The power of monetary policy to affect interest rates and exchange rates depends on the downward slope of the demand function. This column uses the Chinese experiment with parallel currencies to study the impact of sudden increases in money supply. The authors find causal evidence that increases in money supply lead to currency depreciations, and use this to quantify the interest elasticity of reserve demand. The results can be used to understand how the People’s Bank of China maintained the peg between the mainland and parallel currencies.
How do interest rates and exchange rates respond to changes in money supply? These are perhaps the two most central questions in monetarism. With efficient financial markets, demand curves for assets are horizontal, so changes in supply have no effect on their relative value. If money is special, because it provides some service or performs some function, then its demand curve will slope down with its price, whether that price is measured as the opportunity cost of other investments (the interest rate) or the relative value of other currencies (the exchange rate). The power of monetary policy to change the supply of money and affect interest rates and exchange rates (and from these, inflation, output, or other variables) rests on the downward slope of this demand function.
At first sight, it may seem that the answers are obvious. After all, open market operations that increase the supply of bank reserves are traditionally the bread and butter of how central banks lower interest rates. Periods when the supply of money rises quickly are also periods when there is a sharp devaluation of the currency relative to goods or to foreign currencies (see Ward et al. 2019).
And yet, over the last 15 years, very large increases in bank reserves through quantitative easing and tightening have had almost no impact on interest rates. In the ‘ample reserves’ system in which the ECB, the Fed, and most major central banks have been for a long time, the demand for money is saturated and so it is horizontal. In fact, modern monetary policy sets instead the interest rate on these reserves to control the other interest rates, independently of the quantity of money (Reis 2016).
Moreover, as Figure 1 shows, the actual correlation between monthly money growth rates and exchange rates for countries that peg their exchange rate is very close to zero. Pinning down the slope of the demand for money is hard because central banks change money supply almost always in response to shocks that also shift the demand function. In turn, changes in the other country’s monetary policy themselves respond to these and other shocks, making it very hard to control for all that is going on.
Figure 1 The missing link between exchange rates and M2 money growth relative to the US for currencies under a peg (26 countries from February 1979 to December 2015)
Source: Bahaj and Reis (2024a).
The Chinese experiment with parallel currencies
In Bahaj and Reis (2024a), we look to the People’s Bank of China (PBoC) and its experiment with parallel currencies to answer these classic questions. Following its 2009 plan to internationalise the use of the renminbi, the Chinese authorities promoted an offshore currency, the CNH, that can be freely used by anyone for trade, credit, and settlement. The CNH exchanges one-to-one with the mainland currency CNY (and with RMB banknotes) but this exchange is subject to strict limits and restrictions that are imposed to enforce capital controls in and out of mainland China. To prevent these parallel currencies from disappearing under the weight of Gresham’s law, the authorities must keep the market exchange rate between CNY and CNH very close to parity (Bahaj and Reis 2024b).
CNH as a means of payment may seem exotic, but its monetary properties are not so far from the US dollar. Its velocity in 2017-23 (our sample) has been 431. By comparison, the average velocity for the US, using Fedwire transactions and M1 less currency, was 450 on average between 2012 and 2019. In turn, the ratio of CNH deposits to CNH reserves was 3.7. The equivalent money multiplier for the US was 13.3 over 2004–06 (when money was scarce) and 1.6 over 2021–23 (when reserves were ample).
The supply of CNH then offers a useful lab to investigate the classic monetarist questions. This CNH money supply is solely devoted to keeping the peg, while CNY monetary policy pursues many other independent goals.
Rare exogenous transitory increases in the supply of money
Operationally, the People’s Bank of China controls the supply of CNH reserves by issuing CNH bills and redeeming and reissuing them in auctions on a regular, approximately monthly, schedule. Whenever a CNH bill is redeemed, then the CNH money supply goes up; when a new bill is issued, the CNH money supply goes down. The mechanism is the same as in many open market operations around the world, even if here it involves central bank bills as opposed to government bonds. The supply of these bonds is in Figure 2. Noticeably, there are some peculiar short-lived wiggles in this supply. Why?
Figure 2 Stock of CNH bills outstanding and its short-lived fluctuations caused by the shift in maturity structure in June/August 2019 and in November 2020
Source: Bahaj and Reis (2024a).
When it started issuing CNH bills in November 2018, the PBoC planned to have a certain stock of three-month and 12-month bills. Yet, over the next few years it changed both the size and the composition of these bills between three-month, six-month, and 12-month bills. While these changes may have been endogenous, they shifted the maturity structure. Because the auctions for different maturities are on a different schedule, they created nine future dates when certain bills exogenously rolled off and were only replaced five trading days later for five of them, and 25 days later for the other four. The dashed lines in Figure 2 show these dates, where sharp and large declines in the stock of bills outstanding come with temporary increases in CNH money supply. We can use these exogenous changes to measure the elasticity of the demand for money.
The response of the exchange rate to exogenous changes in money supply
Figure 3 shows their impact on the exchange rate between CNY and CNH. Beyond the focus on high frequency, by comparing CNH with its twin parallel currency, the CNY, we can have some confidence that the results are not due to confounding factors that affect the overall value of the yuan during these days. The figure also shows a bootstrapped placebo distribution from other days in the sample to assess statistical significance and distinguishes between the roll-offs reverted in five days or longer.
Figure 3 Response of the CNY-CNH exchange rate to money supply shocks measured as 100 times the cumulative change in the log of the daily exchange rate from the trading day prior to the bill roll off, averaged over the events
Source: Bahaj and Reis (2024a).
On average, the CNH depreciated by 11 basis points during the first ten days. This was in response to an increase in the supply of CNH of ¥11 billion. Since we estimate that the average stock of CNH reserves is ¥196 billion, the elasticity of money demand with respect to the exchange rate is therefore (11/196)/0.0011 = 51. This says that a 1% increase in the exchange rate must be met by an increase in the supply of reserves of approximately 51%, or ¥100 billion to keep the peg.
This number may seem large, but three observations are in order. First, CNH bank deposits during this period were on average ¥730 billion, and RMB banknotes were above three trillion. The required growth rate in M1 is much lower. Second, this is in an environment where CNH and CNY are close substitutes and arbitrage is active. Third, the standard deviation of the CNY-CNH exchange rate during 2017-23 was only 21 basis points, and we find that 64% of the typical transitory demand shocks dissipates by itself within five days. Therefore, keeping the peg in response to the typical shock only actually requires CNH money to move by a few billion.
The interest rate semi elasticity of the demand for reserves
Using a model, we show that this exchange rate elasticity is a weighted sum of the two interest rate semi-elasticities of money demand: the demand for reserves by banks, and the demand for deposits by households. The time series literature has focused on pinning this latter elasticity (see Weber et al. 2017). Using their estimates, we estimate that the semi-elasticity of CNH reserve demand is 48.
We dig further and use a measure of shocks to money demand and an instrument to see how they shift at a monthly frequency the stock of CNH deposits. This provides an alternative estimate of this semi elasticity, using a completely different source of variation. The estimate is 47.
A current debate in advanced economies is by how much a decline in 1% in bank reserves would raise the opportunity cost of reserves, measured by the gap between the interbank rate and the interest rate on reserves. Our estimate says that the answer is 1/47, or two basis points. By comparison, Lopez-Salido and Vissing-Jorgensen (2023) use US data during the ample reserves regime to estimate the answer to be 0.2 basis points. Afonso et al. (2022) use a different identification strategy and a time varying model. In the 2009-12 period, their estimate for the US dollar is two basis points, very close to ours for the CNH. As they discuss, this elasticity is central to understand the optimal size of the central bank balance sheet.
Conclusion
The Chinese experiment with parallel currencies provides a clean identification strategy for the impact of exogenously increasing the money supply on the exchange rate. We discussed it here, and what it implies for key elasticities in controlling the money supply and interest rates. In Bahaj and Reis (2024a), we then use these as an input for the anatomy of the peg between CNH and CNY in a model of money creation and liquidity management by banks.
Source : Voxeu
