effect (or real balance effect) that will save the day.) As Roger shows, actual inflation from 1929 to 1933 was persistently negative. One must presume that inflation expectations were also negative. So clearly although prices were moving, they were not moving in the way required for the anti-Keynesian view to hold. Far from casting doubt on the Keynesian story, falling prices during the Great Depression show how unrealistic the anti-Keynesian view is.

The Zero Lower Bound and Price Flexibility

I’ve only written one of these Socratic/tutorial dialogue type posts before, mainly because I cannot make them as amusing as Tim Harford or Brad DeLong. This one was inspired by these questions.

Q: I get why interest rates cannot go below zero. But why is that such a big deal?

A: Because it means that monetary policy cannot do its job.

Q: But I thought monetary policy was about keeping inflation low.

A: In part. But we also rely on monetary policy to ensure that aggregate demand matches the output the economy as a whole wants to produce.

Q: Isn’t that what the price mechanism is for – matching supply and demand?

A: This is a good example of where thinking about a single market is not a good way of thinking about the macroeconomy. While in the long run you would expect aggregate supply and demand to match, in the short run they need not. People can decide to save more, and investment need not rise to compensate, so demand can fall below supply, leading to unemployment rising.

Q: Yes, I remember our discussion about savings and investment. But unemployment can always be cured by falling wages, surely.

A: Not if prices are falling at the same time. To say unemployment is too high because real wages are too high in this situation just misses the point.

Q: But if both wages and prices start falling, will that not lead to a recovery in demand? What stops that happening?

A: The conventional answer is that prices are sticky, so this process happens only slowly. That certainly seems to be true, but it’s an interesting question whether adjustment would occur even with flexible prices.

Q: Ah yes, sticky prices – that is all this Keynesian stuff that is so controversial. So if you believe prices are sticky, does that mean booms and recessions caused by fluctuations in demand are inevitable?

A: No, as I said earlier, that is monetary policy’s other job. We have very good reasons to think that aggregate demand depends negatively on the real interest rate. So there should always be some real interest rate that brings demand equal to aggregate supply.

Q: And the central bank tries to guess what that interest rate is each month?

A: Basically yes, although they can only determine nominal interest rates, so they also need to estimate what expected inflation will be. I hope you remember the definition of real interest rates.

Q: Of course. And now I see the problem. If the required level of real interest rates is significantly negative, and expected inflation is low, even nominal interest rates at zero may not be enough to get demand high enough.

A: Well done.

Q: But now I’m puzzled. If prices are flexible rather than sticky, surely that would make things worse, not better. In a recession it means negative inflation, and therefore higher real interest rates – it goes in the wrong direction!

A: Careful. I thought you said you remembered the definition of real interest rates. It’s expected inflation that matters, not actual inflation.

Q: Sure, but if inflation is falling, surely expected inflation will fall too.

A: Not if the central bank had a target for the price level, or something else related to it, and people believed the bank could and would achieve that target. Then for every fall in actual prices, expectations about inflation in the future would rise.

Q: Like what goes down, must come up again. That reminds me of the other day ..

A: I’ll stop you right there. You should not even attempt to make these dialogs as amusing as those other bloggers. Stick to the economics.

Q: If you insist. So you are saying flexible prices would work after all. If the price level fell enough, expectations about inflation would rise enough to get real interest rates low enough, even if nominal rates were at zero.

A: Yes, but remember what I said about people needing to believe that the central bank could and would do that. And if people are not fooled, it would require future actual inflation to rise in line with expectations.

Q: Which would conflict with the other goal of the central bank, to keep a lid on inflation.

A: Indeed. Most central banks now have inflation targets, rather than price level targets. So if they were doing their job, they would stop inflation rising enough to get real interest rates low enough.

Q: So with inflation targets, even price flexibility might not be enough to ensure aggregate demand was equal to supply if demand fell by a large amount. So why is this Keynesian stuff controversial – it seems to be important whether prices are sticky or not.

A: I would agree. In the past, before you were born, economists talked about the real balance or Pigou effect saving the day, but that is hardly mentioned nowadays. I’m not entirely sure why.

Q: But my textbook says Keynesian economics is all about the economics of sticky prices.

A: That is the same textbook that says the central bank fixes the money supply.

Q: Yes. You never did explain to me why the textbook says that even though it’s not true.

A: I’m not sure I can. But it helps explain the emphasis on price rigidity when discussing Keynesian economics. Money targets are a variation on a price level target, so in that case price flexibility could be enough as we have just seen. For this reason, and perhaps for other reasons as well, textbooks in my view place too much emphasis on price rigidity as a pre-condition for Keynesian analysis, and too little on good monetary policy as being essential in controlling short run aggregate demand.

Q: You do not seem to like my textbook much. But anyway, whether prices are sticky or not, being at the zero lower bound pretty well proves that there is not enough demand in the economy at the moment, and so we need to focus on ways to increase demand. That cannot be controversial.

A: Oh how I wish you were right. 

Keynesian Economics, Price Rigidity and Demand Denial

Something prompted from revising my second year undergrad lecture notes, and so mainly for economists.

In mainstream macro today, Keynesian economics is synonymous with the macroeconomics of price rigidity. Most of the time I have no problem with that. All the evidence suggests there is significant inertia in aggregate prices, and it is very difficult to tell realistic stories about how inflation moves without taking this into account. Price inertia and imperfect competition are probably essential in understanding why output tends to follow aggregate demand in the short term.

My problem with identifying Keynesian economics with the macroeconomics of price rigidity is that it allows those who would like to ignore Keynesian theory with too easy an opt out. They can argue that, despite appearances to the contrary, prices are in fact pretty flexible. They then conclude that Keynesian economics is irrelevant. Unfortunately far too many academic macroeconomists appear to implicitly or explicitly take this view.

                Is it logically the case that if prices are flexible Keynesian economics can always be ignored? The answer is simply no. Price flexibility alone does not ensure demand always moves quickly towards supply: it is the combination of price flexibility and monetary policy that does this. And when something goes wrong with monetary policy, price flexibility alone may not work.

We are living through exactly such a situation. It is not just the zero lower bound for nominal interest rates that is important here. It is also the fact that monetary policy has an inflation target rather than a price level target. After a large negative demand shock, demand can only be restored in the short term (for a given fiscal stance) by a large reduction in real interest rates. Real interest rates are nominal rates minus expected inflation. The zero lower bound stops nominal rates falling enough, and inflation targeting stops inflation expectations rising enough. No amount of price flexibility can change this. (For a more detailed discussion see here.)

                Students often think that price flexibility must imply that output is supply determined, because if any workers were unemployed, nominal wages would continue falling until they were employed. But just imagine an economy made up of monopolistic competitors where production was linear in labour. In that economy firms would ‘determine’ a constant real wage through their mark-up, and no amount of nominal wage cutting would reduce real wages or increase employment. 

Demand denial is the belief that we can always ignore aggregate demand when analysing short term movements in output and employment. It sometimes seems to be based on a view that price flexibility alone always ensures demand is sufficient for supply. It does not.