Calculating Country Risk Observed by Betas

by: Suna Reyent, CFA

Market participants may observe that world markets have been moving in a synchronized fashion due to technical reasons such as capital flows as well as the increasingly interconnected nature of country-specific macro economic environments.

Can we properly measure individual country risks? Or, has the country risk become less important given the global roller coaster nature of the financial markets?

There are various ways and methodologies of measuring country risks. Yet as far as stock market movements are concerned, we should keep in mind that as markets are more integrated than ever, it is very likely that stocks of many of these countries will go up and down depending on the daily momentum observed in the world markets. When Asian markets have a good day, it may be an extension of the good mood in the U.S., which may or may not be followed by European markets. The trend may continue or change course depending on global economic developments of the day.

For instance, Istanbul Stock Exchange opening depends on how Asian markets have done and on expectations of what Western markets are going to do later in the day. Turkish stock market is influenced by daily expectations related to the U.S. economy in terms of macroeconomic data to be announced later in the day. Also percentage moves up or down are generally sharper and more volatile than those observed in the world markets in general.

If you have been following a foreign market closely, you may have noticed similar observations. Certain country performances have been more volatile than others.

Also as the global markets have become more volatile, the correlations between countries have increased.

The International Capital Asset Pricing Model

In the Capital Asset Pricing Model, a stock’s beta signifies the risk of that stock with respect to the market. Beta is directly proportional to the correlation of the stock return to that of the market as well as its relative volatility with respect to the market. It’s a measure of the stock risk for one unit of market risk. The higher the risk of stock with respect to the market, the higher the expected compensation of the investor with respect to the particular stock.

To keep the matters simple, a high beta signifies high risk, while a low beta signifies low risk. The beta of the market itself is one.

The International Capital Asset Pricing Model is the extension of the Capital Asset Pricing Model where the risk of a specific country can be specified in a likewise fashion with respect to a chosen world index.

I wanted to compare the daily moves of all markets and calculate the observed country risks in terms of betas to see how risky one country is with respect to another. That way I would have a valid methodology to rank countries because I could quantify market risks.

The easiest way for me to come up with the betas was to regress the equity returns of individual countries to the world index of my choice. I wanted to keep the variables as simple as possible in order to include as many countries as possible in this study. I wished to come up with a risk-ranking system for the stock markets of all countries for which there is trading data available.

Calculating country betas is not the only way to compare country risks and it is entirely dependent on past stock returns as well as the global market index returns one chooses to compare the performances to. There are many criticisms of the beta measurement as well as the CAPM theory in general, and I will state a few of them here. For instance, beta will vary depending on the time frame chosen, which may make it unreliable depending on the degree of change. Beta also captures “systematic” risk only, and thus it may give an incomplete picture of the total risk that includes unsystematic risk.

While there is no guarantee that the future beta will look like the past data, it’s a reasonable indicator (as good as any) of the risks one can expect in a certain market.

There are different ways of calculating country betas, and they may involve more sophisticated models than the good-old fashioned OLS regression that readers may have seen in an econometrics class. I wish to keep this article accessible by many, so I will keep such terminology to a minimum. But suffice it to say that such models may explicitly state a country’s degree of non-integration to the world market and involve various country-specific factors ranging from the level of interest rates, inflation and the like. They may also include other global information variables to further specify world risk.

Yet the more sophisticated the model is, the stronger the false sense of security becomes and the more difficult it is to evaluate its shortcomings. Not to mention that such complexities in the model specification would have led me to decrease the number of countries in the study because of data issues, and the depth of the model would have necessitated concentrating on a region or a limited number of countries. I wanted breadth rather than depth.

So what I have done was still data intensive, but more straightforward.

Data, Methodology And Results

I used index data available at the Standard & Poors Index Services where they keep daily data for closing index prices of many countries for which trading data exists. For the series of regressions to be performed, I calculated daily returns of all countries in terms of U.S. dollars.

The country “betas” are calculated by regressing each index return against the global equity portfolio. Beta is estimated as the slope of the fitted line from the linear least-squares calculation.

I wanted to experiment with data that was available. Knowing that markets have become more volatile in the past year or more, I calculated each country beta in two ways: 2-year betas involved 522 observations (daily returns) covering the past two years from December 4, 2006 until December 2, 2008. 10-year betas, on the other hand, involved 2610 observations from December 2, 1998 until December 2, 2008.

Given the lack of data for some of these periods, I had to exclude the following countries: Pakistan, Jordan, Nigeria, Slovenia, Columbia, Iceland, and Luxembourg. I did calculate betas for those countries for which there were adequate observations. However, either due to lack of significant t-values to validate the beta coefficients, or due to other extraordinary issues such as the collapse of the Icelandic market I decided to exclude these countries from the various data matrices I designed. Luxembourg was excluded due to the fact that the equity index houses 10 stocks with 3 firms making up more than 60% of the index and some 45% of the weighting belongs to bank stocks headquartered in Belgium, thus making the calculated beta meaningless. (Note: the weights may have shifted but the idea is that the beta of such an index is not representing the “risk” associated with the country of Luxembourg).

In order to make sure that there were no errors within the Excel regression package, I checked the betas by multiplying the correlation coefficients of all countries (with respect to the world index) by the relative standard deviation of the each country. This was also a good way to see the composition of the beta of each country.

An example of the display for the country of Brazil is given below:

Note: This regression is done with daily beta, thus the standard deviations seen above represent daily numbers. Numbers are rounded to two decimal points, so if you do the math above you may find that the Brazilian beta is 1.79. This error is due to rounding. Brazilian markets indeed exhibited the highest daily beta for the last two years. Series of regressions covering the ten-year time frame also put Brazil at the top of the list in terms of beta.

Another interesting but predictable observation is that two-year betas for almost all countries are considerably higher than their respective ten-year betas. This is due to the fact that both correlations as well as individual country standard deviations have increased recently. But note that the global standard deviation placed at the denominator has also increased. If you look at the composition of the betas, you may see the contribution of various parts of the equation to the betas themselves.

Notable observations include high-beta countries, specifically Brazil, Hungary and Turkey. The emerging nature of these markets attract attention, however some developed markets have also registered riskier betas such as Norway, Austria, Sweden. On the other hand, Mexico, Russia and Argentina are also among the notable emerging markets with higher than average betas. All of these high-beta countries exhibit higher correlations as well as higher than average standard deviations.

A lot of European markets have registered a beta closer to one, which indicates that these markets exhibit an average risk with respect to the rest of the world. The U.S. market with a beta of 1.06 seems to exhibit average risk with respect to the rest of the world as well. However, this is mainly due to the fact that American stock market is the largest constituent of the global market index. Indeed, when the U.S. stock returns are regressed against a global index that excludes the U.S., the resulting beta is 0.57 with an observed correlation of 0.49 as opposed to the high correlation of 0.84 of the original regression. This indicates that the U.S. market beta above is overstating the U.S. market risk. According to the beta of the second regression, the U.S. is actually towards the bottom of the list of these countries.

Asian countries are the winners of this exercise in terms of lower risks associated with their markets. It is remarkable that some of the “emerging” markets of the Asian region have registered lower betas associated with a lower degree of risk, among them Thailand, Philippines, Taiwan, and Malaysia. Hong Kong showed a lower beta than I expected, and the presence of Japan towards the bottom of the list with a beta of 0.41 was another surprise for me.

Among the lower beta countries were those of the Middle East and North Africa, namely Israel, Egypt, and Morocco. I must add that Jordan, Pakistan, and also Nigeria would have made it into the bottom of the list if I had included the results I was able get from the regressions I ran (notwithstanding some insignificant t-statistics associated with the regressions). I may include these countries in a following write-up if I am able to get hold of more complete data sets.

Ten-year table country betas are strikingly lower than their respective two-year betas. There are a few exceptions: Finland occupies the second place in the ten-year list with a beta of 1.29 whereas during the last two years, its beta has decreased to 1.10 placing it towards the middle of the same list. Even more interesting is the fact that Israel registered a beta of 0.78 for the last ten years whereas its beta has fallen to 0.42 during the present time when the world has become a more volatile place. Japan seems to have resisted the trend somewhat but its ten-year beta of 0.46 is not strikingly different from its two-year beta of 0.41.

Brazil occupies the first place in both tables as the riskiest of the bunch. Furthermore, a quick look at the former ranks at the ten-year table shows that Turkey, Hungary, Russia, and Norway have gained significant risk in terms of their respective betas during the last two-years. Turkey and Russia deserve extra attention in terms of their standard deviations for the last ten years. Despite the fact that Turkey registered the highest daily standard deviation of 3.33% during the last ten years, its relatively low correlation of 0.32 managed to keep the country from occupying a higher rank. This was also true for Russia who had the second highest standard deviation of 2.65% and a somewhat lower correlation of 0.41. However, during the last two years, the correlations of these two countries have increased to 0.69 and 0.62 respectively, thus placing their betas higher on the two-year list.

The higher beta associated with the U.S. is misleading in this list as well. Indeed, when ten-year U.S. returns are regressed against global returns excluding the U.S., the resulting beta is 0.57 (same as the two-year beta) with a correlation coefficient of 0.46.

Asian countries occupy lower ranks in the ten-year list as well. Among them are emerging (or so classified) markets of India, Indonesia, Thailand, Taiwan, Philippines, and Malaysia, besides the developed markets of Japan, Hong Kong, and New Zealand.

Egypt and Morocco occupy the last two places in the ten-year list similar to the pattern observed in the two-year list.


An obvious conclusion of this exercise is that country risks have increased over the past two years when compared with ten years of data.

It would be interesting to do the same study with weekly or monthly data as well, and perhaps by going back further in time. However, going back further in time risks capturing certain fundamentals or situations that no longer exist in today’s marketplace. Increased correlations due to the effects of globalization are examples of such fundamental or “regime” changes.

Another notable observation of this exercise is that certain countries manifest higher risk than others on a consistent basis. The results of this exercise have captured what has been familiar to careful observers of world markets.

As the Capital Asset Pricing Theory suggests, higher risk results in higher return expectations on the part of investors. Thus a higher risk should be evaluated according to return expectations of these countries. Conclusively, it may be a valid decision to invest in countries (or stocks) that exhibit higher levels of risk depending on long-term growth expectations, cheap valuations exhibited by fundamentals, or both.

Yet another remarkable observation of this exercise is that there are emerging markets with higher expectations of return that also manifest lower risk. Certain emerging markets of Asia as well as North Africa and the Middle East have exhibited lower risks as measured by their betas.

An inference of this study (or so I reckon) is that if all else is equal, it is more advisable to invest in nations with higher than average growth potential that also show a low degree of risk with respect to the world markets. The diversification effects are also valuable. Of course, all else is never equal and such quantitative studies should always be complemented by fundamental analysis as well as due diligence.

A disclaimer related to this exercise should be announced in the sense that today’s high-risk countries may become tomorrow’s low-risk ones as well. Nevertheless, given what we can observe in the world markets today, investors should evaluate whether their return expectations are worth the risks.