Saturday, March 17, 2012

Bruegel on the Target2 vs Interdistrict Settlement Account debate

The Belgium-based think take Bruegel recently linked to my guide to the Federal Reserve Interdistrict Settlement Account. In my ongoing attempt to ensure the Target2 debate doesn't founder on faulty comparisons to the Interdistrict Settlement Account, here's another post.

Although they have a better grasp on how the ISA works than most, the authors Michiel Bijlsma and Jasper Lukkezen do make an important error:

The important difference between Target 2 and ISA, however, is that in the US all Reserve Banks are owned by the federal government. This means that in the US it is possible to safeguard the integrity of the system by changing the settlement rules. This is as exciting as a game of monopoly among friends. As all Federal Reserve banks are owned by the federal government, a loss in Richmond is irrelevant when there is an equal gain in New York. In the Eurozone, however, the ECB is owned by the national governments via the national central banks, not by the European Union as a whole. When one would change the settlement rules here – for example by discounting claims – this means a transfer across countries.
In actuality, all twelve Federal Reserve district banks are owned by their member private banks. What allows the Fed to change settlement rules is not any ownership position in the Fed (they have none) but the system's constituting articles, or Federal Reserve Act, which put the preservation of the par value of US dollar-denominated banking liabilities above the necessity of the system settling.

As I pointed out in response to commenter John Wittaker here, if the Federal Reserve system was to be dissolved and there existed significant imbalances between districts, final settlement might not be guaranteed. That's because a district Fed's credit is only as good as the credit of its member private banks. If these member were asked to re-capitalize the district bank to enable it to achieve settlement, but they were unable to do so, it would cause problems.

So in that respect, Bijlsma and Lukkezen  are wrong to say that a loss in Richmond and a gain in New York is ultimately irrelevant. When all the chips are down, these imbalances represent transfers between shareholders of district Reserve banks. For instance, if the Richmond Fed failed to settle and its shareholders, including Bank of America, could not support a recapitalization of that district, then the New York Fed would have a massive loss on its hands. This would require member banks like JP Morgan to re-capitalize it. Thus, in comparing the ECB settlement mechanism to the Federal Reserve mechanism, it's not fair to say transfers do not occur - all that can be said is that these transfers are distributed across a different spectrum of actors.

Bijlsma and Lukkezen also have a similiar article here though they don't attribute to the Money View credit for the initial observation nor myself for the underlying research. They note:

Every year in April the average ISA balance over the past 12 months is calculated and netted via transfer of gold certificates between reserve banks.
This also is not entirely correct. Settlement via SOMA transfers, not gold certificates, has been the standard operating procedure since the 1970s.


  1. The Bruegel piece makes a major mistake re:the Interdistrict Settlement Account.
    This settlement has 2 parts, one related to the amount of federal reserve notes outstanding settled once a year according to a formula based on the average daily balance of these notes over the previous year, this involves a minor adjustment in gold certificates, SDR and securities posted as collateral to cover the federal reserve notes.

    The second, much larger settlement (particularly since the Fed pays interest on reserves and the system carries large excess reserves) relates to the portfolio backing reserves, every year the capital key based on the ratio of each reserve bank’s capital and surplus to aggregate capital and surplus is recalculated based on preceding December 31 and a settlement occurs in early April with the new key, that new key is then used every day to allocate assets resulting for open market operations until the next "settlement" i.e calculation of a new key based on the previous december 31 capital position of each reserve bank. The ISA never goes to zero, in normal times (no excess reserves) reserves redistribute naturally roughly in line with member banks capital, in non-normal time (now with vast excess reserves), reserves tend to stay with the money center banks in NY resulting in a surplus in the NY district, and leave the Richmond district, home of the GSEs that selers of fed funds as the fed can;t pay them IOR by law and to a lesse extent the SF district, home of the largest FHLB, also a seller of fed funds.

  2. The Bruegel piece got its facts from my earlier post, see here.

    The ISA is explained on pg 136-138 of the Federal Reserve's accounting manual with a numerical example.

    I've never heard of settlement of the ISA being based on a capital key. You're welcome to show me where in the manual it occurs, but I've already combed it quite closely and haven't stumbled on such an explanation.

    You're right that one component of settlement is related to outstanding notes, specifically each district bank's gold-to-note ratio relative to the system's average. But as my earlier post pointed out, to best understand the ISA it's easiest to assume that each district Fed’s gold-to-note ratio is always equal to the system average.

  3. This is in about every annual report by any reserve district:
    "Activity related to Treasury securities, GSE debt securities, and Federal agency and GSE MBS, including the premiums, discounts, and realized gains and losses, is allocated to each Reserve Bank on a percentage basis derived from an annual settlement of the interdistrict settlement account that occurs in April of each
    year. Activity related to foreign currency denominated assets, including the premiums, discounts, and realized and unrealized gains and losses, is allocated to each Reserve Bank based on the ratio of each Reserve Bank’s capital and surplus to aggregate capital and surplus at the preceding December 31."
    An easy to see to that this settlement occurs daily is to look at H.4.1 report and check for example the allocation of Central bank liquidity swaps.
    The example in the manual is specific to the (once a year) settlement of federal reserve notes and the equalization of gold certificates, there is also an equalization of SDR certificates and treasuries collateral. You can check the H.4.1 and you will see that gold and SDR certificates do not change during the year.

  4. Read pgs 138-139 of the manual, the pages just after the section I suggest you read. It discusses what you are talking about.

    So yes. Foreign assets are apportioned according to capital ratios. Thanks for teaching me that fact. But domestic SOMA securities are not. They are apportioned according to the method described in the manual on pgs 136-138.

    You write that pgs 136-138 describe the yearly settlement of Federal Reserve notes. Yes, that is partly true. But by far the most important part of interdistrict settlement described therein relates not to note imbalances, but payment imbalances. These payment imbalances arise because of different flows of electronic payments and cheques between districts. These imbalances are settled in April by reallocations of the domestic assets in SOMA.

  5. Looking at table 9 of the H.4.1 line "Reverse repurchase agreements" shows that these open market operations are re-allocated daily as per capital key.

  6. I appreciate that this less than clear from the manual, but the data speaks for itself.

  7. Hi jck. I don't disagree that reverse repurchase agreements, which comprise of foreign-currency denominated holdings, are re-allocated daily, and that these are done on the basis of some sort of capital key.

    But these reallocations are not associated with the clearing and settlement of interdistrict flows of payments. Interdistrict flows are settled once a year via reallocations of each district Reserve bank's portion of US-denominated assets held in SOMA. The district banks can't settle with foreign assets.

  8. "Reverse repurchase agreements" have nothing to do with foreign-currency denominated holdings, they are pure open market operations in dollar draining excess (dollar) reserves.
    Looking at the H.4.1 shows that it is clearly the case that the "Securities held outright" (SOMA) are re-allocated daily according to the capital key (see table 9: Statement of Condition of Each Federal Reserve Bank). The balance of the SOMA re-allocated daily is net of "Collateral held against Federal Reserve notes" on table 10, hence the calculation is not so obvious.

  9. Daily allocations have nothing to do with clearing and settlement of the credits and debits owed by district Reserve bank to each other arising from interdistrict payment flows, which is the essence of the Target2 debate. This argument needs to be conducted on the basis of what is written in the manual, you've got too many unsubstantiated claims. I will repeat. Read pgs 136-138. They explain everything.

    For instance... pg 137: "The resulting percentage of each Bank's participation in the System Account is used, until the next reallocation, as the basis for allocating the daily SOMA transactions."

    Couldn't be clearer.

  10. It is very clear indeed that the SOMA adjustment explained in the manual applies to federal reserve notes collateral and in the example given SOMA securities (2000) are clearly equal to federal reserve notes outstanding (2000).

    1. Hi jck!

      It is pure coincidence, that in the example the amount of the SOMA securities is equal to the federal reserve notes outstanding. In the reality it is never the case, for example, as of 14. March 2012:

      SOMA securites: 2.604 trillions
      federal reserve notes outstanding aka currency in circulation: 1.095 trillions



  11. Thanks for pointing out the inaccuracies in our description of the workings of the ISA. We will correct this (of course giving due reference to your work).

    As far as I understand it now, there are two essential differences between ISA in the US and Target 2 in the EMU:

    (1) The federal reserve board can force district banks to rediscount their claims. This possibility does not exist in the EMU.

    (2) In the US the redistribution (if any, because 'the chips are down' should also imply the federal reserve board refusing to force district banks to rediscount) is between private firms instead of countries' governments as in the EMU. This prevents such redistribution from becoming politicized.

    1. Hi Michiel, thanks for stopping by and also for referencing my work.

      1) Yes. Here is the wording in Section 11 of the Federal Reserve Act:

      "Rediscounts by One Reserve Bank for Another
      ...To permit, or, on the affirmative vote of at least five members of the Board of Governors of the Federal Reserve System to require Federal reserve banks to rediscount the discounted paper of other Federal reserve banks at rates of interest to be fixed by the Board of Governors of the Federal Reserve System."

      In my original post, I had mused that this power had been removed. But that is not the case.

      2) Yes, that's how I read it, see <a href=">here</a>.

      Perhaps "less politicized" would be the safest way to describe it - politics has a way of creeping into everything. Also, there is FDIC to consider - deposit insurance. That may muddy the picture, but I can't get my head around it at the moment.

    2. Thinking about this overnight, the fact that the Board can force district banks to rediscount makes it more similar to the Target2 system, since it makes the necessity of settlement a less immediate proposition.

  12. Sorry, it's not a coincidence or an accident, that's how it works in normal times. For most of the life of the Fed, the collateral backing federal reserve notes outstanding has been a huge proportion of SOMA securities, 95% or higher up to a few years ago, so in normal times, the "settling" of the (SOMA -fed notes collateral) portfolio is hardly noticeable, but not now (that was my point about IOER and excess reserves as (one of the) causes of ISA imbalances) and the weekly H.4.1 data ( supports the view that the (SOMA -fed notes collateral) portfolio is "settled" (daily) as per capital key as (badly) explained in various annual reports.

    Re: target2 another angle