Economic forecasting is a ubiquitous but frequently maligned practice, often dismissed by investors as too erratic to be worth the effort of pursuing in their investment strategy. Fathom Consulting explores some ideas on how even the erratic aspects of macroeconomic forecasting can be systematically exploited to build more efficient portfolios.
- More efficient portfolios can be constructed by leveraging the level of disagreement among macroeconomic forecasters polled by Reuters.
- A higher level of disagreement among macro forecasters is associated with higher asset returns when controlling for the level of market risk; portfolios based on that premise exhibit both better absolute and risk-adjusted returns.
- Disagreements among forecasters and market volatility are risks that complement each other. Incorporating information from both risks – for example, through market forecasts conditional on macroeconomic trends – allows investors to create even better portfolios.
For more data-driven insights in your Inbox, subscribe to the Refinitiv Perspectives weekly newsletter.
There is much debate and sometimes outright scepticism in providing exploitable investment signals.
One of the main difficulties in forming accurate forecasts stems from dealing with the inherent human biases of forecasters. Many were slow to catch on to 2021’s inflationary surge, for example, despite clear signs of building price pressures by spring 2021.
The chart below shows how the median forecast for monthly U.S. Consumer Price Index (CPI), as submitted to Reuters, was often well below actual outturns throughout 2021 and some of 2022.
Being human, forecasters are also prone to emotions, being too optimistic when times are good, and too gloomy when things are bad.
Despite these biases, forecasts, just like asset prices, still offer insights into market consensus and what might surprise investors.
For example, macroeconomic surprises are linked to equity returns. Just like any tool, the challenge is to put these surprises to good use.
A sensible starting point is the Reuters Poll database, which can be accessed through the Refinitiv Datastream Data Loader and Datastream Web Service. The database includes quarterly, monthly and weekly macroeconomic forecasts for more than 900 indicators.
Building portfolios based on economic forecast disagreement
Where does one begin with more than 900 macroeconomic indicators? One reasonable starting point is the six U.S. indicators that the National Bureau of Economic Research (NBER) closely monitors to track economic expansions and contractions: nonfarm payrolls, industrial production, unemployment, personal income, personal consumption, and retail sales.
Academic research (e.g., Gao et al, 2019) has found that lower asset returns are often associated with periods of high disagreement among forecasters. Working from this principle, we calculate the dispersion in the macro forecasts for each of the six indicators.
We also build a seventh measure, that collates the level of disagreement across all six variables in a scorecard approach.
We find that this composite metric of disagreements among forecasters is useful in capturing not only periods of extreme market volatility, as measured by the VIX index, but also a broader set of volatile periods than the VIX.
We also find that this composite strategy performs best in tandem with the level of disagreement in payroll forecasts relative to two benchmarks (one that seeks to maximise the historical Sharpe ratio each quarter, and a simple fixed-weight strategy).
From unconditional to conditional use of forecasts
Our analysis shows that there is value in creating portfolios incorporating extreme levels of disagreement among macro forecasters; and that these periods do not all coincide with extreme market volatility.
In other words, market volatility and disagreements among forecasters hold a significantly different set of information, that could be further exploited through the use of conditional forecasts.
We provide an example of this by deploying a forecast of market risk conditional on the state of the economy: FROG (Fathom Risk-Off Gauge). This metric exploits Refinitiv’s data to show in a simple probability whether the markets are in a risk-on or -off regime.
FROG is an attractive indicator, as it mimics the kind of forecaster who is continuously learning about the relationship between markets and the economy, and who is never sure which market regime will prevail at any given point in time, but who is also undeterred in providing clear probability estimates.
We find that a conditional forecast like FROG allows investors to adopt portfolios that can separate between periods of higher and lower returns: risk-off and risk-on periods.
These findings are made clear in the above chart, and can easily be summarised in a couple of maxims for investors seeking to adopt this approach:
- Risk-on and risk-off phases should be as clearly differentiated as possible (little overlap between regimes)
- The relationship between risk and return should be broadly downward sloping in risk-off periods and upward sloping in risk-on phases
Interested in more?
This report highlights some of the key findings from a recent Fathom report. Read the full paper and gain more insights into how economic forecasts and market data can be used to build portfolios that optimise macro-related risk-taking by incorporating information embedded in forecasts.
