Chief Investment Strategist
During the global financial crisis, communication and coordination amongst global central banks was de rigueur. Now that we are six years into the financial market recovery, the focus is more localized and the esprit de corps has faded. National efforts to combat unemployment and deflation have led to negative interest rates and accusations of currency wars. The financial markets, as usual, are discounting future developments – leading to moves such as the more than 20% appreciation in the U.S. dollar since mid-2014. Just as the markets are debating the onset and pace of U.S. Federal Reserve tightening moves for 2015, the Bank of Japan is rapidly expanding its balance sheet and the ECB is preparing to inaugurate quantitative easing. How will the diverging paths of these major central banks, and the impact those diverging paths have on currencies, affect the financial markets? Does the strengthening of the dollar portend trouble to come? From a strategic standpoint, we think that only fixed income investments should be hedged against currency risks. Tactically, we are positioned for the potential of further dollar strength by over-weighting U.S. equities and under-weighting developed market ex-U.S. equities and emerging market equities and debt.
EXHIBIT 1: THE START OF SOMETHING BIG?
Sources: Northern Trust, Bloomberg; data through 2/27/2015. Dollar index is a weighted avg. of major currency FX rates vs. the dollar. CAGR = compound annual growth rate.
Trying to assess the impact of currency moves on economic growth and corporate earnings is a messy and imprecise affair. The Federal Reserve’s model for the U.S. economy includes estimates of the macroeconomic impact of changes in key variables such as interest rates, stock market wealth and oil prices. The FRB/U.S. model from the late 1990s estimated that a 10% reduction in the value of the dollar would have a 1.6% positive impact on growth after two years, while their updated model (November 2014) cuts the impact to 0.74%. Interestingly, the FRB/U.S. model now estimates a negligible overall impact on growth from a $10/barrel increase in oil prices. Looking at only the impact of the appreciating dollar and falling oil prices, the Fed’s model would estimate a negative impact on growth of 1.0% after two years – based on current price levels. One of the challenges in this forecasting process is that the price changes are not happening in a vacuum, as a rising dollar is most likely the result of the U.S. economy outperforming its peers. To understand the impact of a rising dollar on corporate profits, we studied changes in corporate earnings estimates since 2003 (the earliest point at which the data series in local currencies was available). Over the length of the study, earnings estimates actually fell in all markets after periods in which there was a three-month rise in the dollar of at least 5% (Exhibit 2).
EXHIBIT 2: STRONG DOLLAR HAS LED TO LOWER EPS – EVERYWHERE!
Sources: Northern Trust, MSCI, Bloomberg. Study uses monthly consensus 12-month forward earnings per share (EPS) data in local currency from 6/30/2003 to 2/27/2015. "Normal" measured using median. Dollar growth is over 3 months.
While other factors were undoubtedly at play, the trend of slower growth (Germany and Japan) as well as cuts in earnings estimates (U.S., U.K., France and the broader Eurozone) suggests that the dollar strength frequently was occurring at a time of weakness in their respective economies. The much better relative performance of Japan suggests that the export-related earnings offset the impact of softer domestic economic conditions. After the recent episode of dollar strength, Eurozone next-12-month earnings estimates (in euros) were 1.2% lower than they were three months ago, while Japanese estimates (in yen) were 5.4% higher. This data tells us Japanese companies are benefitting more than European ones from currency depreciation – which is logical given the earlier start of Japanese quantitative easing (QE). However, it also raises the sensitivity of forward European earnings growth to improving economic growth. The benefits from improving exports are clearer than domestic demand growth tied to improved credit creation. As shown in the Appendix, during the two major dollar rallies since 1980, U.S. earnings have actually increased at a 2.9% and 8.0% annual rate – reflecting the relative strength of the U.S. economy during these periods.
The impact of dollar movements on different asset classes is unstable over time, with correlations that transition from positive to negative. Exhibit 3 shows how non-U.S. equity returns have been both negatively and positively correlated to movements in the dollar. There appears to be a regime shift with respect to the correlation of non-U.S. relative returns to the dollar starting in the 1990s – likely a result of increased globalization, wherein revenues of non-U.S. firms expanded outside the borders of their home countries. Commodity prices have also had a non-constant relationship to the dollar. Commodity prices were positively correlated to the dollar in the late 1980s, and spent most of the 1990s and the 2000s (up to the 2008 financial crisis) with little relationship to the dollar. Post-financial crisis, both non-U.S. equity returns and commodity returns have shown a negative correlation to dollar movements.
EXHIBIT 3: UNSTABLE RELATIONSHIPS
Sources: Northern Trust, Bloomberg. Data in dollars through 2/27/2015. Dev. ex-U.S. and emerging market returns are relative to U.S. equities, using MSCI World ex-U.S. MSCI Emerging Markets and MSCI U.S. Indexes.
Over the near to intermediate-term, we could expect these negative correlations to persist, more so for commodities than for equities. With commodities, the upward move in the value of the dollar puts downward pressure on the commodity values to compensate. With equities, the downward move in the local currency should have some beneficial impact on growth (through improved exports), improving the corporate profit outlook and offsetting some of the negative currency hit.
There has been significant issuance of dollar-denominated debt by emerging market companies in recent years (a recent estimate from the Bank of International Settlements cite roughly $2 trillion of outstanding debt and up to $3 trillion in bank loans).1 Continued dollar strength represents a considerable risk to these corporate borrowers, where their currencies have correspondingly depreciated as their debt service burden grows commensurately. Chinese issuers are somewhat less at risk due to the effective targeting of the renminbi to the U.S. dollar (the renminbi has depreciated by just 2.3% over the last year, as of 3/5/2015). At the other end of the spectrum would be issuers from countries such as Russia, where the ruble has depreciated by 60% over the same time frame. We expect this increasing debt burden risk to weigh on emerging market company performance over the next year, affecting the performance of both emerging market stocks and bonds.
While we have reviewed the impact of dollar moves on earnings and certain asset classes, investors struggle with trying to assess a fundamental outlook for currency values. We can understand the valuation of currencies by comparing what goods and services cost in various currencies (purchasing power parity). Comparing the value of an economy’s output in real terms versus the value of output in U.S. dollars gives us insight into whether the dollar is over- or undervalued. In 2013, using the most recent data from the International Monetary Fund (IMF), the dollar was slightly undervalued when appraised through a purchasing power parity lens. The fact that the same goods/services bought outside the United States cost 3.7% more than when bought inside the United States implied that the dollar was undervalued. Looking at more specific data points, but still using a purchasing power parity construct, the dollar was undervalued versus other major currencies prior to the recent dollar appreciation (June 30, 2014) by 16%, 7% and 3% versus the pound, euro and yen respectively. However, after the recent move upward, the dollar is now overvalued by 7%, 13% and 14% respectively. Purchasing power parity is thought to occur in the long-term, but it can remain in disequilibrium for long periods due to trade restrictions and the immobility of some services (a haircut that is 10% cheaper in London is not going to motivate much travel). As shown in Exhibit 4, the dollar remained significantly overvalued in the 1980s and early 2000s for multi-year periods.
EXHIBIT 4: AMERICANS IN PARIS … AND TOKYO AND LONDON
Sources: Northern Trust, IMF, OECD, Bloomberg. Left panel: Data through 2013. Right panel: Current as of 2/27/2015.
A parity construct that tends to sit closer to equilibrium in the near-term is interest rate parity. Interest rate parity theorizes that, if the risk-free rate of interest differs within two currencies, the exchange rate between those two currencies should adjust to make up the difference. For instance, the U.S. three-month Treasury currently provides a 0.05% rate of interest. While extremely low (almost zero), it looks great compared with the -0.27% offered by three-month German bunds. Assuming that both vehicles are indeed risk-free, interest rate parity would predict that the U.S. dollar should depreciate by 0.32% versus the euro over the next three months. Otherwise, an investor could make money by borrowing in euro and putting that money in U.S. dollars. However, this theory does not work well in practice. Financial market frictions take away some of the arbitrage opportunities. Also, there are multiple examples of the carry trade phenomenon, wherein currencies with higher interest rates actually continue to show appreciation as investors pile in. The concluding point is fairly simple: because of real world frictions and investor sentiment, theory does not translate well into practice and currency moves can be quite volatile and unpredictable. What does all this mean for portfolio construction? On a tactical basis (12-month time horizon), we do recommend U.S.-dollar investors reduce risks associated with the rising dollar. We have primarily expressed this through recommended underweights to non-dollar equity assets such as developed ex-U.S. and emerging market equities. Additionally, increased currency risk and rising carrying costs leads us to underweight emerging market bonds. Finally, we have been skeptical over a bounce back in natural resources from the recent weakness – based on the potential for further U.S. dollar strength as well as an expectation of moderate global economic growth.
EXHIBIT 5: LOTS OF REASON TO HEDGE YOUR BONDS
Sources: Northern Trust, Bloomberg; last 10 yrs. of monthly returns through 2/27/2015. Fixed income: Barclays Aggregate Global ex USD Index (hedged & unhedged); Equities: MSCI ACWI ex-U.S. gross total return Index (USD & local).
On a strategic basis, we think fixed income portfolios benefit from currency hedging. The increased volatility of the Barclays Aggregate Global ex-USD Index on an unhedged basis, at 8.1% risk compared with 2.5% on a hedged basis, is unacceptably high for the risk control part of a portfolio. This can be demonstrated by looking at the return per unit of risk, shown on the right side of Exhibit 5, which shows significant deterioration in the unhedged fixed income portfolio. With respect to equity exposures, we believe the long-dated nature of these investments diminishes the necessity of currency hedging. While there was some improvement in efficiency over the study period, its benefit is vastly overshadowed by the benefits of hedging the fixed income portfolio.
Finally, we would like to address a topic that is front-and-center in many year-end portfolio reviews (especially for U.S.-based investors). The significant recent outperformance of U.S. equities (including the 13.7% return in 2014 – using the S&P 500 as proxy – compared with U.S. dollar returns of -4.5% for EAFE and -1.8% for Emerging Markets, using MSCI indexes as proxy) has raised questions about the value of global diversification. From a diversification standpoint, the overall risk of an equity portfolio is reduced through globalization. In local currency terms, both a U.S. and European concentrated portfolio have risk just north of 15%, while a Japanese-only portfolio has risk of 18.6%. All of these investors would benefit from reduced risk by moving to a world equity portfolio, where risk is reduced to 13.9%.
EXHIBIT 6: LESS RISK IN THE WORLD
Sources: Northern Trust, Bloomberg. Study uses MSCI indexes and gross total return data from 12/31/1969 – 2/28/2015.
Moving away from diversification to look directly at relative performance, Exhibit 7 shows the performance of the major developed regions versus the MSCI World Index since 1974 and emerging markets since 1992. To smooth out performance, we show the trailing five-year relative performance of each region. There are periods of relatively long-outperformance at the regional level: with Japan and Europe outperforming the United States during the 1970s and 1980s; U.S. and European equities trouncing Japanese shares in the 1990s; and emerging markets outperforming in the early 1990s and mid-2000s. However, if you bought the best-performing region at the end of any of the eight five-year periods studied, that region repeated as the best-performer only three times. Further, while many are worried about Europe, history shows that Europe outperforms 88% of the time over the next five years after underperforming over the previous five years.
EXHIBIT 7: IN AND OUT OF FAVOR
Sources: Northern Trust, Bloomberg. Data through 2/27/2015. Broad equities = MSCI World Index (*ACWI for EM, available since 1992). Regional equities are proxied by MSCI United States, Japan and Europe Indexes. Returns are in USD.
The divergence in global monetary policy that has led to sharp currency swings over the last eight months looks set to continue as both Japan and Europe are committed to improving economic growth and avoiding deflation. The task for the monetary authorities is hardened by the fact that monetary policy is, by itself, insufficient to improve their economic dynamics. This only increases the likely duration of the easy money policies, leading to further currency risks. While making bold currency forecasts is not for the faint of heart, we do believe that continued dollar strength over the next year is the most likely outcome – and that portfolios should be tilted toward U.S.-dollar assets. We have felt that the divergent growth between the United States and other developed markets was unlikely to persist throughout 2015 – and that economies such as Europe would see some pickup in growth and converge toward the United States. This positive convergence remains our most likely outcome, which could limit the eventual magnitude of the currency moves. While European growth data has improved of late, the U.S. data has been more mixed – with the strong February jobs report being a recent positive outlier. The only real certainty seems to be increased volatility in 2015 as the market weighs the incoming economic data and wrestles with the implications for monetary policy.
APPEDIX: ECONOMIC & FINANCIAL DETAILS OF THE EPISODES
Sources: Northern Trust, Northern Trust Quantitative Research, Bloomberg. AR = annual rate. Current as of 2/27/2015.
1Bank for International Settlements Quarterly Review, 7 December 2014
Chief Investment Strategist
While market volatility has been well below average over the last two years, we have seen a spike in daily moves greater than 1% over the last month (as shown in Exhibit 1). In fact, if one annualized the October 1% market moves, the frequency would approach the levels of 2008 and 2009 (although the frequency of moves of greater than 2% would be lower). This pickup in volatility has led some investors to ask what this means for future market returns, and whether there are reliable signals that can help divine the next correction in the market. We recently reviewed the efficacy of various market indicators, like investor sentiment, volatility and margin debt, and found there is only modest value that can be gained from analyzing investor sentiment indicators. We also examined a more fundamental building block valuation as a timing tool, and found that while valuation has little near-term value in predicting market returns, it is valuable on a long-term basis. For this reason, we look to valuations to help forecast long-term returns. The markets that have been underperforming U.S. equities are generally trading at relatively attractive valuations increasing their long-term return potential. This bolsters the case for a strategic allocation to these markets, and cautions against throwing in the towel on underperforming regions.
EXHIBIT 1: VOLATILITY SPIKES
Sources: Northern Trust, Bloomberg. Big move trading days are market moves of more than 1%. Historical averages are from 1972 to present.
In Exhibit 1, we define big move trading days as days when the market moves by more than 1.0% (up or down). For context, this equates to a 175 point day on the Dow Jones Industrial Average at current price levels. Looking at big move days helps put the concept of standard deviation into a more understandable framework. For instance, MSCI World volatility as measured by standard deviation during the 2008-2009 time frame was 25.6%. But perhaps more relevant was the fact that markets suffered 61 and 46 days of greater-than 1.0% declines in 2008 and 2009, respectively versus the historical average of 19 days a year with losses of that magnitude. Lately, we have enjoyed a below-average number of down days with only 12 in 2013 and only 12 in 2014 (through October 31). However, looking at the monthly numbers, we clearly saw a spike in October both in terms of up days and down days (four each). Is there anything volatility and other sentiment indicators can tell us about the future path of the markets?
In Exhibit 2, we analyze the relationship between volatility and subsequent market performance by looking at the monthly Chicago Board Options Exchange S&P 500 Volatility index level (VIX) against subsequent one- and 12-month S&P 500 returns. Splitting the VIX levels into three buckets shows that no significant difference in future returns exists. The R-squared between the VIX and next 30-day returns is just 0.03, meaning that it only explains approximately 3% of S&P 500 return variability. Meanwhile, the longer-term measure is even worse with an R-squared of just 0.004, meaning the VIX explains approximately 0.4% of S&P 500 return variability. The range of realized returns does, however, increase as volatility rises.
EXHIBIT 2: NO SIGNAL IN VOLATILITY
Sources: Northern Trust, Bloomberg. Study uses monthly data from 1/31/1990 through 10/31/2014.
As we dig deeper into the details, we see further evidence that this is not a very predictive model. For example, when the VIX reaches 26, the expected return is around 11%, but ranges between -40% and 20%. Interestingly, when the VIX rises above 40, market returns have been positive 100% of the time on a 12-month basis. However, the 30-day returns illustrate that you would have to take some short-term pain before realizing this longer-term gain.
Along with the strong rise in the markets over the last five years, margin debt has been accelerating and has become a recent concern of investors both because of the elevated levels it has reached on a nominal basis (currently at $464 billion) and the way in which previous peaks in margin debt have coincided with substantial downturns in the markets. However, looking at the level of margin debt when scaled by the size of the U.S. equity markets (using the S&P 500 market cap as our proxy) paints a less dramatic picture (see right panel of Exhibit 3). By this measure, current levels of margin debt are more or less consistent with the trend line over the past 20-plus years. This secular upward trend is most likely a reflection of continually falling interest rates (and, thus, falling margin debt servicing costs). Other secular factors at play include the proliferation of hedge funds (hedge fund data is included in the margin debt metrics) and financial innovation whereby taking on margin debt today comes with substantially less friction than in the past.
EXHIBIT 3: A WORRISOME RISE IN MARGIN DEBT?
Sources: Northern Trust, Bloomberg. Monthly data through 9/30/2014. S&P 500 used to proxy U.S. market cap.
Turning to the prospect of the recent run-up in margin debt signaling an imminent downturn in the markets (as appeared to happen in 2000 and 2007), we are less convinced. Again, on a relative basis, current margin levels are fairly contained. Furthermore, we cannot be certain of the cause and effect of the recent margin debt peaks. We believe it is more likely that the market sell-offs caused the reduction in margin debt and not the other way around. Because margin debt can be used for betting against as well as betting on the market and can even be used for things that have nothing to do with the market (e.g. an individuals desire to consume) it makes sense to dive into indicators that provide a better read on investor sentiment. We discuss three in particular: investor sentiment surveys, put/call data and fund flows.
As a measure of individual investor sentiment, the American Association of Individual Investors (AAII) asks individual investors whether they are bullish, bearish or neutral on stocks over the next 6 months. Individual investors do tend to be a reasonable contrarian indicator as their moods tend to reflect what has already happened in the markets as opposed to what is about to occur. During periods where investors were excessively bearish or bullish, the surveys have been a fair market timing indicator. When theres excessive bearishness in the surveys, the median market return over the next month is 23.3% (annualized), and over the next 12 months is 16.2%. When theres excessive bullishness, the return over the next 30 days is just 3.8% (annualized), and over the next 12 months is 9.4%. Investing during periods of excessive bearishness had a 58% (one month) and 65% (12 months) probability of outperforming the markets normal-sentiment returns. This strategy has a reasonable hit rate, or probability, of success when looking at the 12-month returns.
When we use a stricter definition of excessive bearishness a two standard deviation move the S&P 500 returned 29.8% over the next 30 days (annualized, with a 66% hit rate) and 23.0% over the next year (a 69% hit rate). This is a rare event, happening only 35 weeks in the history of survey. The last time there was a two standard deviation level of bearishness was in March 2009 and the S&P 500 subsequently gained 65% over the next 12 months.
EXHIBIT 4: SENTIMENT INDICATORS LOOK MORE USEFUL THAN THEY ARE
Sources: Northern Trust, Bloomberg, Morningstar. AAII surveys: weekly since July 1987; CBOE puts and calls: daily since January 1997; Morningstar flows: monthly since February 1993. Three-week smoothing used for survey and puts and calls data.
We next turn to daily put and call volume on the Chicago Board Options Exchange (CBOE), focusing on call volume (bullish bets on the markets) as a percent of total put and call volume. To reduce the impact of professional institutional hedging activity, we excluded index option volume. Our analysis shows that analyzing put and call option volumes is a less useful indicator than individual investor sentiment surveys.
When the CBOE reported excessive call volume, the median market return over the next 12 months was about 7.7%, actually above the 7.2% return when there was excessive bearishness. The distribution was non-normal excessive bullishness only happened 4.6% of the time, while excessive bearishness happened 31.7% of the time. Markets tend to have more call writers than put writers and data dont appear to be mean-reverting as there has been a slight downtrend over the last 15 years, possibly reflecting the increasing presence of long-short equity hedge fund strategies. This study performed better on a 30-day basis as the median return during excessive call-writing was a negative 9.8% (annualized) over the next 30 days, versus 15.9% when there was excessive put-writing. The hit rate of reducing equity exposure during a bullish market was nearly 60%, which isnt bad but would require a long-term systematic program to capitalize on it.
As a final gauge of sentiment, we looked at the flow of money into and out of risk assets (e.g. stocks) and risk-control assets (e.g. bonds). We included U.S.-domiciled open-ended mutual funds, ETFs and money market funds. Fund flows can be a solid contrarian indicator on a 12-month horizon. The median market return when investors are taking risk off the table is 13.6% over the next 12 months better than the 11.4% return when risk is on, but with just a 56% hit rate. On a
30-day basis, the market has outperformed when more money is flowing to risk assets (14.7% versus 10.7%, annualized), a logical occurrence since fund flows can be a technical contributor to short-term momentum.
While sentiment can swing markets in the short-term (one year or less), over a long-term horizon (five years or more) the single best predictor of equity market return variation that we can find is valuation. Cash flow yield is our preferred valuation measure to assess valuation levels. Given its recent popularity, we also assess the cyclically adjusted price-to-earnings ratio (CAPE), which uses a 10-year rolling earnings number for earnings. Looking at the data on a one-year basis shows that valuations provide little insight with cash flow yields and CAPE explaining only 11% and 7% of return variability, respectively. However, on a five-year basis, cash flow yields and CAPE explain 43% and 37% of return variability, respectively. Coincidentally, both valuation measures currently predict a 6.4% annual return for U.S. equities over the next five years, below the 10.1% long-term historical average (data back to 1926). As we include valuations in our Capital Markets Assumptions work, this is fairly close to our forecasted 6.6% return for U.S. equities.
EXHIBIT 5: VALUATIONS MATTER LONG TERM
Sources: Northern Trust, MSCI, Bloomberg, Yale University. CAPE data used in Irrational Exuberance by Robert Shiller. U.S. market proxy: S&P 500.
In Exhibit 5, we convert the CAPE to an earnings yield figure to provide comparability with the cash flow yield data. Many investors are concerned about the heights CAPE has reached, currently standing at 26.3. Since 1881, the CAPE metric has only surpassed this level on three occasions, with peaks occurring in 1929, 1999 and 2007 all preceding major market drops. However, some (including us) question the validity of the CAPE in the current environment given the massive fall in earnings during the financial crisis (the 2008 earnings still suppress the 10-year earnings figure). Furthermore, while CAPE is currently stretched, it has been that way for some time with the metric sitting above 20 since 1995, with a brief exception during the financial market crisis. As Robert Shiller, the co-developer of the CAPE ratio, said in an August 17, 2014 NY Times article The United States stock market looks very expensive right now. The CAPE ratio, a stock-price measure I helped develop is hovering at a worrisome level. However, he went on to say The CAPE was never intended to indicate exactly when to buy and to sell. The market could remain at these valuations for years. But we should recognize that we are in an unusual period, and that its time to ask some serious questions about it. In our opinion, it seems as if Professor Shiller also believes the CAPE is a good long-term predictor of returns, but not a short-term trigger.
Valuations are one major component of our Capital Markets Assumption (five-year) return expectations, alongside earnings expectations and dividend yield assumptions. Looking at developed markets, we expect similar earnings growth across the various regions (approximately 5%). While we do expect Europe and Japan to show slower economic growth than the United States, the composition of company revenues in those regions allows better earnings potential than would be suggested by the companys country of domicile. For instance, companies in slow-growing Europe get nearly 50% of their revenues from outside of the European bloc (including nearly a quarter of their revenues from emerging markets) making those companies the most geographically diversified of all regions (by comparison 70% of U.S. company revenues come from the United States).
1Shiller, Robert J. The Mystery of Lofty Elevations. New York Times 16 Aug. 2014: BU3. Print.
EXHIBIT 6: RETURNS SUPPORTED BY GROWTH AND DIVIDENDS
Sources: Northern Trust.
While our earnings expectations across developed markets are similar, the forecasts for higher dividend yields in Europe, and slight multiple expansion, gives Europe a forecasted return advantage over the United States. Looking at emerging markets, the higher earnings growth potential (with 84% of revenues coming from domestic sources) alongside a solid dividend yield and some valuation expansion lead us to continue to expect a return premium out of those equities relative to the developed markets. The current sentiment toward many major equity markets outside the United States is soured by recent underperformance. However, the improved relative valuation increases the relative return potential and we think supports a better 5-year return expectation. We think this justifies a continued commitment to a globally diversified equity portfolio, which helps reduce dependency on any one region and reduces the risk of materially underperforming a global equity universe.
Special thanks to Raymond Luo, Investment Analyst, for data research.