What scares any asset allocator - Correlated markets

October was a rough performance months for investors. Risk-off could be the best explanation; however, the reality of the Fed raising rates by the end of year seems to be sinking into the expectations of all investors. There was no protection from holding the long bond with a decline of over four percent. What is our biggest concern is the correlation between stocks and bonds for the month. There was no diversification gain.

Clearly, the desire to hold less long exposure has become the driver in fixed income with new talk that inflation is around the corner. Yes, inflation has increased and is closing in on the magic 2% target but there is little sense that inflation will overshoot regardless of some comments by Fed officials that this could be desirable. While equities are supposed to be protected from inflation, an inflation shock may be disruptive to both asset classes in the short-run.

The biggest losers were small cap stocks both value and growth. In a rising rate market with limited growth, small caps are usually hit harder from the effect of leverage and sensitivity to top-line revenue growth. Emerging markets both stocks and bonds behaved like a safe haven. DM stocks as measured by the MSCI world index fell more than large cap US stocks. Global DM bonds fell to the same degree as US bonds. The only safe places for investors were mortgages which have shorter duration and limited credit risk and commodities which saw a bump in prices outside of the energy complex. 

Narrative and numbers - You need both for effective forecasting

A provocative post by Peter Lupoff the founder of Tiburon Capital called "When numbers cloud meaning - The fallacy of investment research exactitude" has me thinking about narrative versus the idea of false precision with quantitative analysis. First, something to put the issue into context; a classic joke on false precision, "I am 98.54% certain that you need both precision and narrative to be an effective trader." 

Quant analysis without narrative is lifeless, and narrative without data is just providing opinions without support. We can be precise in pricing derivatives, but we may not know whether these derivatives are valuable. We can forecast exchange rates using past money and interest rate data, but we may not understand the next move by the Fed. We may have data that say stock CAPE is high, but we still may not know whether the market is rich given the environment. 

Quant modeling can provide sensitive to a set of factors. It can generate levels of significance to historical data relationships, but it may not provide unique context with the current environment. The finance and accounting MBA's can give you the numbers, but they may not tell you how to grow sales or develop new products.

Passive indexing, factor investing, and smart beta may help investor find effective base portfolios, but these portfolio tools may not help determine whether stocks or bonds are undervalued and ready for adjustment. To add portfolio value, there needs to be a story or narrative that tells something beyond the data as presented in the model framework. If you don't have anything more to say with a narrative, the fallback position is quant analysis and data. There is nothing wrong with a reliance on quantitative analysis and it may be more effective than a poor narrative not grounded in history, but narrative can address the key issue of uncertainty. 

Quantitative work sets the stage for a good narrative to explain the numbers and add something that has not been measured. The narrative is necessary because there are some current events or expected future events that cannot be effectively expressed in past data. Uncertainty is different than risk, which can be measured with a quant model. Once there is a need to address issues beyond risk there is a need for narrative. Quantitative work handles the risk and narrative address uncertainty. 

Decision-making - The need for a coping strategy

The behavioral finance revolution has added immensely to our knowledge on the aberrations from efficient markets and rational expectations. Mistakes happen because we are sloppy thinkers. However, the wide list of behavioral biases does not help us define what it means to be rational in real life situations or how we learn to be rational under different circumstances. Solving for rational decision-making is a difficult problem. My view is that the first choice is to use quantitative tools given that the investment world is rich with data, yet there have been some interesting attempts at finding broader solutions to incorporate other thinking.

An interesting framework for how to look at problems is referred to as ecological rationality. The tools to be used for a given decision should be related to the environment faced. Determining what to do if you are a fireman at a burning building should be different than if you are analyzing the cheapness of a mortgage. Similarly, analyzing the potential success of a new business strategy is different than looking at the trade-off between stocks and bonds.

There are a number of key researchers in this important area, including Gerd Gigerenzer and his team at the Max Planck Institute for Human Development and the naturalistic decision-making expert, Gary Klein. For them, context is important. They try to solve the puzzles that have been aptly researched by Daniel Kahneman, the Noble Prize winner. Data rich problems lend themselves to quantitative analysis; however, new uncertain problem may need other approaches.

To simplify much of this work, the focus has to be on basic coping strategies for any decision process. This coping can be broken into three parts: categorization, heuristics, and narrative.

The primary strategy issue is categorizing the problem. In the case of uncertainty, is the problem an issue of a lack of information, unreliable information, or an inadequate understanding of the available information? For investment issues, the most likely problem is inadequate understanding and not a lack of information.

The second strategic issue is determining what type of model or heuristic to be used. In a Kahneman world, does this require fast or slow thinking? Does the problem lend itself to a quant tool or does it require different thinking?

The third issue is forming the correct narrative. Decision narratives are critical for managing and understanding actions taken. How should the problem be framed? What is the expected answer? How do you assess whether the process was successful?

Good decision-making requires discipline even if the problem is not well-defined or ambiguous. Good solutions are important, but the process is more critical for success.

Cognitive biases and the need to be quantitative!

The list of cognitive biases that can affect investors keeps growing. An explosion of studies show that observed decision-making under real and test conditions is hard.  Just look at the wheel from Buster Bensen's cognitive bias cheat sheet, the single best graphic I have seen, which lists and categorizes the cognitive biases investors face, to get a flavor of the problem. This work does a good job of reducing all of these biases into four problem categories: 

1. What do you do with too much information?

2. How do you find meaning within all of this information?

3. How do you act quickly given this overload of information?

4. What should you remember to make sure you minimize mistakes?

We generate rules or heuristics to solve these four problems, but shortcuts may generate decision biases. We may be able to make fast decisions, but that does not mean that these choices are good or the best decisions. Solving problems using rules of thumb create biases which means that investors may leave money on the table and mistakes will be made. Performance may suffer and markets are less efficient.

These biases may not always be flaws but rather attempt to solve core problems of information overload. The biases may result in failures in thinking, but problems of information overload and making fast decisions are still real and have to be addressed. The issue with the behavioral finance work is that the number of biases just keep getting larger, yet there seems to be a shortage of answers of how to address these biases in order to minimize mistakes. Our ability to break-away from these biases has not match the growing problem list. We can agree that human decision-making is flawed, but what comes next? 

I think the question of dealing with behavior biases can be answered. It just has not been directly addressed as a core solution. The four problems with information overload and cognitive biases can be solved through embracing quantitative tools. 

• If there is too much information, then use statistics which at its core reduces the need for all information. The focus of statistics is associated with condensing information through finding key measures and reliable relationships. 
• If there is too little meaning in the information, quantitative tools can be used to find significance. Data that are not meaningful are not used. 
• If there is a need to act quickly, models can be pre-loaded and easily updated to provide new one-step ahead forecasts. Bayesian statistics can be used to update information in a systematic fashion. 
• Finally, quantitative tools can be used to provide useful memory. Relationships from the past can be stored and change sin those relationships can be measured objectively.

Quantitative tools can be employed to eliminate specific biases for problems that are repeatable and structured as often found in trading. There may be biases with any quant model, but the biases and errors can be measured and followed. How deeply quantitative tools are embraced is a matter of preferences, but the solutions to cognitive biases are right in front of us if we structure the information problems properly.

Emerging market investing - Play the skew

Skew can be an important component of returns. Obviously, investors would like to avoid negative skew, but if an asset with a positive skewed return distribution can be found, it can potentially generate a nice upside stretch with performance. Still, skew is sensitive to outliers and hard to measure. Skew is often generated from mixed distributions; nevertheless, if you can find positive skew investments and can associate this property with specific factors, portfolios can be structured to generate some extra upside return potential by increasing allocations to these assets.

Positive skew has been found in emerging market equities and it is systematically related to economic factors that can be measured and independent of developed market behavior. See "Why Invest in Emerging Markets? The Role of Conditional Return Asymmetry?" in the Journal of Finance. This work makes intuitive sense and is relatively simple to apply through the use quantile-based measure of skew. (The authors use a variation of Bowley's statistic first developed in 1920.)

This positive skew effect is stronger for EM relative to developed equity markets. The result of these finding is that portfolios should be tilted to EM country exposures that have less negative skew. Moreover, the authors find that the positive skew is related to the degree of financial and trade openness. In particular, those countries that may be less sensitive to "sudden stops" in currency markets will have more positive skew. Financially open but trade closed economies will show more negative skew. When portfolio allocations are made which account for quantile skew measures, EM markets will be given significant higher weight and will result in stronger returns. If you want more distributional upside, hold more emerging markets.

Given these results, it would be interesting to test the same novel approach to measuring skew with assets other than equities. For example, hedge funds strategies that may have more positive skew should also have greater allocation tilts within a portfolio.