All of that information stored on you is useless unless you can make sense of it. If your potential lender asked for a report on you, they might get thousands of pages of information. They would see printouts full of numbers and dates, and they would need weeks, at least, to decipher all of the information. As a result, several number-crunching firms do the work for them.
Fair Isaac Corporation
Perhaps the best-known number cruncher out there is the Fair Isaac Corporation (FICO). FICO is obviously the creator of the FICO Score, the most popular credit score used today. FICO refers to itself as an applied mathematics company, which does not keep any information on hand. It uses consumer information when it builds its mathematical models, but it is not a warehouse for information like the credit-reporting companies.
FICO works more like a consultant. The company works with service providers like lenders, insurance companies, and others to try and predict how people will behave. In the lending industry, for example, FICO helps lenders determine how likely a borrower is to become ninety days late with at least one creditor in the next twenty-four months. Working with lenders, FICO determined that this knowledge would be helpful. To do this, they use a proprietary algorithm (a secret math formula) to find patterns. Once they figure out the formula, FICO sells its analytical software to others. The credit-reporting companies use the software, and they load data into it to create scores.
The FICO Score has changed the way lending works. As a result of automated scoring, lending decisions are faster than ever. Lenders no longer have to read through credit reports and look for hints that suggest a prospective borrower is a good or bad risk. Personal judgment, prejudice, and human error are all but eliminated from the equation. Sometimes a lending decision will be made without any human input; if the model says your score's too low then you can't get a loan. As you can see, automated scoring has its advantages and disadvantages.
FICO has established a strong lead in the credit-scoring business. Most lenders look at your FICO score when they are deciding what kind of terms to offer. There are a lot of other credit scores out there, but the most relevant one is probably the FICO Score. In 2006, the major credit-reporting companies announced that they created a new scoring system called Van-tageScore. This system, a competitor to the FICO Score, would offer an alternative to consumers and financial institutions. However, as of this writing there was concern about the viability of the score: lenders already use the FICO Score and understand it, and learning something new might not be worth it. Furthermore, if lenders want to resell a mortgage in the secondary market, the mortgage is usually evaluated by the borrower's FICO Score.
The Fair Isaac Corporation's analytical skills can go far beyond lending and insurance decisions. The California Penal System reportedly asked the company to build a model that helped predict whether or not convicts would commit additional crimes after being released from prison. The project was never brought to completion, but you can see how these techniques get used elsewhere.
ChoicePoint is another major number cruncher. The company, as previously noted, also keeps a store of data, so it plays several roles. ChoicePoint does not really create credit scores that are used for lending. However, they create scores that may use information from your credit reports. Choice Point is known for insurance scores and an insurance-claims history database. The company might not have much to do with your loans, but they are likely involved in your insurance offerings.
Like FICO, ChoicePoint uses mathematical models to try and predict outcomes. The company helps insurers figure out who might be a profitable customer. In addition, ChoicePoint uses its expertise in a variety of other ways. For example, they work with law-enforcement agencies, and have won some large government contracts. The same concepts apply: they look for cause-and-effect relationships in huge volumes of data, then the model spits out an easy-to-understand answer.
There are plenty of other number crunchers out there. The credit-reporting companies do some scoring, and specialty firms provide scores specific to whatever need comes up. If a product or service provider wants a score, somebody out there will build one. In general, scoring models follow a common set of steps.
Identify the Need. First, the scorer must determine what the score is intended to accomplish. For credit scores, they want to show if somebody is a good borrower. More specifically, you might ask: How likely is it that this borrower will become ninety days late with at least one creditor in the next twenty-four months?
Get the Information. Next, they get all of the information that might be interesting to the question at hand. For credit scores, you might want to collect data on previous credit accounts: how many there are, how much gets borrowed, whether payments are on time, and so on. You might decide to go even deeper: are public records important? Is a person's favorite color important?
Crunch the Numbers. Once you have an objective and all the data you need, you can start to look for relationships. Many software programs do this for you. The real challenge is in gathering the right data, categorizing it appropriately, and asking the right questions. Most tools use regression analysis to find relationships and make them understandable.
Test the Output. After you have a formula or a model, you need to test it to see if it really works. You can look back at the past, or you can test it going forward. With backtesting, you can use historical data on past customers over many years to see what would have happened if you had used that formula or model.
Of course, no scoring system is foolproof, all they can do is use your past credit history to try to predict your future behavior. This is why it's so important to be a responsible borrower — your credit history will indicate that you are a good credit risk.