Too Much Money Can Exacerbate Inflation

The sharp acceleration in money supply and inflation in the aftermath of the pandemic has reignited a debate over the direct role money plays in driving faster price growth. Central banks once paid close attention to money supply. From 1975 to 1982, the Bank of Canada (BoC) directly targeted growth of M1 – a monetary aggregate consisting of currency and bank chequable deposits.¹

The Bank of Canada has long since moved away from targeting money supply as an intermediate step in controlling inflation, opting for direct numerical inflation targets.² While the central bank has largely removed money from its lexicon, ignoring it completely has its downsides. Empirical analysis shows a persistent long-run relationship between money supply and inflation that has stood the test of time. In certain circumstances, it can also be helpful in explaining shorter-term movements. Inflation's relationship with money supply is most evident in situations in which price growth has been pushed out of its narrow "low-and stable" corridor. ^3,4

Money supply surged well before the run up in post-pandemic inflation, but it is now moving in the opposite direction. More narrow definitions of money (M1+; please see Text Box 1 for definitions) have been shrinking for the past four quarters, while broad money (M2++) declined for the first time on record in the second quarter of this year (Chart 1).

A forecasting approach that includes the typical drivers of inflation, as well as money supply suggests that the surge in money supply that contributed to its acceleration will now be helpful in moving it back to target. By removing some of the extensive liquidity provided to the economy in the aftermath of the pandemic, the Bank of Canada is showing its commitment to its 2% target and the policy actions necessary to attain it.

A number of factors conspired to doom central bank's past experiments with monetary targeting. For one, they focused on relatively narrow categorizations of money. Both the Bank of Canada and the Federal Reserve put their focus on "M1" measures that only included chequing deposits and cash outside of banks. The motivation was to focus on narrow money that exhibited the highest sensitivity to interest rates, and thus to the central bank's policy tools. However, financial innovations and changes in product offerings made these relationships increasingly more tenuous.

Focusing on broader monetary aggregates would have reduced some of these earlier forecast misses but even more inclusive aggregates, have had limited success in predicting inflation over the last several decades. In large part this reflects the success of central banks in anchoring inflation expectations and maintaining inflation near their official 2% targets. This credibility had made predicting inflation easier – inflation has, over a relatively short period of time, moved to whatever the central bank targets it to be.

Money Matters (Again)

The more recent period, in which changes to the overnight rate gave way to large increases in Quantitative Easing (QE) and therefore money supply, has thrown this assumption – that the best forecast is simply 2% – into question. In order to shield the economy from the impact of public health measures, policy makers coordinated fiscal and monetary policy in order to provide direct monetary supports to individuals, a welcomed move. At the same time that deficits were expanding by north of 10% of GDP, policy rates were pushed to zero and the BoC engaged in QE, purchasing large amounts of government bonds.

The interventions made by the Bank of Canada and the Federal Government during the pandemic resulted in an immediate surge in both narrow and broad definitions of money. In the second quarter of 2020, M1+ grew by over 50% annualized, while broader monetary aggregates grew close to 20%. Initially inflation remained quiescent, even as a relatively swift economic recovery took place in the second half of the year. However, as nominal spending power moved above pre-crisis levels, price growth began to pick up, peaking in 2022 at a near double-digit rate .

Today narrow money is declining, down 7% from its peak, while broad money growth has slowed considerably. Consumer price inflation has followed suit with a lag, registering a print of 3.5% year-over-year in the second quarter of 2023 (Chart 2). The sequencing of these events is enough to suspect a causal link between the two variables, but it does not prove that they are related. We must consider whether money adds anything to a forecast for inflation when other contributing factors have also been considered.

Including Money Better Explains Post-Pandemic Inflation Acceleration

To test the hypothesis that money supply can matter for inflation, we run two variations of a statistical model of inflation estimated over various sample periods and compare their relative forecast error. The modelling approach we take is a vector error correction model (VECM). The VECM approach allows us to test for both a long-term relationship between economic variables as well as short-term dynamics that move variables from this long-run anchor. We consider all of the variables included in a standard inflation model: the output gap, commodity prices, the Bank of Canada policy rate, and lags of inflation itself.⁶ The only difference between the two models is the inclusion or exclusion of broad money (M2++).

Estimation of the VECM is carried out in two steps. In the first, we test for a long-term (cointegrating) relationship between money supply and inflation. This shows a significant positive relationship between money supply and inflation, as predicted in the economic literature. We then construct the error correction terms from these estimated cointegrating relationships and estimate a vector auto regression (VAR) in first differences. Statistical tests also show a significant positive short-run coefficient of money supply on inflation. Further details of our modelling approach are contained within the Appendix.

Finally, we examine a standard measure of forecast error (Root Mean Square Error) between the VECM including money supply and the one excluding it.⁷ Overall, we find that the model with money supply performs better (Table 1) over the full sample. However, during the period of inflation targeting prior to the pandemic, the model without money performs just as well as the model with it included. In the period immediately prior to the pandemic to the present (we must go back to 2018 to have enough data history for the model to run), the model with money included outperforms once again.

Table 1: Root Mean Square Forecast Error

The Bottom Line

The failure of inflation to materialize after the 2008 downturn contributed to economists downplaying the potential for money supply growth to contribute to future inflation. As growth in money surged well past anything experienced historically, focus remained on the balance of aggregate demand and supply, even as these became increasingly difficult to measure in real time.

The recent experience demands a more careful investigation. Our analysis shows that while money did little to improve model performance over the period prior to the pandemic in which the Bank of Canada targeted inflation (1992-2019), its inclusion improves it in the most recent period. With a good amount of lead time, the surge in money supply in 2020 helps explain the 40-year high in inflation that followed.

The good news is that the Bank of Canada has been successfully targeting inflation since the late 1990s. The success and credibility of the central bank gives confidence that inflation will return to its 2% target. As such, the decline in money supply over the last several months does not imply that the economy is headed for deflation, but rather that the BoC has taken the steps necessary to reduce it. Assuming they continue on this path, we will soon return to a world in which the best forecast for inflation is the central bank's 2% target.