< Back to Research

Beyond Equities: Butterwire Takes on Dynamic Asset Allocation

Raphael Fiorentino
25th June 2019 - 5 min read


An upcoming additional feature in the “Markets” section of the Butterwire app is aimed at informing investors’ allocation decision across asset classes. We introduce a simple framework to assist the timing and nature of dynamic allocation decisions between equities, government bonds, gold, commodities, and real estate.

In the “Markets” section of its app, Butterwire provides a daily update of three proprietary metrics designed to capture the macro context of global financial markets. The first one, iGDP (“i” for implied) reflects the global GDP growth expectations of market participants. The next one, iEMC, gauges the extent that global capital is flowing into Emerging Market economies (or back into Developed Markets), and the last one, iLCI, is a late cycle indicator (also referred to as an indicator of inflationary expectations).

All the stocks covered by butterwire are “clustered” according to their sensitivities to these three metrics, as represented in the graph below.

Profiles Chart

For instance, stocks with high sensitivities to iGDP, iEMC and iLCI will tend to belong to the category of “Inflationary EM growth” (top right corner in the above chart), meaning they tend to benefit the most from expectations of strong global growth, accommodative monetary policies and the building of inflationary pressures.

As such, many industrial commodity stocks belong to this group (demand for industrial commodities is largely derived from investments which in turn tends to favour an EM-led growth tilt. Whereas oil, whose consumption-led demand favours a DM-tilt, will see many energy producers fall in the “Inflationary DM growth” category). Likewise, many gold producers fall in the category of “inflation hedges” which benefits from loose monetary conditions (high iEMC) and depressed growth expectations (low iGDP).

We applied a similar analysis to a sample of asset classes, using global ETFs as proxies and looking at the period 29 March 2005 to 20 June 2019:

  1. iShares MSCI ACWI ETF for equities (ACWI) for Equities
  2. xTrackers Global Government Bonds ETF for bonds (XSGS) for sovereign bonds
  3. SPDR Gold Trust ETF (GLD) for gold
  4. S&P GSCI Commodity Index Trust ETF (GSG) for commodities (from Jul-06)
  5. iShares Global Real Estate Index ETF (CGR) for real estate (from Aug-08)
Performance Graph

Looking back at the performance of these assets, GLD generated the highest return over the period, both in absolute terms and per unit of volatility. Conversely, ACWI generated the 2nd highest return but only the third highest return per unit of volatility, behind XSGS. CGR’s performance was mediocre and GSG’s dismal.

Returns Table

The analysis of each asset class profile in relation to iGDP, iEMC and iLCI confirmed ACWI’s defining characteristics to be its pro-cyclicality, XSGS its defensiveness (ie. best in a low iGDP environment), gold’s profile as an “inflation hedge” (ie. best with low iGDP, high iEMC), commodities’ outstanding hyper pro-cyclicality and pro-inflationary profiles, and real estate’s medium pro-cyclicality and uniquely pro-disinflationary tilt.

The above can be translated in simple dynamic asset allocation rules:

1. Hold maximum (100% in our simulation) Equities whenever iGDP is clearly positive (above 2%)

But hold some Commodities (50% in our simulation) when iLCI is high (above 1%).

2. Mitigate the weight of Equities if iGDP is below 2% and above 0% with Government Bonds (50%-50% in our simulation)

But hold some Real Estate when iLCI is low (below 1%), 33%/33%/33% in our simulation.

3. Replace all Equities and Real Estate holdings by Government Bonds if GDP goes below 0%

But hold some Gold (50% in our simulation) when iEMC is high (above 1.5%)

The chart below represents the composition of the simulated portfolio throughout the period, with Equities averaging 78% weight, Government Bonds (12%), Real Estate (5%), Gold (3%) and Commodities (2%).

Composition Chart

The return and volatility results are presented in the table below. The dynamic allocation to Government Bonds away from equities in low iGDP periods is indeed a major contributor to loss prevention that both improves returns and reduces volatility. The contribution from Gold is marginal over the period and Commodities is a one-off event (between Dec-08 and Mar-09), but Real Estate adds meaningful returns with little impact on portfolio volatility.

Return & Volatility Results

While the impact of such a simple set of allocation rules looks very significant, in effect delivering 2x the (passive) returns of an equity-only tracker (see graph below), it is worth noting that most of the performance uplift takes place between 2007 and 2009 (ie. the period consisting of a commodity super-cycle, followed by a global financial crisis, followed by the launch of zero interest rate policies and large asset repurchase programs).

Comparison Chart

Over the past 10 years, the same dynamic allocation rules only managed to outperform an equity-only strategy on a return per unit of volatility basis (see table below), as Equities enjoyed extraordinarily high returns and low volatilities to the point of almost rendering asset diversification useless.

Performance Comparison Table

We could have changed the nature of the trackers (e.g. different duration and credit risk profiles for bonds) and the type of assets (e.g. Private Equity or Hedge Funds ETFs) but suffice to say that equities will not monotonously run higher for ever, and that “abnormal” market events will no doubt occur again. This will instantly remind us of the benefits of diversification. When this happens and asset covariances assumptions go right out of the window, the simple distillation of macro signals provided by butterwire will hopefully prove to be a quick and useful input to your reflection.

While we believe, perhaps contrary to popular wisdom, that the most important investment decision is that of what positions to hold in an active equity portfolio, it is fair to say that asset allocation is a likely second. As such, we will soon start reflecting these findings in the “Markets” section of the Butterwire app.

The asset classes and underlying rules will be broadly similar to the ones laid out above, and, as illustrated by the asset allocation profile since early 2018 in the graph below, indeed reflects the switch from “Upcycle” to “Downcyle” conditions from early July 2018 and an increasingly defensive and diversified positioning until February 2019.

Since late May 2019, market expectations have been again implying mediocre (0%-2%) global growth and accommodative monetary policies (high iEMC) without inflation (low iLCI), in turn suggesting a highly diversified portfolio as for the Nov-18 to Feb-19 period.

Composition Chart 2