While the subprime mortgage lending crisis has yet to be explained fully, as only time can tell the true causes and effects, a few simple models and an few intuitive observations can lend some valuable insight into why the crisis happened and how it cascaded. As reported here A guide to the subprime mortgage crisis Wall Street Journal or New York Times, the current lending crisis began when banks and other financial institutions extended loans, usually mortgages, to customers with sub-prime, or lower quality, credit ratings. These individuals had histories of late payments, defaults, lower wages, etc. Lenders loaned these riskier individuals money at lower rates on Adjustable Rate Mortgages (ARM’s) , or mortgages that had lower initial rates, but would reset to a market rate after a set period of time. In the past, these new rates were not significantly different than the initial rate because the market for these loans was generally stable, so lenders often pushed these loans over fixed rate loans to those who perhaps could not afford the slightly higher fixed rate. Some of these borrowers defaulted because clearly they were poor candidates for loans in the first place. The more people that defaulted, the less attractive the market for these loans, and the higher the interest rates reset to. Interest rates kept going up and more people were unable to make payments on their loans. Some people had the unfortunate experience of owing significantly more money than what their house was worth, in which case they would leave their house.

The most interesting part of this crisis is how it affected other areas of the credit / debt network. I made a simple network graph of banks, individuals, and government as nodes, and edges with the level of trust, or creditworthiness. Assume that all outward edges from a node have the same edge weight (creditworthiness). This number is displayed inside the node for ease. Creditworthiness is a score from 0 (no creditworthiness) to 100 (the most reliable). The direction of the arrows point toward where debt is owed. Before the credit crisis, the graph looked similar to the one pictured above. However, borrowers with low creditworthiness like the individual with creditworthiness=5, defaulted on their loans. When these people defaulted on their loans, rates increased for others, perhaps affecting those in the 5-10 group the most. When these rates went up, the 5-10 group defaulted, eliminating a total of half of the individual borrowers. The bank on the right who loaned two of these borrowers money has essentially lost a good chunk of money from defaulted loans. It now mandates that all individuals must have credit higher than 15, eliminating the individual with credit=12. This bank is hurting at this point, and the individual with credit = 21 might be inclined to withdraw his money from the bank, which is essentially a loan. This bank can only borrow money at this point from another bank, which is doubtful given its financial distress, or the government. In this situation, it can be argued that the government is the necessary stabilizing agent because it can easily print and lend more money at lower rates, within reason, to banks. Also, the government insures all loans made by individuals to banks up to a certain amount. These typically constitute bank accounts and CD accounts. Perhaps with this insurance, the individual with credit=21 might have kept his money in the right-hand bank. It can also be seen that as the situation progresses, less and less money is lent, which can lead to slowed economic activity due to less available credit. Businesses that may need a loan for improvements might have a problem attaining a loan. While the current crisis is much more complex, involving many more institutions and inputs, some deductions can be made through simple network modeling.

In class today, we started the discussion of the information network. It was mentioned that the world wide web is getting arbitrarily large. An important point to notice is the ways in which it is growing. One way that has become especially popular during recent years is the use of the web for personal services and for the storage of personal data. Today there are shopping services, social networking sites such as MySpace and Facebook, and document services such as Google Docs.

One of the more recent and controversial proposals is the storage of personal medical records on online services. Google [1] and Microsoft [2] are already planning products in the field, and there is a spirited debate about the appropriateness and usefulness of such services.

The two products mentioned above aim to make this  information accessible on the network. Google is developing a front-end for patients, who will be able to add their records and histories to their account and share them with qualified individuals, such as family and doctors. Further integration is planned with other forms of records, such as prescriptions. Google Health, as the product will be known, is planned as an extensible system which will include extensions from third party developers. Microsoft’s offering, HealthVault, includes a secure space to store health information as well as a health-related search engine. Microsoft’s focus, however, appears to be at encouraging outside development, as the software maker has provided $3 million prizes for developers working in the HealthVault framework [3].

On the one hand, people could greatly benefit from immediate access to pertinent medical information. In addition, if allowed to access such systems, doctors and health professionals might be better able to track and treat patients. On the other hand, privacy and security questions dominate the discussion. Can ordinary people trust large companies with their private data? Will the companies abuse or sell this data, or will it target advertisements based on it? What happens if a large system storing this information is hacked or otherwise compromised? Aside from privacy concerns, there are also some legal issues to consider. For example, the Health Insurance Portability and Accountability Act (HIPAA), legislated over a decade ago, normally covers the transport and movement of health information. But where do the new services fit?

At this point, there are more questions than answers regarding this new way of exchanging information. Please see the link below for some more details.

[1] http://www.news.com/8301-10784_3-9880909-7.html?tag=nefd.lede

[2] http://www.pcmag.com/article2/0,1895,2191920,00.asp

[3] http://www.news.com/8301-10784_3-9877656-7.html

Facebook has become one of the largest and most popular social networks among high school and college students. Currently, Facebook has many third party applications. This article talks about an online investment site, Cake Financial, launching a Facebook application that enable to share their portfolios of their actual brokerage accounts. This application allows users to share the trade they have made, and compare actual returns in percentage terms (dollar amounts and net worth are not revealed). Individual investors can also get some insight into how their portfolios have performed over time.

From the very start, Facebook was acting like edges that join individuals (nodes) from one university together.  As it started to open to more universities and high schools, it was becoming like a bridge that joins separate networks together. So now it has become a very large social network consist of mainly high school and college students. On the other hand, Cake Financial is an online social network for investing based upon people’s real portfolio holdings, actual performance and daily trades. Facebook and Cake Financial are two separate networks; Facebook did not have any application that allows you to see real trading accounts before and Cake Financial’s customers are mainly individual investors and do not belong to any network. Having a Cake Financial application on Facebook is a good example of connecting one large connected component to another connected component, and we expect to see that they will merge into a larger social network. 

http://www.voanews.com/english/2008-02-27-voa64.cfm

I never knew much about the federal interest rate until I took Econ 102 my sophomore year. My professor would often talk of Alan Greenspan, the chairman of the Board of Governors of the Federal Reserve at the time, meeting to discuss changing the federal interest rate. I believe at the time, the general trend of the interest rate was increasing, and before we talked about its impact on the economy, all that meant to me was that when I received my monthly interest on the my savings account, it would be a little bit bigger.

Recently, however, the Federal Reserve has been cutting the federal interest rate in order to boost the economy. The reason this works is because with the lower interest rate, businesses are more likely to take out loans and spend money which in theory helps to boost the economy. Lower interest rates also encourage spending in general since savings accounts now will be earning less money through interest. The Federal Reserve decreases the interest rates when businesses are struggling, stocks are falling, and in times of recession.

Really, the whole situation can be viewed through game theory as a two-move sequential move game. The two players are the government/economy and the businesses. The goal of the government is to boost the economy and the goal of the businesses is to make more money - both players know each other’s objectives. The government moves first and lowers the interest rate. They do this because they know the best response for the businesses is to take out loans and spend money, putting it back into the economy. As predicted, the businesses do so. In this game the government has achieved its goal of getting businesses to invest in the economy and at the same time the businesses also come out ahead because they have more money (from loans) to spend. Having more money helps them to increase business through advertisement, buying new equipment, and other means.  In this game, both players are essentially winners in the long run.

We have talked about the prisoner’s dilemma in class but what is also of interest is the effect game theory plays in the “game” played between criminals and the police and the result it has on the unsuspecting public. One of the things Ehud Guttel and Barak Medina discuss in their paper is how criminal sanctions can harm the more vulnerable victims as a result of game theory.

In their model the level of sanctions against a criminal determines the volume of his illegal activity. Say there are two neighborhoods, vulnerable (V) and less vulnerable (LV), with the probability of a criminal’s success in performing his crime is in higher in V than LV. Then a criminal would most definitely act in neighborhood LV if there were no police presence. Since police resources are finite, police cannot be in both neighborhoods at the same time and if captured the burglar bears a sanction that could make his net payoff negative. If the criminal operates only in one neighborhood the police increase their presence there as a counter strategy. Therefore the only way to reach Nash equilibrium is if both players use mixed strategies and randomize which neighborhood they act in.

Since the police’s goal is to reduce the volume of crimes, they can try to do so by decreasing activity in LV such that the criminal will act only in LV where he has a lower chance of success in succeeding. However though his success rate is lower in LV, the reduced police presence means that the probability of apprehension is also low and the number of criminal actswill increase and result in a larger number of victims. The sanction against the criminal restricts the expected payoff and therefore the total number of criminal acts but not the distribution of his activity. In contrast police allocation is affected by the severity of the sanction and a move towards harsher sanctions will induce a police move from V to LV. Now neighborhood LV benefits from both the sanction and the police presence in their neighborhood but neighborhood V are harmed for the exact same reasons. The result is that the more vulnerable victims in neighborhood V will have a spike in criminal activity even with the net decrease in crime.

Guttel and Medina specifically use a burglar as an example and assign numerical values for probabilities, payoffs and sanctions. There game theory logic though is discussed in the first seven pages of their paper. I find that their theory provides a unique look into how game theory may be used to produce arguments that at first may seem counterintuitive.

Source: Less Crime, More (Vulnerable) Victims: Game Theory and the Distributional Effects of Criminal Sanctions

What do zebras fleeing a lion and emailing patterns after the Enron scandal was exposed have in common? Networks! Network theory can be used to identify exactly when a lion comes into the vicinity if a herd of zebras. Their pattern of reaction can be predicted from the specific way that the zebras interact with each other as determined by studies of zebra networks. Biologists have taken advantage of computer science methodology to analyze how associations within a population are made by using well kept records with zebras as nodes and the interaction between them as the edges. This research on zebras has been investigated by Dan Rubenstein of Princeton University, an ecologist who has studied zebras and other horse-like animals for 20 years. He initially embarked upon the use of network theories to potentially enlighten him on the differences between two species of zebras: the plains zebra, which are thriving, while the Grevy’s zebra are endangered. He discovered that the two species are very unalike in their networking behaviors. While Grevy’s zebra spend their time with different acquaintances every day, plains zebra generally form tight social groups with a few outliers that are not a part of the main group. Here is a representation of the herd of plains zebra that Rubenstein studied:

This example illustrates perfectly the reason for the need for such classes such as Networks. Such a systematic study of associations between nodes can be a very important analysis tool that can reveal characteristic that even a biologist cannot discover without the aid of a computer scientist. When Professor Rubenstien received this aid, by giving his data over to a computer scientist to analyze, the computer analyst found that the pattern of zebras fleeing from a lion resembled another networking scenario that she was working on: the circulation of email after the Enron scandal was revealed. This example resembles what we discussed in class about organizing information “as we think” instead of linearly. By doing so, interesting insights can be found from seemingly unrelated topics.

http://www.wired.com/culture/lifestyle/news/2005/03/66813?currentPage=1

In this Wired article, Terdiman discusses MoSoSos or mobile social software which could potentially make a huge cultural impact on our lifestyles. Mobile social software is the technological confluence of social networking tools and either Wi-Fi connection location or global-positioning software. MoSoSos facilitate social encounters by identifying the real-world geographic location of a mobile device (cell phone or laptop) in real time. To put it more simply, it instantly enables you to find like-minded people in your area and at that time for social or business networking. The article runs through a number of different services including Dodgeball.com, Jambo, Playtxt and discusses how each works to connect users.

Unlike online social networks like Friendster and LinkedIn which are relatively passive, MoSoSos are a geographically mobile solution, allowing people to connect with affinity groups or find new friends while on the go. Although there are some privacy and safety concerns, it is clear that the idea is quickly catching on and is spreading by word of mouth. In the UK, Playtxt is very popular and the Google owned dodgeball.com is the US equivalent. The stickiness factor is high, especially on college campuses. For example, Washington University in St. Louis is signed up as a Jambo customer. This means that all university students logged into Wi-Fi hotspots on campus can view others’ profiles and communicate and connect with those members. An initial wireless communication through the digital network can often lead to face-to-face meetings with people of similar interests who you would have not met otherwise.

MoSoSos are directly related to the ideas of closure processes and social-affiliation networks. The MoSoSo network of people is a social affiliation network in which we can consider nodes of two distinct types. There may be some nodes that represent people and other nodes for the activities that these people engage in. The activities may vary from participation in the mobile social network itself to other activities such as the interests and hobbies that people list in their personal profiles. Focal closure or the tendency for new links form between two nodes which engage in a shared activity appears to be at the heart of the operation of a MoSoSo network. If B and C represent people who are users of a mobile social software service (activity A), it is likely that when B and C are in the same vicinity at the same time, they will each be alerted of the presence of the other person. If B and C have both listed rock climbing (activity A) on their profiles, then the likelihood of each of them being notified of the other’s proximity, their decision to meet and ultimately the development of a friendship edge between them will be much higher. Thus, friendship edges develop between members of the network because of shared common interests.

We can look at the network in another way. If A is a user of dodgeball.com (activity C), A’s friend B may also become a member after hearing about it from A. This demonstrates affiliation closure (based on homophily). This is also the way in which the technology spreads and certain ideas catch on quickly. Malcolm Gladwell in The Tipping Point refers to this as the stickiness factor.

More than a year after it’s first available, the Nintendo Wii is still in shortage. Never available in the stores; prices skyrocketed on eBay. Plus many people now use all kinds of online Wii tracking software to try to make money from this unbalanced supply and demand.

The situation was even worse right before Christmas, when parents were trying their hardest to get Wii’s for their crying kids. The article “A Year Later, the Same Scene: Long Lines for the Elusive Wii” on New York Times described that many people waited in front of stores in Manhattan very early in the morning, usually long before they could see the sun.

Have been one of the them myself, I realized game theory can be applied here. Each buyer’s like an individual player of the game. And the game rule is that the earliest few players win Nintendo Wii’s. For each player, the dominant strategy is to show up as early as he/she can and wait in front of the stores since he doesn’t know what time his opponents would show up. Therefore, the Nash Equilibrium of the game is that everybody’s going to show up very very early, say 3AM, just like what was described in the article from New York Times.

So does this maximize the social welfare? John Nash says when everybody is doing the best for themselves, sometimes it’s not good for the society, and I agree with him. Let’s imagine a Non-Nash-Equilibrium of this game. Everybody can have a good sleep and come to the stores late, say 1PM, but in an exact same order of the previous case. Whoever gets Wii’s don’t change, and everybody’s better off because they get more sleep, which is always good, especially for college students.

In class, we briefly discussed how road networks play into game theory. We kept our model of real-world road networks relatively simple, limiting our congestion-reducing strategies to road construction and toll and subsidy implementation. We never delved into other traffic control innovations, such as the traffic signal, to which the following article pertains:

http://www.trafficpd.com/Publication/Signal1.pdf

The article is directed toward a Pennsylvanian audience, but it is nevertheless relevant to our nation’s and even the world’s traffic issues.

We all know traffic is an enormous problem in many major cities, and since the population is always increasing, there’s no mitigating the number of cars on the road until a better, more attractive method of transportation is invented. As we learned in class, adding roads will not necessarily solve traffic congestion (Braess’s Paradox), and this is a commonly accepted fact among traffic planning engineers, who would recommend new road construction only as a last resort. Traffic signals are one of the more efficient or effective solutions to traffic congestion that provide an alternative way to get around Braess’s Paradox. By sequentially assigning “the right-of-way to vehicles approaching an intersection from various directions,” traffic signals allow vehicles from side roads to merge into main roads, in addition to allowing vehicles to more easily make left turns.

In class, we neglected the fact that cars would somehow have to merge where more than two roads intersect, which if included in our analysis would have added a factor to the travel time. Traffic signals address this neglected effect, or oversimplification. Despite the theoretically high effectiveness of traffic signals, location, of course, is what really determines effectiveness. Like road construction, if placed at the wrong location, signals can cause unnecessary delays. They can even increase the likelihood of accidents, which is why sight distance and accident history for the area must be taken into account when deciding whether a signal is necessary.

One big story from a few weeks ago was Microsoft’s big offer to acquire Yahoo! in its bid to expand its power in the market of online services and advertising. Most notably, this would give them a big leg up in their battle against Google, their main competitor in this field.

Consider the situation prior to the possible deal. In this realm, Microsoft, Yahoo! and Google are three competitors, each fighting for a similar online market. Google provides email through GMail, and has the undisputed lead in both online advertising and search engine results. Microsoft has its own version of each of these through Hotmail, MSN Search and a smaller share in online advertising. Yahoo! also provides its own email and search engine service. They all also have smaller initiatives such as maps, online applications and utilities, and so on.

These three bodies can each be considered nodes in a three-way network. In the current sense, each has a negative relationship because they are all competing against each other for market space. These three negative edges is one form of the unbalanced network. Furthermore, these nodes could be considered “supernodes” as each body is slowly eating up each of the other smaller competitors in this market. For example, when Google acquired Doubleclick, one could consider a negative edge between them to become positive forming a bigger party. Yahoo!’s acquisition of del.icio.us and

Now consider what will happen if the deal goes through. Microsoft and Yahoo! will unite to form a gigantic new party, a “super-super-node” of sorts. The battle for online commerce will become even more polarized as Google will now have one other body, Microsoft and Yahoo! to compete against. Thus, all the smaller nodes within the supernodes of Yahoo! and Microsoft will become positive edges and remain negative edges towards all nodes in the Google supernode.

In short, the network of online marketing could undergo a major shift as Microsoft and Yahoo! could merge to make the network more balanced. There will be only two parties competing for services offered on the internet - Google (and all it’s smaller acquired groups) and Microsoft (with all of their smaller acquisitions plus Yahoo! and all of Yahoo!’s). This could lead to a big showdown in the world of online commerce.