The uncertain links between monetary policy and inflation

Look closely at Chart 1 and ask yourself the following question: is there a clear relationship between the pattern of the blue line and that of the red line? More specifically, do changes in the blue line signal future changes in the red line? In a nutshell, does the blue line influence the red line?

For a nation currently obsessed with inflation and its consequences, these are important questions because the chart lines deal with probably the most controversial issue of monetary economics. Back in the inflation-ridden 1970s, this controversy had economists of rival schools of thought duelling with spreadsheets at dawn and once again it is highly topical because it asks, to what extent are changes in a country’s money supply linked to changes in its inflation rate?

The answer depends very much on the prevailing inflation rate when the question is asked. We can even fine-tune this by adding that the question is pretty well only ever put when inflation is rising and getting too high. So, through the 1970s it was posed increasingly often and, with increasing confidence, came the reply that, yes, money supply and inflation are linked; that without excessive growth in money supply inflation won’t accelerate, perhaps even that a rise in money supply causes inflation to rise.

Conversely, when inflation is quiet; when it’s bumbling along at a pace that’s hardly noticed, no one even bothers to ask the question let alone do this week’s exercise, to juxtapose in various ways money supply and inflation and see what emerges. Why would they? During benign periods the opposite notion increasingly gets bandied around, that money supply and inflation are so separate that printing money might be quite happily and safely used to tackle a variety of society’s ills.

Then, three years ago, came the response of governments around the world to the effects of Covid-19 and, before we knew it, a time shift occurred to take us back to the 1970s. Money supply surged and, before long, inflation did the same. Other things being equal, none of us – apart from economists – cares about money supply, but we do care about inflation; given the damage it does to our spending power, currently we obsess about it. Thus the vital question: how much does money-supply growth signal inflation, or even drive it?

For most of the past 30 years shown in Chart 1, you would be hard-pressed to spot a visual link where changes in money supply indicate, or even pre-date, a similar move in inflation, either up or down. From the early 1990s right through to the late 2010s, the two charts lines seem to ramble along in their own sweet way, largely oblivious of the direction of the other.

True, there are dramatic shifts in money supply on either side of 2000 and in the period 2016 to 2018. Yet these shifts had little discernible connection with inflation. In the money-supply spike on either side of the new millennium, they had no noticeable link at all. In the period 2016-18, there is a lazy rise in inflation that both pre-dates and trails the surge in monetary aggregates. It is followed by an equally lazy drop in inflation after the money supply plummets.

However, nothing compares with the symmetry of the twin peaks of 2021 and late 2022 when, first, the growth of the UK’s money supply surged to 15 per cent on the year. Eighteen months later that was followed by a rapid ascent to peak inflation of 11.1 per cent, as measured by the consumer prices index (CPI).

With symmetry like that, surely there must be a link, was the intuitive response. Sure enough, that was a cue to wheel out various ageing monetary economists, whose own peaks coincided with the Thatcher government’s monetary war on inflation from 1979 onwards. Their – predictable – advice was to say that the UK’s monetary aggregates should again be monitored closely, possibly even controlled; no matter that there remains plenty of evidence from the 1970s and ’80s that controlling monetary aggregates via restrictions on banks – some readers will even remember the so-called ‘banking corset’ – is counter-productive.

Besides, earlier this year a paper from three economists at the Bank for International Settlements, caricatured as the bank for central banks, was more circumspect. It found that sometimes there was a firm connection between changes in monetary aggregates and subsequent inflation, but not when the link was needed most.

The economists took long-run data (1951 to 2021) for money growth in excess of growth in national output for 32 mostly advanced economies (including the US and the UK). They juxtaposed those against inflation in a scattergram and did a basic statistical exercise to find out how well the dots joined up. The results were remarkable. “When the observations from all countries are pooled,” they say, “the standard relationship emerges clearly – there is a precisely estimated one-to-one link between excess money growth and inflation.” For every percentage point that money growth rises there is an equal rise in inflation and the data points cluster around their trend line so well that the R-squared, a measure of dispersion that we’ll return to, is 0.98 where 1.0 is perfection.

Then come the caveats. Economists are at pains to point out that their data does nothing to prove that changes to money supply cause changes to inflation. “We are concerned only with the signalling value of the monetary aggregates for inflation,” they say. Even that value may be limited since the signal fails when it is most needed. Hence their chief conclusion: “The strength of the link between money growth and inflation depends on the inflation regime. It is one-to-one when inflation is high and virtually non-existent when it is low.”

Failure may partly be because economists’ attempts to forecast inflation, of which there are many, tend to pay too little attention to monetary aggregates. At least, during the post-pandemic period, forecasts for inflation were worst for those countries where growth in excess money supply was highest. Which draws the economists’ final thought: “Might the neglect of monetary aggregates have gone too far?”

Given that everywhere in the developed world monetary policy – a quintessentially political matter – is in the hands of unelected technocrats, the obvious answer might be ‘yes’. On the other hand, the data for the UK alone makes it easier to say ‘no’. Since data for economic indicators is plentiful and regression analysis is easy with spreadsheets, Bearbull juxtaposed several indicators against the UK’s inflation rate, which, at 7.9 per cent for the basic CPI index and 7.3 per cent for the version including housing costs, is a developed-world outlier, though not in a good way.

To start, take the data from Chart 1 then shift the CPI data 18 months forward so that, in effect, both indicators peak in October 2022. With these two perfectly aligned at the most important inflationary juncture of the past 30 years, it might be reasonable to expect useful insights into the ability of money supply to signal changes in inflation.

Yet Chart 2 shows the depressing result. Based on 342 data points spread across 30 years, money supply’s signalling ability is useless. Okay, the regression line does slope upwards. In other words, it says that, on average, when money supply rises, inflation will also rise. Yet, as the chart’s widely scattered dots indicate, the predictive power of money supply is negligible. The R squared, mentioned earlier, has a value of little more than 0.1. That’s another way of suggesting that only a tenth of inflation’s changes are predicted by money supply, leaving the other nine-tenths predicted by who knows what.

Still, aligning money supply and inflation gets a better result than the unaligned version, as the table quantifies. That’s to be expected, though it’s still surprising that, without alignment, the regression line even slopes the wrong way, indicating that rising money supply will coincide with falling inflation, and vice versa. Using a broader definition of money supply, which includes credit creation, gets much the same result.

With producer prices and average earnings as the so-called independent variables, the exercise brings slightly better results. That’s especially so using output producer prices where the slope of the line is almost 0.4 and the fit of the dots to the regression line is over 0.6. Then again, what else would be expected? After all, output producer prices are basically consumer prices but without retailers’ markup.

So where does this leave us? Sure, inflation remains, as Milton Friedman’s famous quote suggests “always and everywhere a monetary phenomenon”. Yet that’s so general as always to be true. More helpful, if changes to money supply are hopeless indicators of future inflation it is better to know that than to labour under the illusion they are useful. A minor result, but let’s take it.

bearbull@ft.com