Thursday, February 5, 2009

Why are the Fed and the Treasury Working Against Each Other?

It seems like every other day the last couple of weeks I read about how long Treasuries are tanking because markets are “worried about supply.” Then I read about how the Fed is saying they plan to buy long Treasuries, and they’re trying like heck to convince the markets that it’s really gonna happen, but the markets don’t quite believe it.

Maybe there’s a good reason that the markets don’t quite believe it: it doesn’t make any sense. The Treasury is going to issue a crapload of new long bonds, and the Fed is going to buy them back on the open market? Why? If the Treasury just issued short-term bills, the Fed would buy them anyway. At least it would buy enough to keep the federal funds rate near zero, and after that it doesn’t matter. And when the recession, or the potentially deflationary episode, or whatever it is, ends and the Fed wants to stop rolling over the bills, the Treasury can replace them with long bonds if it so chooses.

Now, you might say, “Ah, but when the recession, or the potentially deflationary episode, or whatever it is, ends, long term interest rates will go up, and the Treasury will have to pay more to borrow than it does now, so why not issue them now while the rates are cheap?” That reasoning seems to make sense, until you realize that the Fed, when this episode ends, is going to want to liquidate its long-term bond portfolio, and if rates have risen, it will have to do so at a loss. Since the Fed’s profits go directly into the Treasury, the loss passes right through, and, in terms of present value, it’s (almost) exactly as if the Treasury had issued new debt at a higher yield. Or the Fed could decide to hold the bonds to maturity, in which case it will be just as if the Treasury had never issued them. (In that case, if the Fed wants to reduce the money supply, it could stop rolling over its T-bills, in which case the Treasury will have to issue new ones at higher rates, just as it would have if it had not issued the long bonds in the first place.)

There are only two differences I can see between, on the one hand, having the Treasury issue long bonds and the Fed buy long bonds, and, on the other hand, having the Treasury never issue the bonds in the first place. And neither difference is really a difference. The first difference is that the Fed might buy seasoned bonds in addition to (or instead of) on-the-run issues. The Treasury, by definition, can only issue on-the-run bonds. But this, as I said, is not really a difference: the Treasury can just as well decide to retire its old bonds as to refrain from issuing new ones. Although most Treasury issues are not callable, the Treasury can retire them at market prices, and it will be just as if the Fed had bought them. And in any case, how much difference is there really between on-the-run and seasoned issues? The maturities are slightly different – no big deal in the grand scheme of things – and on-the-run issues are more liquid – a little bit more liquid, but it’s not like seasoned Treasuries sit on your books unintentionally for months because you can’t catch a bid. It’s a technical difference – one that will affect the details that Treasury traders care about, but not the sort of thing that should make headlines.

The other difference is between current policy and future policy. If future policymakers will be somehow influenced by the distribution of securities between the Treasury and the Fed, then that distribution will of course make a difference. But why should they be influenced by it? I can understand that the Fed and the Treasury may perceive themselves as having different interests, but in this case it doesn’t matter, because each one can undo what the other one does. If the Fed sells off its long bonds 5 years from now, the Treasury can stop issuing long bonds. If the Fed holds onto its long bonds, the Treasury can issue more of them. And if the Fed never buys the long bonds in the first place, then the Treasury can do whatever it wants with its finance policy 5 years from now. Again, there are the technical questions about maturity and liquidity, but I don’t see how these are going to make a difference that macro policy makers or investors (as opposed to traders) should care about. And since the timing of these events is unknown, I’m not even sure anyone would be able to make a reasonable guess as to how these technical matters will play out, so perhaps noone at all should care about them.

So what’s going on here? I really don’t get it. Are the Treasury and the Fed really going to insist on working against each other? Is one of them bluffing? (And if so, which one?) Have the markets misinterpreted their signals? Have the markets, in fact, correctly interpreted their signals as being meaningless?


DISCLOSURE: Through my investment and management role in a Treasury directional pooled investment vehicle and through my role as Chief Economist at Atlantic Asset Management, which generally manages fixed income portfolios for its clients, I have direct or indirect interests in various fixed income instruments, which may be impacted by the issues discussed herein. The views expressed herein are entirely my own opinions and may not represent the views of Atlantic Asset Management.

13 comments:

Nick Rowe said...

Interesting question, Andy. Two thoughts, not really an answer:

1. Maybe it doesn't really matter. Except for the minor transactions cost of the Fed buying long with short, we end up in the same place. Especially if the Fed buys the longs directly from Treasury, and then sells shorts into the market.

2. Suppose that the Treasury figures that eventually, when the crisis is past, they will want the deficit to be financed by long bonds. The Treasury sets the long/short mix according to its long-run objective, and lets the Fed deal with the short-run stuff.

I expect what I'm saying is that if you consolidate the balance sheets of Fed and Treasury, which is what you are sort of doing, it's hard to think why we would have a separate Fed in the first place.

gg said...

If I understand correctly the target of the Fed is to lower the risk-free return with which the private sector discounts the cash-flows of projects thus stimulating investment and credit.

The purchase of long maturity bonds is not intended to lower the borrowing cost for the treasury or to enable it to save money or to finance the deficit.
And the question that arises is if the purchase of long treasuries will lower their yields.

If the answer is affirmative, as I believe it is, then it seems as a sensible policy that will stimulate, probably not substantially, the economy with no cost as you describe to the taxpayer

Anonymous said...

It is a matter of conflicting mandates, and short run versus long run priorities.

The dual mandate Fed is not planning in terms of long run strategy. It is in a state of panic, and is frantically engaging in quantitative easing by monetizing anything and everything. With no quant history to calibrate the effects, it is erring on the side of too much stimulation to guard against any possibility of there being too little.

Long run strategy has never been a Fed strong point. There is no exit plan at this point, or much regard for anything beyond short term stimulation at all costs. Fed models basically predicted the current situation would not happen, and they are making things up as they go along at this point.

Treasury is a separate department, with its primary mandate being to fund public spending. Treasury is trying to sell as many long bonds overseas as it can to finance the large scale fiscal stimulus package. That is, Treasury expects interest rates to rise in the next 30 years, and wishes to minimize any interest payments going overseas.

Treasury is thinking more long term, but with different mandates than the Fed, this produces different strategies. That is, if long bond issuance forces the future Fed to regressively extract more resources from domestic citizens, this just increases public funding. Treasury has no official mandate to maintain the quality of life of average citizens, just fund public spending.

Fullcarry said...

Its to maintain money illusion. They can't really let the cat out of the bag now can they.

I really don't mean to be flip here. They really need to maintain the illusion that money is stable. That its creation isn't at the whim of the government. The separation of treasury and FED is one way of maintaining that illusion.

David Pearson said...

Fed losses "pass through" to the Treasury? I looked through the Federal Reserve Act, and I couldn't find anything to this effect.

The act says that earnings from Reserve Bank surplus accounts must be given to the Board of Governors for passing on to the Treasury. Does this mean that Treasury must replenish losses? I don't see why you would interpret it that way. The "earnings" passed on have the character of dividends, and there is not obligation to pay a "negative dividend".

In short, as I read it the Fed is not supposed to have retained earnings from operations, only capital from member banks. This doesn't mean, however, that the Treasury must absorb losses.

What happens in the even of Reserve Bank capital being wiped out by losses? There is no minimum capital requirement stipulated in the Act that I could find. Perhaps you know of one. In the absence of a capital requirement, the Fed could theoretically operate with a negative net worth.

Anonymous said...

What should they do?

Issue 30 years into the market now, and buy 10 years when they want to intervene.

Mortgage rates are more sensitive to 10 years than 30 years, due to better duration matching.

And 30 years is more prudent for funding the deficit, as well as being super cheap for the time being.

Anonymous said...

David Pearson is correct on the capital aspect.

Andy Harless said...

David,

I glossed over the question of what happens to the Fed's losses in order to make the post readable, but perhaps I should have discussed it in a footnote.

You are correct that the Fed's net losses don't pass through directly to the Treasury. However, as long as the Fed is running a net profit, the gross losses on any particular transaction will pass through to the Treasury, in the sense of reducing the net profit. AFAIK in the past, the Fed has always run a net profit, so any losses on any particular transaction have passed through to the Treasury.

It's certainly possible to imagine a case where the Fed would have a net loss in one year. I would argue that any such loss will ultimately be funded by the Treasury even if there is no explicit arrangement. Suppose the Fed wants to achieve a certain path for the base money stock over time. Now, compare the case where it sells a certain asset at breakeven to the case where it sells that asset at a loss. In order to achieve the same path for the base money stock in the second case, it will have to sell some other asset to make up the difference. Therefore it will be losing all the future returns on that other asset. (Note also that, in the breakeven case, the Fed could roll over that other asset indefinitely.) So its future profits will be reduced by an amount equal in present value to loss that it is taking in the current year. Thus as long as the Fed eventually becomes profitable, those profits will be reduced, and the Treasury will suffer the loss.

I suppose one could theoretically imagine a case where the Fed never becomes profitable again, but given the Fed's ability to issue zero-interest liabilities, that seems highly unlikely.

Andy Harless said...

Regarding the question of mandates, I don't think the mandates matter, because each institution must pursue its mandate conditional on the expected actions of the other. If the Treasury knows that the Fed is going to buy up all its long bonds, then there is no point in issuing those bonds in the first place. It's mandate will be served equally well by issuing or not issuing those bonds.

David Pearson said...

Andy,

You're right in the sense that the Fed will lose future income once it decides to shrink its balance sheet. The question is, will it?

In the case where a balance sheet expansion (say, to $3tr) is permanent, then the situation which you describe is unlikely to occur.

Central Bank financing of national debt is typically permanent, for the simple reason that, in the presence of structural fiscal deficits, a reversal of that financing leads to massive crowding out. Japan was one exception to the rule, while the U.S. in 1937 was proof of it: a reversal of balance sheet growth caused the "mini Depression" and helps explain why the Depression lasted so long. Needless to say, I don't believe current Fed policy makers are likely to seek a repeat of this experience.

I like your blog, but much of what you write seems to rest on the non-permanence of quantitative easing.

Andy Harless said...

David,

It's not necessary that the Fed shrink its balance sheet, only that it refrain from expanding the balance sheet faster than it otherwise would have. For example, if the Fed sells some 10-year notes at a loss and replaces them with T-bills, the number of T-bills that it will be able to buy with the proceeds is reduced by the amount of the loss. This is true no matter how many additional T-bills the Fed is buying at the same time.

(Of course, if the Fed holds all the longer-term securities to maturity, then the whole issue never arises; it is just as if the Treasury had issued short-term securities in the first place and the Fed had continued to roll them over.)

Actually, if you look at my first substantive post, I actually suggest specifically that quantitative easing will be permanent. I'm still thinking about a post trying to justify that suggestion on the grounds that, in retrospect, it was a mistake for Japan to undo some of its quantitative easing (and indeed about half of the huge amount of new base money was never withdrawn). If I'm reading the numbers correctly, the amount of extra base money that remained was, relative to GDP, comparable to only the most extreme suggestions about what the US might do.

David Pearson said...

Andy,

In the end you're right: Fed losses have to be borne by its equity holders, the recipients of profits. My argument is that, theoretically, the Fed can postpone the realization of losses because, unlike real firms, its creditors do not care if it has equity or not.

On the previous inflation post you cited, I would ask you to consider something. You argue for a NAIRU model of inflation creation, and certainly that model is valid. But there's an almost completely different model driven entirely by velocity and not capacity. Call it the "debt crisis" model. In it, a government absorbs the non-serviceable liabilities created during a credit boom. The resulting structural budget deficits threaten to launch debt growth into an exponential path. Since governments rarely default on domestic-currency debt, the implication of such a path is exponential growth in the central bank's balance sheet. It is the expectation of such growth that causes velocity and the price level to rise. This dynamic has very little to do with NAIRU, and it has been the prevalent cause of historical, global inflationary episodes ranging from Latin America in the 80's to the U.S. in 1936. The one place where this model broke down was Japan, where for unique reasons, velocity never increased.

Anonymous said...

Could it be simply that the Treasury does not consider the effect on transfers from the Fed when designing its strategy to minimize the costs of financing its debt?