Legal Services Advocacy Project
2324 University Avenue, Suite 101
St. Paul, MN 55114
651-222-3749
October 2000
INTRODUCTION
Companies offering homeowner's insurance in Minnesota are required by statute to submit data annually to the Minnesota Department of Commerce on the number of policies written, canceled and nonrenewed, and the number of applications for policies declined.(1) The data are reported for the 42 zip codes covering the three largest cities in the state: Minneapolis, St. Paul, and Duluth.
During the summer of 2000, the Legal Services Advocacy Project (LSAP)(2) analyzed the data set for the most recent years available(3) in an effort to identify patterns of service for low-income, middle-income and high-income communities, and for communities of color, racially mixed communities and white communities.(4)
The analysis suggests that people with lower incomes face barriers in obtaining and retaining homeowner's insurance. And while there were no indications of any pattern of intentional discrimination or bias against homeowners who live in zip codes of color, the findings suggest that, in effect, homeowners in those areas may be disproportionately underserved in their ability to obtain homeowner's insurance. The findings also make clear that determining the role that policies and practices of insurance companies play in blocking access to and retention of homeowner's insurance in low-income zip codes and in zip codes of color warrants further investigation.
METHODOLOGY
LSAP aggregated the raw data by city and zip code on policies written, canceled, nonrenewed, and declined. The data were further divided into "primary" company activity (defined as companies writing 100 or more homeowner's insurance policies in any one zip during any of the three years) and "secondary" company activity (defined as companies writing less than 100 homeowner's insurance policies in any one zip code during the entire study period).
At the same time, census data obtained from the 1990 Profile Generator at the Missouri State Census Data Center was used to compile background information on each of the 42 zip codes with respect to the following:
- population;
- number and percentage of residents below the poverty level and below 50% of the poverty level;
- the civilian labor force unemployment rate;
- median family income;
- percentage of households with incomes below $10,000 and $25,000;
- number of housing units;
- percentage of units that are owner-occupied, rented and vacant; and
- a percentage breakdown of the population by race and ethnicity.
Based on the census data, zip codes were grouped by income (low-income, middle-income, and high-income) and by race/ethnicity (white, racially mixed, and zip codes of color).
The data was then cross-tabbed to conduct an analysis of patterns of service with respect to homeowner's insurance correlative to income and race/ethnicity.(5) The analysis was conducted to determine whether sufficient evidence exists to conclude whether or not low-income people and people of color are receiving disparate treatment with respect to access to and retention of homeowner's insurance.
FINDINGS
1. Policy Activity Is Dominated by 30% of the Companies
During the reporting period, nearly 440,000 homeowner's insurance policies, divided among 80 companies, were written. Of the 80 companies, only 23 companies had what could be considered "significant" activity (defined as writing more than 100 policies in any one zip code). These companies may be considered "primary" companies, while the remainder may be considered "secondary" companies.
LSAP found that these 23 primary companies, representing only about 30% of all companies doing business in the areas for which data was reported, accounted for 90% of all policy activity. Trends seen in the aggregate data were strongly related to the actions of these primary companies.
2. Significantly More Policies Are Written in High-Income Zip Codes
As illustrated in Tables 1 and 2, on average, there were three and one-half to four times as many policies written in high-income or middle-income zip codes as were written in low-income zip codes. The population differential between the high-income or middle-income zip codes and low-income zip codes was only approximately two to one (when considering the totals from all three cities).
Table 1. Policies Written by Zip Code Groupings: All Companies
| Policies written | Low-income zip codes | Middle-income zip codes | High-income zip codes | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| 1997 | 13,070 | 12.2% | 48,758 | 45.7% | 44,947 | 42.1% | 106,775 | 100.0% |
| 1998 | 19,691 | 12.3% | 74,394 | 46.6% | 65,730 | 41.1% | 159,815 | 100.0% |
| 1999 | 20,231 | 11.7% | 78,730 | 45.7% | 73,373 | 42.6% | 172,334 | 100.0% |
| 3-year aggregate | 52,992 | 12.1% | 201,882 | 46.0% | 184,050 | 41.9% | 438,924 | 100.0% |
Table 2. Population by Zip Code Groupings: Minneapolis, St. Paul and Duluth
| Population | Low-income zip codes | Middle-income zip codes | High-income zip codes | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| Minneapolis | 105,228 | 25.7% | 152,595 | 37.3% | 151,457 | 37.0% | 409,280 | 100.0% |
| Saint Paul | 30,622 | 10.2% | 185,568 | 61.8% | 84,269 | 28.0% | 300,459 | 100.0% |
| Duluth | 23,252 | 25.3% | 39,787 | 43.3% | 28,907 | 31.4% | 91,946 | 100.0% |
| Total | 159,102 | 19.8% | 377,950 | 47.1% | 264,633 | 33.0% | 801,685 | 100.0% |
At the same time, significantly more owner-occupied dwellings were located in middle- and high-income zip codes. Table 3 shows that there were about four times as many owner-occupied dwellings in high-income zip codes as in low-income zip codes.
In all, 21.1% of total housing units were located in low-income zip codes, 32.7% in high-income zip codes, and slightly over 46.2% in middle-income zip codes. However, only 10.3% of owner-occupied units were located in low-income zip codes, with the remainder almost equally divided between middle- and high-income zip codes. At the same time, more rental units were found in low-income zip codes, where almost one-third of total rental units existed. High-income zip codes accounted for only one-fifth of rental units, with the remainder in middle-income zip codes. Thus, for every family that rented in high-income zip codes there were 2.5 families that owned their home. In low-income zip codes, the ratio was exactly the opposite; for every family that owned their home in low-income zip codes, there were 2.6 that rented.
Table 3. Housing Frequency by Income Groups
| Housing units | Low income zip codes | Middle-income zip codes | High-income zip codes | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| Owner- occupied | 18,975 | 10.3% | 85,979 | 46.6% | 79,699 | 43.2% | 184,653 | 100.0% |
| Rental | 48,884 | 32.2% | 70,254 | 46.2% | 32,891 | 21.6% | 152,029 | 100.0% |
| Vacant | 8,101 | 34.8% | 9,936 | 42.6% | 5,276 | 22.6% | 23,313 | 100.0% |
| All units | 75,960 | 21.1% | 166,169 | 46.2% | 117,866 | 32.7% | 359,994 | 100.0% |
3. The More Racially Diverse the Zip Code, the Fewer the Number of Policies Written
As shown in Tables 4 and 5, insurance companies sold more policies in white zip codes than in more racially diverse ones, but the companies did not seem to differentiate substantially between racially mixed zip codes and zip codes of color. On a per person basis, however, the difference was more striking. Only one policy per 1.5 persons was written in zip codes of color, whereas one policy per 2.5 persons was written in white zip codes. In other words, one-and-one-half times as many policies per person were written in white zip codes as in zip codes of color.
Table 4. Policies Written by Year by Racial Composition of Zip Codes: All Companies
| Policies written | Mixed | Zip codes of color | White | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| 1997 | 31,220 | 29.2% | 21,540 | 20.2% | 54,015 | 50.6% | 106,775 | 100.0% |
| 1998 | 47,132 | 29.5% | 32,490 | 20.3% | 80,193 | 50.2% | 159,815 | 100.0% |
| 1999 | 50,343 | 29.2% | 33,845 | 19.6% | 88,146 | 51.1% | 172,334 | 100.0% |
| 3-year aggregate | 128,695 | 29.3% | 87,875 | 20.0% | 222,354 | 50.7% | 438,924 | 100.0% |
Table 5. Policies Per Person Per Zip Code Category
| Mixed | Zip codes of color | White | |
| Population (totals) | 249,968 | 219,700 | 331,795 |
| Policies written (3-year aggregate) | 128,695 | 87,875 | 222,354 |
| Policies per person | 1.9 | 1.5 | 2.5 |
5. Secondary Companies Are More Active in Low-Income Areas than Primary Companies
Graph 1 shows the number of policies written by primary companies, on average, in low-, middle-, and high-income zip codes. Primary companies wrote approximately four times as many policies in high-income zip codes as in low-income zip codes. Additionally, because seven of the ten zip codes of color were also high-income zip codes and every high-income zip code was also a white zip code, that racial bias may be a factor cannot be discounted.
As Tables 7 and 8 show, secondary companies wrote a higher percentage of policies in low-income zip codes than did primary companies. Secondary companies wrote 14.5% of their policies in low-income zip codes and about 39% of their policies in high-income zip codes. By contrast, primary companies wrote 11.8% of their policies in low-income zip codes and about 42% of their policies in high-income zip codes. In other words, secondary companies wrote nearly 23% more of their policies in low-income zip codes than did primary companies.
Table 7. Policies Written by Secondary Companies
| Policies written | Low-income zip codes | Middle-income zip codes | High-income zip codes | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| 1997 | 1,681 | 13.7% | 5,739 | 46.8% | 4,846 | 39.5% | 12,266 | 100.0% |
| 1998 | 2,188 | 15.0% | 6,751 | 46.1% | 5,692 | 38.9% | 14,631 | 100.0% |
| 1999 | 2,161 | 14.6% | 6,840 | 46.3% | 5,774 | 39.1% | 14,775 | 100.0% |
| 3-year aggregate | 6,030 | 14.5% | 19,330 | 46.4% | 16,312 | 39.1% | 41,672 | 100.0% |
Table 8. Policies Written by Primary Companies
| Policies written | Low-income zip codes | Middle-income zip codes | High-income zip codes | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| 1997 | 11,389 | 12.1% | 43,019 | 45.5% | 40,101 | 42.4% | 94,509 | 100.0% |
| 1998 | 17,503 | 12.1% | 67,643 | 46.6% | 60,038 | 41.4% | 145,184 | 100.0% |
| 1999 | 18,070 | 11.5% | 71,890 | 45.6% | 67,599 | 42.9% | 157,559 | 100.0% |
| 3-year aggregate | 46,962 | 11.8% | 182,552 | 46.0% | 167,738 | 42.2% | 397,252 | 100.0% |
5. Denials and Terminations Occur at a Higher Rate in Low-Income Zip Codes
In low-income zip codes, one application was declined for every 114 policies written; in middle-income zip codes, this rate decreased to one declination for every 154 policies written; and in high-income zip codes, the rate decreased further to one application declined for every 218 policies written. In other words, the declination rate was nearly twice as high in low-income zip codes as in high-income zip codes.
With respect to cancellation rates, one policy was canceled for every 15 policies written in low-income zip codes, while, in high-income zip codes, one policy was cancelled every 26 policies were written. Therefore, the cancellation rate was 1.73 times higher in low-income zip codes than in high-income zip codes. The rate of nonrenewals was similar. The nonrenewal rate in low-income zip codes was, whereas the nonrenewal rate in high-income zip codes was 1:214. The non-renewal rate was thus 1.77 times higher in low-income zip codes than in high-income zip codes.
CONCLUSIONS
There appears to be a relationship between income level and the likelihood of both obtaining and retaining homeowner's insurance. This finding is a potential indicator that insurance companies underserve low-income individuals, since it is unclear whether people rent more frequently in lower income zip codes because they cannot afford to own a home or because they have difficulty obtaining homeowner's insurance (or confront other barriers to home ownership).
A relationship between race and an individual's ability to buy homeowners insurance was not immediately evident. Although more racially diverse zip codes did have proportionately fewer policies sold in them than zip codes which had more predominantly white populations, the differences were not large enough conclude from the data alone that bias exists. On the other hand, the predominance of whites in Minnesota means that even the most racially diverse zip codes in the data set are still around 50% white. Minnesota's largely white population, combined with the large size of zip codes, makes it difficult to pinpoint zip codes where people of color live and to determine at what rate people of color are buying or able to buy homeowner's insurance. Smaller geographical areas need to be examined so it can be more accurately determined where people of color live and how they are served by insurance companies. Without further study, race cannot be discounted as a factor in accessing and maintaining homeowner's insurance.
LSAP found that primary companies wrote a smaller percentage of their policies in low-income zip codes than did secondary companies. This might suggest that primary companies are less responsive to insurance needs in lower income zip codes than are secondary companies. Additionally, because seven of the ten zip codes of color were also low-income zip codes and every high-income zip code was a white zip code, these trends could suggest a bias based on race among primary companies. At the same time, a significant discrepancy remained between low-income and high-income zip codes in rates of policies sold by secondary companies. Though this discrepancy was less pronounced than that exhibited by primary companies, the difference raises similar concerns about the ability of low-income people and people of color to obtain homeowner's insurance from companies, whether primary or not. Further investigation is warranted as to the causes of these discrepancies.
Finally, low-income zip codes experienced a greater rate of declinations, policy cancellations and nonrenewals than did high-income zip codes. Applications were declined, canceled and nonrenewed in low-income zip codes at nearly twice the rate as in high-income zip codes. Further investigation is required to determine the cause of this troubling discrepancy.
In sum, LSAP believes that further study of both specific findings and general trends is warranted. Additional investigation is needed to: (1) understand the reasons why secondary companies wrote nearly 23% more of policies in low-income zip codes than did primary companies; (2) identify whether race is a factor in the ability of persons of color to obtain homeowner's insurance, since one and one-half times as many policies were written per person in white zip codes than in zip codes of color; and (3) understand the reasons why the rate of declinations, cancellations and nonrenewals is nearly twice as high in low-income zip codes than in high-income zip codes.
Generally, more research is needed to:
- Understand the economic and racial implications of the policies and practices of companies offering homeowner's insurance;
- Identify the barriers to obtaining and retaining homeowner's insurance; and
- Understand the role that homeowner insurance practices may play in the ability of people in low-income zip codes and zip codes of color to own a home.
APPENDIX A
Methodology
Data on homeowner insurance activity by zip code data from the calender years 1997, 1998, and 1999 was compiled from individual forms sent to the Minnesota Department of Commerce by insurance companies to create a single database. Eighty insurance companies submitted data in one or more of the three years. The aggregate data was organized by year and listed the number of policies written, policies canceled, policies non-renewed, and applications declined for the 42 zip codes that make up Minneapolis (22), Saint Paul (12) , and Duluth (8). A separate database was created for all insurance companies. The eighty companies submitting data were divided into two groups: those that wrote 100 or more homeowner's insurance policies in any one zip code (the "primary companies") and those that wrote fewer than 100 policies in each zip code (the "secondary companies").
Next, the population, income and racial makeup within each zip code for which data were submitted was determined using 1990 Census data. Data was submitted for 42 zip codes: 22 in Minneapolis; 12 in St. Paul; and eight in Duluth.(6) The zip codes were grouped by income into three categories: low income, middle income and high income. A zip code was considered low income if it fulfilled at least three of the following four criteria: a poverty rate equal to or greater than 25%; eight percent or greater below 50% of poverty; eight percent or greater unemployment rate; and/or at or below $25,000 annual median family income.(7) A zip code was considered middle income zip codes if: the poverty rate was between 10% and 25%; between 4% and 8% of its population was below 50% of poverty; the unemployment rate was between 5% and 8%; and/or the annual median family income was between $25,000 and $35,000. A zip code was considered high income if: the poverty rate was 10% or less; 4% or less its population was below 50% of poverty; the unemployment rate was less than or equal to 5%; and/or the annual median family income was $35,000 or more.
The zip codes were then grouped by their racial makeup into three groups: zip codes of color; mixed zip codes; and white zip codes. A zip code was considered a zip code of color if the population was less than or equal to 75% white.(8) A zip code was considered a mixed zip code if the population was between 75% and 90% white. A zip code was considered a white zip code if the population was more than or equal to 90% white.
Once population and racial data for each zip code were determined, the total number of policies written in low income zip codes was compared to the numbers in middle- and high-income zip codes. Additionally, the average number of policies written per company was determined for each income group. Both the totals and averages were then adjusted for population to ensure that differences in population size were not skewing the results. Finally, t-tests were used to check the statistical significance of the differences between the analytic groups, and an analysis of variance (the eta squared test) was conducted. Finally, the rates at which homeowners insurance was either denied or ended were determined for the income groupings.
For the first round of tests, the data from all companies was used. Before the data was grouped by race and income and tests were performed, the data was manipulated in the following ways:
- Total policies written by zip code: this division looked at the sum of all companies' activity in each zip code.
- Total activity by zip code adjusted for population: each zip code total was divided by the population of the particular zip code to determine the total number of policies written per person in each zip code.
- Company averages by zip code: in this division, the average activity of each company was recorded for each zip code. For example, in zip code x each company wrote an average of 34.2 policies.
- Company averages by zip code adjusted for population: this statistic determined, for example, the average number of policies each company wrote per person in a particular zip code.
The next round of tests solely examined the activity of large companies. The divisions of this data are identical to those used for the aggregate data.
APPENDIX B
Primary vs. Secondary Companies
In each of the three years studied, an average of 30% of the companies were "primary," that is, they wrote 100 or more policies in any one zip code. These primary companies control nearly 90% of all transactions involving homeowners insurance over the three-year data set; over the three years, the primary companies wrote approximately 397,000 policies, while the secondary companies wrote only about 42,000.
Table B1. Primary vs. Secondary Companies
| Primary vs. Secondary (1997-1999) | Primary Companies | Secondary Companies | Total | |||
| frequency | % | frequency | % | frequency | % | |
| Policies Written | 397,252 | 90.5% | 41,672 | 9.5% | 438,924 | 100.0% |
| Policies Canceled | 17,160 | 82.3% | 3,680 | 17.7% | 20,840 | 100.0% |
| Policies Non-renewed | 2,142 | 79.6% | 548 | 20.4% | 2,690 | 100.0% |
| Applications Declined | 1,368 | 52.1% | 1,260 | 47.9% | 2,628 | 100.0% |
APPENDIX C
Analytic Groupings
- By Income Status
1. Poverty Status by Income Groups
As income level increases so does average population and average median family income. On the other hand, as income increases, both the number and percentage of persons living below poverty level decreases, as does the unemployment rate. High-income zip codes are on average larger than middle- and low-income zip codes, yet high-income zip codes have fewer people living below poverty level.
As Table C1 below shows, low-income zip codes have nearly three times as many people living in poverty as high-income zip codes. In low-income zip codes, nearly one-third of residents live below the poverty level (i.e., the mean percentage of residents below poverty level is 32.6% ) while only one-sixteenth (6%) of the population in high-income zip codes lives below the poverty level.
Table C1. Poverty Status by Income Groups
| Income Groups | Mean # of residents | Mean # of residents below poverty level | Mean % of residents below poverty level | Mean # of residents below 50% poverty levels | Mean % of residents below 50% poverty levels |
| Low Income (n = 11) | 14,444 | 4,411 | 32.6% | 1,593 | 11.8% |
| Middle Income (n = 19) | 19,892 | 3,219 | 17.0% | 1,153 | 6.3% |
| High-income (n = 11) | 24,058 | 1,509 | 6.0% | 548 | 2.2% |
2. Median Income by Income Groups
As illustrated in Table C2, the average median family income in high-income zip codes was $8,787 higher than in middle-income zip codes, and $21,998 higher than in low-income zip codes. Low-income zip codes had an average median family income that was less than half that of the high-income zip codes. Finally, the percentage of the population with a household income of less than $10,000 varied considerably among the income groups. More than one-third of the households in low-income zip codes fell into this group while not even one-tenth of the households in high-income zip codes had an income below this level.
Table C2. Median Income by Income Groups
| Income Groups | Mean # of residents | Mean median family income | Mean % of households with income below $10,000 |
| Low Income (n = 11) | 14,444 | $21,152 | 33.9% |
| Middle Income (n = 19) | 19,892 | $34,363 | 19.2% |
| High-income (n = 11) | 24,058 | $43,150 | 9.1% |
3. Unemployment Rate by Income Groups
As Table C3 shows, there was also a significant discrepancy among zip code groups with respect to unemployment rates. The rate in low-income zip codes was nearly 11%, while middle- and high-income zip codes had rates of only 6.4% and 4.1% respectively.
Table C3. Unemployment Rate by Income Groups
| Income Groups | Mean # of residents | Mean unemployment rate |
| Low Income (n = 11) | 14,444 | 10.9% |
| Middle Income (n = 19) | 19,892 | 6.4% |
| High-income (n = 11) | 24,058 | 4.1% |
4. Racial Breakdown by Income Groups
As Table C4 indicates, the average percentage of the population which was white increased as income level increased, while the percentage of people of color decreased. The one exception relates to Hispanics, where the highest percentage lived in middle-income zip codes. More specifically, low-income zip codes were on average only 73.1% white. High-income zip codes, on the other hand, were on average 94.7% white.
Table C4. Race by Income Groups
| Income Groups | Mean % White | Mean % Black | Mean % Asian & Pacific Islander | Mean % American Indian, Eskimo, Aluetian | Mean % Hispanic (any race) |
| Low-income (n = 11) | 73.1% | 14.8% | 6.7% | 4.4% | 2.1% |
| Middle Income (n = 19) | 84.5% | 7.4% | 4.9% | 1.9% | 3.2% |
| High-income (n = 11) | 94.7% | 2.3% | 1.7% | 0.9% | 1.2% |
B. Analysis by Racial Groupings
1. All Indicators
As Table C5 illustrates, the more racially diverse a zip code, the higher the poverty rate. The poverty rate was almost three times as high in low-income zip codes as in high-income zip codes. Additionally, low-income zip codes exhibited twice as high an unemployment rate and a mean income that was almost one-third lower than high-income zip codes. Additionally, the more diverse zip codes had more residents on average.
Table C5. All Indicators
| Race Groups | Mean # of residents | Mean # of residents below poverty level | Mean % of residents below poverty level | Mean # of residents below 50% poverty levels | Mean % of residents below 50% poverty levels | Mean unemployment rate | Mean median family income | Mean % of households with income below $10,000 |
| Zip Codes of Color (n = 10) | 21,970 | 5,805 | 30.7% | 2,128 | 11.3% | 10.3% | $23,770 | 28.7% |
| Racially Mixed Zip Codes (n = 12) | 20,831 | 3,298 | 18.2% | 1,183 | 6.8% | 6.1% | $34,695 | 18.6% |
| White Zip Codes (n = 19) | 17,463 | 1,507 | 11.7% | 525 | 4.1% | 5.8% | $37,168 | 17..2% |
2. Racial Breakdown
Table C6 shows that an average zip code of color had a population that was 65.3% white, a racially mixed zip code was on average 83.9% white, and a white zip code was on average 94.3% white. The next most substantial difference was the black population of white zip codes and zip codes of color. White zip codes had a population that was only 2.3% black, while zip codes of color had a population that was nearly 21% black. In addition, the largest concentration of Hispanics was located in racially mixed zip codes.
Table C6. Racial Breakdown
| Race Groups | Mean % White | Mean % Black | Mean % Asian & Pacific Islander | Mean % American Indian, Eskimo, Aluetian | Mean % Hispanic (any race) |
| Zip Codes of Color (n = 10) | 65.3% | 20.7% | 8.4% | 4.3% | 2.6% |
| Racially Mixed Zip Codes (n = 12) | 83.9% | 6.3% | 6.2% | 1.8% | 3.9% |
| White Zip Codes (n = 19) | 94.3% | 2.3% | 1.5% | 1.5% | 1.2% |
C. Housing Characteristics
1. Housing Units by City
As shown in Table C7, half of all the owner-occupied homes in the three cities combined were located in Minneapolis. Saint Paul had about 36% of all owner-occupied housing, and Duluth had the smallest share with only about 14%. Minneapolis had the largest share of rental units, with nearly 57%, Saint Paul ranked second with about 36%, and Duluth had the fewest, with only about 7%.
Table C7. Housing Units by City
| Housing Units | Minneapolis | Saint Paul | Duluth | Total | ||||
| frequency | % | frequency | % | frequency | % | frequency | % | |
| Owner Occupied | 92,312 | 50.0% | 66,750 | 36.1% | 25,592 | 13.9% | 184,653 | 100.0% |
| Rented | 86,366 | 56.8% | 54,807 | 36.1% | 10,856 | 7.1% | 152,029 | 100.0% |
| Vacant | 12,612 | 54.1% | 7,962 | 34.2% | 2,739 | 11.7% | 23,313 | 100.0% |
| All Units | 191,290 | 53.1% | 129,518 | 36.0% | 39,186 | 10.9% | 359,994 | 100.0% |
NOTES
1. Minn. Stat. § 65A.28, Subdivision 1. The statute requires that: "Each insurer writing homeowner's insurance for property located in the metropolitan area or a statutory or home rule charter city of the first class shall compile and file annually with the commissioner on or before May 1 a report for the preceding calendar year. This report shall contain the following information reported by postal zip code areas for each zip code located in a city of the first class which contains property for which the insurer wrote, declined to write, or canceled homeowner's insurance: (a) the number of policies written; (b) the number of policies canceled; (c) the number of policies nonrenewed; and (d) the number of applications for homeowner's insurance declined."
2. The Legal Services Advocacy Project is a statewide division of Mid-Minnesota Legal Assistance, providing legislative and administrative advocacy, research, and training on behalf of Minnesota's low-income citizens.
3. Calendar years 1997, 1998 and 1999.
4. Of the 42 zip codes, 22 are located in Minneapolis, 12 in St. Paul, and eight in Duluth. Only 41 zip codes were analyzed; the 55402 zip code in Minneapolis was discarded because the census information concerning this zip code was incomplete and possibly inaccurate.
5. See Appendix A for more information on the methodology used.
6. Parts of some zip codes lie outside the city for which the data are being reported. Companies with the ability to segregate policies report only those within a given zip code that are inside city limits. Companies without the ability to distinguish, report all policies written within the zip code, even if some of those policyholders reside outside city limits. The zip code 55402 in Minneapolis was thrown out of the data set because the census information concerning it was incomplete and possibly inaccurate. For example, census data for this zip code did not include the civilian labor force unemployment rate and listed the median family income as zero dollars.
7. The civilian labor force unemployment rate was used.
8. Hispanics were categorized as people of color.