Words versus deeds: How central banks manage expectations
Paweł Baranowski, Wirginia Doryń, Tomasz Łyziak, Ewa Stanisławska 22 October 2020
As documented by Blinder et al. (2017), both monetary policymakers and academics assess that central bank communication has intensified substantially since the beginning of the Global Crisis and recent changes in this respect are expected not to revert back in future (see Figure 1). On the one hand, the rising importance of communication might be due to the fact that many central banks have hit the effective lower bound on interest rates, and so are constrained in using the main conventional instrument of monetary policy. On the other hand, the period of low and stable inflation despite output fluctuations has raised a debate on the flattening of the Phillips curve. Some empirical studies suggest a lower impact of economic activity on inflation in advanced economies (e.g. Kuttner and Robinson 2010, Jašová et al. 2020), which can occur in particular when inflation is very low and economic slack high (Forbes et al. 2020). In such circumstances, the traditional interest rate channel is weaker. For the above reasons, monetary policymakers are searching for alternative ways to influence inflation and economic activity, including more active communication.
Figure 1 Central bank communication since the global crisis
Source: Adapted from Blinder et al. (2017).
The empirical literature shows that enhancing communication to the public makes the management of expectations easier for central banks. As shown by Naszodi et al. (2016), greater informational openness by central banks leads to less dispersed and more accurate forecasts by the private sector. One of the most important aspects in this respect is the preparation and publication of central bank forecasts, the importance of which in managing expectations has been confirmed in a number of studies (e.g. Hubert 2015, Hattori et al. 2016, Ehrmann and Fratzscher 2007, Fujiwara 2005).
Assessing the impact of central bank communication on private sector expectations is challenging due to several reasons. First, central bank communication itself is complex and uses various tools, including (among others) announcements of inflation targets, the publication of central bank macroeconomic projections, detailed explanations of monetary policy decisions, and signalling of future decisions (‘forward guidance’). Second, some aspects of central bank communication, such as the tone of policy documents or statements, are difficult to measure. Third, monetary policy is endogenous as it reacts to current or expected economic developments. Fourth, communication aims at explaining monetary policy decisions, and therefore it should be analysed jointly with them in the framework allowing for the effects of words and deeds to be distinguished.
The two dimensions of central bank communication
In a recent paper (Baranowski et al. 2020), we attempt to overcome the above difficulties and empirically examine the impact that both central bank words (communication) and deeds (interest rate decisions) have on the expectations of economists in Poland. It is worth noting that in contrast to many other central banks, the Polish central bank (Narodowy Bank Polski, or NBP) conducted conventional monetary policy, despite the disturbances caused by the global crisis. The fact that the period under investigation (2011-2019) does not cover the use of unconventional monetary policy tools (like quantitative easing or assets purchasing programmes) facilitates the identification of the impact of policy actions on expectations as we can focus solely on interest rate changes. In terms of communication, Narodowy Bank Polski uses a range of tools typical for an inflation targeting regime.
We exploit survey data on inflation and the interest rate expectations of economists and consider various measures available for central banks to shape these expectations. In particular, we consider two dimensions of central bank communication: one related to quantitative information provided by the central bank to the public (namely, inflation and GDP growth projections), and the second related to the qualitative content of its policy documents (i.e. the extent to which it is hawkish or dovish). Central bank tone is quantified based on minutes from meetings of the Monetary Policy Council with the use of the text-mining techniques. More specifically, we apply a dictionary approach which counts words or phrases associated either with positive (describing good economic stance or hawkish policy inclination) or negative (suggesting weak economic stance or dovish policy inclination) connotations. The resulting central bank tone describes the degree of ‘hawkishness’ or ‘dovishness’ of policy minutes. Previous studies found that central bank tone helps in predicting interest rates and has some impact on financial markets and various types of expectations (Apel and Blix-Grimaldi 2014, Picault and Renault 2017, Bennani 2019).
The following examples provide a look into the quantification of the tone of monetary policy documents according to the Apel and Blix Grimaldi (ABG) dictionary and the Bennani and Neuenkirch (BN) dictionary. The bolded words (BN dictionary) or phrases (ABG dictionary) denote matching dictionary entries in each respective case. The sentences conveying positive (hawkish) tone include:
ABG: Those members of the Council pointed out that the risk of accelerating wage growth was boosted by high inflation expectations of households and businesses.
BN: These meeting participants also assessed that the acceleration of interest rate increases might not reduce inflation expectations of households, which are highly adaptive in nature.
On the other hand, the following wordings provide an illustration of negative (or dovish) tone:
ABG: While discussing developments in households’ deposits in 2012, some Council members pointed out that although their lower growth had been, to certain extent, driven by larger inflows of capital to investment funds, this was also indicative of households’ lower propensity to save especially amidst cuts in interest rates on deposits at the end of 2012.
BN: While analysing the external conditions, it was emphasised that the global economic activity remained weak as suggested, in particular, by lower GDP growth in the United States in 2012 Q4 and persisting recession in the euro area.
We model revisions in inflation and interest rate expectations (along a horizon ranging from the current quarter to four quarters ahead) and regress them on news released between the two consecutive survey rounds. Importantly, in the case of interest rate changes, tone, and projections in the estimated model, we employ the unexpected components (the shocks). These shocks are extracted either by regressing raw measures on a large number of real-time data (related to current and future macroeconomic conditions and financial market stance), or by taking deviations from previous predictions. This approach helps us to eliminate endogeneity between monetary policy measures and expectations.
Inflation expectations are more sensitive to policy communication than to policy decisions
Figure 2 summarises our main results and shows how the impact of policy decisions and communications on economists’ expectations differs depending on the variable forecasted by experts and the forecasting horizon. In line with well-documented interest rate smoothing, unexpected changes in short-term interest rates move the whole path of interest rate forecasts. In contrast, they affect inflation expectations only in the ‘four-quarters-ahead’ horizon, corresponding to delays in the monetary transmission mechanism. Moreover, according to our results, the impact of central bank actions is larger than the impact of communication for interest rate expectations, while the opposite holds for inflation expectations.
Figure 2 Responses of expert forecasts to typical monetary policy decisions and communication
Notes: The figure shows a reaction of private sector expectations (in perc. points) to typical (1 standard deviation) changes in various monetary policy measures. Horizon of expectations is given on the horizontal axis. For interest rate expectations the central bank tone is measured with the Apel Blix-Grimaldi (2014) dictionary and central bank projection refers to the GDP growth; for inflation expectations the central bank tone is measured with the Bennani and Neuenkirch (2017) dictionary and central bank projection refers to CPI inflation. Increase of central bank tone corresponds to more hawkish textual content of policy minutes.
Source: Baranowski et al. (2020).
As for the role of central bank communication, we show that a more hawkish tone of policy minutes pushes interest rate forecasts up. By signaling higher inflationary pressure, this raises inflation expectations. Interestingly, the tone of the policy minutes affects expectations more at the shortest horizons (i.e. ‘nowcasts’ and one quarter ahead), while GDP and inflation projections play a greater role for slightly longer horizons. This finding seems plausible given that policy documents provide qualitative information on the view of monetary policymakers on current economic conditions, while central bank projections are more forward-looking.
Last but not least, our analysis suggests that central bank words and deeds are substitutes. This means that to achieve a given change in private sector expectations, the central bank can select the mix of measures to be used – for example, replacing interest rate decisions that elevate volatility in the economy with communication affecting the economy in a more efficient manner (i.e. via a direct impact on the expectations of economic agents).
In summary, our findings imply that central banks have at their disposal a range of measures to affect expectations. However, taking into consideration the differences in response of different kinds of expectations to central bank measures, the management of expectations does not seem a trivial task and constitutes a key challenge for central banks.
Baranowski, P, W Doryń, T Łyziak and E Stanisławska (2020), “Words and deeds in managing expectations: Empirical evidence on an inflation targeting economy”, NBP Working Papers 326, Narodowy Bank Polski.
Bennani, H, P Baranowski and W Doryń (2020), “Do the ECB’s introductory statements help predict monetary policy? Evidence from a tone analysis”, European Journal of Political Economy, forthcoming.
Blinder, A, M Ehrmann, J De Haan and D-J Jansen (2017), “Necessity as the mother of invention: Monetary policy after the crisis”, Economic Policy 32(92): 707-755.
Ehrmann, M and M Fratzscher (2007), “Communication by central bank committee members: different strategies, same effectiveness?”, Journal of Money, Credit and Banking 39(2-3): 509-541.
Forbes K J, J Gagnon and C G Collins (2020), “Low inflation bends the Phillips curve around the world”, Peterson Institute for International Economics, working paper 20-6.
Fujiwara, I (2005), “Is the central bank’s publication of economic forecasts influential?”, Economics Letters 89(3): 255-261.
Hattori, M, S Kong, F Packer and T Sekine (2016), “The effects of a central bank’s inflation forecasts on private sector forecasts: Recent evidence from Japan”, BIS Working Papers 585, Bank for International Settlements.
Hubert, P (2015), “Do central bank forecasts influence private agents? Forecasting performance versus signals”, Journal of Money, Credit and Banking 47(4): 771-789.
Jašová, M, R Moessner and E Takáts (2020), “Domestic and global output gaps as inflation drivers: What does the Phillips curve tell?”, Economic Modelling 87: 238-253.
Kuttner, K and T Robinson (2010), “Understanding the flattening Phillips curve”, The North American Journal of Economics and Finance 21(2): 110-125.
Naszodi, A, C Csavas, S Erhart and D Felcser (2016), “Which aspects of central bank transparency matter? A comprehensive analysis of the effect of transparency on survey forecasts”, International Journal of Central Banking 12(4): 147-192.