Saturday, January 9, 2016

January 2016 Data Update 3: Country Risk and Pricing

I had a long  post on country risk in July 2015, as part of series of posts on the topic. At the time of the post, the Chinese market was in the midst of a meltdown, emerging markets were in turmoil and exchange rates were on the move. It is six months later, and nothing seems to have changed, but I think that the core lesson is worth reemphasizing. In a world of multinational businesses and global investors, there is no place to hide from country risk. 

Country Risk Measurement
I will not bore you by repeating much of what I said in my earlier post on how I view country risk in valuation, but it is built on two presumptions. First, a company's risk exposure is based on where it does business, not where it is incorporated or headquartered. Thus, Coca Cola and Nestle may be incorporated in developed markets (US and Switzerland) but derive a significant portion of their revenues from emerging markets and are thus exposed to risk in those markets. By the same token, Embraer is a Brazilian company that derives a substantial portion of its revenues in developed markets. Second, the risk of investing in equities varies across the world, resulting in higher equity risk premiums in some markets than others.  To estimate these risk premiums, I follow a four-step process:
My paper on equity risk premiums
As an example, let's assume that I want to estimate the equity risk premium for operating in India in January 2016. 
  1. I start with the implied equity risk premium for the S&P in January 2016, which I estimated to be 6.12% in my first data post a few days ago. I use a rounded down estimate of 6% as my mature market premium for the start of 2016.
  2. As a second step, I look up the local currency sovereign rating for India from Moody's and arrive at a Baa3 rating; the typical default spread for a Baa3 rated country at the start of 2016 was 2.44%.  I check this estimate against the sovereign CDS spread for India, which was 2.11% on January 1, 2016. I use the ratings-based spread of 2.44% as the default spread for India, though I would not raise too much of a fight, if you insisted on using the CDS spread.
  3. In the third step, I try to estimate how much riskier equities are than government bonds in emerging markets by using proxies for each one: the S&P Emerging BMI Index (an index of emerging market equities) for stocks, and the S&P Emerging Market Public (government and quasi government) bond index yield. The standard deviation in the former is 17.36% and the coefficient of variation in the latter is 12.91% and the ratio of the former to the latter is 1.34. Multiplying this ratio by the default spread in step 2 yields a country risk premium for India of 3.28%. (CRP for India = 2.44% * 1.34 = 3.28%)
  4. In the fourth step, I add the country risk premium to the implied premium of 6% that I estimated in step 1 to arrive at an equity risk premium for India of 9.28%.
Is this number an estimate? Of course! Would you get a different number if you used the CDS spread as your measure of default risk and different indices for emerging market equities and bonds? The answer is yes. It is for this reason that the spreadsheet that I create for equity risk premiums allows you to replace my defaults with yours for any or all of these variables. Before you exhaust yourself in this effort, I would suggest that small differences in this number will not make or break your valuation. So, make your best estimates and move on!

Country Risk Update - January 2016
Using the approach described for India, I compute equity risk premiums for the 130 countries with a Moody's sovereign rating. For about fourteen more, with no Moody's rating for the country, I was able to find a sovereign rating on S&P that I convert to a Moody's rating and estimate an ERP. Finally, there are about 20 countries, loosely categorized as frontier markets, for which there is no rating or CDS spread; these include the hot spots of the world such as Syria and Iraq. For these, I use the only measure of country risk that I can find, a composite risk score from Political Risk Services (PRS) and use that score to compute an equity risk premium; I create a look up table using the countries that have both PRS scores and ERP to make these judgments. Desperation move? Perhaps, but if you can find a better way of doing it, I would be glad to follow your lead. The resulting equity risk premiums by country are available in the spreadsheet that I referenced earlier but are also in the map below (which adds nothing in terms of content but looks much better):


Country Pricing Update - January 2016
In my July 2016 updates, I also included one on how stocks are priced around the world, using multiples (PE, PBV, EV/Sales, EV/EBITDA, EV/Invested Capital). While that post has a more extensive explanation of why stocks should trade at different multiples around the world, I have updated the multiples, by country, in this spreadsheet. As you peruse these numbers, keep in mind that the number of companies that I have in data set is very small for some countries and the multiples can therefore yield strange values. To prevent outliers from hijacking my estimation, I also compute the multiple using aggregated values; thus, the PE ratio for China is computed by adding the market capitalizations of all companies listed in the market and dividing by the aggregated net income of these companies. 



Much as I would like to read more into this picture (especially about cheap and expensive markets), these country numbers are more a first step in the investment process than a last one. 

Bottom line
I think that we are far too casual in our treatment of country risk, estimating equity risk premiums on auto pilot for countries and attaching these premiums to companies based on where they are incorporated, rather than where they do business. If there is a lesson from the last week's implosion in the Chinese market, it is that the emerging market growth story that so many developed market companies have pushed for the last two decades has a dark side, and that dark side takes the form of higher risk. It is easy to forget this intuitive concept in the good times, but the market lulls us into complacency before shocking us. 

Datasets

8 comments:

Jeff Robinson said...

Such great content should be shared. Could you please add a "shareit" plugin or similar so we can get the word out more to our communities?

Thank you,

Jeff

Martin said...

Just with one look on the map: Netherlands are an outlier. The market cannot be that cheap or are these zoombie banks and insurers?

Anonymous said...

Very timely given the chatter about Aramco of Saudi Arabia possibly looking to sell a minority stake to the public. I just wonder with all the turmoil in the middle east if the ERP for that country might be understated

Marko said...

Dear Professor Damodaran

Your work is, as always, interesting and thought provoking.

Your thoughts on the below country risk considerations would be appreciated.

1. Diversifiable vs. non-diversifiable
a. By including this into the discount rate, the assumption is that ‘country risk’ not diversifiable.
b. From the perspective of the ‘fully diversified portfolio’ surely country risk is diversifiable?
c. How can, for example, expropriation be classed as ‘systematic’ and included into the discount rate?

2. ‘Sovereign Risk’ = ‘Country Risk’
a. You method and calculation seem to make the assumption that ‘Sovereign Risk’ = ‘Country Risk’.
b. Surely the risk of a sovereign defaulting on its bonds and the country risk factors we have to take into account when doing a valuation of, for example - a chicken soup factory in that country, are not the same.
c. Even though I agree there is a correlation in general, I can think of many ‘country risk’ factors that are important that are either completely uncorrelated to ‘sovereign risk’ or even in some cases negatively correlated to it.

3. ‘Contractual Cash Flows’ = ‘Probability Weighted Cash flows’
a. You measure the ‘sovereign risk’ looking at yield-to-maturity on bonds (or alternatively CDS spreads)
b. The implicit assumption here is that the promised / contractual cash flows used in the calculation are equal to the expected cash flows (probability weighted cash flows).
c. Surely this is not the case (maybe it is close for the US but definitely not for Greece or Russia for example).
d. Valuations deal with ‘expected’ / ‘probability weighted’ cash flows so we have to be consistent when constructing the discount rate.

Obviously ‘country risk’ is a real consideration that has to be taken into account. Because of the above I cannot see how, given the financial theory available to us, we can include it into the discount rate. I understand it’s easy and convenient to use this method when compared to the alternative but surely that is not sufficient reason?

The alterative unfortunately is difficult and requires a lot of work. Identify and quantify the actual list of factors that impact and asset in the country it operates (strikes, expropriation, change in regulation, political upheaval, etc.., etc...) and model this in the cash flows.

Thanks in advance,

Marko

Aswath Damodaran said...

Marko,
A friendly suggestion. Don't use the word surely unless you are absolutely certain, and since there is little in finance that is that certain, don't use it in finance. In fact, both your surelies are not so sure. First, country risk is diversifiable only if it is relatively uncorrelated, something that may have been true in the 1980s, but not anymore. There is a systematic country risk factor, where worldwide shocks reverberate across countries and the CRP is my estimate of the exposure to that factor. On the sovereign risk question, you are right that sovereign risk is not equity risk, but can you point me towards one other market measure that I can use instead? (I really am willing to switch, if I can see something better). The sovereign yield is a classic problem with bonds in general. If your point is that bond yields understate country risk, there is an easy solution. Use the CDS. You do talk about an alternative to incorporating country risk into discount rates. I am open again to suggestions but tif by adjusting cash flows for country risk, you just mean incorporating bad scenarios into the expected cash flow, that is still an expected cash flow and not risk adjusted.

Anonymous said...

Hi Professor Damodaran

If I was looking to determine how much assets to allocate to private equity funds in Asia vs. N. America and wanted to use the country risk premium, how would you go about calculating that specifically for PE funds? How would you calculate a return target based on the additional risk that you take in Asia vs. N. America?

What is the best way to risk adjust returns from both US and Asia so I can make them comparable for asset allocation purposes. What sort of risk metrics should I be considering.

Many thanks.

Luiz said...

Hi Professor,

Does currency matter here? For example, an investor that uses USD will invest in equities in Brazil (BRL). How we account for that?

Thanks,

Luiz

Anonymous said...

Thanks for sharing your analysis and update! I was wondering why you use the local currency sovereign rating (instead of the foreign currency one). Is this to avoid increasing the country risk for capital controls or political decisions (given Moody's note on local v foreign currency "A rating gap (a notching between a government’s LC and FC bond ratings) is now only applied in those cases where there is (1) limited capital mobility; and (2) a government which either faces constraints in terms of external liquidity, or, in exceptional cases, shows a material and observable distinction between its ability and willingness to repay creditors in LC versus FC, or vice versa.") Thanks!