Northfield’s 27 Annual Research Conference Monday, October 6

Northfield’s 27th Annual Research Conference
Monday, October 6 th – Wednesday, October 8 th 2014
Stowe Mountain Lodge
7412 Mountain Road
Stowe, VT 05672
This year, Northfield’s annual conference returns to Stowe, VT. Northfield takes pride in offering an excellent
agenda; and this year’s event is an extremely strong program, filled with a wide variety of financial topics, which
are at the forefront of the financial industry. Northfield’s 27th Annual Research Conference will be in a setting
that will allow participants to leave their normal workday and focus on the range of presentations while enjoying
the Stowe area during peak foliage season!
Venue
The Stowe Mountain Lodge is located at the base of Mount Mansfield within The Stowe Mountain Resort. The
area is surrounded with beautiful covered bridges, sparkling lakes, rambling woodlands and majestic
mountain peaks.
Calendar
Our event will begin with a welcome reception on the evening of Sunday, October 5th, while the conference
meeting sessions will begin on Monday, October 6th and finish with lunch on Wednesday, October 8th .
CFA Institute Continuing Education Credit Approved
CFA Institute has approved this program, offered by Northfield Information Services, Inc,
for 12 CE credit hours. If you are a CFA Institute member, CE credit for your participation
in this program will be automatically recorded in your CE tracking tool.
Travel Arrangements and Accommodations
We anticipate a large turnout for this year’s conference, given the desirability of the location. Reservations are on
a first come basis so it is a good idea to register early. Please note - we are accepting registrations via online
registration only for the conference and hotel accommodations. If you have any difficulties registering, please
contact [email protected] for assistance.
Hotel accommodations at the reduced conference rate are nearly gone and must be arranged by contacting The
Stowe Mountain Lodge, at 888-478-6938 , 802-760-4755 or visit, http://www.stowemountainlodge.com.
Monday, October 6, 2014
Agenda
9:00 am
Seminar sessions: Junior Ballroom
9:00 am
AIG Before, During and After the Crisis
Bill Poutsiaka, AIG (CIO)
The story of AIG, the iconic global insurance stalwart, is one that has been told and re-told from a variety of
perspectives, focusing mainly on the company’s crisis during the economic downturn that began in earnest in
2008. The story that deserves to be told in tandem about the company, however, is more objective and begins
long before the crisis sets in. It explores the factors and functions – both internal and external – that brought
the AIG to the brink. This presentation will discuss these factors and functions, and will underscore that this is
a story not exclusive to AIG. It is a story that the company shares with others on the Street --- companies that
created the same situations which, when coupled with the overall economic environment, led to the perfect
storm. We will also discuss how AIG has recovered, talk a bit about the controls put in place to ensure the
company has a sustainable future, and hopefully thereby set positive precedents for the industry.
10:00 am Decomposing Variance Risk for Long-Term Investors – On a Few Elementary Formulas
Yan Ge, CPPIB
Risk is often calculated in return space. To gain insight into this notion of risk, we investigate the difference
between return space analysis and value space analysis. With that insight, we derive some elementary but exact
formulas to decompose the variance risk measure into two terms: a short-horizon term, and a long-horizon term
associated with market trend. Percentage return analysis on financial time series involves a simple
mathematical feature that carries a financial meaning, namely, an embedded rebalancing strategy whereby the
investment is rebalanced to a fixed-dollar (FD) at each time step. Consequently, all risk-return analysis
intentionally or unintentionally overlay a strategy on the financial asset in question. The risk-return analysis is
therefore more about the performance or Profit and Loss of the FD-strategy than that of the original asset, or
that of the buy-and-hold (BH) strategy. Similarly, active risk-return analysis is about the active Profit and Loss
between the two FD-strategies, but often mistaken as the active Profit and Loss of the asset versus a benchmark.
Another consequence of the overlay strategy is that return analysis introduces additional model assumptions,
which may not exist in the actual time series data. Therefore, return space differs from value space in (1)
strategy overlay (2) distribution model assumption. While percentage return is consistent with lognormal or
GBM model assumptions, one cannot take it for granted that return analysis would result in the right statistics
for hedge fund strategies and alternative assets.
3:00pm
To Rebalance or Not to Rebalance: A Statistical Comparison of Terminal Wealth of FixedWeight and Buy-and-Hold Portfolio
Eddie Qian, Panagora Asset Management
“To rebalance or not to rebalance?” This seemingly innocuous question is of fundamental importance to many
important investment issues. For example, it is related to the debate about efficient market theory and market
inefficiency, and by extension, the distinction between traditional cap-weighted indices and alternative betas. It
is also the point of divergence between how many investors carry out their asset allocation policies (to
rebalance) and how they adhere to “passive” cap-weighted indices of underlying asset classes (not to
rebalance). Which approach is better, from the perspective of risk-adjusted returns? In this presentation, we
shall address this question statistically by comparing both expected value and expected variance of terminal
wealth of fixed-weighted portfolios and their buy-and-hold counterparts. We also apply the analysis to longonly as well as long-short portfolios. The theoretical results suggest that overall fixed-weight portfolios with
portfolio rebalancing are more likely to have better risk-adjusted terminal wealth than buy-and-hold portfolios.
4:00pm
Did You Choose Well the When, Where and to Whom of Your Birth?
John O’Brien; University of California at Berkeley
There is universal agreement that the U.S. has traditionally been, and should continue to be, a land of equal
opportunity. Yet, when we observe the nation we see vast inequality of income, wealth, status and influence.
However, do these unequal socioeconomic economic outcomes imply that there is any corresponding inequality
of opportunity? Once born, we each have a level of control over how we develop. But, we have no control over
when, where, and to whom we are born. The philosopher John Rawls suggested that the design of a just and
bountiful society should be developed from the perspective that the designers would not know the when, where
and to whom of their birth. Rawls suggested the resulting society would be “fair”, although it might well evince
a level of inequality of income, etc.; but that level of inequality would be “optimal”. According to the Rawlsian
notion of optimality, the least fortunate in a Rawls-optimal society would be better off than they would be in any
more-uniformly economically equal society. I propose that financial professionals should establish research
programs based on principles of prediction markets to study how to measure and improve the current equality
of opportunity. This is important not only from the Rawlsian viewpoint of “fairness” or “social justice”, but
also from the viewpoint of economic development of the nation. The basis for this belief is that the value of the
nation’s human capital is lessened in proportion to the extent of inequality of opportunity among its existing
and future generations of citizens. Unless the potential for “greatness” is merely hereditary, this must be true.
Tuesday, October 7, 2014
9:00 am
Seminar sessions: Junior Ballroom
9:00 am
Correlations, Diversification, and Hedging: A Critical Review of Portfolio Diversification
Measures
Randy O’Toole, Federated Investors
This paper presents a critical review of two popular approaches that aim to quantify diversification properties
expressly related to correlations and provide frameworks for constructing diversified portfolios: the Portfolio
Diversification Index (PDI) developed by Rudin and Morgan [2006] and the Diversification Ratio (DR)
introduced by Choueifaty and Coignard [2008]. These two measures have garnered much interest from
practitioners and academics, and both have been evaluated and critiqued in a number of studies. Importantly,
the PDI and the DR have become contenders in the quest for achieving diversification through risk-based
portfolio construction methods, which most prominently include minimum variance and risk parity strategies.
We show that the PDI and DR are in fact very closely related to both of these risk-based approaches to
portfolio diversification. First, we establish the link between the PDI and risk parity by showing that the PDI
quantifies diversification properties specifically associated with so-called naïve risk parity portfolios, where
portfolio weights are inversely proportional to each asset’s volatility. This has important implications for the
interpretation of the PDI as a summary measure of diversification and for portfolio construction schemes based
on maximizing the PDI.
10:00 am The Art of Tracking Corporate Bond Indices
Marielle deJong, Amundi
The corporate bond indices, built by index providers to serve as investment benchmarks, contain a great
many securities, and are for that reason difficult to replicate. The art is to construct an investible
portfolio that captures the general price trend among the several thousands of securities in the index,
being limited to selecting few of them. This paper describes a practical approach to this, which combines
a well-established portfolio construction technique known as stratified sampling with a modern bond
risk measure named the Duration Times Spread. The key idea is to divide the index members into
samples related to distinct sources of risk that play in the corporate bond markets, and build small
subsamples that capture those risks. As the Duration Times Spread conveys linear- as well as non-linear
bond price behavior, it proves an effective measure in this portfolio building process.
3:00pm
Valuation of Asset Management Firms
Bernd Scherer, EDHEC
Asset management firms attract investor’s interest in times of recovering markets as they are seen as ideal
recovery plays (due to their aggressive stock market beta). Rising markets will increase assets under
management both indirectly (larger inflows as a function of larger wealth and lower risk aversion) as well as
directly (performance related increase in assets under management). Another often quoted reason for the
interest in asset management firms is the expectation by many market observers that the asset management
industry is ripe for consolidation. Cost synergies are obvious but diseconomies of scale are not. How do these
thoughts enter the rational valuation of asset management firms?
4:00pm
TRC Networks and Systemic Risk
Roger M. Stein, MIT
We introduce a new approach for identifying and monitoring systemic risk by combining network analysis and
tail risk contribution (TRC). Network analysis provides great flexibly in representing and exploring linkages
between institutions, but can be overly general in describing the risk exposures of one entity to another.
Systemic TRC provides a more focused view of key systemic risks along with richer financial intuition, but it
may miss important linkages between financial institutions. Integrating these two methods can provide
information on key relationships between institutions that may become relevant during periods of systemic
stress. We demonstrate this approach using exposures of money market funds to major financial institutions
during July 2011. The results for our example suggest that TRC networks can highlight both institutions and
funds that may become distressed during a financial crisis.
Wednesday, October 8, 2014
9:00 am
Seminar sessions: Junior Ballroom
9:00am
What Would Yale Do If It Were Taxable?
Lisa Goldberg, Aperio Group and the University of California at Berkeley
The phenomenal success of Yale's endowment has been an inspiration to many investors. However, if Yale’s
endowment had to pay the same taxes as individual investors, its portfolio would be constructed very differently.
This paper presents a
simple model for incorporating tax considerations into a pretax asset allocation such as Yale's. With illustrative
examples, we demonstrate the profound impact that taxes can have on optimal portfolio weights as well as the
interplay between taxes and risk. Once taxes are included our model tends to lower allocations to tax-inefficient
asset classes such as hedge funds and increase allocations to tax-efficient strategies. However, with optimal tax
management, hedge fund allocation can still be preserved so long as their returns are uncorrelated with those
of equity.
10:00am
Multiperiod Portfolio Selection and Bayesian Dynamic Models
Petter Kolm, New York University
Planning a sequence of trades extending into the future is a very common problem in finance. Merton (1969)
and Samuelson (1969) considered agents seeking decisions which maximize total anticipated utility over time, a
departure from the one-period portfolio selection theories of the time. All trading is costly, and the need for
inter-temporal optimization is more acute when trading costs are considered. The total cost due to market
impact is known to be super-linear as a function of the trade size (Almgren et al. (2005) measured an exponent
of about 0.6 for impact itself, hence 1.6 for total cost), implying that a large order may be more efficiently
executed as a sequence of small orders. Indeed, optimal liquidation paths had already been studied by Almgren
and Chriss (1999) under an idealized linear impact model, leading to quadratic total cost. Using techniques
inspired by Bayesian statistics, we provide an elegant solution to the classic investment problem of optimally
planning a sequence of trades in the presence of transaction costs.
11:00am
Taking the Art out of Smart Beta
Ed Fishwick, Blackrock
The reasons for the outperformance of “smart-beta” portfolios remains somewhat mysterious. This paper
extends previous literature on the link between portfolio performance and macroeconomic factors by exploring
in detail the relationship between low volatility portfolios and interest rate movements. We propose a method
where we use sign changes in interest rates over the past 60 years as a partial determinant of high and low beta
returns. In a CAPM framework we find strongly heterogeneous returns to high and low beta dependent on the
sign of any interest rate movement, indicating a change in behavior around a zero change. We validate the
existence of this zero threshold with a grid search along the likelihood function of our data. We find that the low
volatility “anomaly” is strongly related to interest rate changes over the period in question.
12:00pm
On a Positive Definition of Asset Specific Risk
Dan diBartolomeo, Northfield
Investment models routinely make distinctions between factor and idiosyncratic (asset specific) risk. This
division is enshrined in theories such as the CAPM and the APT. The estimated magnitudes of stock specific
risks are also a key metric of the opportunity set for active equity managers, and are widely used in the scaling
of alpha expectations. In the conventional process of constructing factor risk models, we arrive at an estimate
of idiosyncratic risk for stocks by virtue of a negative rather than positive specification. We take idiosyncratic
risk of an asset to be closely related to the residual portion of the asset’s observed return variance that we
cannot explain by virtue of our specification of factors rather than actually trying to directly estimate what the
true degree of idiosyncratic risk actually is. Such conventional processes have numerous implications that
should be of interest to investors. For example, it implies that different factor specifications of risk models may
arrive at different estimates of asset specific risk even with the same input data. In this presentation, we will
first examine whether variation of estimates of asset specific risk across models is likely to be statistically
significant or economically material. We will then consider a positive definition of specific risk at both the firm
and individual security level based on imposing a no-arbitrage condition on the capital structure of a firm.
Once we have a prescriptive estimate of specific risk, we will conclude with a discussion of how conditioning
the estimates on alternative information sources such as quantification of text news reports can be used to
capture time series variation in the true, but unobservable level of asset specific risk.
1:00pm
Closing Luncheon –Poolside
Our final meal will be a buffet luncheon to encourage everyone to eat together and enjoy a final dose of local
camaraderie. If you do need to catch a plane and have to run, there will be boxes available so you can get your
sandwich to go.