Friday, June 20, 2008

Evaluating Risk and Reward

We all take risks, especially when investing our hard earned money. And, most of us have heard that risk and return are closely linked. Risk is an important factor in asset selection, because if you select an investment that is more volatile or goes down more than you can handle, you may be tempted to make irrational choices at the wrong time – such is typical of the “sell low and buy high” crowd.

As I approach the brink of financial independence, I want to reduce my risk. I won’t need to take so many chances anymore. I will have enough. Of course, I do not want to eliminate all risk as I believe some is necessary to continue to grow, slow and steady. So, how does one go about assessing risk?

It’s easy to calculate the returns on investments, but evaluating the risk is much more nebulous. Everyone has a different definition of risk and how to quantify it. One approach that I have used in the past applies the statistical measure called standard deviation.

Standard Deviation can be calculated based on returns over several years and provides insight into how often an investment's return is different than its mean return and by how much.

How much return does risk-taking buy you? To help answer that question, I created the following table with data that was pulled from this Matt Krantz article in USA TODAY. The analysis begins with the average annual compound returns and standard deviation for several asset classes over a bunch of years. The return data was derived from the market research firm Global Financial Data.

From there, I calculated the 68% confidence intervals for one standard deviation. For normally distributed data, 68% of the values will be within one standard deviation of the mean, while 90% will be within 1.645 deviations of the mean. So, it is very likely that the return for an investment will fall within the minimum and maximum values provided in the table. I then determined the return to risk ratio for each investment type.

The return/risk ratio is the amount of percentage points of return that can be expected for every percentage point of risk. In other words, how much risk is involved in achieving the return. Volatile investments, with average or poor returns, such as oil and emerging markets, have low ratios.

With a little bit of study, the table can provide some valuable information. For instance, you may notice that value-priced stocks have beaten the S&P 500 and are considerably less risky. In fact, its even money for value stocks. Surprisingly, corporate bonds have an excellent ratio providing good returns for low volatility, while municipal bonds do not appear to pay out enough to compensate for the downside risk.

Housing is another interesting asset class that has an extremely low return to risk ratio. It averages 3.5% return, but it has significant volatility. This is one of those asset classes that is difficult to make a broad assessment because property values vary considerably with location. While I have seen steady appreciation in my neighborhood, others have seen sky-rocketing home values and now plunging prices. Another thing to keep in mind when looking at this data is that some of these assets, like gold and housing, have not historically been considered investments.

One other item of note is the years column. This indicates how far the data series goes back – the more years, means more data and more reliable numbers. Gold is the only asset class with data for more than 200 years.

What other information can you glean from this data? Were you surprised by the return to risk ratios associated with certain asset classes?


  1. Nice table. Lots of interesting info to chew on. Will comment more later on after I've had more time to look at the data.

  2. OK, I'm back.....

    At first glance, this table looks like it can be quite useful in determining the levels of risk and reward in choosing our current investment strategies. However, on further consideration, there are several major caveats that reduce the overall usefulness of this information:

    In order for the table to be really useful, one has to make the assumption that the measurements as well as the items being measured have been consistent throughout the years and will continue as such into the future. The problem is the further back you go, you start losing consistency that makes the data relevant in today's world. Take gold for example. It has 313 years of data, but 300 years ago was an entirely different world with no USA for starters. Investments as we think of today didn't exist as there was no such thing as retirement either- you worked to live with little or no help from the ruling parties and no expectation of not working anymore.

    Next the long time frames themselves pose risk. A rate of return over a period of time can be deceptive in that most of the positive growth moves can be in just a few years with the majority of years having subpar performance. Global Financial Data should have included shorter time frames to give a better perspective of current performance, perhaps breaking out the data in 10-15 year segments. If one enters an investment in a area that has already experienced a big move, it could take decades for that person to participate in above average returns.

    Then we have definition interpretation such as what qualifies for growth, value, emerging, etc.., and whether they have accurately kept apples with apples when measuring long term performance.

    We also have the the probability of future consistency to think of. Oil has been a very risky investment historically with low returns, but I think we are now entering a phase where this will no longer be the case. Increasing demand and limited supply is establishing a floor for this product that will inevitably rise with time and supply/geo-political pressures. Sure there's a bubble going on now that will surely pop, but the pop will just establish reasonable entry points. I think it's safe to say that we won't be seeing cheap oil or $2.00 gas for our foreseeable future.

    On the other side of the spectrum, you have BRKA with has experienced superb performance with minor risk. However the man who was responsible for that great return will most likely not be around to continue managing for another 27 years, and so the risk increases with whether his protege will be able to continue this stellar rate of return.

    I think the table does give some insight to historically stable and volatile investments- which should serve as a guide for what to "trade" (short-term) vs what to "invest" in. But the average Joe or Jane will face the danger of trying to take the "easy" way out by looking at the table and just picking what is listed as the best products without trying to really understand how these numbers tie in to our current investment environment.

  3. Soulfire: You make some very good points. Thanks for the comments.

    You mention that the "long time frames themselves pose risk. A rate of return over a period of time can be deceptive in that most of the positive growth moves can be in just a few years with the majority of years having subpar performance." That is true but I think the data actually helps uncover that. A low average return over numerous years with a high std dev (see oil) indicates a volatile asset that is great for speculation (traders timing the market), but not long term holding.

    I actually consider BRKA moderate risk (std dev is >20), but what this data says is that you get paid back for taking that risk with Buffet. Historically the stock has had lots of upside with only a little downside (min return is -2.1%).