Personalized Asset Allocation
An asset class is a group of investments with similar characteristics and return drivers. We utilize all six high-level liquid, broadly investable asset classes:
U.S. Stocks Equity in companies domiciled in the United States. These assets have high growth expectations and significant volatility.
International Stocks Equity in companies domiciled outside of the United States, includ- ing both developed and emerging markets. These assets are primar- ily recommended for growth, and provide moderate diversification from U.S. stocks. They have high levels of volatility.
U.S. Bonds Debt issued in the United States, primarily by governments and corporations. These assets are recommended for income and strong diversification benefits when coupled with stocks. Inflation- protected bonds add a hedge against inflation.
International Bonds Debt issued outside the United States, primarily by governments and corporations. These assets are recommended for income and diversification benefits.
Alternatives We define alternative assets as hard assets such as real estate (through real estate investment trusts, or REITs) and commodities like gold and energy. Alternative assets are primarily recommended for diversification benefits and as a hedge against inflation. REITs generate income and have high expected total return.
Cash The money you have available in cash – in a high-yield savings account or other savings vehicle. Cash is often overlooked as an asset class.
Each asset class has its own risk and return profile. We consider current interest rates and equity valuations, and their likely impact on future returns, but uses historical risk and return data as an objective starting point for determining an optimal asset class mix. We consider the earliest reliable data available for each asset class, which is 1926 for domestic equities, domestic fixed income, and cash. Data for international equities and alternatives starts in 1970, while international fixed income starts in 2002. Based on that data, we calculate the historical characteristics seen in Figure 1.
Our process for determining the optimal asset class mix is based on a common- sense application of modern portfolio theory (MPT). Developed in the 1950s by Nobel Prize-winning economist Harry Markowitz, MPT attempts to maximize a portfolio’s return for any given level of risk. It does this through a process called mean-variance optimization, or MVO, which finds the optimal combination based on expected return, volatility and covariance.
As seen in the matrix in Figure 2 (below), no two asset classes are perfectly correlated with each other (i.e., correlation = 1.0). Some of the correlations are even negative, meaning those assets tend to move in opposite directions. By combining low or negatively correlated assets it is possible to increase a portfolio’s expected return while simultaneously reducing risk.
The combination of domestic fixed income and alternative asset classes is an example of the kind of risk reduction in which a negatively correlated asset class can act as a counterweight to another asset class. As seen in Figure 1, domestic fixed income has a historical return of 5.3% and a standard deviation (risk) of 5.8%. Alternatives have a return of 6.2% and a standard deviation of 16.3%. As a stand-alone asset class, alternatives are much more volatile, yet when combined with fixed income, they can actually reduce the aggregate portfolio’s standard deviation. A portfolio invested 50% in each asset class would have an expected (historical) return and standard deviation of 5.7% and 7.9%, respectively.
We apply mean-variance optimization to all six asset classes to produce a set of optimal portfolios that maximize return for each level of risk. When plotted on a graph, these portfo- lios represent the efficient frontier. All of Our model portfolios fall on or near the efficient frontier. A portfolio inside the efficient frontier would be suboptimal since it’s possible to achieve a higher return for the same amount of risk.
We combine math and qualitative assessment to categorically dictate asset allocation. While historical results are a good starting point, they can result in data biases, depending on the time period. A “black box” approach favors allocating larger investment amounts to negatively correlated asset classes or those with historically high returns. For example, an investment strategy based solely on data would result in unreasonably heavy weighting in alternatives and emerg- ing markets stock assets. Owning nearly 50% in emerging markets stocks does not pass the “common sense” test and wouldn’t be prudent. Likewise, our investment approach accounts for the current investment environment, which is characterized by low interest rates and cash yields. This means putting constraints on certain asset classes and positioning portfolios to be firmly grounded in reality. History does not always repeat itself.
Methodology Overview For Personalized Asset Allocation All our model portfolios are designed to maximize expected growth for a given level of risk – in other words, they are on or close to the efficient frontier. But that is only half of the equation. Selecting and maintaining the right level of risk for each client is just as important.
Historically, the industry crudely attempted personalized asset allocation by looking at age and risk tolerance alone. Today, most algorithms are driven by a handful of questions, which is an improvement but is insufficient. We take the perspective that asset allocation should be personalized, data-driven, and designed to provide the best chance for each individual to reach his or her goals. To achieve that, our approach combines real-time financial account aggregation, deep investor profile data, a Monte Carlo projection engine and the expertise of financial professionals.
Equal Sector and Style Weighting
An asset class is a group of investments with similar characteristics and return drivers. We utilize all six high-level liquid, broadly investable asset classes:
Increasing Diversification For Better Returns
High-level asset allocation is the most important driver for long-term returns. Construction and security selection within each asset class also make a big difference.
It has largely been proven that picking hot stocks or buying active mutual funds does not provide the highest return. Standard & Poor’s 2014 SPIVA study shows that more than 80% of active large-cap U.S. stock mutual funds lagged the S&P 500. Index investing, in which a bundle of stocks are chosen according to a predetermined set of rules, is a step in the right direction. The most commonly followed indexes, such as the S&P 500, are capitalization weighted. In a capitalization-weighted index, the weight of each stock is equal to the total value of the company divided by the value of all the companies in the index. For example, if Apple is the largest stock in the S&P 500, valued at around $720 billion, and the value of all 500 stocks is around $20 trillion, then Apple represents about 3.6% of the index. For comparison, the weight of the smallest 100 companies in the index is just 4.5% combined.
Academics and institutional investment managers are increasingly recognizing the pitfalls of capitalization weighting. First, it guarantees buying high and selling low. If a stock is overpriced in the market, owners of capitalization-weighted indexes will own more than the “fair value.” Conversely, if a stock is undervalued by the market, it will be owned at a lower amount than “fair value” would suggest. As prices revert to long- term fundamental values, these indexes will suffer.
Another pitfall of capitalization weighting is concentration risk, or the risk of overexposure to a given asset or group of assets. Owning a capitalization-weighted index fund means taking big bets on whatever stocks and sectors happen to be big, for no reason other than they’re already big. This can increase volatility and risk.
The tables in Figure 4 show how S&P 500 owners had their biggest sector bets burst, first in technology before the dot-com bust, then in financials before the sub-prime crisis. In each case, the sector value decreased by 80%.
One of the early alternatives to cap weighting simply weighted each stock equally within an index. The first major equal weighted exchange-traded fund (ETF) to do this was the Guggenheim S&P 500 Equal Weight ETF (RSP). Since its inception in 2003, it has significantly outperformed its cap- weighted parent index, but has also experienced higher volatility.
Approach To Equal Weighting
When it comes to equal weighting, traditional strategies assign the same weight to every stock in a portfolio or index. We go a step further by applying the same concept to a core group of factors. This approach attempts to maintain the performance edge relative to cap weighting while simultaneously reducing overall portfolio volatility.
For one of these factors to be considered, it had to have constituents (stocks) that could easily be classified and didn’t excessively change categories. Economic sector was a logical starting point because performance amongst sectors varies meaningfully year to year, but not much over longer periods of time.
Others have reached similar conclusions on the benefits of equal sector weighting. In a 2010 study by Russell Investments, The Russell Equal Weight Indexes: An Enhancement to Equal Weight Methodology, author Pradeep Velvadapu measured the impact of equal weighting at the sector level from 1978 to 2010. Spanning over 30 years, the study found an investor in the equal-weighted sector approach, based on the Russell 1000 universe, accumulated roughly double the final portfolio value as compared to one who used capitalization weighting.
Some of Russell’s findings include:
- “Sector equal-weighted indexes provided a better absolute return with lower volatility for the time period tested compared to traditional equal-weighted and cap-weighted indexes.”
- “These results are consistent across the domestic large-cap, mid-cap and small-cap spectrum and the global developed and emerging regions.”
In 2013 Andrew Clare, Nick Motson, and Steve Thomas of Cass Business School in London published their study, An Evaluation of Alternative Equity Indices. In it, the authors examined the impact of various weighting schemes for the 1,000 largest U.S. stocks from 1968 to 2011. They found that all of the alternative indexes they created (including equal-weighted) produced better risk-adjusted returns than passive exposure to a market cap-weighted index. One of their primary conclusions was “since the late 1990s the market-capitalization weighted index has proved to be a relatively poor-performing investment strategy.”
We researched several factors relative to the S&P 500 and found:
- More equally weighted economic sectors offered better performance with lower risk
- More equally weighted individual stocks offered better performance but with higher risk
- More equally weighted size offered better performance with similar risk
- More equally weighted style (growth, value, core) offered better performance with lower risk
Using a sampling of individual stocks, our portfolios are designed to ben- efit from all four of these factors. In a hypothetical back-test of the period from 1990 to 2017, the combination was shown to increase returns relative to the S&P 500 by 1.0%, while simultaneously lowering risk. In each case, risk is defined as the standard deviation of annual returns, with portfolios rebalanced annually. Morningstar classifications are used for economic sector and style box.
Our equally weighted portfolios typically contain at least 72 stocks, which sufficiently minimizes stock-specific risk while providing the necessary granularity to achieve desired factor weightings.
Periodic rebalancing is necessary to maintain factor allocations and contain stock-specific risk. It is the act of rebalancing that drives much of the empirical and theoretical outperformance relative to capitaliza- tion-weighted indexes.
The ability to implement this type of portfolio for individual investors was only recently made possible by advances in technology and falling trading costs. Our clients do not pay any trade commissions.
Our goal for equal weighting is to create a portfolio of individual stocks and ETFs which maximizes diversification and spreads risk more effectively than traditional indexes. Stocks are generally selected to be representative of the size, style and sector slot they fill in the portfolio.
Defining The Universe
Our portfolios are broken down into 12 style box categories and 10 economic sectors. This is slightly different than the traditional “nine-box” methodology defined by Morningstar, which consists of three size categories: large, mid and small. The large-cap universe is too diverse to be lumped into one single category. Our approach is to break large cap down further into mega cap and large cap. Mega cap is defined as the biggest stocks, whose sum accounts for 35% of the total value of the market. As of spring 2017, that implies a market capitalization of roughly $95 billion or higher. At the other end of the spectrum, we include a selection of Small Cap index ETFs for a significant portion of our Small Cap equity exposure. Our use of index ETFs in this segment further limits portfolio dispersion by mitigating the higher volatility of individual Small Cap stocks.
Each stock in the portfolio plays a role. The goal is to fill up each of the 12 style boxes in a way that adds up to roughly equal exposure to each of the 10 economic sectors across the portfolio. Within mega cap, the choices are very limited. In large- and mid-cap style boxes, there are significantly more choices. We narrow these down by diversifying at the industry level. It is often possible to remove some options by eliminating niche companies not representative of their industry. For example, we might eliminate Hawaiian Airlines in selecting stocks for exposure to the airline market because it’s more representative of Hawaiian tourism trends than of the airline market overall. Before selecting stocks, we conduct a series of red-flag checks while seeking to maintain diversified exposure to quality, market exposure (beta) and momentum. If there are still multiple options to choose from after finishing red-flag checks, the final selection is randomized.
Our stock selection process
- Does not attempt to “pick winners” – rather, it is designed to find the best representation of the desired factors.
- Performs basic fundamental analysis but does not attempt to predict earnings surprises.
- Follows individual stocks on an ongoing basis to ensure they remain representative of the role they’re intended to play.
This approach is designed to be objective and to minimize potential bias. The overall portfolio is reviewed to ensure there are no unintended themes, such as similar performance trends, balance sheet strengths, betas or dividend yields, and includes a balance of industry leaders and second-tier players. We aim to create a portfolio that captures the intended higher-level factor exposure and otherwise exhibits random results.
How many stocks?
The number of stocks held in a portfolio is an important consideration. Since it isn’t practical to invest in all stocks within the investment universe, a subset of companies is chosen. The goal is to own enough stocks to achieve proper diversification and capture the higher-level factor exposures. This means a portfolio of stocks that eliminates the majority of unsystematic risk (or stock-specific risk) relative to the larger market portfolio. In his 1949 book, The Intelligent Investor, professional investor and Columbia Business School professor Benjamin Graham argued that 10 to 30 stocks were necessary to achieve diversification. This was further refined in the 1987 study by Meir Statman, How Many Stocks Make a Diversified Portfolio? He con- cluded a minimum of 30 stocks was necessary. As can be seen in the table in Figure 9, which is detailed in his 1987 study, holding 30 stocks eliminates roughly 95% of the excess standard deviation over the market portfolio (i.e., “infinity”).
But markets evolve over time, as does the underlying economy. What was true in 1987 is not necessarily true today. A more recent study was published in The Journal of Finance in 2001 by John Campbell, Martin Lettau, Burton Malkiel and Yexiao Xu: Have Individual Stocks Become More Volatile? An Empirical Exploration of Idiosyncratic Risk. The authors point out that the market as a whole has not become more volatile, but in the 35-year period leading up to 1997 the level of volatility for individual stocks increased substantially. This led to lower correlations over time. As a result, it took a portfolio of at least 50 stocks to achieve the same level of diversification previously achieved with 20 stocks. They define this as an excess standard deviation above the market index of 5% or less.
Diversification is one of our highest priorities. Long-term performance should be driven primarily by market exposure (beta) and secondarily by exposure to size, style and sector factors. We typically build portfolios with 90 – 120 stocks in addition to the small cap index ETFs in order to achieve a desired level of expected stock-specific dispersion. As a test, we analyzed a random stock selection over a six year period (2011 – 2016). For each year, we calculated the full-year performance for 100 portfolios consisting of 90 randomly generated stocks from the Mega, Large and Mid Cap stock universes, controlling for size and sector. Small cap exposure was modelled with an index ETF. For each “position” in our model portfolios, a stock was randomly selected from the same size and sector pool. Where none were available (mega-cap utilities, for example), a randomly selected stock from that respective sector was chosen.
After controlling for these factors, the average annual standard deviation among portfolios was 1.8%, which means about 2/3rds of portfolios were within 1.8% of the average. The results are presented in the table in Figure 10. While 1.8% dispersion is relatively modest, it becomes even less impactful when put it into a long-term perspective. Dividing by the square root of “n” years creates an annualized expected stock-specific dispersion number. Assuming a 25-year time horizon (appropriate for most long-term investors), the annualized dispersion drops to: (1.8%) / (√25) = 0.36%.
This number is not a cost. Stock-specific randomness is roughly equally likely to help or hurt and does not meaningfully change the expected return. Since Personal Capital launched in 2011, our actual performance has tracked very closely to the expected result. In the end, we believe that the expected long-term benefits of tactical weighting are easily worth this modest amount of expected stock-specific dispersion. Over time, the factor exposures and the expected benefits of periodic rebalancing are the dominant drivers of relative performance.
In a disciplined rebalancing strategy, applying equal weighting to
- Economic sectors offered better performance with lower risk
- Individual stocks offered better performance but with higher risk
- Size offered better performance with similar risk
- Style (growth, value, core) offered better performance with lower risk
Comprehensive tax optimization can increase after-tax return by up to 1% per year.1 Our sophisticated tax optimization process focuses on three key areas: tax allocation, tax loss harvesting and tax efficiency. For tax allocation and tax loss harvesting, the use of individual securities significantly enhances the ability to add value when compared to use of ETFs or mutual funds alone.
Investors who have both tax-advantaged retirement accounts like individual retirement accounts (IRAs) and traditional taxable investment accounts can improve after- tax return by strategically locating different investments among them. A general rule is to place higher-yield investments in tax-deferred or exempt accounts and low- yield investments in taxable accounts. This is because IRAs are able to shield income from immediate taxation.
Our research has determined that savings can be up to 0.30% per year, depending largely on marginal tax rate. A 2005 study by Daryanani and Cordaro, Asset Location: A Generic Framework for Maximizing After-Tax Wealth, estimated an average of around 0.20% annualized benefit. A Vanguard study showed benefits of “up to 0.75%” per year.
To maximize the benefit, we rank investments based on tax-equivalent yield (TEY), which is automated through our platform. TEY differs from gross yield in that it factors in different tax rates for each investment type. Income from bonds and REITs, for example, is taxed as ordinary income. Dividends from most common stocks, however, are considered “qualified” and are taxed at a more favorable rate. This means if a REIT and stock both exhibit the same percentage gross yield, Personal Capital would likely attempt to place the REIT in the IRA first since its income is less efficient on an after-tax basis.
Tax Loss Harvesting
Tax loss harvesting refers to the intentional selling of securities at a loss to turn an unrealized loss into a realized loss. This may sound counterintuitive, but there are two main ways loss harvesting can save money and improve after-tax return.
- Each individual or family is entitled to deduct up to $3,000 per year in realized losses from ordinary income.
- Losses can be used to offset gains. Loss harvesting allows investors to maintain properly rebalanced portfolios even while deferring gains. By deferring the payment of taxes, more money can appreciate in a compound fashion.
Our loss harvesting approach is tied to each individual’s tax return and only harvests losses when it makes sense.
There are a wide range of claims related to the benefits of loss harvesting. We contend that many are inappropriately overstated. At the upper end, the income deduction for a high-income individual in California with a small portfolio could save 0.78% alone. But more common results lead to savings in the 0.2% to 0.4% range. Even then, if securities are eventually to be sold and proceeds spent, capital gains taxes must eventually be paid. Therefore, aside from the $3,000 deduction, it’s important to realize that the main benefit of loss harvesting is a deferral of capital gains taxes and the opportunity for growth on that deferral. It is not pure profit. Still, tax loss harvesting is a controllable way to boost after-tax return and should not be ignored.
There are thousands of investment vehicles to choose from, and each can have radically different tax implications. Knowing which are tax-efficient is vital to reducing taxes.
Given their greater tax efficiency, Personal Capital portfolios are built on a combination of individual stocks and ETFs. Personal Capital excludes mutual funds.
- Mutual Funds are notoriously bad from a tax perspective. High turnover often creates large annual tax bills. According to Morningstar.com, the 10 largest mutual funds by assets had an average turnover ratio of almost 75%. Most of these are actively managed funds where managers attempt to outperform a benchmark by selling winners to lock in gains. A 2010 study by Lipper (Taxes in the Mutual Funds Industry – 2010; Assessing the Impact of Taxes on Shareholder Return) showed owners of mutual funds in taxable accounts gave up an average of 0.98% to 2.08% in annual return to taxes over the previous 10 years. Profits are also usually distributed to shareholders once per year, so it’s likely that mutual fund owners will have to pay taxes on gains they didn’t individually realize.
- Exchange-traded funds (ETFs) are generally more tax efficient than mutual funds. This is one of the primary reasons they were created. Unlike most mutual funds, ETFs are usually passively managed, which often translates into lower turnover, thus lower tax bills. Certain passively managed mutual funds also fall into this category. But ETFs have another advantage: they trade on the secondary market like stocks and are structured to be easily created and redeemed. In other words, the securities that make up the ETF do not need to be sold to raise cash for redemptions. This largely eliminates the problem of forced distributions and results in greater tax efficiency.
- Individual stocks, when properly managed, are the most tax-efficient way to gain exposure to equities. They leave control over realizing gains entirely in the hands of the investor. Of course, certain stocks pay taxable dividends. But the choice to own dividend-paying stocks is up to the investor – this is not the case with mutual funds or ETFs where investors lack control over underlying securities. Individual stocks can also be tax- located more precisely.
- Bonds, Just like stocks, bond ETFs and passive bond mutual funds are generally more tax- efficient than actively managed bond funds. But the tax treatment of income generated from bonds is different than equities. It is currently taxed as ordinary income, which can be much higher than the rate on qualified stock dividends. There are also exceptions. Municipal bonds are not taxed at the federal level, and if the owner lives in the state in which they’re issued they can avoid state income tax as well.
- Real Estate Investment Trusts (REITs) are companies that invest in physical properties and assets. In general, they tend to focus on specific segments of the market such as retail, healthcare and office properties. To qualify as a REIT, a company must pay out at least 90% of its income (e.g., rental income) in the form of dividends. But unlike stocks, these dividends are generally taxed as ordinary income to shareholders.
A disciplined rebalancing strategy:
- Keeps portfolios on track with long-term goals
- Eliminates costly emotional mistakes
- Enhances risk-adjusted return by creating a systematic way to buy low and sell high
It’s uncomfortable to add money to poor- performing categories, but doing so is proven to add value over time. It works at the asset class level (U.S. stocks, international stocks, US bonds, etc.) and it works within asset classes (individual stocks, certain types of bonds, etc.).
Our software reviews portfolios daily for rebalancing opportunities. Rather than set hard triggers, our approach relies on exception reporting to identify when to evaluate whether a rebalance is beneficial. As a general rule, high-level asset classes will be rebalanced if they deviate more than a few percentage points from target, while specific securities are reviewed if they move more than 0.5% from target. Taxes are strongly considered in the decision. Our goal is to keep turnover under 15% in most years, a threshold that should be sufficient to capture the full power of rebalancing. Depending on market volatility, we may conduct a few or several small rebalances per year.
We performed a hypothetical back-test using historical market returns for the six major liquid asset classes: U.S. stocks, international stocks, US bonds, international bonds, alternatives and cash. We then selected six of the commonly utilized model client asset allocations, ranging from highly aggressive to conservative, and ran a performance analysis from 1980 to 2017. Two sets of returns were calculated for each asset allocation: one with annual rebalancing and one without. As seen in the Figure 11 table, rebalancing added approximately 0.2% to 0.3% to annual returns and reduced volatility.
In any given quarter or year, a multi-asset-class portfolio will perform better than the worst asset class and worse than the best asset class. Yet over time, if properly rebalanced, performance rises toward the top of the group. This is due to the natural tendency of markets to “mean revert.” The end result may not beat whatever happens to be the best asset class in that period, but it’s usually close. Meanwhile, there is less risk along the way.
Figure 12 shows annualized returns for each of the major asset classes, as well as a diversified portfolio similar to one of Personal Capital’s more common investment strategies, which is rebalanced annually. Initially, some asset classes are big winners while others lag badly. The diversified approach sits in the middle. Note that over time, due to the power of rebalancing, the diversified approach slowly floats toward the top. Since 2013, it is ahead of five of the six asset classes and trails only U.S. stocks. Meanwhile, it got there with 20% less volatility. This represents why diversification and rebalancing is such an important part of Personal Capital’s investment process.
Investment methodology is a foundational element of creating strong, efficient portfolios. If you have questions or want more clarification, please contact us.