Monday, July 4, 2011

Ron Paul, Mankiw, and the Government Debt

Ron Paul recently suggested a solution to the debt-ceiling problem
"We owe, like, $1.6 trillion because the Federal Reserve bought that debt, so we have to work hard to pay the interest to the Federal Reserve," Paul said. "We don't, I mean, they're nobody; why do we have to pay them off?
Paul's logic is the following. The debt ceiling applies to all Treasury debt outstanding, but currently $1.6 trillion of the total $14.5 trillion in Treasury debt is held by the Fed. We all know that the Fed is a large leech sucking our blood, so why not call a temporary truce in the debt-ceiling/budget war by simply defaulting on the $1.6 trillion in debt held by the Fed?

Dean Baker likes the idea. He also seems to be thinking like the Fed Chairman of 1937 in that he thinks an increase in reserve requirements would be a great idea. This guy doesn't like Paul's idea as he thinks it would cause a hyperinflation.

Now, along comes Mankiw to the rescue. He poses the problem as an exam question, then gives you the solution, in a few lines. What Ron Paul is suggesting is only an "accounting gimmick," and it is actually irrelevant.

Now, to work through this, consider two alternative scenarios. First, suppose that, in the absence of Paul-default, the Fed holds the Treasury debt it has acquired until maturity. In this case, clearly Paul-default cannot make any difference. Without Paul-default, the Treasury makes the payments on the Fed's Treasury holdings to the Fed, and the Fed sends those payments back to the Treasury. With Paul-default, the net flow between the Fed and the Treasury is the same: zero.

The second scenario is the interesting one. Suppose that, in the absence of Paul-default, the Fed were to sell the Treasury debt before it matures, in a reverse-QE2 program. Under Paul-default, reverse-QE2 is not possible, as the Treasury has defaulted on the Fed's debt and therefore no one else wants it. According to Ben Bernanke and, more recently, Jim Bullard, QE2 works (worked), i.e. purchases of long-maturity Treasury securities by the Fed moves asset prices and increases the inflation rate. If QE2 works, then reverse-QE2 works too. Thus Paul-default, by foreclosing reverse-QE2, matters.

Not so fast, though. Actually the working hypothesis should be that QE2 was irrelevant (see this too), no matter what these people say. In that case, it's not going to matter whether the Fed holds its Treasury portfolio until maturity, or not. Mankiw is right, but the exam question is a little harder than he thought.

28 comments:

  1. The increase in reserve requirements substitutes for the lack of reverse QE2, so Mankiw has taken that into account. He's saying that the proposal is irrelevant except that using reserve requirements would be less efficient than reverse QE2.

    ReplyDelete
  2. "The good news here is that, since QE2 is irrelevant, the Fed can reverse it without cost."

    The real effects of QE2 were nil, but it drove people's expectations in the right direction.

    http://www.econbrowser.com/archives/2011/07/the_effectivene.html

    Is it just me or does that sound like perfect monetary policy?

    ReplyDelete
  3. anonymous 1,

    You could ask him, but my interpretation is that Mankiw is saying that the two ideas (Paul-default and reserve requirement increases) are goofy, independently. Reserve requirements do not "substitute" for lack of reverse-QE2. You can induce banks to hold any quantity of reserves if you make the interest rate on those reserves high enough. A binding reserve requirement just induces a distortion and resulting welfare loss.

    anonymous 2,

    Yes, one interpretation of QE2 is that it just hammered home the idea that the policy rate would at 0.25% for an extended extended period. I'm not sure I would call that perfect. Buying $600 billion in Treasury debt seems a funny way to get that idea across.

    ReplyDelete
  4. "Reserve requirements do not "substitute" for lack of reverse-QE2. You can induce banks to hold any quantity of reserves if you make the interest rate on those reserves high enough"

    They subsitute in the sense that they are an alternative way of controlling inflation, as is increasing the interest on reserves. Are there any drawbacks to paying interest on reserves? Does it not lead to less intermediation in the same way as reserve requirements?

    ReplyDelete
  5. The usual argument for paying interest on reserves is that it improves efficiency. For example, there is a Friedman rule argument that says it would be efficient if you could pay interest on all outside money, including currency. But paying interest on currency is not feasible, so you go for second-best, which is paying interest on the reserves. You can't make a case for a reserve requirement being good for anything, other than creating a captive demand for some government liabilities. As Mankiw says, a reserve requirement is a tax, and I think we can conclude that, among the array of taxes we have available, the reserve requirement tax is not one we should be using.

    ReplyDelete
  6. I agree with Mankiw and (I believe) you that this wouldn't matter much. Obviously a reserve requirement is a bad idea, I don't know of any coherent argument in favor of it.

    What I find very strange about this policy being proposed by Ron Paul is that he has argued that the Fed is an enabler of government spending and big deficits (I think that's wrong, but that's one of his arguments). But isn't Paul's recommend policy here just have the government issue bonds, the Fed prints money to buy the bonds, and then we just destroy the bonds. But, how would this be different than just the Fed directly financing government spending by printing money? So, isn't Paul's implicit policy recommendation that we just switch to a money-printing financed government? Maybe Paul would say this is just a one time deal and so we aren't switching to the regime I described, but it certainty opens the door to that type of regime.

    Also, in response to some of the comments about QE2 above. Why would QE2 act as a signaling device about the future policy rate path? If QE2 doesn't have real effects in and of itself and it's costless to reverse, how could it possibly be a commitment device? If there is no consequences for reneging, why would it be credible?

    ReplyDelete
  7. Ted,

    1. The story of Paul's ideas is one of inconsistency and confusion. We know he does not understand seignorage, for example, and how to measure it. Apparently he does not understand accounting either.

    2. Yes, exactly. It's hard to see QE2 as a commitment to anything important.

    ReplyDelete
  8. @Ted

    "Why would QE2 act as a signaling device about the future policy rate path? If QE2 doesn't have real effects in and of itself and it's costless to reverse, how could it possibly be a commitment device? If there is no consequences for reneging, why would it be credible?"

    For all intents and purposes it appears QE2 did act as a signaling device. Credibility aside, it seems to me the more important question is: if it was irrelevant, why did the markets react at all?

    I guess the easiest answer is: markets think we live in Bernanke's world and markets react accordingly. Reality be damned.

    ReplyDelete
  9. If inflation can be controlled by increasing interest on reserves, what stops the government financing all its spending via seignorage? Money supply continually increases, but is held as reserves due to increasing interest rate.

    ReplyDelete
  10. "if it was irrelevant, why did the markets react at all?"

    1. We usually don't like going down this road, but I guess you have to ask the question. Suppose that, indeed, what you see in the asset prices, pre-QE2, was a reaction to what Bernanke was saying. Maybe the market participants were as clueless as Ben was about QE2's actual implications.

    2. How do you know that what you were seeing in asset price behavior was a reaction to what Fed officials were saying? There was a lot of other stuff going on.

    "what stops the government financing all its spending via seignorage?"

    Good question. On average, the yields on long-maturity Treasuries are higher than interest rates on T-bills. Issuing reserves is just like issuing T-bills, so if the consolidated government effectively just financed its deficit by issuing reserves, then the interest payments on the government debt would be a lot lower. For a government that is worried about being pushed into default, it can help if the government debt is long-maturity, so that it does not have to be rolled over during a sovereign debt crisis. However, if you think that no one would ever worry about the US government defaulting, why doesn't the US government just borrow short and roll the debt over?

    ReplyDelete
  11. Ted, you ask if "Paul's implicit policy recommendation [is that] we just switch to a money-printing financed government?" Yes, it is a step in that direction. But there is no essential difference between debt-financing and money financing. Remember that the money we are talking about here is interest-bearing central bank deposits. Functionally interest-bearing deposits are equivalent to T-bills, i.e. they are F-bills. I’m not sure if Mr. Williamson agree with me here, but in my view the prerequisite for inflation control would be the same as now: a) a policy interest rate that tracks the market-clearing real interest rate plus a spread for the inflation target and b) appropriate fiscal backing, i.e. a present value of expected real primary public surpluses that equals the real value of the debt at the price level that correspond to a). Milton Friedman once (1948) proposed a monetary-fiscal policy reform along the lines you think rightly follows from the Paulian logic. You'll not end up in hyperinflation.

    ReplyDelete
  12. Torgeir,

    This sounds like fiscal theory of the price level. What's the "market-clearing real interest rate?" Is this like Woodford's Wicksellian natural rate?

    ReplyDelete
  13. Stephen, yes it's the fiscal theory of the price level combined with a Fisherine approach to inflation control. And yes, the real market clearing interst rate is the Wicksellian short-term natural rate, which I think is also what Wooodford has in mind.

    ReplyDelete
  14. Sometimes when I hear John Cochrane talking about the fiscal theory, it seems he is saying that monetary policy, i.e. asset swaps by the central bank, never matters for anything. Everything, including the inflation rate, is driven by fiscal policy. I don't think you are saying that, but maybe you are.

    ReplyDelete
  15. Stephen, let me share with you and your readers my comments to Tyler Cowen's post on Greg's post
    http://marginalrevolution.com/marginalrevolution/2011/07/greg-mankiws-exam-question.html

    First I wrote this:

    Tyler, I cannot ignore your sloppy scholarship on monetary issues. It’s too much.
    I’m from Minnesota where I got my Ph.D. But I’m Argentine and I have seen all kinds of gimmicks to deal with fiscal crises. Contrary to what Greg M. says and to your nonsense comment, the Treasury can collect the seignorage from the issue of currency and therefore Paul’s proposal should be modified to limit the reduction in the amount of Treasury debt accounted as a FRS’s asset to the amount of currency in circulation, that is, to around $1 trillion at the end of June 2011. Now you should explain to your readers why my proposal makes sense and they should ignore what Greg and you wrote.

    Second, in a comment to one of MR reader's comment I said:

    I understand your frustration. It’s the same frustration that serious economists should feel about the failure to explain clearly monetary issues. Unfortunately most economists are still addicted to the theoretical approach of monetary aggregates that combine means of payments with financial assets. Today in all economies there is a government-issued currency and for accounting purposes it is a liability of a central bank, but since this currency is no longer backed by gold or any commodity, for accounting purposes the double-entry principle requires to enter an asset (in the Fed they call it U.S. Treasury bill but I prefer to call it confetti). The modified proposal I suggested in a previous comment amounts to replace the confetti with toilet paper for accounting purposes. It doesn’t have any economic effect because the seignorage was already collected by government (in the past 100 years). The only effect is to circumvent the debt ceiling because someone made the mistake long ago of considering the confetti as debt when it was not government debt –of course, let us hope that accountants and lawyers agree that toilet paper is not government debt, otherwise it would be stupid to replace the confetti.
    BTW, seignorage is the rent that the issuer of currency earns at the time of issuing it. You cannot earn rent twice on currency already issued (in other words, if you need additional rent, you have to issue more currency).
    Also, don’t get excited by Paul’s proposal because most likely there will be a simple agreement to increase the debt ceiling by a small amount, enough to finance the government for one or two months.
    It’s unfortunate that Tyler has failed consistently in explaining monetary issues.

    Finally, to avoid mistakes about the Fed's holdings of U.S. Treasury securities as reported in the Fed's H.4.1. stat form (as of June 29, they amount to $1.6 trillion), let us make clear that the many classes of securities aggregated in that entry are not perfect substitutes and perhaps they are not even close substitutes. I don't know the details and composition of that aggregate but since currency in circulation (as of June 29, $1.0 trillion) is lower that the total amount of those securities, I'm willing to argue that there is a lot of confetti reported as securities.

    ReplyDelete
  16. Stephen, Woodford’s paper “Monetary Policy in the Information Economy” convinced me that monetary policy is about nominal interest determination and that central bank assets swaps in public debt essentially are irrelevant for inflation control. Reading Leeper, Woodford, Sims and Cochrane on the fiscal theory of the price level, and trying to integrate this with the Woodfordian conception of nominal interest rate determination, has led me to think of monetary policy as fiscal policy in disguise.

    Here is an attempt at a brief explanation: Central banks steer inflation by determining the nominal interest on the public debt. Public debt is privately held governmental nominal debt and central bank liabilities to the private sector. Nominal interest expenses on public debt are continuous helicopter drops of public debt, and such transfers are inherently fiscal. The price level then has to grow gradually over time, at a pace determined by the differential between the policy rate and the Wicksellian natural rate, in order to keep the real value of the public debt equal to an assumed constant present value of public real primary surpluses. (Implicitly I’m assuming that all public debt is overnight debt; it gets more complicated with long term debt).

    I too get the impression from reading Cochrane that he thinks of monetary policy as asset swaps. If such assets swaps are irrelevant for inflation, then it seems difficult to explain how monetary policy can steer inflation. It might just be semantics, that is, it might be that Cochrane thinks the active ingredient in central bank policy is its fiscal aspect. But in his “Understanding the Great Recession”, he writes that “changing the interest on reserves from 0 to the overnight rate [i.e. begin paying interest on central bank deposits and using this interest rate as the policy rate] is exactly the same thing as a[n] … open-market operation”. This, if I understand him rightly, is wrong. Federal Reserve interest expenses to the private sector is not an asset swap, it is a transfer.

    I said that central bank asset swaps are “essentially” irrelevant for inflation control. In the pre-October 2008 US regime open-market operations were necessary in order to create appropriate scarcity of federal funds and keep the interbank federal funds rate on the target rate. If what I’ve tried to explain above is correct, though, this was just a roundabout way of controlling inflation by indirectly determining the public sector’s nominal interest expenditures. The post-October 2008 regime makes the causal links more transparent.

    ReplyDelete
  17. 1. I'm not sure how Woodford convinced you that monetary policy is about nominal interest rate determination. In Woodford's model that's just an assumption. In practice, of course, monetary policy is about asset swaps. The central bank is a financial intermediary, and the way it intervenes is by issuing some liabilities to buy some assets, or selling some assets and retiring some liabilities.

    2. When Cochrane says “changing the interest on reserves from 0 to the overnight rate [i.e. begin paying interest on central bank deposits and using this interest rate as the policy rate] is exactly the same thing as a[n] … open-market operation," that's exactly right. In the US system, when the stock of excess reserves is zero overnight, an open market operation moves the overnight rate (the fed funds rate). When there is a positive stock of excess reserves, the interest rate on reserves determines the overnight rate, open market operations are irrelevant, but changing the interest rate on reserves is equivalent to conducting an open market operation when the quantity of excess reserves is zero.

    ReplyDelete
  18. "First, suppose that, in the absence of Paul-default, the Fed holds the Treasury debt it has acquired until maturity. In this case, clearly Paul-default cannot make any difference. Without Paul-default, the Treasury makes the payments on the Fed's Treasury holdings to the Fed, and the Fed sends those payments back to the Treasury."

    I don’t think I agree with this. The Fed pays interest income earned on Treasury holdings to the Treasury, but not principal. Upon maturity, the principle is paid by the Treasury to the Fed, the debt extinguished, and corresponding reserves extinguished. Nothing flows back to the Treasury. On the other hand, in a default by the Treasury, upon maturity there is no principle paid to the Fed and, as such, no extinguishment of reserves. The flows are very different; in the first, the Fed enjoys a net inflow (so as to net out assets and liabilities) while in the second the Fed endures an outflow (there is no netting out of assets and liabilities, only a writedown of assets). It would be very dangerous to think these operations were identical.

    ReplyDelete
  19. JP,

    Three alternatives:

    (i) No Paul-default. The debt matures. The Treasury makes a payment to the Fed. The Fed sends the payment back to the Treasury. The Fed does not retire any liabilities.

    (ii) No Paul default. The debt matures. The Treasury makes a payment to the Fed. The Fed retires reserves.

    (iii) Paul default. The debt matures. No payment to the Fed.

    These are all identical. Where's the danger you have in mind?

    ReplyDelete
  20. Stephen, some of your readers are asking questions that I asked 50 years ago to my first professor of monetary economics in Buenos Aires. He couldn't answer them clearly so I've been "working" on them since then --first as student, then as a professor, and finally as an adviser to central banks. Indeed it'd take too long to give a detailed explanation of how monetary systems of payments have changed over time, from commodity money to fiat money, and what a government monopoly of fiat money implies.

    My 2011 version goes like this. Based on the functional approach to financial intermediation (as attempted by R. Merton and his MIT colleagues) I separate the payments function from all other functions of financial intermediaries. Then, based on Fama's distinction between monetary and accounting systems of payments, I argue that financial intermediaries provide only accounting systems.

    For a long time now governments have monopolized monetary systems of payments in all countries. Because it is legal tender and fiat money, any government-issued currency is not government debt, and therefore once it is issued, it is expected to be used and accepted as a means of payment. When government pays employees, beneficiaries and suppliers with money just issued, the stock of currency in circulation increases and there is no expectation that the government will have to service a debt because of that issue. To simplify, let us assume that the Currency Office is responsible for the new currency. It'd be no different from the Confetti Office responsible for providing small pieces of colored paper to celebrate the country's Great Leader. Following a long tradition economists say (or should say) that the government's budget revenue has increased in the amount of seignorage.

    Now to make things interesting let us assume that there are three state banks (I love this part because I worked several years in the reform of China's state banking system). One is to manage a Government-Security Fund in which deposits and borrowed funds from the public are invested in a variety of government securities. The second is to manage a Cherry-Picking Business Fund in which deposits and borrowed funds from the public are invested in a variety of private businesses. And the third one is to manage a Sovereign Wealth Fund in which deposits and borrowed funds from the government and the public are invested abroad in a variety of financial assets. Indeed, we can justify the creation of these three state banks on the basis that they pursue different objectives --say, the first one to manipulate interest rates, the second to reward cronies, and the third one to manipulate the exchange rate of the government currency (FYI, in China the purpose of the third one is to provide liquidity to the domestic banking system not to manipulate the exchange rate as most economists claim). Albeit disguised in many ways, today these three state banks ARE part of the financial systems of most countries.

    So how effective can monetary policy be? Well, there are at least two definitions of monetary policy. One refers only at the flow-supply of currency, that is, at how the Currency Office generates seignorage to finance government expenditure and therefore it is part of fiscal policy. The second one refers to how the three state banks are managed --that is, their objectives and the specific instruments used to achieve them. Unfortunately there is no other name that can express exactly what this second definition implies (banking or financial policies have a much broader scope).

    Once we agree on how to simplify correctly that significant part of the modern economy, perhaps we may have a good discussion about monetary policy.

    ReplyDelete
  21. All I’m saying is that (i) doesn’t exist. That’s not how monetary operations work.

    That leaves (ii) and (iii). In (ii) the Fed gets a payment and retires the corresponding debt held on the asset side of its balance sheet. In (iii) it doesn’t get a payment, the debt does not get retired (it goes bad), and a hole appears on the Fed's balance sheet.

    The danger is that holes in the asset side of a bank’s balance sheet destroy the quality of that bank’s public liabilities. i.e. inflation results.

    ReplyDelete
  22. JP,

    "All I’m saying is that (i) doesn’t exist. That’s not how monetary operations work."

    Wrong. The asset matures. The Fed debits the account of the Treasury with the Fed. Then the Fed does whatever it wants. It's just accounting.

    Holes in the asset side of a bank's balance sheet? You'll have to explain as I have no idea what you mean.

    ReplyDelete
  23. I think the difference between us is that I see Peter stealing from Paul, whereas you see Peter stealing from himself.

    From my perspective, (i) and (iii) are the same. (ii) is the only one in which there is quid pro quo and a proper netting out. In all my scenarios, Paul-default is bad for anyone holding US dollar denominated assets.

    ReplyDelete
  24. JP,

    This has no consequences at all for the resources extracted from the private sector by the public sector, or how those resources are extracted.

    ReplyDelete
  25. From the perspective of security analysis, it most definitely has consequences. The "public sector" disaggregated is no more than a collection of public debt-issuing institutions, some of whom have promised (explicitly or implicitly) to guarantee/backstop/fund each other. Because the quality of that promise is never certain, it is the job of the security analyst to handicap all scenarios so as to properly price each institution's issued debt. From a securities analysis perspective, a fiscal transfer from the Fed to the Treasury strengthens the liabilities of the Treasury relative to the Fed.

    James Hamilton commented on your post yesterday:

    http://www.econbrowser.com/archives/2011/07/ron_pauls_debt.html

    ReplyDelete
  26. "From a securities analysis perspective..."

    Maybe you need to change how you think about securities analysis.

    ReplyDelete
  27. "Because the quality of that promise is never certain, it is the job of the security analyst to handicap all scenarios so as to properly price each institution's issued debt. From a securities analysis perspective, a fiscal transfer from the Fed to the Treasury strengthens the liabilities of the Treasury relative to the Fed."

    This makes little sense -- if both institutions are fundamentally guaranteed by the same people (the federal government's taxation powers), then it doesn't matter who "looks stronger." This is the problem with noneconomist-businessman types -- they don't understand arbitrage.

    ReplyDelete
  28. "Maybe you need to change how you think about securities analysis."

    Stephen, that's a good bit of advice that I adopted a long time ago. I'm always evaluating different ways of thinking about securities analysis so as to improve my investment performance, suggestions always appreciated.

    ReplyDelete