Currency Choices in Valuation

I am currently in Bogota, Colombia, doing a seminar in risk. One of the topics that came up yesterday was about the choice of currency to do a valuation in, and how it affects your inputs. In particular, the question that I was asked was whether an analyst should value a Colombian company in Colombian pesos or US dollars, and the implications of this choice. Here is how I responded:

Which currency should I do my valuation in?
If you do your valuation right, it should not matter. Your value for a company should be the same, no matter what currency you choose to value it in. Thus, a company that is under valued by 20%, when you do your valuation in pesos, should remain under valued by 20%, when you do your valuation in US dollars.
Given this proposition, you should pick the currency with which you are most comfortable with and where it is easiest to get the financial information. My instinct, given the latter requirement, is to do valuations in the local currency since most financial statements are richer and more detailed in the local currency. Your choice of currency should not be a function of the investor for whom you are doing the valuation. Thus, you should not try to value a Colombian company in US dollars, just because the investor for whom you are doing the valuation is dollar based.
 
How is my discount rate affected by my currency choice?
In the context of discount rates, the input that is most influenced by the currency choice is the riskfree rate. If you work with a higher inflation currency, the riskfree rate will be higher. In the Colombian context, the Colombian peso riskfree rate was 6.5% and the US dollar riskfree rate was 4% last week. The difference of 2.5% is entirely attributable to differences in expected inflation.

Just as a side note, while getting a US dollar riskfree rate is easy (I used the T.Bond rate), I had to work a little harder to get the riskfree rate in pesos, since the peso-denominated Colombian government bond does have some default risk embedded in it. In particular, I subtracted out the default spread for the Colombian government (about 2%) from the bond rate (8.5%) to get to the riskfree rate.

The other inputs remain pretty stable. Betas should measure the business risk of the company. I have never understood the rationale of a widely used practice of using betas against the S&P 500, when doing dollar based analysis, and switching to betas against local indices, for local currency analysis. Those of you who follow my work know that I am firm believer in using sector or bottom up betas. For Ecopetrol, the Colombian company, I estimated a beta of about 0.80, based on the fact that it was an oil company, and used that beta for both US dollar and Colombian peso analysis. Even more dangerous is the practice of using the US equity risk premium, for US dollar analysis, and the much larger Colombian equity risk premium, for peso analysis. The company is a Colombian company and you cannot make the country risk go away by switching currencies. Both the dollar and the peso analysis therefore should use the higher Colombian risk premium.

As a final note, the cost of debt should be in the same currency that you estimate the cost of equity in and this is true no matter what currency the company actually borrows in. Therefore, if the company borrows in US dollars but you are doing your analysis in pesos, you will have to restate the cost of debt in peso terms.

How are my cash flows affected by my choice of currency?
The key rule here is that your cash flows have to be in the same currency as your discount rate. Thus, if you decide to do your analysis in pesos, you cash flows have to be in nominal pesos. If you decide to do your analysis in dollars, your cash flows have to be in nominal dollars. If it is a company with Colombian operations, this will often mean that you have estimate the cash flows in pesos and convert them into dollars. You have to use forward or expected exchange rates (and not the current spot rate) to make the conversion. In fact, if you want to preserve consistency, your expected exchange rate has to be computed from either interest rate or purchasing power parity. In the context of Colombia, for instance, the 2.5% higher inflation in Colombia that I have built into the riskfree rate will translate into an expected devaluation in the peso of about 2.5% a year.


Can I avoid this currency choice altogether?
You could, if you do your analysis in real terms. Thus, your discount rate has to be a real discount rate; the real riskfree rate is about 2% (I used the inflation-indexed US treasury to get this) and you can build the rest of the inputs on top of this rate. Your expected cash flows should be real cash flows; thus, you cannot count the inflation component of growth. Again, if you do it right, you should get the same value.

The bottom line: Make your choice of currencies at the start of the process and stay consistent with that choice all the way through. If you are wrong about expected inflation, it will cancel out - both your discount rates and cash flows will change. If you are inconsistent about inflation, applying one rate to cash flows and another to discount rates, your valuation cannot be salvaged.