ALASKA STATE LEGISLATURE  SENATE LABOR AND COMMERCE STANDING COMMITTEE  April 22, 2003 1:35 p.m. MEMBERS PRESENT Senator Con Bunde, Chair Senator Ralph Seekins, Vice Chair Senator Gary Stevens Senator Bettye Davis Senator Hollis French MEMBERS ABSENT  All members present COMMITTEE CALENDAR CS FOR SENATE BILL NO. 13(JUD) "An Act relating to using credit history or insurance scoring for insurance purposes; and providing for an effective date." HEARD AND HELD PREVIOUS ACTION SB 13 - See Labor and Commerce minutes dated 4/8/03. WITNESS REGISTER Mr. Jesse Kiehl Staff to Senator Kim Elton Alaska State Capitol Juneau, AK 99801-1182 POSITION STATEMENT: Commented on SB 13 for the sponsor. Mr. Birny Birnbaum, Executive Director Center for Economic Justice Texas POSITION STATEMENT: Supported SB 13. Mr. Sam Sorich, Vice President National Association of Independent Insurers 980 Ninth Street, Suite 1600 Sacramento CA 95814 POSITION STATEMENT: Opposed SB 13. Mr. Eddy Lo, Insurance Manager Fair, Isaac and Company 200 Smith Range Rd. San Rafael CA 94903 POSITION STATEMENT: Opposed SB 13. Mr. Arthur Parks Underwriting Administrator State Farm Insurance 2500 25th Ave., NE Salem OR 97317 POSITION STATEMENT: Opposed SB 13. Ms. Elizabeth Moceri Regional Counsel Allstate Insurance Company Bothell WA POSITION STATEMENT: Opposed SB 13. Mr. Thom Buzard 2910 Linda Juneau AK 99801 POSITION STATEMENT: Supported SB 13. ACTION NARRATIVE TAPE 03-22, SIDE A  SB 13-INSURANCE DISCRIMINATION BY CREDIT RATING  CHAIR CON BUNDE called the Senate Labor and Commerce Standing Committee meeting to order at 1:35 p.m. Senator Seekins was present. Chair Bunde announced SB 13 to be up for consideration. MR. JESSE KIEHL, staff to Senator Elton, sponsor, said he would answer any questions or shed further light on the discussion as it continues today. MR. BIRNY BIRNBAUM, Executive Director, Center for Economic Justice, said he advocates on behalf of low-income consumers. He has been asked to testify on credit scoring because of his many years of experience and expertise on the topic. His position is that insurers' use of credit information should be prohibited because it is inherently unfair, discriminates against certain classes of consumers, undermines the insurance mechanism, undermines regulatory oversight of rates, and violates actuarial principles. He said that insurers say that if you have good credit, you get a good score and that simply is not the case. The vast majority of consumers don't have any delinquencies or bankruptcies and yet a much greater percentage is penalized because of a bad credit score. He wanted to respond to some comments made by Mr. Niehaus from Progressive Insurance [at a previous hearing] who said that a ban on credit scoring would force Progressive to raise rates for two-thirds of its policyholders. He said that is not true. A ban on credit scoring would simply prevent insurers from using an unfair classification tool. Insurers choose the rate they charge subject to review by the commissioner [of the Department of Community and Economic Development] and nobody is forcing them to raise rates. Insurers are saying that because a majority of consumers benefit, it should be allowed. However, in America, minorities are supposed to be protected against unfair practices by the majority. He suggested including a provision in SB 13 that requires notice be given to consumers after the statute is enacted that says: Your legislature has banned the use of your personal credit information by insurers to determine if you are eligible for coverage [and] how much you will pay. No insurer is required to raise rates or charge you more as a result of this law. MR. BIRNBAUM said Mr. Niehaus argued that good drivers would subsidize bad drivers, but the Texas study points out that even in the worst group of credit scorers, there were 12 claims out of 100 policies, while in the best group of credit scorers there were only 9 claims out of 100. The bottom line is that the majority of consumers simply don't file claims in a given year. The notion that a group of really bad drivers is being subsidized by a group of really good drivers is just not true. MR. BIRNBAUM said that Mr. Niehaus also argued that a ban on credit scoring would create an uneven playing field where direct writers would have an advantage. That's not correct because the Fair Credit Reporting Act allows insurers to use credit to get mailing lists to consumers, but once the solicitation is sent, the insurer is subject to the insurance laws of the state. All sorts of insurers use credit for direct mail - like Allstate, State Farm, Progressive and others. Mr. Niehaus also failed to identify Maryland as a state that has banned the use of credit scoring for homeowners and severely restricted its use for auto insurance. He also argued that a ban on credit scoring would damage an Alaska market that is already in bad shape. He said Alaska ranks 47th out of the 50 states and the District of Columbia in profitability and that insurers had lost money in three out of the last four years. Those are the years during which insurers started using credit and it's hard to see how a ban on credit could make the market any worse. He said that Mr. Niehaus suggested that the Legislature look at the National Conference of Insurance Legislators (NCIL) model as an alternative. He cautioned that model was developed as a compromise between agents and insurers and doesn't have essential consumer protections. It simply endorses the practices of most insurers. MR. BIRNBAUM provided the nine factors in the scoring model that Progressive uses: 1. Months you have managed credit 2. Age at which you first established credit 3. Number of times a payment was past due more than 30 days 4. Number of loans and accounts with a satisfactory current payment record 5. Number of credit card accounts currently past due more than 30 days 6. Percent of available credit limit currently being used on revolving accounts 7. Percent of viable credit limit currently being used on all open accounts 8. Months since your most recent auto loan was made 9. Credit inquiries you initiated in the past 25 months He said a lot of those factors don't distinguish between a problem someone may have had with credit five or six years ago and current problems. They are factors that have nothing to do with how a person manages credit and whether that person is going to be a good driver. MR. BIRNBAUM said insurers have argued that the Texas study is a conclusive independent study that demonstrates the correlation between credit and risk of loss. However, he is very familiar with the study and the methodology that was used was considered and rejected by the National Association of Insurance Commissioners (NAIC) in 1996. It didn't get at the core issue, which was whether credit has actually any unique contribution to explaining risk of loss or whether credit is simply correlated with some other factor. The study did show that the average credit score for the standard market was significantly higher than the average score for the non-standard market. He advised: But, this shouldn't have happened because the credit scores were taken before the insurers were using credit information. So, the expectation was that the average credit scores would be about the same. What this shows conclusively is that the use of credit by insurers is duplicating some other factor that insurers were already using to distinguish among consumers by risk. MR. BIRNBAUM said the Texas study shows that credit replicates some other factor to some extent, but it doesn't show whether it has any unique contribution to explaining risk of loss. MR. SAM SORICH, Vice President, National Association of Independent Insurers (NAII), said about 100 of NAII's member companies are doing business in Alaska and are responsible for about 60% of the homeowners and car insurance written in Alaska. He said when they sell their product, they don't know what the ultimate cost will be; they can only estimate it. They could do that and charge everyone the same rate, but that wouldn't be fair. Over the years, insurance companies have tried to identify factors that show the likelihood that a person would have an insurance loss. They use things like the number of miles driven, the type of car a person has, how old a person is, their experience, etc. Those factors go into determining the appropriate rate for a person. They try to make sure a person pays a premium that is commensurate with the risk of loss, because it would be unfair to require them to pay more for insurance than their risk of loss factor warrants. Some people use credit information to further the goal of making sure that the premiums charged are commensurate with the risk of loss. MR. SORICH commented that the methodology that was used by the University of Texas business school was not paid for by the industry. The business school took 150,000 policies written in 1998 that did not use credit scoring, because insurance companies didn't use credit scoring in 1998. The business school looked at the loss experience and the credit scores of those policyholders and found that the people who had a high experience of loss had poor credit scores and people who had a lower incidence of loss had good credit scores. The regression analysis of a relative loss ratio of a credit score was highly significant and indicated a less than 1 in 10,000 chance that the relationship observed could be due to chance alone. MR. SORICH said Mr. Birnbaum mentioned that the study did not look at whether credit scoring just duplicates other factors, but that is incorrect. He explained: The study says logistical and multiple regression analysis examined whether the review of the relationship between credit scoring and incurred losses was explainable by existing underwriting variables or whether the credit score added new information about losses not contained in the existing underwriting variables. It was determined that credit scores did yield new information not contained in the existing underwriting variables. Insurance companies have a responsibility to our policyholders, our customers, to consider this evidence, because if we are forced to ignore this evidence, it will mean that we will have to charge people more than they should be paying for insurance. SENATOR DAVIS arrived at 1:48 p.m. MR. SORICH pointed out that the legal basis for an insurer's use of credit information, the federal Fair Credit Reporting Act (FCRA) of 1970, lists seven permissible purposes for credit information. Insurance underwriting is expressly detailed in the FCRA. States can enact legislation restricting the use of credit information by insurance companies, but the legislation may not be inconsistent with the federal law. No state has enacted legislation similar to SB 13 and no state has enacted an absolute ban on insurers' use of credit information. He explained the Hawaii law limits insurance companies' use of credit for rating of auto insurance. It does nothing about homeowners insurance. Hawaii's system is completely different than Alaska's system. Hawaii has a no-fault auto insurance system, which means that lawsuits are limited. Medical costs are limited according to a worker's compensation medical fee schedule and insurance companies are required to offer after- market parts. All of those provisions help to lower costs in Hawaii. The use of credit scoring provides the consumer with basically five benefits. It makes the insurance companies' decisions more objective. A credit scores doesn't have a good day or a bad day; it doesn't have a gut feeling about a person. A credit score does not consider a person's race, age, income, where they live or what kind of car or house they have. Second, the insurance business is based on information. Insurance companies look at driving records and past claims history. The use of credit information helps them have more information about the individuals, which enables them to make a more informed decision. Third, credit information helps to lower the cost of insurance because it expedites the underwriting process. Fourth, based on the evidence in Texas and other studies, it's clear that the use of credit information helps to make insurance premiums more equitable. It helps insurance companies achieve the goal of making sure that the rates they charge are commensurate with the risk of loss. Finally, the availability of credit information helps insurance companies to write more insurance. They are not in the business of not selling insurance. MR. SORICH said this practice is subject to some controversies and clearly there are regulations and restrictions that are appropriate in this area, but SB 13 takes the wrong approach by imposing an absolute prohibition. There are other alternatives. Insurance companies or agents did not develop the NCIL model, it was adopted by a group of state legislators. It was adopted last year and is being considered and passed in a number of states: North Dakota, Nebraska, Kansas and Oklahoma. He encouraged the Alaska Legislature to consider this model. CHAIR BUNDE asked if someone had an accident record but a good credit score, that might help the person get insurance at a lower rate than someone without a good credit score. MR. SORICH replied: In some cases, it could. If we only have the driving record to work on, we have to make our decisions based on the driving record. The addition of additional information gives us more options. CHAIR BUNDE responded that Mr. Sorich had mentioned insurance companies have a responsibility to their customers and asked if it wouldn't be accurate to say their greatest responsibility is to their investors, because they must maintain a profit or they won't stay in business. MR. SORICH replied that is true, but many companies are not profit organizations. They are mutual insurance companies or reciprocals. They are in the business to stay solvent. 1:58 p.m. SENATOR SEEKINS asked when he last bought a vehicle. MR. SORICH replied about five years ago. SENATOR SEEKINS asked if he would have a better credit score on his insurance rating than someone who bought a car three months ago. MR. SORICH replied the fact that he pays his car payments on time is a valid indicator of how he handles his finances. SENATOR SEEKINS asked if the number of months he made payments made him think that. MR. SORICH replied that different scoring models consider payment history on loans, such as car loans. The credit report does not go back beyond three years for car payments. So, if a person just bought a car, that piece of information may not be available to develop the score. SENATOR SEEKINS said he looked at one insurance company's credit model and it had an 11-point difference based on how long ago a person bought his or her vehicle. MR. SORICH responded that he thought the Fair Isaac model does not consider all the inquiries. The NCIL specifically states that all those inquiries are considered as one inquiry. SENATOR SEEKINS said his responsibility model (from a major company) indicated that as the number of non-insurance inquiries went up, the points changed. MR. SORICH said that Mr. Lo could probably answer that better than he could and that the Progressive model considers all inquiries for loans made over a certain period of time as one inquiry. MR. EDDY LO, Insurance Manager, Fair, Isaac and Company, used a power point presentation called "A Discussion on the Use of Credit Information and Scoring for Insurance" that was also presented in the form of a handout to members. He pointed out that there are legal grounds for using credit information with insurance. Fair Isaac was a pioneer in modeling the use of credit information over 30 years ago. The FCRA specifically allowed the use of credit scores for setting insurance rates, but before that, the FCRA required the consumer-reporting agency to adopt procedures governing the accuracy, relevancy, access to and utilization of consumer reports. He showed a picture illustrating credit information being summarized by a numeric score. It the score value is low, there is a corresponding very high loss ratio (the losses in relationship to premium collected under an insurance policy), which has been proved and validated many times. It shows that you can separate risk by percentage compared to an average number. MR. LO said they make sure that policy premiums and losses are matched to credit information up front. Using credit information at the beginning of a policy shows nothing about the subsequent losses of that particular policy's performance. Starting with that premise, they tracked actual losses at 12-month intervals. They were able to see that when the credit information was summarized by a numeric model and the score was low, the loss ratios were very high. Very high credit scores correlated with low loss ratios. This is a Fair Isaac convention, but he thought the model Senator Seekins was basing his questions on used a different convention. He said based on looking back in history when credit had nothing [to do with] losses and trying to correlate what happened since that time, where there are even errors, the resulting losses are very predictable. Underwriters like to look at objective information and there are many third party validations of the information. MR. LO said the Fair, Isaac and Company model does not use age, disability, gender, health status, income, location, marital status, nationality, net worth, occupation, race, religion, sexual orientation and zip codes. They are not the only ones saying the use of credit information is fair. From 2 - 6% of people dispute what is in their credit report and that is the starting point of the estimate of an error rate; but the problem is, while even 6% is acceptable, the database in the U.S. is humongous so if it applies to 260 million people, that would mean millions of consumers' reports need to be corrected. The FCRA has a very stringent process to insure the integrity of the data; heavy penalties are involved to insure accuracy is high. They found for all insurance inquiries, the net result is that every time an inquiry is generated, statistical data shows that higher and higher losses are expected. MR. LO said that Fair Isaac builds the models, but they do not calculate the scores themselves. Three major reporting agencies do that and Fair Isaac has contracts with them all. The type of information they show are adverse public interest records like bankruptcies, foreclosures, judgments. Most people, 96 percent, do not have adverse public records and have good credit. They have found that any delinquencies over 60 days in the last two years indicate losses are worse; but once again, the majority of people have no delinquencies in their credit reports. They are not penalized by the use of credit information. Ninety-seven percent of the time, people do not have collections generated. Increases in financial applications indicate and predict higher losses in the future or in the past, but they are not an absolute indication. TAPE 03-22, SIDE B  SENATOR STEVENS arrived at 2:25 p.m. MR. LO continued his explanation by showing five different ways consumers are statistically separated by underwriters in terms of risk. There is legal ground for the use of credit for insurance as defined by FCRA and there are indications that their model is very predictive statistically. CHAIR BUNDE asked if credit scoring was banned, whether that would create any conflict between state and federal law. MR. LO replied yes, because it is clearly stated in FCRA that one of the permissible purposes of credit information is underwriting of insurance. SENATOR FRENCH arrived at 2:30 p.m. SENATOR SEEKINS said in the model he was provided, the lower the score is, the lower the insurance policy will cost. MR. LO responded that it is exactly the opposite of his convention. SENATOR DAVIS arrived at 2:35 p.m. SENATOR SEEKINS asked if a person bought a car today or in the last 11 months versus more than 48 months ago, whether that provides any predictability and, therefore, affects a person's credit score. MR. LO replied that it does and, from his general understanding of that particular model, the longer you use credit information, the longer you practice financial management skill with it and get better at it. That generates lower losses. A person who recently bought a car will not have gone through as much financial exercising and proven on record what it looks like. He said one particular instance does not dominate a credit score; the overall profile predicts the final score and the future losses. "That has been proven to be accurate." SENATOR DAVIS asked what he believes causes the correlation between credit scoring and insurance loss and why he feels that it works. MR. LO replied that he assumed she didn't mean the statistical correlation relationship, but rather why someone who managed their finances well would produce lower losses in the future. He could only give her a layman's interpretation, because he is not a social scientist. He said a person who uses credit well and makes payments on time is more likely to make sure the tire pressure in the car is correct, that the battery and lights are working, the oil level is properly maintained, the furnace is working properly, and nothing is left laying around to trip over. SENATOR DAVIS asked what happens to people who don't believe in credit. She asked if they get penalized under his model. MR. LO replied no; if you can't find a person on a credit data base, no score can be generated for that person. It has to go back to the underwriting guideline that clearly establishes traditional underwriting factors. That is not a new phenomenon. MR. ARTHUR PARKS, Underwriting Administrator, State Farm Insurance, said State Farm has developed its own model. State Farm sampled 1.5 million policyholders by scoring them on experience and length of time and looked at what the loss ratios were. They found a 99% correlation between what the underwriting score was and what the ultimate loss ratio was. Their model incorporates certain components from the credit report and the prior losses an individual has incurred. Those are combined to come up with a score that indicates the propensity for future losses. MR. PARKS explained that having certain credit components doesn't cause future losses, like good grades in school do not cause a good driving record, but those qualities are a predictor. He said they are looking at the probability of missed or late payments, as well as the financial risk of incurring a loss down the road - two different things. The way people fit into the overall formula is different. You can have someone who has an excellent credit score for lending purposes, but doesn't have a good score for underwriting purposes. They have found that credit records are extremely complete. Recent studies have shown that anywhere from 10-22% of motor vehicle records are inaccurate, because people hire attorneys to clear their record. Seventy-nine percent of the losses they get from comprehensive loss underwriting exchange are inaccurate. That means that approximately 21% of the losses don't show up on the exchange. In summary, MR. PARKS said, banning credit for underwriting purposes will do several things. It will force those with a better risk ratio to subsidize those with a poor loss history and allow insurance companies to use less accurate information to select and price business. It will decrease the ability to fully distribute the cost of insurance amongst those who are driving the costs upwards. CHAIR BUNDE announced that members were being called back to the floor and asked if everyone would be available to accept written questions. There was general assent. MS. ELIZABETH MOCERI, regional counsel, Allstate Insurance, said she was concerned about the viability of writing insurance in Alaska and that using credit is a very valuable tool for Allstate. At this time, Allstate uses it to accept and reject applicants for insurance. Allstate does not use it to price insurance right now and is waiting to find out if it will be able to. In the states where Allstate does, it is able to have a rate for folks it normally would not take. Allstate is in the business to write more insurance, not less. Alaska has unique rules, such as Rule 82 attorney fees and unique uninsured motorist provisions, that make writing insurance and paying claims very expensive. CHAIR BUNDE said the committee heard that if someone's credit score has negatively impacted their insurance rates, they are notified. He wanted to know if that occurs with Allstate as well. MR. THOM BUZARD said he is a Juneau resident and is in favor of SB 13, which outlaws the use of one's credit scores. He related his history of 18 years ago when his son was born with a major birth defect and hospitalized for 10 days. Mr. Buzard ended up being $50,000 in debt. If that happened now, six months from now, he would not be able to get insurance from anybody. He lost everything he owned trying to pay all his bills. He argued that people who are high risk are the ones that need to be insured for society's protection. He said the insurance companies have so much power over their customers that they sometimes don't report claims. He didn't think it was reasonable to use credit scoring because some people don't use credit and some are too young to establish it - and some of them have medical bankruptcies. Some have fallen prey to easy credit and failed to realize until it's too late the mess they had gotten themselves into. CHAIR BUNDE thanked him for his brevity and said his written testimony would be made part of the record. He said there would be another opportunity to discuss this bill and adjourned the meeting at 2:55 p.m.