Understanding the fundamental link between risk and return is critical to intelligent investing. But the intuition for how the two relate is not often explained. How are prices set for risky assets? Why should the return of an asset be related to risk? Should you be able to find assets with high rates of return and little risk? What do the trade-offs look like for typical investments? In the investing world, risk and return are always related.

Now I have nothing against complimentary mid-day meals, but as with most things in life, the concept of getting something for nothing is rarely realized. This is particularly true in the investing world. Whenever you seek higher rates of return, risk always comes along for the ride. The risk may not be visible, nor will it necessarily be realized, but it is always there, lurking in the shadows. Investing without recognizing and accounting for risk can be disastrous, a truism for both individual and institutional investors. History is littered with examples where financial firms have lost billions of dollars by forgetting this reality. For instance, in late 2007 and early 2008, Citigroup, Merrill Lynch, and UBS had to write-off losses of more than $50 billion due to bad bets on the U.S. mortgage market. The golden rule with investments is to take on risk that you get paid for and avoid risk that comes with no expected reward. In addition, you should be careful to expose yourself to only those risks for which you are willing to bear the consequences if things do not go your way.

Institutional investors generally spend a great deal of time and effort analyzing the amount and types of investment risk in their portfolios (for very good reasons). In fact, some of the most important contributions from financial economics have to do with the role of risk in markets, especially how to evaluate risk and mitigate its impact. Surprisingly, in the deliberations of many individual investors, it is not obvious that risk enters into the decision at all.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

Different types of investment securities have different levels of risk. While risk can be measured in a number of ways, the most common measure used by institutional investors is the volatility of returns, typically expressed on an annual basis. The least-risky assets are those with low volatilities, and the highest-risk assets are those that expose investors to high levels of returns' volatility.

For most individual investors, the range of investable securities is broadly divided into two categories: fixed income, or bonds, and equities, or stocks. Bonds are securities that involve one or more periodic payments at specific times in the future by a government or corporate entity. Unlike stocks, bonds do not imply any ownership of an entity, but rather represent a loan of money. When you purchase a bond you are in effect loaning money to that organization in return for interest payments. Equities or stocks, on the other hand, represent ownership shares of a corporation. While some companies pay regular payments to stockholders in the form of dividends, many stocks have no regular payments at all. The prices of stocks vary with the outlook for the overall market and the fortunes of that specific company.

The reason that we divide investable securities into categories is that different types of assets have very different risk and return characteristics. For example, bonds tend to have less correlation with the overall market and thus have lower expected returns than equities.

Within the categories of fixed income and equities, assets are often broken down into subcategories called asset classes. For fixed-income asset classes, the categorization is usually based on how far out the payments extend (the maturity of the bond), and who is guaranteeing the payments (usually a bank, corporation, or government). Fixed-income assets include cash, money market funds, government bonds of various maturities, corporate bonds, municipal bonds, mortgage bonds, and foreign government bonds.

For equities, the division into asset classes is typically based on the size of the company (in terms of its market value), the market value of the company relative to the accounting value of its assets, and whether the equity is foreign or domestic. In the case of the Financial Engines asset class model, domestic equities are divided into six asset classes: large, medium, and small capitalization stocks (often referred to as large-cap, mid-cap, and small-cap, respectively) in both value and growth dimensions. Growth stocks have relatively high prices compared to the value of their corporate assets, while value stocks have relatively low prices compared to their assets. Since the prices of growth stocks imply a higher ability to generate future profits from their assets, their returns tend to be more sensitive to the performance of the overall market. Value stocks tend to be a bit less sensitive to the movements in the overall market. For foreign equities, the Financial Engine model divides the world into asset classes for Europe, Pacific Rim countries, and emerging markets.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

History is an extremely poor guide to future expected returns for financial assets. While it might seem reasonable that the average stock market return over the last 50 years is a logical estimate for what the next 50 years may hold, there is a large range of error in the historical average. The reason is that  stock returns are very volatile relative to their average value.  If a variable is very consistent, then only a relatively few observations are needed to accurately estimate its expected value. However, if the uncertainty in the number is large relative to its average value, then a much larger number of observations are required to accurately estimate its expected value.

For example, if U.S. stock market returns (adjusted for inflation) have a true long-run annualized average of 7.5 percent  with a volatility of 1 percent, then only a small number of years would be required to estimate the average expected return to within plus or minus 1 percent. In fact, with only four years of data you could be quite sure that your estimate of the expected return was within plus or minus 1 percent  of the true value.  In this example, the uncertainty of stock returns (volatility) is about one-eighth as large as the average value. While the historical compounded annualized average for inflation-adjusted U.S. equity returns may be in the range of 7.5 percent, the volatility of these returns is approximately 20 percent-or roughly two-and-a-half times larger than the average. This means you need a lot more data to accurately estimate the average value with any confidence.

Here is an interesting question that I have sometimes posed in conference presentations: How many years of U.S. stock returns would you need to observe in order to estimate the expected return on stocks to within plus or minus 1 percent with high confidence? Ask this question of most individual investors or even financial advisors, and you will typically get numbers in the range of 50-100 years of data. But if you believe that the actual volatility of stock returns is about 20 percent  per year (consistent with historical estimates), then you would need an astounding 1,500 years of historical stock returns in order to estimate the expected return on stocks to within plus or minus 1 percent  accuracy.  Obviously we don't have stock return histories going back to the end of the Roman Empire, and even if we did, it is doubtful that the returns on the stocks of chariot makers in 400 a.d. would have any relevance to today's modern market.

Accordingly, historical equity returns are not a very valuable guide to future expected returns. With the 100 years or so of data we are able to observe for stock returns in the United States, the historical average is only accurate to within plus or minus about 4 percent  of the long-term expected value.  This is a huge range of possible expected returns when doing future wealth projections. The consequence of overestimating or underestimating expected returns on stocks by 4 percent  over a 25-year investment horizon results in about a factor of six difference in ending wealth values. That is, the high-end return estimate would yield wealth that is over six times more than the low-end estimate (adjusted for inflation). Clearly any forecast that could be off by this magnitude is going to be problematic-hence the futility of generating accurate forward-looking returns estimates solely from historical averages. Remember this the next time you are presented with a chart of historical stock returns and told that they represent the best forecast of what might happen in the future. In reality, there is a great deal of uncertainty about the level of returns we can expect to see in the future.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

In its starkest terms, the argument over market efficiency is really kind of silly. Of course the market is efficient (or I suppose, inefficient, depending on your point of view); the real debate revolves around the degree of efficiency. The overwhelming consensus among financial economists and most institutional investors is that the market is very good at incorporating public information into prices, usually within very short periods of time. There is little doubt that prices react quickly to new information about the future value of assets and that there are few obvious opportunities to make easy money with little or no risk. The more interesting debate involves just what kinds of information are incorporated into prices by the trading process, and just how that information impacts the value of assets. This remains an active and ongoing area of research among financial economists.

A slightly less austere view of the market efficiency theory is that the market is very good at assimilating information into prices, making it quite difficult to spot and exploit erroneous valuations placed on assets. Adherents to this point of view make the obvious observation that if the market were widely and predictably making mistakes about the value of assets, then there would be informed traders ready to exploit their superior knowledge to make money from these errors. However, in so doing, these informed traders would eliminate the very errors that created the profit opportunity in the first place. For instance, if you believe a stock is priced too low by the market, then you would buy the stock. But purchasing the stock drives its price up, and if you purchase enough of it, you eliminate any opportunity for further gain. Given the literally hundreds of billions of dollars chasing profit opportunities globally on a second-by-second basis, it seems likely that if such pricing inefficiencies exist they are going to be difficult and perhaps costly to find and exploit. And if they do exist, they sure won't stick around for very long.

My own point of view is that markets are highly efficient, particularly at the aggregate level of overall asset classes. While there may be examples of individual securities that have prices that are too high or too low for short periods of time, it is highly unlikely that the market is making consistent mistakes that result in predictable and easy-to-exploit profit opportunities. There is an active debate among economists over the precise degree of market efficiency, but few would argue that there are a large number of figurative $20 bills lying around on the sidewalk just waiting to be picked up by observant investors. It is much more likely that if such market inefficiencies exist, they are costly to find and fleeting in nature. Trying to beat the market is a very difficult game and perhaps only worth pursuing if you have substantial resources and uncommon access to information (for instance, the ability to directly observe trading activity in real time at a high level of detail, or superior information about nonpublic companies and the value of their assets). It is no exaggeration to say that there are hundreds of billions of dollars and thousands of the world's brightest minds devoted to finding just such market inefficiencies. As a consequence, it is highly unlikely that individual investors on their own will be able to find consistently profitable opportunities to beat the market. While this view may be at odds with hundreds of personal investing books out there pushing various investing systems, it is a much more realistic and pragmatic view of the investing world-one that is far more likely to result in good outcomes for your investments.

Even if you admit the possibility that there may be inefficiencies in the market, it may be the case that the cost of finding and exploiting such opportunities may just offset the expected profit. Perhaps the transaction costs to exploit such an inefficiency eat up most of the potential profit. Or those with access to valuable insights about future asset values may charge an amount that offsets some or all of the potential for excess gains. In any event, there is strong evidence that reliably beating the market is a very tough thing to do.

Furthermore, with the rapid development and application of technology and computing to financial trading, markets are almost certainly becoming more efficient over time. Information that once took days to be disseminated is now globally transmitted in microseconds, with near instantaneous impact on market prices. I would not argue that markets are perfectly efficient at all times, but merely that the prices for traded financial assets reflect relevant public information about their future value under the vast majority of circumstances. Perhaps intelligent traders with superior knowledge and analytic tools can find opportunities for excess profit beyond market returns, but it is not easy for the every-day investor to do this. And the strategies that are employed to exploit such opportunities almost always incorporate significant additional investment risks and transaction costs. It pays to be skeptical when considering the claims of someone who says he or she has a sure-fire way to beat the market. The market is a very tough competitor and the playing field is littered with those who, to their own detriment, underestimated its remarkable wisdom.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

When selecting funds and managing your assets, it pays to devote substantial attention to the costs of investing. Unlike many types of products, the fees associated with investment funds directly detract from the net value you receive. This does not imply that the cheapest investment fund will always be the best choice (in Chapter 9 of The Intelligent Portfolio, you'll learn there is more to selecting good funds than fees alone). But it does mean that paying incremental investment fees without comparable value being provided in return is something to be avoided. Make sure the fees you pay add up. if you decide to purchase a fund with a higher expense ratio or a load charge, be sure you understand the value that you are receiving for the additional fees.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

To learn more, delve into Chapter 7 of The Intelligent Portfolio, "How fees eat your lunch" 

Optimization is a mathematical technique for figuring out the best solution to a particular problem, given a range of possible alternatives. In the context of investing, the purpose of optimization is to select investments that do the best job of meeting a set of objectives. These objectives may be fairly simple (give me the combination of funds with the highest expected return for a desired level of risk), or they may be very complex, with numerous competing goals and constraints.

In the institutional world of pension funds and university endowments, portfolio optimization is used to create alternative investment strategies. Data on each of the investment options, along with information about desired risk levels and other constraints, are fed into an optimization algorithm to yield a recommended portfolio allocation. The most popular forms of portfolio optimization are based on the concept of mean-variance portfolio theory. Very simply, this means an optimization where the goal is to maximize the expected return of the portfolio for a given level of variance (volatility). The optimizer takes information on each of the securities available to you and determines the portfolio that offers the highest expected return for the level of risk that you are willing to assume.

Of course, there are many ways to measure risk other than looking at just the volatility of returns. More complex optimization methods are able to accommodate different measures for risk, and the inclusion of various types of constraints. For instance, a given investment may have a minimum investment amount, or you may wish to impose a maximum percentage allocation to a particular security. Moreover, there may be factors other than expected return that you may wish to optimize. For instance, capital gains taxes can impact the desirability of buying or selling certain securities. There may also be preferences for the number of securities in the portfolio, or for concentration in a single security. Each of these constraints or preferences can add significant complexity to the optimization algorithm. Modern portfolio optimization engines are able to handle a wide variety of such real-world complexities. The bottom line is that optimization engines are able to help you build portfolios that achieve desired objectives while abiding by constraints and preferences that you believe are important. They are an invaluable tool for modern portfolio management, both for institutions and for individual investors.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

Monte Carlo simulation provides a powerful way to analyze problems that involve uncertainty. The formal development of the Monte Carlo method dates from the Manhattan Project to develop the atom bomb during World War II and is based on the work of two famous mathematicians, Stanislaus Ulam and John von Neumann.

From its early use in nuclear physics, Monte Carlo simulation made its way into other fields of study in subsequent years and is now used in a wide range of domains, including weather forecasting, traffic analysis, chemistry, genetics, statistics, and investment analysis. The method is particularly useful in solving problems where the probabilistic behavior of a system is very complex. The probabilistic behavior of an investment portfolio consisting of a mix of different securities represents one such complex system. That is why institutional investors such as defined benefit pension managers have used Monte Carlo simulation for more than two decades to evaluate the likelihood of meeting future liabilities. More recently, Monte Carlo simulation has rapidly evolved from a mathematical curiosity into something of a marketing buzzword in the financial services industry. The purpose of Monte Carlo investment simulation is to provide a better understanding of the range of possible outcomes when the future values of the assets in the portfolio are uncertain. That is, Monte Carlo simulation can't help you predict the future (unfortunately no one can do that), but it can show you the range of possible future outcomes you might expect with different market performance.

These simulation tools provide important benefits for investors, including the ability to:

• Calculate the probability of reaching investment objectives
• Provide forward-looking measures of investment risk
• Test drive investment strategies prior to implementation

Virtually all types of financial assets have some degree of uncertainty in their future values. For instance, the future value of a portfolio of stocks may be worth more or less than its current value based on whether the stock market moves up or down. Even guaranteed securities like bank CDs, which are advertised as being very low (or no) risk, have some uncertainty in their future values. While the rate of interest may be guaranteed at the end of the period (say, 1 year), the actual value of the assets is dependent on how inflation behaves over the holding period. If inflation is low, the value of the CD account will be worth correspondingly more; if inflation spikes up unexpectedly, the value of the CD will be lower (in today's dollars), since the value of the accumulated returns will have been eroded by inflation.

The simulation of investment portfolios provides a powerful tool for investors. It allows the evaluation of different decisions (such as how much risk to take, how much to save, etc.) and how they impact the range of possible investment outcomes. Rather than evaluating such choices in the dark, we can directly observe how different decisions impact the outcomes that you ultimately care about. For instance, you can calculate how additional savings of $100 per month impacts the probability of reaching your desired retirement income goal. Such simulations can be quite realistic, taking into consideration the impact of market performance, fees, specific security risks, taxes, account distribution rules, and even uncertainty in how long you might live.

The mechanics of Monte Carlo simulation are straightforward in principle but often complex in implementation. The Monte Carlo method is predicated on repeated randomized experiments of a statistic of interest (for instance, the future value of an investment portfolio). The idea is to model the random behavior of a complex system (such as a portfolio of investments) by repeatedly sampling different possible random values of the variables that drive the outcomes of the system (such as the future values of each security in the portfolio).

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

If you believe that the market is mostly efficient, at least at the level of asset classes such as large-cap stocks or corporate bonds, then there is significant information in the market's consensus view of the future embedded in the prices of traded assets. One way of getting at that information about the future is to add up what all investors hold in their investment portfolios. The market portfolio consists of the aggregation of all the assets in the world, weighted in proportion to their total market values. For instance, at a high level, the market portfolio consists of all the bonds and stocks in the world, with the portfolio allocations determined by the current total market value of stocks and bonds, respectively. Of course, you can specify the composition of the market portfolio in a greater degree of detail by splitting up the various asset classes into finer categories (for instance into large-, medium-, and small-capitalization stocks, value and growth stocks, or different types of bonds).

You can also think of the market portfolio as representing the average portfolio allocation of all investors, since it reflects the average portfolio held by all investors. In principle, the market portfolio should consist of all financial assets, though usually approximations are made using indexes that track the mostly widely held asset classes. Standard financial economic theory dictates that the market portfolio is efficient and that it has the highest expected return of any portfolio for that level of volatility. You can achieve higher expected return portfolios, but only if you are willing to assume more market volatility. As we will see, this is an important characteristic of the market portfolio.

Another aspect of the market portfolio is that it represents an efficient allocation of asset classes for an investor with an average tolerance for risk. Investors often ask me what a "typical" efficient investment allocation looks like. If you have risk tolerance similar to the average investor, you have the ability to invest in all asset classes at a reasonable cost, and you are not forced to hold any particular assets or positions (for instance, holding restricted positions in your employer's stock), then the market portfolio is a natural benchmark for an efficient allocation to the various asset classes. Of course, your circumstances may be different and you may require something other than what is appropriate for the average investor.

Furthermore, the market portfolio is based on market consensus expectations about the future risk and return of each asset class. Since it represents the combined expectations of all market participants, it embeds an enormous amount of information about the future. Accordingly, any investment methodology that ignores this important source of information is likely to be flawed. What is surprising is how many investors, even experienced professional ones, completely ignore this important source of information about future expected returns. The value of this information does not depend on you adhering strictly to the market consensus view. Even if you wish to make bets that differ from the overall market consensus, it is still highly valuable to understand what the market consensus is projecting so that you can know how to act upon your predictions based on how they differ from the market's perspective.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

Another prominent technique in the institutional investor toolkit is the use of analytic methods to determine how specific investments are related to market returns, and how they have performed relative to what we would expect given the type of investment. When selecting investments for a portfolio, it is very useful to know what their expected risk and return characteristics are and how they have performed compared to other similar assets. Financial Engines has adopted these methods used by institutional investors to help individual investors do a better job of investment selection. One important technique for determining what an investment acts like is investment style analysis.

Investment style analysis, or returns-based style analysis, was developed by William F. Sharpe, a 1990 Nobel laureate  and cofounder of Financial Engines, for use in assessing and measuring the performance of investment managers for pension funds and other institutional money managers. The technique was first published in a paper by Sharpe in 1988. Since then, the method has gained widespread use throughout the investing world due to its unique ability to look through a fund's underlying investments to understand how it behaves.

The key to the technique is to recognize that the performance of a fund is actually driven by a combination of exposures to different parts of the economy. For instance, a large-capitalization growth equity mutual fund may have exposures to growth stocks with large capitalizations. However, most funds have exposures to other parts of the market as well. Perhaps some of the stocks held by the fund may be from small or midsized companies, or from value-oriented stocks. If the fund is actively managed (as opposed to an index fund), it may keep a small percentage of assets in cash in order to better handle daily inflows and outflows. Style analysis of such a fund would yield exposures to large-cap growth equity, and perhaps various other equity exposures, or a small percentage in cash. The important point is that most funds are not easily categorized into a single asset class. In the real world, funds have several exposures and often behave significantly different from their stated investment objective. It's kind of like the difference between the name on the label and the list of ingredients on the back of the can (for instance, check out the ingredients listed on a can of Cheez Whiz).

Style analysis works by examining how a fund's returns move with various asset classes in the market. For example, some funds might move more closely with changes in the value of smaller company stocks, while other funds might be more related to movements in corporate bonds. Sharpe discovered that by setting up the problem a particular way, one could derive accurate estimates of a fund's investment exposures with a relatively small amount of data on historical returns. This opened up a whole new window of analysis of funds and their investment behavior. Instead of relying on a single designated benchmark like the S&P 500 Index, one could construct a custom benchmark for each fund that was representative of types of investments selected by their investment strategy. A manager who mostly invested in value stocks but held 20 percent  of the portfolio in growth stocks could be measured against the performance of a benchmark based on 20 percent  growth equity and 80 percent  value equity. With the use of such custom benchmarks, it is possible to more accurately ascertain the extent an investment manager is actually adding value. The use of style analysis techniques among professional investors has greatly expanded in recent years due to its flexibility and utility.

Knowledge of a fund's investment style, as we shall see, is critical in building effective portfolios that maximize the expected return for a given level of risk.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

The Financial Engines' investment database includes over 17,000 mutual funds at the time of this writing (including all share classes of each fund). Each one of those funds has an associated scorecard that ranks the fund on several measures relative to its peer group (defined by funds with similar investment styles). Each fund is ranked on four separate measures:

•  Fund-specific risk (risk in addition to its investment style)
•  Expenses (the total cost of the fund)
•  Turnover (how often the fund manager trades)
•  Historical alpha (the degree to which the fund over- or underperformed relative to its investment style)

 For each of these measures, the fund's ranking relative to its peer group is charted in terms of 20 percent  increments (represented by five yellow dots). A rating of one out of five dots indicates that the fund ranks below 80 percent  of its peers for this measure, while a rating of four dots indicates that the fund is in the top 40 percent  of its peer group. A rating of five dots indicates that the fund is in the top 20 percent  of its peer group.

The fund-specific risk measure evaluates how much active management, concentration, or investment style rotation the fund displays. Actively managed funds have more fund-specific risk than index- or passively managed funds, sometimes much more. Funds have higher fund-specific risk when they concentrate their investments into a small number of stocks, or when they frequently change their investment style over time.

The expenses metric evaluates the cost of owning the fund. Higher expenses paid to the fund manager imply less money that you get to keep. As we will see in Chapter 9, of The Intelligent Portfolio, expenses are an extremely important factor in selecting good investments.

The turnover metric evaluates how often the fund buys and sells securities in its portfolio. This is important because any time a fund manager purchases or sells a position in its portfolio, they incur costs in the form of brokerage commissions and market spreads (the difference between the price that you can buy and sell a security in the stock market). While these costs are generally small for a given transaction, they can add up quickly for an actively managed fund that is doing lots of trading. Furthermore, a fund that trades often is less tax efficient than one that trades infrequently, due to the fact that the manager must distribute capital gains to its shareholders. This can be an important consideration for funds held in a taxable account.

Finally, the historical alpha metric evaluates the historical performance of the fund relative to its specific investment style. The alpha measure can be positive or negative depending on whether the manager showed performance above or below the returns of its investment style benchmark. The historical alpha measures the difference between the returns of the funds and the returns of the underlying investment-style benchmark.

Now that may have been more analysis than some of you were counting on. But the core ideas in this chapter are really pretty simple. Unlike most endeavors, the link between observed performance and skill in investing is very noisy. When someone hits 50 home runs during a major league season, you can be reasonably sure that they must have some uncommon skill in hitting baseballs to have accomplished such a feat. But a person who randomly flips eight heads in a row is no more skilled at coin flipping than anyone else. They just got lucky. Given the amount of uncertainty in the performance of investment managers, it is very difficult to separate out the good from the lucky (or the bad from the unlucky, for that matter). Wall Street knows that a good track record is a big selling point for investment funds. They put a lot more marketing muscle behind those products with the gaudy performance records than those with only average returns. Very few investment performance histories fall outside of what can be attributed to random variation. Make sure you consider this carefully before committing your hard-earned dollars to a speculative bet on future performance matching that of the past.

* Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

In many 401(k) retirement plans, particularly for larger employers, investors will have access to institutional fund managers who often charge far lower fees than comparable retail mutual funds. Often, institutional fund managers charge fees that are only 10 percent to 50 percent of those charged by mutual funds. Clearly, these investments can be a very good deal for participants in the plan. Moreover, the presence of such funds can make it profitable to keep your money in the plan after you leave the company, even if you might have greater fund choice in a retail IRA rollout account. Don't let the brand names of retail mutual funds distract you from taking advantage of a great deal with less well-known, but very cost-effective institutional fund products.

Copyright (c) 2008 by Christopher L. Jones.  All rights reserved.  This material is used by permission of John Wiley & Sons, Inc.

To learn more, delve into Chapter 7 of The Intelligent Portfolio, "How fees eat your lunch"