Speech by Isabel Schnabel, Member of the Executive Board of the ECB, at the 25th Jacques Polak Annual Research Conference

Washington, D.C., 14 November 2024

Over the past few years, monetary policy has had to change course in an unprecedented way. Faced with persistently low inflation, central banks in many advanced economies had lowered policy rates to record low levels and bought significant amounts of assets to bring inflation back to target.

When the pandemic hit and inflation surged as our economies reopened, central banks responded by raising policy rates sharply to restore price stability and anchor inflation expectations. They also started to gradually reduce their balance sheets, further removing policy accommodation.

Although the fight against inflation has not yet been won, with domestic price pressures remaining high, it is time for central banks to start reflecting on the lessons that can be learned from these episodes of low and high inflation.

My main message today is that, in a volatile macroeconomic environment, central banks need to be more agile, meaning they should draw primarily on instruments that can be reversed quickly and that allow monetary policy to be adjusted swiftly to changing economic conditions.^[1]

Central banks cannot rely on stable macroeconomic relationships

Since the pandemic, the macroeconomic environment has changed in three important ways.

First, the recent surge in inflation showed that large and persistent inflationary episodes are not confined to the past, when central banks had not yet adopted inflation targeting as a framework for monetary policy.

Large inflation outbursts can occur even when central banks have a clear mandate and a strong commitment to achieving price stability.^[2]

With the benefit of hindsight, the long period of low and stable inflation during the great moderation was the result not only of good policies but also of good luck, in the sense that the shocks hitting our economies were, on balance, disinflationary.^[3] Today, this balance may have changed signs.

The change in the macroeconomic landscape is also reflected in financial markets. While inflation risk premia had become negative in the years before the pandemic, they started to rise already in the second half of 2020 and returned to positive territory in 2022 (Slide 2, left-hand side).

In the same vein, we have seen a shift in the Survey of Professional Forecasters. In the pre-pandemic period, the distribution of inflation expectations was skewed to the downside. In the high inflation period, the distribution shifted markedly to the right, showing a fat, right-hand tail. And then, as inflation slowed, it became more balanced (Slide 2, right-hand side).

So, within a few years, we may have moved from a regime of chronically low inflation to one where adverse supply-side shocks dominate – owing, for example, to the growing fragmentation of the global economy or climate change. This may give rise to upside risks, as some of the factors that had contributed to low inflation during the 2010s may now be working in the opposite direction.

Second, the level of the natural rate of interest, or r* – the real short-term rate at which monetary policy is neither contractionary nor expansionary – has become more uncertain.^[4]

Before the pandemic, a secular decline in productivity growth and an ageing society had put downward pressure on r*.^[5] In recent years, however, the range of model-based estimates of the natural rate of interest has moved upwards in tandem with market-based measures (Slide 3, left-hand side).^[6]

Whether real interest rates will ultimately be higher than during the 2010s depends critically on how our society addresses the structural challenges it is currently facing.

If firms and governments are serious about fighting climate change and making the euro area economy more competitive, private and public investments will need to rise measurably so that capital deepening and higher productivity growth offset the drag from an ageing society on potential output and real interest rates.

Given the global nature of r*, the future level of real rates will also depend on changes in the supply of savings and the demand for investments in the rest of the world. A global surge in investment in response to digitalisation, decarbonisation and deglobalisation would raise r*, while higher global uncertainty in an era of geopolitical shifts may have the opposite effect.

This uncertainty about the level of r* is reflected not only in the wide range of estimates across different models but also in the even larger parameter uncertainty within models (Slide 3, right-hand side).

The latter has recently increased notably. For example, according to the frequently used Holston-Laubach-Williams model, the 68% confidence interval for the estimate of the euro area’s current natural rate has a width of almost 12 percentage points around its point estimate.

Given this uncertainty, central banks need to be prepared for all scenarios.

Third, before the pandemic, a wide range of research suggested that the Phillips curve had flattened considerably, implying a weaker relationship between economic slack and inflation.^[7]

Today, there is increasing evidence across major economies that the slope of the Phillips curve is highly state-dependent.^[8] When marginal costs increase rapidly and threaten to erode profit margins, firms tend to raise their prices more frequently, as seen in past years (Slide 4, left-hand side).^[9]

More recently, however, firms have started absorbing a substantial part of the increase in their marginal costs – mainly reflecting strong wage growth – in their profit margins, causing the frequency of price increases to slow measurably. This change in the pass-through of marginal costs to prices also reflects the success of monetary policy in restoring the balance between supply and demand.

Research has also shown that, beyond certain levels of labour market tightness, the relationship between inflation and economic slack can change (Slide 4, right-hand side).^[10] In the United States, for example, the number of job vacancies exceeding the number of unemployed workers was accompanied by a rapid surge in inflation, resulting in a steepening of the Phillips curve.^[11]

A state-contingent or non-linear Phillips curve implies that conditions can change quickly if a large shock hits the economy.

A reassessment of policy tools

All this suggests that the macroeconomic environment in which central banks are operating today has become more uncertain and more volatile. In a world more prone to shocks, the inflation regime can shift quickly, real equilibrium rates can move up or down and macroeconomic relationships like the Phillips curve can become highly unstable.

One key takeaway from this increase in volatility is that monetary policy needs to be more agile. It needs to be able to change course quickly when circumstances change.

However, not all policy tools are equally agile. The experience of the past few years suggests that some instruments can be unwound quickly, while others have more persistent effects on financial markets and the economy.

I see three lessons from the episodes of low and high inflation for central banks’ toolkits.

Interest rates remain the primary monetary policy tool

First, policy rates clearly pass the agility test.

The recent tightening cycle is a case in point. Raising policy rates quickly has proven highly effective in reigning in inflation by dampening the growth in aggregate demand and anchoring inflation expectations. This has helped dispel any doubts about central banks’ determination to fight inflation and fears of them being subject to fiscal dominance.^[12]

Our approach to adjusting policy rates was in line with the outcome of our 2021 monetary policy strategy review, in which we confirmed that policy rates are our main policy instrument. Constraints on reacting swiftly are usually self-imposed in the form of forward guidance.

Being able to change course quickly also suggests that negative policy rates, which are more controversial, remain a useful policy instrument should we become constrained again by the zero lower bound.

While negative rates come with undisputable costs, they can be a powerful instrument for easing the policy stance when inflation is too low. Most notably, breaking the zero lower bound allows central banks to shift the entire distribution of the expected future interest rate path, thereby providing more policy accommodation when needed.^[13]

And when the tide changes, this accommodation can be quickly removed, as it was in July 2022 when the ECB unwound negative rates in just one step, raising the deposit facility rate (DFR) from -0.5% to 0%.

Asset purchases are more powerful for stabilising markets than for stimulating the economy

Second, asset purchases need to be used more cautiously and selectively.

Asset purchases are a powerful tool for stabilising financial markets during periods of stress. When markets dry up as intermediaries pull back from market making, central banks can quickly provide liquidity and instil confidence, thereby restoring market functioning.

Stabilising financial markets typically requires only short-lived interventions. During the pandemic, for example, systemic stress in the euro area receded rapidly once the ECB started making purchases under the pandemic emergency purchase programme (PEPP) (Slide 5, left-hand side).

In particular, use of the PEPP’s flexibility, allowing – if necessary – for deviations from the ECB’s capital key, which serves as a benchmark for allocating government bond purchases in the euro area, turned out to be limited and transitory (Slide 5, right-hand side).

Similarly, during the liability-driven investment (LDI) crisis in the United Kingdom, the Bank of England purchased assets over just 13 trading days and began unwinding the portfolio six weeks later. Holdings were fully unwound after a few weeks without causing any market dysfunction.

Hence, asset purchases used for financial stability purposes also pass the agility test, if calibrated and communicated appropriately. Then, purchases can be limited in volume and unwound quickly, which also helps limit moral hazard. They therefore need to remain part and parcel of central banks’ toolkits.

This is particularly important in the euro area, which has been prone to fragmentation and has therefore introduced tools, such as the Transmission Protection Instrument (TPI), to ensure a smooth transmission across the entire euro area.

The assessment is less clear for asset purchases conducted for quantitative easing (QE) when central banks buy bonds to lower long-term interest rates to stimulate the economy.

Experience shows that, in this case, purchasing volumes have to be considerable for central banks to compress the term premium to an extent that affects prices and activity.

For the euro area, for example, ECB staff estimate that between January 2015 and the end of 2018 purchases worth €2.6 trillion were needed to lower euro area five-year sovereign yields by only 0.8 percentage points.^[14]

The past few years have demonstrated that such purchase volumes can come with significant costs.

One such cost is the constraint put on central banks when policy needs to change course.

Large bond holdings cannot be reversed quickly. If quantitative tightening proceeded at the same pace as quantitative easing, there would be the risk of absorption bottlenecks and a sharp rise in interest rates that could undo the effects on activity and inflation for which asset purchases were made in the first place. It could also disturb the smooth transmission of monetary policy throughout the euro area.

In other words, quantitative tightening (QT) imposes a certain gradualism on central banks, as asset holdings can be built up much faster than they can be shrunk.

This has several implications.

One is that it may entail losses for central banks.

When central banks need to raise interest rates in the pursuit of their price stability mandate, interest rate risk materialises, causing large losses on bond holdings. While these losses need to be viewed against the gains made during the purchasing phase, there can be a hit to credibility, even if central banks are not profit-maximising institutions. Depending on the accounting treatment, losses can materialise over a long period of time.

Another implication relates to the impact on the monetary policy stance. As bond holdings can be unwound only gradually, asset prices will remain distorted for a long time.

We saw this occur during the most recent tightening cycle. Term premia remain compressed and risk premia in many asset markets stand at the lower end of the historical distribution, potentially distorting risk-taking behaviour (Slide 6, left-hand side and middle).

For monetary policy, this could mean that central banks needed to raise policy rates more forcefully to compensate for the persistent effects of bond holdings on long-term interest rates. Indeed, scaled by the magnitude of rate hikes, the tightening in financial conditions seems to have been more muted than in previous episodes (Slide 6, right-hand side).

So, what does this imply for the use of QE in the future?

In short, it means that the bar for starting QE should be higher than in the past. There needs to be a clear threat to medium-term price stability for central banks to activate large-scale asset purchases.

In particular, we need to better understand to what extent an extended period of inflation that is moderately below target risks a de-anchoring of inflation expectations that could unleash a truly harmful deflationary spiral.

Downward price and wage rigidities imply that it probably requires a significant shock for low inflation to turn into outright deflation. This can be seen when looking at the pricing behaviour of firms.

Despite persistently low inflation during the 2010s, the share of industrial firms planning to reduce prices was limited and about the same as before the global financial crisis, and the net balance of selling price expectations remained positive on average (Slide 7, left-hand side).

Households, on balance, also did not expect deflation in the low inflation period following the global financial crisis, with the net balance of inflation expectations remaining positive throughout (Slide 7, right-hand side). At the peak, just 5% of households expected prices to fall during the 2010s.

Therefore, should monetary policy in the future be once again constrained by the effective lower bound, central banks would need to carefully assess whether the benefits of QE really outweigh the costs, which often only materialise after a considerable delay.

The ECB’s experience highlights that there are other tools, such as targeted longer-term refinancing operations (TLTROs), that are effective in stimulating bank lending and therefore growth, and can be unwound significantly faster (Slide 8).

Forward guidance should be used sparingly and should be state contingent

The third lesson is that forward guidance must not excessively tie policymakers’ hands.

In the pre-pandemic years, explicit forward guidance signalling a “low-for-longer” policy was effective in providing additional stimulus when needed at the effective lower bound.^[15]

But forward guidance is useful only when macroeconomic volatility is low – that is, when the risk of central banks having to renege on previous commitments is limited.

In a high-volatility world, by contrast, forward guidance almost mechanically implies that central banks either take the risk of falling behind the curve or take a course of action that is inconsistent with their previous guidance.^[16]

Indeed, with the benefit of hindsight, forward guidance, including on the sequencing of the use of policy tools, was arguably one reason why central banks initially reacted too slowly to emerging price pressures, before they changed course sharply.

As forward guidance acted as a credible coordination device for investors, the markets in December 2021 still expected rates to remain negative for the foreseeable future, despite the marked rise in inflation (Slide 9).

In today’s volatile environment, forward guidance is therefore of limited use to central banks. And even if conditions changed in the future, implying a return to a low-inflation, low-volatility world, central banks should only communicate about the likely future direction of monetary policy in a “Delphic” way, that is, conditional on how the economy evolves.

Importantly, the conditions should be formulated in a qualitative rather than a quantitative manner to account for the uncertainty around the inflation and economic outlook.

Conclusion

Let me conclude.

In recent years, the macroeconomic landscape has become considerably more volatile. To effectively manage inflation in this environment, central banks need to prioritise agility and flexibility when choosing their instruments in the pursuit of their mandate.

Short-term interest rates therefore remain the instrument of choice in most circumstances.

Asset purchases have proven to be a powerful tool for market stabilisation, while their cost-benefit ratio is less favourable when it comes to stimulating the economy near the effective lower bound. QE should therefore be used more cautiously in the future. Other tools, such as TLTROs, have been effective in reviving bank lending and can be reversed more quickly.

Our recent experience has also taught us that central bankers should be careful not to tie their hands too much by providing explicit forward guidance. In a volatile macroeconomic environment, they should remain agile and retain the ability to change course quickly when circumstances change.

Thank you.