articles : http://www.entrepreneur.com/tradejournals/article/132511164.html

http://www.freerepublic.com/focus/f-news/1828706/posts

According to the article, at the end of 2001, the price of real estate, especially that of apartment, has been skyrocketed in some metropolitan area of Seoul, South Korea. That area is called Gangnam district where many private institutes gathered. At that time, many people and some media said that students in Gangnam district were accepted to top colleges in Korea by far more than those living in other districts or cities. Although the information is not officially approved, many people thought that if they moved to Gangnam district, their children would be highly likely to go top colleges in Korea.

Since then, many people tried to move to Gangnam district. Because the supply was limited, the price of apartment in Gangnam had been raised much more than that of other districts. Therefore, the price of apartment was exactly tripled in about two years. This led people to fall in another information cascades that if they bought apartments in Gangnam district, not only would their children go to top colleges, but also they would make a lot more fortune due to increasing price of apartment. People didn’t actually confirm what they heard was true, and even people who didn’t have much fortune borrowed money from bank to buy apartment in Gangnam.

This trend had continued two more years, but in August of 2005, as mentioned in the article, president Roh announced a new policy to stop the price increase of apartments in Gangnam. Since then, the price has not been raised at all. Actually, the price dropped a little. This leads people, who bought apartment between the end of 2003 and 2005, got into a huge debt. And even worse, if the real estate bubble in Gangnam district would end, their debt would become even more. This kind of problem happens everyday in the world, because of huge amount of uncertain information people get. To prevent this, people should make their decisions not based on what they heard or on uncertain information, but on the certain information they get by themselves or that from trusted resources.

In class, we discussed information cascade in terms of economic behavior of the mass. An information cascade is a situation in which every subsequent actor, based on the observations of others, makes the same choice independent of his/her private signal. This may result in erroneous mass behavior. Information cascade is similar to herding behavior of the crowds. Benign herding behaviors may be frequent in everyday decisions based on learning from the information of others.

For example, when a person decides which of two restaurants to dine in, the person may choose restaurant A. If this was earlier in the day, both restaurants may be empty, so the choice in the restaurants is at random. Later, when other people pass by in search of a place to eat, they will see that restaurant A has customers while B is empty. Most likely, they will choose A on the assumption that the restaurant having customers already makes it the better choice. Economically speaking, restaurant A will get more business that night than restaurant B.

Lisa Anderson and Charles Holt set up an experiment to prove this theory. Two marbles of one color, blue, will be placed in one of the cups (cup A) along with one marble of the other color, red. The other cup (cup B) will contain two red and one blue. Random cup is chosen for all the subjects. The subjects will pick a marble out of the cup, and guess which of the cup it is. After the first subject, the marble is replaced, and the second subject will pick a marble. It is found that after the first couple subjects have conformed prediction, subjects tend to conform to this prediction regardless of the color of their marble.

This is similar to what is going on with mp3 player market. Once the iPods became popular, and became a “conformity,” more people tend to buy the iPods, following the trend.

http://web.centre.edu/econed/informat.htm

We’re starting to cover in class the topic of network effects, where goods consumed have positive externalities (e.g. the more people have cellphones, the more utility the people buying them will enjoy). Of course, negative externalities exist as well: imagine a river, a factory upstream and a farm downstream. In this case, the water of the river acts like a good that the factory uses by dumping its industrial waste into, and the farm uses the water to irrigate its plants. One can clearly see that the utility of the farmer is affected negatively by the factory, since polluted water isn’t exactly condusive to watering plants.  In this case, it seems somewhat fair that the factory pays some sort of compensation to the farmer for reducing his utility, and using that intuition, we can examine a theorem that’s used very often in modern law in dealing with matters of negative externalities: Coase Theorem.

A result from classical economics shows that goods with positive externalities are underproduced, and goods with negative externalities are overproduced. In this case, underproduced and overproduced are measured from the standpoint of society. When we say the good is underproduced, we mean that the net benefit to society (measured as the sum of utility of members of the society)  from producing one more good is greater than the cost of producing that good, and conversely for the case of overproduced goods. An intuitive way of looking at this is as follows: In the case of cellphones, if I buy one, than the utility my friends get from having their cellphones increases since they can contact me. Yet the amount they benefit doesn’t factor into the decision of whether I should buy one - the only thing that matters is weighing the cost of the phone versus the amount of utility I get from buying the phone. In this sense, even though I’m increasing the utility of society as a whole, my own increase in utility is less than the overall increase in society’s utility.  This means that there will exist people for which even though their own utility gained from having a phone is less than the cost of buying it, the benefit to society is at least as great as the cost of the phone : This is an outcome where society as a whole would have benefitted if the person bought the phone, but the result is that the phone is not bought, hence the phone is underproduced. The arguement works exactly in the same way for goods with negative externalities.

From that, one can see that the problem arises because the benefits and costs to society are not the same as the benefits and costs to the individual consuming the good. If we had some way to make the two the same, then the problem of overproducing and underproducing would disappear.

In 1960,  economist Robert Coase published his paper “The Problem of Social Cost”, and was awarded the Nobel Prize in 1991 for it, along with his  other work on transaction costs. What Coase basically proposes is that property rights be established as a way of accounting for this negative externality. Setting aside the perhaps impractical nature of this for the moment, suppose we somehow decided that ownership of the river could be assigned totally to either the factory or the farm. Suppose further, for the sake of illustration, that we could quantify a) the amount of utility the factory gets from dumping its waste into the river, and b) the amount of utilty lost by the farm as a result of the factory dumping the waste,

Now, say the ownership of the river was assigned to the farm. If the factory gains $100 in utility from dumping while the farm loses $50 in utility as a result, one can intuitively see that this would result in the factory wanting to pay the farm some amount between $50-$100  for permission for the factory to continue dumping waste into the river. Conversely, assume that the factory gains $50 in utility from dumping while the farm loses $100 in utilty. In this case, the farm would exercise its ownership of the river to stop the factory from dumping its waste into the river, and since the factory gains less from dumping than the farmer does from the factory not dumping, the factory would not be able to rationally compensate an amount to the farm to allow the factory to keep dumping.

This also works in the reverse direction: Assume the ownership of the river was assigned to the factory. If the factory gains $100 in utility from dumping while the farm loses $50 as a result, the factory gains more from dumping than the farm gains from not dumping, hence the dumping will continue. If the factory gains $50 in utility from dumping while the farm loses $100 utility as a result, one can see that the farmer would want to pay the factory some amount between $50-$100 to make it stop dumping.

The interesting result from this is that, regardless of who the property rights of the river is assigned to, the result is that the actions that cause a greater net benefit to society ( measured as the sum of the utility of the farm and the factory)  are allowed to occur. That is to say, if the factory gains more from dumping than the farm does from not dumping, then as a result of the property right established, dumping will occur. The reverse is also true: if the farm gains more from not dumping than the factory does from dumping, than dumping will not occur. Both these results are independent of whoever the river’s property rights get assigned to.

Of course, that is not to say that the Theorem is without its shortcomings. For one thing, notice that we assumed that there were no bargaining costs - it doesn’t cost anything for the factory or farm to negotiate with each other. This is not entirely realistic since we expect most of this bargaining to occur in real life in the presence of lawyers representing the factory and farm, and  who don’t work for free. For another, as we established earlier, the concept of establishing property rights may not make sense entirely either: imagine if the factory dumped its waste into the air instead. Establishing property rights for air would seem like a much less viable solution.

For further reading: http://en.wikipedia.org/wiki/Coase_theorem

As a company built on search, it comes as no surprise that Google is heavily dependent on keyword-based advertising. A large chunk of Google’s revenue is generated from clicks on its search ads. Recently, a report from researcher comScore showed that Google’s search-related ads are experiencing a decline in the number of clicks. The research firm said that “clicks on ads declined 3% in February… and were up just 3% compared to last year.” In the wake of this news, investors are wary of buying additional Google shares. The day after the report was issued, the stock price of Google declined 3%. In response to this report, Google officials are pushing forward with the idea that fewer clicks can actually lead to more revenue for Google. By improving the “quality” of clicks, Google hopes to offset the decrease in volume by increasing the bid price per click. The “quality” of clicks can be improved by reducing the number of clicks that don’t lead to revenue for advertisers. Basically, Google wants to generate more with less, which is always a good strategy.

To improve the quality of clicks, Google has reduced the clickable area around an ad to reduce the number of accidental clicks. In January, it also introduced a “conversion optimizer” for its AdWords advertising platform to “enable its ad customers to buy ads based on the number of sales those ads generate.”

As learned in class, “ad quality” is a very ambiguous subject. By quantifying the effects of quality, Google is once again advancing the business of search. Keyword-based advertising is much more complex than the simple model that we work with in this class, and more investigation into this topic is crucial for a full understanding of the rapidly expanding internet advertising industry. Currently, the search advertisement agreement between Google and its ad customers is very inefficient. The customers pay Google per click, and this money goes to waste when people click without actually purchasing from the advertiser. However, if Google can continue to find ways to improve the chance that a person clicking on the ad will actually purchase from the advertiser, it can charge a higher price per click, thus generating more revenue. At the same time, excess clicks that do not generate sales will be reduced, which benefits the advertisers.

“Google’s Gamble” http://www.businessweek.com/technology/content/mar2008/tc20080328_837834.htm

A very interesting question was asked in a previous lecture about whether or not people can see past choices. According to the results of an experimental study based on the Bikhchandani, Hirshleifer and Welch (BHW) model, many individuals do not in fact recognize cascade behaviors of others. The BHW model shows that in decisive situations where there is incomplete information, it may be rational to follow predecessors and to ignore our own private information.

In this experiment, a prediction game was run where one marble was selected out of 2 urns, and the subjects were asked to set prices on these games to maximize their income. Each participant was randomly placed in one of seven cascade positions. The subjects were able to see the decisions, pro or contra, of their predecessors, who were actually artificial agents of the experiment. The positions were divided up into three types: from position -3 to -1 no cascade has started, position 0 was where the cascade starts, and positions 1 to 3 are within the cascade. The focus of this experiment was on the price setting behaviors of the subjects, and how it was affected by their cascade positions.

The data collected showed that, for a majority of the subjects, as they are placed further in the cascade chain, whether they receive a high or low signal from their predecessors, their tendency to veer away from the choice established by the cascade decreases. And the longer the cascade, the more likely the subjects is to set a higher price. This price setting behavior shows that cascade behavior is not really recognized.

Although these cascade behaviors are rational, especially in the case like this experiment where people are trying to maximize their own personal benefit, they can be harmful especially since it has been demonstrated that people don’t often realize the presence of these cascades. If we do not realize the determinant of our decisions are the choices of others rather than what we feel is right for ourselves, then what we might perceive as the correct choice might do more bad than good.

Robert Shiller recently made an application about the incompleteness of information cascades and its dangerous attributes to the recent housing bubble. But what if the information that cascades is more opinionated than informational? An example of this is the critic industry. How many of us have seen a movie or eaten at a restaurant at the recommendation of a professional critic? More importantly, how many of us realized how much we hated the movie and were disgusted by the same food that this professional critic had upheld in such high regard? Each person develops their own opinions and tastes, but the influence of a published critic may affect enough to try otherwise.

There is a small population of people who have built their careers on forming their opinions. Movies are a good example of how people will make the decision to see a movie as a result of a good review only to be severely disappointed afterwards. It is interesting that people will trust complete strangers to have the same taste as they do and what is more interesting is that most people will still be affected by the opinion of that same critic later on. The truth is that it is hard to ignore another person’s opinion. It is almost impossible to make an independent decision.

Of course, there are famous critics who are famous for a reason: their opinions and evaluations are consistent and agreed on by many people who followed their recommendations. Critics eventually build a reputation and their information cascade may disappear because their fans may just trust the critic’s to make their decisions for them. Robert Parker is a renowned wine critic whose rating system can mean instant celebrity for producers overnight.                        

There are ways people try to resist this information cascade by creating sites like imdb.com. These sites allow people to sound off their opinions and for people to decide based on individual opinions as well as a “popular” opinion. Imdb.com even creates an aggregate user rating curve.

Shiller, Robert. “How a bubble stayed under the radar”, New York Times. March 2, 2008.

Gross, Jeffrey. “Don’t listen to critics”, The217. February 26, 2008.

http://www.ohio.com/editorial/commentary/16246947.html?page=1&c=y

The Akron Beacon Journal discusses the recent housing bubble and the failure of economic experts to recognize the problem. They quote Alan Greenspan’s autobiography, saying, “I’d come to realize that we’d never be able to identify irrational exuberance with certainty, much less act on it, until after the fact”. The article introduces the concept of information cascades as the major reason for the housing bubble and tech bubble of the laste 1990’s. A buyer’s willingness to pay a high price for a house will influence later buyers into (perhaps irrationally) thinking that similar houses have a high value as well.

The reasoning presented in this article is very similar to the model for information cascades that was presented in class. It seems like quite a strong statement to say that we’ll never be able to identify irrational exuberance. When conducting analysis on the housing or stock markets, perhaps we should always assume there is a level of irrational exuberance caused by information cascades. A model could be developed to determine how much the apparent market figures need to be modified in order to counteract the effects of information cascades. In such a large economy, the prevelance of information cascades (presented as a discrete phenomenon in class) would average out to some computable value. It seems like this sort of reasoning would be very beneficial in recognizing bubbles before they burst.

According to class material, Web search was initially supposed to only help users find relevant and useful pages. However, due to the exponential growth of the Web, it is impossible to prevent search engines from turning into profitable money-making machines. For search sites like Google, this industry makes up most of its revenue.

To increase effectiveness, search engines use key-word based advertising; advertisers pay for results based on certain queries specific to their businesses. These ads are based on a cost-per-click model, where advertisers only pay when users click on their ads. Prices can range from $0.60 per click to over $50. These costs are determined by an auction procedure called Generalized Second Price auction, since sites have multiple slots for paid ads but these are valued at different prices.

Since this business is so lucrative and competitive, there will naturally be those who try to take advantage. These fraudulent clicks, which increase a Web publisher’s advertising revenue and “drain” a competing advertiser’s budget, are becoming more frequent. Interestingly, the larger search engines rebuff data that about these clicks. The following article describes the new partnership between one of the major search engines, Yahoo!, and a click-fraud auditing firm, Click Forensics. There have been tensions between Click Forensics and other search engines due to accusations that these search engines underestimate the frequency of fraudulent clicks. According to the firm, the overall click-fraud rate for the online advertising industry in the last quarter of 2007 was 16.6%, up from 14.2% a year ago. In terms of contextual advertising, Click Forensics estimated the volume to be 28.3%, a sharp increase from 10.2% in 2006.

Part of this increase comes from botnets, which are PCs that are infected with software. These number in the millions and are programmed to flood certain sites with clicks. The leader in search and advertising, Google, has repeatedly rejected these claims. Google already offers free clicks for the 10% of their clicks that are “questionable”, and state that fraudulent clicks number far less than that.

There is obviously a conflict in information because Google claims that Click Forensics has access to only 1% of their data, certainly not enough to make accurate reports, while Click Forensics believes their network of more than 4000 advertisers validates their information. Withholding of data between advertisers and search sites is the norm. Larger search engines like Google and Yahoo! avoid sharing this information with advertisers directly, because this would only lead to more fraud while advertisers want to prevent compromising their users’ privacy as well as high advertising rates. Nonetheless, Yahoo! believes their partnership will help reduce their own number of fraudulent clicks, which they estimate to be 12-15%.

http://www.forbes.com/technology/2008/03/17/click-fraud-yahoo-tech-security-cx_ag_0317click.html?boxes=relstories

“’Q. Why would I want to join an inexplicable mob?

A. Tons of other people are doing it.’”

–Bill Wasik, “My Crowd” Harpers Magazine. Mar 2006, Vol. 312 Issue 1870, p56-66.

A monument both to the potential vacuousness of “information” cascades and the way individuals and other actors learn to coopt the social environment surrounding them, the Flash Mob fad of 2003 to early 2006 was once the next big thing. Beginning in the summer of 2003, New York seemed to be full of flash mobs, inexplicable semi-impromptu gatherings of hip urbanites who would, anonymously, collectively, stage largely meaningless spectacles and then disperse. Naturally, they spread and this is turned triggered massive, largely empty media attention, aggrandizing what amounted to a uniquely empty trend into a much-scrutinized catalyst for cultural or political organization. The hype was unbelievable. Then in March 2006, Harpers editor Bill Wasik published “My Crowd” and revealed the truth. Two main facts emerged over the Harpers piece, which detailed the origins and implications of the phenomena: One, Wasik was in fact the until-then-semi-anonymous “Bill” who created the “flash mob” back in 2003, and secondly, flash mobs were even more interesting than the hype had supposed, the main component of Wasik’s Milgramesque sociological experiment on and parody of conformity among the young, the hip, and the urban.

Flash Mobs were in a sense an information cascade with truly no information to impart, pure distillations of the herd instinct. Like Milgram’s conformity experiments–which Wasik cites as an inspiration–there was really no conceivable reason to join in beyond social pressure, something Wasik was quite upfront about with his “subjects.” As the one-question FAQ he would circulate with his other announcements put it, the one reason to participate was that “Tons of other people are doing it” (Wasik). As Wasik writes, “Not only was the flash mob a vacuous fad; it was, in its very form (pointless aggregation and then dispersal), intended as a metaphor for the hollow hipster culture that spawned it” (Wasik). If we like to tell ourselves that information cascades really do provide a useful signal, Wasik’s “experiment” challenges such assumptions by laying bear the kernel at the heart of such behavior. We may tell ourselves that following the crowd is smart, and oftentimes it can be, but in our hearts following the crowd is often its own payoff.

Also interesting were two other groups’ responses to the flash mob fad: the media and corporations like Ford. The first quite naturally leap on the storm, creating a proper information cascade of information as bloggers and journalists all reported on it. Even more interesting was the way companies like Ford Motors tried to capitalize on the fad, hosting “Fusion Flash Concerts” to promote its new automobiles. Information cascades do not convey information after all, they only really produce it–What a given individual learns is that others seem to prefer a given choice, not the reasons behind those choices, so that every individual is in fact making decisions based on different information–so it’s only natural for actors to try to attatch their own information to the transmission.

Works Cited

Bill Wasik, “My Crowd” Harpers Magazine. Mar 2006, Vol. 312 Issue 1870, p56-66.

http://www.harpers.org/archive/2006/03/0080963

http://www.nytimes.com/2008/03/02/business/02view.html?_r=1&oref=slogin This article is about some of the reasons the experts did not detect the housing bubble before it happened.  Much of what happened was because many of the top experts, including Alan Greenspan, thought the housing bubble was not going to be a bubble, but a ”froth”.  This is a term used by Greenspan to explain that it would not be a full bubble, but a series of small bubbles that he thought would never grow to a scale that would effect the whole economy.  The failure of recognizing this housing bubble is the cause of the failing United States Financial Institutions (many other reasons, of course).  This is because average investors did not see perceived risk, and trusted the expert’s views.  They saw the chance to profit and did not think enough about risk.   In this article it explains,”Three economists, Sushil Bikhchandani, David Hirshleifer and Ivo Welch, in a classic 1992 article, defined what they call “information cascades” that can lead people into serious error. They found that these cascades can affect even perfectly rational people and cause bubblelike phenomena. Why? Ultimately, people sometimes need to rely on the judgment of others, and therein lies the problem. The theory provides a framework for understanding the real estate turbulence we are now observing.”   Basically the problem was too many experts underestimated the severity of the situation and told people everything was fine, which increased people’s desires to invest thinking of profit not risk, which put the wrong pressures on the market, eventually leading to a seriously failing market and many foreclosures on peoples homes.  In other words, the experts had incomplete information that caused people to have a fake sense of confidence causing them to enter the housing market and bid up prices, which resulted in a bubble that is hurting the whole economy. This is how information cascade can cause huge problems, when people don’t make their own decision and rely too heavily on the advice of so called “experts”.