Friday, November 5, 2010

QE 2

The Fed will be purchasing $600 billion of long term Treasury bonds. This is a 30% increase in the monetary base. For money expenditures to return to the growth path of the Great Moderation over the next year, the increase in Final Sales of Domestic Product must be 21 percent.

At first pass, then, the amount of quantitative easing looks excessive--about 50 percent too high. However, it is possible that the expansion in the monetary base will not result in a proportional increase in money expenditures--that the real demand for the monetary base will grow faster than real income. There is no guarantee that the demand for the money base will remain proportional to money expenditures over the next year.

Unfortunately, the Fed doesn't target a growth path of money expenditures. Much of the discussion of the Fed's policy, including much of what the Fed itself says, suggests that the purpose of the policy is to lower long term interest rates. The Fed will buy long term bonds, raising their prices, and lowering their nominal yields. The lower nominal interest rates, given expected inflation, will result in lower real interest rates. The lower real interest rates are supposed to encourage additional borrowing. How does that work?

Apparently, the idea is that those who sell the government bonds to the Fed will use the funds they receive to purchase privately issued financial assets, bidding up their prices and lowering their yields as well. Because of the lower interest rates, firms will issue more such bonds--borrow more. These could be firms issuing new corporate bonds. Or perhaps more mortgate backed securities would be issued, allowing for lower mortgate rates. Perhaps other asset backed securities could be issued, allowing banks to take credit card debt off their balance sheets. Perhaps finance companies could issue more bonds, and make more loans to small business. Regardless, those directly or indirectly borrowing the money will spend at least some of it on consumer and capital goods. The firms producing the consumer and capital goods will obtain additional sales. As firms see expanded sales, they will expand production and hire workers.

It is very likely that firms have been limiting price increases because of slow sales. If sales pick up as planned, then firms will raise their prices more quickly. The inflation rate will rise. The Fed says that current levels of inflation are too low. This pick up in inflation will move it closer to what the Fed thinks is the right level. While the Fed continues to be less than perfectly clear, a 2 percent inflation rate appears to be what is being proposed.

To the degree the Fed's policy is expected to slightly increase inflation, this will reinforce the effect of the lower nominal interest rates and further reduce real interest rates. This entices more borrowing and more spending, more sales and more production.

Maybe.. but not necessarily.

I favor targeting a growth path for Final Sales of Domestic Product. I favor some opportunistic disinflation, with a shift in the growth path to one increasing 3 percent per year from the old growth path for the third quarter of 2008. I believe the Fed should commit to hitting that adjusted growth one year in the future--a 13 percent increase.

Given the current unsettled conditions, I think the Fed should purchase $260 billion of Treasury bonds. This is a 13 percent increase in the monetary base, which is currently approximatly $2 trillion.

They should buy bonds that have yields greater than the interest rate the Fed is paying on reserve balances (.25 percent.) Currently, Treasuries with maturities 2 and 3 years into the future meet that criterion. However, if the Fed's purchases push down those yields, to .25 percent or below, the Fed should purchase longer term Treasury bills as needed. Of course, the Fed can and should reduce the interest rate it pays reserve balances. I believe it should be reduced to slightly less than zero, which would allow the Fed to purchase shorter term Treasuries again.

The implied target for the monetary base, of $2,260 billion should remain tentative. The "policy" should be to buy (or sell) whatever amount and maturities of Treasury bonds needed to reach the target for money expenditures-- $16,415 billion, in my view. In particular, if $260 billion of bonds is not enough, then the Fed should commit to buying more. But, of course, if money expenditures are above target, then the Fed should be committed to selling off whatever amount is necessary.

What of the transmission mechanism? Should the goal be to lower interest rates on long term Treasury bonds? Should the goal be to raise inflation expectations so that these lower nominal interest rates entail an even lower real interest rate? More importantly, should the goal be to encourage more borrowing?

My answer is no.

Clearly, as an advocate of money expenditure targeting, I believe the goal should be to increase both expected and actual money expenditures. However, what happens to short term or long term, nominal or real interest rates is not important. Whether borrowing increases or deceases is not important.

In my view, the ideal response to quantitative easing is for households and firms to sell off already existing bonds that they currently hold and use the funds they receive to purchase consumer goods or capital goods. This is not an increase in borrowing, but rather a decrease in lending. While not directly a decrease in the flow of new lending, those who end up with the already existing bonds won't be purchasing newly issued bonds. The sale of existing bonds quickly translates into a contraction in the flow of new lending.

Unfortunately, bad macroeconomic analysis is common. Decreased lending is nearly always associated with reduced expenditures on goods and services. Less lending implies higher interest rates, less borrowing, and less spending on goods and services. While plausible enough, the implicit assumption is that the alternative to lending is to hold money. Less lending is implicitly assumed to be an increase in the demand to hold money. Or, perhaps, the assumption is that the quantity of money drops because any lending is funded by newly created money.

However, if the alternative to lending is spending on goods and services, then less lending does not entail less spending. Less lending is more spending.

How can quantitative easing result in less lending and more spending? One possible pathway would be lower nominal interest rates. The Fed buys bonds, raising their prices and lowering their yields. The lower nominal interest rate reduces the quantity of bonds demanded, and rather than hold the money they receive, the previous bond holders purchase consumer goods or capital goods.

As explained above, the increased spending on consumer or capital goods will make firms less restrained in raising prices. The result should be higher inflation, and expecting this, the policy should raise expected inflation. This higher expected inflation could add to the effect of lower nominal interest rates to further lower real interest rates, and so, further reduce the real quantity of bonds demanded, and spur a larger increase in spending.

However, it is possible that existing bonds holders could sell even more bonds than the Fed is purchasing, resulting in higher nominal interest rates, yet because of the higher expected inflation, the result is lower real interest rates. That is what is motivating the decrease in the real quantity of bonds demanded, and so, the decrease in lending and an increase in the demand for consumer goods and capital goods.

This pathway also provides an incentive for those holding short term bonds, including various bank deposits, whose yields are already very low, to sell them off and use the proceeds to purchase consumer and capital goods. Again, this involves no increase in borrowing, but rather a decrease in short term lending. While the nominal yields on these short term securities are unlikely to fall much, an increase in expected inflation will reduce the real interest rates on them and so motivate a decrease in the quantity of short bonds demanded and an increase in spending on consumer and capital goods. Further, as explained above, it is possible that higher expected inflation could result in higher nominal interest rates for short term bonds and deposits, but still allow for lower real rates, and so motivate the reduced quantity of bonds demanded, and increase in the demand for consumer goods and capital goods.

Finally, and most importantly, the expectations of future inflation are not the only pathway by which the impact of quantitative easing could raise interest rates. The expected increase in sales provides an incentive for firms to sell off their existing holdings of long and short term bonds and use the proceeds to purchase capital goods to be able to expand capacity. The expected increase in production, real income, and employment provide an incentive for households to sell off existing holdings of bonds to fund the purchase of consumer goods.

Consider a well off household that is refraining from purchasing a new sailboat, instead holding on to T-bills, because of fear that poor business performance is going to result in a reorganization and an extended period of unemployment. Finding a new, high paying VP job may take some time with the slow economy. Turn expectations around, and the household sells off the T-bills to fund a purchase of the sailboat.

Expectations of higher real sales, real income and real output, and higher employment could motivate more sales of bonds than the amount purchased by the Fed, and so higher nominal interest rates on the long(er) term bonds the Fed purchases as well as the shorter term ones that have yields already at or below the interest rate the Fed pays on bank reserves. What is interesting, however, is that the reduced demand for those bonds (and increased demand for consumer and capital goods,) could result in higher real interest rates as well.

It is possible, and perhaps in some way desirable, that quantitative easing result in higher nominal and real interest rates and less lending--but still more spending. Aside from that final impact, the result is almost the opposite of the conventional account of its effect. Not lower nominal interest rates, lower real interest rates, more borrowing and so, more spending. But rather, higher nominal interest rates, higher real interest rates, less lending, and more spending.

If the Fed were to return to the growth path of money expenditures from the Great Moderation, the most likely result would be for the price level to return to its growth path. This would involve higher than 2 percent inflation during the adjustment period, and to the degree inflation has been less than 2 percent during the Great Recession, a higher inflation rate in the long run.

My preferred noninflationary growth path is a bit more complicated. My judgement is that the result of a prompt move to that growth path would involve an increase in the price level, and so some inflation during the adjustment period, but the lower growth rate of money expenditures would result in lower--actually zero--trend inflation in the long run.

Finally, I am not really opposed to real interest rates falling. And higher expected sales, real output, and employment would certainly motivate increased borrowing--a higher supply of bonds. I am not opposed to increased borrowing. There are financially sound households and firms that might benefit by borrowing and would be good credit risks for lenders. But increased borrowing is not necessary. And most importantly, imagining that quantitative easing is about (too) low interest rates and already (excessively) indebted households and firms borrowing more, and so increasing spending is WRONG!

So, it looks to me that the Fed's quantitative easing is excessive and its explanation of what it is trying to do is mistaken. But most importantly, the Fed needs to start targeting money expenditures and stop thinking about interest rates--short term or long term.

8 comments:

  1. Great post Bill. I'm curious though, how do you reach your (tentative) target? It seems to be the amount of the expansion in monetary aggregates that you think will boost nominal final sales to your target. Is it just a one-for-one relationship or did you use a regression?

    ReplyDelete
  2. Great post. Two minor problems:

    "Of course, the Fed can and should reduce the interest rate it pays reserve balances. I believe it should be reduced to slightly less than zero, which would allow the Fed to purchase shorter term Treasuries again."

    The Fed should reduce IOR. But it is very likely that the yield on short term Treasuries would fall below IOR, as those short term Treasuries have lower liquidity risk premium than monetary base.

    "They should buy bonds that have yields greater than the interest rate the Fed is paying on reserve balances (.25 percent.) Currently, Treasuries with maturities 2 and 3 years into the future meet that criterion."

    In the presence of arbitrage, it is not terribly important which maturities the Fed is buying, the most important problem is the lack of commitment to a money expenditure path target. The Fed should choose a money expenditure target, and it should obtain an estimate (preferably based on market mechanism) of future path of IOR rates that is consistent with money expenditure target. The risk adjusted difference between the term structure of interest rates and an estimate of optimal future path of IOR rates should give a good indication about which treasuries should be purchased by the Fed. It is quite possible that a result of such exercise would produce a list of Treasury securities similar to what NY Fed has produced.

    ReplyDelete
  3. Bill, Good post. A few quibbles. I'm not sure I'd start off with the 50% increase in the base statistic. With the base being 1/2 interest free currency, and 1/2 interest bearing reserves, it's no longer clear there is a meaningful definition of the base. It's a hodgepodge, and the QT of money no longer applies. If anything, it is currency where the QT may still have some application. But it's not clear how much currency will increase.

    I agree with 123 on maturities. I'm not sure it's that important. Even if the yield on a 3 year is above 1/4 percent, the expected return on holding the 3 year may be less than 1/4 point for the next 12 months (according to the expectations hypothesis of the term structure.

    ReplyDelete
  4. Scott,

    There is no meaningful definition of the base? That's a bit strong, isn't it? Sure, IOR complicates matters, but 'no meaningful definition', really!? If banks charge different interest rates on chequing deposits (and currency earns no interest), then is there no meaningful definition of the money supply? If anything, despite IOR, the monetary base is much less of a 'hodgepodge' than any of the broader monetary aggregates. Even NGDP must also be a meaningless hodgepodge, right? I mean, what is being spent?

    Perhaps I do not understand you point.

    ReplyDelete
  5. "Of course, the Fed can and should reduce the interest rate it pays reserve balances. I believe it should be reduced to slightly less than zero, which would allow the Fed to purchase shorter term Treasuries again." Mightn't such a reduction of IOR be in itself sufficient monetary stimulus, without the need to buy Treasury securities?

    "However, if the alternative to lending is spending on goods and services, then less lending does not entail less spending. Less lending is more spending." This analysis seems flawed. I can spend a sum of money that I have, or I can lend it. (I could also hoard it, but we're leaving that possibility aside. I could invest it in capital goods, but that is already being counted as "spending.") If I lend it, does that entail less spending? No, because the person or firm to whom I lend it may (and *must*, if we are excluding hoarding) spend it: the amount of spending ends up the same as if *I* had spent it.

    I do not see why you have any confidence in your estimate of *how much* QE the Fed needs.

    ReplyDelete
  6. Lee, I meant that monetary models typically assume the base does not earn interest. So we can't talk about what we'd expect to have if the base grows X%, without also considering that ERs pay higher interest than T-bills. Maybe my statement was a bit too strong, but I can emphasize enough how normal quantity theory of money predictions go right out the window when you start paying IOR.

    ReplyDelete
  7. Scott:

    No, they don't go out the window.

    They are always ceteris paribus, and one of the things is that is left unchanged with the change in the quantity of money is the interest rate paid on reserves.

    As Lee said, all of this ground has been covered with interest rates on deposits. If you think about deposit only money, it is pretty obvious.

    ReplyDelete
  8. I have no confidence in my estimate of the amount of base money that needs to be created for money expenditures to return to target.

    I don't think about it in terms of a target for base money that will determine some level of money expenditures that I hope will be the target.

    The target is primary. If a proportional increase in base money doesn't lead to money expenditures on target, then increase it some more. I am certain that as soon as money expenditures start moving rapidly to the target, that the quantity of base money will need to contract.

    Anyone who has confidence that past relationships between nominal interest rates and money expenditures is in error.

    I favor index futures targeting. But that just means that the level of base money is bounded by market expectations--what market participants think is the proper level of base money.

    ReplyDelete