A rational currency allocation

  • Do market capitalisation driven weights make sense from a currency perspective? If not, how can we go about getting closer to a more balanced and optimal currency mix as part of international asset allocation?

Much has been written about the long-term, zero-sum game nature of a random allocation to currencies (via international equity and bond exposures). We do not wish to revisit this debate here suffice to say that, in general, an allocation to developed market currencies arising as a consequence of an international investment will tend to introduce an extra source of volatility for no discernible return over the long run, certainly from a US Dollar perspective.

What about for emerging market currencies?

We would contend that because of the Balassa-Samuelson effect, countries that are on a growth convergence path to developed market GDP (per capita) levels will see their exchange rates appreciate in real terms over time. Thus, any particular allocation to EM currencies via international debt and equity investing will also add value to, for example, an American or Japanese investor over time. While this may indeed be the case, can we do better? To put it slightly differently, do market capitalisation benchmark driven weights make sense from a currency perspective? We would argue they do not, simply because currencies are fungible and the likely appreciation path of a currency is independent of its relative economic size if there are indeed productivity advances to be experienced.

Consider the table below which ranks the constituent countries of the MSCI EM index by their weight in such an index (first column) and associates these with a “fair value” valuation (second column) versus the US Dollar.

How do we determine valuation or fair value here? We estimate this using a cross-sectional linear regression of 59 countries. The regression estimates the observed relationship between productivity growth rates and the spot exchange rate relative to Purchasing Power Parity (PPP is sourced from the OECD). Using the regression outputs (intercept and coefficient), we generate ‘productivity adjusted’ purchasing power parity exchange rates for MSCI EM countries using respective GDP per capita levels as inputs. The newly adjusted fair values are then evaluated against prevailing spot exchange rates to arrive at a final valuation estimate. This methodology attempts to correct for the main pitfall in Purchasing Power Parity valuations for emerging market currencies, whereby less developed countries exhibit structurally weaker exchange rates. In the table below, valuations are calculated as the overvaluation of the real exchange rate versus the “fair value” as described above and a negative valuation indicates undervaluation.

Now, if we look at the MSCI EM weighted valuation (valuation times country weight in the index) for each of the countries in the table, we arrive at an overall overvaluation of EM currencies to the tune of 1.7% as of September 2016 (fourth 4). What is more revealing is that a simple re-allocation of these weights to equal weights results in an undervaluation of EM currencies to the tune of 8.5%. In other words, passively buying the MSCI EM index results in a much more expensive currency mix than would otherwise be implied by a simple, equally weighed exposure.

The key takeaway here is that while an equally weighted EM currency exposure is by no means definitive or optimal in itself, exposure to a more diversified and growth oriented currency mix is a big step towards achieving that optimality. We refer to this exercise, in general terms, as currency recalibration : it has spurned further and fruitful research into combinations of valuation, carry and momentum (core factors) that are able to deliver enhanced portfolio efficiency relative to the currency mix embedded in market cap driven investing.



Source: Macrobond, OECD, IMF WEO, Record.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *