Index Investor's  Instablog

PIMCO has published a very interesting research note on “Passive Versus Active Management of TIPS.”  As one of the best active fixed income managers in the world, their views are always interesting.  When it comes to TIPS, their key arguments in favor of active management are that (a) relatively illiquid markets and predictable index fund activity (e.g., “market on close” buy and sell orders, and index rebalancing around auctions) facilitate arbitrage by active managers and reduce the return to passive investors.  We don’t doubt that these costs are real. However, PIMCO’s note fails to put them into any type of context.  So we’ll do that for them.  PIMCO’s actively managed Real Return Bond Fund (PRTNX) has a 3% front end load and a 1.15% annual management fee. Over the past three years, it delivered average annual nominal returns of 5.56%, with a standard deviation of 10.08%.  In comparison, Vanguard’s Inflation Protected Securities Fund (VIPSX) has no front end load and annual expenses of .20%.  Over the same three year period, its average annual return has been 5.20% with a standard deviation of 8.66%.  On balance, while we respect the arguments made by PIMCO, when you put the additional costs they cite in context, it is hard not to conclude that the Vanguard fund is the superior offering.

Disclosure: No positions.

In its August 10, 2009 issue, Barron’s ran an article about “Trouble in the Forest.”  It concluded that timber is overvalued, and “could decline by as much as 50% in coming years.”  We have a very different view.

The underlying diversification logic for investing in timber is quite simple: the key return driver is biological growth, which has essentially no correlation with factors driving returns on other asset classes.  That said, the correlation of timber returns with other asset classes should be different from zero, as it also depends on the price of timber products (which depends, in part, on GDP growth) as well as changes in real interest rates and investor behavior – factors affect returns on other asset classes as well as timber.
 
However, in valuing timber as a global asset class, we face a number of significant challenges.  First, the underlying assets are not uniform – they are divided between softwoods and hardwoods, at different stages of maturity, located in different countries, face different supply conditions (e.g., development, harvesting, and environmental regulations and pest risks), and different demand conditions in end-user markets.  Second, the majority of investment vehicles containing these assets are illiquid limited partnerships, and the few publicly traded timber investment vehicles (e.g., timber REITs) provide insufficient liquidity to serve as the basis for indexed investment products.  Finally, the two indexes that attempt to measure returns from timberland investing (the NCREIF Index in North America, and IPD Index in Europe) are regional in coverage and utilize an appraisal based valuation methodology based on timber limited partnerships, which tends to understate the volatility of returns and their correlation with other asset classes. Given these challenges, the result of any valuation estimate for timber as a global asset class must be regarded as, at best, a rough approximation.

Our valuation approach is based on two timber REITs that are traded in the United States: Plum Creek (PCL) and Rayonier (RYN).  We chose this approach because both of these REITs are liquid, publicly traded vehicles, and both derive most of their revenues from their timberland operations.  This avoids many of the problems created by appraisal based approaches such as the NCREIF and IPD indexes.  That said, tor the reasons noted above, this approach is still far from a perfect solution to the asset class valuation problem presented by timber.
 
As in the case of equities, we compare the returns that a weighted mix of PCL and RYN are expected to supply (defined as their current dividend yield plus the expected growth rate of those dividends) to the equilibrium return investors should rationally demand for holding timber assets (defined as the current yield on real return bonds plus an appropriate risk premium for this asset class).  We note that, since PCL and RYN are listed securities, investors should not demand a liquidity premium for holding them, as they would in the case of an investment in a TIMO Limited Partnership (Timber Management Organization). Two of the variables we use in our valuation analysis are readily available: the dividend yields on the timber REITS and the yield on real return bonds.  The other two variables, the expected rate of growth and the appropriate risk premium, have to be estimated. The former presents a particularly difficult challenge.

In broad terms, the rate of dividend growth results from the interaction of physical, economic, and regulatory processes.  Physically, trees grow, adding a certain amount of mass each year.  The exact rate depends on the mix of trees (e.g., southern pine grows much faster than northern hardwoods), on silviculture techniques employed (e.g., fertilization, thinning, etc.), and weather and other natural factors (e.g., fires, drought, and beetle invasions).  Another aspect of the physical process is that a certain number of trees are harvested each year, and sold to provide revenue to the timber REIT.  A third aspect of the physical process is that trees are exposed to certain risks, such as fire, drought, or disease (e.g., the mountain pine beetle in the northwest United States and Canada).  And fourth physical process is that, through photosynthesis, trees sequester a portion of the carbon dioxide that would otherwise be added to the earth’s atmosphere.

In the economic area, three processes are important. First, as trees grow, they can be harvested to make increasingly valuable products, starting with pulpwood when they are young, and sawtimber when they reach full maturity.  This value-increasing process is known as “in-growth.” The speed and extent to which in-growth occurs depends on the type of tree; in general, this process produces greater value growth for hardwoods (whose physical growth is slower) than it does for pines and other fast-growing softwoods.  At the level of individual timber investments, the rate of in-growth is a key driver of returns; however, at the asset class level, we have decided to assume a constant mix of grades over time.  The second economic process (or, more accurately, processes) is the interaction of supply and demand that determines changes in real prices for different types and grades of timber. As is true in the case of commodities, there is likely to be an asymmetry at work with respect to the impact of these processes, with prices reacting more quickly to more visible changes in demand, while changes in supply side factors (which only happen with a significant time delay) are more likely to generate surprises. In North America., a good example of this may be the eventual supply side and price impact of the mountain pine beetle epidemic that has been spreading through the northwestern forests of the United States and Canada.  The IMF produces a global timber price index that captures the net impact of demand and supply fluctuations, which is further broken down into hardwood and softwood.  The average annual change in real prices (derived by adjusting the IMF series for changes in U.S. inflation) between 1981 and 2007 are shown in the following table:

	Average	Standard Deviation
Hardwoood	0.4%	11.8%
Softwood	1.7%	21.6%
All Timber	0.1%	9.2%

As you can see, over the long term, prices have been quite stable in real terms, though with a high degree of volatility from year to year (and additional volatility across different regional markets).

The third set of economic processes that affects the growth rate of dividends includes changes in a timber REIT’s cost structure, and in its non-timber related revenue streams (e.g., proceeds from selling timber land for real estate development or conservation easements).  For example, if wood prices decline, and non-timber sources of revenue dry up (as is happening during the current recession), a timber REIT (or timber LP) will have to either cut operating costs and/or distributions to investors, increase leverage, or increase the physical volume of trees that are harvested.

Regulatory processes also affect the future growth rate for timber REIT dividends.  In the past, the most important of these included restrictions on harvesting or land development.  In the future, the most important regulatory factor is likely to be the imposition of carbon taxes or a cap and trade systems to limit carbon emissions. These new environmental regulations could provide an additional source of revenue for timber REITs in the future. For example, estimates of the amount of CO2 sequestered each year per acre of growing timberland range from 84 to 172 metric tons.  Current forecasts call for CO2 emissions allowances and offsets to trade at between USD 25 and USD 50 per metric ton, depending on the final shape of a cap and trade system t. At this level of pricing per metric ton of sequestered CO2, the potential new revenues to owners of timberlands would be significant relative to their current revenue (for an early attempt at establishing the CO2 sequestration value of timberland, see “Economic Valuation of Forest Ecosystem Services” by Chiabai, Travisi, Ding, Markandya and Nunes. For a review of similar studies, see “Estimates of Carbon Mitigation Potential from Agricultural and Forestry Activities” by the U.S. Congressional Research Service. Most recently, see “Forging the Climate Consensus: Domestic and International Offsets” by the National Commission on Energy Policy).

The following table summarizes the assumptions we make about these physical and economic variables in our valuation model:

Growth Driver	Assumption
Biological growth of trees	We assume 6% as the long term average for a diversified timberland portfolio. We stress that biological growth rates can vary widely for different types of timber investment (with softwoods and timber located in tropical countries delivering the highest growth, and hardwoods and timber in more temperate climates delivering the slowest growth rates).  We have also changed our valuation model to assume a constant mix of product grades, to present a better approximation for timber as a global asset class.
Harvesting rate	As a long term average, we assume that 5% of tree volume is harvested each year. As a practical matter, this should vary with timber prices and the REITs prevailing dividend level.  So 5% is a “noisy” long-term estimate for timber as a global asset class.
Change in prices of timber products	In line with IMF data, we assume that over the long term, average timber prices will just keep pace with inflation. Again, this is a “noisy” estimate, because the IMF data also shows that real prices are highly volatile. Moreover,  there are indications that climate change is causing increasing tree deaths in some areas, which should lead to future real price increases (see “Western U.S. Forests Suffer Death by Degrees” by E. Pennisi, Science, 23Jan09). Hence we believe our long term price change assumption is conservative.
Carbon credits	Until more comprehensive regulations are enacted, we assume no additional return to timberland owners from the CO2 sequestration service they provide.  Again, given the high level of global concern with limiting the increase in atmospheric CO2 levels, we believe this is a conservative assumption.

This leaves the question of the appropriate return premium that investors should demand to compensate them for bearing the risk of investing in timber as an asset class.  Historically, the difference between returns on the NCRIEF timberland index and those on real return bonds has averaged around six percent.  However, since the timber REITS are much more liquid than the properties included in the NCRIEF index, and since timber has displayed a very low correlation with returns on other asset classes (particularly during the worst of the 2008 crisis, even in the case of liquid timber vehicles), we use three percent as the required return premium for investing in liquid timberland assets. Given these assumptions, our assessment of the valuation of the timber asset class at 31 August 2009 is shown in the following table.  We use the dividend discount model approach to produce our estimate of whether timber is over, under, or fairly valued today.  The specific formula is (Current Dividend Yield x 100) x (1+ Forecast Dividend Growth) divided by (Current Yield on Real Return Bonds + Timber Risk Premium - Forecast Dividend Growth). A value greater than 100% implies overvaluation, and less than 100% implies undervaluation.

Average Dividend Yield (70% PCL + 30% RYN)	5.10%
Plus Long Term Annual Biological Growth	6.00%
Less Percent of Physical Timber Stock Harvested Each Year	(5.00%)
Plus Long Term Real Annual Price Change	0.00%
Plus Other Sources of Annual Value Increase (e.g., Carbon Credits)	0.00%
Equals Average Annual Real Return Supplied	6.10%
Real Bond Yield	1.95%
Plus Risk Premium for Timber	3.00%
Equals Average Annual Real Return Demanded	4.95%
Ratio of Returns Demanded/Returns Supplied Equals Valuation Ratio (less than 100% implies undervaluation)	77%

We stress that this is a long-term valuation estimate that contains a higher degree of uncertainty that valuation estimates for larger and more liquid asset classes.  Over a one year time horizon, you could easily reach a different valuation conclusion. For example, if you believe that real timber prices will decline over the next year, and/or that physical harvesting rates will increase to cover costs and dividends, then you could argue that, in so far as PCL and RYN are roughly accurate proxies for the asset class as a whole, timber is likely overvalued today.  On the other hand, whether looking over a short or long-term time horizon, if you believe that new revenues from timber’s CO2 sequestration service are likely to be significant, and/or that three percent is too high a risk premium to use, then you could argue that timber is actually undervalued today on a medium term view, and possibly on a short-term view, depending on your outlook for cap and trade legislation.  Finally, you could also argue (as Robert Hagler does in “Re-Allocating Timber Investment Portfolios for the Decade Ahead”) that timber remains a relatively inefficient asset class in which it is still possible for active managers to generate significant additional returns.
 
In sum, timber valuation is an issue upon which reasonable people can and do disagree, in no small measure because of their different time horizons and the different underlying assumptions and methodologies they use to reach their conclusions.  On balance, taking a long-term view, we continue to believe that timberland is likely undervalued today, for three reasons: (1) future revenue growth related to CO2 sequestration is likely to be significant; (2) the negative impact on timber prices caused by the recession and long-term slowdown in North American housing construction will be moderated or offset by the impact of supply side changes, such as the mountain pine beetle problem, and by rising demand for wood products that will accompany rising incomes in China.  On a one year view, however, we are neutral, with downward price risk balanced against the upside potential inherent in pending environmental legislation.

Disclosure: No positions

One of the most frequently heard comments about the crash of 2008 is, “I didn’t see it coming.”  This raises a critical question: How can you improve the accuracy of your financial forecasts, or, more broadly, the quality of your foresight?

We believe the answer to this question begins with understanding the nature of the system whose behavior we are trying to predict. At one extreme, physical systems are characterized by relationships defined by the laws of physics and chemistry that are stable over time.  It should therefore be possible to use a single model to forecast the behavior of such a system with a high level of confidence over both short and long time horizons.  Moreover, knowledge of this system’s past behavior can be used to accurately specify the values for the variables used to model its future behavior.

At the other extreme, social systems – like financial markets -- are populated by thinking, feeling, and socially interacting agents who adapt their behavior and goals as events unfold, causing the underlying relationships that drive system behavior to be both complex (e.g., multiple causes for an effect, positive feedback loops and non-linear relationships between causes and effects, and wide time separation between causes and effects) and unstable over time. This system presents forecasters with a far more difficult challenge.  First, because of the system’s complexity, there is an irreducible level of uncertainty associated with the identification of the variables to include in a forecasting model, and the specification of the relationships between them.  Second, once one has developed a forecasting model,  accurately estimating the future values of the included variables and relationships  presents a further challenge – because the system constantly evolves, knowledge of historical values may provide a poor guide to what lies ahead, particularly as the forecast time horizon lengthens.  Third, it is often the case that forecasting models and their users are themselves part of the process that drives the evolution of a complex adaptive system.  For example, a model that accurately forecasts the price of an asset can be discovered by others, whose subsequent use of the model changes the underlying relationships and competes away its ability to generate profitable predictions.

More »

In The Index Investor, May, 2009 journal we reviewed three key fear triggers – loss, uncertainty, and social isolation – that have a powerful impact on investor behavior.  In our June edition, we look at a closely related topic – regret.  Regret is the feeling we experience when we compare the outcome of a previous decision to what would have happened had we chosen another course of action.  It is distinct from disappointment, which is what we feel when confronted with an unexpected negative outcome for which we do not believe our previous decision was responsible.  In terms of neurobiology, regret is produced by the activation of the orbitofrontal cortex, a region of the brain that is associated with cognitive processing (see “The Involvement of the Orbitofrontal Cortex in the Experience of Regret” by Camille, Coricelli, Sallet et al).  However, repeated experiences of regret (and increasing regret aversion) have been shown to activate the amygdala as well, indicating that there is a fear component involved as well as a cognitive one (see “Regret and Its Avoidance: A Neuroimaging Study of Choice Behavior” by Coricelli, Citchley Joffily, et al).

Research has found that the desire to avoid regret has a strong influence on human decision making (see, for example, “Predicting Human Interactive Learning by Regret-Driven Neural Networks” by Marchiori and Warglien). Broadly speaking, the nature of the regret experience seems to depend on two factors: whether it involved an error of commission or omission, and whether it is being viewed from a near term or longer term time perspective.  Errors of commission involve taking actions that later turn out to have worse consequences than an alternative course of action. Errors of omission involve not taking an action that would have produced a better result than the one obtained by not acting.  These are closely related to, and often confused with the Type 1 and Type 2 errors found in statistics.  In the statistical field of hypothesis testing, one usually compares a hypothesis that some action has a statistically significant effect with the so-called “null hypothesis” that it does not.  In a Type 1 error, the null hypothesis (no effect) is rejected when it is true – hence, this type of error is also knwon as a “false positive.”  In a Type 2 error, the test hypothesis is rejected (and the null accepted) when the test hypothesis is actually statistically significant – hence, this error is also known as a “false negative.”  As you can see, the more you try to limit the chance of one type of error, the more you increase the chance of making the other.

Confusion usually arises when errors of commission and omission are used interchangeably with Type 1 and Type 2 errors.  The underlying – and usually unstated – issue is what constitutes the null hypothesis.  Consider a manager who decides to make an investment that later declines in value.  Clearly, this is an error of commission. But is it a Type 1 or a Type 2 error?  It depends. If the null hypothesis was “this is not a good investment” then it is a Type 1 error.  But if the null hypothesis was “this is a good investment” and the test hypothesis “this is a bad investment” is rejected, it is a Type 2 error.  Do you see how this can get confusing?  After struggling for years with how to apply Type 1 and Type 2 error concepts to practical (non-statistical) decision problems, I’ve come to think of the null hypothesis as whatever in the situation in question constitutes the conventional wisdom.  Hence, in my view of the world, a Type 1 error involves accepting a thesis at odds with the conventional wisdom when the latter is correct, while a Type 2 error involves accepting the conventional wisdom when it is actually not correct.  Perhaps more important, this helps to make it clear why people tend to place more emphasis on avoiding errors of commission (Type 1) than they do on avoiding errors of omission (Type 2) – the first involves going against the crowd, while the second requires only that you go along with the crowd.  This nicely aligns with the findings we reviewed in last month’s issue that social isolation is a powerful fear trigger.

More »

In response to the global recession, money supply growth rates are now at record levels in many parts of the world, which has significantly raised the chances of higher inflation in the years ahead.  A number of recent research papers have re-examined the inflation hedging properties of different asset classes, and we will summarize their key findings here.

In “Inflation Hedging for Long-Term Investors”, Attie and Roache of the IMF begin with two important distinctions: first, between the one year and longer term response of nominal asset class returns to an increase in inflation, and second, between an increase in expected inflation and an unexpected increase in inflation.  From our perspective, for a long-term investor, the key issue is the evolution of longer term asset class returns to both expected and unexpected increases in inflation.

The IMF paper focuses on U.S. markets (where data availability is best) and examines the inflation hedging properties of cash (i.e., short term government securities), nominal return government bonds, equities, commodities and gold. They also include two SDR weighted indices of global equity and global government bonds (i.e., these country weights are proportionate to the weights of different currencies in the Special Drawing Rights basket).  Let’s start with the twelve month change in returns on different asset classes (between 1973 and 2008) in response to a one percent increase in the rate of inflation (i.e., the short-term response).  The IMF finds that the two best hedges were commodities (a 9.87% increase in the GSCI index) and gold (a 6.87% increase).  Cash was next best, with a fall of 57 basis points, followed by short term foreign bonds with a fall of 69 basis points.  In contrast, domestic equities fell by (2.59%), global equities by (3.48%), domestic government bonds (all maturities) by (1.33%) and global bonds (all maturities) by (2.36%).

However, for long-term investors, the one year return response to a rise in inflation is less important than the five year response.  As the IMF notes, “inflation shocks persist...After one year, the cumulative increase in price level is nearly three times the size of an initial shock, and after five years this has risen to five times.”  Hence, the long-run return response of different asset classes is critical.  To capture this, the IMF calculates a long-run return multiplier, which essentially measures the extent to which the effects of an inflation shock are offset by a rise in nominal asset class returns.  A multiplier of 1.0 signifies that the inflation shock is completely offset by higher asset class returns; greater than 1.0 signifies more than offset, and less than 1.0 (or negative) signifies a failure to fully offset the effects of inflation.

More »

For our subscribers to The Index Investor, we have regularly reviewed the asset class valuation and return impact of a “wild card” influenza pandemic scenario, and in particular, a step-function increase in the transmissibility of H5N1 – so called “bird flu.”  Given the headlines over the past few days about a new strain of H1N1 influenza that is apparently spreading from Mexico, we have prepared this short background memo for our subscribers.  It covers three issues: (1) Background on influenza, and its potential economic impact; (2) Warning Indicators to monitor; and (3) Our estimate of the possible implications of H1N1 Mexican influenza for asset class valuations and returns over the next twelve months.

Background on Influenza

Influenza viruses are classified first by type (A, B, or C); then by subtype, and then by strain.  Most influenza viruses, including the most recent Mexican “swine flu” and so called “bird flu” (or, more technically, “Highly Pathogenetic Avian Influenza” or HPAI) are type A influenzas.  Viruses are subtyped based two of the eight strands of RNA found on their genome: HA (which affects the production of the glycoprotein hemagluttin) and NA (which affects the production of the glycosylate enzyme neuraminidase).  Hence, HPAI is of the subtype H5N1, and the latest Mexican swine flu is of the H1N1 subtype.  Currently, 15 HA subtypes and 9 NA subtypes have been identified.  These subtypes are further classified according to their so-called “strain”, which is based on the genetic heritage of the different strands of RNA they contain.  In between periodic outbreaks in humans, the world’s population of influenza viruses resides in the intestinal tract of waterfowl, which are usually not affected by them.  In contrast, human influenza viruses have a marked preference for the upper respiratory tract.  Hence, in order for an avian influenza virus to attain the capability to infect humans, its genome must change, so that it develops a preference for attaching itself to the human upper respiratory tract rather than the intestinal tract of aquatic birds.  There are different theories about how these changes happen.  Some treat them as random accidents, produced by the tendency of the influenza virus to replicate itself in great numbers, but with poor fidelity between generations (i.e., to randomly mutate different aspects of its eight RNA strands).  Another theory is that the creation of new virus types is facilitated when a host becomes infected with more than one type of influenza virus.  Pigs are the prime suspect for this mechanism, because their intestinal tracts are similar to waterfowl (in that influenza viruses that bind to the latter can bind to the former), while the upper respiratory tracts are similar to humans’.  Hence the reassortment of influenza RNA in pigs can produce new “swine” viruses with both avian and human characteristics.  Yet another theory posits that the evolution of the influenza virus is driven more purposefully, in that variants with higher fitness (i.e., ability to attach to a host, replicate, and be transmitted) are (through some mechanism) selected as different subtypes recombine (e.g., this seems to account for the rapid spread of antiviral resistance through multiple types of flu viruses around the world in the past two years).  

Three different terms are critical when it comes to assessing the danger posed by an influenza virus.  The first is its transmissibility, or the ease with which it is passed from human to human (abbreviated as H2H), without any common exposure to aquatic or other birds (e.g., chickens have become a reservoir for HPAI) or pigs.  The second is referred to as either the virus’s “virulence” or its “pathogenicity.&... Both of these terms refer to the degree of sickness (and, ultimately, the death rate) produced by a given strain of influenza.  Finally, you may hear the term “tissue tropism” in the same context as virulence or pathogenicity.  This refers to the specific body organs that are affected by an influenza virus.  The typical influenza virus affects the upper respiratory tract.  It kills via a number of mechanisms, including aggravation of preexisting respiratory and cardiopulmonary conditions, and weakening a host so as to allow the development of a secondary bacterial pneumonia infection.  Less often, an influenza virus can directly cause a type of viral pneumonia (which, unlike bacterial pneumonia, cannot be treated with antibiotics).  This was the main way that the 1918 pandemic influenza (which was also of the H1N1 subtype) killed its victims, via rapid lung inflammation and associated hemorrhaging. What has made many medical professionals particularly fearful of H5N1 has been the evidence of its broad tropism, with apparently severe effects on a range of organs, including the brain, liver, and intestinal tract.  Last (but certainly not least), history has shown that in most cases (1918 being an exception) there is an apparent evolutionary tradeoff between transmissibility and virulence – for example, while easily transmissible, seasonal flu is not particularly deadly; in contrast, while quite virulent, H5N1 has thus far shown (in humans) very weak transmissibility.

Let us now turn to the economic impact of influenza. One thing to keep in mind is that our knowledge of these issues is limited by the weakness of the underlying data we have to work with.  For example, records from the 1918 pandemic are quite poor.  More surprising is that even more recent data has significant weaknesses. For example, there is an ongoing controversy about the measurement in the United States of “flu related deaths.”  The narrower definition is based on influenza and pneumonia related deaths, leading to estimates of on the order of 36,000 annual deaths from seasonal flu in the United States. Yet on its website, the Center for Disease Control also offers a higher annual estimate (51,000) that also includes deaths from other causes (e.g., cardio-pulmonary and other respiratory diseases) that are aggravated by influenza. 

One commonly used assumption is that each year in the United States, 15% to 20% of the population is infected with seasonal influenza. Based on a population of 306 million, this amounts to about 61 million infections per year. However, since the strains of seasonal flu in circulation are usually relatively mild, only 1% of infected people (about 610,000) end up being hospitalized. The highest hospitalization rates are typically found among the very young and the very old.  Of those who are hospitalized because of influenza, roughly 8% die (which yields 49,000 deaths, or about 0.08% -- i.e., eight one hundredths of one percent -- of those infected, or 0.016% of the overall population).  As noted above, data on the 1918 pandemic are limited.  However, available estimates suggest that 675,000 people died in the United States, out of a population of about 103 million, for an overall death rate of about 0.66% of the population.  Of those infected, an estimated 2.5% died.  To put that into current terms, out of a 2009 population of 306 million, an exact repetition of the Spanish flu would lead to just over 2 million deaths.

However, many things have changed since 1918, and it is therefore highly unlikely that we would see such an exact repetition.  Specifically, three factors seem likely to reduce the death rate from any pandemic. First, influenza vaccines exist today. To be sure, the 2008 vaccine does not appear to give any immunity to the latest Mexican swine flu.  But vaccine development and production technology is sufficiently advanced that significant dosage volumes could be available about six months after the outbreak of a highly virulent new strain of influenza (there is a caveat here, which is that H5N1 is lethal to chicken eggs, which is a primary production technique for traditional influenza vaccines; however, the latest Mexican H1N1 strain has not been reported to be lethal to eggs).  Second, much more sophisticated modeling methodologies are available to help devise policies (e.g., school closings and travel bans) that can help to limit the spread of a virus until large volumes of vaccine become available (of course, the caveat here is that globalization enables viruses to move around the world much more quickly, as we are seeing with the Mexican case).  Third, modern medicine has more treatments at its disposal than were available in 1918, including antivirals (though rising levels of virus resistance to amantadine and Tamiflu have limited the effectiveness of this line of attack), mechanical ventilators, and antibiotics to control secondary infections.  So it is unlikely (though not impossible) that we would again see the high death rates associated with the 1918 influenza pandemic.

Warning Indicators to Monitor

Thus far, based on available media reports, the Mexican swine flu does not appear to be highly virulent.  The cases outside of Mexico appear to have been mild, with few hospitalizations required and no deaths. However, the data from within Mexico paint a different picture, with more than 143 deaths now reported.  Since we don’t have an estimate of underlying infection rates (which are at best very rough, even under ideal conditions), we can’t reach any conclusions about the meaning of this figure.  Moreover, we have very little information on the cause of death – though the good news here is that there are no reports of unusual tropisms – apparently, deaths are caused by traditional (for flu) respiratory tract complications (and Mexico City’s high level of pollution and pre-existing respiratory conditions would logically elevate its death rate from these).

That said, we are looking for the following warning signs that this outbreak represents a more serious threat than it now appears to be:

1.    Reports that the Mexican swine flu affects other organs – e.g., that it is neurotopic, or that it affects the digestive tract, liver or kidneys. 

2.    Also with respect to virulence, we are looking for any reports of coinfection (e.g., in swine) with Mexican H5N2 poultry influenza, which was associated with heart, pancreas and kidney tropism.  Similarly, we are looking for any reports of Mexican swine H1N1 reaching Indonesia or Egypt, where H5N1 infections in poultry (and possibly other animals) have reached high levels (it is no coincidence that two of the United States premier infectious disease research organizations – Naval Medical Research Units 2 and 3, are, respectively, deployed to Indonesia and Egypt). The analogy we have in mind is 1918, when the initial mild wave of flu infections was soon followed by a subsequent wave of much more serious infections (which could have been caused by reassortment or recombination with more dangerous strains of the influenza virus).

3.    Reports that it is associated with viral pneumonia, and cases of severe inflammation (which produce so-called “cytokine storms”, in which inflammation sets off a positive feedback loop, sending the body’ immune system into overdrive, and filling the lungs with white blood cells and other fluids). This may be associated with an unusually high death rate for 19 – 64 year olds, relative to the death rates for younger and older infected patients

4.    Reports that the virus is characterized by unusually high replication rates in a host.

5.    Rising rates of hospitalizations – above 1 – 2% of infected patients.

6.    Reports of more than 10% of those hospitalized with Mexican swine flu dying from the disease.


Economic and Asset Allocation Implications

In recent years, there have been a large number of estimates of the amount of economic damage that could result from a serious global influenza pandemic (see, for example, “Pandemic Economics: The 1918 Influenza and its Modern Day Implications” by Thomas Garrett, or “A Potential Influenza Pandemic: Possible Macroeconomic Effects and Policy Issues” by the U.S. Congressional Budget Office).  All of them agree that the impact on a normally functioning global economy could be quite serious – e.g., a reduction in global GDP of more than 2.5%.  However, that is already happening, even in the absence of an influenza pandemic.  The real question is whether a pandemic would make things much worse.  Our guess is that while it would worsen the situation somewhat in the short term, it might actually help it in medium term. This view rests on the key assumption that a flu pandemic might move the world back towards our cooperative scenario, and off the track towards increased conflict that we seem to be on today.

In terms of asset class valuations, our previous analysis was that the primary impact of an influenza pandemic would be a sharp rise in uncertainty, and an associated increase in demand for appropriate hedges, such as short term government securities and gold. Differential demand for different currencies could be driven by perceptions that one or more areas were coping significantly better or worse with the flu outbreak.  The reduced economic output associated with a flu pandemic would obviously be bad for equities, as well as commodities, assuming that the fall in demand for them would be much greater than any offsetting fall in supply.  The impact on commercial property would depend on the severity of the influenza outbreak, with the more severe scenarios associated with lower valuations for commercial property, due to reduced demand.  However, as noted with respect to the economic impact of pandemic flu, these negative asset allocation effects have already occurred due to the financial panic of 2008.  So rather than a substantial effect, at this point we estimate that the most likely result of the Mexican swine flu (assuming it doesn’t become much worse) is a damping of the (quite possibly premature) rally in global equity markets, and some further upward pressure on gold and short-term government security prices.

Tom Coyne, Strategist
www.indexinvestor.com