This interview originally appeared in Motley Fool Inside Value.

Value investor William Nygren has 34 years of investment experience, including eight years as director of research at Harris Associates. Now a partner at Harris, he is also portfolio manager at the Oakmark Fund, Oakmark Select Fund, and Oakmark Global Select Fund. As of December 31, assets under management at Harris Associates were about $123 billion, and the Oakmark fund family had about $75 billion. Nygren's many accolades include being named Morningstar's domestic stock manager of the year in 2001.

Fool analyst John Rotonti recently interviewed him about the qualities he looks for in companies and management teams, the best indicators of stock outperformance, and how he likes to value businesses.

John Rotonti: What's your definition of a high-quality business?

Bill Nygren: Oakmark would consider a business to be high-quality if it had many of the following attributes:

  • high return on incremental invested capital
  • high free cash generation
  • above-average growth opportunities
  • competitive advantage not subject to replication
  • low cyclicality
  • low risk of obsolescence

I'm no doubt missing other important characteristics, but you get the idea. More importantly, however, is that there is a big difference between a high-quality business and a great stock, and that difference is stock price. Just as consumers do, we are always making trade-offs between price and quality.

John Rotonti: What's your definition of a high-quality management team?

Bill Nygren: I think all investors would agree that a very important measure of a management team is its ability to maximize long-term returns for the business. It is rare that we disagree with the consensus on how well management runs the business. The issue we spend a great deal of time analyzing — that many other investors gloss over — is how a manager allocates capital.

If you buy a stock expecting to sell it after the next earnings release, capital allocation barely matters. But if you buy a stock expecting to hold it for five years, like we do, capital allocation becomes critically important. We define good capital allocators as always searching for the option that maximizes long-term risk-adjusted return. That means thinking like an owner, not a professional manager. A professional manager is almost always focused on making the company bigger — effectively focusing on the numerator. An owner pays just as much attention to the denominator, meaning maximizing per-share values.

John Rotonti: Warren Buffett has often defined a good business as one that generates a high return on tangible equity, and Joel Greenblatt uses tangible capital to calculate return on invested capital (ROIC) in The Little Book That Beats the MarketWhen you calculate ROIC or return on equity (ROE), do you prefer to keep the intangibles in or take them out?

Bill Nygren: I don't see it as an either/or selection. If you are assessing how well the company allocated its capital when it made acquisitions, of course you want the denominator to include the full cost of those acquisitions, not just the tangible assets that were acquired. If, however, you are making projections about returns from future investment for organic growth, then you would not want the denominator to include the acquisition premium.

John Rotonti: Do you think there's one source of competitive advantage that's stronger and more enduring than others?

Bill Nygren: As soon as it is obvious that a business is successful (meaning it earns a higher return than its cost of capital), other businesses begin trying to replicate that success. The most obvious enduring protection is a patent that can't be bypassed (think pharmaceuticals), which gives relative certainty to monopoly status until expiration. Unfortunately, most investors recognize and are willing to pay for that advantage. Other competitive advantages we like are brand names, high switching costs, and scale. The longer I've been at this, the more respect I gain for corporate culture being a sustainable advantage. One would think that Liberty Media's (FWONA) singular focus on per-share value, 3G's focus on cost reduction, or Goldman Sachs's (GS 0.83%) focus on hiring the best and brightest ought to be easily copied, but it isn't. I think that an advantaged corporate culture is the "moat" that is most often available for free. 

John Rotonti: Now a few questions on valuation. In a previous conversation, you told me that your preferred method for estimating intrinsic value is to look at recent acquisition multiples for comparable businesses and that you use discounted cash flow analysis just as a sanity check. For businesses that don't have comparable acquisition multiples (such as Apple (AAPL 1.97%), Google holding company Alphabet (GOOG 0.23%) (GOOGL 0.26%), or Amazon (AMZN 0.22%)), what's your next preferred method?

Bill Nygren: Our preferred method is not encompassed by one summary statistic. We want to apply the statistic(s) that best captures each company's unique value. We owned Amazon most of last year — an odd holding for a value manager given a P/E ratio of several hundred. The metric that gave us confidence that Amazon was cheaper than other retailers was enterprise value divided by sales. It seemed odd to us that Amazon, growing sales at 20%-plus annually, was priced at a smaller percentage of sales than were average brick-and-mortar stores. Other investors were focused on Amazon's low margin; we believed forfeiting a year of earnings was a small price to pay to grow more than 2000 basis points faster than competitors.

In the case of Google, now Alphabet, we use a sum-of-the-parts approach explicitly valuing cash, cumulative investments in venture cap-like projects, and YouTube valued at a similar price-to-hours-watched ratio as other media companies. We then apply an appropriate multiple to search EBITA considering its low incremental capital needs, high market share, and strong industry tailwind.

John Rotonti: You have sometimes supplemented your analysis with the price-to-sales ratio comparing companies in the same industry. What are your thoughts on using the P/S ratio when looking at cyclical companies in some of the heavier industries? Does the P/S multiple eliminate some of the operating leverage component inherent in cyclical industries?

Bill Nygren: I think price-to-sales is most useful when comparing very similar companies. If you told me Wal-Mart (WMT 0.60%) was selling at a lower price-to-sales than GM (GM 1.35%), that doesn't shout to me that one is mispriced relative to the other. However, if you said GM is much lower than Ford (F 1.20%) — that starts to sound interesting. Underlying a price-to-sales comparison is the implicit statement that margins will trend toward the same number for both companies.

Using price-to-sales to claim that a cyclical is cheap is really no different than using a price-to-earnings ratio, but substituting "normal" margins for those currently earned. We try to smooth the earnings peaks and valleys in our valuations but tend more often to use price-to-normalized-earnings because it can also be used to compare across industries.

John Rotonti: For companies with high ROIC, do you think it's useful to incorporate enterprise value/invested capital as a valuation metric? I have sometimes found that companies that have high ROIC and are trading at lower multiples of invested capital (let's say EV/invested capital less than 3) tend to also look attractive using other valuation methodologies.

Bill Nygren: That's a metric we haven't used. The value of a high ROIC really comes from high incremental returns, which can be quite different than the base returns. Back when I was taking finance courses, they taught that a quick way to forecast a company's growth rate was to take its ROE and multiply by the earnings retention rate (one less the dividend payout ratio). So a company that had a 25% ROE and paid out 20% of earnings would be expected to grow at a 20% rate. The problem with that, of course, is that a company might have no ability to invest new capital in projects that are as attractive as the base. For example, think of how much cash Apple (AAPL 1.97%) generates versus its reinvestment opportunities.

The ideal situation for a value investor would be a company that was priced appropriately for its mediocre return on existing capital but had a very high return on incremental capital and had ample opportunities to invest for organic growth. An example could be a company that has market-share growth opportunities in an industry with large economies of scale.

John Rotonti: In your experience, which factors lead to share outperformance over a long period of time?

Bill Nygren: Your first question asked what defined a high-quality business, and I answered by listing a number of characteristics such as high return on capital, high growth, high free cash flow, and so on. A company that can enjoy those characteristics for a long time will perform significantly better than the average company. But to also be a significantly above-average stock, it has to benefit from those characteristics for a longer time than is implied by the current stock price. I don't think you can separate the positive fundamental attributes from the stock price when trying to predict returns. A great company purchased at too high of a price will be a lousy investment, and a mediocre company purchased at a very cheap price can be a great investment.

John Rotonti: Are there ways of identifying what I'll call non-GAAP compounders, and is this something you try to do? These are fast-growth companies that are investing heavily today so they may not be generating a GAAP profit. I think some metrics to focus on may be customer retention and market share.

Bill Nygren: Any time you can identify a measure of value that isn't earnings-based, you have the potential to see value compounding where investors focused on reported earnings don't see it. Examples we have profited from include using price-to-sales to identify value growth at Amazon, using price-to-subscriber to value cable TV or programming companies, price-to- EBITDA plus R&D for drug companies. All can be examples of businesses where growth capital is invested through the income statement rather than being capitalized.

John Rotonti: From what I can tell, you have tended to sell once a stock price approaches your estimate of fair value. Is this always the case, or are there some companies that you consider of such high quality that you would consider holding as long as the valuation does not balloon outside a zone of reasonableness?

Bill Nygren: When we purchase a stock at Oakmark, we set a sell target that is based on our best estimate of the value of that company. Our value estimate incorporates our belief about how the business quality compares to other businesses — higher-quality companies would benefit from higher multiples used to set sell targets. While we own a stock, the analyst's job is to constantly refine and update our estimate of fundamental value, and thus the sell target. When that target is reached, unless we are delaying the sale to allow a gain to benefit from lower long-term capital gain tax rates, we sell.

Investing is always about opportunity cost — by definition, having your assets invested in a specific portfolio of stocks means you have no capital remaining for the stocks not in the portfolio. One of the all-time great investors, Charlie Munger, is constantly reminding us that some conclusions are easier to reach if we invert our thinking. So let's assume we want to hold stocks of great businesses when they appear fully valued. The capital used to make room for holding those stocks means you no longer have as much capital to buy all the stocks you have identified as undervalued. You would effectively be saying that a fully valued stock is a superior investment to an undervalued stock. The only way that could be true is if you didn't build in an appropriate multiple premium for a great business when setting your sell target. I think it is a very risky game for a value investor to become a momentum investor once a stock has reached its sell target.

John Rotonti: How do you approach diversification?

Bill Nygren: The goal of diversification in a portfolio is to take advantage of the fact that risk is reduced by adding a stock that is not perfectly correlated with the other stocks owned. If we make an extreme example by assuming that Apple and Samsung will share the smartphone market, but we don't know what their relative shares will be, an investor could eliminate the company-specific market-share risk by owning both stocks, but would still have the smartphone industry risk. But real-world examples are never that precise; there is always the risk of a new competitor.

We think about portfolio risk across industry lines. The portfolio risks we want to consciously consider are those where a certain macro event would require revisions to our value estimates for multiple portfolio holdings. For example, MasterCard (MA 0.78%) (technology), QVC (QVCA) (media or retailing) and Fiat Chrysler (FCAU) (auto) are all owned in the Oakmark Select Fund and all are considered to be in different industries. But all three will react negatively if consumer spending is expected to decline. So instead of just looking at our industry weightings, we think about the percentage of the portfolio that is exposed to consumer spending, or interest rates, or housing, or changing currency rates and so on. When we consider which stock should be added to the portfolio when we sell an existing holding, we look at how it changes the risk to the total portfolio. If we already own stocks exposed to the same macro risks, we set a higher hurdle for adding to that risk. For example, Oakmark Select today is very heavily invested in financials that are deemed to be systemically important and therefore are subject to additional regulation. Despite believing that some systemically important financials we don't own are also very attractively valued, if the undervaluation is even close, we'd prefer to add a stock that had a different risk profile. Industry classification can give a quick indication of overlapping risks, but we believe a more careful look across industry lines is also necessary. And remember, the goal isn't to eliminate all company-specific or industry-specific risk from the portfolio. That would basically turn it into an index fund. The goal is to stay in the game even when you are wrong and make sure you are getting paid adequately for the risks you are taking.

John Rotonti: Can you discuss the research process at Oakmark?

Bill Nygren: Our analysts are all generalists. They will only be hired at Oakmark if they are value investors. We tell them the same thing we put at the start of each of our quarterly reports:

"At Oakmark, we are long-term investors. We attempt to identify growing businesses that are managed to benefit their shareholders. We will purchase stock in those businesses only when priced substantially below our estimate of intrinsic value. After purchase, we patiently wait for the gap between stock price and intrinsic value to close."

So the analyst's job is to find stocks that are cheap and well-managed, with growing per-share values. Analysts can look in whatever industries they want to find those stocks. By requiring such discipline for the type of stocks we will buy, we believe we can give the analysts a great deal of freedom and creativity for where they find such stocks. Further, we believe we attract better analysts to Oakmark than we would if we had a specialist structure.

We have a weekly meeting attended by the entire investment team. Reports of about four pages plus attachments are prepared a day ahead and distributed to everyone. Sometimes reports on new ideas are a little longer, while most updates on existing holdings are a little shorter. We don't use PowerPoint. Everyone at the meeting tries to identify flaws in the analyst's work, and additionally, one person is charged with constructing the bear case for any stock we consider investing in. At the end of the discussion, our three most experienced investment professionals vote as to whether or not the stock meets our criteria. If two or three say yes, the stock is added to our approved list and available for portfolio managers to purchase. Research is centrally stored and accessible to all investment professionals. I describe our process as being much like sibling rivalry: We can bitterly fight with each other, and then after it is over, enjoy going out to lunch together. For this process to work, attacks have to be made on ideas, not on individuals.

John Rotonti: Do you meet with management?

Bill Nygren: Yes, we meet with almost every management we invest with.

When I joined Harris Associates (the advisor to the Oakmark Funds) in 1983, my boss asked me to go to a due diligence meeting for an IPO and report back. Upon returning, I told him the CEO was extremely impressive. He leaned back in his chair, took a big puff of his cigar, and then said, "I don't want you giving your opinion on a CEO until you've seen 100 of them. Only then can you start to tell the top from the bottom decile. Of course they are impressive, how do you think they get to be CEOs?"

Though I didn't appreciate his comment at the time, the more years that have passed, the more I see the wisdom in what he said. It is too easy to complete all your work on a company, then meet management and convince yourself they are exceptional. Rating managements effectively requires just as much rigor as does a valuation model.

We have specific goals of what we want to learn from management meetings. As long-term holders, our questions are never about next quarter or the outlook for the year. We are trying to learn how they think, what their long-term goals are, how they are incentivized, and how they will judge their own success or failure. It is a very different conversation than we would have if we were trying to refine our earnings estimate. Think about the questions you would ask someone you didn't know personally who wanted you to become their business partner — that is the direction our conversations normally go.

Just like our valuation models aren't always perfect, neither are our qualitative assessments of managements. But if you judge our success against a goal of being directionally right more often than not, I believe getting to know management teams has been very additive to our process.

John Rotonti: Do you have any performance metrics that you prefer management compensation be based on?

Bill Nygren: We have one objective when looking at management compensation: We want to believe that they will maximize their personal economics by maximizing the long-term return on the stock. For that reason, we prefer performance metrics that drive fundamental value and that they be measured on a per-share basis. It is easy to make a company bigger by diluting the shareholders. Every management team says they want to maximize shareholder value; the elite management teams maximize per-share value.

John Rotonti: Are there any industries you tend to prefer? Or avoid?

Bill Nygren: Part of being a value investor is that very few investments are defined as off-limits. Instead, almost everything is a trade-off between price and quality. We won't invest in a business if we don't believe per-share value growth plus dividend yield can match the same for the S&P 500. To do so makes time your enemy. But that eliminates fewer companies than you might think because even stagnant or slowly declining businesses can be good cash-generators and decapitalize rapidly enough to meet our goal. We also won't invest with a management team that is content destroying per-share value.

Some less exciting industries rarely get priced at levels we find attractive — like utilities. But other industries that share commodity-like negatives are frequently priced low enough to merit our interest. Industries with strong growth and exciting new products typically attract enough interest from investors that we won't think they are cheap. But over the years, industries go in and out of favor. Back in the late 1990s, people thought Oakmark would never own tech stocks because we had deemed them too expensive for so long. Today, tech is one of our largest commitments. I think it is important for value investors to remember that the term "value" is not synonymous with "below-average." Companies don't get tagged as "value" companies. Value is never independent of price. Based on price, sometimes a company represents value and sometimes it doesn't. A value investor has to look at a wide array of possible investments and be comfortable saying "no" most of the time.

John Rotonti: When we experience a broad market sell-off (say 20% or 30% or more), what are you looking to buy first? Do you buy more of your core holdings? Do you look to purchase shares of businesses that may have been on your watch list but the valuation never made sense? Do you go first to the companies offering the largest margin of safety?

Bill Nygren: People who own the Oakmark Funds expect us to be reasonably fully invested most of the time. They are hiring us to find stocks that are better than the average stock, not to guess future stock market direction. In an ideal world, our investors would respond to a 20% correction by giving us more money to capitalize on the price decline. But unfortunately, very few investors behave that way. The result is that when the market corrects, we rarely have much new cash to invest. So our process is the same whether stocks are down 20% or up 20%. We want our assets invested in the stocks that have the best risk-adjusted expected returns. We are always looking for opportunities to sell stocks at or near our estimate of value and reinvest those funds in stocks selling well below our value estimate. Large market moves often create more pricing variations one stock to the next, so we tend to be more active after bigger market moves.

In a June 2015 email exchange, Rotonti asked how Nygren calculates the discount rate that his team uses for DCF valuation models. Here was his answer:

Bill Nygren: So the first challenge is identifying the right benchmark. In finance class, we all learned that stocks are very long-duration assets, so one would be inclined to consider a long-term bond as the appropriate risk-free benchmark. But the real (as opposed to nominal) return on a long bond is highly exposed to the risk that inflation differs from current expectations. Though businesses don't perfectly pass through inflation, they do at least lessen the risk of high inflation by having the ability to increase prices. So we conclude that rather than comparing an equity to a long-term bond, the more appropriate comparison is an intermediate-term bond, and we use a seven-year maturity. Historically, investors have earned a little more than inflation on a seven-year government, so we set a floor for that benchmark of 100 basis points above inflation expectations. If the bond drops below that yield, we would not lower our discount rate.

Since we are investing in corporations, our next step is to examine the yield premium for A-rated industrials. We add that historical average premium to the Treasury yield (or if higher, our floor yield). If the current A vs Treasury premium is higher than the historical average, we use that premium instead. (That prevents the possibility of having an equity-required return that is less than a current bond yield – not even close to an issue today, but was an issue six years ago during the credit crunch.) Finally, we add the historical average premium that equities have achieved relative to A-rated bonds. That becomes our discount rate for an average equity. Then we adjust from -100bp to +500 bp for the relative riskiness of a specific stock versus the market. (The main reason it isn't symmetrical is leverage.)

In ballpark terms, inflation expectations today are about 2%; adding 100bp premium puts our seven-year government floor at 3%. Since the current yield on a seven-year is less than 3%, we use the floor. To that 3% we add the historical average A-rated bond premium, and then the historical average premium equities have achieved versus A-rated bonds. All in, that gives us about a 7% discount rate for the S&P 500. (Note that since the S&P has yielded a little over 2% and we believe it can grow earnings 5% to 6% including share repurchase, for a long-term total return of 7% to 8%, prices today appear to be at or slightly below a fair-value level.) So our DCF discount rates today range from 6% for what we believe are the least-risky equities up to 12% for the most risky.

Having said all that, our preferred model for estimating intrinsic value is one that uses recent acquisition multiples for comparable businesses, and our DCFs are used just as a sanity check on what we believe is a more reliable model.