My recent post, Endogenous Money, A Reconsideration, has sparked quite a bit of interest. In a nutshell, the money-creation view of banks applies only to assets with zero-duration such as credit lines and credit card debt. When it comes to assets with duration (eg, 30-year fixed-rate mortgage), banks truly intermediate between borrowers (duration-suppliers) and savers (duration-buyers). I‘ve received several very good questions on Twitter. I will use this post to respond in more detail.
- What term point separates money markets from capital markets?
This is a very important question! It determines which A/Ls expand the money supply (and as such, give rise to NGDP fluctuations) and which ones don’t.
Repos are collateralized loans that pay fixed interest over some term. The longer the term, the greater the duration. To classify repos as part of either money markets or capital markets, we need to know if banks hedge the respective duration. Clearly, overnight repos are part of money markets because the interest indexes daily (effectively, zero duration). However, A/Ls with near-zero duration such as 1-month, 2-month or 3-month are more difficult to classify and may vary from bank to bank.
That’s because banks do not perfectly duration-match assets and liabilities and often macro-hedge the balance sheet. In other words, banks can retain a certain amount of interest rate risk, provided it’s within pre-defined tolerance limits. A 3-month repo has so little duration that it may not push the bank duration profile outside its tolerance limits, hence the bank may choose to leave it unhedged. If that’s the case, a 3-month repo would be considered a money-market instrument, meaning that banks fund such repos with newly-created money ex-nihilo.
A more definitive classification would involve a survey of the duration profile of all banks in the economy and the respective interest-rate-risk tolerance limits. In aggregate, the unhedged duration on bank balance sheets marks the dividing line between capital markets and money markets. A/Ls with duration less than that would be money markets; A/Ls with greater duration would be capital markets.
I am not aware of such survey and would love if someone could point me to one. Understanding bank duration profile, especially tracked over time, will reveal to what extent bank duration appetites drive the business cycle.
2. Repo markets connect capital markets with money markets. What is the macro significance?
I would classify repos in three categories based on the respective borrower. I don’t see any macro significance when banks borrow from repo markets or issue other types of collateralized debt. It is just a way of minimizing the cost of bank liabilities. From the standpoint of the macro economy, deposit liabilities and bank repo liabilities are perfect substitutes.
If the borrower is a non-bank that enjoys explicit or implicit bank guarantees, it is basically a bank that is engaged in regulatory arbitrage, aka a shadow-bank. Such off-balance sheet exposure allows banks to operate below leverage and funding ratios mandated by regulators. From a macro standpoint, skirting regulatory requirements increases the fragility of the banking system, a risk that became painfully evident during the Great Financial Crisis. Since then, regulators have attempted to clam-down on such off-balance sheet exposure.
I think the third repo category is the most interesting one. Repos allow non-banks, such as REITs, to use leverage to buy assets with duration. This risk appetite will be there with or without repos — repos are just an efficient way to deliver leverage to such risk-seekers. The interesting part is that repos have short duration. For practical purposes, repos can be considered zero-duration financing for assets with duration. From a macro standpoints, the relevant question is whether the non-banks that finance with repos hedge the resulting duration mismatch. If they don’t, this willingness to take on duration risk has both pros and cons for the macro economy.
On the positive side, willingness to borrow at a variable rate of interest provides the raw materials for money creation by banks. For the most part, such borrowings are very short-term (eg, credit lines to finance production of inventory or credit card debt). However, repos provide a long-term source for money creation because the underlying assets have longer term. Such long-term source of money has very important stabilization effects because it can satisfy asset money demand (to read more, please refer to this post: Alternative View of the Business Cycle, Part II).
On the negative side, repos being short-term instruments, require constant refinancing. This increases fragility and systemic risk. If, for whatever reason, such refinancing pipeline were to be disrupted, you end-up with asset liquidations and market busts. This is what happened during the Great Depression when home financing relied on short-term, variable-interest loans.
Australia makes an interesting case study. Home purchases in Australia are financed with variable-interest mortgages. Just like repos, such mortgages are zero-duration liabilities providing banks with a long-term supply of the raw materials needed for money creation. However, unlike repos, such mortgages do not require constant refinancing. In other words, Australia has found a way to both maximize the benefits of repo financing and minimize the downside (to read more, please refer to this post: Do expansions die of old age and the curious case of Australia).
3. Does the duration of the central bank balance sheet matter for monetary policy transmission?
The main take-away from the proposed framework is that monetary policy transmits through the central bank balance sheet. Basically, by absorbing duration, the central bank enables banks to finance duration borrowers (be it public or private) with new money creation, making the respective spending additive to NGDP (to read more, please read this post: Thoughts on Fed Targets and Tools).
The question is what drives policy transmission of central bank asset purchases— asset duration or balance sheet size. In other words, does it make a difference if the Fed buys 3-year or 10-year Treasuries?
This is a very difficult question and not an easy one to answer, and I have to admit I am still thinking through it. For starters, the duration of central bank purchases must be greater than the duration of commercial bank balance sheets, which is the threshold I discuss in point 1 above. To illustrate, if banks are comfortable financing 3-month T-bills with new money creation without hedging, the central bank stepping-in and buying such near-zero duration securities makes no difference for monetary condition. Simply, money creation by the central bank substitutes for money creation by private banks.
However, for assets purchases with duration greater than the threshold, the proposed framework suggests that the actual duration does not matter. The only thing that drives policy transmission is the size of the central bank balance sheet. Basically, interest rates ensure that duration markets (aka, capital markets) are always in equilibrium, which means that for every willing borrowing, there must be a willing saver (for empirical evidence, please refer to these two posts: The Empirical Case for Crowding-Out and Japan — evidence of crowding-out and the importance of central bank balance sheets).
In conclusion, the framework suggests that whether the central bank buys 10 vs 3-year duration may affect the shape of the yield curve, but not NGDP. That’s because a disequilibrium between desired borrowings and savings, which in turn gives rise to NGDP fluctuations, can only occur in money markets. Which is why only the liability side of the central bank balance sheet matters, not the duration of its assets.
4. Why does an interest rate swap cancel the money created upon the origination of the loan? What does going long or short money mean?
Most of the comments to my original post are related to this question. Let me try to explain in more detail.
First, I will point-out that banks must find duration buyers before originating loans with duration. For example, as soon as a bank enters into an interest rate lock for a 30-year fixed-rate mortgage, it hedges that commitment by selling duration in capital markets — either by selling bonds in forward markets or by entering into an interest rate swap and/or swaption. This fact alone proves that banks are not in position to create money ex-nihilo when it comes to assets with duration. In other words, the endogenous money view does not apply.
Let’s take a closer look at who the duration buyers are. In a forward bond sale, the seller agrees to sell and the buyer agrees to buy a bond in the future at a pre-determined price. This means that the duration buyer must have sufficient money balances, in the form of demand deposits, to pay for the bond. In the absence of the bond purchase, she will continue to receive variable interest on her money. In terms of risk exposure, she is long money, which means she expects to receive variable interest on some notional. However, she doesn’t want variable interest. She prefers locked expected return, which is why she enters into the forward bond purchase. Once the forward bond trade settles, the bank delivers a bond, and she pays for it with her money. Effectively, she sells her money back to the bank in exchange for the bond. As a result, her money is destroyed, and she is left with the bond. This money destruction cancels 1 for 1 the money created upon the origination of the mortgage.
Alternatively, the duration buyer could enter into an interest rate swap. In an interest rate swap, one party agrees to pay variable interest on some notional in exchange for fixed interest from the other party. The variable payer is short money, because she commits to paying variable interest, and long duration, because she expects to receive fixed interest. In the example above, the duration buyer passes the variable interest she receives on her money holdings to the bank in exchange for fixed interest. The swap allows her to synthetically own a bond without taking on bond-specific risks such as credit risk or option cost.
The question at hand is how the swap changes her money position. While the money continues to persist, it is effectively locked-out of circulation. To illustrate why, think in terms of interest rate risk exposure. Her money balances mean she is long money, but the swap means she is also short money. Basically, her long and short positions in money cancel out, leaving her with net zero exposure to money as if she had none. If she were to spend the money on something else, she will be taking on interest rate risk because of her continued obligations under the swap. There is no reason for her to take on such risk (or as I call it, have net short position in money). This is why I say her money is locked-out of circulation. In other words, the swap “synthetically” cancels her money.
The chart above illustrates how interest rate swaps effectuate duration transfers. Agents A uses a credit card to buy $1,000 worth of goods from Agent B. The Bank facilitates by debiting A and crediting B. The corresponding entries on the bank balance sheet are $1,000 zero-duration asset (the credit card loan) and $1,000 zero-duration liability (Agent B’s deposit). Since there is no duration mismatch between the bank’s assets and liabilities, no hedges are needed.
Agent B now has $1,000 deposit that pays variable interest, but she would like to lock her expected return. Agent C would like to borrow $1,000 at a fixed rate of interest. Agent B could buy a bond from Agent C, but she is not comfortable taking on credit risk. The Bank is comfortable with the credit risk but is not willing to originate a fixed-rate loan unless it can transfer the duration to someone else. That’s where the interest rate swap comes-in. Agent B and the Bank enter into a swap whereby Agent B agrees to pay to the Bank variable interest on $1,000 notional in exchange for fixed interest. The Bank is now in position to originate the fixed-rate loan to Agent C. The Bank adds $1,000 duration asset (Agent C’s Promissory Note) and $1,000 zero-duration liability (Agent C’s deposit). The duration mismatch on the bank’s balance sheet is offset by the notional assets and liabilities related to the interest rate swap.
At the end of the day, there are $2,000 of outstanding bank liabilities, corresponding to the combined deposits of Agent C and B. It appears as though the Bank did create new money upon originating the fixed-rate loan to Agent C. However, after accounting for the notional liability of Agent B under the interest rate swap, only $1,000 of unencumbered money holdings remain. Agent B’s money is synthetically cancelled by the variable leg of the swap, meaning that Agent B has committed to not spending that money for the duration of the swap. Another way to make the same point is to say that Agent B’s notional liabilities give rise to $1,000 of incremental money demand that offsets the $1,000 created upon the origination of the fixed-rate loan.
Several readers have objected that swaps involve notional balances that do not have to be pre-funded. True, but duration buyers are hedging long-money positions elsewhere. The aggregate capacity of the non-bank private sector to pay variable interest on swaps is limited by the existing money supply. In other words, people use swaps to hedge and don’t just take on variable interest rate risk for no reason. Furthermore, even if someone were to go naked on the variable leg of the swap, without any offsetting long-money position elsewhere, they have to post collateral, usually in cash. A tiny movement in rates could wipe that position, unless they keep posting more collateral, ultimately forcing them to fully fund the swap.
We can argue how much risk appetite is out there for the variable leg of the swap, but that is not the point. The point is that it is this risk appetite that constrains bank money creation with respect to assets with duration. In other words, the endogenous money view does not apply. The private non-bank sector must go short money first before banks are in position to create money later.
