Article

Episodes change the risk properties of assets: Brazil edition

The last two weeks have seen a relatively sharp increase in developed market government bond yields, seemingly driven by a perceived change in tone from central banks.

The moves are actually relatively modest, merely bringing yields back to levels seen at the start of the year.

However, phases in which a single issue seems to dominate market behaviour can provide information about perceptions of risk, correlations, and the potential for diversification. For example, if we interpret recent moves as an indicator of perceived upward pressure on global real rates, we might expect most government bonds around the world to share the same behaviour. This is indeed the case, with one notable exception, Brazil:

Similarly, when real rates increase in the absence of stronger growth news, you may also expect other assets like equities to be impacted negatively, given a rising discount rate.  Again recent price behaviour seems to partly illustrate this, but Brazil, along with several other emerging markets bucked the trend of equity weakness.

It is plausible that this is in part due to the role of resources in the Brazilian economy;  South Africa and Russia also outperformed, while a look at performance by global sector shows that materials was the only sector other than banks to generate a positive return.

However, many resource-focused economies also saw their currencies weaken quite materially versus the US Dollar, while the Brazilian Real was one of few to strengthen.

Why might Brazil have displayed such resilience across asset classes? In short, because path dependency and the starting point of valuation matter. It is often the case that idiosyncratic episodes in pricing can change the characteristics of an asset going forward. In the case of Brazil, there was a clear event in May surrounding a corruption scandal involving President Temer. This impacted equities, bonds and the currency, all of which fell fairly sharply.

Because Brazil had already had its own episode, there was greater scope for it to behave independently in the face of the moves we have just seen.

Theoretically this shouldn’t be the case. It runs entirely contrary to efficient markets thinking, whereby all assets are priced correctly and prices can only move if ‘something new’ happens. According to efficient markets views there is no scope for an asset to go up ‘just because it had gone down too much beforehand.’

However, if we believe that episodes can alter the likely future characteristics of an asset, it challenges two important aspects of how we tend to think about risk:

  1. The simplistic ‘risk spectrum’ of assets that we tend to hold in our heads is flawed. The view that sovereign bonds are safe, corporate bonds less so, and equities and emerging market currencies are risky is true with regards to the inherent qualities of the assets (sovereigns are less likely to default, corporate bond holders have priority over equity holders in the event of bankruptcy etc.). However, when it comes to price behaviour, this is not always true. Brazilian assets were actually the ‘safety asset’ in this recent phase, as was banks sector equity, which has been seen as very risky ever since the financial crisis.
  1. Backward looking risk models can create a false impression. Because Brazilian assets were highly volatile in May, risk models which overweight the recent past would have identified them as risky positions to add to a portfolio. However, the very fact that they had just experienced that volatility actually increased the propensity for diversifying behaviour. If such episodic volatility is accompanied by improving valuations then the ‘margin of safety’ that is created can also reduce risk. (The converse of this idea is also true – with markets having just gone through a period of very low volatility and a re-rating in many assets, many simple risk models might indicate that there is room to add to exposures today – a very dangerous assumption!)

The diversification properties of positions in a portfolio are therefore highly dynamic. Nor is this limited to single assets. Emerging market currencies are traditionally (and rightly) perceived as risky but many provided positive return while equities sold off amidst the recession fears in early 2016. In part this was due to the legacy of price weakness that many of these currencies had already seen in the ‘taper tantrum’ of 2013 and especially the commodity weakness of 2014 and 2015. More importantly, as we and others have noted, the negative correlation between bonds and equities that has made bonds an ‘insurance policy’ for multi asset portfolio is a relatively recent phenomenon, in part driven by how badly damaged bonds had been in the 1960s and 70s.

Of course, the key is to achieve diversification when you truly need it, and not just in phases of moderate volatility such as we have just seen. The relationship between risk events and correlation patterns is non-linear; Brazil might have been resilient in this phase, but a fiercer market shock (particularly one sparked by global growth fears) could well have hit the currency hard as investors rush for liquidity. However, this example does illustrate that portfolio construction is an ongoing and forward-looking process. You cannot add diversification to a multi-asset portfolio in the same way at all times, and history is not always your best guide.