the effects of monetary policy – Bank Underground

Natalie Burr

In economic theory, expectations of future inflation are an important determinant of inflation, making them a key variable of interest for monetary policy makers. But is there empirical evidence to suggest monetary policy can help determine inflation expectations? I answer this question in a recent paper by applying a Bayesian proxy vector autoregression (BVAR) model to summary measures of inflation expectations for households, firms, professional forecasters and financial markets, derived using principal component analysis (PCA). I find that median inflation expectations respond to contractionary monetary policy, with heterogeneity across groups: financial markets and firms’ expectations fall, while households’ expectations rise. I also document that monetary policy shocks reduce the dispersion of expectations in the 12–18 months following a shock.

To start, do inflation expectations matter for monetary policy? In theory, yes! For households and firms, inflation expectations part-determine the real discount rate, which affects consumption, saving and investment decisions. They also feed into household wage demands, and firms’ price-setting. Financial market expectations affect asset prices and financial conditions that households and businesses face when making consumption, investment and financing decisions. While professional forecasters are not economic agents per se (ECB (2021)), their expectations may inform others.

Data

Since inflation expectations are not directly observable, economists rely either on survey-based measures or market prices (eg derived from the difference between nominal and inflation-linked government bonds). I construct a panel data set for the UK from June 1997 (the first monetary policy meeting after the Bank of England gained operational independence for the conduct of monetary policy), to December 2019. I omit the post-2019 period to avoid structural breaks in the data following the high inflation period. I exploit the cross-sectional and time-series variation of the data, in addition to moments of the distribution, looking at the central tendency and dispersion. Reis (2021a, 2021b) and Meeks and Monti (2023) document how information across the distribution matters for inflation outcomes.

I collect short- and medium-term inflation expectations data across economic agents (summarised in Table A). For households, I use the quarterly Bank of England/Ipsos Inflation Attitudes Survey and expectations from the Bank of England Millennium database. For firms, the availability, quality and scope of surveys has been limited historically (Coibion et al (2020)) despite their importance to monetary policy makers in price- and wage-setting. I use evidence from the monthly Decision Maker Panel and the quarterly CBI Distributive Trades survey. For professional forecasters, I collect evidence from HMT’s quarterly independent forecasts and the Bank’s Survey of External Forecasters. Finally, for financial markets I combine a survey-based measure from the Bank’s Market Participants Survey with break-even inflation rates derived from inflation-linked bonds. Extracting inflation expectations from financial markets is challenging, as they contain time-varying liquidity and risk premia, and UK instruments reference the Retail Price Index (RPI), not the Consumer Prices Index (CPI) (I adjust for the RPI-CPI wedge using its historical average).

Table A: Summary of inflation expectations metrics

Methodology – PCA

It can be difficult to know what to make of the wide range of measures, available over different but overlapping time horizons, at various frequencies, for different groups of economic agents and various moments of the distribution. I therefore construct a summary statistic of inflation expectations using PCA, following Ahn and Fulton (2020), for households, firms, financial markets and professional forecasters individually. As a dimensionality reduction technique, PCA decomposes the covariance structure of variables into factors that are common to all, and idiosyncratic ‘noise’. It maximises the common information across indicators and assigns weights to individual data series based on the degree of comovement with other variables in the model. Due to varying sample lengths and frequency of the data, I apply a methodology proposed by Stock and Watson (2002), specifically for an unbalanced panel data set. I first estimate principal components (PCs) for the balanced panel. The variables with missing observations (including those at frequencies lower than monthly) are then linearly projected on the PCs of variables available over the full sample period. This process is repeated until convergence of PCs across iterations (Erdem and Tsatsaronis (2013)).

Chart 1 shows the first PC for median inflation expectations. To gain economic interpretation, I adjust the series to have the same mean as a medium-term inflation expectation measure from each sample, such that this can be interpreted as the ‘level’ factor of inflation expectations.

Chart 1: First principal component of median inflation expectations

Source: Author’s calculations.

The volatility in firms’ expectations stands out, which Candia et al (2024) document too. They suggest this is due to firms’ inattention to inflation and monetary policy (instead, focusing on sectoral dynamics or competitor behaviour).

Chart 2 shows the summary index for the dispersion of inflation expectations (measured as the difference between the 80th and 20th percentile of the distribution). The larger the dispersion, the wider the distribution and the higher the disagreement among individuals on future inflation outcomes. We see how periods of relatively high inflation (post-financial crisis and Covid) coincide with an increase in dispersion.

Chart 2: First principal component of the dispersion of inflation expectations

Source: Author’s calculations.

How might monetary policy affect inflation expectations? A contractionary shock should reduce median inflation expectations. Agents observe the central bank action, update their beliefs, and anticipating that contractionary policy reduces inflation, inflation expectations fall. For dispersion, it is more ambiguous, though Grigoli et al (2020), using US data, find that a monetary policy shock increased the dispersion of professional forecasters’ expectations for up to nine months following the shock.

Methodology – BVAR

To empirically estimate the effects of monetary policy on inflation expectations, I use a monetary policy surprise measure (Cesa-Bianchi et al (2020)) which I introduce into a BVAR as an external instrument for the monetary policy shock. I am interested in the causal impacts of monetary policy on inflation expectations, but monetary policy also reacts to changes in expectations – the causality goes both ways. I follow Cesa-Bianchi et al by including a small set of variables that intend to capture the various transmission channels of monetary policy. I use CPI as the measure of the aggregate price level; real GDP; GDP expectations; the nominal effective sterling exchange rate; investment-grade UK and US corporate bond spreads, and UK mortgage spreads; the one-year nominal UK government bond yield; and the PC of inflation expectations.

Results

Chart 3 shows the impulse response functions of median inflation expectations to a contractionary monetary policy shock that increases the one-year gilt yield by 1 percentage point.

Chart 3: Impulse response functions of median inflation expectations to a 1 percentage point monetary policy shock

Notes: Each panel shows the impulse response function of the inflation expectations PC to a monetary policy shock that increases the one-year gilt yield by 1 percentage point on impact. Shaded areas show the 68% credibility bands, and solid lines show the median response. The model is estimated with two lags and a constant from 1997 M6–2019 M12. For professional forecasters and firms, the sample is shortened to 2000 M3–2019 M12 and 2008 M6–2019 M12 respectively.

Source: Author’s calculations.

Financial market and firms’ expectations fall quickly in response to the monetary policy shock (in line with results by Di Pace et al (2025) for firms). Financial market expectations remain below baseline for 18 months, whereas firms’ expectations fall persistently. This corroborates the idea that financial markets are forward-looking and incorporate the future contractionary effect of monetary policy on inflation into contemporaneous expectations.

Professional forecasters do not react significantly to monetary policy shocks, likely a feature of little variation in the data. In contrast, households’ inflation expectations rise in response to a monetary policy surprise. This could be explained by suggestions in the literature that households are inattentive to inflation and monetary policy, do not fully internalise the general equilibrium link between them (De Fiore et al (2022)), and are more backward-looking.

Chart 4 shows the impulse response functions of the dispersion of inflation expectations to a contractionary monetary policy shock. Dispersion increases on impact, reflecting an initial increase in uncertainty around the inflation outlook. This might be surprising in the context of monetary policy as a macroeconomic stabilisation tool. However, it is notable that despite the shock, dispersion is actually below baseline in the subsequent 12–18 months, reflecting the ability of monetary policy to reduce the dispersion among agents in the economy about future inflation outturns.

Chart 4: Impulse response functions of the dispersion of inflation expectations to a 1 percentage point monetary policy shock

Source: Author’s calculations.

Policy implications and conclusion

In light of evidence that monetary policy can affect inflation expectations, is managing them an effective monetary policy tool? In practice it can be challenging for a monetary policy maker to try and steer them effectively with actions or communications (Rudd (2021)). However, the challenge of influencing expectations is not new for central bankers. Expectations about the future path of the economy and interest rates are what monetary policy makers influence when providing forward guidance (Sutherland (2022)) – a widely used monetary policy tool.

The results presented in this paper suggest that monetary policy does significantly influence inflation expectations, albeit with considerable heterogeneity across economic agents. However, to the extent monetary policy makers rely on expectations to transmit changes in the monetary policy stance, the Lucas critique applies: if policy were calibrated to target inflation expectations through communications, the way inflation expectations are formed, and possibly how they transmit would change. Model results estimated on historical data would no longer be applicable, given this change in policy regime.

To conclude, the empirical evidence presented in this post has important policy implications and is consistent with an expectations channel of monetary policy – suggesting that monetary policy makers have the potential to impact inflation expectations by changing their monetary policy stance. Further research is needed on communication strategies that could enable policymakers to maximise the effectiveness of this channel and use this as a credible policy tool to support the effective transmission of monetary policy.

Natalie Burr works in the Bank’s External Monetary Policy Committee Unit.

If you want to get in touch, please email us at [email protected] or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.