Jump to Body Text

An evaluation of Minnesota's shelter program for battered women

A report to the Minnesota legislature

Stephen Coleman, Ph.D.
Center for Applied Research & Policy Analysis, School of Law Enforcement, Criminal Justice, and Public Safety, Metropolitan State University, St. Paul, Minnesota

Published: March 2001

Executive summary

This report is about Minnesota's shelters for battered women and their children. Shelters include residential facilities and a network of hotels, motels, and safe homes used by shelter providers. Residential shelters offer, at a minimum, security, lodging, and meals, and are available 24 hours per day. Shelter staff and volunteers also advocate for battered women and offer a variety of services to help women escape abuse and move on to a better life. Although shelters are independent, nonprofit entities, they receive most of their financial support from the state.

The state reimburses shelters for the daily (per diem) costs of women and their children who stay in a shelter. State payments are limited to expenses for food, lodging, and safety. Up to FY2001, there was no limit on total state spending, however, and it grew rapidly. Total per diem reimbursements were $3.2 million in FY1990 but $20.9 million in FY2000, with a rapid acceleration in costs after 1997. This led the Legislature in 1999 to put a cap on total annual spending at $17.979 million. As part of welfare reform, responsibility for the reimbursement program was transferred from the Department of Human Services to the Minnesota Center for Crime Victim Services (MCCVS) in the Department of Public Safety. MCCVS then set a fixed reimbursement rate for all shelters, whereas previously some had much higher costs than others and received higher per diem reimbursements. The Legislature called for this study to learn more about what affects the demand for shelter beds, whether the state has sufficient shelter capacity, and why the costs have increased so dramatically.

To address these issues, we examined statistical, financial, and survey data about the state's shelter system and compared it with other states. The analysis led to these main conclusions: (1) The state generally has enough shelter capacity except in some regions. (2) The rapid increase in costs resulted mostly from an increase in how long women were staying in shelters, but the widespread shortage of low-income housing and the changing characteristics of shelter clients contributed to the increase in length of stay. And (3) the funding cap has created serious financial problems for many shelters.

The report concludes with several recommendation to improve the financial stability of the shelter system, including a partial restoration of funding, more authority for MCCVS to manage payments to shelters, and a change from per diem funding to contracts with shelter providers to pay for their services. The report also reviews research evidence on the effectiveness of shelters and the criminal justice system as deterrents to domestic violence.


The advocacy movement for battered women began in the 1970s with women in Minnesota among the leaders and first to establish a shelter for battered women in the United States. The state's independent, non-profit residential shelters, safe houses, and similar programs that use hotels and motels, give refuge and other services to any woman who has been assaulted by her spouse, male relatives, or other males with whom she has had a significant relationship. [1]The shelters also take in the children of a battered woman. At a minimum, residential shelters provide security, lodging, and meals, and are available 24 hours per day.

State government became a financial supporter of shelters in 1977, which increased the number and capacity of shelters through the 1980s and 1990s. Almost all residential shelters receive both an annual grant of $103,000 or, in rural areas $104,000, and per diem payments to cover basic costs for each woman and child who spends a night in a shelter. Other types of shelter providers receive only per diem payments, but they also have grants for community-based advocacy to support their activities [2]. In the past, per diem payments were essentially an entitlement program for shelters, and state expenditures increased steadily to about $20.9 million in FY2000. In its 1999 session, however, the Legislature capped total state spending at $17.979 annually for fiscal years 2001 to 2003, and a new payment system was set up to limit the funds available to each shelter in order to stay within the $17.979 million [3]. This situation has caused serious financial problems for some shelters and created uncertainty as to how much financial support they will receive in the future, what level of services they will be able to offer, and how the shelter system might be expanded. Despite its financial support for shelters, under current law state government has little control over their establishment, expansion, and operation.

In 2000, the Legislature requested the Center for Applied Research and Policy Analysis at Metropolitan State University to do a comprehensive study of Minnesota's shelter system in cooperation with the Minnesota Center for Crime Victim Services (MCCVS) and report the results by March 15, 2001 [4]. This report fulfills the requirement. The issues dealt with in the report include the following:

  1. Does Minnesota have enough shelter capacity?

  2. Are all areas of the state adequately served by the shelter system?

  3. What accounts for the steady increase in costs of the shelter system despite a recent decline in the number of women sheltered?

  4. Are shelters meeting their primary goal of providing refuge to women in crises?

  5. How effective are shelters?

  6. How does Minnesota compare with other states in shelter costs and capacity?

  7. What has been the impact of the change in funding and would other funding arrangements work better?

To answer these questions, we examined statistical, financial, and survey data on shelters and their residents. We analyzed data on the incidence of domestic assault and the proportion of abused women who seek help in shelters. We reviewed how shelters establish their per diem expense rates and collected comparative data on shelter usage and costs from other states and Canada. To assess effectiveness, we reviewed research that others have done on shelters. And we held three public meetings to discuss the current shelter situation with persons who operate shelters and with other advocates of battered women. We begin the report with a general description of the state's shelter system and recent trends in shelter usage.

Overview of Minnesota's shelter system

The main purpose of a shelter is to provide safety to a woman and her children in a crisis situation of domestic abuse. Some women who come to shelters, however, are not in an immediate crisis but were victims of earlier domestic abuse or violence. Shelters determine a woman's eligibility to stay and for how long. In addition to safety, shelters typically offer their clients help obtaining social services, protection orders, medical services, transportation, and housing. Shelters also assist many women by phone and through outreach or educational programs, and some provide training to personnel in the criminal justice system. Shelters are nonprofit agencies supporting themselves with a combination of state and federal funds, local contributions, and volunteer efforts. The majority of financial support comes from state funds, however, and some shelters rely on state funds almost exclusively.

Minnesota has 25 residential shelters (and one soon to open), 33 hotel-motel networks, and 15 safe homes, mostly in less-populated areas of the state. These facilities have a licensed capacity of 653 beds, but some shelters routinely operate above capacity [5]Most women seeking refuge are placed in a residential shelter; less than 3 percent of state per diem funding goes to safe homes and hotel-motel networks. A few shelters specialize in certain social groups, such as Asian-American women, Native-American women on reservations, and Hispanic women. In the past, the state also reimbursed shelter costs for women from Minnesota who stayed in any of five shelters in North Dakota and Wisconsin. Reimbursement for shelters in Wisconsin is tentatively being phased out, however; only the shelter in Fargo may still be available for reimbursement for Minnesota residents in the future. There are no reciprocity agreements between Minnesota and other states for shelter expenses of nonresidents.

The number of women eligible for state reimbursement who stayed in a shelter or other safe facility increased steadily and substantially through the 1980s and up to 1995 when it reached a peak of 6,100 women [6]. The trend from 1990 to 1999 is shown in Figure 1. Since 1995, the number of women using shelters has decreased each year, falling to about 4,900 in 1999. Partial year data for FY2001 indicates a similar level of usage [7]. The number of children in shelters is about 36 percent greater than the number of women; average family size is about 2.36 [8].

Figure 1. Shelter Trends 1990-1999

line graph

Figure 1 also shows the trend in cost per family staying in a shelter. These costs were adjusted for inflation to 1999 dollars. In 1999, the average cost per family was $3,519. Up to 1997, costs showed only a slight increase when adjusted for inflation, but after 1997 they increased much more rapidly. Total per diem state expenditures under General Assistance increased steadily from $3.2 million in 1990 to $20.9 million in FY2000.

The reasons for the decline in women coming to shelters are not clear. It is possible that there has been a decrease in domestic abuse, but data are inadequate to verify that. Another possibility is that the strong economy gave women more opportunities to get out of abusive situations. It is well known that economic independence is one of the most important factors in a woman's ability to do that [9]. The idea of a decline in abuse is contradicted, however, by a steady increase in arrests for violations of orders for protection in the 1990s [10]. The fact that shelter costs increased at a time when the number of women sheltered was declining raises important questions that we take up later.

Legislative history

State support for battered women's shelters began in 1977 with direct funding to shelters [11]. The Legislature also began the practice of making battered women in shelters eligible for per diem payments as General Assistance recipients [12]. County human service departments reimbursed shelters directly for their expenses. This created an entitlement program for shelters, as there was no fixed limit on state spending. The more people who stayed in a shelter, and the longer their stays, the more money that a shelter received in reimbursement. In 1980, Minnesota statutes were amended to clarify that the only costs eligible for reimbursement were for "maintenance and security" [13]. Payments to individual shelters were not uniform or limited to a specific per diem rate under this payment system. In fact, each shelter had a different per diem rate. This resulted in some shelters with higher costs receiving significantly higher per diem payments.

In 1997, responsibility for per diem payments was transferred from the Department of Human Services (DHS) to the Department of Corrections, and in 1998 the battered women's shelter program was transferred to the Minnesota Center for Crime Victim Services (MCCVS), now in the Department of Public Safety. To allow time for an orderly transition, however, administration of per diem payments by MCCVS was delayed by the Legislature until July 1, 2000 (FY2001). In 1999, the Legislature capped spending at $17.979 million for three years beginning with FY2001. This amount was based on a forecast for FY2001 by DHS before actual spending of $20.9 million in FY2000 was known [14].

Under the new funding arrangement effective on July 1, 2000, payments to shelters do not depend on a woman's eligibility for General Assistance or other welfare payments, and county human service departments are no longer involved in per diem payments to shelters. Payments go directly from MCCVS to shelters, and reimbursements are limited to costs of food, lodging, and safety [15]. In place of the previous unlimited entitlement to reimbursement, MCCVS sets up an annual allocation or reserve for each shelter depending on the total of available funds, the type of shelter, the shelter's historical average occupancy rate, and licensed capacity [16]. All but five shelters received the same per diem rate of $75 for FY2001; the others had lower rates [17]. A shelter invoices MCCVS for reimbursement, but if it exceeds its allotment for the year, it has to bear the unreimbursed expenses. Because some shelters previously received higher per diem rates, they must now operate on reduced funding, which has created significant programming and financial problems for them. MCCVS also has federal funds that it can distribute to shelters, but this money generally cannot be used in place of state funds.

Under the revised payment system, if a new shelter were opened it would be eligible for payment from the general appropriation. Because the appropriation is capped, however, its per diem share would come at the expense of other shelters. The same might happen if an existing shelter increased its capacity. MCCVS does not believe that it has legal authority to limit expansion in the shelter system or to deny reimbursement to a new or expanded shelter. A shelter only needs to apply for state designation and agree to meet certain requirements to be reimbursed [18]. About half of the shelter operators who responded to a survey from MCCVS in December 2000 expressed the opinion that the state should put a moratorium on additional shelter beds pending a restoration of funds. These issues make it important to know whether Minnesota has enough capacity in its shelter system and whether the current capacity is well used.

Estimated need for state shelter capacity

Nationally, there is no information on how many shelter beds are needed for a given population or area. In many places, the demand far exceeds the supply, and women have great difficulty getting into a shelter. Minnesota, however, has a very well developed shelter system and space is almost always available in some shelters. So it is a reasonable question whether Minnesota needs additional capacity.

The need for additional shelters has been greatly reduced by the success of the Day One program [19]This is a computer network of most of the state's shelters that tracks the availability of shelter space. Day One began in 1995 with funding from Allina and the United Way, and it continues to be supported financially by Allina. It started with shelters in the Twin Cities area, but has expanded to virtually all the state's shelters, with additional shelters and safe homes expected to join in the near future. If a woman seeks refuge at a shelter that is full, the system can immediately locate available beds in other shelters and link the woman to another shelter over the phone. If the available bed is a good distance away, Day One can assist with transportation. A distant placement, however, might cause additional difficulties for a woman who needs to be in a certain location for a job, court appearances, or other appointments, or has children in school.

According to Day One, no woman has been turned away from a shelter in the Day One system for lack of a bed since this program has been in operation [20]. In a survey of shelter providers by MCCVS, however, some shelters in the Twin Cities area reported turning away women who were eligible to stay. It is not clear how many of these women subsequently found a bed through Day One, or whether some may have declined to go to another shelter. Day One's success might suggest there is no need for additional shelter beds in the state. But battered women who live in areas of the state without shelters might be deterred from seeking help in a shelter if none is nearby; they would not show up in the Day One statistics.

Here we make two estimates of the need for shelter beds. The first estimate is based on the incidence of domestic violence and the proportion of abused women who seek refuge in a shelter. The second estimate is based on the relationship between past occupancy rates in Minnesota shelters and the populations of the counties where the shelters are located.

Incidence-Based Estimate.The incidence rate for domestic violence refers to the percentage of a population who have been victims within a specific interval of time. Sometimes a "lifetime" victimization rate is also calculated. Here we use the incidence rate for intimate partner violence against women during the previous 12 months. This data is from a representative national sample of 8,000 women 18 years of age or older surveyed in 1995 and 1996 by the National Violence Against Women (NVAW) Survey [21]. Incidents were counted if the perpetrators were current or former dates, spouses, or cohabiting partners, including partners of the same sex. The inclusion of dates and same-sex partners makes this a relatively inclusive survey of violence.

The national survey reported that 1.5 percent of women were victims of rape or physical assault by an intimate partner during a 12 month period. The lifetime rate, which is to say, up to the ages of the respondents, was 25 percent. These rates are greater than estimates from a different survey, the National Crime Victimization Survey, which put the annual rate at 0.77 percent in 1998 [22]. The methodology of the latter survey, however, probably leads to an underestimate of the incidence rate, so we will use the 1.5 percent rate in further analysis. None of these statistics include threatening behavior that did not lead to physical contact.

We can compare the 12-month national victimization rate with data for Minnesota. Minnesota Planning did a survey of Minnesota citizens about their experiences as crime victims in 1998 [23]The survey drew on a sample of 4,500 randomly selected people 16 years of age or older who had a driver's license. People were asked: "In 1998, did your spouse, significant other, partner or other family member injure you with an object or weapon or hit, slap, push or kick you?" About 3 percent of women answered "Yes" to this question. This is about twice the rate observed nationally, but there are important differences between the state and national surveys. The Minnesota assaults included incidents when the perpetrator was a family member but not an intimate partner. And 23 percent of the perpetrators were juveniles (under age 18). So the Minnesota survey appears to somewhat overestimate the incidence of intimate partner violence.

National and state incidence rates are slightly lower than those of other countries. Rates of physical assaults by spouses and former spouses within a 12-month period in Canada are 3 percent, in Australia 2.6 percent, and in England and Wales 4.2 percent [24]. Statistical errors from survey sampling and differences in how the surveys were done may account for some of the differences in the statistics across countries.

The NVAW survey shows that it is typical for an abused woman to be assaulted many times by the same intimate partner. The average number of physical assaults is about 3.4 (+/- 0.6) per year, and the average duration of an abusive relationship is 4.5 years. One-fourth of abusive relationships last more than five years. About 20 percent of abused women are victims of 10 or more physical assaults by the same person. Among women who are raped, 15 percent are victims 10 or more times. Most physical assaults are relatively minor and consist of pushing, grabbing, shoving, slapping or hitting. Of women assaulted, 41 percent were injured and of those injured, 28 percent received medical care (11 percent of victims), typically in a hospital emergency room or outpatient clinic.

In 1999, the population of women in Minnesota 18 years of age or older was 1,800,000. At a 1.5 percent incidence rate, we estimate that about 27,000 women are battered each year. Of this number, about 11,070 are injured and 3,100 seek medical care. Of the 27,000 annual victims, we know that about 5,000 to 6,000 have been sheltered in recent years. This means that roughly 20 percent of abused women are in shelters during the year in Minnesota.

Does a 20 percent shelter usage rate mean that Minnesota has enough shelter space? Not much information is available on the percentage of abused women who seek refuge in a shelter. The national and state surveys on victimization did not ask women if they had been in a shelter. Good data is available on Canadian shelters, however, from a 1997/98 Transition Home Survey [25]. Canada has an extensive system of 422 shelter and transitional housing facilities for battered women, and the rate of shelter usage is not likely to be much constrained by a shortage of shelters. Over a 12-month period in 1997 and 1998, a total of 43,600 women and 47,200 children were admitted to shelters out of a population of 12.3 million Canadian women ages 15 and above. As Canadian surveys show, about 3 percent of women-or 369,000-are victims of violence from a current or former spouse within the last 12 months. Therefore about 12 percent of women victims (43,600 of 369,000) had entered a shelter during the 12 months. This is a significantly lower rate of shelter usage than we estimate for Minnesota. Even if we have underestimated the abuse rate for Minnesota, the state still compares favorably with Canada in the proportions of abused women sheltered.

Wisconsin also has a lower rate of shelter usage than Minnesota. Wisconsin has 35 shelters, which gave refuge to 3,072 women in 1999 [26]. This total is 63 percent of the number of women sheltered in Minnesota that year, although the population of Wisconsin is about 10 percent greater than Minnesota. Wisconsin also reported that about 1,700 women were turned away from shelters. If these were added to the total, however, the usage rate would still be lower than Minnesota's.

A study of 6,600 women in shelters in Texas found that 14 percent had previously visited a shelter, and this happened in only 3 percent of all abuse incidents these women experienced [27]. The women tried, on average, five other help-seeking strategies before coming to a shelter, and 71 percent had left home previously before coming to a shelter. In general, research shows that women are much more likely to contact or seek help from relatives, friends, or the police than to seek refuge in a shelter. To cite another example, a study of over 400 abused women in North Carolina found that a shelter was the least used of services available to the women [28]. When battered women were asked in another survey why they did not seek help in a shelter, only a minority said that it was because shelter space was unavailable or because they did not know of a shelter [29]

County Occupancy-Based Estimate.A second method of evaluating shelter availability is to develop a statistical model of the relationship between county population and shelter use in counties that usually have shelter beds available. We based the analysis on nine counties: Anoka, Beltrami, Blue Earth, Crow Wing, Dakota, Goodhue, Olmsted, Otter Tail, and Washington. We compared population to average shelter occupancy rates for 1998 and 1999 as reported to MCCVS. The average occupancy rate was 82 percent for shelters in these counties, suggesting that there was usually enough capacity to meet demand. (Hennepin and Ramsey counties were not included because they are less representative of other counties and their shelters are usually full or above capacity.) Statistical regression analysis gives an estimate that, on average, the number of beds used equals a base number of 9.2 beds plus an additional 0.82 beds for each 10,000 population. Statistically speaking, this mathematical model fits the recent use of shelter beds in the nine counties very well [30]. The estimated need for beds is represented by the line in Figure 3; the closeness of county points to the line indicates the accuracy of the model. Note that Olmsted County is somewhat above the line, meaning it used more beds than one might expect for its population. But Olmsted has a unique International Shelter for immigrant battered women and their children in addition to a conventional residential shelter. The statistical model may slightly overestimate the needs of the smallest counties. For example, the shelter in Otter Tail County has a capacity of 10 but has had an occupancy rate of 58 percent.

Figure 2. Populations and Shelter Beds

line graph

From the national victimization survey data, a hypothetical Minnesota county with a population of 100,000 would have about 555 women over the age of 18 who are domestic violence victims per year. The statistical model of shelter occupancy for a county of 100,000 population suggests a demand for at least 17 beds. A shelter of capacity 17 that is full for a year would have 17 x 365 = 6,205 person-days capacity per year. Given the current average stay in a shelter of about 19 days and an average family size of 2.36, one can calculate the potential number of women served each year. It would be 6,205/ (19 x 2.36) = 138 women. So based on the experience of shelters in counties outside the Twin Cities, about 25 percent (138/555) of battered women would be sheltered each year in this hypothetical county at current occupancy rates. This estimate is a bit larger than the 20 percent estimate from state-level data, suggesting that there may be a need for some additional bed capacity statewide, although shelter capacity is adequate in many areas of the state where there are residential shelters.

Regional needs for shelter capacity

The need for additional shelter capacity seems most apparent in two or three large regions of the state. This can be seen on a map of the location of shelters (Figure 3). One region includes counties along the western border of the state. Another is a group of fast-growing counties partially encircling the Twin Cities metropolitan area, including Chisago, Isanti, Sherburne, Wright, Carver, and Scott counties. Counties in the southeastern corner of the state may also be underserved. Residents of counties without residential shelters can seek refuge in other counties, or use the Day One program to find a shelter opening, but as the distance to a shelter increases women are less likely to use it.

In a survey by MCCVS, shelter providers were asked about the need for additional shelters in the state. Although some respondents hesitated to recommend additional shelters because of current funding problems, several areas of the state were cited as needing more shelters. The reported needs were for more residential shelter capacity in the Twin Cities metropolitan area-where shelters tend to operate at or near capacity-and more safe homes and hotel-motel networks in rural Minnesota.

Figure 3. Battered Women's Shelter Programs

map of minnesota indicating shelters

Southwestern Minnesota was specifically mentioned by a few respondents as needing more shelter beds, although a shelter provider at a public meeting suggested that transitional housing for battered women might be more useful in that region. There was a residential shelter in Marshall at one time, but it was closed in favor of the current hotel-motel services. Shelter providers also stated that in some parts of the state it is becoming more difficult to find volunteers willing to open safe homes. This may require more spending for motels.

To estimate the potential demand for shelter beds regionally, we use the statistical model that relates shelter occupancy to county population, as discussed previously. The estimate is that a county with a population of 100,000 needs at least 17 beds. For a population of 50,000, it would be about 14 beds, which is a few beds more than are currently being used in counties that have approximately that population. In practice, these size estimates should be increased slightly so that the occupancy rate is not always at 100 percent.

The need for a shelter also depends on the minimum number of beds that a shelter must have to operate cost-effectively, assuming that occupancy is near 100 percent. It's not clear, for example, if 10 beds would be a cost-effective size, considering that a shelter must be staffed around the clock. If the smallest shelter were to have 15 beds, the statistical analysis implies that a county would have to have a population of about 70,000 to support a shelter of that size.

Let's first consider six counties in the outer ring of the Twin Cites metropolitan area: Chisago, Isanti, Sherburne, Wright, Carver, and Scott. With a total population of 368,000 (Table 1), none of these counties has a residential shelter or access to a hotel-motel shelter in the county; Scott County has only a small safe-home program. On the basis of population, Wright and Scott counties have the greatest need for shelters. Each could support a residential shelter of about 16 or 17 beds. In placing new shelters, however, one must also consider accessibility. By careful siting of two or three shelters across these six counties, it might be possible to strike a good balance between need and accessibility, while keeping costs down.

Table 1. Population in "Outer-Ring" Counties

County Population
Wright 87,800
Scott 77,900
Carver 66,200
Sherburne 63,200
Chisago 42,000
Isanti 30,800
Total 368,000

Another area of the state not well served by residential shelters, and with minimal access to safe homes or shelter in a hotel-motel network, are counties along the western border of the state, including both the northwest and southwest corners. Clay County with a population of 53,000 is the largest of these counties but is probably not quite large enough to support a shelter. Residents of Clay County and neighboring counties, however, can use the shelter in Fargo. This shelter, therefore, should continue to receive support from the Minnesota shelter program. But if the time comes when Minnesota residents are no longer able to gain access to the Fargo shelter, consideration should be given to establishing a shelter in Moorhead.

The aggregate population of nine counties in the southwest corner of the state-Lincoln, Lyon, Redwood, Pipestone, Murray, Cottonwood, Rock, Nobles, and Jackson-is 125,000. Potentially this area would support a shelter of about 20 beds. Two hotel-motel programs in Lyon and Nobles counties now serve these counties. For residential shelter, a woman typically must go quite a distance-to Willmar, for example. A residential shelter, however, would be able to offer more and better services to battered women than a hotel-motel program. If a residential shelter were opened, the hotel-motel program might still be continued in order to have some additional capacity and for instances when distance to shelter would be a factor.

In the northwest corner, the combined populations of Roseau, Kittson, Marshall, Polk and Pennington counties are 77,000. There is a shelter in Pennington now, as well as safe home and hotel-motel programs, which appear to have adequate capacity given the population of the regions. The low population density of the northwest region means, however, that some women have to travel a good distance to find refuge.

The southeast corner of the state-Winona, Houston and Fillmore counties-has a combined population of about 81,000. Minnesota women from this area have been able to use a shelter in LaCrosse, but MCCVS is ceasing financial support for this program. The effects of that decision are as yet unknown. These counties continue to be served by hotel-motel programs and safe homes, and have access to shelters in other counties through the Day One program, but the population is large enough to support a permanent shelter. Whether residents would find the development of a residential shelter better than the current programs is an open question.

Overall, the analysis suggests that Minnesota has close to enough shelter beds except in some areas. The fact that the number of women staying in shelters has declined since 1996 by about 1,200 is further indication that adequate space is potentially available for more women. Some small population groups, however, may not be adequately served by the shelter system. These include older women, who may not feel comfortable in a shelter population of young women with children, women with disabilities, women with large numbers of children, and women of color living outside the metropolitan area, who may need a more culturally sensitive shelter than would be available locally [31].

The availability of space in shelters also depends on how long people stay there and the relative priority given to women who are in an immediate crisis when determining admission to a shelter-the next issues that we consider.

Length of stay

From 1998 to 1999, the state's per diem payments to battered women's shelters increased from $14.295 million to $17.163 million-a 20 percent increase. Over the same time, the number of women sheltered decreased from 5,184 to 4,872-a 6 percent drop. What accounts for these changes? Table 2 has a comparison of shelter spending and usage for these two years. Analysis of shelter occupancy and cost data from MCCVS shows that most of the increase in costs was the result of an increase in average length of stay of shelter residents from 15.7 days to 18.7 days-a 19 percent jump. If per diem costs had stayed the same and only the average length of stay increased, state spending for the number of residents in 1999 would still have increased by about $2.6 million, which is 90 percent of the actual increase.

Length of stay remains a major component of cost under the new funding plan for shelters. The cost to the state of women staying one more day in a shelter, on average, amounts to about $885,000 per year. This figure is based on 5,000 women sheltered per year with an average family size of 2.36 and a per diem expense of $75 per person.

The increase in average length of stay has been a continuing pattern since the 1980s. Table 3 offers a comparison of the state shelter system in 1999 and 1983, when the average length of stay was 11 days. The increase in state expenditures over this time has been primarily a result of increases in the number of shelters, the number of women staying in shelters, and average length of stay. In earlier years, there was also a shift of costs from some counties to the state.

Table 2. Shelter comparison, 1998 and 1999.

Measure 1998 1999 Change
Total State Spending (millions) $14.295 $17.163+20%
Average Monthly Cases432406- 6%
Total Cases (Women Sheltered)5,1844,872- 6%
Average Cost per Case $2,761 $3,519+27%
Total Person-Days 192,000 215,000+12%
Cost per Person per Day $74.50 $79.80+ 7%
Average Person-Days per Case (Women and Children)37.044.2+19%
Estimated Average Length of Stay15.7 days18.7 days+19%

Notes: Analysis includes residential shelters, safe homes, and hotel-motel shelters. A case refers to a woman and her children, if any, staying in a shelter, and whose per diem costs were paid to shelters by the Department of Human Services. Average length of stay was calculated by dividing days per case by the average family size of 2.36 from MCCVS data. It was assumed that this number remained approximately the same through 1999. Data sources: Minnesota Department of Human Services and MCCVS.

Table 3. The state's battered women's shelter program in 1983 and 1999.

Measure 1983 1999
Women sheltered2,8004,900
Average length of stay11 days17 days
Average family size2.32.4
Number of shelters*1724

*In 1999 there were also five shelters used in border states and 27 additional safe homes and hotel-motel shelter programs. Data sources: Minnesota Department of Corrections, "Data Summary Report for Calendar Year 1983, Update 2." Winter, 1984; and MCCVS.

One explanation for the increase in length of stay in Minnesota shelters might be that shelters were trying to keep their occupancy up to gain the most revenue from the state. Per diem reimbursement creates an incentive for shelters to allow women to stay for longer periods, at least while there is no pressure to admit new residents. If fewer women are seeking refuge, a shelter may be less aggressive about moving clients into other housing. We have no evidence that this is happening, but the possibility raises questions about incentives in the current reimbursement system. A more likely explanation emerges as we analyze changes in the characteristics and problems of women staying in shelters and the lack of affordable low-income housing.

Characteristics of shelter residents

To understand better what is causing the increases in length of stay, we examined the characteristics of women who stay in shelters. Periodically, the Wilder Research Center has conducted surveys of battered women's shelter residents on a single day of the year. Table 4 compares women in metropolitan area shelters on October 24, 1991 with those in shelters on October 23, 1997. (We limit this discussion to the metropolitan area because there were only 36 women in shelters on that day in 1997 in Greater Minnesota-too few for statistical analysis.)

Table 4. Comparison of women in battered women's shelters in the metropolitan area in 1991 and 1997.

Characteristic 1991 1997
Count of Women92126
Count of Children126192
Never Married49%66%
Lived Outside MN Before Age 1645%64%
Lived in MN Less Than 1 Year13%26%
Abuse Was 1 of 2 Main Reasons Left Last Housing80%57%

Note: Differences from 1991 to 1997 are statistically significant at p was less than 05. Data sources: surveys administered in battered women's shelters on October 24, 1991 and October 23, 1997. See Greg Owen, June A. Heineman, and Michelle R. Decker, "Homelessness in Minnesota," Wilder Research Center, June 1992; and Greg Owen, et al., "Minnesota Statewide Survey of Persons Without Permanent Shelter," Wilder Research Center, June 1998.

The comparison shows an increase from 1991 to 1997 in women who were Black, never married, and more recently came to Minnesota. In 1997, the women also were less likely to report that domestic abuse was one of two main reasons that they had left their previous housing. In fact, over 40 percent of women in metropolitan area battered women?s shelters did not cite abuse as a primary reason for leaving their last housing, although they reported being in a prior abusive relationship.

In Table 5 are more characteristics of women in metropolitan area battered women's shelters as gleaned from the 1997 Wilder survey. Some women reported drug and alcohol problems and 30 percent had a diagnosis of major depression in the past two years. About 40 percent or more reported various credit or eviction problems related to their previous housing, and 39 percent said that their rental history contributed to their being in the shelter. About 79 percent did not have a job, and of those unemployed, one-third had been unemployed for over a year. The main source of income of shelter residents was AFDC, TANF, or other government assistance, and the median income in the previous month was about $600-too little for most women to support their families.

Difficulty finding affordable housing seems to be the main reason for the increasing length of stay-an opinion voiced by many shelter providers [32]. And a shortage of low-income housing exists throughout much of Minnesota. This problem is compounded by the low income of most women in shelters and, in the metropolitan area, by the eviction records of many shelter clients. Some of their difficulties were likely caused by the men who were abusing them, but is also clear that many women had serious problems that would make it difficult for them to find and keep housing in the best of times. Shelter advocates are generally able to find housing for residential clients with poor rental histories, but it takes longer, resulting in more costly stays in a shelter.

Table 5. Characteristics of women in battered women's shelters in the metropolitan area in 1997.

Characteristic Percentage
Had lived in an emergency shelter as an adult32%
Had lived in a drug or alcohol treatment facility as an adult18%
Had lived in a detention center as an adult10%
Had lived in a mental hospital as an adult5%
Had diagnosis of major depression in past 2 years by doctor or nurse30%
Had been without a permanent place to live 4 to 12 months19%
Had been without a permanent place to live more than one year13%
Had difficulty getting or keeping housing because of credit problems48%
Had difficulty getting or keeping housing because of eviction (unlawful detainer)41%
Had difficulty getting or keeping housing because of number of children30%
Had difficulty getting or keeping housing because of criminal history10%
Rental history problem contributed to being in battered women's shelter39%
In the past 12 months was slapped or threatened by intimate partner91%

Data source: a survey administered to 126 women in battered women's shelters in the metropolitan area on October 23, 1997. See Greg Owen, et al., "Minnesota Statewide Survey of Persons Without Permanent Shelter," Wilder Research Center, June 1998.

As in Minnesota, a one-day survey of residents in battered women's shelters Canada found that 77 percent of women said they were there to escape abuse, but the remainder were there mostly because of housing problems [33]A study of women in battered women's shelters and homeless shelters in Phoenix, Arizona reached a similar conclusion [34]Many women in battered women's shelters were there mainly because they needed a place to live and financial support for themselves and their children. Other states have also experienced increases in length of stay related to housing shortages. In Wisconsin in 1989 the average stay was 11 days [35], but in 1999 it was 16 days [36]. In Seattle, battered women's shelters are full and stays are getting longer because of the time it takes to find an affordable apartment [37]. Most Seattle shelters now allow a three-month stay instead of one month. The result is fewer women served and many turned away in emergency situations.

Homeless shelters in Hennepin County have also reported an increase in length of stay for families over the last few years. Average length of stay increased from 23 days in 1997 to 44 days in 2000. At the same time, the average number of families in emergency shelters increased from 201 to 401.

One might hope that having women stay for a longer time in a shelter creates more opportunity to deal with their problems. If this is true, the time spent might be worth the added expense. This brings us to the question of the effectiveness of shelters.

Shelter effectiveness

When a woman enters a shelter, the shelter staff works with her to develop a plan of action for her time in the shelter. The plan addresses the issues that she cites as important to enhance her safety and move her on to a better living situation. To help her meet her goals, she will be offered assistance, support, options for safe housing, and advocacy.

Not much research has been done on the effectiveness of the services provided to women by battered women's shelters or whether shelters reduce the incidence of domestic violence. Research by Berk, Newton and Berk found that when a battered woman seeks help, it reduces her chances of being a victim again-this includes shelters as well as other help-seeking actions [38]. But they concluded that the effectiveness of a shelter at reducing future violence also depends on whether a woman is taking control of her life. In this case, "a shelter stay can dramatically reduce the likelihood of new violence. Otherwise, shelters may have no impact or perhaps even trigger retaliation for disobedience." [39]

The effectiveness of shelters at reducing violence, however, must be viewed more comprehensively than by what happens just to women who come to shelters. Some researchers believe that the availability of a shelter can itself deter violence, because a woman in an abusive situation may be able to use the threat to leave to reduce her partner's aggression. [40]And many women contact shelters for information and help without seeking refuge. Beyond that, shelters are a source of advocacy and education to the community at large and work toward reducing domestic violence.

Shelters have a long-term deterrent effect on violence when they help women change their lifestyles, keeping them and their children safe from future abusive situations. There is substantial research evidence that children who witness domestic violence suffer short- and long-term adverse consequences and may be victims of abuse themselves. [41]These benefits of shelters can be augmented by cooperative efforts between child protection services and domestic violence advocates. The U.S. Department of Justice has identified several exemplary programs that better protect children while dealing with domestic violence. [42]In Massachusetts, for example, domestic violence specialists consult with child protection workers throughout the state to improve decision-making in these cases. And Dade County (Florida) has a unique court program designed to address the co-occurrence of child maltreatment and other forms of family violence. In 2000, the Minnesota Legislature approved a pilot program for the Ramsey County Attorney and St. Paul City Attorney to set up a joint domestic abuse prosecution unit that would coordinate with child protection attorneys. One of the program?s goals is to reduce the exposure to domestic violence for both adult and child victims, recognizing the interests of children in prosecuting domestic abuse. [43]

Only one study has tried to calculate the cost-effectiveness of a battered women's shelter. [44]The analysis compared the costs of a shelter in Arizona with the estimated savings it might bring about, primarily by helping women avoid violent assaults. Assaults often lead to expensive hospitalizations and lost wages. Social, judicial, and mental health costs and benefits were also included in the analysis. The analysis concluded that the benefits were greater than the costs.

Bowker surveyed 1,000 battered women about the effectiveness of eight formal services available to help them. [45]These were police, physicians and nurses, clergy, lawyers, district attorneys, social service or counseling agencies, women's groups, and battered women's shelters. Of these services, women gave the highest ratings (percentage "very" or "somewhat effective") to women's groups (60%) and battered women's shelters (56%).

In their extensive study of women in battered women's shelters in Texas, Gondolf and Fisher discussed how shelters have evolved from simply providing refuge to offering many other services. [46]Because many women return to a batterer after a stay in a shelter, they contended that a shelter is part of a long-term change process. They found only two significant predictors of whether a battered woman in a shelter would return to a batterer or not. The more economic independence a woman has, the less likely she is to return. But if the batterer is in counseling, a woman is more likely to return to him. That is, women are more likely to return if they believe the man will change his behavior. Unfortunately, men can also use counseling to manipulate women into returning, and there is strong research evidence that batterer treatment is not effective at reducing violence except during the treatment period. [47]

Gondolf and Fisher concluded: "Basically, our findings support the initial conception of shelters as a haven that provides battered women a temporary residence and supportive environment." [48]Shelters help women who are already seeking to change their lives and give leverage or act as a bargaining chip for women who return to the men who abused them. Their research did not find substantial support for the efficacy of other shelter services in reducing battering. They proposed that a more comprehensive service delivery model is necessary to deal with the multiple social and economic needs of battered women in shelters. Such a model would have to give priority to such services as long-term housing needs and child-care.

The effectiveness of advocacy services was also called into question in a well-constructed social experiment of services for women leaving a battered women's shelter in an unnamed Midwestern city. [49]The study had about 140 study participants; half were assigned at random to receive 4-6 hours of advocacy services per week for 10 weeks after they left the shelter, while the other half did not receive these services. The advocacy was designed to provide support and to help women access community services. After 6 months, there were no differences in the outcomes for women who got the services and those who did not. In either case, about two-thirds of the women were no longer involved with the abuser, but one-fourth of the women who had ended their relationships had experienced continuing abuse. The researchers concluded that the advocacy services by themselves were not sufficient to create long-term change or reduce the likelihood of further domestic violence for abused women.

Battered women's shelters prevent abuse, but it is not clear how long a woman needs to stay in a shelter to obtain that benefit. Research offers no guidelines. One can conclude, however, that the main benefit of a shelter is refuge, not services that try to solve a clients? social and economic problems-something not likely to happen in a shelter stay of two or three weeks. Shelter operators in Minnesota attested to the fact that shelter clients increasingly need a wider range of services and present more challenges to the staff, who may not be trained to deal with the range of economic, personal, psychological, and medical issues. [50]If women were allowed to stay in a shelter until their needs for services were met, however, it would run the risk that space might not be available for women in an emergency, and the shelter would begin to lose its effectiveness. This has happened in other states. As other researchers have suggested, a different model of service delivery is needed to meet the multiple social, economic, and housing needs of battered women who come to shelters.

We next look at how other jurisdictions have financed their shelter systems, and examine some alternative models for providing services.

State comparisons

To understand Minnesota's shelter system, it is helpful to compare it with other states as to the method and level of funding. Exact comparisons are difficult, however, because states use different mixes of local, state and federal funds, and pay for different services. In 1999, MCCVS surveyed other state governments about these issues. We reviewed responses from 13 states and found that most had contracts with shelters that controlled payments. Sometimes contracts were for fixed amounts; others allowed for reimbursement of expenses up to the amount allowed by contract. A contract typically specifies a range of services that a shelter is to provide, but payments are not tied to the number of women and children served. Only two states seem to make per diem payments like Minnesota (Oklahoma and Hawaii). Some states pay each shelter the same amount, while in others the amounts vary. Some states also fund transitional housing for battered women, which may be included in the total appropriations.

Wisconsin offers an interesting comparison with Minnesota as a neighboring state with a strong shelter system but with very different funding. It has a shelter program in each of its 72 counties and 11 tribal areas-a more extensive network than in Minnesota counties. [51]It has 35 residential shelters with a total of 750 beds-about 15 percent more beds than Minnesota. But state spending is less than a third of Minnesota's. Annual funding in Wisconsin was $7.4 million in 1999-$5 million state general purpose revenue, $1.1 million in federal Family Violence Prevention funds, $1 million from Temporary Assistance to Needy Families (TANF) funds, and $0.3 million from surcharges attached to sentences for domestic abuse-related crimes. How can it cost Wisconsin so much less? In contrast to Minnesota, Wisconsin provides shelters with only about one-third of their operating expenses. Shelters obtain the rest of their funds locally, from a mix of public and private sources.

All Wisconsin shelter programs have contracts for fixed amounts but are reimbursed for expenditures monthly. Shelters receive different contract amounts based mainly on historical costs, as in Minnesota. Residential shelters receive basic grants between $65,000 and $116,000; nonresidential shelter services receive between $32,000 and $63,000. Additional grants are provided for children's services ($20,000), support services ($25,000), and rural outreach offices ($30,000). The variation in spending across shelters is much less than in Minnesota. For example, in Minnesota in FY2000 several shelters received state funds in amounts of $1.5 million to over $2 million, while nonresidential programs may have received as little as a few thousand dollars in per diem funds. Although about 1,400 Wisconsin women were turned away from shelters in 1999, mostly in Milwaukee and Madison, advocates for battered women have not reached a consensus as to whether more beds are needed. Many feel that any additional resources should be directed toward keeping women safe in their own homes. [52]But some rural counties would like to have a residential shelter instead of a nonresidential program.

A comparison with other states confirms that Minnesota's spending for battered women's shelters is among the most generous, though exact comparisons are not possible. In recent years, Louisiana, spent $3.2 million [53]; Oklahoma, $4.6 million state and federal; Georgia, $4.9 million state and federal; Massachusetts, $13.6 million; Texas, $13.6 million state and federal [54]; California, $16 million state [55]; Pennsylvania, $18 million state and federal, some for nonshelter services; and New York, about $30 million. At $17.979 million in state per diem funds plus $2.485 million in 24 base grants for residential shelters, Minnesota has perhaps the highest spending in relation to population. On a per capita basis, Minnesota's spending is similar to Canada's. In 1997 operating costs of shelters in Canada were $170 million (Can.) or about $110 million (U.S.). [56]If Canada's spending level were applied to Minnesota it would be equivalent to $17 million, given that the ratio of Canada's population to Minnesota's is 6.5:1. We must also note that many of the states that spend less per capita than Minnesota do not have adequate shelter beds. Shelters in New York, Massachusetts, Oregon, and Texas, for example, report turning away many women. [57]

We located per diem costs for several places, which also show that Minnesota's level of payment is one of the most generous. Hawaii allows a per diem of $79, compared to Minnesota's current $75. But shelters in Los Angeles in 1997 had a per diem cost of $51 [58], and in New York City in 1996 it was $64 (50% federal, 25% state, 25% city). [59]Alberta, Canada reported a per diem cost of $51 (Can.) in 1995. [60]

An important difference between Minnesota and some other jurisdictions is the availability of transitional housing for battered women. Minnesota has transitional housing but only a few places focus on battered women. Los Angeles, for example, has transitional or second-stage shelters that allow battered women to stay for longer periods than would be possible in a crisis shelter, typically from 6 to 14 months, while still receiving various social services. [61]The average cost for the transitional housing is much less than for a crisis shelter, about $32 per day compared to $51 for the shelter. A study of the Los Angeles shelter system argued that, ideally, there should be two or three transitional housing beds for each crisis bed. [62]Some local advocates expressed the view, however, that it is more important for Minnesota to expand its system of crisis shelters to areas of the state that are underserved than to spend more money on transitional housing for battered women. [63]

Justice system response

For many women, a shelter is a last resort after other means of dealing with an abusive partner have failed. Demands on the state's shelters would be less if other social systems were more effective at reducing domestic violence. The criminal justice system is at the forefront of this effort, but has received much criticism from advocates for not protecting women.

Police are often the first public response to domestic violence. A new, exceptionally well-done study of 492 battered women in Chicago found that 39 percent had contacted the police or someone else had called the police for them. [64]This was greater than the 26 percent who had sought medical attention and 18 percent who had asked for help from a counselor or social service agency. Further analysis showed that those with police contact were more likely to experience an increased level of violence after the contact. This was also true to a lesser degree for women who sought counseling but not true for those who got medical help. The researchers concluded that it was not the police contact itself that increased violence, but that women who call the police are often at greater risk of escalating violence than other abused women.

A key question is whether police should arrest abusers, and there has been much research on whether an arrest deters future incidents of violence. One of the earliest and most famous studies on this was done in Minneapolis. It seemed to prove that arrest of an offender would deter violence. But later attempts to replicate the Minneapolis study in several other cities did not find the same impact. Nevertheless, early research results and advocates for battered women prompted every state to make it possible for officers to make arrests in domestic violence cases without a warrant; and some states have a mandatory arrest policy. [65]Minnesota law permits a police officer to arrest a domestic violence offender within 12 hours of an incident if the officer has probable cause to believe it happened. [66]

The current view of many researchers is that arrests reduce violence by some abusers who are employed and whose victims are white or Hispanic but may increase violence by unemployed abusers and those whose victims are black. [67]These conclusions should be viewed tentatively, however, pending further research. Researchers also have found that although arrests may decrease violence in the short run, they may increase it in the long run. Further, if citations to appear in court are given to offenders, that may cause more violence than arrests. (Minnesota law prohibits the use of citations in domestic abuse cases. [68]) It was also found that court orders for counseling and court orders for protection made no improvement over arrests or not arresting in reducing violence.

Court orders for protection (restraining orders) are another justice system remedy for domestic violence. Research on their effectiveness has found that they do not adequately protect women from further abuse, and women and the courts should not rely on them as a primary defense against violence. [69]As with a shelter, a protective order is a signal of a failure to prevent violence and a last try to avoid it. The perpetrator under court order typically has a long criminal record that includes other violent crimes, some against men as well as women. In a study of over 600 such cases in Massachusetts, almost half of abusers attacked their victims again within two years of the restraining order. The recidivism rate did not differ between those women who had maintained the order and those who dropped it. Researchers concluded that a protection order should be used in conjunction with vigorous prosecution and sanctioning of offenders, not as a substitute for prosecution. If offenders are allowed to remain in the community, they should have the same scrutiny as other violent offenders, whom they in fact resemble.

Research on domestic abuse is hampered because it is not a separate type of crime in the FBI classification system, and charges against an offender may be for various crimes arising from an abusive incident. In Minnesota, missing or incomplete records are also a problem, especially for misdemeanor assaults. Minnesota Planning did a study of almost 11,000 domestic abuse arrests between 1992 and 1996 based on records for murder, violations of orders for protection, and fifth-degree assault. [70]Ten percent of the cases were felonies-the most serious level of crime-28 percent were gross misdemeanors, and 68 percent were misdemeanors. About half (46%) of these cases were prosecuted, leading to a conviction in 77 percent of prosecutions.

The most common sentence was jail (81%) followed by no incarceration or probation (15%) and prison (4%). So, of the 11,000 arrests, about 30 percent led to a jail or prison sentence. The median jail sentence was 90 days from 1992 through 1995 but increased to 95 days in 1996. Imposed prison time dropped from 903 days in 1992 to 450 days in 1996. [71]The researchers speculated that the decrease in prison time might have resulted from more offenders going to prison who previously would have gone to jail and received a shorter sentence. Although the jail sentences imposed might be typical for assault cases, they do not reflect the fact that domestic violence is usually a long-standing series of incidents. Minnesota law stipulates that repeated convictions for domestic abuse can lead to stiffer charges and sentences, but a more likely scenario is that previous assaults against a woman were not reported to the police or charged. An offender can also try for a plea bargain to avoid a domestic abuse conviction and the risk of a more serious charge on a subsequent offense.

Overall, the justice system seems to have little deterrent effect on many abusers. Nevertheless, Minnesota laws on domestic violence seem to be exemplary and do not call for major changes-a view generally expressed by shelter operators. [72]The Institute for Law and Justice, a non-profit organization specializing in criminal justice, has reviewed domestic violence laws in all states and compared them with ideal practices. The Institute?s review put Minnesota among the top four states having the best laws for dealing with domestic violence. [73]


Now that we have examined the shelter system from many directions, we can turn to the issues of shelter cost and funding, which precipitated this study. The funding issue is both a result of previous state policies and a current problem. Our review shows that Minnesota has a generous level of funding for battered women?s shelters compared with other states. But the historical funding arrangement, which was like an entitlement program, led to a high-cost system, especially for some shelter providers. The limitation on reimbursement to maintenance and security costs was interpreted differently by shelters, with little scrutiny by the state, contributing to the wide range of costs for different shelter providers. Still, none of this might have led to a funding problem if the characteristics of shelter residents had stayed the same and there were no shortage of low-income housing-conditions that dramatically increased costs in the last few years.

The funding cap and uniform per diem rate have caused financial difficulties for shelters and limit the addition of shelter capacity in regions of the state that are underserved. Shelters are also at risk for spending their reimbursement allotments before the end of a fiscal year. Even with a cap on total spending, the per diem funding method continues to be a possible incentive for shelters to let women stay longer than necessary. In short, the funding issue calls for both a short-term remedy to deal with current financial problems and a long-term solution to put the whole shelter system on a better foundation.

Impact of the funding cap

Based on a spending forecast by DHS, the Legislature in 1999 capped per diem funding at $17.979 million per year for three years starting with FY2001. At that time, DHS had also forecast that per diem costs would climb to about $19.3 million in FY2002 and $20.6 million in FY2003. [74]But actual spending in FY2000-after the forecast was made-was $20.9 million. In 2000, the Legislature also appropriated a one-time amount of $1.2 million to cover any unreimbursed per diem shelter expenses from the previous year. [75]

The DHS forecast suggests that shelters will have a total shortfall of almost $4 million in their budgets in FY2002 and FY2003, given the funding cap (but excluding the $1.2 million), and assuming that they continue to operate as before. Is this accurate? Clearly, the forecast underestimated actual expenses in FY2000 alone by nearly $3 million. DHS used a mathematical technique called exponential smoothing to make its forecast of shelter costs. [76]In this method, a forecast is made from the most recent observation to the next period in time, as from one year to the next. The next forecast is a weighted average of the previous forecast and the actual cost at that time; that is, the next forecast is corrected for error in the previous forecast. If the actual budget was higher than forecast, for example, the next forecast would be increased from the previous forecast. In this method, the new forecast is always set between the previous forecast and the previous observed amount. This is a well-established forecasting method that works well in situations when there are no trends. [77]If there is a trend, however, the forecast will always lag behind the true value. Note that this method of forecasting does not use any other information about shelters, such as length of stay, or other changes in the shelter system to adjust its forecasts. In particular, the impact of the Day One program on raising costs by helping more women find open shelter beds was not anticipated.

Expenditures for the shelter program had been fairly stable up to 1997, so the forecasting method worked well. But after 1997, with expenses increasing rapidly, the forecasting method necessarily produced unreliable estimates. One cannot determine what the true shelter cost would have been beyond FY2000 absent the cap. One can say that almost certainly the forecast underestimated what the expenditure would have been, because this type of forecast will always lag behind a trend.

At public meetings and in a survey by MCCVS, shelter providers reported the impact of the new per diem rate on their operations after about six months experience. As one might expect, some shelters made substantial cuts in services and budgets; others found additional funds or drew on savings to keep nearly the same level of services. The largest prospective budget cuts are in the largest residential shelters, which are in the metropolitan area. Many shelter providers outside the metropolitan area reported no adverse consequences yet, though some have used cheaper (and less desirable) motels. Several large shelters expect shortfalls in the range of $300,000 to $500,000 in FY2001 and increasing shortfalls in FY2002.

Typically, shelters have cut staffing for advocacy and community outreach services to reduce their budgets. No shelter has closed its doors, but at least one or two shelter providers indicated they might be forced to quit if they must continue at the current per diem levels for another year. Shelter providers warn that by cutting community outreach their effectiveness at preventing domestic violence is substantially diminished, as is their ability to help women who prefer to stay in their homes or not come to a shelter. The financial problems will grow next year as shelter costs inevitably increase for salaries, health insurance, transportation, and housing. Shelter providers also say that they are not able to replace the lost state dollars with grants from foundations. Generally, foundations are more interested in starting new programs than putting money to operate existing programs, and battered women's shelters are no longer seen as new programs. In any case, the abruptness of the change in funding left shelter providers little time to find new sources of funds.

Funding options

In the near term, no realistic option exists but to restore some of the state funding for shelters with higher costs. Although their cost basis might be too high, it was accepted uncritically by the state for many years, and there is no fair way to simply ignore it now. The new funding method, which was devised by MCCVS, has advantages in being easy to understand and administer. But it does not spread the pain of budget cuts fairly among the shelters. This analysis, however, cannot determine exactly how much additional funding from the state is necessary or how much relief might be achieved by a reallocation among shelters or by shelters reducing their costs. Inaccuracy of the forecast used in setting the cap also suggests that an increase in funding is warranted. An increase in state funds, however, should not imply a return to the earlier funding system but rather a more controlled and gradual downward adjustment of budgets by high-cost shelters. Or it might give shelters more time to find other sources of funds to maintain their spending levels. As one option, MCCVS might set up a tiered funding system for different types of shelters, depending on their previous per diem rates, but without a unique rate for each shelter.

In the longer term, the state needs a better way to fund the shelter system and to support new shelters where needed. A new funding system should also remove the financial incentives inherent in per diem payments. This could be done, as it is in other states, if MCCVS would contract with individual shelters as to what services they provide after a careful review of their costs. Similarly, if the Legislature appropriates funds for new residential shelters, they could be set up under contracts from the beginning. Another possibility is for the Legislature to fund some additional transitional housing specifically for battered women coming out of shelters. This would take some of the pressure off residential shelters while helping more battered women continue their transition to a better life.


The analysis leads to several recommendations-some to the Legislature and others to MCCVS.

  • The Legislature should consider a substantial increase in funding to alleviate financial problems of several large shelter providers.The cap on state spending, and the abruptness in how it was applied, have caused significant financial problems for several shelters. A strong argument can be made for a substantial increase in legislative appropriations. Without legislative relief, increasing costs will further exacerbate the financial problems of shelters over the next few years. Because the cap was based, in part, on inaccurate forecasts, a higher spending cap would also be more consistent with actual expenditures when the cap was put in place. It is reasonable to expect, too, that high-cost shelters try to reduce their costs or seek alternative funding.

  • In the near term, MCCVS should consider a tiered per diem funding plan.Although the new fixed-rate per diem funding arrangement is easy to understand and administer, it does not fairly distribute the pain of budget cuts. As a short-term solution, MCCVS should consider a tiered funding system for shelters, with reimbursement rates depending partly on their earlier expense rates. This would allow a more gradual downward adjustment of budgets by high-cost shelters relative to other shelters.

  • The Legislature should consider funding up to two or three new residential shelters in areas that are currently underserved.Generally, the state has an adequate amount of shelter space, but our analysis shows several large and highly populated regions that have no close access to residential shelters and limited access to other safe facilities. These regions include counties in an outer ring around the metropolitan area (Chisago, Isanti, Sherburne, Wright, Carver, and Scott), and Southwestern Minnesota. These regions should be priority areas for new shelters.

  • MCCVS should consider continuing state reimbursement for Minnesota women in the shelter in LaCrosse, Wisconsin.MCCVS plans to cut off per diem payments to this shelter. But shelter services are limited in Southeastern Minnesota-an area served by the LaCrosse shelter. Similarly, state funding for the shelter in Fargo should continue, as planned by MCCVS.

  • MCCVS should have statutory authority to restrict per diem payments to new shelters or for new beds in existing shelters if such payments would compromise the financial viability of the existing shelter system.MCCVS does not believe it has the legal authority to limit payments to eligible shelters. About half of shelter providers favor a moratorium on new beds or shelters, at least until the funding situation is stabilized.

  • MCCVS should monitor the average length of stay in shelters and have the authority to limit per diem payments for women who stay beyond a certain period of time in a shelter if the financial viability of the shelter system is threatened.Because the state pays most of the cost of the shelter system, it ought to have more control over funding and be able to react swiftly to changes in the system that may affect overall system viability. Length of stay is a major component of cost over which the state has no control. An increase of one day in average length of stay represents a potential increase of about $885,000 per year in per diem costs. MCCVS does not collect data from shelters on average length of stay, but shelters generally have this data available.

  • To put the entire shelter system on a sound financial basis in the long term, MCCVS should move away from per diem funding and begin to write contracts with shelters for the necessary services and terms of operation, after a careful review of their cost bases.Minnesota's method of funding shelters is unusual. It is more typical for states to contract with shelter providers as to the range of services that states will pay for. This is also a strong control on spending. Per diem payments are a left-over from the former funding system that acted as an entitlement and allowed the shelter system to reach high levels of spending with little control by the state over how the money was spent. Per diem payments remain a potential incentive for shelters to let women stay longer than necessary, although there is no direct evidence of this happening. The range of services under contract might differ among shelter providers, and state payments could go for expenses beyond the current limitation to food, lodging, and safety. For example, a particular shelter might be paid to provide training or services to other shelters. Thus the entire system might be tailored to meet a broader range of objectives than possible under current law.

  • The state should develop a comprehensive service model to help battered women after they leave shelters, including their needs for transitional housing, employment, child-care, and social and medical services.Many of the women who come to shelters have multiple serious and long-term problems that cannot be solved by shelters. If shelters devote too much time and energy to these problems, it will limit their ability to give refuge to abused women in emergency situations and unnecessarily increase costs. A comprehensive solution is called for.

[1] For a complete definition of domestic abuse, see the Domestic Abuse Act, M.S. 518B.01.

[2] M.S. 611A.32.

[3] Minnesota Laws 1999, Chap. 216, art. 1, sec. 8, subd. 3. .

[4] Minnesota Laws 2000, Chap. 488, art. 4, sec. 6, and art. 6, sec. 11; HF 2699. "By March 15, 2001 . . . the study must estimate the relative impact of the following, as it relates to providing shelter for victims of domestic violence: (1) the incidence of domestic violence; (2) law enforcement practices in response to domestic violence; (3) the number of victims seeking shelter and whether adequate shelter space exists, and trends regarding this; (4) the number of victims who have children also needing shelter; (5) the financial status of domestic violence victims; (6) the necessary length of stay in shelter; and (7) opportunities for victims to leave shelters. In studying these issues, the center shall analyze costs and demand for shelters in other states having programs comparable to Minnesota's."

[5] Shelter statistics are from MCCVS.

[6] Data from Department of Human Services on General Assistance payments to battered women's shelters, November 5, 1999. Some women may have stayed in a shelter more than once in a year; they will be included multiple times in the annual count. These numbers include only women eligible for general assistance payments, who are most shelter residents, but some other women may also have stayed in shelters without state reimbursement.

[7] Quarterly reimbursement data from MCCVS.

[8] Ibid.

[9] Edward W. Gondolf and Ellen R. Fisher, Battered Women as Survivors, An Alternative to Treating Learned Helplessness, Lexington Books, Lexington, MA., 1988.

[10] According to data from Minnesota Planning, Criminal Justice Center, arrests for violations of protection orders increased from 50 in 1993 to 301 in 1998.

[11] Minnesota Department of Corrections and Department of Human Services, "Funding Structure of General Assistance for Emergency Shelters for Battered Women-A Report to the Legislature," March 1991.

[12] M.S. 256D.05; the relevant section is now repealed.

[13] M.S. 256D.05, Subd. 3 repealed 1999, Chap. 216, art. 6, sec. 26, effective 1 July 2000.

[14] Governor' Budget, 1999, change item for FY2000-01 for the Crime Victim Services Center, p. H-292. The estimate for FY2001 was $17.979 million, which was where the cap was set. The forecast for FY2002 was $19.3 million, and for FY2003, $20.6 million.

[15] M.S. 611A.371 or Laws 2000, Chap. 445, art. 2, sec. 23.

[16] M.S. 611A.373 or Laws 2000, Chap. 445, art. 2, sec. 25.

[17] . M.S. 611A.373. For a more detailed explanation of the funding scheme, see "Emergency Shelter Services for Battered Women Per Diem Funding Administration Plan," Minnesota Center for Crime Victim Services and Minnesota Department of Human Services, October 1999.

[18] Shelters must also meet local licensing regulations to operate.

[19] Personal communication from Pat Prinzevalle, Alexandra House, January 4, 2001; see also the Allina Health System Foundation at www.allina.com/ahs/foundation.nsf/page/AF_dayone (This URL is no longer active 11/15/02); or Day One .

[20] Ibid. Some women have been turned away for not meeting admission criteria. Alcohol and drug use would also cause a woman to be turned away

[21] Patricia Tjaden and Nancy Thoennes, "Extent, Nature, and Consequence of Intimate Partner Violence, Findings from the National Violence Against Women Survey," National Institute of Justice, July 2000

[22] "Keeping Watch, 1999 Minnesota Crime Survey," Minnesota Planning, March 2000.

[23] "Keeping Watch, 1999 Minnesota Crime Survey," Minnesota Planning, March 2000

[24] "Family Violence in Canada: A Statistical Profile 2000," Statistics Canada, July 2000

[25] "Shelters for Abused Women," The Daily, June 11, 1999, Statistics Canada

[26] Personal communication from Sharon Lewandowski, Domestic Abuse Program Coordinator, Wisconsin Department of Health and Family Services, Nov. 2000.

[27] Edward W. Gondolf and Ellen R. Fisher, Battered Women as Survivors, An Alternative to Treating Learned Helplessness, Lexington Books, Lexington, MA, p. 31, 1988

[28] Ira W. Hutchinson and David J. Hirschel, "Abused Women Help-Seeking Strategies and Police Utilization," Violence Against Women 4: 436-56, 1988

[29] Leslie M. Tutty, "Residents' Views of the Efficacy of Shelter Services for Assaulted Women," Violence Against Women 5: 898-925, 1999.

[30] The model is Beds Used = 9.15 + 0.000082 * County Population, with R square = 0.78, F = 25.2, p = 0.0015. Standard error of the constant is 2.83 and of the coefficient is 0.000016. The constant term, or base level, may partly reflect transfers of women from other counties through the Day One program.

[31] These groups were mentioned by advocates at a public meeting on December 18, 2000 in St. Paul

[32] Statements made at a meeting of the Battered Women and Domestic Abuse Advisory Council, MCCVS, December 5, 2000, and at public meetings of shelter providers and advocates on December 18, 2000 in St. Paul, and on January 9, 2001 in St. Cloud.

[33] "Family Violence in Canada: A Statistical Profile 2000," p. 42.

[34] Jean Calterone Williams, "Domestic Violence and Poverty: The Narratives of Homeless Women," Frontiers 143, 1998

[35] Bonnie Brandl, "Programs for Batterers: A Discussion Paper," Wisconsin Department of Health and Social Services, August 1990

[36] Lewandowski

[37] Ruth Teichrob, "More and more battered women are finding county shelters full," Seattle Post-Intelligencer, March 26, 2000

[38] Richard Berk, Phyllis J. Newton, and Sarah Fenstermaker Berk, "What a Difference a Day Makes: An Empirical Study of the Impact of Shelters for Battered Women," Journal of Marriage and the Family 48: 481-90, 1986.

[39] Ibid.: 488

[40] Lee M. Bowker, Beating Wife Beating, Lexington, MA: Lexington Books, 1983. Amy Farmer and Jill Tiefenthaler, "Domestic Violence: The Value of Services as Signals," American Economic Review 86: 274-79, 1996.

[41] For a review, see Zena H. Rudo and Diane S. Powell, "Family Violence: A Review of the Literature," Florida Mental Health Institute, University of South Florida, August 1996; www.fmhi.usf.edu/institute/pubs/rudo-powell-violence.htm (Editor's Note: As of 11/11/02, this link is no longer active)

[42] safe from the Start,U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention, November 2000: 33-34.

[43] Laws 2000, Chap. 471, sec. 3, and Chap. 488, art. 4, sec. 5, subd.

[44] Sharon Chanley and Jesse J. Chanley, Jr., "Providing Refuge: The Value of Domestic Violence Shelter Services," paper presented at the Annual Meeting of the Western Political Science Assoc., Seattle, WA, March 24-27, 1999.

[45] Lee H. Bowker, "A Battered Woman's Problems Are Social, Not Psychological," in Current Controversies on Family Violence, edited by Richard J. Gelles and Donileen R. Loseke, pp 133-53, Newbury Park: Sage, 1993

[46] Edward W. Gondolf and Ellen R. Fisher, Battered Women as Survivors, An Alternative to Treating Learned Helplessness, Lexington Books, Lexington, MA, 1988

[47] Robert C. Davis, Bruce G. Taylor, and Christopher D. Maxwell, "Does Batterer Treatment Reduce Violence?: A Randomized Experiment in Brooklyn," National Institute of Justice, report # 180772, January 2000

[48] Ibid.: 96.

[49] Chris M. Sullivan, Rebecca Campbell, Holly Angelique, Kimberly K. Eby, and William S. Davidson II, "An Advocacy Intervention Program for Women with Abusive Partners: Six-Month Follow-Up," American Journal of Community Psychology 22: 101-21, 1994

[50] Public meeting of shelter providers, St. Cloud

[51] Lewandowski, personal communication

[52] Lewandowski, personal communication

[53] "Louisiana Executive Budget, Program D: Family Violence," www.state.la.us/opb/exec-bud00/01-exec/01-114D.htm.

[54] "Texas Performance Review, Chapter 7: Public Safety", www.window.state.tx.us/tpr/tpr5/7ps/

[55] Kate Sproul, "California's Response to Domestic Violence," Senate Publications #928-S, Sacramento, CA, November 1997.

[56] Cathy Trainor, "Canada's Shelters for Abused Women," Juristat, No. 85-002-XIE Vol. 19, No. 6, Statistics Canada

[57] Doris Sue Wong, "Shelters for Battered Women Found Lacking," The Boston Globe, May 10, 1999, Metro/Region, p. B1

[58] "A Report on Domestic Violence Shelters in the City and County of Los Angeles," Shelter Partnership, Inc., 523 West Sixth St., Suite 616, Los Angeles, CA 90014, January 1997

[59] Linda Stasi, "Women Abused as City, State Fiddle," Daily News (New York), Feb. 19, 1996, News, p. 8

[60] "1995 Statistics for Alberta's Women's Shelters," Government of Alberta News Release, Aug. 2, 1996

[61] "A Report on Domestic Violence Shelters in the City and the County of Los Angeles," Shelter Partnership, Inc

[62] Ibid.

[63] Public meeting, St. Paul, December 18, 2000

[64] Carolyn Rebecca Block, The Chicago Women?s Health Risk Study, Report to the National Institute of Justice, Illinois Criminal Justice Authority, Chicago IL, June 2000

[65] See Institute for Law and Justice, Alexandria, VA for a comparison of state laws and training requirements at www.ilj.org/dv

[66] M.S. 629.341.

[67] Janell D. Schmidt and Lawrence W. Sherman, "Does Arrest Deter Domestic Violence?" in Legal Interventions in Family Violence: Research Findings and Policy Implication, National Institute of Justice and American Bar Association, July 1998, www.ncjrs.org/pdfiles/171666.pdf

[68] M.S. 629.72

[69] Andrew R. Klein, "Re-abuse in a Population of Court-Restrained Male Batterers: Why Restraining Orders Don't Work," in Legal Interventions in Family Violence.

[70] Minnesota Planning, "Tracking Crime," September 1998

[71] A new report by the Office of the Legislative Auditor, "Chronic Offenders," February 2001, pp. 62-64, found that about 80 percent of persons convicted of misdemeanor domestic assault in 1999 went to jail, and their average jail time was about 70 days.

[72] Views expressed at a public meeting in St. Cloud on January 9, 2001.

[73] . Institute for Law and Justice, www.ilj.org The other top states are California, Rhode Island, and Wisconsin.

[74] Governor's Budget, change item for FY 2000-01 for the Crime Victim Services Center, p. H-292. Expected savings for the FY2002-3 biennium were $3.893 million

[75] Laws 2000, Chap. 488, art. 4, sec. 6. Generally, there was a lag of a few months before the county human service department processed all the reimbursements. The appropriation was an estimate of this amount yet to be paid out.

[76] Personal communication from George Hoffman, DHS, January 2, 2001. For more on this method, see Robert Goodell Brown, Smoothing, Forecasting and Prediction of Discrete Time Series, Prentice-Hall, 1963

[77] Brown.