Using emerging market currencies to exploit the ‘Balassa-Samuelson’ hypothesis is a seductive idea, writes Martin Steward. But it might not work in a new world of ‘inflation tolerance’
It’s been a tough summer for many emerging market currencies. Over the first nine months of 2011, the Indian rupee is down 10%, the Polish zloty, Brazilian real and Mexican peso, around 13% and the South African rand, 22%. And that’s against the euro, which hasn’t exactly been in a happy place since mid-year.
Global growth expectations have taken a severe knock and anything exposed to it is getting sold off. “We’ve been cutting back some emerging market currencies over the summer because we began to worry about severe underperformance by risk assets that need liquidity and global trade to remain strong,” says Paul Brain, head of fixed income at Newton Investment Management. “A lot of the trade surplus currencies will come under pressure in that situation.”
That might feel like a bit of a disaster if you have awarded emerging currency or bond mandates over the past few years - but it shouldn’t. The rationale most often given for these mandates is to make a claim on emerging economic growth. In the May issue of IPE, Bob Arends, head of currency at Henderson Global Investors, wrote that currency is an “efficient” way to capture the “major macroeconomic shift” as emerging economies capture a greater share of global growth and productivity. It should not be a shock that emerging market currencies are a high-beta investment. That was the whole point. The more pertinent question is: what kind of beta are they?
“Ask the average currency manager,” observes Thanos Papasavvas, head of currency management at Investec Asset Management, “and they either say: ‘We like EM currencies, so we select a portfolio of majors that are underweight and EMs that are overweight’ - the beta strategy - or: ‘we like EM currencies, and while we are long and short in these markets we tend to be overweight high-yielders’ - the carry strategy.”
Few would quibble with that formulation. But it contrasts with the way developed-market currency management is characterised. There, it is generally assumed that profiting from spot appreciation is zero-sum alpha; and that if there is any beta to be had, at all, it is through buying interest rate differentials (carry). In emerging currencies, we are asked to accept that there is some other systematic source of beta risk - but its nature remains ambiguous.
For many pension funds the first move into active currency management was spurred by the interpretation of carry as beta, leading to big inflows for carry-based strategies. Perhaps because of the bad experience of 2008, no-one wants to lay too much emphasis on carry as the source of systematic risk from emerging market currencies.
“We would certainly emphasise currency as one of the most interesting components of the emerging wealth theme,” says Matt Roberts, a senior investment consultant at Towers Watson. Like Papasavvas, he also refers to a non-carry, spot appreciation beta related to faster emerging growth rates - and stresses the importance of tilting portfolios towards it.
“Most of the return differential between developed and emerging market currencies over the past 20 years or so has come from real interest rate spread - carry - rather than spot appreciation,” he concedes. “But the reason you might want to avoid simply shorting the low yielders and buying the high yielders in a portfolio that includes emerging markets is that you could end up being short one EM currency and long another, or even long the Australian dollar.”
To get the non-carry beta, Roberts suggests, you probably want a portfolio that is broadly short developed and long emerging currencies.
Currency managers tend to agree. “Carry would be factor in our portfolios but it certainly wouldn’t be the only factor,” says Alan Wilde, head of fixed income and currency at Baring Asset Management, for example. “We have neither a carry bias nor an objection against carry.”
At Investec, carry does feed into the strategy’s fundamental quantitative analysis, but it is first adjusted for trend and volatility, and then corrected according to long-term measures of economic strength which indicate the probability of crisis (and consequent interventions).
“The Brazilian real has the most positive real interest rate differential, but we’ve been underweight,” says Papasavvas. “We were also underweight the South African rand and the Indonesian rupiah. [A concentration in higher-yielding currencies] would raise concerns because we would know that we would be taking on too much carry risk and portfolio construction limits would signal for us to re-think - just as it would if we were concentrated in oil producers, for example. I would like to think that our process is the most carry-neutral out there.”
But Investec is not exactly bullish about the putative non-carry beta, either. Papassavas rationalises his Brazil underweight as a judgement that it is simply “expensive”, but also as a “hedge against the overall ‘beta-ness’ of the portfolio”. His strategy (like many others) does precisely what Roberts at Towers Watson suggests his clients should avoid - playing emerging currencies against one another, as well as against the majors. Alongside its LIBOR-plus Emerging Market Currency Alpha fund, it offers an ELMI+ benchmarked active Emerging Markets Currency fund - no doubt to appeal to the type of client represented by Roberts - but even that only takes on beta tactically.
No other emerging market beta is articulated so confidently and yet treated so sceptically by asset managers. Is there a suspicion that perhaps there is no non-carry beta in emerging currencies - or that it might simply be carry in disguise?
Jerome Booth, head of research at Ashmore Investment Management, sees the whole story through the prism of emerging economies’ central banks. “It will be the decisions of central banks in the emerging world that decide what happens to global FX, not anything the G7 get up to,” he says, “and they have four main motives for allowing currency appreciation.”
Two of those are about buying fewer US dollars, which depresses their currencies: more of an endowment motive for their reserves means greater incentive to diversify; and they are concerned about liquidity conditions in US Treasury markets, given the proportion held by central banks. The other two are to wean their economies off of an export model of growth (where that is the case); and to control inflation.
Clearly, this all depends upon the actions of central banks - but it is hardly a carry trade. Moreover, while some aspects of the story might be coincident with growth differentials between emerging and developed economies, it is not dependent upon them. “That is a very strong argument in the very long term - but you can get that exposure just as well through other emerging market asset classes,” says Booth. “In the meantime we have this much more powerful effect of global rebalancing, the stock of reserves and inflation control. Central banks are massive net creditors working for high-saving economies. Even if there is no shift towards domestic consumption, merely the fact that they change the stock of their existing reserves will have a 30% impact on the US dollar.”
There is another emerging currency beta hypothesis which does link directly with growth differentials, purportedly drawing on a robust economic model: “If you want pure exposure to the Balassa-Samuelson effect you probably want to structure a portfolio to be long EM and short developed,” as Roberts puts it.
Developed independently in the 1960s by Béla Balassa & Paul Samuelson, the Balassa-Samuelson hypothesis seeks to explain the ‘Penn Effect’ - the fact that the real effective exchange rate between domestically-produced and consumed, non-tradable goods in various countries tends not to be 1.0, as it would be under purchasing power parity. A Big Mac (or taxi ride or electricity, or most goods that feed into consumer price indices) costs less in a low-income country than in a richer one.
The hypothesis holds that, because work that is less responsive to improvements, innovations tends to be in domestically-consumed services which cannot be outsourced, productivity growth rates vary more by country in the traded goods sectors (non-perishable, easily-shipped commodities). To maintain some wage level equality between providers of these services and workers in sectors that are more mobile and more responsive to productivity improvements, they have to be paid more than they would be in an economy with fewer people working in the more mobile and productive sectors. This is why your New York taxi driver is likely to be from Asia and your London plumber from Poland.
GDP growth, net of the growing pool of workers, indicates improving productivity, and as more people work in productivity-sensitive sectors, the prices of productivity-insensitive goods and services increases: the cost of the Asian Big Mac closes in on the cost of the European Big Mac. In other words the real effective exchange rate of Asia versus Europe strengthens. This explains the consistent 0.4% appreciation we see for each 1% of GDP growth differential between emerging and developed economies.
The problem with extrapolating Balassa-Samuelson out to an emerging currency beta strategy is that the hypothesis is concerned exclusively with real effective exchange rates. “When you do real growth against nominal appreciation, the relationship is a little less robust,” says Koon Chow, a director in emerging markets research at Barclays Capital.
This is important because getting exposure to the real effective exchange rate is not as simple as getting exposure to the nominal rate.
Think of it this way. Instead of treating yourself to a Big Mac in Antwerp, you save €5 and buy 208 Thai bhat at the prevailing nominal EUR/TBT rate of 41.6. You go to Bangkok and buy three Big Macs at just under 70 bhat each. Over the next 10 years Thailand improves its labour productivity and grows at four-times the rate of Belgium: consistent with the Penn Effect, the Big Mac now sells for 165 bhat in Bangkok and €6.70 in Antwerp. You sell your three Bangkok Big Macs for 495 bhat, buy €11.90 at the nominal EUR/TBT rate of 41.6, return to Antwerp and buy that long-awaited Belgian Big Mac for €6.70. That leaves a profit of €5.20, an annualised return of about 7.5%, or Antwerp’s inflation plus 4.5%. Beer money!
You can see the problem. No one wants to buy 10-year old hamburgers in Bangkok. You can’t store a haircut or taxi ride for 10 years, either. Unfortunately, it’s this stuff with which the Balassa-Samuelson hypothesis is concerned. Non-perishable and tradable goods tend not to be cheaper in lower-income economies precisely because of those characteristics, and this is why I deliberately left the nominal exchange rate the same across the 10 years of the scenario above: to show that the real effective rate can change markedly without the nominal rate budging an inch. It might well move, but that doesn’t necessarily have anything to do with changing relative productivity.
Basically, our theoretical Big Mac profit derives from going short Belgian inflation and long Thai inflation. Without any emerging currencies you are probably already doing that: buy shares in MacDonalds and your Bangkok revenues will rise in line with local hamburger price inflation. The shorter-term equity-price correlation with growth is quite weak, but at least the economics are clear.
Record Currency Management is among those currency managers that cite the Penn Effect and Balassa-Samuelson hypothesis most explicitly. CEO James Wood-Collins argues that returns to his firm’s essentially short-developed, long-emerging strategy come from two systematic beta sources: carry, and nominal effective spot appreciation as a result of “price-level convergence” (Penn/Balassa-Samuelson). However, it turns out to be tricky to draw the distinction between the two.
“Price convergence can occur in a number of ways, the most obvious being either domestic inflation or currency appreciation, or a combination of the two” says Wood-Collins. “But domestic inflation as a result of price convergence does not automatically impact the nominal effective exchange rate: it is possible that the convergence scenario could play out with no return accruing to external investors holding the spot currency at all - other than the return accruing through the interest rate structure embedded in a forward currency contract.”
In other words, if price convergence does not deliver a return through the nominal exhange rate, it will do so via carry. But, of course, that depends upon central banks responding to that inflation, maintaining a real interest-rate differential over your funding currency. But what if you have a central bank that has no credibility and does not care about inflation?
“Pre-financial crisis Russia was focused on supporting growth and keeping the exchange rate stable,” recalls Chow. “It had 8% GDP growth and inflation flirting with double digits but the central bank wasn’t hiking and there was no linker bond. So how did you make money from this high growth and inflation? You didn’t - you lost it. You’d have been buying the bonds, and when they eventually got hit in the financial crisis on the FX you got hit as well.” Here Chow is describing the tail event that threatens all carry trades: spot volatility wiping out accreted yield.
This Russian situation was a real-world example of the conditions described in the Big Mac scenario above. As Chow observes: “The irony is that while the nominal effective exchange rate barely moved, the real effective measure of the rouble increased pre-financial crisis: there was just no obvious way of making money from it.”
But while Wood-Collins argues that carry will capture price-convergence, Record’s strategy is not solely about carry: Record overweights countries it likes according to their real interest-rate differentials, but it also excludes, for example, Argentina - whose relatively high interest rates are about funding the deficits of a flagging economy rather than fighting inflation. Indeed, Wood-Collins actually expects future returns from price convergence to come as much through spot appreciation as through carry. But again, this is only because most central banks have adopted inflation-targeting, leaving nominal exchange rates as their only inflation-pressure release valve. So, carry or spot, as Wood-Collins concedes: “This does require some domestic interest-rate responsiveness to inflation - which is why we screen countries in our emerging universe for monetary policy.”
These mechanisms might explain why the past 20 years or so (the age of hawkish central banks) have been kind to the correlation between nominal effective exchange rates and GDP growth - “less robust” than that between real rates and growth in any case, as Chow reminds us.
“Over the past 100 years, or so, the evidence that currencies are good for exposure to GDP appreciation is not good - but it is for the past 10-15 years,” notes Rob Drijkoningen, head of global emerging markets at ING Investment Management. “Are the last 15 years what you consider most relevant?”
The answer is ‘no’, if we suspect that emerging economies’ central banks might become more dovish to compensate for a generation of sluggish growth in the deleveraging developed world. The carry available from Singapore or Malaysia - even Mexico - is limited. Turkey is getting there. Poland, South Korea, Indonesia, Peru and the Philippines all recently put rates on hold. Brazil is already cutting, and it has previously had one of the world’s most hawkish central banks. “We are seeing a shift away from inflation-targeting, globally,” notes Chow. “This is an environment for CPI-linked bonds: you may not get the appreciation from the FX, but you certainly will through the inflation.”
Booth does not have a problem with inflation-linked bonds but he disagrees profoundly that we might be entering a new era of inflation tolerance that could arrest the appreciation of emerging currencies. Central banks “prudently” cut rates in 2008 despite that being inappropriate for their cycle, he concedes, but they stood ready to allow currency appreciation to take the strain of the inflation they knew would result. Indonesia’s central bank, he points out, has been explicit that refusing to buy US dollars is what enables it to hold down rates.
“Only Turkey and Brazil have loosened and they are both pursuing high-risk strategies,” he argues. “Most have been raising rates, allowing currencies to appreciate, or both.”
Ashmore’s position on emerging currencies depends emphatically on central banks selling US dollars and staying on top of inflation - Booth acknowledges and embraces that. “This has nothing to do with Balassa-Samuelson, and nothing to do with carry trades,” he says. But as we have seen, the Record strategy is equally dependent on central bank decisions, and it is therefore no surprise that Wood-Collins is equally certain that we are not seeing a drift into a new era of long-term dovishness. “If you don’t have domestic interest-rate responsiveness to inflation, you will have sustained negative real rates - which is not feasible or sustainable for the long term in a system of global capital flows,” as he puts it.
Whichever side of that debate you sit on, it is clear that currency portfolio risk is not somehow directly exposed to emerging growth, in any immediate sense. As such, investors need to focus their emerging currency allocation decisions not on their expectations for emerging growth, or even inflation, but on their expectations for central bank responses to those things.