As a CIO, you\u2019re familiar with the following scenario:You\u2019ve sunk $900,000 into a forecasting system that\u2019s expected to cost $1 million but has yet to deliver on its promises. In hindsight, you\u2019ve realized that all that money would have been better spent on a sales-force automation (SFA) system. Nevertheless, you\u2019ve still got $100,000 in your coffers, so instead of scrapping the forecasting system and starting anew with SFA tools, you sink the remainder of your money into the forecasting system and hope for the best.A foolish mistake? Perhaps, but not at all unusual, according to Dan Ariely, the Luis Alvarez Renta professor of behavioral economics at MIT\u2019s Sloan School of Management and director of the MIT Media Lab\u2019s e-rationality research group. As a behavioral economist, Ariely studies how people make decisions in real life and why their decisions often deviate from classical economic models, which assume that people act rationally and in their own best interests.Ariely, 36, takes a self-described "armchair" approach to that discipline."I look at how I behave. When I find something curious or bizarre about myself, I tend to look at it in more detail," he says.For example, Ariely\u2019s own tendency to procrastinate led him and a colleague to study different measures for overcoming procrastination. They found that people are willing to set their own deadlines so that they don\u2019t wait until the last minute. Yet while self-imposed deadlines help people improve their performance, they are not as effective as deadlines set by others.Ariely, with his self-examining style, is one of a handful of academics challenging the intellectual foundation of economics. For years, classical economists maintained that individuals make decisions based on their own self-interests. Ariely\u2019s research on procrastination, how people value goods, how they perceive pain and the effects of female physical beauty on the male brain reveals the exact opposite: Individuals make irrational decisions that are not in their best interests.\nWhat may interest CIOs most are Ariely\u2019s investigations into how people value goods and how people\u2019s experiences of physical and psychological pain affect their decisions. CIO Senior Writer Meridith Levinson caught up with Ariely at his home in Cambridge, Mass., where he explained his research, discussed it in the context of the irrational decisions CIOs make and shared his thoughts on what CIOs can do to make more rational financial decisions.\n\n \n\n\n\n\nCIO: Your research shows that people, including CIOs, don\u2019t know how to value or set prices for various goods. How did you come to that conclusion?\n \n\nDan Ariely: We [Ariely and fellow researchers Drazen Prelec and George Loewenstein] conducted a second-price auction in class. We sold MBA students keyboards, [computer] mice, bottles of wine, DVDs, books and chocolate. We explained the procedure for a second-price auction where the highest bidder pays the second highest bidder\u2019s price. We described each of the products and said to the students, "Before you tell us your bids, please tell us the last two digits of your Social Security numbers and translate those last two digits into a dollar figure." So if your last two digits are 44, that\u2019s $44. Then we asked them to tell us whether they would pay the amount of money indicated by the last two digits of their Social Security numbers for the various products.\n\n \n\n\n\n\nWhat did you find?\n \n\nThe people with the highest Social Security numbers, in most cases, bid 100 percent more than people with the lowest Social Security numbers.\n\n \n\n\n\n\nIn that experiment, you were also testing a concept called coherent arbitrariness. Can you explain what that is?\n \n\nIt\u2019s the idea that people do not know how much to pay for anything. They will rely on their own arbitrary judgments to generate a value, and the value they generate will be coherent in the sense that it will be based on the value of a similar item. For example, when we sold a large box of chocolates and a small box of chocolates everybody said, "We\u2019ll pay $X more for the big box than the small box." Once the price of a product [in this case, the small box of chocolates] was fixed, the prices of similar products were set in a relative manner.\n\n \n\n\n\n\nHow does that research apply to prices that CIOs pay for IT?\n \n\nImagine that an organization is thinking of creating a database. Why does it care what Oracle thinks [the database] should cost? Theoretically, the company should be able to say, "How good is this [database] for us?" Often it relies on prices that it or others have paid for similar products in the past. "That database was $2 million. This database is twice as big so we should be able to pay at least $4 million." It\u2019s hard to figure out the value of an IT investment. If what you\u2019re willing to pay is a function of what you paid before, that\u2019s a problem because it means you don\u2019t know what it\u2019s worth.\n\n \n\n\n\n\nIs hardware an exception to what you\u2019ve found, since its price decreases every year?\n \n\nThe fact that the cost of hardware decreases every year is an additional factor people take into consideration when estimating a price. They still set prices in a relative manner. They begin with the starting level price?what they paid last year?then they adjust it based on factors that seem relevant. In the case of hardware, you\u2019re expecting prices to go down.\n\n \n\n\n\n\nHow should CIOs evaluate the value of a database or any other IT investment?\n \n\nThey should not compare databases to databases. They should compare the database with all possible IT investments. People have a tendency to make decisions in silos. We think about whether or not to invest in a new database or to allocate salaries separately. It\u2019s better to think across categories to find out where the best value is. Should we update software? Hardware? Move to a new operating system?\n\n \n\n\n\n\nCan you explain why people rely on those apples-to-apples comparisons to determine or rationalize what they\u2019re spending?\n \n\nIt\u2019s so simple. It\u2019s seductive. It appears rational.\nEvaluating these things is hard. I don\u2019t want to say people are stupid. Go back to the example I gave you with the box of chocolates. How do you evaluate the value of a piece of chocolate? You know what its sweet taste and soft melting texture is like, but how you translate that into money is quite complex. And chocolate is supposed to be easy. Theoretically, you do this all the time. So how do you deal with a database? How do you take all of its complexities into account? People use these heuristics, these shortcuts, these relative comparisons that seem to give us the reason to make choices. A lot of times you don\u2019t compute; you just look for reasons to do one thing over another. Those are reason-based choices.\n\n \n\n\n\n\nA lot of companies are holding back on IT investments because of the weak economy. Are they acting irrationally?\n \n\nIt\u2019s hard to say for sure. It\u2019s a question of how you look at your portfolio of decisions. Do you look at the decisions one at a time or in total? For example, if I asked you to play a game where I flip a coin and if it lands heads, I give you $140. But if it lands tails, you give me $100. If I asked you, "Would you play this game once," you\u2019d say no. If I asked you, "Would you play this game 100 times," you\u2019d say yes. Why should things be different if you play something one time or 100 times? This is something called Samuelson\u2019s Paradox. When you focus on one decision and realize the downside, you focus on the downside and don\u2019t take the risk. You\u2019re risk averse. If an organization thinks of 10 decisions separately, it might not take enough risk. When you think about a lot of gambles, or when you look at all the decisions together, you take on much more risk because the fear of losing a lot on a single project is mitigated.\n\n \n\n\n\n\nIn essence, companies are being irrational if they are evaluating IT investments in a silo; whereas if they are evaluating them as a portfolio, they are being rational?\n \n\nIt\u2019s irrational when the outcome of their decisions is different when they evaluate their investments one at a time as opposed to all together. It\u2019s fine to look at investments separately if the outcome is the same as when you look at all of them together. The problem is, I suspect the outcome is not the same.\n\n \n\n\n\n\nIn light of your explanation, is taking a portfolio management approach to IT investments a good idea for CIOs?\n\n \n\n\n\n\nYes. When companies look at investments one at a time, they\u2019re more risk averse. Being risk averse can be a very bad thing for a company. (For more on portfolio management, see "Portfolio Management: How to Do It Right," Page 56.)\nRelated to being risk averse, CIOs also fall victim to making IT investment decisions based on money they\u2019ve spent in the past, a theory you refer to as the sunk cost. Can you explain what\u2019s irrational about it?\nImagine this scenario: You work in a company. You have a project. The project is established to cost $1 million. You\u2019ve already spent $900,000. The sunk cost is what you\u2019ve already spent. It\u2019s the idea of throwing good money after the bad. Then you find out there was a better project to pursue. Would you spend the next $100,000 to finish the project? Would your answer be different if you hadn\u2019t [already] spent $900,000? You made a decision. It didn\u2019t turn out as you wanted. If you were an economically rational person, what you\u2019ve spent in the past wouldn\u2019t influence your behavior in the present. It\u2019s the same if you invest in some legacy system and the payments will last five years, and after three years, you discover it\u2019s not really what you need anymore. Are you likely to switch if you\u2019re still making payments on the old project? It might be very painful to purchase the new one if you\u2019re still making payments on the old one.\n\n \n\n\n\n\nWhat should CIOs do when a more worthwhile project or investment comes along after they\u2019ve already spent a lot of money on something else?\n \n\nAgain, they should think about what\u2019s the best use of a dollar and ignore the past. I\u2019m saying it as if it\u2019s an easy thing to do. It\u2019s not.\n\n \n\n\n\n\nCIOs often hire consultants that tell them what they already know or what they want to hear. How does your research explain that seemingly irrational decision?\n \n\nIt\u2019s an issue of accountability. People don\u2019t want to make mistakes. They\u2019re afraid to be fired. They\u2019re afraid their bonuses will be affected. People are trying to cover their tracks. They hire consultants to tell them what they already think. It\u2019s like IBM used to say, "Nobody ever got fired for buying IBM."\nYou can also ask the question, are consultants really free to find what they want? It\u2019s an issue of conflict of interest. If I\u2019m being paid by you, I like you, and you want me to find X, am I really going to find that X is wrong? There are dependencies in these relationships.\n\n \n\n\n\n\nYou\u2019ve found in your research that pain and people\u2019s experience of pain, whether physical or psychological, often cause them to make knee-jerk decisions. Could you elaborate on the experiment you did in this area?\n \n\nWe put someone in a wet suit with tubes running through it that carry cold and warm water. We ran cold water through the wet suit, cooling him down to 46 degrees Fahrenheit. That\u2019s very cold. We tell him that we\u2019re starting the experiment, and we change the water temperature. So he started off being very cold, now the water becomes wonderfully warm. Then we cool the water off again. Then we stop changing the water temperature and ask him, "How painful was this?"\n\n \n\n\n\n\nWhat did you find?\n \n\nThat people\u2019s perception of pain depended on the pattern of pain.\n\n \n\n\n\n\nWhat are the implications of that experiment, and how do they tie in with the research you\u2019ve done on irrationality?\n \n\nSay you use an ISP that improves its reliability from 80 percent to 90 percent. That ISP\u2019s reliability is perceived as better than an ISP whose reliability is consistent or started at 90 percent and went down to 80 percent. If things improve, we are happy.