- Money growth has surged in most economies, the result of aggressive quantitative easing.
- If money growth is a harbinger of inflation, that would be concerning.
- In high-inflation 1970s and 80s, money growth was believed inevitably to cause inflation.
- However, experience since 2000 does not seem to support that view.
- Fast-growing money and low inflation can coexist for long periods.
There is concern, especially in the financial community that the recent rapid growth of the stock of money in many economies presages faster inflation – if not immediately, then in due course.
Financial markets are worrying about inflation accelerating
Such a view was prominent in the 1980s in particular, when extremely rapid inflation in most OECD economies was accompanied by fast growth of the main monetary aggregates – as it is virtually always bound to be – and when, in addition, it was quite widely reckoned that the direction of causality ran primarily from money to inflation.
Taken further, that view credits central banks with the generally low inflation that has prevailed since. That could be right: but a counterargument is that central banks did not cause the slowdown of inflation but merely presided over a slowdown that had other causes. Those included intense competition in both labour and product markets as a result of brisk technological progress; and increased globalisation, not least the admission of China into the international trading system.
Hence the question being asked today: does the post-2008, post pandemic, surge in the money aggregates in many countries risk leading to an upsurge of inflation?
And see recent rapid money growth as a particular concern
The basic concern is not so much with spikes in the price level, whether caused by bottlenecks that result from changes in trading relationships (such as Brexit, or a more general move to reduce dependence on China), or changes in the structure of demand (such as may arise from the pandemic, or policies in respect of climate change). Rather, the concern is with the possibility of a sustained, generalised inflation, of the sort that changes inflation expectations permanently.
Concern is amplified in financial markets because even an inflation blip, especially when combined with continuing uncertainty about its future course, introduces a significant inflation risk premium into bond yields, and puts some other asset prices under pressure. Bond yields went to exceptional lows in anticipation of recession, so a bond market sell-off could happen before an economic upturn, without requiring a sustained, substantial rise in inflation.
The procedure that we have adopted to consider this issue in the current context is to look at the history since 2000 in four countries – the US, the UK, the Euro19, and Japan.
We have looked at money growth and inflation post 2000 …
… but not in the UK, Euro19, or Japan
Basic time series data for the US, the UK, EUR19, and Japan are shown in Figures 1 through 4. All four sets of data tell basically the same story. In each case the two series of course have upward trends. But money growth has consistently outstripped nominal GDP growth; and more importantly there is no obvious tendency for one series to influence the other.
In case appearances are deceptive, we ran a statistical test, regressing the (logarithm) of quarterly nominal GDP on: a time trend (to eliminate the trend dimension); one lag of itself; and two lags of the log of M3 (one and two quarters back). The money terms were found to be not significantly different from zero. 1
The evidence that money, particularly broad money, is not currently a good leading indicator of inflation is supported by the fact no money aggregate now appears in the OECD’s composite leading indicator for these four monetary jurisdictions. In fact money appears in leading indicators for only five of the OECD’s 36 countries.
These indicators are compiled without theoretical prejudice on a ‘what-works’ basis. Series are detrended and cyclical components examined to find any ability to forecast other series. If they work they are in; if they add nothing to other indicators they are out. The absence of money aggregates is thus the dog that doesn’t bark.
A second approach
Our second approach was to consider the evolution of broad money (as measured by M3) relative to that of base money (as measured by M0).
The thinking here is that, when a central bank engages in quantitative easing, it thereby creates bank deposits and reserve deposits of banks at the central bank. Hence the first-round consequence will be that M0 and M3 increase by similar absolute amounts. However, because M3 is typically a multiple of M0, the ratio M3/M0 will fall.
Whether or not the policy becomes really stimulative depends on what happens thereafter:
- If the banks start to lend aggressively then, through the monetary multiplier, M3 will start to rise relative to M0. Hence if that ratio is observed to rise, it implies either that the central bank has eased back on QE, or that bank lending has picked up, or some combination of the two. Either way the policy can, in some basic sense, be held to have been successful – or, at least, to have accompanied a successful outturn. 2
- If however, the ratio M3/M0 does not rise – or more tellingly, if it continues to fall – this implies that the central bank has continued throughout with QE, and/or that bank lending has not picked up materially. Either way, the policy cannot be held to have been particularly successful. 3
QE spawned a modest increase in US lending …
What the data show is that indeed the ratio M3/M0 did fall, in all four countries/regions, both after the financial crisis and again recently. But there is a difference between what happened thereafter, in the US on the one hand, and in the UK, the Euro area, and Japan on the other:
… but not in the UK, Euro19, or Japan
- The US. Following a steep fall in the ratio as the central bank aggressively bought government debt following the 2008 financial crisis, M3/M0 bottomed in late 2014. Thereafter M3/M0 rose. QE was proceeding intermittently, so the rise in the ratio could coincide with some tailing off, but could also indicate that banks managed to increase their lending somewhat. (See Figure 5.)
- The UK. The ratio began declining from a very high level in 2006 before falling steeply with the onset of QE. There was a brief rally in 2010 and 2011 before further QE precipitated a continued decline. (See Figure 6.) 4
- In Euro19. There has been a downward drift in the ratio, irrespective of QE, which began only in March 2015. It was suspended at the beginning of 2019, when the ratio stabilised, but there has been no meaningful upturn. (See Figure 7.)
- Japan. QE ran from 2001 to 2006. There was a blip in the ratio when it stopped, but no rise, just a drift thereafter. When QE resumed in 2013 the ratio fell again: and, particularly importantly, it has continued to do so. (See Figure 8.)
While there were brief periods of upturn, the ratio in all countries trended down overall over the post-2000 period. The expansion of central bank money – commercial bank reserves – has evidently not triggered a commensurate increase in bank lending to the public. On the face of it, monetary policy has failed to achieve the desired objective.
While money is at least a permissive factor for inflation, growth of base money is and of itself evidently not sufficient to produce a sustained inflation; that typically requires a wage-price-wage spiral, as employees resist real-wage reductions.
That has not happened since the mid-1980s, as unionisation has fallen in the private sectors of all Western countries; as the proportion of the economy accounted for by the manufacturing sector has diminished; as participation rates have fallen in some; 5 and as wages have either stagnated or grown slower than labour productivity. In those circumstances it is not surprising that QE has inflated asset prices without causing a more general inflation.
We therefore conclude that the rapid growth in base money seen in a range of countries as a result of QE after the 2008 Global Financial Crisis does not, in and of itself, presage a sustained increase in inflation. Moreover, the probable rise in unemployment following the Covid epidemic makes a strong wage response even less likely. And the chances of such an inflation in either Japan or the Euro area seem particularly remote. Even a modest pick-up in nominal GDP growth, however, could well trigger turbulence and rotation in asset markets.
Temporary increases in prices as a result of bottlenecks during recovery would not invalidate our basic conclusion. What would falsify it would be were inflation to pick up in a sustained way
without there having been a major increase in bank lending that led to overall pressure on productive capacity and wages. So watch for:
- Bank lending and broad monetary aggregates rising relative to base money.
- Capacity utilisation; and rising job vacancies that may precede wage pressures.
- An increase in earnings or rising wage share in GDP, betokening pressure in labour markets.
- Sustained pressure on commodity prices, particularly ‘hard’ commodities or oil.◼
Our Associate Gerald Holtham is Hodge Professor of Regional Economy at Cardiff Metropolitan University, and Managing Partner at Cadwyn Capital. Prior to that he was a Fellow of Magdalen College, Oxford, Chief Investment Officer at Morley Fund Management, Director of the Institute of Public Policy Research, Chief Economist at Lehman Brothers, and Head of the General Economics Division in the OECD Economics Directorate. He is a graduate of the University of Oxford.
1 Regression result for United States:
Sample, quarterly 2000-2020. Dependent Variable log(GDP) current prices.
2 Unavoidably there is the age-old problem of the counterfactual: that it is not possible to be certain what would have happened in the absence of the quantitative easing.
3 Of course, credit creation via the banks is only one possible transmission channel from QE to economic activity. Purchase of bonds raises their price and depresses bond yields, flattening the yield curve. This could encourage others to borrow via the bond market for mortgages or corporate investment.
4 The M0 data used in this Figure have been constructed, by the authors, from two Bank of England series. This was necessary for the following reason. Since 2006 the Bank of England has been publishing the data needed for this exercise, viz. notes and coin plus what the Bank calls “Banking Department sterling reserves balance liabilities”, monthly and non-seasonally adjusted). However, before 2006 published M0 consisted essentially only of notes and coin: hence it was necessary to add in bank reserves for the pre-2006 period. These are available on the BoE’s Research datasets, but only on an annual basis. Hence it was necessary to interpolate these monthly data and then add them to the pre2006 monthly data for notes and coin. Any error introduced by the interpolation procedure is likely to be slight: the reserves in that period averaged only about 0.5% the value of notes and coin.
5 In the United States, for example, the civilian labour force participation rate has fallen by over 5 percentage points since its peak around 2000, and the employment-to-population ratio has fallen similarly. And union membership has declined steadily, since 1983. (In that year, the earliest for which strictly comparable data are available, union membership rate was 20.1% of total wage and salary workers. In 2020, the union membership rate was 10.8%.) See U.S. Bureau of Labor Statistics, 2021. Union Members Summary. 22 January. Available at [Accessed 10 February 2021]. For a discussion of the evolution, see Dunn, M., and Walker, J., 2016. Union Membership In the United States. US Bureau of Labor Statistics. Available at [Accessed 10 February 2021]