The September 2006 issue of the Harvard Business Review has a brief piece by Columbia sociology professor Duncan Watts and Steve Hasker of McKinsey about marketing in environments where social influence is important. Watts and Hasker argue that when a consumer’s interest in a given product is driven by how popular the product seems to be with others in the consumer’s social network, predicting whether a product will be a success or not becomes very difficult. To cope effectively with this uncertainty, marketers should spend less time and money trying to predict big-budget blockbusters, and instead develop “portfolios” of products, and the ability to rapidly shift marketing resources to emerging successes based on customer feedback.

The implication for marketing executives is that they should de-emphasize designing, making, and selling would-be hits and focus instead on creating portfolios of products that can be marketed using real-time measurement of and rapid response to consumer feedback.

The aurhors recommend five measures for more effective marketing campaigns which take social network effects into account:

1. Increase the number of bets, and decrease their size
2. Focus on detection, measurement, and feedback
3. Follow through with flexible marketing budgets
4. Exploit naturally emerging social influence
5. Build flexibility into supply chains and contracts

Their results are based on academic work published earlier this year by Watts as well as Matthew Salganik and Peter Dodds: Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market

Read more: Marketing in an Unpredictable World, by Duncan J. Watts and Steve Hasker

UPDATE: I’ve heard from some readers that the Harvard Business Review link didn’t work for them. Here is an alternate link to the paper, hosted by Columbia: Marketing in an Unpredictable World

UPDATE 2: For discussion of a very similar “portfolio” approach to dealing with complex or uncertain environments, this time in the context of business strategy, see my earlier post Strategy in an Unknowable Universe

Columbia professor Ray Fisman was interviewed by Hermes magazine about his research on speed dating. Some interesting observations relavent to behavioral decision making:

3. Does your study measure how well what people say they look for matches with what they actually look for?

Most of what I do is work on corruption in poor countries. If I want to know how much someone is paying in bribes, I’m not going to ask them, “How much did you pay in bribes last year?” I’m going to say, “The guy down the street from you, who looks pretty much like you, how much did he pay?” Similarly, in the speed-dating study we ask people, “What do you care about?” We also ask them, “The average man, what do you think he cares about?” But then we actually see how they behave in the game. And, not at all surprisingly, what they say the average man cares about lines up much more closely with what they actually reveal through their actions than what they claimed they cared about beforehand. In particular, everyone — both men and women — says they care less about physical attractiveness than the average.

4. Do you think speed-dating is more efficient than traditional search methods?

In some sense, it’s efficient: there are all these slice studies on how 10 seconds’ worth of observation is as predictive of your experience with a professor as a semester’s worth, and they’ve reduced it to 2 seconds and that’s just as good; and they’ve reduced it to just a photo and that’s pretty good, too. So you learn a lot in four minutes, perhaps as much in four minutes as you do in a much longer superficial interaction like, say, a date. So, this does meaningfully provide you with 20 rapid-fire dates, to the extent that we form as much of an impression in 4 minutes, or 10 seconds, as we do in 4 hours. The thing that’s left out of this neat decomposition of people into attributes, though, is actually learning to love someone. And that’s what I think is kind of missing. Focusing on people as a bundle of attributes almost makes people think about this decision in the wrong frame of mind.

5. Do you think people become unwilling to commit because of all the choices dating services enable?

Yes. And the way that you can make these choices — just the very fact that it’s set up in this way — distorts the way people choose. There was an article in the New York Times on a backlash against Internet dating, and I wonder to what degree that’s at least partly as a result of these sorts of realizations.

6. The results of your speed-dating studies, particularly with regard to intelligence and physical appearance, seem to reinforce gender stereotypes. Why do you think this is?

Well, they are stereotypes for a reason. However, it’s not as simple as, “I avoid all women who are ambitious or intelligent.” It’s about, “Intelligence and ambition is OK until it supersedes my own.” It’s also worth mentioning that these are average effects — there are surely men who do not have this property. I like to think I’m one of them: my significant other is definitely a lot smarter than I am. When her grandmother heard about me, she said, “I told your mother this, and now I’m going to tell you: never let a man think you’re smarter than he is. Men don’t like that.” Everyone laughed and thought this was so anachronistic, but it shows up in our data. Grandma’s views on dating aren’t so dated after all!

Read more: Dating Data: Economic Theory and the Search for a Mate

Much of the process of sound decision making rests on our ability to perform appropriate comparisons. Which is a better investment: Google or Yahoo? Which is safer: flying or driving? Which business school is best? Our answers to all of these questions hinge crucially on the basis we use for comparison. Which features are really salient, and which are just noise? Are we looking at a large, objective collection of evidence, or just the recent evidence we have at hand? Are we using our instincts, and predictions of the future, or looking at statistical data from the past? Are we focusing on the ways in which competing alternatives are similar, or the ways in which they differ? What is the relevant timeframe we’re analyzing? Do we care about absolute performance, or relative performance?

Our answers to each of these questions can radically change the outcome of a decision making process, for better or for worse. In his latest Mauboussin on Strategy article, Michael Mauboussin surveys the many behavioral factors that go into forming comparisons, and offers some advice for making comparisons which are appropriate to the situation.

Read more: Mauboussin on Strategy: How Do You Compare?

Previously:

• More Than You Know
• Mauboussin on Discounted Cash Flow Models
• Mauboussin on Strategy: Size Matters
• All posts tagged Michael Mauboussin

Michael Mauboussin speaks with Columbia’s Ideas at Work magazine about some of the ideas in his recent book, More Than You Know: Finding Financial Wisdom in Unconventional Places.

In the book’s conclusion you mention some of the things the experts still don’t understand about investing. Can you talk about the directions for future research?

If you look at the world of finance, there are many, many open questions. For example, we don’t really understand how capital markets get to efficiency. There are some theories that are widely used in the world of finance, including mean-variance and no-arbitrage assumptions. I suspect these traditional ideas will eventually be superseded by this idea of complex adaptive systems, or the wisdom of crowds.

I think that the recent developments in neuroscience and decision making are absolutely fantastic. Another area that is really intriguing are the statistical regularities, like the power laws, that have come out of the study of physical systems, like earthquakes. In biological science, we know things like body mass and metabolic rate also follow a power law, a scaling property, and we have ways to explain those phenomena reasonably well. We see many of those same power laws in social sciences, yet we really have no causal mechanisms. So we don’t know why city sizes follow a power law or why the sizes of corporations follow a power law.

The last idea I’d mention is the flight simulator for the mind. One of the challenging things about investing is it’s very difficult to get timely and clear-cut feedback. If you’re a handicapper at the racetrack or you’re a weather forecaster, you get feedback pretty immediately on the decisions that you make, and that helps you calibrate and improve your decision-making process. When you purchase or sell a stock, you really don’t know in a timely fashion whether that decision was a good or a bad one. So an interesting question is whether we could create some sort of artificial environment that allows people to get better feedback on their decisions.

Read more: Guppies, ants and golf swings: Mental models for investors

Previously:

• Mauboussin on Discounted Cash Flow Models
• Mauboussin on Strategy: Size Matters
• All posts tagged Michael Mauboussin

Last week I had the opporunity to see a presentation given by Dan Reingold, author of the recent book Confessions of a Wall Street Analyst: A True Story of Inside Information and Corruption in the Stock Market, which details his experiences as a telecom analyst for a number of the major investment banks on Wall Street. The talk primarily covered the question of analysts manipulating their recommendations in order to boost profits for their bank, and as a top analyst himself, Reingold was in a central position to watch (and participate in) the problems as they unfolded throughout the dot-com boom.

According to Mr. Reingold, there are three primary factors which work to compromise the objectivity of analysts’ ratings and recommendations. The first is peer pressure from their banking colleagues. Suppose an analyst has a coworker on the investment banking side who is working on a deal with Company XYZ to do an IPO for them which will generate substantial fees for the bank. That coworker is likely to place alot of pressure on the analyst not to make negative remarks about Company XYZ’s prospects publically, which would likely anger the potential client and threaten the deal and associated revenue. The second factors is pure self-interest or greed. According to Reingold, it is a common practice today for analysts to be offered compensation packages which explicitly include a percentage of the investment banking revenue in their target industry. If an analyst has this type of compensation package, they clearly have a strong financial incentive to skew their recommendations in the direction which will help generate the most revenue from the other side of the bank. Finally, Reingold suggested that simple human error, in combination with lax independence rules, works against a fair and objective analyst marketplace. When analysts are brought “over the wall” to consult on pending investment deals, they often are made privy to insider information that is valuable to potential investors. Reingold said that once an analyst knows certain inside information about specific companies within their target industry, it is very difficult to prevent that information from coloring his or her otherwise independent judgement, or even subconsciously leaking information to clients.

Reingold was also critical of the recent resolution of Attorney General Eliot Spitzer’s investigation into conflict of interest problems at the major investment banks. A few points of interest:

• Over a 4-5 year period, about 10 of the top banks made $80 BILLION in profits. The resolution calls for those banks to collectively pay $1.4 billion in fines. A fine which is only a tiny percentage of profits may send the message that crime pays and fines are a necessary cost of doing business on Wall Street.
• Jack Grubman, a high profile target of the investigation, was ordered to pay $15 million in fines personally. His severance package from Citigroup totaled $34 million. Those numbers don’t seem to provide a very strong personal incentive for avoiding conflicts of interest.

Reingold went on to offer his suggestions for a stronger set of reforms which would, in his view, do more to curb conflict of interest problems. I’d like to ask a somewhat different question — is it possible to estimate how big of a problem these cases represent, and to what extent Wall Street analysts are biased? Some statistical work has been done on measuring the significance, or accuracy, of analyst buy and sell recommendations, and it shows that sell ratings are significantly more informative than buy ratings. This gives some evidence for analyst bias, although there are other potential explanations for the data. Perhaps analysts are prone to irrational streaks of optimism, causing them to issue unwarranted buy ratings. I float that possibility somewhat in jest, but readers of this blog know that humans face many psychological and cognitive biases which can cause them to make errors in decision making, and even though analysts are well-paid professionals, they are not immune from these biases altogether. Perhaps a well-crafted statistical study can shed more light on the objectivity of individual analysts.

Previously: Analyst Recommendations and Insider Trading

UPDATE: In the comments, “bronxite” suggests another bias that could be effecting the skew in buy/sell ratings, which is that analysts have an endogenous preference to cover more exciting, growth-oriented companies. More detail can be found in the paper Do Security Analysts Speak in Two Tongues? by Ulrike Malmendier of Stanford and Devin Shanthikumar of Harvard. A related issue is what I think is an innate human distaste for naysaying. Most people would shy away from a position which required publically badmouthing other organizations day in and day out. We can see similar psychological factors at play in the persistent suspicion and hostility against short sellers in the market, who are portrayed as vultures preying on the misfortune of others.