Addendum: Here is a review of the book.
Sunday, June 10, 2012
On my to-read pile
Addendum: Here is a review of the book.
Tuesday, June 5, 2012
Barro on the Slow Recovery
Robert says incentives are the key:
Consider the expansion of social-safety-net programs, including food stamps, unemployment insurance, Medicaid (prospectively) and housing and mortgage programs. In a study published last month by the National Bureau of Economic Research, University of Chicago economist Casey Mulligan observed that, because these programs were means-tested (falling or ending as income rises), expanding them raised the effective marginal tax rate on labor income.
Specifically, Mr. Mulligan estimates that the effective marginal tax rate for low-income households went from around 40% in 2007, before the recession started, to about 48% in 2009, at the start of the recovery. Thus, while these programs may be attractive from the standpoint of assisting poor families, they dilute incentives to work.
To achieve a real recovery, government policy should focus on individual incentives to work, produce and invest. Central here are tax rates and regulations, including especially clarity about future policies. In a successful policy package, the government would get its fiscal house in order and make meaningful long-term reforms to entitlement programs and the tax structure.
Summers on Quantitative Easing
Larry is skeptical:
There is also an oddity in this renewed emphasis on quantitative easing. The essential aim of such policies is to shorten the debt held by the public or issued by the consolidated public sector comprising both the government and central bank. Any rational chief financial officer in the private sector would see this as a moment to extend debt maturities and lock in low rates – exactly the opposite of what central banks are doing. In the U.S. Treasury, for example, discussions of debt-management policy have had exactly this emphasis. But the Treasury does not alone control the maturity of debt when the central bank is active in all debt markets.
Monday, June 4, 2012
More on Unconventional Open Market Operations
Miles Kimball may have been the first New Monetarist (actually, if you read the paper, he may have been the first New-Keynesian; he's basically outlining the basic NK model in 1995). Fortunately for us, Miles wants to blog, which is guaranteed to increase the average quality of discourse in the medium. From a dismal low, you might say, but progress is progress.
Miles has taken the trouble to write at length in reply to my comments on one of his posts, so I'm encouraged to carry on the discussion.
For some background on liquidity traps, quantitative easing (QE) and what it can and cannot do, you can browse my archive, or read these particular pieces:
1. QE irrelevance.
2. An example.
3. Liquidity traps.
To understand QE, we need to know the difference between a channel system and a floor system. A good example of a channel system is the central banking framework within which the Bank of Canada works. The Bank sets a deposit rate (the interest rate on reserve accounts) and a higher rate at which it lends to financial institutions, and targets an overnight rate that lies between those two rates (i.e. the target rate lies in the "channel"). In a channel system, as long as the overnight rate lies within the channel, open market operations matter - purchases or sales of assets by the central bank will move the overnight rate.
Prior to the financial crisis, the Fed worked within what was essentially a channel system. The interest rate on reserves (IROR) was zero, excess reserves were essentially zero overnight, and the overnight fed funds rate fell between zero and the discount rate (there are some complications involving what the "fed funds rate" is, and how lending at the discount window takes place, but ignore that for now). Pre-financial crisis, if the Fed bought or sold assets (T-bills, long Treasuries, whatever), that would move the fed funds rate.
A floor system is different. Under such a system, the central bank sets an interest rate on reserves and a central bank lending rate, and plans to have a positive supply of reserves in the system overnight. As a result, the overnight rate must be equal to the IROR, by arbitrage. An open market operation in short-term government debt in a floor system will have no effect, at the margin, as the central bank is simply swapping one interest-bearing short-term asset for another. The instrument of monetary policy in a floor system is the IROR, which determines short-term nominal interest rates.
Currently, the Fed operates under a floor system. The supply of excess reserves is enormous, and the IROR determines short-term interest rates. There are some weird features of the system, such as the fact that the GSEs receive no interest on their reserve accounts, and there is some lack of arbitrage which results in a fed funds rate less than the IROR, but I think those weird features are irrelevant to how monetary policy works.
Under a floor system, we are effectively in a perpetual liquidity trap. Conventional open market operations in short-term government debt do not matter, whether the IROR is 5%, 0.25%, 10%, or zero. But, not to worry, the central bank can always change the IROR except, of course, when it hits the zero lower bound (neglecting the possibility of taxation of reserve balances, which is another issue altogether).
My irrelevance argument goes one step further. First, we have to understand why open market operations move the overnight rate in a channel system. An open market purchase of T-bills under a channel system essentially involves the transformation by the central bank of T-bills into currency. Remember that, in a channel system, overnight reserve balances are zero. The increase in outside money has to show up somewhere, so currency outstanding has to increase. Private financial intermediaries cannot convert T-bills into currency, as they are not permitted (either explicitly or implicitly in the US) to issue currency. That's why monetary policy matters in a channel system - central bank intervention works by varying the quantity of liquidity transformation.
But what happens if, for example, the central bank purchases long Treasury bonds under a floor system? The Fed issues reserves, and purchases Treasury bonds, thus transforming T-bonds into overnight reserves. Does that do anything? Why should it? A private financial institution can create a special purpose vehicle (SPV) whose only function is to hold T-bonds as assets, and finance that portfolio by rolling over overnight repos. The SPV performs exactly the same asset transformation as the central bank is performing when it purchases T-bonds. Imagine, for example, that there is an SPV that sells T-bonds to the Fed, and suppose that the overnight repos of the SPV were held by a financial institution with a reserve account. Suppose further that this financial institution increases its reserve account balance by exactly the amount of the reduction in its repo holdings after the Fed purchases the T bonds. Clearly, the financial institution does not care whether the T-bonds that back its overnight assets are held by the Fed or the SPV. Nothing changes.
This is just a more elaborate liquidity trap. Under a floor system, the only instrument the central bank has is the IROR. Asset purchases - of whatever - are irrelevant. Under current circumstances, this means that, as long as the Fed does not change the IROR, the inflation rate is passive. Changes in the price level are determined by the demand and supply of liquid assets - the whole array of that stuff, i.e. currency, reserves, interest-bearing government debt of all maturities, liquid asset-backed securities. The Fed can only change the composition of the total stock of liquid assets under a floor system, and so it can't change the price level without changing the IROR. What will move the price level, as long as the IROR is fixed? First, if the private sector creates more liquid assets that compete with reserves, that will increase the price level, and we will get more inflation. That is what I was worried about a while back. Not right now. If it looks unlikely that the private sector will be creating more liquid assets, and if the demand for US-dollar-denominated liquid assets rises, the price level falls and we get less inflation. That's our problem now. Of course there's nothing the Fed can do about that, as the IROR does not have far to go to reach zero.
Some specific comments in reply to Miles's post:
1. Miles argues that, even if we have some doubts about QE, why not try it? At worst it's irrelevant, so no big deal. The problem is that the Fed wants to believe it works, otherwise it looks silly, given the massive QE operations that it has engaged in. Various economists in the Fed system have been falling all over themselves to justify the actions of their superiors, and smart people like Miles are buying the arguments. The Fed has now convinced itself that QE works. In particular, the Fed thinks it works both ways. Thus, if we ultimately see what we think is too much inflation (a more remote possibility at the moment, obviously), the Fed will think it can control it through asset sales. It certainly can, but the sales only start to bite at the point where excess reserves get very close to zero.
2. Miles isn't sure exactly what friction makes QE work, but he seems confident that the friction is small, and so QE must be carried out on a very large scale in order to do much. In fact, as he says:
Miles has taken the trouble to write at length in reply to my comments on one of his posts, so I'm encouraged to carry on the discussion.
For some background on liquidity traps, quantitative easing (QE) and what it can and cannot do, you can browse my archive, or read these particular pieces:
1. QE irrelevance.
2. An example.
3. Liquidity traps.
To understand QE, we need to know the difference between a channel system and a floor system. A good example of a channel system is the central banking framework within which the Bank of Canada works. The Bank sets a deposit rate (the interest rate on reserve accounts) and a higher rate at which it lends to financial institutions, and targets an overnight rate that lies between those two rates (i.e. the target rate lies in the "channel"). In a channel system, as long as the overnight rate lies within the channel, open market operations matter - purchases or sales of assets by the central bank will move the overnight rate.
Prior to the financial crisis, the Fed worked within what was essentially a channel system. The interest rate on reserves (IROR) was zero, excess reserves were essentially zero overnight, and the overnight fed funds rate fell between zero and the discount rate (there are some complications involving what the "fed funds rate" is, and how lending at the discount window takes place, but ignore that for now). Pre-financial crisis, if the Fed bought or sold assets (T-bills, long Treasuries, whatever), that would move the fed funds rate.
A floor system is different. Under such a system, the central bank sets an interest rate on reserves and a central bank lending rate, and plans to have a positive supply of reserves in the system overnight. As a result, the overnight rate must be equal to the IROR, by arbitrage. An open market operation in short-term government debt in a floor system will have no effect, at the margin, as the central bank is simply swapping one interest-bearing short-term asset for another. The instrument of monetary policy in a floor system is the IROR, which determines short-term nominal interest rates.
Currently, the Fed operates under a floor system. The supply of excess reserves is enormous, and the IROR determines short-term interest rates. There are some weird features of the system, such as the fact that the GSEs receive no interest on their reserve accounts, and there is some lack of arbitrage which results in a fed funds rate less than the IROR, but I think those weird features are irrelevant to how monetary policy works.
Under a floor system, we are effectively in a perpetual liquidity trap. Conventional open market operations in short-term government debt do not matter, whether the IROR is 5%, 0.25%, 10%, or zero. But, not to worry, the central bank can always change the IROR except, of course, when it hits the zero lower bound (neglecting the possibility of taxation of reserve balances, which is another issue altogether).
My irrelevance argument goes one step further. First, we have to understand why open market operations move the overnight rate in a channel system. An open market purchase of T-bills under a channel system essentially involves the transformation by the central bank of T-bills into currency. Remember that, in a channel system, overnight reserve balances are zero. The increase in outside money has to show up somewhere, so currency outstanding has to increase. Private financial intermediaries cannot convert T-bills into currency, as they are not permitted (either explicitly or implicitly in the US) to issue currency. That's why monetary policy matters in a channel system - central bank intervention works by varying the quantity of liquidity transformation.
But what happens if, for example, the central bank purchases long Treasury bonds under a floor system? The Fed issues reserves, and purchases Treasury bonds, thus transforming T-bonds into overnight reserves. Does that do anything? Why should it? A private financial institution can create a special purpose vehicle (SPV) whose only function is to hold T-bonds as assets, and finance that portfolio by rolling over overnight repos. The SPV performs exactly the same asset transformation as the central bank is performing when it purchases T-bonds. Imagine, for example, that there is an SPV that sells T-bonds to the Fed, and suppose that the overnight repos of the SPV were held by a financial institution with a reserve account. Suppose further that this financial institution increases its reserve account balance by exactly the amount of the reduction in its repo holdings after the Fed purchases the T bonds. Clearly, the financial institution does not care whether the T-bonds that back its overnight assets are held by the Fed or the SPV. Nothing changes.
This is just a more elaborate liquidity trap. Under a floor system, the only instrument the central bank has is the IROR. Asset purchases - of whatever - are irrelevant. Under current circumstances, this means that, as long as the Fed does not change the IROR, the inflation rate is passive. Changes in the price level are determined by the demand and supply of liquid assets - the whole array of that stuff, i.e. currency, reserves, interest-bearing government debt of all maturities, liquid asset-backed securities. The Fed can only change the composition of the total stock of liquid assets under a floor system, and so it can't change the price level without changing the IROR. What will move the price level, as long as the IROR is fixed? First, if the private sector creates more liquid assets that compete with reserves, that will increase the price level, and we will get more inflation. That is what I was worried about a while back. Not right now. If it looks unlikely that the private sector will be creating more liquid assets, and if the demand for US-dollar-denominated liquid assets rises, the price level falls and we get less inflation. That's our problem now. Of course there's nothing the Fed can do about that, as the IROR does not have far to go to reach zero.
Some specific comments in reply to Miles's post:
1. Miles argues that, even if we have some doubts about QE, why not try it? At worst it's irrelevant, so no big deal. The problem is that the Fed wants to believe it works, otherwise it looks silly, given the massive QE operations that it has engaged in. Various economists in the Fed system have been falling all over themselves to justify the actions of their superiors, and smart people like Miles are buying the arguments. The Fed has now convinced itself that QE works. In particular, the Fed thinks it works both ways. Thus, if we ultimately see what we think is too much inflation (a more remote possibility at the moment, obviously), the Fed will think it can control it through asset sales. It certainly can, but the sales only start to bite at the point where excess reserves get very close to zero.
2. Miles isn't sure exactly what friction makes QE work, but he seems confident that the friction is small, and so QE must be carried out on a very large scale in order to do much. In fact, as he says:
Balance sheet monetary policy can powerfully stimulate the economy if the Fed does enough.That seems inconsistent with the rest of the argument. Suddenly we go from statements about how little we know to confident predictions about what we can accomplish if only we do "enough." You have to be more specific about the "enough" to give that content.
Sunday, June 3, 2012
The Beveridge Curve and the Long-Term Unemployed
It's well known that the Beveridge curve relationship (the negative correlation between the unemployment rate and the vacancy rate) shifted out as the unemployment rate began to come down from its peak of 10% in late 2009. The first chart shows the most recent update. 
The data in the chart runs up to March 2012, the latest date for which we have JOLTS data. The unemployment/vacancies data from December 2000 to December 2007 (the beginning of the last recession) trace out what is apparently a stable Beveridge curve. As well, the data from December 2007 through October 2009 is consistent with that stable Beveridge curve, but the post-October 2009 observations make it appear that the relationship has shifted. If we had thought that the 2000-2007 Beveridge curve was structural, we would have predicted an unemployment rate of 5.25-5.5%, conditional on the observed vacancy rate in March 2012. But the actual unemployment rate in March 2012 was 8.2%.
Suppose that we disaggregate, and look for Beveridge relationships in terms of duration of unemployment. The second chart is a scatter plot of the vacancy rate vs. the unemployment rate for those unemployed less than 5 weeks (those unemployed less than five weeks divided by the total labor force).
In this chart, you don't see any correlation at all. But in the next two charts, you get nice Beveridge curve correlations, for those unemployed 5 to 14 weeks, and 15 to 26 weeks respectively. 
So, now you know what group has to be explaining the shift in the Beveridge curve in the first chart. The next chart is for those unemployed 27 weeks or more. 
In this chart I have connected the dots and supplied some dates. The shifts you see in the Beveridge curve in the first chart seem to be entirely due to what is going on with the long-term unemployed. Further, in this last chart you can see an earlier shift, which occurred after the 2001 recession.
Finally, let's take a look at the number of those unemployed 27 weeks or more, as a fraction of total unemployed, since 1948.
Long-term unemployed as a fraction of total unemployed increases in all recessions, for obvious reasons, but the behavior of the time series after the last 3 recessions is striking. After the last 3 "jobless recovery" recessions, the fraction of unemployed who are long-term unemployed has remained persistently high. The most recent recession is just an exaggerated version the the previous two - the long-term unemployed account for an extremely high fraction of total unemployed, and that fraction is persistent, just as in the previous two recessions.
So, where do these observations lead us? The unemployment rate is currently unusually high, and high in a way that does not appear to be consistent with posted vacancies. But if we disaggregate, it seems that the characteristics of the time series might have a lot to do with the fact that there are much more long-term unemployed now than is typically the case. But why has that happened? Especially since the phenomenon appears not to be new (going back to 1990 at least), it's hard to avoid thinking about mismatch. But in order to evaluate that story, we need more information about the long-term unemployed. How many of these are former construction workers? How many are David Autor's middle-skill people? Possibly the financial crisis merely increased the rate of structural change that was already occurring in labor markets? Are there other features of the long-term unemployed we need to be thinking about? The long-term unemployed may have depreciated skills; they may have been picked over as a group and be of perceived low average quality. Their search effort may be low. All of these things matter for policy, particularly for unemployment insurance programs.
What is clear is that conventional models typically have insufficient heterogeneity to explain these facts. In Mortensen-Pissarides models, for example, labor is homogeneous, and mismatch is embedded in reduced-form matching functions. I'm interested in learning about work you know about that either already captures this stuff, or could potentially do so.
Suppose that we disaggregate, and look for Beveridge relationships in terms of duration of unemployment. The second chart is a scatter plot of the vacancy rate vs. the unemployment rate for those unemployed less than 5 weeks (those unemployed less than five weeks divided by the total labor force).
Finally, let's take a look at the number of those unemployed 27 weeks or more, as a fraction of total unemployed, since 1948.
So, where do these observations lead us? The unemployment rate is currently unusually high, and high in a way that does not appear to be consistent with posted vacancies. But if we disaggregate, it seems that the characteristics of the time series might have a lot to do with the fact that there are much more long-term unemployed now than is typically the case. But why has that happened? Especially since the phenomenon appears not to be new (going back to 1990 at least), it's hard to avoid thinking about mismatch. But in order to evaluate that story, we need more information about the long-term unemployed. How many of these are former construction workers? How many are David Autor's middle-skill people? Possibly the financial crisis merely increased the rate of structural change that was already occurring in labor markets? Are there other features of the long-term unemployed we need to be thinking about? The long-term unemployed may have depreciated skills; they may have been picked over as a group and be of perceived low average quality. Their search effort may be low. All of these things matter for policy, particularly for unemployment insurance programs.
What is clear is that conventional models typically have insufficient heterogeneity to explain these facts. In Mortensen-Pissarides models, for example, labor is homogeneous, and mismatch is embedded in reduced-form matching functions. I'm interested in learning about work you know about that either already captures this stuff, or could potentially do so.
Saturday, June 2, 2012
Friday, June 1, 2012
Quantitative Easing: The Conventional View
I ran across two pieces by well-known macroeconomists that support - wholeheartedly - the Fed's view of why it conducts quantitative easing (QE) exercises, and why QE is supposed to work. The first is by Miles Kimball, in a blog piece intended for a wide audience. The second is this paper by Roger Farmer, which is a quasi-formal approach to the question.
Kimball's narrative would make any Old Keynesian or New Keynesian comfortable. Here's what he says:
1. There is a Phillips curve:
Core inflation has been rising (mostly) by either measure since late 2010. Does Kimball think the US economy is above the "natural level of output?" I doubt it. If not, he should re-think his definition. Maybe he could tell us how to measure the natural level of output while he is at it.
2. What should a central bank do?
3. Lowering interest rates increases aggregate demand, and there are no practical limits on the stimulative effects the Fed can have:
So, Kimball needs to think harder about QE. What about Farmer? The empirical part of the paper we have mostly seen before. These are the kind of event-study-type pictures that are sometimes used to "prove" that QE works. Any self-respecting economist will take more convincing than that. A serious model and some solid structural work would be nice. The theory part of Farmer's paper is supposed to be model free. It's basic intertemporal asset pricing, and the idea is the following. Suppose there exists nominal government debt of different maturities. In Woodford fashion, we can think of monetary policy as a contingent rule for setting the one-period nominal interest rate. But what if our rule is not feasible, i.e. the one-period nominal interest rate can't be nonnegative in all states of the world, given the rule we would like to impose. Then, it must be optimal for the one-period nominal rate to be zero in some states, but we then need to choose the short nominal rates in the other states. Given term structure relationships, the thought experiment we conduct in determining the optimal policy rule effectively involves changing some long-term nominal rates. Apparently that's how Farmer thinks about QE. Problem: To say how QE works, we have to have the relevant asset swaps in the model. There are no asset quantities in siqht in the framework that Farmer lays out. That isn't much help.
Kimball's narrative would make any Old Keynesian or New Keynesian comfortable. Here's what he says:
1. There is a Phillips curve:
The “natural level of output” is the level of output at which core inflation will be steady. Above the natural level of output, core inflation rises. Below the natural level of output, core inflation falls.Here's core inflation for the last 5 years, as measured using the core CPI and core PCE deflator.
2. What should a central bank do?
WHEN BELOW NATURAL OUTPUT: PRINT MONEY AND BUY ASSETS!According to Kimball, this prescription need not be modified given the current state of affairs. He says it does not matter if short-term nominal interest rates are zero, or close to it, because:
Whenever the Fed buys any asset, its price goes up.Kimball makes this seem like a simple application of what we learned in Econ 101, but he thinks we might need a little Finance too:
One of the most useful facts in all of Finance comes into play. For assets, a higher price is basically the same thing as a lower interest rate.Another useful piece of finance is the Modigliani Miller theorem. Kimball might want to explain to us why, if the Fed issues reserves (overnight liabilities), and buys some other assets, and private financial intermediaries are perfectly capable of issuing overnight liabilities and buying the same assets, that the Fed's QE is not undone.
3. Lowering interest rates increases aggregate demand, and there are no practical limits on the stimulative effects the Fed can have:
What if all the assets in the world got down to a zero nominal interest rate and the economy still didn’t have enough stimulus? Then, and only then, we would be in deep, deep trouble on the aggregate demand front from which there would be no escape through monetary policy. But we are far, far away from that situation. Simple economic models studied by economists often have this happen because they have so few types of assets in them, but the real world has a huge number of different types of assets, some with nominal interest rates that are still very far from zero.What's wrong with that paragraph? What's right about it? A central bank is a financial intermediary. Its power to alter the allocation of resources and economic welfare derives from its monopoly over the issue of some special kinds of liabilities (currency and reserves) which are used in retail transactions and large-value financial transactions. As Kimball notes, all but a small quantity of the reserves currently outstanding are currently "asleep," i.e. they sit during the day and overnight, and are not so different from T-bills (except that more economic agents can hold T-bills than have reserve accounts). If the Fed issues reserves and buys long-term Treasury bonds under these conditions, that can have no effect, as that's a process of intermediating Treasury bonds that is no different from what can be done by a shadow bank. If the Fed issues reserves and buys mortgage-backed securities issued by Fannie Mae or Freddie Mac, that amounts to the same thing - no effect. However, if the Fed were to, for example, buy mortgages directly, that would be an entirely different game. How good is the Fed at screening mortgage borrowers? Is the Fed going to target particular segments of the mortgage market? Maybe Congress wants some say in how the Fed does that? Maybe some people will be lobbying Congress in a serious way to make sure that their segment of the credit market gets the intervention? Sounds like opening a can of worms, doesn't it?
So, Kimball needs to think harder about QE. What about Farmer? The empirical part of the paper we have mostly seen before. These are the kind of event-study-type pictures that are sometimes used to "prove" that QE works. Any self-respecting economist will take more convincing than that. A serious model and some solid structural work would be nice. The theory part of Farmer's paper is supposed to be model free. It's basic intertemporal asset pricing, and the idea is the following. Suppose there exists nominal government debt of different maturities. In Woodford fashion, we can think of monetary policy as a contingent rule for setting the one-period nominal interest rate. But what if our rule is not feasible, i.e. the one-period nominal interest rate can't be nonnegative in all states of the world, given the rule we would like to impose. Then, it must be optimal for the one-period nominal rate to be zero in some states, but we then need to choose the short nominal rates in the other states. Given term structure relationships, the thought experiment we conduct in determining the optimal policy rule effectively involves changing some long-term nominal rates. Apparently that's how Farmer thinks about QE. Problem: To say how QE works, we have to have the relevant asset swaps in the model. There are no asset quantities in siqht in the framework that Farmer lays out. That isn't much help.
Subscribe to:
Posts (Atom)