One argument against reducing the interest rate on reserves to zero (as there were for decades before 2008,) is that this would cause problems for money market mutual funds.
What is the problem exactly?
Presumably it would involve banks buying money market instruments (short and safe bonds,) rather than hold reserve balances. This would tend to drive up the prices and lower the yields on those money market instruments. And then, the money market mutual funds would earn less on their asset portfolios.
So? Why can't money market mutual funds handle this situation?
Tuesday, September 11, 2012
An Alternative Monetary Regime
Consider the following monetary regime. There is a central bank that solely issues reserve balances and uses them to fund private securities. The central bank is owned by the banks. Any profits or losses on the central bank's asset portfolio are distributed to the banks. The central bank issues no hand-to-hand currency. Hand-to-hand currency is solely issued by the banks. The banks also issue checkable deposits.
The government regulates the central bank, requiring it to keep nominal GDP on target. (Maybe it could require index futures convertibility.)
The government runs budget deficits and has a national debt. I funds the national debt with a variety of securities, some with short terms to maturity. The government has very good credit and is expected to always pay off bonds. The government accepts checks drawn on private banks in payment of taxes. It generally deposits those checks in the banks against they were drawn. It follows the same policy when it sells bonds. It accepts checks in return for bonds and deposits those checks in the banks against they are drawn. The government funds expenditures and debt repayments by writing checks from these various private banks.
The stock market crashes and those who previously were buying stocks buy short term to maturity government bonds. The yields on those government bonds are driven to zero.
Does monetary policy now become ineffective?
The nominal interest rate on some government bonds is zero. Further, the interest rate is the same as the interest rate on hand-to-hand currency.
I think the answer is no. What is it about the existing monetary regime that causes some economists to believe that a zero yield on some government bonds makes monetary policy ineffective? (That is, to increase nominal expenditure.)
The government regulates the central bank, requiring it to keep nominal GDP on target. (Maybe it could require index futures convertibility.)
The government runs budget deficits and has a national debt. I funds the national debt with a variety of securities, some with short terms to maturity. The government has very good credit and is expected to always pay off bonds. The government accepts checks drawn on private banks in payment of taxes. It generally deposits those checks in the banks against they were drawn. It follows the same policy when it sells bonds. It accepts checks in return for bonds and deposits those checks in the banks against they are drawn. The government funds expenditures and debt repayments by writing checks from these various private banks.
The stock market crashes and those who previously were buying stocks buy short term to maturity government bonds. The yields on those government bonds are driven to zero.
Does monetary policy now become ineffective?
The nominal interest rate on some government bonds is zero. Further, the interest rate is the same as the interest rate on hand-to-hand currency.
I think the answer is no. What is it about the existing monetary regime that causes some economists to believe that a zero yield on some government bonds makes monetary policy ineffective? (That is, to increase nominal expenditure.)
Sunday, September 9, 2012
Interest Rates, new Keynesians, and Market Monetarists
When Market Monetarists read new Keynesians, we are often puzzled by their focus on the policy interest rate. They argue as if (and sometimes say,) that the only interest rate that really counts is the expected future path of the policy interest rate.
This approach is institution specific. What would happen if there were no policy interest rate? Sure, central banks like targeting monetary market interest rates, but they don't have to do that. Plenty of economists favor letting all interest rates depend on supply and demand.
Suppose that a central bank targets the monetary base. All interest rates depend on supply and demand in credit markets. There is no policy interest rate. There is no expected future path of the policy interest rate. It is impossible for a policy formula to spit out a policy interest rate that is below zero. Comparing spending on output over time with a hypothetical scenario where the policy rate falls below zero according to the rule to one where it fails to fall enough but rather just stays at zero for a protracted period of time is impossible.
With a base money rule, if spending on output is too low, the target for base money is increased. If spending is too high, the target for base money is decreased.
Further, consider the scenario where spending is too low, and so base money is being increased. What do we make of the zero nominal bound in this scenario? The only obvious zero nominal bound is that base money cannot be negative, and with spending being too low and base money increasing, that is hardly a problem. The other bound is slightly more relevant. The central bank might purchase all the assets that it can, and so it cannot raise base money any more. Let's ignore that "upper bound" for now.
No, the problem must be that some financial asset other than the central bank's own liabilities has a zero nominal yield. I think the most plausible candidate would be that the central bank is purchasing some kind of asset with newly created base money, and the central bank is bidding up its price by doing this, and so driving down its yield. At some point, the yield is driven down to zero. The yield on the asset the central bank is buying to create base money has a zero nominal yield.
However, it is certainly possible that private investors might demand more of the asset the central bank usually purchases and push up its price and drive down its yield. For example, suppose the central bank generally buys T-bills, and private investors worried about losses on other financial assets like stocks sell them and buy T-bills. This effect would be independent of purchases of T-bills in anticipation of central bank purchases to raise base money in response to inadequate spending on output. Conceivably, private investors might force the T-bill yield to zero, and so now if the central bank purchased T-bills to expand base money, it would be buying an asset with a zero nominal yield.
So here we have a possible zero-nominal bound problem. The central bank is purchasing assets that have a zero yield. Now, Market Monetarists and new Keynesians agree that spending on output today will be influenced by spending on output in the future and the level of base money in the future will impact spending on output in the future. For the new Keynesians this occurs because at some point in the future, keeping the policy interest rate at zero will result in growing expenditure--presumably at ever increasing rates. Keeping spending from growing in the future would require increases in the policy rate. By increasing the policy rate less than that, spending can increase to whatever amount is desired. (Only if nominal interest rates are expected to be zero forever would this be impossible. Let's leave that possibility aside.)
From a Market Monetarist perspective, a level of base money consistent with some higher level of nominal expenditure on output would start to cause spending to rise to that higher level once the demand for base money at that future date falls below the quantity at the future date. So the question then has nothing to do with the impact of different targets. Nominal GDP level targeting will tend to keep spending on output from falling as compared to a pure inflation target or even a Taylor-type rule.
From a Market Monetarist perspective, a level of base money consistent with some higher level of nominal expenditure on output would start to cause spending to rise to that higher level once the demand for base money at that future date falls below the quantity at the future date. So the question then has nothing to do with the impact of different targets. Nominal GDP level targeting will tend to keep spending on output from falling as compared to a pure inflation target or even a Taylor-type rule.
However, it is also true that Market Monetarists agree with new Keynesians that spending on output can fall below target, even with a nominal GDP rule. The issue, then, is whether "extra" increases in base money will increase spending on output beyond the increases in spending that would occur anyway due to the expectation that eventually spending on output will return to the target path. To clarify, assuming there is a nominal GDP level target, assuming that spending will not be permitted to rise above target at any point, will increases in base money while spending on output is below target cause spending on output to be closer to the target or reach the target more rapidly?
Suppose we knew what level of base money will be necessary in 2015 to get nominal GDP back to the target path. Base money has reached that point. The interest rate on assets the central bank buys is already zero. The new Keynesian argument is that purchasing more of that asset now, knowing that the additional purchases will be reversed and base money will return to the proper 2015 target, when spending will also reach its proper target, will do nothing for spending today. Extra purchases of this asset, and the extra increases in base money, will do nothing to raise spending closer to target than it otherwise would be or hasten the return of spending on output to target. As long as people believe that base money will be adequate in 2015 for spending to return to target in 2015, then spending today will be as high as the central bank can make it. Any additional increase in base money is pointless.
Why is it that open market purchases of an asset with a zero interest rate create no extra increase in spending now. Market Monetarists and new Keynesians would agree that ephemeral changes in base money or reductions in policy interest rates would have little effect. So, the question is whether extra purchases of zero rate assets with newly created base money that will persist for some extended period of time will help. Again, will they either cause spending on output to be closer to target while remaining below target or will they hasten the return of spending on output to target?
Woodford gives a very traditional Keynesian argument for the ineffectiveness of open market operations when the interest rate is zero. The demand for base money becomes perfectly elastic. Now, when the supply of a good increases and demand is perfectly elastic, then the price remains the same--it doesn't fall as would occur if demand has the usual downward slope. And so, Woodford appears to be saying that when the nominal interest rate is zero, an increase in the quantity of money cannot result in a lower nominal interest rate. But perhaps the real significance of the argument is that the quantity of money demanded shifts dollar-for-dollar with the increase in the supply of money.
The Barro argument is that when interest rates on the assets the central bank purchases are equal to zero, then they become perfect substitutes for base money. If the central bank purchases such an asset with newly created base money, it increases the quantity of money in the form of base money and reduces the quantity of money in the form of these zero interest bonds. The total quantity of money, inclusive of both base money and the zero-interest bonds remains the same. It would be like replacing a $20 bill (unit of currency) with two $10 bills. The quantity of money is not effected.
The Yeager argument is that when there is a shortage of some bond at a yield of zero, the excess demand for the bonds might be shifted to an increased demand for money. If a central bank seeks to accommodate the added demand for money by increasing the quantity of money, and does so by purchasing more of the very same bond that is already facing an excess demand, then while it is creating more money, it worsens the shortage of the bonds by purchasing more of them. When that is shifted over to an added demand for money, the central bank is left where it began. It has failed to correct the underlying shortage of money.
From a Market Monetarist perspective, the only reason why spending on output falls below target is an excess demand for money. That doesn't mean that the problem must be caused by a reduction in the quantity of base money or that an increase in the demand to hold money beyond an increase in the quantity of money occurs as a "bolt from the blue." It could occur for any number of reasons. Still, increasing the quantity of base money should relieve the problem unless there is some special situation where the very process by which base money increases results in a matching increase in the demand to hold base money, leaving the underlying shortage unchanged--or conceivably worsened.
The Woodford, Barro, and Yeager arguments all suggest that a central bank purchasing assets with a yield of zero would not relieve the underlying excess demand for money. Yes, Market Monetarists and New Keynesians agree that at some future time, this won't be an issue, an appropriate path for base money or policy interest rates will return spending on output to target, and that expectation of this recovery will limit current deviations of spending from target. But, increases in base money created by purchases of an asset with an interest rate of zero are likely to do nothing now directly.
In other words, if base money is at the level now that will bring nominal GDP back to the target growth path at some future date, increasing base money beyond that point now will not cause spending to rise closer to the current target during that period it remains below target nor will it help reach the target sooner. There is no point to raising base money. (Of course, no one really knows what the level base money needs to be to reach the target in the future, so raising base money during the period when spending is too low might still be sensible.)
Of course, from a Market Monetarist approach, the question is by how much base money needs to increase in order to get nominal GDP back to target. If increasing base money by purchasing assets with a zero nominal interest rate has little effect, and those assets are going to be purchased anyway, then this simply implies that a larger increase in base money is necessary. All of the zero yield securities must be purchased and then other securities must be purchased as well.
Once the central bank is purchasing securities that don't have a zero yield, then the Barro and Yeager arguments no longer apply. The securities being purchased would not be perfect substitutes for base money, and so, the quantity of money would be increasing. There is not a shortage of those securities at their current positive yield, and as long as the purchases don't go so far as to create a shortage, there is no shortage to be shifted to an increase in the demand for money. The quantity of money rises, without any self-inflicted increase in the demand to hold money, and so the underlying excess demand for money is relieved.
This suggests that a sufficient expansion of base money now, which would require that on the margin assets be purchased that have yields greater than zero, would result in an increase in spending on output. The shortfall in spending on output could be reduced, and spending on output could return to target more quickly. Since expectations that spending on output will be higher sooner should result in more spending due to the effect expectations of future spending, this would be a second avenue by which the extra increases in base money would help raise spending.
What about the Woodford account? The demand for money is supposedly perfectly elastic at an interest rate of zero. But the interest rate for the securities the central bank is purchasing is not zero. Of course, Woodford's assumption is that the interest rate on those securities would be reduced by the central bank's purchases, lowering the opportunity cost of holding money, so that in the end, the quantity of money demanded rises to match the added quantity. From Woodford's interest-centric view, what is happening is that once the interest rate on one particular asset is driven to zero, then the interest rate on some other asset is lowered. If that one is also driven to zero, then another is lowered.
This process of driving one interest rate and then another to zero occur until the central bank has lowered the interest rate on all assets it can buy to zero. This is very much parallel to the "upper bound" for base money targeting. Once the central bank has purchased every asset it is permitted to purchase, then no further increase in base money is possible.
This points to another institutionally specific element of new Keynesian thinking. Central banks like to target interest rates, and new Keynesian economists narrowly focuses on interest rate targeting. Central banks like to purchase safe money market instruments. New Keynesian economics focuses on just such instruments.
Market Monetarists insist that the problem of a shortfall of spending from target is that the quantity of money is less than the demand to hold money. If small modest increases in base money used to purchase safe money market instruments are inadequate to expand the quantity of money enough to meet the demand under some circumstances, then heroic purchases of longer term and riskier securities should be used.
Woodford argues that heroic open market purchases of assets with interest rates above zero will not be effective. They will not raise spending closer to target or allow it to reach the target sooner. The effect is the same as if base money simply expands (or contracts) whatever amount is needed to keep the policy rate equal to zero until spending returns to target. But what does this mean with base money targeting? Spending will not rise closer to target or reach the target sooner due to any increase in base money beyond the level needed to keep spending on output at target when the interest rate on the asset the central bank usually buys rises above zero.
But surely this depends on what particular bond the central bank typically buys. Suppose the central bank usually buys government bonds with maturities between 2 and 5 years. Now suppose further that the interest rate on 3 month bill T-bills falls to zero. The central bank never buys any of those anyway. It continues to purchase the longer term bonds as usual. Is it impossible to relieve an excess demand for base money because the 3 month T-bills has a zero yield? Suppose instead that the central bank targets a yield for bonds with 4 years to go. Can it not continue to make adjustments in that interest rate even if the 3 month T-bills is zero?
Now, suppose that the Treasury never sold T-bills at all. Whatever market conditions would have caused that yield to fall to zero still occur, but there are no T-bills. Of course, the central bank doesn't buy nonexistent T-bills. It buys government bonds with 3 to 5 year maturities. Do market conditions that would create a zero yield on nonexistent T-bills make increases in base money ineffective?
Suppose now that there is no national debt at all and the central bank only purchases private securities that bear default risk. (Perhaps the liabilities of the central bank should count as government debt.) Suppose market conditions that would have caused the interest rate on T-bills to fall to zero occurs, but there are no government bonds at all. Does the central bank's expansion of base money by purchasing risky private securities become ineffective?
Suppose one business finds that it is able to borrow at a zero nominal rate. Does this make increases in base money by the central bank when it purchases bonds issued by other firms ineffective?
Now consider a cashless (or currency-less) monetary regime. All payments are made by transfer of bank balances. The interest rate on balances at the central bank are slightly less than the interest rate on short term government debt. If the yield on that debt falls, then the interest rate the banks receive on their balances at the central bank fall in proportion.
There is no zero nominal bound on interest rates. Interest can be as negative as necessary to keep spending on output on target. What is it about these negative interest rates on short term government bonds that solves the problem? If we have an economy that only has consumer goods and everyone is the same, then the only way for spending to rise is for everyone to consume more now. The lower interest rate on government debt, central bank balances, money balances held in banks, and everything else, have to deter desired saving in the current period and so raise desired consumption. There is nothing else.
But in reality, when there are different people, then a sufficiently negative interest rate on government bonds would motivate households that save to lend more to households that dissave. This involves bearing more risk, on the assumption that households that dissave are more likely to default than the government. Further, in an economy with investment, then some of those holding government bonds might sell their bonds due to the low yields and spend the money on capital goods now in order to get consumer goods in the future. Others might lend to firms that will purchase capital goods rather than lend to the government. Of course, this involves more risk, assuming that investments in capital goods are more likely to fail than the probability the government will default on its debt.
Something happens that makes market-clearing interest rates on short and safe securities negative. If the central bank's own liabilities are short and safe and it keeps the interest rate on them at zero, then no interest rates are going to fall to the market clearing rates because people will hold the liabilities of the central bank instead. But if the central bank expands the quantity of base money enough, it will bear the risk and so all of the impact of hypothetically negative nominal interest rates on short and safe assets in terms of motivating purchases of capital goods or debt financed consumer goods are still obtained. Only by assuming those are nonexistent, that is, in a consumer good only, representative agent economy, will it look like the expansions in base money have no effect.
Finally, suppose that all hand-to-hand currency is issued by risky private firms. The central bank only offers deposit accounts to banks, and the banks using the central bank only issue deposit accounts. The interest rate on short-and-safe government bonds can turn negative. The interest rates on bank deposits can turn negative. The zero-nominal interest rate on privately-issued currency is irrelevant, because it is quite risky. It is only held for its "liquidity" yield. Presumably criminals still use it as do people making small face-to-face transactions. This currency is not base money. It is like travelers checks issued by fly-by-night firms.
If the central bank makes open market purchases of T-bills, and drives their prices above par and their yields below zero, then the interest rates paid on reserves and the interest rates banks pay on deposits are also driven below zero. Saving is less attractive and consumption more attractive, but also there is a motivation to bear more risk by lending money to dissavers and financing capital goods either through lending to firms or selling of short and safe securities to buy capital goods.
Now, suppose the risk of these investments in capital goods or loans to dissaving households approaches that of the risky hand-to-hand currency. Then, there will be an increase in the demand for that currency and the firms issuing it presumably will have less rather than more incentive to issue it. Currency becomes more difficult to obtain for small face-to-face transactions and criminal activities. To keep spending on target, other sorts of spending must expand. (Presumably the criminal activities weren't measured anyway.) This would be spending undertaken by check or electronic payment.
Of course, if all deposits were made redeemable in this hand-to-hand currency, even though it is risky and issued by private firms, then it would be impossible to expand spending by check or electronic payment to offset the reduction in spending of hand-to-hand currency. On the contrary, the shortage of hand-to-hand currency would generate a shortage of deposits and reduced spending on output too.
So, central banks are governmental institutions issuing hand-to-hand currency that is safe, short, and has a zero nominal yield. All other money is redeemable in that currency. The central banks like to make periodic changes in money market interest rates implemented by modest changes in purchase of short and safe assets--generally short term to maturity government bonds. If there is a large increase in the demand for those short and safe government bonds, then central banks must do something different to keep spending growing at a slow and steady growth path. Bearing more risk itself to motivate more investment in capital goods and more consumer loans is possible. This raises spending on output now, and when the market clearing rates on short and safe assets rise to zero, then it can go back to holding those and bear less risk. While it may appear undesirable for central banks to bear such risk, it is the least bad alternative if they insist on issuing zero-risk, zero-nominal-yield, hand-to-hand currency, or rather, in making all other money redeemable in their currency.
Woodford gives a very traditional Keynesian argument for the ineffectiveness of open market operations when the interest rate is zero. The demand for base money becomes perfectly elastic. Now, when the supply of a good increases and demand is perfectly elastic, then the price remains the same--it doesn't fall as would occur if demand has the usual downward slope. And so, Woodford appears to be saying that when the nominal interest rate is zero, an increase in the quantity of money cannot result in a lower nominal interest rate. But perhaps the real significance of the argument is that the quantity of money demanded shifts dollar-for-dollar with the increase in the supply of money.
The Barro argument is that when interest rates on the assets the central bank purchases are equal to zero, then they become perfect substitutes for base money. If the central bank purchases such an asset with newly created base money, it increases the quantity of money in the form of base money and reduces the quantity of money in the form of these zero interest bonds. The total quantity of money, inclusive of both base money and the zero-interest bonds remains the same. It would be like replacing a $20 bill (unit of currency) with two $10 bills. The quantity of money is not effected.
The Yeager argument is that when there is a shortage of some bond at a yield of zero, the excess demand for the bonds might be shifted to an increased demand for money. If a central bank seeks to accommodate the added demand for money by increasing the quantity of money, and does so by purchasing more of the very same bond that is already facing an excess demand, then while it is creating more money, it worsens the shortage of the bonds by purchasing more of them. When that is shifted over to an added demand for money, the central bank is left where it began. It has failed to correct the underlying shortage of money.
From a Market Monetarist perspective, the only reason why spending on output falls below target is an excess demand for money. That doesn't mean that the problem must be caused by a reduction in the quantity of base money or that an increase in the demand to hold money beyond an increase in the quantity of money occurs as a "bolt from the blue." It could occur for any number of reasons. Still, increasing the quantity of base money should relieve the problem unless there is some special situation where the very process by which base money increases results in a matching increase in the demand to hold base money, leaving the underlying shortage unchanged--or conceivably worsened.
The Woodford, Barro, and Yeager arguments all suggest that a central bank purchasing assets with a yield of zero would not relieve the underlying excess demand for money. Yes, Market Monetarists and New Keynesians agree that at some future time, this won't be an issue, an appropriate path for base money or policy interest rates will return spending on output to target, and that expectation of this recovery will limit current deviations of spending from target. But, increases in base money created by purchases of an asset with an interest rate of zero are likely to do nothing now directly.
In other words, if base money is at the level now that will bring nominal GDP back to the target growth path at some future date, increasing base money beyond that point now will not cause spending to rise closer to the current target during that period it remains below target nor will it help reach the target sooner. There is no point to raising base money. (Of course, no one really knows what the level base money needs to be to reach the target in the future, so raising base money during the period when spending is too low might still be sensible.)
Of course, from a Market Monetarist approach, the question is by how much base money needs to increase in order to get nominal GDP back to target. If increasing base money by purchasing assets with a zero nominal interest rate has little effect, and those assets are going to be purchased anyway, then this simply implies that a larger increase in base money is necessary. All of the zero yield securities must be purchased and then other securities must be purchased as well.
Once the central bank is purchasing securities that don't have a zero yield, then the Barro and Yeager arguments no longer apply. The securities being purchased would not be perfect substitutes for base money, and so, the quantity of money would be increasing. There is not a shortage of those securities at their current positive yield, and as long as the purchases don't go so far as to create a shortage, there is no shortage to be shifted to an increase in the demand for money. The quantity of money rises, without any self-inflicted increase in the demand to hold money, and so the underlying excess demand for money is relieved.
This suggests that a sufficient expansion of base money now, which would require that on the margin assets be purchased that have yields greater than zero, would result in an increase in spending on output. The shortfall in spending on output could be reduced, and spending on output could return to target more quickly. Since expectations that spending on output will be higher sooner should result in more spending due to the effect expectations of future spending, this would be a second avenue by which the extra increases in base money would help raise spending.
What about the Woodford account? The demand for money is supposedly perfectly elastic at an interest rate of zero. But the interest rate for the securities the central bank is purchasing is not zero. Of course, Woodford's assumption is that the interest rate on those securities would be reduced by the central bank's purchases, lowering the opportunity cost of holding money, so that in the end, the quantity of money demanded rises to match the added quantity. From Woodford's interest-centric view, what is happening is that once the interest rate on one particular asset is driven to zero, then the interest rate on some other asset is lowered. If that one is also driven to zero, then another is lowered.
This process of driving one interest rate and then another to zero occur until the central bank has lowered the interest rate on all assets it can buy to zero. This is very much parallel to the "upper bound" for base money targeting. Once the central bank has purchased every asset it is permitted to purchase, then no further increase in base money is possible.
This points to another institutionally specific element of new Keynesian thinking. Central banks like to target interest rates, and new Keynesian economists narrowly focuses on interest rate targeting. Central banks like to purchase safe money market instruments. New Keynesian economics focuses on just such instruments.
Market Monetarists insist that the problem of a shortfall of spending from target is that the quantity of money is less than the demand to hold money. If small modest increases in base money used to purchase safe money market instruments are inadequate to expand the quantity of money enough to meet the demand under some circumstances, then heroic purchases of longer term and riskier securities should be used.
Woodford argues that heroic open market purchases of assets with interest rates above zero will not be effective. They will not raise spending closer to target or allow it to reach the target sooner. The effect is the same as if base money simply expands (or contracts) whatever amount is needed to keep the policy rate equal to zero until spending returns to target. But what does this mean with base money targeting? Spending will not rise closer to target or reach the target sooner due to any increase in base money beyond the level needed to keep spending on output at target when the interest rate on the asset the central bank usually buys rises above zero.
But surely this depends on what particular bond the central bank typically buys. Suppose the central bank usually buys government bonds with maturities between 2 and 5 years. Now suppose further that the interest rate on 3 month bill T-bills falls to zero. The central bank never buys any of those anyway. It continues to purchase the longer term bonds as usual. Is it impossible to relieve an excess demand for base money because the 3 month T-bills has a zero yield? Suppose instead that the central bank targets a yield for bonds with 4 years to go. Can it not continue to make adjustments in that interest rate even if the 3 month T-bills is zero?
Now, suppose that the Treasury never sold T-bills at all. Whatever market conditions would have caused that yield to fall to zero still occur, but there are no T-bills. Of course, the central bank doesn't buy nonexistent T-bills. It buys government bonds with 3 to 5 year maturities. Do market conditions that would create a zero yield on nonexistent T-bills make increases in base money ineffective?
Suppose now that there is no national debt at all and the central bank only purchases private securities that bear default risk. (Perhaps the liabilities of the central bank should count as government debt.) Suppose market conditions that would have caused the interest rate on T-bills to fall to zero occurs, but there are no government bonds at all. Does the central bank's expansion of base money by purchasing risky private securities become ineffective?
Suppose one business finds that it is able to borrow at a zero nominal rate. Does this make increases in base money by the central bank when it purchases bonds issued by other firms ineffective?
Now consider a cashless (or currency-less) monetary regime. All payments are made by transfer of bank balances. The interest rate on balances at the central bank are slightly less than the interest rate on short term government debt. If the yield on that debt falls, then the interest rate the banks receive on their balances at the central bank fall in proportion.
There is no zero nominal bound on interest rates. Interest can be as negative as necessary to keep spending on output on target. What is it about these negative interest rates on short term government bonds that solves the problem? If we have an economy that only has consumer goods and everyone is the same, then the only way for spending to rise is for everyone to consume more now. The lower interest rate on government debt, central bank balances, money balances held in banks, and everything else, have to deter desired saving in the current period and so raise desired consumption. There is nothing else.
But in reality, when there are different people, then a sufficiently negative interest rate on government bonds would motivate households that save to lend more to households that dissave. This involves bearing more risk, on the assumption that households that dissave are more likely to default than the government. Further, in an economy with investment, then some of those holding government bonds might sell their bonds due to the low yields and spend the money on capital goods now in order to get consumer goods in the future. Others might lend to firms that will purchase capital goods rather than lend to the government. Of course, this involves more risk, assuming that investments in capital goods are more likely to fail than the probability the government will default on its debt.
Something happens that makes market-clearing interest rates on short and safe securities negative. If the central bank's own liabilities are short and safe and it keeps the interest rate on them at zero, then no interest rates are going to fall to the market clearing rates because people will hold the liabilities of the central bank instead. But if the central bank expands the quantity of base money enough, it will bear the risk and so all of the impact of hypothetically negative nominal interest rates on short and safe assets in terms of motivating purchases of capital goods or debt financed consumer goods are still obtained. Only by assuming those are nonexistent, that is, in a consumer good only, representative agent economy, will it look like the expansions in base money have no effect.
Finally, suppose that all hand-to-hand currency is issued by risky private firms. The central bank only offers deposit accounts to banks, and the banks using the central bank only issue deposit accounts. The interest rate on short-and-safe government bonds can turn negative. The interest rates on bank deposits can turn negative. The zero-nominal interest rate on privately-issued currency is irrelevant, because it is quite risky. It is only held for its "liquidity" yield. Presumably criminals still use it as do people making small face-to-face transactions. This currency is not base money. It is like travelers checks issued by fly-by-night firms.
If the central bank makes open market purchases of T-bills, and drives their prices above par and their yields below zero, then the interest rates paid on reserves and the interest rates banks pay on deposits are also driven below zero. Saving is less attractive and consumption more attractive, but also there is a motivation to bear more risk by lending money to dissavers and financing capital goods either through lending to firms or selling of short and safe securities to buy capital goods.
Now, suppose the risk of these investments in capital goods or loans to dissaving households approaches that of the risky hand-to-hand currency. Then, there will be an increase in the demand for that currency and the firms issuing it presumably will have less rather than more incentive to issue it. Currency becomes more difficult to obtain for small face-to-face transactions and criminal activities. To keep spending on target, other sorts of spending must expand. (Presumably the criminal activities weren't measured anyway.) This would be spending undertaken by check or electronic payment.
Of course, if all deposits were made redeemable in this hand-to-hand currency, even though it is risky and issued by private firms, then it would be impossible to expand spending by check or electronic payment to offset the reduction in spending of hand-to-hand currency. On the contrary, the shortage of hand-to-hand currency would generate a shortage of deposits and reduced spending on output too.
So, central banks are governmental institutions issuing hand-to-hand currency that is safe, short, and has a zero nominal yield. All other money is redeemable in that currency. The central banks like to make periodic changes in money market interest rates implemented by modest changes in purchase of short and safe assets--generally short term to maturity government bonds. If there is a large increase in the demand for those short and safe government bonds, then central banks must do something different to keep spending growing at a slow and steady growth path. Bearing more risk itself to motivate more investment in capital goods and more consumer loans is possible. This raises spending on output now, and when the market clearing rates on short and safe assets rise to zero, then it can go back to holding those and bear less risk. While it may appear undesirable for central banks to bear such risk, it is the least bad alternative if they insist on issuing zero-risk, zero-nominal-yield, hand-to-hand currency, or rather, in making all other money redeemable in their currency.
Thursday, September 6, 2012
Wren-Lewis on Zero Lower Bound "Denial."
Wren-Lewis claims that some macroeconomists are in Zero Lower Bound "Denial." This would be the view that monetary policy can overcome the problem of the zero nominal bound on interest rates so that fiscal policy is unnecessary.
In my view, we Market Monetarists are a tiny minority struggling against a conventional wisdom that the Fed is "out of ammunition." I do believe that an appropriate monetary regime can do all that is needed to provide for a stable macroeconomic environment. And I certainly don't favor expanding government spending at this time.
Wren-Lewis argues that while a nominal GDP level target can help the economy recover, it cannot do it alone and needs help from fiscal policy. His fundamental error here is to confuse what is really necessary today and what might be necessary under some hypothetical circumstance.
If the Fed had a nominal GDP level target all along, it is possible that it would have never found it necessary to lower its policy interest rate to near zero. Further, if it shifts to a nominal GDP level target now, it is possible that a rapid recovery in spending on output will promptly occur and that keeping the policy rate at zero would be undesirable.
Wren-Lewis imagines a hypothetical where there is a negative shock to demand so large that it results in the target for the policy interest rate to go to zero whether there is an inflation target or a nominal GDP level target. What he fails to make clear is that it takes a larger negative shock to create this problem with a nominal GDP level target than with an inflation target. Presumably, given any particular negative demand shock, the period of time for which the policy rate would be at the zero bound would be longer with an inflation target than with a nominal GDP level target.
However, it is true that even with nominal GDP level targeting, it is possible that there could be a negative demand shock that is so severe than the zero nominal bound issue would be relevant.
Wren Lewis also claims that a nominal GDP level target only helps a recovery now by imposing costs in the future. Spending now is higher than it would have been because inflation and real output are higher than would be desirable in the future.
There are two errors here. The first is to assume that inflation targeting is the "true" goal. In my view, the goal should be for spending on output to remain on a steady growth path. In other words, at any future date, the goal is for total spending on output to be at a certain level. The rate of change of nominal GDP and inflation are really irrelevant.
For example, suppose the target for spending on output for the first quarter of 2013 is $17 trillion. There is no "goal" for spending growth really. The proper growth rate depends on the difference between where nominal GDP happens to be and the target, along with the date of the current period. Now, it is true, that if the target is a 5 percent growth path, and the economy is currently at target level, then the appropriate growth rate is 5 percent. But if nominal GDP is below target, then that is the problem, and the greater than 5 percent of nominal GDP growth needed to get back to target is the solution.
As for inflation, that is just the wrong target. The price level should be the target level of nominal GDP divided by the potential output. The inflation rate then, should be based upon the difference between the current price level and that what the price level should be. Now, if nominal GDP is remains on target, then the inflation rate will equal the difference between the growth rate of the target path and the growth rate of potential output. But if nominal GDP is below target, and the current price level is below where it should be, then that the inflation rate is higher is not the problem, it is part of the solution.
Suppose the target were a constant price level of 100. If the price level remained on target, the inflation rate would remain zero. But if the price level falls to 98, then the "inflation" of prices from 98 back to 100 isn't a problem, it is the solution. Of course, a price level target is a disaster when there is a shock to supply, and that is why nominal GDP is the proper goal. If the target for nominal GDP is $10,000 billion and the target for next year is $10,300 billion, but actual nominal GDP is only $9,900 billion, that nominal GDP grows roughly 4 percent rather than 3 percent is the solution, not a problem.
While Wren-Lewis' argument that inflation will be too high is mistaken, he also argues that real output must be too high. Nominal GDP targeting supposedly involves forcing real output to rise beyond potential to get nominal GDP back to the target growth path. This follows (I think) from assuming that the policy interest rate this period causes an output gap in the next period which causes inflation to shift from its expected level in the period after that. The assumption, then, is that to get nominal GDP back up to trend, the inflation rate will have to be higher, and to get that to happen, real output must rise above potential.
I am sure that there are some purposes for which models where the policy rate this period causes output gaps in the next period and then changes in inflation in the the next period are useful, but I don't think they are realistic. A more plausible neo-Wicksellian story would be that today's policy rate has little impact, but expected future policy rates influence spending on output in the current period and future periods. And then spending on output during this and future periods determine both prices and real output in this and future periods.
If spending on output is too low this period, the likely result will be that both prices and production are too low. When spending rises more quickly to return to target, both output and prices rise more quickly. While it is certainly possible that output would rise too much and prices too little, (or vice versa,) it not necessary.
Wren-Lewis also is suffers from a fundamental confusion between something being "less" effective and something not being effective. He writes:
So, let's say that a temporary purchase of $20 billion of T-bills yielding 5 percent causes $200 billion of increased spending on output. A temporary purchase of T-bonds is an order of magnitude less effective. That is, it will only increase spending on output by a tenth as much. A $20 billion temporary purchase of Treasury bonds yielding 2 percent will only raise spending on output by $20 billion.
Now, let us suppose that nominal GDP is $ 2 trillion below target. If T-bills were yielding 5 percent, then it would be necessary to purchase only $200 billion worth of them. The yield on T-bills and other money market rates would presumably be slightly lower for a time and, in fact, the new Keynesian approach to describing this policy would be that the central bank should lower the yields on those money market instruments that amount. Whether these T-bill purchases are permanent or temporary isn't really specified. Anyway, if T-bill rates are zero, so that purchases of T-bills make no difference, and instead Treasury bonds yielding 2 percent purchased instead, then the fact that they are an order of magnitude less effective means that $2 trillion of Treasury bonds must be purchased.
So what?
As long as temporary open market operates in long term assets are effective at all, then the question of "effectiveness" just tells us how much must be purchased. Under "normal" circumstances, only a small amount of T-bills must be purchased, but under unusual circumstances, a large amount of Treasury bonds must be purchased? How does this make monetary policy ineffective?
I think that expanding the government's budget deficit and national debt reduces national savings and raises the natural interest rate. Alternatively, have the government expand its national debt and sell lots of T-bills will tend to lower the equilibrium price and raise the equilibrium yield on T-bills. However, I think the government is already spending too much, and it is also borrowing too much. And so, expanding government spending is a bad idea, and funding more of the existing spending by borrowing rather than taxes is also a bad idea.
Fortunately, it is possible that a nominal GDP level target might make it possible for the central bank to return to targeting short term interest rates using purchases and sales of short term assets. And even if it is not enough, the central bank can always purchase other sorts of assets and create spending even if the interest rates on T-bills and other short and safe assets remain near zero.
When the central bank has purchased all the assets that it can, then it is time to look to more radical remedies. Perhaps deficit-financed tax cuts. Or for an even more extreme example, government spending programs justified by there being no opportunity cost. But I would instead go with privatization of hand-to-hand currency, ending the zero bound on nominal interest rates.
In my view, we Market Monetarists are a tiny minority struggling against a conventional wisdom that the Fed is "out of ammunition." I do believe that an appropriate monetary regime can do all that is needed to provide for a stable macroeconomic environment. And I certainly don't favor expanding government spending at this time.
Wren-Lewis argues that while a nominal GDP level target can help the economy recover, it cannot do it alone and needs help from fiscal policy. His fundamental error here is to confuse what is really necessary today and what might be necessary under some hypothetical circumstance.
If the Fed had a nominal GDP level target all along, it is possible that it would have never found it necessary to lower its policy interest rate to near zero. Further, if it shifts to a nominal GDP level target now, it is possible that a rapid recovery in spending on output will promptly occur and that keeping the policy rate at zero would be undesirable.
Wren-Lewis imagines a hypothetical where there is a negative shock to demand so large that it results in the target for the policy interest rate to go to zero whether there is an inflation target or a nominal GDP level target. What he fails to make clear is that it takes a larger negative shock to create this problem with a nominal GDP level target than with an inflation target. Presumably, given any particular negative demand shock, the period of time for which the policy rate would be at the zero bound would be longer with an inflation target than with a nominal GDP level target.
However, it is true that even with nominal GDP level targeting, it is possible that there could be a negative demand shock that is so severe than the zero nominal bound issue would be relevant.
Wren Lewis also claims that a nominal GDP level target only helps a recovery now by imposing costs in the future. Spending now is higher than it would have been because inflation and real output are higher than would be desirable in the future.
There are two errors here. The first is to assume that inflation targeting is the "true" goal. In my view, the goal should be for spending on output to remain on a steady growth path. In other words, at any future date, the goal is for total spending on output to be at a certain level. The rate of change of nominal GDP and inflation are really irrelevant.
For example, suppose the target for spending on output for the first quarter of 2013 is $17 trillion. There is no "goal" for spending growth really. The proper growth rate depends on the difference between where nominal GDP happens to be and the target, along with the date of the current period. Now, it is true, that if the target is a 5 percent growth path, and the economy is currently at target level, then the appropriate growth rate is 5 percent. But if nominal GDP is below target, then that is the problem, and the greater than 5 percent of nominal GDP growth needed to get back to target is the solution.
As for inflation, that is just the wrong target. The price level should be the target level of nominal GDP divided by the potential output. The inflation rate then, should be based upon the difference between the current price level and that what the price level should be. Now, if nominal GDP is remains on target, then the inflation rate will equal the difference between the growth rate of the target path and the growth rate of potential output. But if nominal GDP is below target, and the current price level is below where it should be, then that the inflation rate is higher is not the problem, it is part of the solution.
Suppose the target were a constant price level of 100. If the price level remained on target, the inflation rate would remain zero. But if the price level falls to 98, then the "inflation" of prices from 98 back to 100 isn't a problem, it is the solution. Of course, a price level target is a disaster when there is a shock to supply, and that is why nominal GDP is the proper goal. If the target for nominal GDP is $10,000 billion and the target for next year is $10,300 billion, but actual nominal GDP is only $9,900 billion, that nominal GDP grows roughly 4 percent rather than 3 percent is the solution, not a problem.
While Wren-Lewis' argument that inflation will be too high is mistaken, he also argues that real output must be too high. Nominal GDP targeting supposedly involves forcing real output to rise beyond potential to get nominal GDP back to the target growth path. This follows (I think) from assuming that the policy interest rate this period causes an output gap in the next period which causes inflation to shift from its expected level in the period after that. The assumption, then, is that to get nominal GDP back up to trend, the inflation rate will have to be higher, and to get that to happen, real output must rise above potential.
I am sure that there are some purposes for which models where the policy rate this period causes output gaps in the next period and then changes in inflation in the the next period are useful, but I don't think they are realistic. A more plausible neo-Wicksellian story would be that today's policy rate has little impact, but expected future policy rates influence spending on output in the current period and future periods. And then spending on output during this and future periods determine both prices and real output in this and future periods.
If spending on output is too low this period, the likely result will be that both prices and production are too low. When spending rises more quickly to return to target, both output and prices rise more quickly. While it is certainly possible that output would rise too much and prices too little, (or vice versa,) it not necessary.
Wren-Lewis also is suffers from a fundamental confusion between something being "less" effective and something not being effective. He writes:
Monetary policy that involves temporarily creating money to buy financial assets is of an order less effective and reliable than conventional monetary policy, or fiscal policy.
So, let's say that a temporary purchase of $20 billion of T-bills yielding 5 percent causes $200 billion of increased spending on output. A temporary purchase of T-bonds is an order of magnitude less effective. That is, it will only increase spending on output by a tenth as much. A $20 billion temporary purchase of Treasury bonds yielding 2 percent will only raise spending on output by $20 billion.
Now, let us suppose that nominal GDP is $ 2 trillion below target. If T-bills were yielding 5 percent, then it would be necessary to purchase only $200 billion worth of them. The yield on T-bills and other money market rates would presumably be slightly lower for a time and, in fact, the new Keynesian approach to describing this policy would be that the central bank should lower the yields on those money market instruments that amount. Whether these T-bill purchases are permanent or temporary isn't really specified. Anyway, if T-bill rates are zero, so that purchases of T-bills make no difference, and instead Treasury bonds yielding 2 percent purchased instead, then the fact that they are an order of magnitude less effective means that $2 trillion of Treasury bonds must be purchased.
So what?
As long as temporary open market operates in long term assets are effective at all, then the question of "effectiveness" just tells us how much must be purchased. Under "normal" circumstances, only a small amount of T-bills must be purchased, but under unusual circumstances, a large amount of Treasury bonds must be purchased? How does this make monetary policy ineffective?
I think that expanding the government's budget deficit and national debt reduces national savings and raises the natural interest rate. Alternatively, have the government expand its national debt and sell lots of T-bills will tend to lower the equilibrium price and raise the equilibrium yield on T-bills. However, I think the government is already spending too much, and it is also borrowing too much. And so, expanding government spending is a bad idea, and funding more of the existing spending by borrowing rather than taxes is also a bad idea.
Fortunately, it is possible that a nominal GDP level target might make it possible for the central bank to return to targeting short term interest rates using purchases and sales of short term assets. And even if it is not enough, the central bank can always purchase other sorts of assets and create spending even if the interest rates on T-bills and other short and safe assets remain near zero.
When the central bank has purchased all the assets that it can, then it is time to look to more radical remedies. Perhaps deficit-financed tax cuts. Or for an even more extreme example, government spending programs justified by there being no opportunity cost. But I would instead go with privatization of hand-to-hand currency, ending the zero bound on nominal interest rates.
Wednesday, September 5, 2012
Cochrane on Woodford -- Too Depressing
John Cochrane commented on Woodford's recent call of nominal GDP level targeting. While he claims to agree with 99 percent of what Woodford wrote, he rejects nominal GDP level targeting. My guess is that the "99 percent" he agrees with are the parts that I found wildly implausible, particularly the "Wallace Equivalence" arguments that make all open market operations irrelevant.
But let's get to why Cochrane disagrees with Woodford, and Market Monetarists:
So much wrong here.
Now, I don't really want to say that the problem is that Treasury interest rates are too high, and I believe that an appropriate nominal GDP level target would result in higher real and nominal interest rates on Treasuries. But really, I just don't think the price and yield on Treasuries should be a first order concern. The right price is the market price.
I do believe that there is a problem of a lack of demand, and while I don't think that all demand is the same, I do think that monetary stimulus can fix a problem of too little demand. The argument is simple. If the demand for money increases, and the quantity of money fails to increase the same amount, then the demand for something else must fall, and that frequently includes newly produced output. If that situation arises, then increasing the quantity of money to match the increase in the demand to hold money will result in increased demand for other things, which will frequently include newly produced goods and services.
I fully recognize that an alternative solution to this problems is for the price level (including resource prices like wages) to fall enough so the real quantity of money rises to meet the demand. This will raise the real demand for other goods, frequently including newly-produced goods. I consider this a less desirable approach to correcting the imbalance between the quantity of money and the demand to hold it.
So, the alternatives are a higher nominal quantity of money and stable (or recovered) nominal spending on output with prices and wages continuing on their trend trajectories, or else, a nominal quantity of money that fails to meet the additional money demand, and so a lower trend growth path for nominal spending on output, along with lower growth paths for prices and wages, allowing the real demand for output to recover. This second alternative is less desirable. Worse, having the Fed tell everyone that it is trying to keep prices rising 2 percent per year hardly helps coordinate the needed shift to a lower growth path for prices and wages.
Cochrane wants Woodford to distinguish between "nominal stimulus" and "real stimulus." (Tyler Cowen seems to like this distinction as well.) I personally live in a nominal world. I earn money and spend it. I don't live in a world of barter. In my world, all stimulus is nominal stimulus.
Consider the apple market. When the demand for apples increases, this can be understood as a horizontal shift in the demand curve. A larger physical quantity of apples is demanded at any given price. However, by multiplying that given price by the added quantity, that "real" increase in demand is given a nominal value--the desired added dollar spending on apples.
It is also possible to look at the added demand for apples as an upward shift in the demand curve. For any given quantity of apples, the amount that people are willing to pay increases. This could be simply a dollar amount (dollars per apple.) But it is also possible to multiply the given quantity of apples by the added amount people are willing to pay per apple and get the increase in dollar spending on apples.
An increase in the demand for apples is an increase the willingness to spend money on apples--at least in a money economy. To me, talk about "real stimulus," seems like imagining that the demand curve is perfectly inelastic. An increase in the "real" demand for apples shifts that vertical demand curve to the right. Nominal demand, on the other hand, doesn't matter. That is just moving the price up and down the vertical demand curve. In other words, it makes next to no sense to economists who internalize the law of demand, and certainly would require some account of "supply" to make sense of the changing price in face of a perfectly inelastic demand.
Of course, with an isolated individual who uses labor to grow apples, what counts is the value of leisure, the value of apples, and the productivity of labor. "Real" stimulus must involve being willing to work more (or harder) to get more apples. Demanding more apples is supplying more labor (and demanding less leisure.) If the "real" demand for apples decreases, that is supplying less labor (and demanding more leisure.)
Putting numbers on the hours worked or the apples (wages and prices) would be pointless and irrelevant. Nominal stimulus would pointless. It is only "real stimulus" that counts. Unfortunately, imagining that a real market economy is more or less the same thing as an isolated individual is just about as absurd as perfectly inelastic demand.
Unfortunately, while I think of economics as an effort to understand a market economy--millions of ordinary people exchanging goods and resources for money, others think of economics as people doing incredibly complicated optimization problems.
Now, what about the Phillips curve? Cochrane sees the Phillips curve as higher inflation causing higher output. And he has doubts about it. According to him, we don't want more inflation for its own sake, but only for it to "goose" output. And he is skeptical that deliberately causing inflation raises output. He further argues that higher expected inflation should raise inflation while reducing output and employment.
What a mess. And sadly, there are probably many economists who suffer from just these confusions. The positive relationship between prices and output comes from increased spending on output. Firms respond to the added demand by both raising prices and producing more output. It is not that the higher prices are causing higher output or that the increased output is causing higher prices. It is the additional nominal stimulus that is causing both the higher output and the higher prices.
In a growing economy, with growing spending, growing output, and inflation, then more rapid growth in spending on output results in more rapid growth in real output and higher inflation. A higher inflation rate is associated with more rapid output growth, but the higher inflation rate doesn't cause the higher real output growth, nor does the more rapid real output growth cause higher inflation. Both are caused by a third thing, the more rapid growth in spending on output.
If, instead, we take spending on output as given, and consider what happens with higher prices, then the result is lower real sales and reduced production. Similarly, if we consider firms expanding real output, given spending on output, then they must lower their prices to sell the added output.
With a growing economy, with growing spending, growing output, and inflation, if we take the growth of spending as constant, then more rapid inflation results in slower growth in real output. Firms raise their prices more quickly, real expenditures grow more slowly and firms expand their production more slowly. And if we assume firms expand their production more quickly, if spending on output is growing at a constant rate, then they can only sell that added output if they raise their prices more slowly.
So, given spending on output, higher inflation causes lower real growth. And higher real growth causes lower inflation. Again, it is more rapid growth in spending on output that causes higher inflation and higher real growth.
Now, it is likely that higher expected inflation will cause firms to raise their prices more quickly--cause higher inflation. And given growth of spending on output, that will result in slower growth in output.
But the actual scenario is more rapid growth in spending on output.
Suppose spending on output is expected to grow 4%, and real output is expected to grow 2%, and inflation is expected to be 2%. Then due to the new policy of nominal GDP level targeting, spending on output is expected to grow 7%. And that causes expected inflation to rise to 3%. Suppose this causes actual inflation to rise to 3% too. If spending on output actually grows by more than 5%, then while inflation will be higher, so will real output growth. If spending on output grows 7% as expected, then real output would grow 4%.
Cochrane makes the following claim:
Well, apparently Cochrane didn't read Woodford's paper. While Woodford doesn't really speak to whether or not keeping nominal GDP on the trend would have been possible he certainly says that by committing to return nominal GDP to trend in the future, it will raise spending on output today.
Here is the relevant section of Woodford's paper:
Woodford clearly says that expected future levels of inflation and real income effect expenditures on output today. So, if the central bank can influence expected future inflation and real income, it can impact nominal expenditures now. That there is a zero-bound today imposes little constraint. By committing to sufficiently high spending on output in the future, when there is no zero bound, it could raise spending today to any level.
Of course, it may be that the future level of spending on output needed to get current spending up to the trend now would be higher than that trend extended into the future, but we don't know that. In other words, as many Market Monetarists often point out, a commitment to raise nominal GDP back to the previous trend might require short and safe interest rates to rise well above zero immediately.
If we consider the reverse situation, where the central bank is committing to cause some massive drop in nominal GDP in a a few years, its is very likely that the result would be nominal interest rates on short and safe assets being pegged at zero, and current spending on output collapsing. If the proposed massive deflation in a few years was recognized to be insane and dropped, then nominal interest rates as well as current spending on output should at least partially recover.
More to the point, if the Fed had always following a nominal GDP level target, there is good reason to believe that any shortfall from that trend would have been smaller. And by shifting to that target even now, a stronger recovery in spending on output could be generated. And while a recovery in spending on output will probably result in somewhat higher inflation (both realized and expected,) it will likely also cause a more rapid recovery in production and employment.
Of course, if Cochrane wants to argue that if increased output and employment were possible just because there was an expansion in spending on output, the prices and wages would already be lower, so that real expenditure would have already increased enough to result in the needed volume of sales of output, then he should stay that. I would still favor keeping nominal GDP on a slow steady growth path, even if reduced productivity resulted in more rapid inflation of output prices. And so, if nominal GDP falls below target, I would favor getting it back up to the target growth path-as soon as possible--even if that was associated with higher inflation.
Nominal GDP targeting is a regime. It isn't just an excuse to cause inflation, "goose output," and reduce real interest rates on Treasuries.
But let's get to why Cochrane disagrees with Woodford, and Market Monetarists:
Treasury rates are at 50 year lows. The 10 year Treasury rate is 1.5%. At 2% inflation, they pay you half a percent per year to borrow at that rate. Yes, the economy is in the toilet, but surely too-high Treasury interest rates are not the crucial economic problem right now.
So the case for "stimulus" must be that some other, unstated lack of "demand" is the problem, and that all "demand" is the same so that monetary "stimulus" will cure that problem. I disagree on that one.
Mike's enthusiasm for deliberate inflation is even more puzzling to me. Mike uses the word "stimulus," never differentiating between real and nominal stimulus. Surely, we don't want to cook up some inflation just for its own sake -- we want to cook up some inflation because we think it will goose output. But why? Why especially will increasing expected inflation help? Because that is the aim of all the policies under discussion here -- promising to keep rates low even once inflation rises, adopting "nominal GDP targets," helicopter drops, or similar policies such as raising the inflation target.
I don't put much faith in Phillips curves to start with -- the idea that deliberate inflation raises output. I put less faith in the idea floating around Jackson hole that a little inflation will set us permanently back on the trend line, not just be a little sugar rush and then back to sclerosis.
But it's a rare Phillips curve in which raising expected inflation is a good thing. It just gives you more inflation, with if anything less output and employment.
So much wrong here.
Now, I don't really want to say that the problem is that Treasury interest rates are too high, and I believe that an appropriate nominal GDP level target would result in higher real and nominal interest rates on Treasuries. But really, I just don't think the price and yield on Treasuries should be a first order concern. The right price is the market price.
I do believe that there is a problem of a lack of demand, and while I don't think that all demand is the same, I do think that monetary stimulus can fix a problem of too little demand. The argument is simple. If the demand for money increases, and the quantity of money fails to increase the same amount, then the demand for something else must fall, and that frequently includes newly produced output. If that situation arises, then increasing the quantity of money to match the increase in the demand to hold money will result in increased demand for other things, which will frequently include newly produced goods and services.
I fully recognize that an alternative solution to this problems is for the price level (including resource prices like wages) to fall enough so the real quantity of money rises to meet the demand. This will raise the real demand for other goods, frequently including newly-produced goods. I consider this a less desirable approach to correcting the imbalance between the quantity of money and the demand to hold it.
So, the alternatives are a higher nominal quantity of money and stable (or recovered) nominal spending on output with prices and wages continuing on their trend trajectories, or else, a nominal quantity of money that fails to meet the additional money demand, and so a lower trend growth path for nominal spending on output, along with lower growth paths for prices and wages, allowing the real demand for output to recover. This second alternative is less desirable. Worse, having the Fed tell everyone that it is trying to keep prices rising 2 percent per year hardly helps coordinate the needed shift to a lower growth path for prices and wages.
Cochrane wants Woodford to distinguish between "nominal stimulus" and "real stimulus." (Tyler Cowen seems to like this distinction as well.) I personally live in a nominal world. I earn money and spend it. I don't live in a world of barter. In my world, all stimulus is nominal stimulus.
Consider the apple market. When the demand for apples increases, this can be understood as a horizontal shift in the demand curve. A larger physical quantity of apples is demanded at any given price. However, by multiplying that given price by the added quantity, that "real" increase in demand is given a nominal value--the desired added dollar spending on apples.
It is also possible to look at the added demand for apples as an upward shift in the demand curve. For any given quantity of apples, the amount that people are willing to pay increases. This could be simply a dollar amount (dollars per apple.) But it is also possible to multiply the given quantity of apples by the added amount people are willing to pay per apple and get the increase in dollar spending on apples.
An increase in the demand for apples is an increase the willingness to spend money on apples--at least in a money economy. To me, talk about "real stimulus," seems like imagining that the demand curve is perfectly inelastic. An increase in the "real" demand for apples shifts that vertical demand curve to the right. Nominal demand, on the other hand, doesn't matter. That is just moving the price up and down the vertical demand curve. In other words, it makes next to no sense to economists who internalize the law of demand, and certainly would require some account of "supply" to make sense of the changing price in face of a perfectly inelastic demand.
Of course, with an isolated individual who uses labor to grow apples, what counts is the value of leisure, the value of apples, and the productivity of labor. "Real" stimulus must involve being willing to work more (or harder) to get more apples. Demanding more apples is supplying more labor (and demanding less leisure.) If the "real" demand for apples decreases, that is supplying less labor (and demanding more leisure.)
Putting numbers on the hours worked or the apples (wages and prices) would be pointless and irrelevant. Nominal stimulus would pointless. It is only "real stimulus" that counts. Unfortunately, imagining that a real market economy is more or less the same thing as an isolated individual is just about as absurd as perfectly inelastic demand.
Unfortunately, while I think of economics as an effort to understand a market economy--millions of ordinary people exchanging goods and resources for money, others think of economics as people doing incredibly complicated optimization problems.
Now, what about the Phillips curve? Cochrane sees the Phillips curve as higher inflation causing higher output. And he has doubts about it. According to him, we don't want more inflation for its own sake, but only for it to "goose" output. And he is skeptical that deliberately causing inflation raises output. He further argues that higher expected inflation should raise inflation while reducing output and employment.
What a mess. And sadly, there are probably many economists who suffer from just these confusions. The positive relationship between prices and output comes from increased spending on output. Firms respond to the added demand by both raising prices and producing more output. It is not that the higher prices are causing higher output or that the increased output is causing higher prices. It is the additional nominal stimulus that is causing both the higher output and the higher prices.
In a growing economy, with growing spending, growing output, and inflation, then more rapid growth in spending on output results in more rapid growth in real output and higher inflation. A higher inflation rate is associated with more rapid output growth, but the higher inflation rate doesn't cause the higher real output growth, nor does the more rapid real output growth cause higher inflation. Both are caused by a third thing, the more rapid growth in spending on output.
If, instead, we take spending on output as given, and consider what happens with higher prices, then the result is lower real sales and reduced production. Similarly, if we consider firms expanding real output, given spending on output, then they must lower their prices to sell the added output.
With a growing economy, with growing spending, growing output, and inflation, if we take the growth of spending as constant, then more rapid inflation results in slower growth in real output. Firms raise their prices more quickly, real expenditures grow more slowly and firms expand their production more slowly. And if we assume firms expand their production more quickly, if spending on output is growing at a constant rate, then they can only sell that added output if they raise their prices more slowly.
So, given spending on output, higher inflation causes lower real growth. And higher real growth causes lower inflation. Again, it is more rapid growth in spending on output that causes higher inflation and higher real growth.
Now, it is likely that higher expected inflation will cause firms to raise their prices more quickly--cause higher inflation. And given growth of spending on output, that will result in slower growth in output.
But the actual scenario is more rapid growth in spending on output.
Suppose spending on output is expected to grow 4%, and real output is expected to grow 2%, and inflation is expected to be 2%. Then due to the new policy of nominal GDP level targeting, spending on output is expected to grow 7%. And that causes expected inflation to rise to 3%. Suppose this causes actual inflation to rise to 3% too. If spending on output actually grows by more than 5%, then while inflation will be higher, so will real output growth. If spending on output grows 7% as expected, then real output would grow 4%.
Cochrane makes the following claim:
Now, let's be clear what a nominal GDP target is and is and is not. Many people (and a few persistent commenters on this blog!) urge nominal GDP targeting by looking at a graph like this and saying "see, if the Fed had kept nominal GDP on trend, we wouldn't have had such a huge recession. Sure, part of it might have been more inflation, but surely part of a steady nominal GDP would have been less recession." This is NOT what Mike is talking about.
Mike recognizes, as I do, that the Fed can do nothing more to raise nominal GDP today. Rates are at zero. The Fed has did what it could. The trend line was not achievable.
Well, apparently Cochrane didn't read Woodford's paper. While Woodford doesn't really speak to whether or not keeping nominal GDP on the trend would have been possible he certainly says that by committing to return nominal GDP to trend in the future, it will raise spending on output today.
Here is the relevant section of Woodford's paper:
Standard New Keynesian models imply that a higher level of expected real incomeor inflation in the future creates incentives for greater real expenditure and larger price increases now; but in the case of a conventional interest-rate reaction function for the central bank, short-term interest rates should increase, and the disincentive that this provides to current expenditure will attenuate (without completely eliminating) the sensitivity of current conditions to expectations. If nominal interest rates instead remain unchanged, the degree to which higher expected real income and inflation later produce higher real income and inflation now is amplified. If the situation is expected to persist for a period of time, the degree of amplification should increase exponentially. Hence it is precisely when the interest-rate lower bound is expected to be a binding constraint for some time to come that expectations about the conduct of policy after the constraint ceases to bind should have a particularly large effect on current economic conditions
Woodford clearly says that expected future levels of inflation and real income effect expenditures on output today. So, if the central bank can influence expected future inflation and real income, it can impact nominal expenditures now. That there is a zero-bound today imposes little constraint. By committing to sufficiently high spending on output in the future, when there is no zero bound, it could raise spending today to any level.
Of course, it may be that the future level of spending on output needed to get current spending up to the trend now would be higher than that trend extended into the future, but we don't know that. In other words, as many Market Monetarists often point out, a commitment to raise nominal GDP back to the previous trend might require short and safe interest rates to rise well above zero immediately.
If we consider the reverse situation, where the central bank is committing to cause some massive drop in nominal GDP in a a few years, its is very likely that the result would be nominal interest rates on short and safe assets being pegged at zero, and current spending on output collapsing. If the proposed massive deflation in a few years was recognized to be insane and dropped, then nominal interest rates as well as current spending on output should at least partially recover.
More to the point, if the Fed had always following a nominal GDP level target, there is good reason to believe that any shortfall from that trend would have been smaller. And by shifting to that target even now, a stronger recovery in spending on output could be generated. And while a recovery in spending on output will probably result in somewhat higher inflation (both realized and expected,) it will likely also cause a more rapid recovery in production and employment.
Of course, if Cochrane wants to argue that if increased output and employment were possible just because there was an expansion in spending on output, the prices and wages would already be lower, so that real expenditure would have already increased enough to result in the needed volume of sales of output, then he should stay that. I would still favor keeping nominal GDP on a slow steady growth path, even if reduced productivity resulted in more rapid inflation of output prices. And so, if nominal GDP falls below target, I would favor getting it back up to the target growth path-as soon as possible--even if that was associated with higher inflation.
Nominal GDP targeting is a regime. It isn't just an excuse to cause inflation, "goose output," and reduce real interest rates on Treasuries.
Monday, September 3, 2012
Woodford at Jackson Hole
Key new Keynesian macroeconomist, Micheal Woodford, was invited to give a paper at the Federal Reserve's Jackson Hole conference. The paper, "Methods of Policy Accommodation at the Interest-Rate Lower Bound," is very heartening from a Market Monetarist perspective. Most importantly, he gives support to a nominal GDP level target. He writes:
Woodford considers a nominal GDP target inferior to a "gap adjusted price level target," which he has advocated before. While I was aware that Woodford was among those who argued that a price level target is superior to an inflation target when there are problems with the zero bound on interest rates, I was not aware that he advocated a gap adjusted price level target. "Gap adjusted" means that the target for the price level is adjusted based upon the "output gap." For example, an adverse supply shock would likely generate a negative output gap if the price level were held constant--real output would fall below potential output. Woodford's proposal would raise the target for the price level in proportion and then, even further, to close the output gap.
Why does Woodford suggest that a nominal GDP level target is superior? He gives exactly the reasons that I would give. Complicated formulas are difficult to explain to the public, and the output gap is subject to controversy. What is the exact value of potential output?
Woodford also takes a stand on the "What would Milton Friedman do" issue:
Woodford specifically rejects what I have been describing as the "the new Keynesian" approach to the zero bound problem--promising to keep interest rates at zero for an extended period of time. He insists that his position has always been that the policy rate should be kept at (or near) zero until some nominal economic value returns to target--such as nominal GDP. He even points out what many Market Monetarists have been saying, that talk about it being appropriate to keep interest rates near zero for an extended period of time can be immediately contractionary if this is taken to be a forecast that the economy will perform poorly for an extended period of time. In other words, the Federal Reserve's actual policy response has been useless if not counterproductive.
Woodford also avoids the "inflationista" bias that has run through too much new Keynesian commentary--particularly Krugman and DeLong. Woodford writes:
Woodford reviews empirical evidence that central bank talk about keeping short term rates low in the future causes market participants to expect interest rates to be low in the future. And while the opposite result, that no financial traders paid any mind to the Fed, would be pretty damning, there is still little evidence of the key new Keynesian and Market Monetarist claim--if the central bank promises to get nominal GDP back up to the target growth path in the future, this will stimulate spending now.
Woodford is very skeptical about the effectiveness of quantitative easing. While many Market Monetarists agree that it is pointless without fixing the target--a nominal GDP growth path--Woodford's unfortunate continued allegiance to interest rate targeting shows through in this section of his paper.
On a positive note, Woodford describes "pure" quantitative easing as a focus on the quantity of money--the central bank's liabilities--rather than the purchase of particular assets to direct credit to particular markets. Unfortunately, his brief explanation of the quantity theory suggests a fixed relationship between base money, some broader measure of the quantity of money, and then nominal expenditure on output, which is hardly essential to the quantity theory perspective and specifically rejected by Market Monetarists.
Interestingly, Woodford agrees that it is only permanent changes in base money that solve the "zero bound" problem. Or more exactly, it is the future levels of base money that do the trick. He then rather quickly insists that promises of future policy interest rates are the equivalent.
While there is nothing really new here, to what degree has the quantity theory of money always included an implicit assumption that quantity of money is "permanent." That is, the thought experiments are always about fixed levels or growth paths of some "quantity of money." And to what degree do "long and variable lags" and "adaptive expectations," assume that people are learning about new permanent growth paths?
Consider a traditional monetarist story. Velocity is fixed, and the money multiplier falls 10 percent. This will tend to cause the M2 money supply and nominal GDP to fall to a 10 percent lower growth path, with a likely recession of real output and then a gradual recovery with the price level permanently shifting to a 10 percent lower growth path. To avoid that recession in output and deflation of prices, the central bank would permanently expand base money, offsetting the decrease in the quantity of money, and ideally, leaving the M2 measure of the quantity of money on target. Nominal GDP would hardly be effected, and the recession in output and the deflation of prices would be avoided.
Now, what would happen if the central bank increased base money for a few months and then reduced it again? I think the traditional monetarist approach would suggest a postponement of the recession and deflation. On the other hand, if the money multiplier shifted again, then responding with a further change in base money would be entirely appropriate. Of course, with money supply targeting, it is some measure of the quantity of money that is being kept on a permanently stable growth path.
Market Monetarists don't favor any commitment to a future level of base money. It is rather that base money in the future will follow a path that will keep nominal GDP on target. How is this any better than promising that a policy interest rate will follow a future path consistent with keeping nominal GDP on target? At the very least, we don't have to worry about anyone thinking that a low nominal interest rate will cause falling nominal GDP and deflation enough to keep the real interest rate at equilibrium. While there are people who seem to think that the quantity of base money doesn't matter, I don't know of anyone who thinks that increases in base money lead to deflation.
Woodford argues that expanding base money by purchasing Treasury bills does not directly impact spending on output at the zero bound for interest rates, because the demand for base money becomes infinitely elastic. Only if the the quantity of base money is expected to be higher after the zero bound no long applies (after the natural real interest rate rises above zero, or rather becomes less negative than the expected inflation rate) will it be expected to expand spending on output in the future, and thereby impact spending now. This argument is consistent with Market Monetarist thinking.
However, Woodford also casts doubt on the effectiveness of expanding the quantity of money by the purchase of assets whose interest rates are not near zero. That is, assets that are not near perfect substitutes for money, so that their purchase simply changes the form rather than quantity of some more inclusive concept of money.
This is important because while Market Monetarists are very supportive of a target for a growth path for nominal GDP, most are not entirely comfortable with an expectations-only approach. What if no one was paying attention to the Federal Reserve's announcements? Or what if no one believes that anyone else is responding to them? If an increase in the quantity of base money would result in increased expenditure despite naive expectations, that would provide a reason to expect that the central bank can eventually get nominal GDP on target. It would be a little worrisome if it only can work if people believe it will work.
Here Woodford provides an accessible explanation of Wallace equivalence--a reason why open market operations of long term to maturity and risky bonds do not matter. Interestingly, in Woodford's account, there is a connection with Ricardian equivalence. Suppose the Fed purchases long term and risky securities and issues short and safe securities. This increases the risk of loss to the Fed and a risk of higher taxes for someone to make up for the shortfall of revenue going from the Fed to the Treasury. Investors will then want to hold fewer risky securities and more safe securities to offset the risk of these tax hikes. And so, investors increase their demand for these safe securities created by the Fed. There is no excess supply of them and so no motivation to spend more on output.
I don't believe in Ricardian equivalence and I find this even more incredible. That is, I don't believe that people expand saving enough to earn enough interest to pay the added taxes that they and their descendants will have to pay to cover interest and principal payments on government debt. And so, adding more complications through risk of loss by the central bank and so a risk of reduced transfers to the treasury and so higher taxes in the future seems even more implausible.
Aside from skepticism about the desirability of using models of God-like knowledge and optimization abilities to analyse real market economies, I think that another problem with the application of Woodford's analysis to truly heroic open market operations is corner solutions. When there are no other existing asset holders to sell to the Fed, and that the Fed is instead solely purchasing new issues, then it seems to me that spending on output increases.
Of course, the Fed has not been following a strategy of quantitative easing until a nominal target is hit, but rather making purchases of specific amounts of assets. Further the Fed has claimed that its purpose is to lower interest rates on those assets. Woodford argues that much of the effect on interest rates associated with these purchases was instead due to the Fed's promise to keep policy rates lower in the future. Interestingly, he also reports on evidence that the asset purchases resulted in higher expected inflation, which would tend to raise nominal interest rates.
Market Monetarists have long insisted that the purpose of quantitative easing should be to raise spending on output, and if this is expected to work, then the result would higher real and nominal interest rates. This would involve market participants selling more securities than the Fed is buying, using the funds to purchase capital goods or consumer goods.
Rather than making open ended commitments to quantitative easing combined with a nominal GDP level target, which is the Market Monetarist approach, Woodford seems committed to instead keeping the policy rate at zero until nominal GDP reaches the target level. Rather than depend on the effects of such quantitative easing to expand spending, he favors fiscal policy--tax cuts or expanded government spending--to directly raise demand for output.
While I don't doubt that government borrowing and spending in the context of a central bank willing to create whatever amount of money people want to borrow can expand spending, I don't think that is a wise policy when government spending and debt is already too high.
An alternative that I believe should be equally easy to explain to the general public, but that would preserve more of the advantages of the adjusted price-level target path, would be a criterion based on a nominal GDP target path, as proposed by Romer (2011) among others. Under this proposal, the FOMC would pledge to maintain the funds rate target at its lower bound as long as nominal GDP remains below a deterministic target path, representing the path that the FOMC would have kept it on (or near) if the interest-rate lower bound had not constrained policy since late 2008. Once nominal GDP again reaches the level of this path, it will be appropriate to raise nominal interest rates, to the level necessary to maintain a steadyHe also includes a diagram, very familiar to Market Monetarists, showing nominal GDP falling far below the trend of the Great Moderation. Options
growth rate of nominal GDP thereafter.
Woodford considers a nominal GDP target inferior to a "gap adjusted price level target," which he has advocated before. While I was aware that Woodford was among those who argued that a price level target is superior to an inflation target when there are problems with the zero bound on interest rates, I was not aware that he advocated a gap adjusted price level target. "Gap adjusted" means that the target for the price level is adjusted based upon the "output gap." For example, an adverse supply shock would likely generate a negative output gap if the price level were held constant--real output would fall below potential output. Woodford's proposal would raise the target for the price level in proportion and then, even further, to close the output gap.
Why does Woodford suggest that a nominal GDP level target is superior? He gives exactly the reasons that I would give. Complicated formulas are difficult to explain to the public, and the output gap is subject to controversy. What is the exact value of potential output?
Woodford also takes a stand on the "What would Milton Friedman do" issue:
On the one hand, the Fed’s ability to directly control broad monetary aggregates (the ones more directly related to nominal income in the way that Friedman assumed) can no longer be taken for granted, under current conditions; and on the other hand, modern methods of forecastHe even cites David Beckworth's blog posts on the "Milton Friedman" question. While it would have been nice to see some more general citation of Market Monetarism, rather than perhaps including us in the "others" who support nominal GDP targeting, perhaps Beckworth is the best citation right now, since as editor, his name appears on Boom and Bust Banking: The Causes and Cures of the Great Recession.
targeting make a commitment to the pursuit of a target defined in terms of variables that are not under the short-run control of the central bank more credible. Under these circumstances, a case can be made that a nominal GDP target path would remain true to Friedman’s fundamental concerns.
Woodford specifically rejects what I have been describing as the "the new Keynesian" approach to the zero bound problem--promising to keep interest rates at zero for an extended period of time. He insists that his position has always been that the policy rate should be kept at (or near) zero until some nominal economic value returns to target--such as nominal GDP. He even points out what many Market Monetarists have been saying, that talk about it being appropriate to keep interest rates near zero for an extended period of time can be immediately contractionary if this is taken to be a forecast that the economy will perform poorly for an extended period of time. In other words, the Federal Reserve's actual policy response has been useless if not counterproductive.
Woodford also avoids the "inflationista" bias that has run through too much new Keynesian commentary--particularly Krugman and DeLong. Woodford writes:
Standard New Keynesian models imply that a higher level of expected real income or inflation in the future creates incentives for greater real expenditure and larger price increases now; but in the case of a conventional interest-rate reaction function for the central bank, short-term interest rates should increase, and the disincentive that this provides to current expenditure will attenuate (without completely eliminating) the sensitivity of current conditions to expectations. If nominal interest rates instead remain unchanged, the degree to which higher expected real income and inflation later produce higher real income and inflation now is amplified. If the situation is expected to persist for a period of time, the degree of amplification should increase exponentially. Hence it is precisely when the interest-rate lower bound is expected to be a binding constraint for some time to come that expectations about the conduct of policy after the constraint ceases to bind should have a particularly large effect on current economic conditions — to the extent, that is, that it is possible to shift expectations about conditions that far in the future.Yes, the inflation is there, but the increase in real income is given equal weight. It is close to the Market Monetarist view that expectations of increased nominal expenditure in the future results in increased nominal expenditure in the present, with the precise mix of inflation and real output being secondary. And, of course, for purposes of human welfare, the more real output and less inflation the better.
Woodford reviews empirical evidence that central bank talk about keeping short term rates low in the future causes market participants to expect interest rates to be low in the future. And while the opposite result, that no financial traders paid any mind to the Fed, would be pretty damning, there is still little evidence of the key new Keynesian and Market Monetarist claim--if the central bank promises to get nominal GDP back up to the target growth path in the future, this will stimulate spending now.
Woodford is very skeptical about the effectiveness of quantitative easing. While many Market Monetarists agree that it is pointless without fixing the target--a nominal GDP growth path--Woodford's unfortunate continued allegiance to interest rate targeting shows through in this section of his paper.
On a positive note, Woodford describes "pure" quantitative easing as a focus on the quantity of money--the central bank's liabilities--rather than the purchase of particular assets to direct credit to particular markets. Unfortunately, his brief explanation of the quantity theory suggests a fixed relationship between base money, some broader measure of the quantity of money, and then nominal expenditure on output, which is hardly essential to the quantity theory perspective and specifically rejected by Market Monetarists.
Interestingly, Woodford agrees that it is only permanent changes in base money that solve the "zero bound" problem. Or more exactly, it is the future levels of base money that do the trick. He then rather quickly insists that promises of future policy interest rates are the equivalent.
While there is nothing really new here, to what degree has the quantity theory of money always included an implicit assumption that quantity of money is "permanent." That is, the thought experiments are always about fixed levels or growth paths of some "quantity of money." And to what degree do "long and variable lags" and "adaptive expectations," assume that people are learning about new permanent growth paths?
Consider a traditional monetarist story. Velocity is fixed, and the money multiplier falls 10 percent. This will tend to cause the M2 money supply and nominal GDP to fall to a 10 percent lower growth path, with a likely recession of real output and then a gradual recovery with the price level permanently shifting to a 10 percent lower growth path. To avoid that recession in output and deflation of prices, the central bank would permanently expand base money, offsetting the decrease in the quantity of money, and ideally, leaving the M2 measure of the quantity of money on target. Nominal GDP would hardly be effected, and the recession in output and the deflation of prices would be avoided.
Now, what would happen if the central bank increased base money for a few months and then reduced it again? I think the traditional monetarist approach would suggest a postponement of the recession and deflation. On the other hand, if the money multiplier shifted again, then responding with a further change in base money would be entirely appropriate. Of course, with money supply targeting, it is some measure of the quantity of money that is being kept on a permanently stable growth path.
Market Monetarists don't favor any commitment to a future level of base money. It is rather that base money in the future will follow a path that will keep nominal GDP on target. How is this any better than promising that a policy interest rate will follow a future path consistent with keeping nominal GDP on target? At the very least, we don't have to worry about anyone thinking that a low nominal interest rate will cause falling nominal GDP and deflation enough to keep the real interest rate at equilibrium. While there are people who seem to think that the quantity of base money doesn't matter, I don't know of anyone who thinks that increases in base money lead to deflation.
Woodford argues that expanding base money by purchasing Treasury bills does not directly impact spending on output at the zero bound for interest rates, because the demand for base money becomes infinitely elastic. Only if the the quantity of base money is expected to be higher after the zero bound no long applies (after the natural real interest rate rises above zero, or rather becomes less negative than the expected inflation rate) will it be expected to expand spending on output in the future, and thereby impact spending now. This argument is consistent with Market Monetarist thinking.
However, Woodford also casts doubt on the effectiveness of expanding the quantity of money by the purchase of assets whose interest rates are not near zero. That is, assets that are not near perfect substitutes for money, so that their purchase simply changes the form rather than quantity of some more inclusive concept of money.
This is important because while Market Monetarists are very supportive of a target for a growth path for nominal GDP, most are not entirely comfortable with an expectations-only approach. What if no one was paying attention to the Federal Reserve's announcements? Or what if no one believes that anyone else is responding to them? If an increase in the quantity of base money would result in increased expenditure despite naive expectations, that would provide a reason to expect that the central bank can eventually get nominal GDP on target. It would be a little worrisome if it only can work if people believe it will work.
Here Woodford provides an accessible explanation of Wallace equivalence--a reason why open market operations of long term to maturity and risky bonds do not matter. Interestingly, in Woodford's account, there is a connection with Ricardian equivalence. Suppose the Fed purchases long term and risky securities and issues short and safe securities. This increases the risk of loss to the Fed and a risk of higher taxes for someone to make up for the shortfall of revenue going from the Fed to the Treasury. Investors will then want to hold fewer risky securities and more safe securities to offset the risk of these tax hikes. And so, investors increase their demand for these safe securities created by the Fed. There is no excess supply of them and so no motivation to spend more on output.
I don't believe in Ricardian equivalence and I find this even more incredible. That is, I don't believe that people expand saving enough to earn enough interest to pay the added taxes that they and their descendants will have to pay to cover interest and principal payments on government debt. And so, adding more complications through risk of loss by the central bank and so a risk of reduced transfers to the treasury and so higher taxes in the future seems even more implausible.
Aside from skepticism about the desirability of using models of God-like knowledge and optimization abilities to analyse real market economies, I think that another problem with the application of Woodford's analysis to truly heroic open market operations is corner solutions. When there are no other existing asset holders to sell to the Fed, and that the Fed is instead solely purchasing new issues, then it seems to me that spending on output increases.
Of course, the Fed has not been following a strategy of quantitative easing until a nominal target is hit, but rather making purchases of specific amounts of assets. Further the Fed has claimed that its purpose is to lower interest rates on those assets. Woodford argues that much of the effect on interest rates associated with these purchases was instead due to the Fed's promise to keep policy rates lower in the future. Interestingly, he also reports on evidence that the asset purchases resulted in higher expected inflation, which would tend to raise nominal interest rates.
Market Monetarists have long insisted that the purpose of quantitative easing should be to raise spending on output, and if this is expected to work, then the result would higher real and nominal interest rates. This would involve market participants selling more securities than the Fed is buying, using the funds to purchase capital goods or consumer goods.
Rather than making open ended commitments to quantitative easing combined with a nominal GDP level target, which is the Market Monetarist approach, Woodford seems committed to instead keeping the policy rate at zero until nominal GDP reaches the target level. Rather than depend on the effects of such quantitative easing to expand spending, he favors fiscal policy--tax cuts or expanded government spending--to directly raise demand for output.
While I don't doubt that government borrowing and spending in the context of a central bank willing to create whatever amount of money people want to borrow can expand spending, I don't think that is a wise policy when government spending and debt is already too high.
Thursday, August 23, 2012
Boom and Bust Banking--In Print
It is available from the Independent Institute here.
Subscribe to:
Posts (Atom)