ClearMacro Forecasts
Introduction
We believe that in order to make better investment decisions, you need to start with a comprehensive set of views about how you expect markets to behave in the future, given everything you know today.
Whether used to directly drive portfolio construction, provide insights or simply to surface questions that lead to insights our platform equips our clients to achieve better outcomes.
Even though there is a high degree of uncertainty that comes naturally with trying to forecast the future, investment managers need to have some basis on which they make their decisions. Our belief is that it is important to have a transparent and consistently applied approach which can provide a degree of confidence and accountability to internal decision making and support communication with stakeholders.
Our approach uses multiple models to incorporate the changing drivers of returns over different time horizons. We combine them together into a term structure to help reconcile long term and short term market outlooks across horizons from 1 month through to 10 years.
Our global multi-asset forecasts can quickly help you to expand investment coverage, making it easier to perform relative selection across a wider pool of investment opportunities.
Methodology
We use short-term (Tactical), medium-term (Cyclical) and long-term (Trend) forecasting models as shown in this illustration:
Trend is a measure of the long term returns a market is capable of producing.
Cyclical measures the effect of the current business cycle at a given point in time and it's previously observed effect on market returns.
Tactical encompasses short-term market effects in response to current events.
By building our forecasts up in this way, we are able to provide tangible explanations about what the underlying drivers for a given forecast are. Our customers find this extremely useful when speaking with their internal and external stakeholders.
We forecast these components over different forward-looking horizons from 1 month to 10 years.
Collectively, we refer to them as a Term Structure of Expected Returns - a forecasting roadmap of how the market will behave over the next 10 years:
The chart above is a somewhat stylized illustration to give you the general idea. The overall roadmap seamlessly consolidates long, medium and short term views.
To make things a little more real, here's what our term structures for the US Equity and Bond markets look like right now for the next 10 years:
Now let's take a look at how each component of the overall forecast is put together.
Trend
The first component of our forecast we call Trend. It is a measure of the long term returns we believe a given market is able to produce. We use relevant and credible academic models such as Building Blocks to forecast the 3 most important return drivers over the long term: Value, Income and Growth.
Value is the incremental annual return that can be captured by buying or selling when a market has moved significantly from its long-term equilibrium (or “fair value”). To estimate it, we track a given market’s current deviation from “fair-value”, and then apply a consistent time-frame for that deviation to normalize.
Income is the expected annual cash-flow yield of an asset over the long-term. To estimate it, we forecast a market’s net annual cash yield to investors. Different metrics are applied to track expected cash-flows depending on the asset class. In the case of equities for example, it is the current dividend yield adjusted for a reversion of pay-out ratios to their long-term average over the next 10 years.
Growth is the expected annual nominal growth in net cash-flows of an asset over the long-term. To estimate it, we forecast how much real revenues are likely to grow over the next 10 years based on the outlook for real economic growth. We then adjust this by accounting for expected inflation over the same period. Different metrics are applied to track expected cash-flow growth depending on the asset class, excluding fixed income assets, where there is no Growth building block.
The overall Trend forecast is the sum of these three Building Blocks. Here's what our 10 year Trend forecast looks like for the US Equity market and where the contributions are coming from:
Cyclical
The cyclical model splits the business cycle into 4 distinct phases: Expansion, Slowdown, Contraction and Recovery.
We use the level and momentum of OECD’s Composite Lead Indicator to determine where a given market currently lies within the business cycle. Here is an example using the US Equity market:
For any asset, we analyse historical returns in each phase compared to average historical returns across all phases to determine whether an asset has performed better or worse than the historical average in each phase, and by how much.
Using this information we forecast the expected return from cyclical effects relative to the Trend forecast for horizons out to 5 years. Here's what our 1 year Cyclical forecast (relative to Trend) for the US Equity market looks like right now:
Tactical
Tactical is a measure of the short-term impact on returns caused by temporary macro and technical factors. To forecast this, we utilise our Signal Library product containing over 130 signals applied to 6000 different markets:
We run historical tests to find Signals which have the ability to forecast returns for a given market over 1, 3, 6 and 12 month horizons.
The results give us an indication of what market returns are generally like under different conditions. We then filter those results down to only those which pass a 99% significance test before including them in our forecasts.
As an example, here's our US Inflation Signal, which posits that the level and momentum of inflation correlates negatively with subsequent short to medium-term equity returns. It combines data such as headline/core CPI rates, money supply, output gaps, capacity utilisation, real FX rates, inflation surprises, earnings, and PPI.
Here's a look at the actual Signal history, using a Z-score based approach to categorise the Signal into Bullish (low inflation) vs. Bearish (high inflation) conditions:
We can test this signal to see if it has any utility for predicting forward looking US Equity returns over a 12 month horizon by analysing how changes in the Signal relate to changes in the distribution of the realised returns:
Though the distributions look very similar, the subtle shift tells us that there is a pattern that emerges when we look back over the history of US Equity returns.
To create our forecasts we perform this analysis in-sample using 70% of the historical returns data. We apply a statistical test on the results to discard any that do not meet a 90% statistical significance threshold. We also discard any results at this stage that do not align with the hypothesis of the Signal being tested.
In order to increase confidence further, we then repeat the analysis out-of-sample using the remaining 30% of the historical data and apply a second statistical test to discard any results that do not meet a 90% statistical significance threshold.
Of course this is only one Signal, so we repeat this process hundreds of thousands of times to generate a more complete picture of what all of our Signals are telling us about forward looking returns.
The qualifying results across all of our individual Signals are then combined, taking into account their correlations with each other, and summarised into two categories: Fundamental and Technical.
Here's our current 12 month Tactical return forecast for the US Equity market:
Another benefit of constructing the forecast in this way is that we also have detailed information about exactly which Signals from our library contribute to a particular forecast, and how those drivers have changed over time:
Here's what the breakdown looks like right now for our latest forecast. Note if this section is empty, it means there are currently no high-conviction active tilts being applied:
Summary
To recap, for every market:
Depending on the forecasting horizon, our overall market forecast will be a combination of these components.