Retirement Security Amid Global Crises: The Problem Of Cutting Risk And Maintaining Yield

by: Kevin Wilson


A question of general interest from a reader called “Mander” seems worth exploring in this time of great uncertainty; the question is, can we maintain retirement income while cutting risk?

An absolutely critical first step is to set aside two years worth of living expenses as a cash reserve so that the retiree is not forced to chase yield.

We can then evaluate personal risk tolerance, and use that to modify a model allocation that matches investment horizon to portfolio duration; equities would be about 45% of assets invested.

To that basic allocation we add layers of equity and bond funds at different weightings to be back-tested for hypothetical sensitivity to drawdowns, risk-adjusted returns, and yields.

Using just two equity CEFs (BXMX, HQH), three equity ETFs (SDY, DVY, PFF), one equity MF (FLPSX), and four bond ETFs (TLT, BLV, IEF, AGG), we produce a reasonable portfolio.

Editor's Note: SA's senior editor for financial advisor content tapped Kevin Wilson, a regular contributor and a professional financial advisor, to answer a question from "Mander," who, in commenting on an article, wanted to know how to secure income while preserving capital in today's markets. Wilson's response follows. Submit your own questions to senior editor Gil Weinreich (

A question of general interest from a reader called "Mander" seems well worth exploring in this time of great uncertainty and potential crisis. "Mander" asks: "…how [does one] avoid market crash exposure while still finding enough income to support retirement? What (outside financials) can one do to source income? Like investment real estate, small business, micro-financing?"

These are broad questions that are best approached first in terms of risk, and then in terms of income. The first issue is to determine the tactical allocation strategy that is reasonable to adopt in retirement when one is expecting a major market downturn. As I have written elsewhere, I think preparing defensively for a market downturn is in order; there is abundant evidence that a significant drawdown could occur in the next few months or quarters. Valuations aren't as extreme as in 2000, but are worse than in 2008 (Chart 1). Measuring from the low of last week, the market itself has been flat since late November of 2014, with high volatility amidst a series of sell-offs since the fall of 2014. This is strongly reminiscent of the extended top that was formed in 2000, which was followed by a 49% drop.

Of course, we can't predict what will happen, or when it will happen, but we can recognize the inherent risk environment that we are in right now. It is fairly likely that the next 15% move is down rather than up, and that if this is wrong, we will then be at the second-most extreme overvaluations in market history (i.e., at even higher risk). There are a number of different ways to proceed now, but since the questions from "Mander" are generic rather than specific, and since my intent is to provide a general approach that can be adapted by individual investors to meet their needs, it may be best to simply generate a tentative allocation model that attempts to meet the twin goals of minimizing losses to the extent possible, plus maximizing retirement income.

Chart 1: Buffet's Valuation Indicator (Market Value/GDP)


In practice, the first thing to look at in constructing an allocation model is the investor's risk tolerance, but we don't know this in a generic case, so we will operate on a different footing. That is, we will generate a tentative long-term or strategic allocation that we can modify tactically as needed (subject to further fine-tuning by each individual investor based on their risk tolerance). This tentative strategic allocation should attempt to match one's investment horizon with one's portfolio duration, which I determined in this example by using John Hussman's method. This involves the simple and reasonable idea that stock duration should be accounted for first, as follows: 1) assume a retirement duration of 20 years; 2) assume a stock duration of 44 years; and 3) 20/44 = 45% equity (maximum). This can be adjusted by investors as needed. Note that at peak valuations the assumed stock duration was more like 50 years.

The second thing to put in a generic model would seem to be an evaluation of the risk of loss, or the expected drawdown for a given allocation. Here we run into a problem, because "Mander" was interested in whether real estate, small business investments, or micro-financing could be used to boost income. These are extremely difficult types of investments to model for risk evaluation, and in any case, I cannot make any recommendations as to private equity holdings. Although there is little doubt that there are privately-held businesses with great IRRs and very good potential dividends, based on my experience, the majority of them are substantially higher in risk, especially if there is a recession on the horizon, which is the base assumption here. Thus, even buying what seems like a simple real estate idea (even many publicly or privately traded REITs) can go badly wrong in a recession and bear market (which again is our base case here). Many other private equity or alternative investments, including some very large, publicly-traded partnerships, generally have huge volatility.

Even a diversified collection of global publicly traded partnerships and small companies in an equity ETF portfolio (Powershares Global Listed Private Equity ETF [PSP]) also has a history of very high volatility (Chart 2). If one has a great deal of experience already in doing these kinds of private equity deals, it might work to carefully evaluate local or regional opportunities, but then it's likely that the question wouldn't even have been asked in that case. I have seen so much money lost on private equity deals that sounded promising to investors, but were eventually shown to be high-risk speculations with poor liquidity; therefore I don't think they're appropriate for most investors. Novices, and especially retired novices, should avoid private equity like the plague.

Chart 2: Volatility of an ETF (NYSEARCA:PSP) with Listed Private Equity Assets

Source: Author; Y-Charts

Alternatively, if we instead try to work with publicly traded conventional companies and funds, we might be able to achieve our generic allocation goals in a first order model with a lower level of idiosyncratic risk. Note however, that current conditions (extreme valuations, falling earnings, falling revenues, falling profit margins, increasing debt loads, etc.) are generating fierce headwinds in the form of negative expected returns (-2.7%) for S&P 500 equities, low returns for other types of equities, and negative returns (-1.6%) for US bonds (Chart 3) over the next 7 years. This is a very tough set of conditions for an allocation model, because it is highly likely that asset classes will not act very much like they did in the last ten years, which is what our back-testing model uses as a database. In particular, it will be difficult for bonds to provide the level of protection in a downturn that they have in the past. Indeed, only the highly volatile emerging markets, and investments in timber, are expected to have positive returns over the long term. Normally, the way such negative returns are resolved though, is for a major drawdown to occur, which is what "Mander" and probably others are concerned about in the first place. Thus, taking note that drawdowns will eventually remove the problem of negative expected returns once they are deep enough, the key will be to put together a portfolio that cuts these hypothetical drawdowns to acceptable levels while generating a decent yield. It is assumed that any tactical aspects of the model can be adjusted to a more growth-oriented stance after a major drawdown occurs.

Chart 3: GMO's Estimate of Expected 7-Year Returns

Source: GMO

One other thing: one of the best ways to protect a portfolio is to set aside enough money for a cash reserve, so that one can fund retirement expenses for a year or two (i.e., the average length of time that a bear market lasts). That way an investor doesn't have to sell anything at an awkward moment during a downturn, and dividend reinvestments allow one to obtain the advantage of dollar cost-averaging. The following assumes that "Mander" and other investors of like mind will be able to fund this cash reserve with current bank accounts or by selling a portion of current holdings. Below is an example for illustrative purposes only:

Amount Available for Investments: $1,000,000.

Annual Income Needed from Investments: $50,000.

Two-Year Cash Reserve (Plus Inflation): ($104,000).

Net Investable Assets for Model: $896,000.

Modeling this amount using Morningstar software, on a purely hypothetical (back-tested) basis, I used the I-Shares 20+ Yr. Treasury Bond ETF (TLT; Chart 4), the I-Shares 7-10 Yr. Treasury Bond ETF (IEF; Chart 5), and the I-Shares Core Aggregate Bond ETF (AGG; Chart 6) for the 55% bond allocation. For the 45% equity allocation, I used two Closed-End Funds (CEFs), i.e., the Nuveen S&P 500 Buy-Write Fund ETF (BXMX; Chart 7), and the Tekla Healthcare Investors Fund (HQH; Chart 8), and two ETFs, i.e., the SPDR S&P High Yield Dividend Aristocrats ETF (SDY; Chart 9) and the I-Shares Select Dividend ETF (DVY; Chart 10).

Note that I have not specified the amounts or percentages allocated to each holding; this is because the model is intended to be an example of how to approach the problem, not an actual solution that can just be plugged in and used. Anyway, this gave me a first order (rough) model that hypothetically would have produced a decent total return over the last ten years, compared to the S&P 500 (assuming all dividends are re-invested; not always the right assumption, but that's the way the model works). The estimated yield (1.86%) is too low perhaps, but let's defer that problem until later and look at the drawdown that hypothetically would have occurred in the worst 12 months (i.e., March 1, 2008 - March 1, 2009) of the Great Financial Crisis. Note that using this drawdown approach does not in any way guarantee that similar results will be obtained in the future, since past performance is not indicative of future returns. Rather, we use this kind of data to stress-test a portfolio for a general evaluation of its risk exposure.

Chart 4: Weekly Price Action for Long Bond (NYSEARCA:TLT) ETF

Source: Author; Y-Charts

Chart 5: Weekly Price Action for Intermediate Bond (NYSEARCA:IEF) ETF

Source: Author; Y-Charts

Chart 6: Weekly Price Action for Aggregate Bond (NYSEARCA:AGG) ETF

Source: Author; Y-Charts

Chart 7: Weekly Price Action for Buy-Write (NYSE:BXMX) CEF

Source: Author; Y-Charts

Chart 8: Weekly Price Action for Healthcare (NYSE:HQH) CEF

Source: Author; Y-Charts

Chart 9: Weekly Price Action for HY Dividend Aristocrat (NYSEARCA:SDY) ETF

Source: Author; Y-Charts

Chart 10: Weekly Price Action for Select Dividend (NYSEARCA:DVY) ETF

Source: Author; Y-Charts

That first model drawdown turns out to be only about 12% in 2008, an outstanding result compared to the 45% drawdown for the S&P 500 over the same time range. The Sharpe Ratio (risk-adjusted return) was 1.08 over the last 10 years, compared to only 0.48 for the S&P 500. This suggests, but does not prove, that the model has partially compensated for the risk of a major drawdown. This first model is massively overweight healthcare and utilities, and somewhat overweight industrials and real estate. It is neutral weight on consumer defensives and basic materials, and fairly underweight communication services, energy, technology, consumer cyclicals, and financials. There are almost no foreign stocks, which doesn't hurt much because foreign stock correlations with the S&P 500 are high in general nowadays, and extremely high (with negative alpha) during bear markets.

So in this first pass at a generic tactical allocation model, we achieved our goal of limited hypothetical paper losses (drawdown, or risk) but we fell short on the portfolio yield. If we assume that the retiree, in this case, has taken my advice and set aside a large cash reserve, this model could work just as it is, because the current income for retirement expenses comes from that cash, and we are able to focus on risk management without chasing yields.

Experience tells me that some investors will object to holding so much cash out of the markets for so long, and will probably think they can do better on the income side. The problem is that there is no way to do that without increasing the risk. An example or two will illustrate this problem fairly well. If instead of a two-year cash reserve this hypothetical investor stays fully invested, they would then likely seek to boost the yield with some fine-tuning of the rough model. The following represents an attempt to tweak the model (under these constraints) to try to make it retain its approximate safety level while experimenting with boosting the yield. This is not so easy, because we have to proceed now by trial and error, and because chasing yield is generally quite risky, and especially so right now due to the Fed's suppression of rates and normal price discovery.

It seems reasonable for the purposes of this example to throw in some discounted, high-yielding individual stocks here that have decent fundamentals, to try to boost the estimated yield. Using a screening tool (I have my own) greatly cuts down the work here. I generated this second iteration of the model by adding 9 stocks from a range of sectors and industries to the first model: readers do not need to know their names because I don't recommend this approach: it's just an illustration of the problem. The 9 stocks all have yields in excess of 4%. Adding this small number of stocks is fine from a diversification point of view because we are already well diversified through the ETFs, CEFs, and a mutual fund. Note that as a final modification we added the Fidelity Low-Priced Stock mutual fund (FLPSX; Chart 11) in order to capture a contribution from midcap stocks.

Chart 11: Total return Graph for Fidelity Low-Priced Stock (MUTF:FLPSX)

Source: Author; Y-Charts

The result for the second iteration model is better, in that the estimated yield jumps up to about 2.70% and the hypothetical drawdown remains pretty good at about 14%. Sector weightings have changed substantially, with a strong overweight now in both utilities and basic materials, and somewhat overweight allocations in healthcare, communication services, and energy. All other sectors are underweight to strongly underweight. This is an unconventional sector allocation and may have its own additional risks. A final tweaking of the allocations to boost yields involves adding two more ETFs, the I-Shares US Preferred Stock ETF (PFF; Chart 12) and the Vanguard Long Term Bond Fund (BLV; Chart 13); we also had to cut back on the relatively low-yielding AGG bond ETF to make room for the new allocations. Some investors might also want to add in something with a better expected return, as shown by GMO for the timber asset class on Chart 3 above; a proxy for this is the I-Shares Global Timber & Forestry ETF (WOOD; Chart 14). Note, however, that since WOOD doesn't have a 10-year history, we can't add it to our model for 10-year drawdowns and statistics. Anyway, this third iteration tactical model has an estimated yield of about 3.10%, much better than the first model, but the hypothetical drawdown rises a quite a bit more, to around 18%. This is still fairly good compared to the S&P 500's drawdown in 2008, but substantially worse than the first model. Hypothetical performance remains decent over the last ten years, again assuming dividends were reinvested, and the Sharpe Ratio drops slightly to about 0.95, still pretty good.

Chart 12: Weekly Price Action for Preferred Stock (NYSEARCA:PFF) ETF

Source: Author; Y-Charts

Chart 13: Weekly Price Action for Long Term Bond (NYSEARCA:BLV) ETF

Source: Author; Y-Charts

Chart 14: Weekly Price Action for Timber (NASDAQ:WOOD) ETF

Source: Author; Y-Charts

As you can see, there is a distinct and non-trivial trade-off for each incremental unit of dividend yield we try to add. Pushing more money into fewer preferred names or funds with very high yields to try and do even better on yield just exacerbates the risk. Even the 3.10% yield we ended up with doesn't match what almost all investors want or need to spend, which is 4%-5% usually. If we try to boost the hypothetical portfolio yield to 4% or higher, which we can, of course, do easily given the recent 20% sell-off for the median stock, the price for that will be even higher model volatility and an even higher level of hypothetical drawdowns and risk. And that assumes that the future will act just like the past, which is extremely unlikely and certainly should not be expected; in other words, chasing yield could potentially be a disaster at some point.

This is the problem of course: expected returns and yields cannot produce the income that people need. There is nothing one can do about this, except either: 1) accept substantially higher risk, or 2) accept much lower returns, spending down assets for a time until things (hopefully) improve. We have Ben Bernanke, Janet Yellen, and many other central bankers to thank for the financial repression retirees are fighting now. My suggestion is that investors hold out a cash reserve and stop chasing yield, no matter what the provocation from central bankers; history tells us that such yield chasing always ends in tears.

Disclosure: I am/we are long TLT, AGG, BXMX, HQH, SDY, PFF.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: This article is intended to provide information to interested parties. As I have no knowledge of individual investor circumstances, goals, and/or portfolio concentration or diversification, readers are expected to complete their own due diligence before purchasing any stocks or other securities mentioned or recommended. This post is illustrative and educational and is not a specific recommendation or an offer of products or services. Past performance is not an indicator of future performance.