VoxEU Column Macroeconomic policy Monetary Policy

Friedman vs Phillips: A historic divide

Milton Friedman and Bill Phillips most likely assumed that their separate methods for predicting inflation would lead to much the same outcomes. Recently, however, monetary aggregates and the Phillips curve have provided extremely disparate signals. This column discusses recent economic developments leading to these disparate signals, concluding that inflation will most likely end up somewhere between the predictions of the two models – which is almost certainly higher than what central banks and the IMF are expecting.

Milton Friedman and Bill Phillips most likely assumed that their separate methods for predicting inflation would lead to much the same outcomes.  However, monetary aggregates and the Phillips curve, the two main analytical drivers of inflation, have never before provided such extremely disparate signals, and uniformly so for most developed economies.  

In several of these economies, and particularly the US and Japan, growth in the monetary aggregates has recently reached a high not seen over the last four decades, whereas the labour market and output gaps have remained weak (Figure 1). While the recent spikes in M2 growth are not unprecedented in the euro area and the UK, past spikes were never accompanied by an output gap as deeply negative as today’s.

Figure 1 Spike in M2 growth with negative output gap 





Money growth has also risen in those aggregates that matter for sustained spending. Unlike the period after the Global Crisis, the increase in M1 and M2 has outstripped the relative increase in the monetary base.  Commercial and industrial (C&I) loans have been in demand when cash flows were at their worst, and have receded since their peak in May after lockdowns eased.

Figure 2 Growth in M1 and M2 


Despite becoming more receptive to the risk of inflation, both markets and policy-makers are sceptical about any sustained inflation. It is with this bias in mind that we argue our case for inflation to be lower than monetary aggregate growth might suggest, but higher and earlier than the Phillips curve would signal.

Both models have been abandoned by central banks

“We didn’t abandon monetary aggregates, they abandoned us”, Gerald Bouey’s remarked famously when central banks shifted from targeting monetary aggregates to inflation targets instead. However, the underlying reason for the shift was not that  the link between monetary aggregates and inflation became non-existent, but rather uncertain. And by targeting inflation, central banks were moving to the ‘targets’ of policy, rather than ‘intermediate targets’ like money.

Since, the role of money has also changed structurally, due to financial innovation and financial globalisation, but episodes like the current surge in monetary aggregates are still big enough to have an economic impact.

The Phillips curve, like Schrodinger’s cat, continues to raise questions about its mortal status

The Phillips curve didn’t survive the Fed’s review – its flattening or comatose state provides the Federal Open Market Committee (FOMC) with too little information, in its assessment, to merit a place in the forecasting toolbox. But as with monetary aggregates, this has more to do with the link between the Phillips curve and inflation becoming uncertain, not non-existent.

But there is no doubting the evidence – inflation has been far less receptive to changes in labour market and growth conditions. But why? In our view, demography and China have been intrinsic to the story of disinflation over the last few decades, but central banks have not factored in either of these stories adequately. If we’re right, the global Phillips curve may yet be alive and kicking.

Overall, getting a clean signal out of either model is very difficult today given the fall in velocity when it comes to monetary aggregates and the threat of scarring within the Phillips curve.

Surging monetary aggregates and an impending recovery in velocity

Do monetary aggregates still matter?

A historic surge or collapse in an important macroeconomic indicator is always likely to create ripples. When the economy is functioning smoothly, prices can help clear markets and much of the change in money growth is likely to be endogenous, i.e. rising when economic growth is strong, and falling when growth and transactions falter. However, when the economy faces distortions, quantity signals assume greater importance. Put together, M2 growth of twice the level seen in the previous four decades in the US, in a heavily distorted economy, is likely to contain macroeconomic information (Decker and Goodhart 2018).

The nature of the surge in monetary aggregates is just as important as the quantum

In the post-Global Crisis period, central banks injected money into the financial system. Financial institutions, needing to strengthen their balance sheets, found buying financial assets a far more reliable route to achieving that protection than lending to the private sector. As a result, those injections found their way into the monetary base rather than broad money or credit. During the pandemic, the increase in the monetary base has been outstripped by the increase in M1 and M2 (Figure 3 below). Redirecting those funds into the economy required the creation of additional incentives, like the Bank of England’s ‘Funding for Lending’ programme, and even then credit flows from the banking sector to the private sector were grudging at best for quite a while.

Figure 3 Monetary base, M1 and M2 post-Global crisis and during the Covid-19 pandemic


Monetary aggregates are far more likely to translate into spending because the nature of their rise is almost directly linked to more spending.

Velocity has collapsed for the same reason that personal savings have shot up

The lockdown, mandated or voluntary, has meant that we can’t (or won’t) go anywhere and non-essential businesses had to shut down by government mandate at various times. Consequently, income that would normally have flowed to the services sector has instead stayed on the household balance sheet as a stockpile of personal savings. Revenue interruptions in the services sector, the largest part of the economy, account for a sharp fall in the entire chain of transactions within and related to services.

If we expect personal savings to normalise post-vaccination, as most do, then we must also think of velocity reclaiming some of the ground it has lost in the last year.

And hence as long as money growth doesn’t turn negative, so that the stock of money remains higher than it was at the beginning of the pandemic, normalization of the economy and money velocity will result in higher prices or output, or both.

Can the Phillips curve, and the scarring in the economy, hence provide an offset to engorged monetary aggregates? Yes, but to a lesser extent than the data seem to suggest. Here’s why.

A Phillips curve with far less scarring than the data suggest

Loss of income has been far smaller than the fall in employment would normally entail

Thanks to furlough schemes and government transfers, the loss of consumption that results from high unemployment simply hasn’t materialized. If measured by the ability of households to spend, unemployment is probably at historical lows.

Dossche and Zlatanos (2020) show that the bulk of household savings in the euro area are ‘forced’ rather than precautionary, suggesting that disposable income has far exceeded the consumption needs of households during the pandemic.

Nor have bankruptcies come anywhere close to what one might have expected after such a deep shock to growth, thanks to the various business protection schemes.

Personal savings could flow to various forms of expenditure 

A significant portion of the decrease in the demand for services that occurred in the course of the pandemic is most likely permanent– vacations and restaurant visits could see upgrades, but probably won’t double to make up for what was lost in the last year. Personal savings might instead move into real assets, explaining the boom in housing across the world.

Many believe, incorrectly, that personal savings are simply the dual of the fiscal deficit. That link, however, is far less direct.

The high personal savings rate comes from middle-income/upper-middle-income families as they currently aren’t consuming services. Stimulus cheques from a fiscal expansion will have little to no impact on the consumption/savings decision of these families. For richer households, consumption growth and the personal savings rate are primarily a function of whether normal life is resuming, and not as much of whether fiscal policy continues to give stimulus cheques.

Lower-income households, on the other hand, tend to consume their entire disposable income. They are not part of the high personal savings rate story around the world. Uncertainty around future stimulus cheques (in the US, in particular) could very well lead to lower consumption growth among lower-income households. Consumption growth here depends on the path of fiscal policy and employment.

Fiscal support is likely to remain in place (Landau 2021) – to help the corporate sector and the unemployed, less as an overall support for all disposable incomes (but that’s not what the economy needs anyway). Thus, the combination of fiscal and household saving dynamics together implies strong spending in 2021.

In summary, neither of the scenarios that monetary aggregates or the Phillips curve are suggesting are likely to play out. Instead, inflation will probably end up somewhere between the predictions of the two models, which is almost certainly higher than what central banks and the IMF are expecting. Financial markets, however, are slowly beginning to think seriously about this scenario, as is a small but growing body of academics.


Decker, F and C Goodhart (2018), “Credit mechanics: a precursor to the current money supply debate”,, 14 December.

Dossche, M and S Zlatanos (2020), “COVID-19 and the increase in household savings: precautionary or forced?”, ECB Economic Bulletin 6.

Landau, J-P (2021), “Inflation and the Biden stimulus”,, 8 February.

7,034 Reads