Measuring inflation expectations: the devil is in the detail

The expectations of economic agents regarding inflation constitute a key factor to how this variable behaves. For some time now both theoretical economists and monetary policymakers have devoted time and effort to studying this relationship in some detail. Ben Bernanke, who has performed both functions, summarised the issue when he said that «an essential prerequisite for controlling inflation is controlling inflation expectations». But achieving control involves at least four big challenges: understanding how inflation expectations interact with the rest of the relevant macroeconomic variables; identifying how they are formed; measuring how they evolve to determine whether they are as we would wish; and evaluating the policies a central bank might adopt to influence them. Below some points are presented regarding the third of these challenges.

Inflation expectations are important because of the powerful influence they have on the decisions taken by individuals. It therefore comes as no surprise that the desire or aim of monetary authorities whose mandate is to safeguard price stability is to keep inflation expectations anchored. In other words, fluctuations in inflation expectations over the medium to long term must be moderate and removed from any temporary shocks to current prices. Any significantly different behaviour to this would pose a serious threat to price stability and, as far as possible, would need to be corrected. Being able to measure inflation expectations is therefore of the utmost importance. However, some details are complex and capable of causing far-reaching errors.

There are two types of measure, very different in nature: those based on surveys and those inferred from the prices of certain financial instruments. One of the main strengths of the former is that they directly provide the inflation expectations of those being surveyed, be they consumers, business people or economic-financial analysts. Consequently the data hardly need to be processed afterwards. The European Central Bank (ECB) systematically monitors several types of survey, the most important one being the quarterly survey on analysts. The historical and recent results are very interesting given the debate currently raging on the risk of deflation in the euro area (see the graph). Firstly, of note is the convergence in analysts' replies regarding the ECB's inflation target (close to but below 2%) over time, as well as its lower degree of fluctuation. This suggests that inflation expectations are considerably anchored and that the ECB's monetary policy is gradually gaining credibility. Secondly, volatility in inflation expectations falls as the horizon for the forecast increases, which means that the impact of temporary shocks is noted in the short term but disappears over the long term. For example, the sharp fall in Europe's inflation over the last few months, driven by the notable slump in oil prices, has led analysts to lower their inflation forecast for 2015 from 1.2% in June 2014 to 0.3% in December. On the other hand, the long-term forecasts (five years' time) have only been lowered by 0.1 pps between these two dates, standing at 1.8% in December, very close to the ECB's target. This pattern can also be seen in similar surveys in the US.

There are three broad types of product that can be used to measure expectations via financial instruments: inflation-linked or I/L bonds, inflation swaps and inflation caps and floors.1 Due to its size and liquidity, the inflation-linked government bond market is the most important benchmark for analysing and evaluating the trend in inflation expectations among participants in the financial markets. This kind of asset, issued by an increasing number of governments, offers holders protection against purchasing power being eroded by inflation on the (nominal) return for any investment. This is achieved by updating the principal in line with the inflation recorded during the lifetime of the bond. Consequently, the spread between the yield on a nominal bond (nominal IRR) and that of an I/L bond (real IRR) with equal maturity and by the same issuer, the so-called break-even inflation rate (BEIR), provides an approximate indicator of the average inflation rate expected by investors over that period. An inflation swap is an instrument in which one of the parties undertakes to pay a fixed interest rate in exchange for receiving the average inflation rate observed during the lifetime of the swap. The fixed interest rate for a swap therefore indicates the average inflation expected over the lifetime of the swap, similar to the BEIR in the absence of arbitrage.

Comparatively, both ways of measuring inflation expectations have been equally successful at predicting the actual inflation rate (see the second graph). However, as they are dissimilar in nature, significant differences arise at certain moments. The current situation is a case in point. The main advantage of inflation measures contained in financial asset prices lies in the fact that these are observed on a very frequent basis: daily or from one day to another in most cases, as opposed to the monthly or quarterly frequency of surveys. Also of great importance is the availability of measures over different horizons, from one year to 30, at least in the main countries. On the other hand, the main drawback of the BEIR and the rest of the variables of financial origin lies in two interrelated factors: their greater volatility compared with survey-based measures and the distortion resulting from the presence of risk premia, making it difficult to calculate the «true» expected inflation. More specifically, in addition to a liquidity risk premium,2 there is also an inflation risk premium (IRP hereafter). This is essentially the compensation received by investors for the risk taken regarding possible deviations between the actual inflation rate and the inflation they expected. In this way, higher risk (greater deviation from expectations) leads to a higher IRP. This premium must therefore be taken into account before interpreting the level and fluctuations of the BEIR and inflation swap rates.3

The initial interpretation of inflation currently given by financial products as a whole (bonds, swaps and options) shows a striking trend: the decrease in long-term inflation expectations is appreciably greater than that observed in surveys, both in the euro area and the US. As seen in the first graph, in the fourth quarter of 2014 the euro area's inflation expectations for 2019 implied in inflation swaps had fallen to 1.2% compared with 1.8% from the survey of analysts at the same time. What can explain this divergence between market and survey-based measures? A large part of the answer lies in the IRP. In fact, over the last few quarters the IRP has shrunk considerably, falling below the average for the last decade in the US and the euro area (40 and 50 bps, respectively).4 It should also be noted that the influence exercised by this drop in the IRP on fluctuations in inflation expectations is greater over longer horizons.5 This last point is crucial in order to accurately judge the severity of the recent downward path observed in the BEIR and the contained rates in inflation swaps. It therefore seems that survey-based measures have a certain advantage within the current disinflationary context.

In short, both survey-based inflation measures and those inferred from the price of financial instruments are subject to limitations that must be taken into account. It is therefore of the utmost importance to adjust their interpretation to the prevailing economic situation at any given time as this will determine the effectiveness of the decisions taken by central banks and consequently their credibility.

Carlos Martínez Sarnago and Joan Daniel Pina

Financial Markets Unit, Strategic Planning and Research Department, CaixaBank