Recent Posts

Tuesday, November 28, 2023

Central Bank Balance Sheets

This article continues my series of articles on central banks as banks.

Central bank balance sheets (in the modern era, at least) are relatively simple. There is a split between banks with a currency peg and those without. After that, the key point to keep in mind that the minimum size of the central bank balance sheet is not under the control of the central bank — other actors create a minimal demand for their liabilities. The only freedom of action for central bankers is growing beyond the minimum, which they did not do before the days of Quantitative Easing (QE). The article finishes off with a discussion of consolidation.

This text overlaps material found in my book Understanding Government Finance.

Central Bank Liabilities

The key driver of a central bank’s balance sheet size is its liabilities. The central bank will have a small sliver of common equity, which in the modern era is owned by the central government. Since some central banks were historically private banks, there are a few cases where some common shares of the central bank are still traded on the stock market — but the central government typically owns the vast majority of the equity, and has control of the institution. There is a lot of silliness about the Federal Reserve being owned by private banks — they own preferred equity, which does not confer “ownership rights” (e.g.. a claim on central bank profits).

The main typical classes of liabilities for the central bank include (not including any trade credit they incur as part of their operations).

  • The central government, possibly other governmental entities keep deposits at the central bank. These balances are not part of the “monetary base” and are typically ignored in discussions. Since the central government owns the central bank, the deposit is just one arm of the government owing to an other, so its economic significance to outsiders is effectively nil. (This will be returned to when consolidation is discussed.)

  • Government-issued currency (banknotes) are typically liabilities of the central bank. (Private banknotes still exist, such as those issued by some banks in Scotland.) These banknotes can be returned to the central bank in exchange for a deposit liability (by banks). The amount of banknotes in circulation is driven by the needs of households and consumer-facing businesses (as well as the underground economy). Although the level of interest rates might influence the amount of money held as banknotes (they are assumed to do so in neoclassical models), nobody who has not been brainwashed by neoclassical theory believes that the central bank can directly determine the amount of banknotes demanded (and supplied by the government).

  • Private banks might need to hold deposits at the central bank as a result of reserve requirements (which are based on the amount of demand deposits). The amount of required reserves is fixed within a accounting period, and the amount of reserves required is driven by the balance sheet decisions of the private bank and its customers. Once again, even if interest rate policy influences the size of private bank balance sheets, it does so indirectly and with a lag — the amount of required reserves in an accounting period is not under the direct control of the central bank. The central bank has no choice but to supply those deposits, as otherwise one or more private banks must fail their required reserve test. That is, the central bank will have caused a bank failure(s) as a result of peculiar policy decision — which is not politically tenable.

  • Private banks might not trust each other, and decide to hold excess liquidity in the form of claims on the central bank (excess deposits, a positive balance with the payments system). Private banks doing this is a sign of a financial crisis — a sign that the central bank messed up regulating behaviour. These balances are quite obviously not under the control of the central bank.

  • Finally, the central bank can force excess deposits onto private banks (beyond what they need for required reserves and any crisis demand) via purchasing assets. (When the central bank purchases something via an electronic transfer, it creates a deposit balance to the bank of the recipient. That bank can send that balance to another bank, but it remains a deposit at the central bank. The only way to “delete” that balance is to exchange it for banknotes.) These are the only liabilities whose magnitude is under the direct control of the central bank. This means that when a central bank engages in “QE” it can inflate its balance sheet — but can only reverse back to whatever the minimum liability demand is.

  • Central banks might “borrow” money in open market operations to reduce reserve balances. This is typically structured as a paired sell-buy transaction on a bond, generically known as a repurchase (“repo”) transaction. This is a change of the mix of liabilities, but does not change the amount outstanding.

  • Central banks could issue bonds — but they generally do not. One way of ending the silliness of governments pretending that they have to listen to “bond vigilantes” is to have the central bank issue bonds. This is seen as unacceptable as it would put a lot of fiscal conservatives out of work.

  • Central banks owe each other money when they enter into swap line agreements. (Which I ignore for the rest of this article.)

If we look at the Bank of Canada for (most of) the 1994-2020 period, almost the only liability held by the private sector was banknotes — there were no “reserves.” This corresponds to the “simplified system of government finance” I described in Understanding Government Finance.

In summary, most of the liabilities are the result of the decisions of other actors, and the central bank has to supply them. They can crank up the size of their balance sheet beyond the minimum, but it is not clear why that is a good idea.

Assets — Depends on the Peg Status

What assets a central bank holds has an important dependence upon whether the currency is pegged. If the currency is pegged, it is possible for selected counter-parties to redeem the local currency for the peg asset. Since the central bank probably does not want the riff-raff showing up at the front door, this is going to generally be private banks or foreign central banks. In the Bretton Woods system, it was only foreign central banks that had the right to redeem U.S. dollars for gold, but older versions of the gold standard would extend the right to the private sector.

If your currency is redeemable for an external instrument — gold, hard currency — the central bank has little choice but to hold that instrument as an asset on its balance sheet for the redemption promise to be credible. Some academic monetarists argue that the central bank can pin the price of anything via “expectations,” but nobody in their right mind takes those people seriously. How much of its balance sheet needs to be held in the form of the peg asset is an exercise in psychology. If the peg is credible, nobody redeems their currency, and the currency needs little backing. But as soon as the peg credibility is questioned, the central bank needs a lot of the peg asset.

My expertise and interest is with developed countries with non-pegged currencies. (The euro is not really an exception — the member countries are pegged to a non-pegged currency. The member national banks face challenges that are both similar and different to a country with an external peg.) For these central banks, there is no need to hold a “peg asset,” although most have legacy holdings of gold, as well as foreign currency reserves (that are typically with a small weight relative to countries with currency pegs). I will now discuss their asset allocation decision.

Central Bank Assets For Non-Pegged Currencies

The main liabilities of central banks are as liquid as they can get — demand deposits, and bearer banknotes that are redeemable on demand. As a result, the central bank needs liquid assets if there was demand to reduce its balance sheet.

The natural match for the liability structure is a fixed income portfolio. And like a private bank, a central bank can have a good idea how many of its liabilities could plausibly be redeemed, so it can get away with extending the duration of its assets somewhat. That is, it has a portfolio of short-dated fixed income assets that it can sell at close to their balance sheet carrying value to meet redemptions, and another portfolio with more duration that could be liquidated slowly if necessary.

We then run into the question: should the assets be debts issued by the central government or the private sector? What type of fixed income instruments?

This used to be a controversy, leading to ideas like the “real bills doctrine” (which I am going to let the reader look up on the internet if they are not familiar with it). However, World War II and mainstream economic thinking led to central banks holding balance sheets that were 100% central government bonds and bills, as well as repurchase agreements on those bonds. Then, the Financial Crisis of 2008 hit, and central banks ended up buying a variety of private sector assets (of varying dubiousness) as part of a programme of “de-risking” key parts of the financial system.

I might return to the debate about the purchasing private sector assets by the central bank later. I will just finish off this line of thought with the concept of seigneurage (which has alternate spellings “seigniorage” and “seignorage”). The rate of interest on currency notes is 0%, as also was the case for required reserves at the Federal Reserve. The central bank is holding a portfolio of fixed income assets that people hope earn a positive rate of interest. This implied that the bank should run a steady profit — at least as long as the New Keynesians with their negative interest rates are kept away from the rate decision committee. This profit was called seigneurage, harking back to the cut the sovereign took when precious metals (including foreign coins) were minted into coins at the royal mint. However, this is not a profit created by the minting of new money, rather it is the carry generated by the central bank portfolio versus its 0%-costing liabilities. (The advent of QE has meant that central banks were forced to pay interest at close to the policy rate on deposits at the central bank, turning them into de facto overnight bills issued by the central bank.)


Consolidation is an accounting term referring to merging the balance sheet of a wholly-owned subsidiary onto its parent company. (Wholly owned means that the parent company owns 100% of the common equity of the subsidiary.) Instead of presenting two balance sheets, the two balance sheets are added together — but with intra-company entries netted out. This netting needs to be done to avoid double-counting entries, and to ignore “ghost debt” that is purely an intra-company affair.

  • If the subsidiary has $100 in equity, that would show up as a $100 asset on the parent’s non-consolidated balance sheet, which would imply a corresponding $100 in equity in order for the balance sheet to in fact balance. If we just merged all the balance sheet entries, that $100 equity would show up twice, as it appears on both non-consolidated balance sheets.

  • If the parent lends money to the subsidiary (or vice-versa), it might matter for tax purposes or in a bankruptcy, but otherwise, it is the parent company lending money to itself. If the subsidiary defaults on the debt, so what? The parent owns it already.

Any time you pick up the balance sheet of a multinational corporation, you are looking at a consolidated balance sheet. Even bush-league multinationals will end up with thousands of corporate entities across multiple jurisdiction due to the magic of international tax accounting and financial engineers run amok.

We then get to central banks. Modern developed central banks are wholly-owned subsidiaries of the central government (or almost wholly owned in oddball cases like the Bank of Japan). Under generally accepted financial accounting standards, the central should be consolidated with the central government. This means that intra-governmental debts — like deposits — should be netted out to zero. And for the purposes of economic modelling, we should consolidate since the intra-governmental transactions have no effect on the rest of the economy.

We can interpret seigneurage in two ways.

  1. If we do not consolidate, the central government issues debt, the central bank buys some in order to supply money, then the central bank returns the carry profits to the central government. This carry offsets some of the total debt issuance.

  2. If we consolidate, we just look at the amount of government debt and money held by the private sector, since the central bank’s holdings of government bonds nets out with the associated liabilities of the central bank. The interest expense is reduced relative to the gross issuance of debt, some of which was bought back by the central bank. There is no need to add in “seigneurage income” into projections of government revenue — we just calculate interest expense based on the mix of money/debt held by the private sector.

There is no doubt that consolidation is cleaner and eliminates problems in economic models. One of the issues with mathematical modelling of the economy is that there is a large number of variables to track, and non-consolidated balance sheets add yet more variables. Furthermore, we need to add behavioural relationships that determine the intra-governmental transactions. The problem with that is that those transactions have no effect on the rest of the economy, and so there is no means to pin down those behavioural relationships.

Please note that the previous statements are less true if the central bank is managing a currency peg. In that situation, the central bank can default (on its redemption pledge), and so we need to model the default process. Similarly, the only reason to care about the non-consolidated balance sheet for a non-pegged currency is if you can come up with a plausible default scenario (which I generally cannot).

Nevertheless, whether or not one can consolidate the central bank with the central government is a stupid economic debate. The objections generally come from critics of Modern Monetary Theory (even though other people like central bankers will consolidate for convenience) based on crackpot theories or bad faith attacks. As stated, it is a stupid debate — I certainly can consolidate the central bank, because I know how accounting works. Consolidation is not necessary for any conclusions, since a well-defined model will give exactly the same results for the non-governmental variables regardless of whether you consolidate or not. You consolidate because it is easier, and you do not do it if you are worried about the central bank/government defaulting.

Up Next?

I have run through most of the basics, but I have at least one more background primer before I get to more exciting central bank debates.

Email subscription: Go to 

(c) Brian Romanchuk 2023

No comments:

Post a Comment

Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.