The fundamental question when making any investment is always – how much extra expected return do I require for my expected risk?
Why are inflation expectations so important?
The fundamental question when making any investment is always – how much extra expected return do I require for my expected risk? While the expected risk and expected return inputs among market participants are widely varied, the benchmark of almost any asset starts with the risk-free rate (typically three-month US Treasury Bills), plus a premium to compensate for expected risk. The yield on a risk-free instrument is made up of two components: the inflation adjusted yield (otherwise known as "real" yield") plus expected inflation. On this basis, the market’s expectation of future inflation underpins the benchmarking and subsequently, pricing of nearly every security and asset class in global capital markets.
With inflation expectations so low across most of the developed world, we explore the contributors to, and accuracy of, these expectations. Chart 1 highlights the trend in one-, five-, and ten-year inflation expectations since 1982.
To measure this, we take the longest-available inflation expectations series from the Cleveland Federal Reserve. This series is compiled by combining the market-implied inflation expectations from a number of securities. These securities include inflation swaps, interest rates, and inflation-linked securities (TIPS). We will examine how these inflation expectations are intertwined over three timeframes: one, five and ten years.
Policy-makers are well aware that inflation expectations can become dangerously self-reinforcing given they are largely informed by recently experienced inflation. This view is confirmed by the correlation between expected (future) inflation and experienced inflation (see Chart 2).
The correlation between one-, five-and ten-year inflation expectations to one-, five- and ten-year trailing inflation respectively are 76%, 85% and 91%. This relationship can be seen in Chart 3.
This strongly suggests that what we’ve just experienced is mirrored as the expectations of the future. This phenomenon can be found in every facet of capital markets be it volatility, economic growth or equity market performance forecasts.
Knowing that current inflation informs our expectations for the future, should we put any weight in these forecasts? For this manner of analysis to be effective, future inflation would have to look a great deal like past inflation. Below we test to see if this framework holds any water.
Spoiler Alert: The relationship between future and expected inflation is not so good – especially in the short-term. The correlation between one-, five- and ten-year inflation expectations to one-, five- and ten-year forward realized inflation respectively are 38%, 53% and 83%.
While 83% is quite a high correlation, it must be examined where this comes from and if it is a useful forecasting tool. The 83% correlation coefficient on the ten-year horizon fails to capture the fairly consistent error in the future inflation expectations. Namely, as long as the over- or under-estimation is consistent, the correlation coefficient can be 100%, but not be accurate at all.
In fact, historically, the median over-estimation of inflation averaged 0.71% per year. This is a large sum, particularly when: a) inflation has averaged 3.6% since 1947; and b) 0.71% is compounded over ten years.
Low dispersion of estimates across timeframes
The correlation between inflation expectations across timeframes is particularly strong (as shown by Chart 4). The lowest correlation exists between the one-year and five-year inflation expectations at 97%, and 99% between five-year and ten-year expectations.
The similarity is a possible source of over/ under-estimation of inflation, it is also a nod at behavioral bias we all exhibit called Recency Bias; the tendency to project current experiences far out into the future. Logically, one would not expect the economic environment today will be the same as ten years in the future. Yet this is exactly how short-term and long-term inflation expectations have clustered. The reason for why expectations on one timeframe would bleed so much into the others is perhaps the subject of future study.
While inflation expectations are vital to investment success, the market seems to do a poor job of forecasting it. Our practical understanding of the drivers of inflation remains a "work in progress". While inflation has been consistently over-estimated in the Cleveland Federal Reserve’s series, the sample may be biased by the overall downward trend of inflation. Because Recency Bias is so clearly evident, it seems likely that serial under-estimation would also be possible when a period of very low inflation transitions into higher inflation.
From a portfolio perspective, the main take away is that one must hold a variety of assets to be able to provide a positive return in a variety of inflationary outcomes. This diversification is critical, given how often markets misprice inflation and, subsequently, assets.