As an asset manager having worked in almost every asset class from venture writing pitch books for securities offerings; to capital markets trading – equity, bonds and derivatives; to alternative classes such as private equity, hedge funds and real estate, it is easy to observe the interdependence and adverse consequences of government intervention in market pricing.
The cause of most financial assets mis-pricing is due to misperception of the true risk. As a former Director of Global Association of Risk professionals in providing educational seminars to fellow FRMs, I learned from risk managers the commonalities and disparities of risk measurement relating to their respective asset classes. In common, pricing trends to define risk. Professionals chain these data observations to model risk that, in turn, establishes pricing.
To chain a data series of observations to establish risk and pricing is sometimes a bad assumption. When a certain divergent observation does not fit the data series the observation may be termed a “black swan.” However, I propose that there is no such thing in pricing the risk of financial assets. Consider that an outlier observation may be part of its own or another data series and is erroneously chained because of its reference to an underlying security only. Can we say that observations should be chained together merely because they refer to the same security even in the face of different economic criteria, different time periods, different political environments, different geopolitical circumstances? To assume risk is based on chaining return observations would mis-price risk causing mis-priced assets such as what happened to Internet stocks in the late 90’s and real estate leading to the 2008 collapse. The risk of financial and real assets is not captured by probability similar to hitting a bullseye or picking a lotto winning set of numbers. Neither are the mean-variance optimization charts accurate in depicting an optimal mix of assets. The reason is because risk is mis-priced assuming correlations among the asset classes to be consistent. That is a bad assumption. Using correlations to price risk is like tuning a guitar turning one tuning peg to tune all six strings. Correlations also do not capture scale. That is, one asset may have a 100% correlation to another asset but the first asset increases 1bps and the other increases 5%! Also, correlation assumes linearity in that correlation is assumed to be the same on all measurements.
Ray Dalio (Bridgewater) – perhaps the top hedge fund ranked manager in the world – calls correlation a moving target. For hedge funds risk is evaluated more properly in higher moment measurements unlike what is typically shown to you by your adviser in mean-variance presentations. Consider that increasing observations (degrees of freedom) allows for a lower more narrow confidence intervals and a higher significance to an observation. In other words, by chaining observations as though they belong together because they refer to the same referenced asset results in a perceived lower risk and higher confidence of a mean return. This is how asset bubbles are created.
As an example, I just saw today (02Sep16) a CNBC spot on student housing how the asset class (among other residential real estate classes) is becoming popular with institutions according to a student housing firm, considering it has become highly upscale with schools luring out-of-state students under pressure to increase enrollments. The asset class is touted as a “low correlation” investment to financial assets and even to real estate in general including to multi family residential. There may be reasons for this such as, one year leases, parent guarantees and utility of proximity to school. For several of the last years cap rates have decreased as the properties may have appreciated in the face of low interest rates. NOI has been fairly steady for most properties (though there was a student housing REIT that folded).
However, low correlations can turn quickly and a steady NOI can turn negative if interest rates increase. Let’s say interest costs increase on variable rate mortgages, occupancy declines due to competition, maintenance on upscale units and amenities increases and government policy changes to cap costs or encourage local annexes or distance education. This sector of multi family housing would nowhere recognize the changing correlations or the true risks. Consider that the distributions may fall precipitously in an out year coincidentally with higher interest rates with an extension risk that the property may not be flipped for any appreciation anytime soon. Also, consider that some tax benefits could change. How about business risk? Perhaps the sponsor improperly manages the company and begins to dip into vendor payments thereby alienating contractors and causing lawsuits. What if the sponsor sells out the best performing properties to generate cash to regenerate new acquisition fees and you are “dragged-along” forced into another less desirous property for the sake of preserving a tax free distribution? What if the sponsor changes from their set of competencies to grow by becoming vertically integrated and due to a lack of deal flow for their size they move to ground up development from value add, or move to “as is” from value add due to capital constraints?
As you can see, in any asset class risk may be easily mis-priced by looking at observations and statistical analyses more than qualitative factors and sensitivity/scenario analyses. Risk is path dependent and fluid and is affected more by government policy intervening market price discovery.
Pj de Marigny