Sunday, March 3, 2013

Monetary policy with large debts

This is a Wall Street Journal Op-Ed, March 4 2013. They titled it "Treasury needs a better long game," but the most important question is whether the Fed can keep any independence, if 5% interest rates will cause $900 billion interest costs. There is a pdf version of the oped on my webpage. 


Sooner or later, the Federal Reserve will want to raise interest rates. Maybe next year. Maybe when unemployment declines below 6.5%. Maybe when inflation creeps up to 3%. But it will happen.

Can the Fed tighten without shedding much of the record $3 trillion of Treasury bonds and mortgage-backed securities on its balance sheet, and soaking up $2 trillion of excess reserves? Yes. The Fed can easily raise short-term interest rates by changing the rate it pays banks on reserves and the discount rate at which it lends.

But this comforting thought leaves out a vital consideration: Monetary policy depends on fiscal policy in an era of large debts and deficits. Suppose that the Fed raises interest rates to 5% over the next few years. This is a reversion to normal, not a big tightening. Yet with $18 trillion of debt outstanding, the federal government will have to pay $900 billion more in annual interest.

Will Congress and the public really agree to spend $900 billion a year for monetary tightening? Or will Congress simply command the Fed to keep down interest payments, as it did after World War II, reasoning that "Fed independence" isn't worth that huge sum of money?


This additional expenditure would double the deficit, which tempts a tipping point. Bond markets can accept fairly big temporary deficits without charging higher interest rates—buyers understand that bigger deficits for a few years can be made up by slightly larger tax revenues or spending cuts over decades to follow. But once markets sense that deficits may be unsustainable, and that bond buyers may face default, restructuring or inflation, they will demand still-higher interest rates. Higher rates mean higher deficits—leading to a fiscal death spiral.

Many economists think the tipping point starts when total government debt (federal, state and local) exceeds 90% of GDP. We are past that value, with large state and local debts, continuing sclerotic growth and a looming entitlements crisis to boot. This, not the "balance sheet" or other monetary or institutional constraints, will be the Fed's quandary—can the monetary authority really dare to risk a fiscal crisis?

The obvious answer is to fix the long-run deficit problem, with the reform of runaway spending, entitlement programs and a pro-growth tax policy. So far that is not happening.

Still, the Fed and the Treasury can buy a lot of time by lengthening the maturity of U.S. debt. Suppose all U.S. debt were converted to 30-year bonds. Then, if interest rates rose, Treasury would pay no more on its outstanding debt for 30 years. And if the country couldn't solve its fiscal problems by that time, it would deserve a Greek crisis.

Maturity structure of US debt.
Alas, the maturity structure of U.S. debt is quite short. I estimate that our government rolls over 40% of its debt every year, and 65% within three years. (I account for Federal Reserve holdings, coupon payments and use market values.) Thus the fiscal impact of higher interest rates will come quickly.

Mr. and Mrs. Smith shopping for a mortgage understand this trade-off. Mr. Smith: "Let's get the adjustable rate, we only have to pay 1%." Mrs. Smith: "No, honey, that is just the teaser rate. If we get the 30-year fixed at 3%, then we won't get kicked out of the house if rates go up."

Amazingly, nobody in the federal government is thinking about this trade-off. Instead, each agency thinks only for itself.

The Fed is still buying long-term bonds in an effort to temporarily drive down long-term interest rates by a few basis points. It has concluded it can survive the loss in mark-to-market value of its bond portfolio that higher interest rates will imply, when they come, by suspending its customary interest-rebate payments to the Treasury. If the Treasury was counting on that roughly $80 billion per year, that is Treasury's problem. If higher rates cost the Treasury $900 billion a year, that is Congress's problem.

The Treasury's Bureau of Public Debt controls the maturity of federal debt issues. It has been gently borrowing longer in response to low long-term rates, but not enough to substantially alter the government's interest-rate risk. The bureau also views its job narrowly—which is to finance whatever deficits Congress determines, not to take actions that mitigate future deficits. Congress and the administration are busy with other matters.

Ironically, the Fed's buying and the Treasury bureau's selling have neatly offset, leaving very little change in the maturity structure of debt in private hands.

What to do? First, the Treasury and Fed need a new "accord" to decide who is in charge of interest-rate risk, most likely the Treasury, and then grant it clear legal authority to manage that risk. The Fed should then swap its portfolio of long-term bonds for a portfolio of short-term Treasuries and forswear meddling in the maturity structure again.

Second, the Treasury should seize its once-in-a lifetime opportunity to go long. Thirty-year interest rates are at 2.8%, a 60-year low. Many corporations and homeowners are borrowing long to lock in low funding costs. So should the Treasury.

You may complain that if the Treasury borrows long, then long-term rates will rise. If so, it is better that everyone knows that now. It means that markets aren't really willing to buy long-term government debt, that the 2.8% yield is only a fiction of the Fed's current buying, and that it won't last long anyway. Better fix the fiscal hole, fast.

[WSJ cut: Moreover, if buying and selling a lot of bonds is a problem, the Treasury should engage in an aggressive swap program. In a swap, the Treasury pays a counterparty a fixed rate (say, 2.8%) and receives floating rates, with no bond changing hands. The First Bank of Podunk uses swaps to manage interest rate risk, when it doesn’t want to buy and sell assets. So can the Treasury.

You may complain about counterparty risk. But swaps are collateralized, so each counterparty does not lose if the other one defaults. And if the thousands of pages of Dodd-Frank regulation, and the army of stress testers can’t ensure that too-big-to-fail banks properly manage simple interest rate risk, then we really should have a law-book bonfire. ]

You also may argue that 2.8% long term-debt is more expensive than 0.16% one-year debt. There are two fallacies here. First, the 2.8% long-term yield reflects an expectation that short rates will rise in the future, so the expected cost over 30 years, as well as the true annual cost, are much closer to the same. Second, to the extent that long-term bonds really do pay more interest over their life span, this is the premium for insurance. Sure, running a restaurant is cheaper if you don't pay fire insurance. Until there is a fire.

A much longer maturity structure for government debt will buy a lot of insurance at a very low premium. It will buy the Fed control over monetary policy and preserve its independence. If Fed officials realized the risks, they would be screaming for longer maturities now.

But we don't have long to act. All forecasts say long-term rates will rise soon. As the car dealer says, this is a great deal, but only for today.



Notes: See this earlier blog post or pdf essay for more. I hope to sell WSJ on the second part, keeping the balance sheet big and Treasury floating-rate debt, sometime soon.

If you think I'm pessimistic or making up numbers, here are the CBO's baseline budget projections 


By 2023, the CBO thinks interest payments on the debt will be $857 billion, essentially the entire deficit!  The CBO assumes (see p. 67 here) that interest rates start rising in 2016 and in 2018 stop at 4% one year and 5.2% ten year rates. What if the Fed really wants to tighten?

Moreover, the cheery forecast of return to a balanced primary budget relies on an assumption of spectacular growth (p. 68) -- 3 years of 4% growth bringing us back to "potential" (see p.36) -- in-your-dreams tax revenues, and the rosy-scenario "baseline" expenditure cuts. What if the primary deficit is also $900 billion dollars?     
 

26 comments:

  1. Professor,

    I am not sure I see your argument for the strong recommendation below.

    "The Fed should then swap its portfolio of long-term bonds for a portfolio of short-term Treasuries and forswear meddling in the maturity structure again."

    In other news, didn't I read in the WSJ a month ago that the Treasury plans on issuing floating rate debt? Professor Harvey tried to tackle that last year

    http://gardenofecon.com/2012/05/wrong-time-wrong-place-for-floating-rate-debt/

    ReplyDelete
  2. John,

    "Yet with $18 trillion of debt outstanding, the federal government will have to pay $900 billion more in annual interest."

    The federal government incurs an annual $900 billion in interest payments only because it chooses to sell coupon bonds instead of accrual bonds.

    "Still, the Fed and the Treasury can buy a lot of time by lengthening the maturity of U.S. debt. Suppose all U.S. debt were converted to 30-year bonds. Then, if interest rates rose, Treasury would pay no more on its outstanding debt for 30 years."

    Suppose the Treasury began lengthening the term structure of its debt 10% a year AND converting that portion of its debt to accrual securities instead of coupon securities. That means that interest and principle are repaid when the bond matures instead of the bond making regular coupon interest payments.

    "The obvious answer is to fix the long-run deficit problem, with the reform of runaway spending, entitlement programs and a pro-growth tax policy."

    The obvious answer is to fix the debt problem is to sell equity instead of debt. Government debt is a guaranteed claim on future tax revenue. There is nothing stopping the federal government from selling equity claims on the same future tax revenue.

    ReplyDelete
    Replies
    1. That would be politically unfeasible. Think about it. twenty years from now the government still cannot live within it's means, we have a recession, and now the public notices that a lot of wealthy people are taking in part of the tax money before it even gets to the government. The calls for confiscation would be enormous.

      Delete
    2. KyleN,

      Which part would be politically unfeasible - converting from coupon to accrual debt securities or selling equity instead of debt?

      Delete
  3. In the absence of a credible commitment device, it is difficult for the FED to keep down long term rates. When the future changes, current promises are out of the window. Isn't the balance sheet that they constructed (in collaboration with the Treasure) simply providing that commitment device? "You can believe us. We won't raise rates (ever) because it would bankrupt us all." I guess I believe them.

    ReplyDelete
    Replies
    1. Yes, this is my fear, the Fed may not be able to keep down rates. In fact if we ever do see a real recovery then inflation will reappear as banks begin to loan out this huge hoard of cash they are sitting on.

      Delete
  4. John,

    "What if the primary deficit is also $900 dollars?"

    I think you meant to ask - what if the primary deficit is also $900 billion dollars.

    ReplyDelete
    Replies
    1. I believe he meant $900 billion dollars. I would not want him to be my accountant if he can screw up by that amount.

      Seriously, the amount of money involved in all this is just staggering and mind numbing. Small businesses would be hard pressed to survive in this kind of economic crisis, let alone grow to become a bug business, I meant big business.

      Delete
  5. "Moreover, the Treasury should engage in an aggressive swap program. In a swap, the Treasury pays a counterparty a fixed rate (say, 2.8%) and receives floating rates, with no bond changing hands. The First Bank of Podunk uses swaps to manage interest rate risk, when it doesn’t want to buy and sell assets. So can the Treasury."

    The U. S. Treasury is not in the money lending business. Banks use swaps to manage interest rate risk (specifically interest spread risk) because they both borrow money (usually short term) and lend money (usually long term). The U. S. Treasury is a borrower only.

    What the WSJ is advocating is for the US Treasury to take the opposite side of a bank's trade - a bank sells its long term fixed interest payments to Treasury for short term floating rate payments.

    In essence, the tax payer is on the hook when a bank cannot properly manage its interest rate risk.

    ReplyDelete
  6. When Britain faced a similar dilemma under William Pitt the Younger, two solutions were adopted: the Bank of England suspended redemption for the next 3 decades and existing government debt was consolidated into IOUs that had no maturity dates. All new borrowing by the government became, in practical terms, non-recourse; the Crown's only obligation was to "pay" interest in notes that the courts would force commerce to accept as legal tender. Official interest rates became a constant; what fluctuated were the discounts, volumes and maturities of private credit.

    ReplyDelete
  7. So we have the following scenarios:

    1) growth accelerates and we have too much inflation - Fed raises rates
    2) stagflation - Fed may or may not raise rates
    3) slow grind/deflation - Fed does not raise rates

    The 1st scenario is one of fiscal solvency
    the 3rd scenario is one where rates either fall further or stay low

    the 2nd is basically a tossup - hire interest costs, but the real value of our debt will rapidly diminish

    In other words, I have a hard time seeing a scenario in which we need to worry about interest rates rising

    ReplyDelete
    Replies
    1. I wonder what his students at Chicago will think of this?

      Delete
  8. Sure - no problem. That interest is going to bondholders who will now be richer. Just call it stimulus and everthing is fine.

    ReplyDelete
  9. Excellent piece based on substance and realistic assumptions (if conservative). It is just too bad that the WSJ chose not to include the CBO Baseline projections chart (including the "in-your-dreams" spending projections, not shown), it would make readers realize how utterly hilarious - and useless - the CBO is. I wonder if there is a serious economist anywhere who believes we will have a deficit of only 430 billion in 2015? Same with the 10-year debt levels in which case it is safe to assume that the $900 billion interest cost could in all probability be higher.

    ReplyDelete
  10. "All forecasts say long-term rates will rise soon."

    If that statement were true then the long-term interest rates would already have risen. :-)

    ReplyDelete
    Replies
    1. good call...

      Delete
    2. Funny. But do not see it as you do. You need to think in terms of probabilities. In fact, this phrase means that the probability is very high to have higher interest rates soon. But, of course, this is not certain.

      Delete
    3. That's if you assume there's no such thing as a liquidity premium. And no effects from collateral requirements -- both by counterparties and banking regs.

      Delete
  11. I don't think it will be as easy as anyone thinks for the Fed to exit the market. Can you see what happens on a Wall Street trading desk on the first day?

    Fed-"get me a bid for $75 billion"
    Desk-"what, checking." Covers mouthpiece on the phone and screams, "They are covering" to the trading room.

    Everyone texts their algos and the markets run.

    ReplyDelete
  12. Forgot to add, the Fed is going to get front run by all the banks they bailed out. Turkey shoot in the bond market.

    ReplyDelete
  13. $900 billion?

    Not counting inter-agency debt, you are assuming $18 trillion in outstanding federal IOUs all at 5 percent?

    I wonder if interest rates will ever go up again. They did not in Japan despite five years of QE (2001-2006).

    John Taylor gushed about the positive effects of that five years of Japanese QE in a paper he authored in 2006, and available on his website, btw.

    Is it imperative that the Fed sell its holdings of Treasury bonds? Or can they hold until maturity? If they hold until maturity (and funnel interest to the Treasury) they are, of course, essentially deleveraging the USA (assuming they buy Treasuries).

    Why not do this? So far, QE is associated with deflation, in anything. Since the Fed started QE, gold prices have cracked, oil dumped. Now, seven of the last nine CPI readings have been flat to down. Inflation is dead. The Fed doesn;t seem to cause inflation with QE, that's the observation, if not the theory.

    Given that QE worked in Japan and did not lead to inflation, even after five years, should we consider a 10 year program of QE, and deleveraging as we go along? Combined with minor inflation, that would effectively deleverage the USA.

    BTW, yet I favor a balanced federal budget, and I favor it with federal outlays at lass than 17 percent of GDP.

    But we live in a democracy.....

    ReplyDelete
    Replies
    1. “$900 billion?

      

Not counting inter-agency debt, you are assuming $18 trillion in outstanding federal IOUs all at 5 percent?”

      According to the referenced CBO report, the projected public debt in 2023 (that's excluding intra-government debt) will be $19.994 trillion. So, $18 trillion is actually an understatement unless it excludes the part of the "public debt" held by the Federal Reserve. And, I think the CBO is using a current law baseline in that report which would somewhat understate spending and deficits.

      The current average maturity of outstanding debt is about 65 months per a recent WSJ article and Treasury hopes to lengthen that. Also, interesting in that article was this quote which is relevant to Cochrane's post:

      “Fed Chairman Ben Bernanke last summer acknowledged that Treasury and Fed policies were at odds.

      "To the extent that the Treasury actively sought to lengthen the duration of its borrowing, it would to some extent offset the benefits of [the Fed's] policies," Mr. Bernanke said at a June press conference. But the impact of the Treasury's strategy won't wash out Fed policy, he said. "On the margin, the effects of the Fed's actions can be felt."”

      http://online.wsj.com/article/BT-CO-20130206-711894.html

      The CBO, per the referenced report and the above chart, has pegged net interest expense at $857 billion in 2023. Thus, it appears they are assuming a lower average interest expense on outstanding debt than 5 percent, but $857 billion is in the ballpark of $900 billion. Also, the rough average before the financial crisis (i.e. for years 2006 and 2007) was about 5 percent.

      http://www.treasurydirect.gov/govt/rates/pd/avg/2007/2007_06.htm

      Of course, a number of things could happen to change either the public debt or the average interest rate, both of which affect the level of “net interest expense”. If we suffer a recession before 2023, interest rates may remain low, but the public debt will exceed expectations. If the economy is strong, public debt may come in lower than estimated, but the interest rate would be higher…. $900 billion thus strikes me as somewhere between Sylla and Charybdis.

      I would, however, also welcome Cochrane’s thoughts on the numbers.

      Delete
    2. The Japanese QE policy has so many exemptions that it is not really comparable to the US. The most important one is that Japanese debt load has been mostly financed by domestic savings. Not such thing in the US. We still depend on foreign buyers of US Debt.

      Delete
  14. This might have been covered, but isn't it possible that big sales at 30 years might dramatically raise long rates and cause some kind of financial crisis by itself? I can see a sitution where the yields run away while they're selling causing huge losses for bondholders and causing some kind of bond death spiral that infects the whole curve.

    ReplyDelete
  15. Lengthening the maturity structure of the Treasury debt makes a lot of sense but will exacerbate the short-term deficit some ($100 billion?). The bigger issue which I would like to see discussed is how this debt will get repaid. There are really only a few possible ways.

    1. Raise tax rates. This has very bad effects on growth (see Alesina's work on this) and could actually make the debt worse.

    2. Lower spending. This may have negative short-run effects and thus is probably politically infeasible. But Alesina's evidence suggests that this is the preferred path to a fiscal stabilization.

    3. Inflate it away. I think the first few paragraphs of John's op-ed lay it out pretty clearly. The Fed will come under increasing pressure to keep buying the Treasury's debt. The result will eventually be increasing inflation with all of its real costs.

    I think that the Fed would love to be able to generate a little inflation right now to help reduce the real debt burden in the U.S. But it is a slippery slope and (expected) inflation gains made in the past came at a high cost. But the NK idea is to lower real rates enough to tilt the intertemporal see-saw towards current spending. The downside to this (IMHO) is that current monetary policy is simply accommodating fiscal spending. If you believe in large fiscal multipliers then this is all to the good, but if you think that government spending is so wasteful and inefficient that the actual multipliers are very small, then the current monetary policy is (potentially) doing double harm.

    4. There is one last path out and that is through higher growth. And the really crazy part is we know how to get the economy to grow faster, there is simply no political will to do it.

    ReplyDelete
    Replies
    1. Anonymous,

      "Lengthening the maturity structure of the Treasury debt makes a lot of sense but will exacerbate the short-term deficit some ($100 billion?)."

      Not necessarily. Your presumption is that the federal government must sell coupon securities to fund deficits. Meaning that government debt must make regular interest payments to persuade anyone to buy it. This is not the case. The US Treasury can sell accrual securities where the repayment of interest and principle are made at the time the bond matures ( 6 months from now to 30 years from now ).

      The interest payments affect the short term deficit only because they are regularly made (semi-annual payments).

      Delete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.