Burton Flynn and Ivan Nechunaev
What happened in Argentina last month didn’t surprise us. Following a comprehensive study of emerging market currency crises over the last 30 years, we sold our holdings in Argentina earlier this year. Here’s why.
As value investors focused on small emerging markets, we scour dozens of less covered countries searching for quality companies that others might not fully appreciate. This bottom-up approach led our fund to return 62% over its first four years, a track record which fewer than 3% of EM-focused funds managed to exceed. Even though the fund’s fifth year was a difficult one for emerging markets broadly (the median fund down -7%), we might have been one of only 3% of EM funds to post a positive return that year if it were not for our 25% exposure to the Turkish lira and the Pakistan rupee which devalued 40% and 20% respectively that year. Instead, we ended up losing -11%, a result that put us in the bottom 20th percentile for the year.
For the first time, we came to the painful but admittedly necessary conclusion that we needed to engage more deliberately into macroeconomic analysis. We know how to identify good companies at great prices, but we think predicting which countries will perform well is extremely difficult. For a fundamental bottom-up value investor, making macroeconomic predictions is akin to asking a detective out looking for clues on the street to drop his magnifying glass to take a helicopter ride. Nonetheless, we set out to develop a macro risk management process to protect our investors against future avoidable currency risk by taking four steps.
First, we built a robust quantitative risk model which helps objectively warn us of potential trouble. Out of the 32 emerging markets we ranked in our model, only Turkey ranked riskier than Argentina. Furthermore, Argentina was at the bottom of the list in terms of inflation, central bank qualities (independence, transparency, and credibility), and overall sentiment which we measured through recent FX devaluation, expected devaluation (using forward rates), FX volatility, foreign fund flows, credit ratings, and sovereign yield. It also ranked in the bottom 20th percentile in terms of balance of payments concerns which we measure using current account balance, change in reserves, and the overall level of reserves. Additionally, we noticed that Argentina’s ratio of reserves to short-term external debt ratio was at 0.7x, indicating that it lacked the resources necessary to cover its short-term foreign denominated debt obligations. There was only one country with a lower ratio.
Second, we undertook an empirical study of currency crises in 18 emerging market currencies over the past 30 years. For the purpose of the study, we defined a currency crisis as a devaluation of 50% or more in any 52-week period. We were surprised to find that there have been only 12 such instances, four of which were associated with the Asian Financial Crisis of 1997. What was more revealing was that of the remaining eight crises, four were different instances of the Turkish lira and two were the Argentine peso (and two of these six occurred in 2018). In other words, Turkey and Argentina together have accounted for 75% of non-systematic currency crises. If we relax the threshold for devaluation, these two currencies show up many times over. It wasn’t hard to conclude that there are chronic problems in these two economies, and that we would be wise to give more scrutiny to these markets.
Third, we performed case studies on each crisis to identify characteristics which have been associated with past crises, and which could inform our macro risk management process. Abdoulaye Toure, an MBA student at Harvard Business School, helped us research and profile these crises as part of the the course Projects in Investing in which students partner with portfolio managers to execute an impactful project. Among other characteristics, we found that countries with high levels of dependence on foreign investment and low central bank credibility or independence are also more prone to currency corrections. However, one of the most important conclusions from this research was that major devaluations were much more likely to occur around important political transitions.
Finally, we developed a 30-slide country profile in which we systematically consider all macroeconomic trends which could help us identify potential macro risks. In addition to compiling the data for variables such as inflation, unemployment, GDP growth, fiscal deficits, interest rates, debt levels, credit ratings, trade restrictions, capital controls, balance of payments, currency movements, and so forth, we read all Financial Times articles from the past 12 months and all Economist articles from the past 5 years written about the country to search for explanations of the movements in the variables. Our country profile process helped us determine that Argentina has a high level of dependence on foreign investment and low central bank credibility, a red flag identified through our case study project. However, most importantly, Argentina was preparing for the primaries for its presidential elections – a potential political transition – and Argentines have a history of radically changing governments frequently.
We felt that investors were overweighting the probability that the sitting president would be reelected. While Macri has followed the economic crisis playbook and the consensus among foreign investors was that he would be reelected, during our recent on-the-ground trip we couldn’t find a single taxi driver or store clerk who supported the president due to the massive inflation, unemployment, and economic recession experienced during his tenure in office.
As a result, we decided to sell our holdings in Argentina. This decision saved our investors millions of euros of losses in a single day which they would have suffered had we been tracking the index that most of our closest peers follow. The index lost 4% on August 12, 2019 due to the 50% drop in Argentine stocks following President Macri’s defeat in the primaries.
We proceed to expand our study to include other macroeconomic and governance-related data which we believe can further strengthen our evolving macro risk management process.