Food for thought on political cycles and stock returns

There is a well know stock effect in the US based on the political party in power. Periods of Democratic presidencies see higher stock returns than Republican administrations. The excess return difference is significant and greater than 10 percent as measure between 1927 and 2015. The last four years, while different, does not change the long-term relationship. A recent working paper has tried to answer this political cycle effect through an endogenous model of economic behavior. See "Political Cycles and Stock Returns" by Pastor and Veronesi.

Their simple model is based on the market response to time varying risk aversion. It is not what president do but when they are elected. When risk aversion is high such during a crisis, voters will more likely elect a Democratic president who will more likely provide greater social insurance. When risk aversion is low during periods of economic boom, voters will more likely elect a Republican because they are willing to take more business risk. 

Higher (lower) risk aversion under Democratic (Republican) administrations will result in a higher (lower) equity risk premium and thus higher (lower) stock returns. The authors show that there is direct link between risk aversion and voter preference. 

Is this year going to be different from the endogenous political cycle story? Economic growth has rebounded, uncertainty is still high, the pandemic crisis has not ended, and the need for social insurance still exist. This environment fits the story for a Democratic administration which is consistent with even late poll numbers, yet the uniqueness of the current crisis and strong voter turnout may make this election closer than forecast. 

 

Back to basics - Always hold increasing returns to scale businesses

What should drive your equity choices for the rest of 2020? The classic factor style approach focuses on value, quality, size, momentum, or volatility which may often seem abstract from the underlying businesses. For example, value or quality is a factor which is sorted and not a strategy or a review of the business and the environment it works in. 

Simply put, successful long-term businesses that produce high returns should be those that have increasing returns to scale that can be exploited, serve as a tailwind, and can stop competitors. 

Most good firm will have at least constant returns to  scale. If a 10% increase in resource usage of labor and capital leads to a 10% increase in production, there is constant return to scale. If the output increases by more than 10%, there is increasing returns to scale. We can look at the classic Cobb-Douglas production function Q = K^a*L^b and work through the math to find conditions for increasing or decreasing returns to scale. Increasing returns to scale will lead to economies of scale or decreasing long-run average costs which support higher profits.

Increasing returns to scale is not the same as just getting large. Many firms think they should merge or acquire others to get scale by sharing supplies, management, or platform costs, but that is not the same as increasing return to scale nor will it create economies of scale with the underlying activities of the firm.

There are number of businesses that will have increasing returns to scale, those with high upfront costs or R&D, network effects, customer learning curves, branding for quality and location. Meaningful increasing returns to scale exist for businesses that create barriers to entry from their scale. For example, look for network scale. The attractiveness of a network product grows through the use by more customers, and more network users create value for all so that even moving to a cheaper product that does not have network is expensive. Firms that set an industry standard and gain a network are especially attractive and will have increasing returns to scale that will be hard to overcome.

However, these increasing return to scale business have issues of winner-take-all competition. There is a degree of gambling with finding these firms especially during their early life stage. Hence, there should be desire for those that have broken-out ahead of other competitors. Investors may be late to early gains, but they have the opportunity for sustained returns. 

The result of an increasing returns to scale business is high industry concentration which leads to demands for some regulatory control. We are seeing this now with the network tech companies. Government has to be large to monitor this concentration. There are political issues if the economy is dominated by large firms and large government for oversight. Small businesses may suffer because they cannot compete or be subject to a single service provider. There are spill-over effects beyond profit and loss but as an investment strategy it will always have merit.

Trend-following strategy as the new safe haven asset

 

Trend-following has often described as a "crisis alpha" strategy, but there has always been a simpler crisis alpha strategy over the last two decades, buying Treasury bonds. Given their strong negative correlation and positive total return during a market decline, Treasuries have all of the characteristics of a great crisis alpha strategy. That story may not hold going forward.

Treasuries have been viewed as the ultimate safe asset. Investors have the minimal credit risk from the US Treasury. Investors trade in the most liquidity market in the world, and if there was a crisis, fund flows would drive the price of the safe asset higher. Treasuries have been negatively correlated with equities; more negative than the correlation between equities and trend-following. The cost was low and in stable times you still received the coupon payments. What was not to like from using Treasuries as your hedge and crisis alpha?

Those halcyon days of Treasuries as a safe asset strategy are over. Liquidity can be a problem as seen in March. The Fed is now the largest holder of Treasury debt which changes the competitive nature of the market. Rates are low so you don't receive any yield especially inside 5-years to maturity, and the lower zero bound places a restriction on the potential for gain. You will not get a return kick to offset losses from an equity decline. Treasuries are not the safe asset as in years past. There is no total return offset to losses from risky assets.

So, what can an investor do to find an alternative safe asset? You can buy other forms of fixed income to offset the low yield, but safety becomes a question. Many other perceived safe assets have high risk characteristics so it is not clear they will provide safety when needed. 

Looking back in time, Treasuries have done a great job of providing portfolio safety and protection. Looking forward it is not clear that this will be repeated. On a forward looking basis, trend following may serve as a superior safe asset strategy. Instead of buying a zero return asset like bonds, investors should hold a strategy that will take positions based market price direction either up or down. Trend-following may still hold bond positions, but only if price direction warrants the exposure.

Buy the positive convexity strategy since the upside gain from holding bonds in a down market will not provide the upside needed to offset losses. Buy the diversified portfolio of trends because it offers a better mix of protection alternatives. It is a simple strategy. There are no guarantees it will do better in the short-run versus bonds, but in the current environment, it can still be the best alternative to a bond hedge.  

Trend-following - Deep learning can help but burden still on setting up the problem correctly

Deep learning techniques are a natural for quantitative investing and more specifically for trend-following, but just because the concept of deep learning is good for finding improved models does not mean it should be used. I am not anti-deep learning but realize that there are high barriers to entry to get it right as well as barriers to success based on costs of transacting. There is no simple answer with using these techniques because the value-added is all conditional on the set-up of the problem and the type of techniques used. 

A recent paper applying deep learning to trend-following shows that it can have a significant improvement in Sharpe ratio and other key statistics, yet there can be high variation in success based on problem formatting. See "Enhancing Time Series Momentum Strategies Using Deep Neural Networks". 

No one should be surprised with this result. There have been hundreds of trend-followers through time that have all had their own variation on trend identification, position sizing, and portfolio construction. There is no single model solution. There is high variation in performance based on the techniques used and assumptions made. 

The researchers form a control strategy without any deep learning and then compare against four different approaches: 

Lasso regression (LINEAR)

Multi-layer perceptron (MLP)

Wavenet based on convolutional networks (CNN) 

Long short-term Memory (LSTM).

The techniques employ increasing complexity with LINEAR being a relatively simple linear model, MLP using a 2-layer neural net, wavenet  accounting for long history, and LSTM using sequence prediction based on memory structures.

These are some of the best techniques for use with time series data and are fitted against a number of objective functions such as Sharpe ratio, return, MSE, and binary choice. All returns are rescaled to a target volatility to ensure fair comparisons. The following two exhibits show the cumulative returns for the set of models. The box plots show the range of results across individual assets. There can be large differences between asset return performance across models.

Deep learning techniques can improve gross performance statistics, but it comes at a cost. More turnover translates to greater drag based on the transaction costs of constant adjustment. The impact of transaction costs is not trivial and can turn the great strategy into a marginal or losing model. The exhibits below show the added turnover and the impact on Sharpe ratio as transaction costs increase. 

Deep learning or any strategy with constant adjustments work effectively when transaction costs are not included or are very low. Increases in cost will have a much greater impact on deep learning models than simple approaches. Hence, transaction costs must be accounted for when building models. 

The challenges of trend-following research are the same across time. Improvements can be made through better and new techniques but costs matter and may lead the investor back to simple models.