580 URBAN AFFAIRS QUARTERLY/June 1991

representing 1 of these 32 insurers. By 1980 this number was down to 18; 5 areas had no agency location representing any of these firms. In 1960, 28 of the 36 suburban communities had no agency locations representing these firms. In 1980 only 7 had no such representation.

Consistent growth in number of locations or ratio of locations to housing units was experienced in six Milwaukee communities from 1960 to 1970 and again from 1970 to 1980. These communities - Granville, Jackson Park, Kosciusko, Lake, Tippecanoe, and Valley - are mainly at the city's periphery bordering the suburbs. Neither the number of locations nor the ratio increased in seven community areas that are concentrated near the central part of the city-Garfield, Grand Avenue, Halyard, Lakeside, Midtown, Muskego Avenue, and Riverside West. Other areas within the city experienced an inconsistent pattern during the two decades, but most of these lost agencies between 1970 and 1980.

RACIAL IMPLICATIONS

As previously discussed, one evident trend in the changing location of insurance agencies is growth in the suburbs juxtaposed to decline in the city. Another feature of recent change is growth in predominantly white areas juxtaposed to decline in minority areas. Insurance companies display a preference for predominantly white communities as locations for their agencies. The racial effect is in part a result of the concentration of the metropolitan area's minority population in the city coupled with the industry's emerging preference for suburban locations. But even within the city, racial composition of neighborhoods is associated with agent location. Like the prosuburban bias, the racial bias cannot be explained strictly in terms of other population or housing characteristics.

A striking feature of the 15 community areas in the city of Milwaukee that lost agency locations during either decade is that their populations in 1980 were 31% nonwhite compared to 16% for the remaining 8 areas. Sherman Park is illustrative. Many community organizations in this neighborhood have worked hard to retain its identity as one of the few racially integrated areas in the city. Residents frequently have charged local realtors, lenders, and insurers with redlining the neighborhood and undermining their efforts to stabilize their community (Saltman 1990). Between 1970 and 1980 Sherman Park's minority population increased from 1% to 24%. The number of insurance agency locations dropped from 22 to 9 during these same years.

The racial effects of agency location appear to differ between the city of Milwaukee and its suburban ring. In the suburbs, the range in percentage

Squires et al. / INSURANCE REDLINING 581

nonwhite is only from 0.3 to 6.4, and the correlation between the number of agency locations and race is statistically insignificant for 1980. Among the 23 city community areas, the correlation between agency locations and race for 1980 is -.50 and is statistically significant. This relationship also was negative and significant (-.46) for 1970; the correlation (-.23) for 1960 was not significant.

The number of agency locations within the city of Milwaukee is associated with the age and condition of housing, the number of owner-occupied housing units, and median family income. The most consistent predictor of agency location across time in both the city and suburbs is the number of owner-occupied dwellings. A question that arises is whether the association between race and agency location is an artifact of income, housing condition, or related demographic characteristics of the city's neighborhoods. Multipleregression equations incorporating these variables and race were calculated for each year, separately for city community areas and suburbs (see Table 2). The equations show that race indeed had an independent effect in the city in 1980 (beta = −.673) and in 1970 (beta = -1.09). In 1960, however, the racial effect was not statistically significant.

There is evidence that the racial disturbances that Milwaukee and many other large cities experienced in the 1960s discouraged insurers from selling policies in inner-city markets (President's National Advisory Panel on Insurance in Riot Affected Areas 1968). It appears these disturbances (or other racial factors) also influenced the agency location decisions of the property/ casualty industry in Milwaukee, and the effects continued to be felt in subsequent years.

In the city of Milwaukee, the nonwhite population increased from 6.0% of the total in 1960 to 25.6% in 1980; for the metropolitan area, the increase was from 2.5% to 10.9%. The rapid increase in the minority population could have assisted insurers attempting to serve the minority market. Instead, that growth was apparently perceived as a threat to the stability of the market in the inner city, thus further discouraging service to minority areas. Since 1960 race has become a significant factor in explaining insurance agency location in the city.

Race clearly is associated with agency location, and the effect of race remains significant even after accounting for the effects of income, housing condition, owner-occupancy, and related neighborhood characteristics that presumably influence the location of insurance agencies. In the city of Milwaukee, racial composition of neighborhoods is significantly associated with the pattern of insurance agency location.