Stephen P. Barnes, CFP®, CFA, is a principal and portfolio manager for Barnes Investment Advisory Inc., a private investment and financial planning firm headquartered in Phoenix, Arizona.

David Yeske, CFP®, is principal of Yeske & Company in San Francisco, California, and 2003 president of the Financial Planning Association.

Harold Evensky, CFP®, is chairman of Evensky, Brown and Katz in Coral Gables, Florida. He is the author of Wealth Management, published by McGraw/Hill.

Stephen Barnes: Let’s start off with forward rates of return. Give me your thoughts on that, Harold.

Harold Evensky: Our long-term planning has been based on the long-term Ibbotson data—roughly two percent real return in fixed income, seven percent in equities. For the most part, our asset allocation, which is based on a five- or ten-year rise, has used fairly similar numbers. Well over a year ago, we reached the conclusion that the market was fundamentally overvalued—simplistically looking at the price-to-earnings, price-to-book. We concluded that there was going to be a period of adjustment—roughly a decade or so to work itself out. Even after the correction, we still feel that the market is fundamentally overvalued. The recent run-up just makes no economic sense.

We’re using 2.5 percent real return for fixed income. Originally, we increased it to a three percent real rate of return, because of the historic low level of fixed income to that yield, but we’ve adjusted that back down to 2.5 percent.

We cut equities back by one percent, to six percent. The only thing that the recent correction has done has made us feel a little more comfortable with the six percent. We’re still on the optimistic side of what I would consider most credible observers, both practitioners and academics. I think anyone arguing much above that would have a tough road to hoe in trying to demonstrate that.

Ibbotson has completely revised the way in which they’re developing forward-looking expectations, so Roger [Ibbotson] is very much in that range. And Jeremy Siegel, who is rah-rah, stocks for the long run, is talking about maybe seven percent real return. So there’s tantamount to no one out there except the daily talking heads on Wall Street suggesting anything more. We think that there are fundamental changes in the numbers planners should be using.

David Yeske: I’m not quite so convinced that we have a problem. When I hear about the shrinking equity risk premium, I have to ask, which one? I believe there’s more than one priced source of risk in the market. We’ve all heard about the Fama-French three-factor model: size, value, market exposure. And the market exposure part of that model is, to a large degree, what Ibbotson and others are looking at with their numbers. Ibbotson’s numbers are predominantly the S&P 500, so that’s your beta factor. But if you believe that there are other priced sources of risk—if you believe there is a value factor or a size factor—there may not be so big a problem if you’re loading up on those risk factors.

If you’re building global portfolios, I think the aggregate weighted price/earnings ratio for the developed part of Europe is about 14 right now. That’s the number that’s often quoted as the long-run “average” for the U.S. market. Speaking of which, I also have a problem looking at the U.S. market and saying that the long-run average is 14 and we’re at a little over 20, so we’re above that long-run average. The problem with that kind of simplistic approach is that the price-to-earnings ratio does not emerge in a vacuum.

Interest and inflation rates move in the opposite direction from price-to-earnings ratios. So the average price-to-earnings ratio includes periods in which you have double-digit inflation, in which you had 19 percent mortgage rates. In the late ’70s, the price-to-earnings ratio was often eight or nine, because of the presence of extremely high interest rates. Right now we’re in a period of extremely low interest rates, so in an environmental sense, it would be really weird if the price-to-earnings ratio wasn’t higher than the long-run average.

The other thing is that when you go into a recession and into a bear market as we have, the price-to-earnings ratio tends to soar in relation to where you think valuation should be. Earnings often fall faster than the price. So, for a while, until you bottom out and come out, the trailing 12-month’s P/E looks large in relation to where you think it should be. But I wouldn’t bother spending a lot of time arguing for that. If you’re investing globally, there are lots of markets with better valuations than the United States, and within the U.S. and overseas markets, you can load up on these other risk factors. In the end, I have no idea what the risk premium is. It’s like expected return—it’s unobservable. I do know that, all other things being equal, risky assets in an efficient market and a diversified portfolio are going to have a higher return than non-risky assets, and if I do everything I possibly can to minimize expenses, we’ll do as well as we can and it will be what it will be.

Barnes: Are you suggesting we should try to tilt portfolios toward those risk factors that you believe offer the higher returns and away from those that offer the lower returns?

Yeske: If you do a simple Markowitz mean variance optimization, using data for those risk factors, where you include value stocks and small company stocks, you almost invariably end up with an efficient frontier that has various combinations of cash, domestic and foreign small-caps. That’s what I would consider to be an aesthetically displeasing portfolio. It scares me. It’s a demonstration of some of the dangers of mean variance optimization.

But if you put in some kind of a judgment overlay and say, “No, we’re going to have some large company stocks, here and abroad, some bonds, some of those potentially boring things,” and re-plot it against your efficient frontier, you’ll find that the judgment overlay only pulls you a few basis points off your efficient frontier. So it doesn’t cost you much to add that extra layer of comfort. I’m giving my clients what could be called suboptimal portfolios, but I’m willing to pull a few basis points off the efficient frontier in order to have that extra comfort, that extra diversification.

Barnes: What you’re talking about is a portfolio that theoretically has a tremendous amount of alpha due to the overweighting of some of the so-called anomalies—small cap, the value tilt—which would make the portfolio look entirely different from an S&P 500 portfolio. And because that additional alpha is not consistent over time, there will be periods where the client will likely forget the original attractiveness of the portfolio and wonder why their portfolio is not keeping up with their golf partner’s portfolio and therefore will want to change the strategy.

Yeske: Right. The only portfolio that will survive the theoretical long run we’re all talking about is the one that the client can live with in the short run. We optimize risk and return using annual returns, but the problem is that clients look at this stuff more often than annually, and the fact that it was only down 5 percent over the course of a year doesn’t help if it was down 15 percent in February. They’ll freak.

Evensky: Our conclusion is, when we felt returns were going to be lower, we simply estimated what the net net return was going to be—net meaning net after expenses, taxes and inflation. We concluded that the net net real return was going to be astronomically lower than anything we had in mind. As a consequence, expenses and taxes were going to mushroom in importance. Before, we paid attention to expenses and taxes, but they were secondary in our consideration. Now they are major. As a result, all the costs of trying to capture these marginal extra returns, that alpha, get subsumed by the cost. And that’s the reason that we think planners need to be making major changes in portfolio design.

If we’re wrong and indeed, instead of the 2.5 percent real returns it’s an 8 percent real return, by our inefficient management, by not slicing and dicing and using traditional multi-asset class, multi-style, we lose a half point, our clients are going to get net net real return 9 percent instead of 9.5. Oh, well. If we’re right, and someone doesn’t do it and they lose 50 basis points, they’ve just cut out about 20 percent of their effective return. The risk of not making a change, if you remotely think these numbers are right, is going to be critical for your clients. The risk of making the change if you’re wrong is going to be fairly modest.

Yeske: We’re actually not that far apart in terms of a lot of the assumptions we use. I never used an assumption of a 14 percent gross return, so I don’t have to make a big adjustment The other problem is, when you’re talking about risk premiums, you can’t get quite as precise as double-entry bookkeeping. It’s a random variable. The only part of this that’s controllable is the expense number. You can make a reasonable projection of the expense, but that’s an expense coming off of a wildly varying random variable, so I tend not to get too caught up in adding it up and making precise calculations of what I think the net’s going to be.

You can’t run retirement projections without a number, but I also would suggest that retirement projections are pretty problematic anyway—anytime you project anything much beyond tomorrow, you’re into some interesting territory. We try to control that by using Monte Carlo. I look at the full spread of future outcomes, allowing returns to vary widely. I have a fairly dispersed distribution from which the Monte Carlo simulation pulls portfolio returns, allowing inflation rates to vary. But as far as I’m concerned, this stuff is only minimally controllable by us. All of this planning we do, all of these long-run projections—in reality, all we’re really doing is trying to answer the question, “What do we do next?” What we do next is the only thing we really control.

Evensky: I would argue what we control is taxes and expenses. In terms of implementation, I thought that we could be fairly loose in controlling them, but now I think we need to focus significantly on the control of that. And the designs we’ve been using, what I wrote about in my book, while intellectually are perfect for institutions, no longer make sense for the retail investor because of the taxes.

Yeske: I couldn’t agree more. I’ve always focused on expenses. As for the mean variance optimization, I think that’s a useful exercise in that it captures the covariance matrix. You want to bring together assets that have some reasonable expectation of a positive return and where you have as low a covariance as you can possibly find in order to minimize variance, because we want to maximize the compound return.

Barnes: I like the way Harold framed the portfolio construction or the forward returns assumption by asking, “What if we’re wrong?” We use an optimizer, not to find the theoretical efficient frontier, but to do some sensitivity analysis to understand what happens if we’re wrong and how can we best construct a portfolio. It may not reside on that theoretical efficient frontier, but in terms of the combination of potential outcomes, we have what we believe is the best assumption for the weighted probability of those outcomes.

Yeske: I fear that as financial planners we sometimes are subject to the same behavioral characteristics as clients. At some point, the pain gets great enough for us planners that we start looking for alternatives. I fear we have only a marginally greater ability to withstand short-term volatility compared with our clients. And a lot of what we’re seeing with alternative investments, such as hedge funds, falls into that category.

I just ran my own numbers for the last ten years and the portfolios of the majority of my clients who’ve been with me for ten years have average annual rates of return between nine and ten percent, and they never got that big. Those portfolios didn’t have the 25, 27 percent returns of the late ’90s, but they didn’t suffer the huge declines in the last three years. They’re running about what I originally projected.

Evensky: We reached the same conclusion. Our clients have come through the last decade extremely well. We’ve got little turnover. They’re happy. What we think is broke is that we don’t necessarily think the next decade is going to be the same. We don’t think it’s going to be a substantive change. We think the next-decade returns are likely to be marginally less. And we think there is going to be a fundamental difference in base market returns at least over the next decade.

Barnes: So we’re talking about at least moving from left to right. Harold, in your case, you’re thinking seven or eight percent after expenses, but before taxes.

Evensky: That’s correct.

Barnes: And Dave, you’re talking about something that’s not different from what you’ve been using in the past in terms of forward expectations: nine to ten percent for diversified portfolios.

Yeske: Correct.

Barnes: I would end up somewhere in between the two. The biggest problem I have with equity risk premium is the difficulty in understanding the risk premium for which equities, for which asset class—which sub-asset class of equities and, taking it further, for which equity? The market is probably less homogenous now than at any point I’ve seen over the last couple of decades. The last three or four years have probably been the least homogenous in terms of valuations and what I believe the forward returns are for various market categories.

The other problem with this entire debate is the concept of an average risk premium starting from what point. Because as you were pointing out, Harold, the forward return expectations, at least as far as my perspective on them goes, were significantly different on October 12 [1992], or whatever the day was of the latest trough in market valuations, than it is today after a pretty significant rise in valuation. So the point being, the equity risk premium is something that’s not stable over time or, a better way to explain it is that the forward expectations for returns are not stable over time. They change depending on what starting point we’re using and the valuations at that point.

Yeske: I’ve been hearing about the shrinking equity risk premium for the last two to three years, and I have to say that over that period, peak to trough, the S&P 500 has fallen by 50 percent. It is axiomatic that, all other things being equal, as the price falls, the risk premium goes up. I’m not saying what it is, but if it was a problem two or three years ago, how could it continue to be the same magnitude of problem when we fell by 50 percent? And that’s assuming you think the S&P 500 is the relevant index.

Evensky: Who said it was the same magnitude? I just said it was a problem. That’s the key. The question that it came down to for me as a practitioner is the design of my portfolio. Our equity portfolio design has been a very traditional institution. We had domestic, split it into large growth core and value, and small growth core and value. Internationally, we had core, value, small-cap, and emerging markets. We had a three percent allocation and used three managers. I used to have 12 or 15 managers on a portfolio.

My point is, if you conclude that returns are going to be somewhat less, and you calculate the cost of having all these different pieces, trying to capture little marginal pieces of alpha, and you factor in the rebalancing, the sales of managers, all of the various associated costs, we’ve concluded it doesn’t work. There’s not enough there to capture it, which is why I’m arguing that I think everyone needs to consider a core and satellite, an institutional design. Fewer moving parts. You’re giving up a certain amount of the ability to capture alpha at the margins, but we still have a value overlay, we still have a small-cap overlay, we still have an international. We still build in elements of inefficiency, but we think that it’s a much more practical, rational approach if we’re right in our returns in a retail client portfolio.

Yeske: The difference for me is that I never believed alpha existed. Alpha is just an artifact of a mis-specified model. The minute you move to a multi-factor model, a three-factor model, alpha is gone. So I have always built portfolios that were intended to load up on priced sources of risk and were intended to have some factor loadings based on a multi-factor model of return. They were not intended to add any alpha through an individual security selection, active asset allocation or active market timing. As a consequence, my portfolios at the management fund level have never had an aggregate weighted expense ratio of even 50 basis points. My portfolios, before you add in anything for me, tend to run below 50 basis points. And when you add in my fee, it adds less than 100 basis points, and actually about 90 percent of my time goes into financial planning, so I would argue that only a small part of that should be allocable to the portfolio anyway. So Harold and I are absolutely in agreement about it.

Nontraditional Assets

Barnes (referring to audience question): The question relates to nontraditional asset classes, those we haven’t normally or historically included in our textbook asset allocation efforts, and how they might be used.

Evensky: We think they’re wonderful, terrific, great. We still consider them, for the most part, institutional investments. But we’ve not seen vehicles that we think are appropriate and reasonable for the retail market.

Yeske: I don’t use them either. It seems that the search process is potentially too expensive. The lack of liquidity is a real issue, so I stick with traded securities.

Barnes: It’s the holy grail of what we do. We want an asset class that’s uncorrelated or negatively correlated with our other asset classes, and yet offer an attractive rate of return. The problem is, in each of these categories we’re talking about—direct ownership of real estate, venture capital, private equity, et cetera—there are other issues that preclude us from using them. The real-estate issue is, we can’t diversify with direct ownership in real estate sufficiently to be comfortable with the risk it introduces to the client’s portfolio. In private equity or venture capital, after ten years of reviewing hundreds of opportunities in this category, it’s become clear to me that it is no different from the IPO marketplace. So the hurdle is that we want a much higher potential return than we normally associate with our publicly traded securities because that’s what we expect to be paid for the illiquidity—and the reality is, the deals just aren’t out there.

Tax Law Changes

Barnes (referring to audience question): The question is, to what extent are the changes in tax law affecting the generation of income or additional income to the portfolio? From our perspective, they’ve made dividends a much more attractive portion of the return to the client’s portfolio and we have completely reconsidered the construction of a client’s portfolio. In the past, we would normally tilt the equity exposure into tax-deferred accounts to the extent they’re available, and place the fixed-income in the taxable account in state-tax-free municipal bonds.

With the change in the tax treatment of dividends, and with much lower tax rates applicable to long-term capital gains, we have reversed our stance on that. The problem is, we have filled portfolios that we cannot just go in and reverse the positions. Our fixed-income exposure in municipal bonds is a laddered portfolio of individual bonds. So, the prospect of taking those all out of the portfolio immediately and switching it into the tax-deferred account in taxable bonds is just not something that’s realistic in terms of the after-cost effect on the client’s portfolio. It’s something that we’re only doing gradually as bonds mature and as we have the opportunity to change the positions.

Yeske: I’ve done the projections, and it doesn’t appear to make a huge difference if I shift. I’ve spent the last three or four years shifting away from investments in taxable accounts that generated a lot of capital gains distributions or dividends, and now all of a sudden a change in the tax law is going to whip us back in the other direction. They’re doing exactly what you’re not supposed to do with tax policy: they’re giving us nothing predictable to plan around. But the difference looks marginal to me. With new investments or reallocations where we’re rebalancing, I’m allowing it to go into funds that may have a higher dividend payout or may have stocks that collectively pay a higher dividend rate than what I was doing before, but I’m only making changes at the margin.

Evensky: For us it’s been fairly marginal. In terms of income, we look at a total return portfolio and figure out how to get cash flow out of it. This whole thing about dividend-paying stocks is insane. Yes, it’s more attractive than it was before, but dividends are 15 percent, capital gains are 15 percent. I can defer a capital gain, I can’t defer a dividend. So why would I want to stock up on dividends? It makes no sense. If I have a “value manager,” and they’re buying high-dividend-paying stocks because they like the company, that’s great, and it makes it a little more attractive if it’s taxable. But I’m not buying a stock because it pays dividends or hiring a manager because they’re buying dividend-paying stocks. I’m hiring a manager because I like their style and where they’re playing. All these new dividend stories are insulting investments and they’re stupid.

Yeske: And the whole thing’s going to sunset in a few years, anyway.

Barnes: Playing the devil’s advocate with Harold, let me make one point on dividends, at least in our analysis of individual companies. The dividend is something we generally will have a higher degree of confidence in than the capital gain. And so to the extent that the rest of the story for two competing alternatives is the same, if one pays a higher dividend, we’re going to favor that one as opposed to the lower dividend company because we hopefully will have a higher level of confidence in the receipt of that dividend versus the receipt of potential capital gains down the road.

Evensky: If a company has consistently paid dividends because it’s good corporate management, I’m hoping my manager will buy the company—not because it’s paying the dividend, but because there’s good corporate management. What scares me is all these new companies that are paying dividends. There’s no earthly reason to believe they’re going to keep this up. We like, or we expect that our managers will like, companies that pay dividends. That’s an indication historically that the company has faith in its ability to maintain the dividends and we have little faith in the company keeping the money and spending it. We’d much rather they go out in the capital markets and compete for new projects. That’s not an issue of dividends, it’s an issue of corporate management.

Barnes: Higher-dividend-paying companies do not offer higher returns. Higher-returning companies pay higher dividends.

Minimizing Expenses in Portfolios

Barnes (referring to audience question): The question relates to the minimization of expenses in clients’ portfolios. Any change in how you handle that?

Evensky: We’ve always had a strong bias toward passive management. In a perfect world, we would have probably been 100 percent passive a long time ago. Hope springs eternal, so we’ve continually had pockets of active management. For as long as I can remember, our value has been consistently passive. For the most part, large-cap domestic has been passive. We’ve had active in international, largely because we’ve considered the EAFE a flawed index. But it’s been changed, so we may revisit that at some point. Emerging markets is an area where we’ve used active managers. To the extent we have had growth allocations in our portfolio, particularly small-cap growth, that’s been an active manager. Right now in equities, it’s maybe 60 or 70 percent passive, up from 50 percent.

The biggest change we’ve made is in the number of managers. We had multiple managers for large-cap growth, core, and value. The value manager’s stock did well, they sold it. Core bought it, did better, so we didn’t see it. The same stock moving back and forth—very small cost, but nevertheless there’s cost. When you’ve got that many pieces, you’re rebalancing more often, and that’s the antithesis of good tax planning because you’re always selling gains and keeping losers. It was tax-inefficient. No matter how hard we tried, we kept hiring and firing managers, so that’s the reason for this core and satellite. It’s cutting out a lot of moving parts, but it’s not a substantive change, and our orientation’s basically toward passive management.

Advisor-Guided Mutual Fund Portfolio

Barnes (referring to audience question): The question relates to the implications of an advisor-guided mutual fund portfolio, where the total expense structure can be in the 3 to 3.5 percent neighborhood, and what the implications are for that, given your expectations for returns.

Evensky: I don’t think it’s a viable model. Any kind of active management is tough to defend anyway.

Yeske: I agree. Low turnover funds, low expense funds, doing everything you can to control expenses makes sense in every way. I did a run of the Morningstar database, sorting out all of the funds in the large blend category. Once I removed those for which there was no expense ratio information available, I was left with almost exactly 1,000 funds. I took the quintile that had the highest expenses and the quintile with the lowest expenses, and I compared the 15-year performance. I found two things. One, the high-expense funds had total expenses of about three percent, not counting transaction fees, because those don’t get reported, as you know. Two, the 200 funds out of that 1,000, the bottom quintile that had the highest fees, underperformed the quintile that had the lowest fees, gross of fees. Not by much—they were close. What I find interesting about that is that there’s always justification that you’re adding value at least equal to the fees. But it turns out that it doesn’t.

Barnes: The presence of these kinds of fees, for which there is not, in our opinion, a reasonable return on those fees, is a big reason that we don’t use mutual funds, by and large, in our client portfolio.

Evensky: I think you’re particularly right, except for that large universe of purported planners who present themselves largely as money managers. As Dave said before, most of our fees are not related to the investment issue, and part of our job is to educate our clients on what they’re paying for.

Fixed Income

Barnes (referring to audience question): The question is, what is our view on fixed-income at this time?

Evensky: We have made some changes, adjusted our investment policy statement. In the past we’ve used a fairly traditional ladder portfolio, one to ten years—roughly about a 4 1/2-year duration. We’ve adjusted it so that the target’s about a 3 1/2-year duration. We may never make significant changes, but those are the kind of changes within the policy we do make. Rarely do we make a stock/bond change. We have occasionally made a small change in large versus small, but more frequently it’s a change in duration.

Yeske: I haven’t really made any changes because primarily I use funds for fixed-income and primarily I use global bond funds with average maturities of about five years. All you get is volatility above that. Beyond five years, that’s a good place to be market timing if you want to make a call on where you are in the interest rate cycle, but I don’t want to do that.

Evensky: We use a lot of funds but we ladder the funds. We have a short-term, a short-intermediate, and intermediate, so that’s how we target duration in the funds.

Barnes: We have seen our fixed-income exposure decline as a percentage of the portfolio. We build laddered sub-portfolios of bonds, state-tax-free bonds, one to five years out, and the problem is that as bond positions are maturing right now, we are not finding acceptable alternatives for that cash coming into the portfolios. So because we’re dragging our feet on reinvesting the proceeds, we have a little bit lower exposure to fixed income than what we had two years ago.

What will ultimately begin to happen is that we will have too large a cash position and we will have to turn to something, and at that point we’ll probably be looking at something like short-term bond funds or a core bond-fund type of a portfolio. Unlike Dave, we kind of split the bond exposure in client portfolios. We have a foreign slice of the pie as well, and in that category we’re using a mutual fund, and that’s remained pretty constant.

Commodities

Barnes (referring to audience question): The question is about the use of commodities in portfolios.

Evensky: We’ve used the Oppenheimer Real Asset, which basically mimics the Goldman Sachs Commodity Index. To really believe that someone’s going to be prepared to have 20 percent in commodities, 20 percent REITs—I mean, the day we went into it, it dropped 50 percent that year, so luckily, we had a fairly small allocation. The other problem is the cost with the rebalancing. So, intellectually I agree with what Roger [Gibson] says, but I don’t think practically in the retail world that that’s a reasonable solution.

Yeske: I don’t use commodities or commodity funds per se. Obviously some of the companies that are held in the mutual funds we use are sensitive to changes in commodity prices. This is the beauty, in my opinion, of having some kind of a value bias. Some people say, “That’s just a sector rotation strategy,” and my answer is, “Well, yeah, probably.” Because if you look at value, whatever segments of the economy are most depressed end up in a value fund, and that tends to rotate from sector to sector. But the key is that I’m not ginning up some big massive model of the economy and trying to figure out where it’s going to be, I just let the market reveal that to me.

Barnes: We’ve not used a specific commodity exposure in client portfolios at this point. But PIMCO has come out with a couple of funds lately that are, at least at this early stage, extremely interesting. So we’re playing around with the idea of whether we would introduce something like that and to what extent it makes sense in client portfolios.

Again it’s part of the process that we are constantly going through, which is, what else can we include in the portfolio that would add value? We have a specific category in our asset allocation that’s called “other” because I have no idea what we can call it. But what we’re trying to fill that category with are assets that offer an attractive rate of return, but do so with a pattern of returns that will be substantially different from the other assets in the portfolio. Some of the things that have been included in that category are TIPS [Treasury inflation-protection securities] and gold mining companies. So if we were to introduce commodities, they would fit into that category in our strategy.

Satellite Component

Barnes (referring to audience question): The question relates to the selection process for the satellite component of a portfolio.

Evensky: I’m blushing a little bit because it’s sort of counter-intuitive for us. The satellite criteria—we’ve currently set 20 percent of our equity to satellite—that’s our alpha-adding, opportunistic part of the portfolio. The core is just four positions: 50 percent of that, in most cases, is a Russell 2000 ETF. One criterion for the satellite is that it’s a fundamentally sound economic investment. We want to avoid some story sale. Second, we think, net of expenses and taxes, that it will generate a return better than our core net return. Third, it does not have to be negatively correlated or poorly correlated with the core. The nature of anything we put in there is likely to be, but that’s not one of the criteria. It’s not there to balance the risk, it’s there purely as a value-added. We split it in at least three parts, simply because we’re just not prepared to make that kind of a big bet. After that, basically anything could conceivably fit.

We’ve looked at high yield bonds, bond funds, hedge funds, commodities, but right now the three positions we have in there are a micro-cap growth, emerging markets, and a short position on Treasuries. We had the commodities for a variety of reasons. We have a tactical overlay on the satellite, which is also new for us, and we were never prepared to do that on the whole portfolio. We believe in the concept of basic relative market returns and of reversion to the mean, so at the point that we had 2 1/2 standard deviations above our expected returns in commodities, we said, “This is great—this is new for us. We don’t have to stay there.” In the past we always had to stay there. So we said, “Let’s take our money and look for something else.”