Category Archives: Risk

Hedge Fund Crowding Costs – Energy

Leave a reply

In recent articles we discussed shared long bets among U.S. hedge funds – a phenomenon called “crowding” – without quantifying its consequences. Crowding is costly to investors, fund managers, and allocators. A prime example of this is the Energy Sector:

  • Over the past 10 years the aggregate hedge fund long energy portfolio (HF Energy Aggregate) delivered negative risk-adjusted returns.
  • Since the energy cycle peaked in mid-2008, HF Energy Aggregate took more risk than the market energy portfolio (Market Energy Aggregate), yet it underperformed.
  • Since mid-2008, HF Energy Aggregate has lost 20% on a risk-adjusted basis.

Hedge Fund Energy Crowding

Our previous article discussed hedge fund (HF) energy herding, We created an aggregate position-weighted portfolio (HF Energy Aggregate) consisting of all long energy equity positions reported by over 400 U.S. hedge funds with medium to low turnover. We then evaluated HF Energy Aggregate’s risk relative to the capitalization-weighted portfolio of energy equities (Market Energy Aggregate) using AlphaBetaWorks’ Statistical Equity Risk Model. The exercise revealed factor (systematic/market) and residual (idiosyncratic/security-specific) crowding.

We mentioned that consensus hedge fund long energy bets tend to disappoint and carry higher risk, but we did not quantify these costs.

Crowded Energy Stocks Underperform

Hedge fund crowding hurts performance. HF Energy Aggregate had poor returns following the peak of the last energy cycle in 2008, even without taking into account its higher risk:

Chart of the historical cumulative returns of Hedge Fund Energy Aggregate and Market Energy Aggregate

Historical Return for Hedge Fund Energy Aggregate vs. Market Energy Aggregate

The spectacular relative performance of HF Energy Aggregate during the commodity boom and the disastrous relative performance in the subsequent commodity crash suggest herding into higher-risk stocks. This is consistent with the aggregate systematic crowding of hedge funds towards higher market beta.

Crowded Energy Stocks Are Riskier

HF Energy Aggregate tends to carry approximately 20% more market exposure than Market Energy Aggregate:

Chart of the exposure of Hedge Fund Energy Aggregate’ to most significant risk factors

Hedge Fund Energy Aggregate’s Exposure to Significant Risk Factors

Crowded Energy Stocks Have Poor Risk-Adjusted Returns

Due to the higher risk of HF Energy Aggregate, its residual return (risk-adjusted performance due to security selection) is even worse. Investors would have made approximately 20% more over the past 10 years holding an ETF portfolio with similar factor risk (factor portfolio):

Chart of the historical cumulative returns, factor returns, and risk-adjusted returns from security selection of Hedge Fund Energy Aggregate

Historical Hedge Fund Energy Aggregate Factor and Residual Returns

Investors would have also made approximately 20% more since the mid-2008 energy cycle peak with this factor portfolio:

Chart of the historical cumulative returns, factor returns, and risk-adjusted returns from security selection of Hedge Fund Energy Aggregate since the 2008 energy cycle peak

Historical Hedge Fund Energy Aggregate Factor and Residual Returns since Cycle Peak

Summary

  • Hedge Fund Energy Aggregate tends to have higher risk than Market Energy Aggregate.
  • Crowded Hedge Fund Energy stocks tend to generate negative risk-adjusted returns.
  • Crowded Hedge Fund Energy stocks tend to outperform in a boom and underperform in a bust: Since mid-2008 energy cycle peak, HF Energy Aggregate underperformed nominally and lost 20% on a risk-adjusted basis.
The information herein is not represented or warranted to be accurate, correct, complete or timely.
Past performance is no guarantee of future results.
Copyright © 2012-2014, AlphaBetaWorks, a division of Alpha Beta Analytics, LLC. All rights reserved.
Content may not be republished without express written consent.

Pure Sector Factors and the Energy Cycle

1 Reply

Separating the Signal from the Noise in the Energy Sector

In an article introducing pure sector factors, we illustrated how market noise obscures relationships among individual sectors, concealing industry-specific performance. As a result, most investors are oblivious of the underlying secular trends and are frequently blindsided by them. This obliviousness can be costly. For example, investors aware of the pure oil and gas producer performance would have been forewarned of the recent energy sector selloff.

Oil and Gas Producer Sector Performance

The selloff in oil and gas producer equities in September and October of 2014 caught some by surprise. It should not have.

A superficial analysis of oil and gas producers’ performance would simply consider historical returns. The sector roughly tracked the broad market, increasing over 200% since 2009 lows:

Chart of the cumulative return of the oil and gas producer sector index

Oil and Gas Producers Sector Index Return

This simple chart conceals powerful secular trends. Obscured by market noise, the underlying energy sector cycle is concealed.

Oil and Gas Producer Pure Sector Factor

We reveal the underlying trends by removing market and macroeconomic effects from security returns and calculating the performance of pure oil and gas producer sector factor. Contrary to the chart above, the sector’s pure performance since 2009 lows was -20%:

Chart of the cumulative historical return of the oil and gas producers pure sector factor

Oil and Gas Producers Pure Sector Factor Return

The pure sector factor tracks the energy cycle, revealing a story of broad economic and geopolitical trends:

  • Overcapacity and crash of the late-90s;
  • Supply shortages, emerging-market commodity boom, and unconventional revolution of 2000-2008;
  • Low-cost production growth, over-investment, and a return to overcapacity post-2008.

Since the energy cycle peaked in 2008, the oil and gas producer sector has lagged the market, on a risk-adjusted basis, by over 35%. Investors in the sector have missed out on over 35% in gains. This deterioration has been concealed by the broad market rally.

Deterioration in sector-specific performance began around the same time as rapid growth in natural gas supply from Haynesville and Marcellus and oil supply from Bakken. The pure return of producer equities, free from market noise, has been negative even as oil prices advanced. Consequently, pure oil and gas producer performance has been a powerful leading indicator.

Oil and Gas Pipelines Pure Sector Factor

The flip-side of the growth in supply has been the shortage of pipeline capacity. The following are the pure returns for oil and gas pipelines, with market effects removed. This boom in pipelines is a direct consequence of the glut of hydrocarbon production evident in the pure producer performance:

Chart of the pure sector factor return for the oil and gas pipelines sector

Oil and Gas Pipelines Pure Sector Factor Return

Recent Pure Oil and Gas Producer Performance

The poor performance of oil and gas producers in October 2014 was a continuation of a long trend. When the broad market experienced hiccups, the underlying trends became evident:

Chart of the daily return of oil and gas producers pure sector factor

Oil and Gas Producers Pure Sector Factor Daily Return

Conclusions

  • By stripping away the effects of market and macroeconomic variables, we can calculate the returns of pure sector factors.
  • Pure oil and gas producer performance has been a powerful leading indicator of the energy sector selloff.
  • The energy cycle peaked in mid-2008 and producers have been in decline since.
  • Despite sector index return of over 200% since 2009 lows, sector investors missed out on 35% in returns since 2008 – pure sector performance was a decline of over 35%.
  • Failure to isolate pure sector factors left many energy investors complacent and ignorant of the underlying trends as the broad market advanced.

 

The information herein is not represented or warranted to be accurate, correct, complete or timely.
Past performance is no guarantee of future results.
Copyright © 2012-2014, AlphaBetaWorks, a division of Alpha Beta Analytics, LLC. All rights reserved.
Content may not be republished without express written consent.

Vanguard Wellington Fund (VWENX) – Market Timing

Leave a reply

Vanguard Wellington Fund (VWENX) is an excellent market timer with a track record of increasing exposures to factors that subsequently generate larger-than-typical returns.

Over the past 10 years, the fund returned approximately 7% solely through varying its factor exposures. This compares to an approximately 4% market-timing loss for the average medium turnover U.S. equity mutual fund:

Chart of the historical return due to market timing of Vanguard Wellington Fund (VWENX)

Vanguard Wellington Fund (VWENX) Market Timing Return History

VWENX’s 3-year marker timing return (βReturn3) is well ahead of the group:

Chart of the return distribution of market timing returns of medium turnover U.S. Mutual Funds

U.S. Mutual Fund Market Timing Return Distribution

Given the excellent market-timing performance, VWENX’s bets are worth watching.

Chart of the historical factor exposures of Vanguard Wellington Fund (VWENX)

Vanguard Wellington Fund (VWENX) Historical Factor Exposures

  • The fund’s exposures to finance sector, health sector, and large-caps are larger than typical.
  • The fund’s exposures to consumer sector, energy sector, and utilities are smaller than typical.

Parnassus Core Equity Fund (PRILX) – Returns-Based Analysis

Leave a reply

Returns-Based Analysis and Fund Beta

In an earlier article, we discussed the failures of returns-based style analysis – a common method of estimating fund risk:

  • Returns-based analysis is usually flawed when portfolio risk varies over time.
  • Returns-based analysis may not even accurately estimate average or representative risk of the portfolio.
  • Flaws are most pronounced for the most active funds – precisely the ones most in need of accurate analysis.

These flaws can be addressed by analyzing historical portfolio holdings, estimating their factor exposures, and aggregating these estimates. This approach requires a robust risk model and is favored by AlphaBetaWorks.

Parnassus Core Equity Fund (PRILX) Returns-Based Beta

Returns-based analysis estimates market exposure (beta) using one or more linear regressions, possibly with several independent variables. We use one fund, the Parnassus Core Equity Fund (PRILX) as an example. For PRILX, a simple single-factor linear regression estimates beta near 0.84:

Chart of the relationship of the returns of U.S. Market and Parnassus Core Equity Fund (PRILX)

U.S. Market and Parnassus Core Equity Fund (PRILX) Returns

This type of regression is the foundation of returns-based style analysis.

Parnassus Core Equity Fund (PRILX) Beta History

A key assumption of returns-based analysis is that beta does not vary over the regression period.

To test this U.S. Market beta estimate, we used the AlphaBetaWorks Statistical Equity Risk Model to estimate monthly U.S. Market exposures of PRILX. The model estimated the exposures of individual holdings. It then combined these into aggregate portfolio beta. Over the past 10 years, the fund has varied U.S. Market exposure between 68% and 95% (0.68 to 0.95 beta):

Chart of the historical U.S. Market Exposure (Beta) of Parnassus Core Equity Fund (PRILX)

Parnassus Core Equity Fund (PRILX) – Historical U.S. Market Exposure (Beta)

PRILX’s average U.S. Market exposure was approximately 79% (0.79 beta) during the period. However, actual beta was rarely near the average:

Chart of the distribution of U.S. Market Exposure (Beta) of Parnassus Core Equity Fund (PRILX)

Parnassus Core Equity Fund (PRILX) – U.S. Market Exposure (Beta) Distribution

Implications of Beta Estimation Errors

For low-volatility funds whose risk profiles vary little over time, the shortcomings of returns-based attribution are relatively minor. For higher volatility strategies, they are severe.

The 5% difference in estimate exposure (0.05 difference in estimated beta) above is not trivial: U.S. Market is up approximately 140% over the past 10 years. A 0.05 difference in estimated average beta translates into a 7% difference in returns attributable to beta over this period.

In addition, returns-based analysis obscured the variation in beta over history: Since 2012 beta has been below 0.75. Consequently, returns-bases analysis’ attribution of performance to market and non-market sources is further off the mark.

Pure Small Cap Performance

Leave a reply

In recent months, the poor performance of small companies has led to a flurry of analysis. Unfortunately, most analysis overlooks small caps’ broad market risk, or beta, and fails to identify pure small cap performance. Consequently, this analysis is imprecise and its conclusions are inaccurate. Much more telling is Size Risk Factor – a pure small cap performance indicator. It reveals that small cap performance has been poor for some time.

Small Cap and Large Cap Index Returns

A common approach to analyzing the relative performance of small caps is comparing a small cap index or ETF to a large cap index or ETF. Performance is compared by plotting price ratios, or relative returns. For instance, the chart below is a comparison of the iShares Russell 2000 ETF (IWM) relative to the SPDR S&P 500 ETF Trust (SPY):

Chart of the cumulative relative return of iShares Russell 2000 ETF (IWM) relative to SPDR S&P 500 ETF Trust (SPY)

iShares Russell 2000 ETF (IWM) Cumulative Relative Return vs. SPDR S&P 500 ETF Trust (SPY)

While this is an accurate portrayal of the relative performance of the two ETFs, it does not capture the pure effect of constituents’ sizes. There are a number of differences between the two funds. Size is one of the most significant. The other is market exposure, or market beta.

Small caps tend to have higher beta than large caps: Recent U.S. Market exposure (beta) of IWM is approximately 1.23. It is approximately 0.95 for SPY. Because of this difference in beta, market is often the dominant factor of relative performance. Hence, analyzing trends in small caps requires more than simply looking at a particular index.

The Size Factor – Pure Small Cap Performance

To isolate the pure small cap performance, we must remove the effect of market and sector risks. AlphaBetaWorks’ Size Factor (ABW Size Factor) strips market and sector effects from security returns, revealing performance purely due to size. The ABW Size Factor is the difference in returns, net of market and sector effects, between the largest and the smallest stocks. The opposite of the ABW Size Factor is the ABW Small-Cap Factor.

ABW Size Factor is closely related to the Fama–French SMB Factor, but with enhancements: SMB Factor captures size risk, but it also picks up market and sector returns. ABW Size Factor strips out market and sector effects for a pure size measurement.

With market effect filtered out, the picture looks different. U.S. small caps have been under-performing since 2010. 2014 has been an especially bad year:

Chart of the cumulative return of U.S. small cap (negative of the Size factor)

U.S. -Size (Small Cap) Factor Cumulative Return History

On a market-risk-adjusted basis, IWM underperformed by approximately 12% over the 3 year period ending 12/31/2013, even before the steep losses of 2014. Investors aware of the pure small cap performance would not have been surprised by the 2014 returns of small caps.

Unlike the performance of small cap indices, buoyed by their high beta and the advancing market,  Size Factor has been flashing warning signs for small cap investors for a few years.

The information herein is not represented or warranted to be accurate, correct, complete or timely.
Past performance is no guarantee of future results.
Copyright © 2012-2014, AlphaBetaWorks, a division of Alpha Beta Analytics, LLC. All rights reserved.
Content may not be republished without express written consent.

iShares Core High Dividend ETF (HDV) Alpha and Beta

Leave a reply

iShares Core High Dividend ETF (HDV) is a low-risk fund that has consistently delivered positive risk-adjusted return.

iShares Core High Dividend ETF (HDV) – Historical Beta

U.S. Market exposure (beta) of the fund has varied in the narrow window between 0.52 and 0.60:

Chart of the history of U.S. Market Beta for iShares Core High Dividend ETF (HDV)

iShares Core High Dividend ETF (HDV) U.S. Market Beta History

iShares Core High Dividend ETF (HDV) – Historical Alpha

The fund has consistently generated positive risk-adjusted return from security selection (αReturn):

Chart of the risk-adjusted returns from security selection of iShares Core High Dividend ETF (HDV)

iShares Core High Dividend ETF (HDV) Security Selection Return History

This αReturn exceeds the results of 95% of U.S. mutual funds with medium and lower portfolio turnover rates:

Chart of the security selection return of iShares Core High Dividend ETF (HDV) compared to the peer group of all U.S. mutual funds with medium turnover

iShares Core High Dividend ETF (HDV) Security Selection Return Vs Peers

 

Currency Risk and Local Currency Beta

Leave a reply

Risk models and portfolio analysis tools often assume that all foreign securities have local currency exposure (beta) of one, under the convenient belief that reference currency performance due to local currency changes is simply local currency appreciation.

Reality is rarely this convenient. This standard approach misrepresents the true currency risk of most foreign investments. For example, an exporter’s margins will expand when its domestic currency depreciates; thus its local currency value will increase as the currency declines. Importers will act in the opposite fashion. Meanwhile, security valuations in emerging markets are vulnerable to external capital flows, amplifying currency risk.

The following figure illustrates the relationship between the JPYUSD exchange rate and the local (JPY) share price for Toyota – an example representative of other Japanese exporters:

The Correlation Between Toyota's Residual Return, in Local Currency, and JPYUSD FX Rate

Toyota’s Local Currency Beta

When Toyota Motor Corporation sells cars outside of Japan, it receives dollars from U.S. sales, Euros for European sales, etc. Meanwhile, some of the costs of these sales are incurred in Japan. As JPY declines relative to USD or EUR, costs decline relative to sales and margins expand. As margins expand and profits increase, Toyota’s shares appreciate. The result is a negative relationship between Toyota’s share price and JPY.

It is tempting to think that a U.S. investor who holds Toyota un-hedged has long JPY exposure. In reality, this investor is short JPY. The local currency beta of Toyota more than offsets the JPYUSD FX risk. $1 Invested in Toyota, un-hedged, creates approximately $0.37 short JPY exposure:

Chart of the Correlation Between Toyota Price, in USD, and JPYUSD FX Rate

Toyota’s USD Price Local Currency Beta

In general, local currency betas vary among and within markets. Simple (and widely used) estimates of currency risk can thus be misleading:

Charts of the Distribution of Local Currency Betas Across Various Markets

Local Currency Beta Distribution Across Markets

Currency risk can be especially complex in emerging markets that are broadly dependent on external capital flows, yet have a number of exporters that benefit from currency declines.

The AlphaBetaWorks World Equity Risk Model encompasses local currency beta and the translation from local to reference currency. This compound process provides an edge to most international investors and was valuable, for example, to U.S. investors looking to be long Japanese equities in 2012 and 2013.

The information herein is not represented or warranted to be accurate, correct, complete or timely.
Past performance is no guarantee of future results.
Copyright © 2012-2014, AlphaBetaWorks, a division of Alpha Beta Analytics, LLC. All rights reserved.
Content may not be republished without express written consent.

The Trend Towards U.S. Hedge Fund Passivity

We charted the historical cross sectional standard deviation of factor (systematic) and α (security selection) returns for U.S. hedge funds:

US Hedge Fund Long Portfolio Performance Dispersion

US Hedge Fund Long Portfolio Performance Dispersion

A surprising finding is that, even in midst of the 2008-2009 market volatility spike, the long positions of U.S. hedge funds were less differentiated from each other that they had been in the relatively lower-volatility regime of 2003-2004.