Why Central Bankers are Like Sheepdogs

Central bankers and economic policy makers often talk about ‘anchoring expectations’.  What do they mean by that?  The narrative of anchored expectations is difficult to square with modern New Keynesian (NK) theory. The problem with NK theory is that expectations are supposed to be forward looking and rational. And although nobody would deny that central bank announcements do move markets, it is difficult to believe that expectations are forward looking and rational in quite the way that the modern theory of rational expectations requires them to be. Expectations, like sheep, have a life of their own. And like sheep dogs herding sheep, central banks must herd expectations.

In modern NK Dynamic Stochastic Dynamic General Equilibrium Models, (DSGE), expectations are anchored by the so-called rational expectations (RE) assumption. According to that assumption, people today make guesses about what people tomorrow will do. And these people make still further guesses about the actions of other people in the even more distant future. These guesses, so the NK story goes, must be right on average. That of course is poppycock, akin to medieval religious debates about how many angels can dance on the head of a pin.

The problem with the RE assumption, is that there is no obvious right way to guess what will happen in the future. Not only is the future unknown and unknowable; but even in the equilibrium models that NK economists favor, there are many possible rational outcomes each of which would lead to a different expectation today. And each of these expectations would turn out to be rational if only our descendants would  behave in the way we conjecture that they might. As I explained, beginning with my 1993 graduate textbook The Macroeconomics of Self-Fulfilling Prophecies, the existence of multiple equilibria in RE models need not be a 'problem'; it is an opportunity.

NK theorists have tried to tackle the ‘problem’ of multiple equilibria by ruling out some of the possible outcomes in their models by appealing to purely logical arguments. Those attempts to refine equilibria using logic can lead to some pretty absurd conclusions. For example, NK models predict that a fiscal expansion will have a bigger effect on current employment if  it occurs in ten years time, after interest rate policy has normalized, than if it occurs today. Better still, postpone it until the next moon landing. Don’t laugh too hard: it is arguments like these that persuaded otherwise sensible central bankers to use so called ‘forward guidance’.

According to the NK DSGE model, we  all make sensible guesses about what we think the Treasury and the Central Bank will do in the distant future and we alter our existing plans accordingly. In a particularly bizarre manifestation of this theory called the Fiscal Theory of the Price Level (FTPL), we are all supposed to act in a way that causes the current value of government debt to magically equal the value of planned future deficits. For example, if the Obama administration were to announce a new unfunded entitlement, the price level today would increase immediately to reduce the real value of existing Treasury debt. I am not making this up: the FTPL was the content of the warmly accepted lunchtime speech by Nobel Laureate Christopher Sims at the 2016 Jackson Hole conference.

It is not enough, of course, to laugh at the absurdities of existing theories. It takes a model to beat a model. And I have an alternative that  makes a lot more sense. Expectations are not irrational: they are fundamental. What we believe influences what happens. One way to implement this idea is to assume that beliefs about future nominal income growth are equal to current nominal income growth (see here). When that assumption is combined with the Keynesian insight that there are many possible equilibrium unemployment rates, we arrive at a more plausible description of why monetary policy matters than the unlikely fairy tales that are told to justify the broken NK paradigm.

Expectations are like sheep. They must be herded by central bank actions that alter current economic conditions. And like sheep, expectations are stubborn and persistent. Weaving that idea into a consistent macroeconomic narrative leads to a very different prescription for how we should conduct future policy.  The NK model which guides existing macroeconomic policy cannot simply be tweaked as some have claimed. It must be torn up and reconstructed from scratch. That is what I offer in my book Prosperity for All that is now available from a bookseller near you. Also now available on Amazon kindle.

Message to the Fed: We’re not in Kansas Anymore

In a recent post, I argued that the Fed should raise the interest rate to return inflation to positive territory. This post is about the possible cost of that action on the nascent recovery and how that cost can be mitigated, or avoided. My views are based on the theory expounded in my academic papers and in my book, Prosperity for All, that will be available from your favorite bookseller in a couple of weeks.

I will argue that a rate rise may have negative consequences on output and employment. But those negative consequences can be mitigated in two ways. First, by raising the interest rate paid on excess reserves (IOR) at the same time as operating in the repo market to raise the Federal Funds Rate (FFR). And second, by operating in the asset markets to offset a potential market crash by standing ready to buy and sell an exchange traded fund in the stock market.

There is a lasting and stable connection in data between changes in the interest rate and changes in the unemployment rate. Past data suggest, that if the Fed were to raise the interest rate at its next meeting, unemployment would increase and output growth would slow. It is fear of that outcome that causes central bank doves to be reluctant to raise the interest rate.

But although an interest rate increase has preceded a slowdown by approximately three months in past data, there is a connection at longer horizons between inflation and the T-bill rate. That connection, sometimes called the Fisher relationship after the American economist Irving Fisher, arises from the fact that, risk-adjusted, T-bills and equities should pay the same rate of return.

The one-year real return on a T-bill is the difference between the interest rate and the expected one-year inflation rate. The one-year real return on holding the S&P 500 is the gain you can expect to make from buying the market today and selling it one year later. Economic theory suggests that the gap between those two expected returns arises from the fact that equities are riskier than T-bills, and importantly, the gap cannot be too big.

Therein lies the policy maker’s conundrum. To hit an inflation target of 2%, the T-bill rate must be 2% higher than the underlying risk adjusted real rate: policy makers call this rate r*. There is some evidence that r* is currently very low currently, possibly zero or even negative.  But if the Fed were to raise the policy rate to 2% at the next meeting, they are terrified that they might trigger a recession.  Let’s examine that argument.

The fact that a rate rise caused a slowdown in past data does not mean that a rate rise will cause a slowdown in future data. This time really is different. It is different because in 2008 the Fed expanded its policy options. Before 2008 the interest rate set by the Fed was the Federal Funds Rate (FFR). That is the overnight rate at which commercial banks can borrow or lend to each other. Before 2008, there was a large and active Fed funds market used by commercial banks to meet reserve requirements.

Commercial banks are required to hold roughly 10% of their balance sheets in the form of reserves. In the past, because reserves did not pay interest, banks kept them to a minimum. Excess reserves for much of the post-war period were essentially zero. Firms and households hold cash because they need liquid assets to facilitate trade. But cash is costly to hold because a firm must forgo investment opportunities.  In the parlance of economic theory, we say that the FFR is the opportunity cost of holding money.

After 2008, the Fed began to pay interest on reserves (IOR) at a rate slightly higher than the FFR. To a first approximation, in the new environment the FFR and the IOR are equal. That fact has a profound effect on the economic theory of the transmission mechanism of interest rate changes on the real economy. In our brave new world, the opportunity cost of holding money is zero.

Let me make a strong assumption. That $100 bills and reserves held by commercial banks at the Fed are perfect substitutes. That is not quite correct, but it is not a bad approximation. To the extent it is correct to assume that cash and reserves are perfect substitutes, an increase in the IOR and a simultaneous increase in the FFR will not change the opportunity cost of holding money. That is an important observation because one of the principal reasons that past interest rate increases caused recessions is that past interest rate increases were also increases in the opportunity cost of holding money.

In a world where the opportunity cost of holding money is zero, households and firms will continue to increase their liquidity to the point where they are satiated. Like a glutton enjoying a feast, there will come a point when holding additional cash in the form of reserves has no additional benefit. And that is exactly where we are now. U.S. corporations are sitting on huge piles of cash because the cost of holding that cash is zero. And those cash piles are reflected in reserve balance of the commercial banks that are currently holding more than $3 trillion in excess reserves.

If the Fed were to raise the FFR and leave the IOR unchanged, there would be an immediate flight from cash. Commercial banks would expand their balance sheets by buying up T-bills. To maintain its control over the FFR, the Fed would be forced to sell off assets to meet this demand and there would be an immediate and massive contraction in the money supply. Cash would once again become scarce and that scarcity would have repercussions in the goods markets as households and firms cut back on expenditure plans to maintain liquidity. The outcome would be a Fed induced recession.

But if the Fed were to raise the IOR and the FFR at the same time, the negative effects of an interest rate rise on economic activity, operating through a possible tightening of liquidity, could be mitigated. But although a simultaneous increase in IOR and FFR would be less contractionary than an increase in the FFR alone, it is likely that a rate increase would still be contractionary. There is a second channel by which an interest rate rise might trigger a recession. That second channel works through wealth effects that could potentially follow a stock market crash.

Let me be clear. I am by no means certain that an interest rate increase would trigger a market decline. In my view, the markets have a life of their own that is governed by the animal spirits of market traders. But it is wise to plan for all contingencies. And because I am concerned that a wealth effect, operating through animal spirits, may still be operative, I have recommended that the Fed be given the power to intervene in the asset markets to prevent the market crash that might otherwise follow a rate rise.  I do not want the Federal government to own private companies, and for that reason, I continue to advocate that the Fed should buy and sell non-voting shares in Exchange Traded Funds.

Let me close by repeating a mantra: two targets need two instruments. To hit a 2% inflation target and maintain ‘maximum employment’, the Fed must not only have authority to set the interest rate. It must also be given the authority to intervene in markets in some other way. An immediate recession need not follow a rate rise, providing the Fed acts to mitigate the effects of a rate rise on the real economy. My preferred central bank action is direct intervention in the asset markets, as opposed to a more traditional fiscal policy enacted by the Treasury. Financial policy is more direct and faster acting and my research on the connection between the stock market and the unemployment rate suggests that it will be more effective. Of one thing I am certain. Inaction or worse, moving the interest rate into negative territory, will not lead to higher inflation. It will lead to stagnation and another lost decade.

Reinventing IS-LM: The IS-LM-NAC Model and How to Use It

I have been critical of the IS-LM model in earlier posts. My paper with Konstantin Platonov fixes some of the more salient problems of IS-LM by reintroducing two key ideas from Keynes. 1. The confidence fairy is real. 2. If confidence remains depressed, high unemployment can exist forever.  My new Vox piece, coauthored with Konstantin Platonov, presents the key findings of the paper in simple language. Here are some excerpts from my VOX piece...

Larry Summers has argued that market economies may get stuck in permanently inefficient equilibria. He calls this 'secular stagnation' (Summers 2014). In this equilibrium, unemployment may be permanently ‘too high’ and output may remain permanently below potential, because private investors are pessimistic about the prospects for future growth. Our most recent research attempts to explain why secular stagnation occurs and how economic policy may be used to escape it (Farmer and Platonov 2016)

In the wake of the Great Recession, macroeconomic orthodoxy is under attack. Paul Krugman (2011) has called for a return to the IS-LM model, an approach that was developed by Sir John Hicks (1937). We are sympathetic to that call but we believe that the IS-LM model needs to be redesigned. We suggest a different way of thinking about the effect of monetary policy that we call the 'IS-LM-NAC' model. It is part of a broader research agenda ( Farmer 2010, 2012, 2014, 2016a, 2018) that studies models in which beliefs independently influence outcomes.... continue reading

Here is a link to our paper, Animal Spirits in a Monetary Economy.

The Economics of the Grim Reaper

I am on a roll with wonkish posts. So here is another one.  I am going to explain the difference between two kinds of macroeconomic models, representative agent (RA) and overlapping generations (OG) models. And I will explain why OG models provide an explanation of why central banks should intervene in asset markets that is different, and in my view stronger, than any argument that might be leveled using the RA model as an organizing principle. Because the OG model assumes that people die and new people are born, I have called this post: The Economics of the Grim Reaper.

Macroeconomists have two "workhorse" models. In one, they assume that the world works "as if" there were a single representative family making all decisions. The head of that family has superhuman perceptions of all future possible outcomes and he/she makes plans accordingly. This is called the RA or "representative agent" model. In the second workhorse model, we assume that population consists of a sequence of overlapping generations of selfish individuals. This is called the "overlapping generations", or OG model. You can find a simple (as simple as any graduate text can get) introduction to these two kinds of models in my graduate textbook The Macroeconomics of Self-Fulfilling Prophecies.

The RA model may sound unrealistic, but it is more reasonable than it at first seems. For instance, many of the strong results that are known to hold in the RA model, also hold if there is a finite number of infinitely lived families all of whom trade with each other in the financial markets. The really strong assumption, which is added on top, is that people know the probabilities of all future events and they trade with each other, contingent on these events.

If we want to understand why government should intervene in financial markets, and I do, we must explain why government can do something that private markets cannot. Perhaps the reason is that the public really does not have superhuman perception of future events. Perhaps, that is, the rational expectations assumption is wrong. But that, in itself, is not a reason for governments to intervene in financial markets.

Hold your nose for a moment, and suppose that people can, and do, trade securities contingent on all future events. In order to argue that government should intervene in the financial markets, you must be able to show that treasury or central bank mandarins have superior knowledge. That seems a stretch. After all, we do not insist on shutting down gambling on horse races. But people only bet on horse races because they have different opinions as to the probabilities as to which horse will win. Rational expectations, in this context, is obviously false. If we assume the same is true of the financial markets, that people disagree about future probabilities, that in itself is not an argument for intervention. Government must be shown to have an advantage that allows them to intervene in a way that improves social welfare. And in my view, individual human beings are still the best judges of what is, and what is not, in their own best interests.

Absence of rational expectations, I will argue, is not a reason for government to intervene in financial markets. Death, on the other hand, is such an argument. To make that case, I need to say more about the OG model.

It is well known that the OG model behaves very differently from the RA model, For example, Paul Samuelson, in a seminal article that introduced this model to the English speaking world, showed that the overlapping generations model sometimes allows a role for governments to improve the arrangements that are made by private agents. That role relies on what economists call 'dynamic inefficiency' and it occurs when interest rates, in the model, are 'too low'.

If interest rates are too low, government can take resources from future generations, give them to current generations, and make everyone happier. Although the real world is looking more and more as if that might be a good assumption, it is not the reason I will emphasize in this post, for treasury and central bank intervention in financial markets. There is a second reason why death matters that was first pointed to by David Cass and Karl Shell in article that appeared in 1983 called "Do Sunspots Matter?"

The second reason why death matters is quite distinct from the property of dynamic inefficiency and it occurs, even when interest rates are high and all market exchange is dynamically efficient. This second reason occurs because, in OG models, economic activity can fluctuate as a consequence of self-fulfilling beliefs. Financial cycles arise in these models because people engage in self-fulfilling bouts of optimism and pessimism that cause fluctuations in output and employment. And the government DOES have an advantage over private agents (wonkish example here, and here). It can make trades on behalf of the unborn that can smooth out these cycles.

In my forthcoming book, Prosperity for All, I make the case that this second reason for inefficiency is important and is the main reason that financial cycles are so destructive for human welfare. By implementing the financial reform that I propose in my book, we can ensure that next time really will be different. 

Beliefs and the Monetary Transmission Mechanism

View fullsize

In my last post, I urged the Fed to raise the interest rate at its next meeting. In response to that post, David Andolfatto asks, 

Suppose a CB was to raise its policy rate permanently in an attempt to raise the inflation rate (without the corresponding policies you favor [in your post]). You claim that this would ultimately raise the rate of inflation (possibly after a recession). Can you describe the economic mechanism that would work to produce this result? (I mean, an explanation that extends beyond simply stating that the Fisher equation must hold.)

Here is my necessary, (but Wonkish) response. 

If we write a 'Fisher like equation' of the form

EQ 1) i - rho = E[Pi_t+1] + E[y_t+1] - y_t

Here, i is the money interest rate, y_t is GDP, Pi_t is inflation between dates t-1 and t and rho is the real rate of interest. E[ ] is the subjective expectation formed at date t. An equation like this would arise in a representative agent model with logarithmic preferences.

Now we can agree, I think, that if the Fed sets i equal to some constant rate, if we impose rational expectations, and if we write down a process that determines {y_t}; then an increase in i will lead to an increase in inflation. You want to know the adjustment mechanism from one steady state rational expectations equilibrium to another.

Let's assume that the Fed operates an interest rate peg, and that at some date, it announces an increase in that peg. At date t, one of two things can happen. Either, the increase in i can lead to an increase in E[Pi_t+1] + E[y_t+1], OR it can lead to a fall in y_t. If expectations are slow to adjust, the increase in the peg will lead to a fall in current GDP.

In my view, the equations needed to close this model are equations that determine E[Pi_t+1] and E[y_t+1], as functions of Pi_t, y_t i_t, and, other observables. Those equations are fundamentals that behave like preferences. It is the fact that expectations are 'sticky' that leads to nominal shocks, such as an increase in a purely nominal variable, like i, to have real effects.

In a slightly more complicated model, where the coefficients on output growth and inflation in equation 1, are different, the following expectations rule works well at explaining data

EQ 2) E[Pi_t_1] + E[y_t+1] - y_t = Pi_t + y_t - y_t-1

in words, expected nominal income growth is a martingale. By closing a three-equation model with that specification of the belief function, and by writing down a slightly more sophisticated version of a Taylor rule (not just a simple interest rate peg) I showed here that a model closed with a belief function, outperforms the NK Phillips curve model.

The idea of making the belief function 'fundamental' is consistent with equilibrium and rational expectations because of my assumption that the labor market operates as what I call, in my forthcoming book, Prosperity for All,  a 'Keynesian search market'. Employment and GDP growth are not uniquely pinned down in the steady state: they depend on initial conditions and the entire history of shocks.