Liquidity of Last Resort

A common thread amongst investors and the media at present, is bitcoin’s apparent high correlation to risk assets. Simply eyeballing the data, it looks to be the case for the March 2020 crash, which pushed down asset prices across the board including bitcoin. But it shouldn’t be concluded from this datapoint alone, that bitcoin is a risk asset.

It is true that bitcoin’s correlation to equities has risen in recent years. The chart below, taking trailing 12-month correlation of daily returns, highlights a sharp rise in correlation at the beginning of the COVID crisis in early 2020.

Correlations peaked at a positive 52% to the Nasdaq 100 and MSCI World indices, but have since declined sharply to around 20%, a similar level to gold. Interestingly, this has not impacted the long-term average correlation to equities of 0.04%. It is also worth pointing out that US 10-year treasuries have risen to a 40% correlation against the Nasdaq over the course of 2021, severely threatening the classic 60/40 equity-bond portfolio’s historical diversification merits.

A fair critique of this analysis is that it lacks sufficient nuance. While for most of the time correlation is very low, it is at those crucial periods of market stress where bitcoin’s diversification merits are needed the most. In order to replicate periods of market stress, we have isolated periods where weekly price returns are greater than a 3x standard deviation move in what we define as a “stress event”. For the S&P 500 this has identified 13 stress events since 2013.

The chart above highlights these 13 events for the S&P 500, while overlaying the equivalent stress events for bitcoin prices since 2013. Bitcoin certainly has more stress events than the S&P 500 and to a much greater magnitude due to its volatility. However, the data also highlights that the events between the S&P 500 and bitcoin rarely coincide, indicating that during periods of stress, the assets are not correlated. The March 2020 COVID crisis being the only exception to this pattern.

By isolating the risk asset’s (S&P 500) stress events, we can quantify how bitcoin prices behaved over the same periods. We can also differentiate these events between when prices were rising and when prices were falling.

The data highlights that during stress events, correlations to bitcoin prices rise from 13% to 27% for the S&P 500. Splitting those events into positive and negative momentum events, highlight that bitcoin has its highest correlation during price falls. Incidentally, this is when gold has its highest correlation to bitcoin, this results in some interesting inferences. As we saw during the great financial crisis in 2009, gold fell in-line with risk assets, but this wasn’t due to investors’ fear of falling prices in gold. It was a time of severe liquidity constraints, and gold being one of the only liquid assets prompted investors to sell their holdings to raise cash, which they couldn’t obtain from selling other illiquid assets, thereby helping them fund redemptions or margin calls.

We believe the March 2020 price falls in bitcoin happened for a similar reason, with bitcoin being the liquidity of last resort for the modern age. Bitcoin can undertake this role due to its ability to be traded both during and outside the trading hours of traditional markets, and its speed of settlement compared to traditional markets.

Bitcoin is often compared to tech stocks, but contrary to popular belief, statistics highlight that the Nasdaq has a lower correlation to bitcoin than the S&P 500. Furthermore, a rising Nasdaq does not necessarily mean bitcoin prices will rise as the correlation in positive momentum events are typically negative.

The highest correlation for both positive and negative momentum events is with the Solactive Social Media Index, which includes companies such as Facebook, Tencent, Snap and Alphabet, all companies closely linked to the rising use of the Internet and networking.

As we have written before, bitcoin has a negative correlation to US 10-year treasuries. This negative correlation has become more prevalent in recent years. From this one could infer that bitcoin is a risk asset, however, as we have seen a significant rise in correlation between equities and treasuries over the last year it is difficult to make that conclusion.

While our analysis window began in 2013, bringing the window forwards to the beginning of 2017 had little impact to the overall results. It is also worth considering how bitcoin may behave in a portfolio. Our analysis highlights that since 2015 bitcoin has been much more effective in reducing correlation to the 60/40 equity/bond portfolio than other alternative investments.

In conclusion, the overall correlations are lower than many make them out to be, but during periods of market stress correlations do rise. This is understandable as bitcoin is establishing itself as a store of value with high liquidity, essentially becoming the liquidity of last resort during periods of crisis, rather than being misidentified as a risk asset