Loading...
pc-agenda-jul-26-21       REGULAR MEETING AGENDA    Planning Commission meetings are being conducted in a hybrid format with in‐person and remote  options for attending, participating, and commenting. The public can make statements in this meeting  during the planned public comment sections.     Remote Attendance/Comment Options: Members of the public may attend this meeting by watching  on cable channel 16, streaming on CCXmedia.org, or via Webex by calling 1‐415‐655‐0001 and entering  access code 177 094 5354. Members of the public wishing to address the Planning Commission during  public comment sections should call 763‐593‐8060.    1. Call to Order    2. Approval of Agenda    3. Approval of Minutes  July 12, 2021, Planning Commission Meeting    4. Update and Discussion – Major Amendment to PUD 90  Applicant: ISD #270 ‐ Hopkins School District  Address: 5430 and 5300 Glenwood Ave, Golden Valley, MN 55422    5. Discussion – Pawn Shops, Precious Metal Dealers, and Payday Lenders    6. Discussion – Accessory Dwelling Unit Introduction    – End of Televised Portion of Meeting –  To listen to this portion, please call 1‐415‐655‐0001 and enter meeting access code 177 094 5354.      7. Council Liaison Report    8. Other Business  a. Reports on Board of Zoning Appeals and Other Meetings    9. Adjournment  July 26, 2021 – 7 pm         REGULAR MEETING MINUTES    This meeting was held via Webex in accordance with the local emergency declaration made by  the City under Minn. Stat. § 12.37. In accordance with that declaration, beginning on March 16,  2020, all Planning Commission meetings held during the emergency were conducted  electronically. The City used Webex to conduct this meeting and members of the public were  able to monitor the meetings by watching it on Comcast cable channel 16, by streaming it on  CCXmedia.org, or by dialing in to the public call‐in line.     1. Call to Order  The meeting was called to order at 7:00 by Vice‐Chair Brookins.    Roll Call  Commissioners in person:  Ron Blum, Adam Brookins, Andy Johnson   Commissioners virtual:    Rich Baker, Ryan Sadeghi  Commissioners absent:    Noah Orloff, Lauren Pockl, Chuck Segelbaum  Staff present:     Jason Zimmerman – Planning Manager, Myles Campbell – Planner   Council Liaison present:   Gillian Rosenquist    2. Approval of Agenda  Vice‐Chair Brookins asked for a motion to approve the agenda.  MOTION made by Commissioner Johnson, seconded by Commissioner Blum, to approve the agenda  of July 12, 2021. Motion carried.    3. Approval of Minutes  Vice‐Chair Brookins asked for a motion to approve the minutes from June 28, 2021.  Commissioner Blum noted details to a statement he made and asked they be added.   MOTION made by Commissioner Johnson, seconded by Commissioner Blum, to approve minutes,  with additions. Motion carried.    4. Public Hearing – Major Amendment to PUD 90  Applicant: ISD #270 ‐ Hopkins School District  Address: 5430 and 5300 Glenwood Ave, Golden Valley, MN 55422    Jason Zimmerman, Planning Manager, informed Commissioners that staff is waiting for  updated information from the applicant and recommending tabling the hearing to July 26. This  is the last date the applicant will have to submit information before their window for approvals  is closed.     July 12, 2021 – 7 pm    City of Golden Valley    Planning Commission Regular Meeting  July 12, 2021, 2021 – 7 pm       2  MOTION made by Commissioner Blum, seconded by Commissioner Johnson, to table the item to  the July 26th, 2021 meeting, as requested by the applicant. The motion carried unanimously.      5. Discussion – Accessory Dwelling Unit Introduction  Val Quarles, Community Development Intern, introduced herself to the Commissioners and  introduced Accessory Dwelling Structures.   ADU: additional living quarters on residential lots that are independent of the primary unit(s). They  are also referred to as mother‐in‐law apartments, granny flats, carriage houses, or secondary  suites.   Council, residents, and the comp plan have all expressed desire for an ADU policy in Golden Valley.  This is a good opportunity for the city to densify without the need for land acquisition.     Quarles provided details on ADUs and stated they can be attached or detached. The 2040  Comprehensive Plan leans toward ADUs in its goal to expand the variety of housing options.  Quarles went more in depth on this portion of the housing chapter in the comp plan, and discussed  independent surveys on folks considering ADUs. In conjunction with providing a homeowner  perspective, a policymaker perspective was provided and benefits were discussed.   There are 19 municipalities within the Twin Cities area that have ADU policies; this will be helpful  for staff as City Code already allows for some ADU‐like arrangements. Specific limits in current  regulations prevent the creation of high quality ADUs that best serve the needs of the community.     Quarles opened the discussion with two questions:   Could ADUs be a good fit for Golden Valley? What possibilities does this open up?   What additional questions do you have, as we move forward, that staff may research and  report back on?     Commissioner Blum asked if Council wants to see ADU legislation or if the question is if ADUs are  appropriate. Jason Zimmerman, Planning Manager, stated he feels there is a sincere interest in  legislature but the goal right now is to do the research, understand what may be involved, and  report back.   Blum stated he felt the cornerstone of the ADU conversation has been how it will impact residents  in Golden Valley. He noted the presentation thus far has been exclusively positive and asked  Quarles to consider the items that are less positive. Quarles responded that it has seemed folks  view ADUs in a positive light until they note the cost. While researching this item, the question to  other cities was posed if ADUs had an affordable housing component. The cost to build is very high,  creating sewer and water access can be expensive, and adding off street parking can add another  element.   Commissioner Sadeghi stated he’s generally in support of ADUs but ultimately it may come down  to the build form guidance and how that affects Golden Valley. Commissioner Johnson asked if  adding an ADU would turn and R‐1 into an R‐2. Zimmerman responded that in some ways it does  but the City wouldn’t just apply R‐2 regulations to R‐1, there would be other considerations.  City of Golden Valley    Planning Commission Regular Meeting  July 12, 2021, 2021 – 7 pm       3  Johnson brought up changes in circumstances regarding the intended use for the ADU by the  homeowner and potential unintended consequences.   Commissioner Baker added that while this will not solve all housing equity concerns, creating  smaller housing units and increasing rental properties both lead to greater housing equity. Sadeghi  agreed with Baker’s points and mentioned the greater value in multi‐generational living.     The discussion continued on to revolve around rent abilities, affordability, land as a finite resource,  private/public green space, use, and regulations in other cities.     In conclusion, the following questions are being requested of staff to research and answer:  1. What are potential impacts of ADUs?  2. What happens with ADUs after the initial use has expired?  3. ADU rental cost versus apartment?  4. How many ADUs are being built in the surrounding cities?      6. Discussion – Use Table Updates    Myles Campbell, Planner, started by reviewing the history of the Use Table discussion.   The Uses reviewed in the study are:   Commercial Uses   Industrial Uses   Office Uses   Economic & Business Land Use Table   Institutional   Residential    Prior to a public hearing, edits will be made based on Commission direction. There will need to be  some formatting changes with Municode, and staff needs to confirm potential rezoning actions  with the City Attorney.   Vice‐Chair Brookins opened the discussion.   Commissioner Blum thanked staff for the gradual amendments so the Use Tables have a higher  readability. Commissioner Baker echoed this. Commissioner Johnson asked about the places of  assembly and recalled it being a conditional use in the Light Industrial but stated he couldn’t find  that use on the tables. Campbell responded that it was left off so the discussion could continue at  the public hearing. Staff recommendation was to make it a permitted use due to legal  recommendations around RLUIPA. The conversation continued on the regulations around RLUIPA,  updates, and appropriate zoning designation.             City of Golden Valley    Planning Commission Regular Meeting  July 12, 2021, 2021 – 7 pm       4  Televised portion of the meeting concluded at 8:08 pm    7. Council Liaison Report   Council Member Rosenquist noted that July was Park and Recreation month and that the City Council  has approved a proclamation in that regard, that they approved Home Energy Squad audits and hosting  a Green Corps member, and previewed the upcoming Council/Manager meeting where the Police  Commission, Section 8 non‐discrimination, and the STAR program would be discussed.     8. Other Business  None.    9. Adjournment  MOTION by Commissioner Blum to adjourn, seconded by Commissioner Johnson, and approved  unanimously. Meeting adjourned at 8:15 pm.                                                                                                            ________________________________                                                                                                  Andy Johnson, Secretary  ________________________________  Amie Kolesar, Planning Assistant      1      Date:  July 26, 2021  To:  Golden Valley Planning Commission  From:  Jason Zimmerman, Planning Manager  Subject:    Informal Public Hearing – Meadowbrook School PUD No. 90, Amendment #5 –  5430 and 5300 Glenwood Avenue      Summary  Hopkins Public Schools, represented by Neil Tessier, applied for a Major Planned Unit Development  (PUD) Amendment in order to expand the boundary of the PUD and to incorporate additional land  area currently addressed as 5300 Glenwood Avenue. This expansion would allow for the  completion of a traffic control plan initiated as part of Amendment #4 in 2018 as well as provide  future space for district offices or special services educational programming.    On April 26, 2021, the Planning Commission unanimously recommended denial of this request. On  May 18, the City Council heard this request and, after conducting a public hearing, voted to return  the application to the Planning Commission and encouraged the applicant to work with them  further to address outstanding concerns.    A new public hearing was targeted for June 14, for June 28, for July 12, and then for July 26. On July  23, the applicant asked that the three applications (a change in land use for 5300 Glenwood  Avenue, a rezoning of 5300 Glenwood Avenue, and a Major PUD Amendment to incorporate 5300  Glenwood Avenue into the existing PUD) be withdrawn in order to allow for additional work to be  carried out before resubmitting all three at a future date.    Without the PUD amendment, Meadowbrook will still be able to continue to use the unimproved  drive connecting the two lots as an alternative queue throughout the school year, though any use  of the 5300 building under its current zoning designation – Office – would require close scrutiny  by the City to ensure adequate parking is being provided.    Recommendation  Because the application has been withdrawn, there is no action needed at this time.      1      Date:  July 26, 2021  To:  Golden Valley Planning Commission  From:  Myles Campbell, Planner  Subject:  Pawnshops, Precious Metal Dealers, and Payday Lenders        Summary  In August 2020, the City Council enacted an interim ordinance (moratorium) on any new  pawnshops, precious metal dealers, or payday lenders from locating within the City for a 12‐ month period. The Council indicated this was to study the potential adverse impacts on the  health and welfare of community members by these types of businesses. This interim ordinance  would allow time for the Council and Planning Commission to review the City code licensing  requirements and zoning code allowances respectively. A preliminary discussion on the topic of  these uses with Planning Commission was held in March of this year.     Following the Planning Commission’s discussion of these uses, the City Council indicated they  would like to take additional time to explore how new licensing requirements for payday lenders  could be put in place to strengthen consumer protections. While staff had been planning to table  the zoning discussion on these items until such licensing requirements were further developed,  review of the state statute requirements for interim ordinances has indicated that an extension  to this moratorium is not possible. As such, the City Manager has directed staff and the Planning  Commission to proceed with adopting reasonable zoning regulations that will be in place shortly  after the expiration of the interim ordinance.    This memo will present a summary of the original topic discussion with Planning Commissioners  and the draft language proposed by staff. From there this memo will address those concerns and  questions voiced by the Commissioners at the time, as well as providing an updated set of  language given the modifications suggested by Planning Commissioners.     Current Code  To remind Commissioners, these three uses are, in staff’s perspective, under‐regulated by the  current zoning code. Pawnshops and precious metal dealers are additionally regulated by a set of  strict licensing requirements regarding their operation, however neither is explicitly defined in  our zoning code, or in any zoning districts. Payday Lenders are similarly undefined.   2    Because these uses are not defined outright, staff’s interpretation would be to classify  pawnshops and precious metal dealers as a “general retail use” as provided for in our  Commercial and Mixed Use Districts. Staff would also read code as to classify payday lenders as  “financial institutions” and as such they would be allowed in a number of zoning districts:  Commercial, Office, and all Mixed‐Use districts.    While broadly permitted in zoning, the major restrictive factor for a new pawnshop currently  would be the City licensing requirements. A license is required to operate such a business, and  includes a number of stipulations such as:     A background check for the business owner and their affiliations (spouses, partners,  organizations)   Restrictions on hours of operation   Itemization of individual properties received  o Daily reporting to the GVPD on these records    The City Code does not currently have any licensing requirements for payday lenders, although  staff believes it likely these will be put in place within the next year, given the support from City  Council for such regulations.     Original Discussion – March 22, 2021    The Planning Commission previously discussed the topic of pawnshops, precious metal dealers,  and payday lenders in March of this year – then as part of the ongoing process of reorganizing  the City’s allowed land uses into a table format. The meeting minutes from this discussion have  been included with this memo for review. At that meeting, staff provided some background on  why the City was considering these changes to code, and also provided a number of examples of  similar ordinances in comparable municipalities such as St. Louis Park and Brooklyn Park.     Generally, Commissioners were open to the idea of defining these uses under the zoning code  and treating them as restricted uses, rather than permitted or conditional uses. Commissioners  did have some discussion and questions on the reasons why these uses were being singled out as  having a harmful impact on community members. These questions will be addressed more fully  later in this memo but at a high level, staff would argue that the reasons are two‐fold. One being  the capacity for pawnshops and precious metal dealers to facilitate the movement of potential  stolen goods, and secondly the capacity for all of these “fringe banking” type businesses to have  predatory lending practices damaging to their users health and financial well‐being.     In addition to questions on the purpose for the new use regulations, Commissioners also asked  for mapping examples of the proximity buffers proposed by staff. These have been included with  this memo for review, and discussed in greater detail further in this memo. Finally,  Commissioners and staff discussed some modifications to the proposed language of the text  amendments, and specifically in regard to the definitions for payday lenders and currency  exchanges.   3      In terms of what had been presented by staff, the following draft language was provided for  Commissioners at this meeting:    Pawnshops and Precious Metal Dealers  Refer to the definitions for Pawnbrokers and Precious Metal Dealers as provided in the licensing  code, Article XIV –Pawnbrokers and Precious Metal Dealers, Sec. 16‐389 –Definitions  Include these uses as restricted uses in the Commercial (C) zoning district, subject to the  following restrictions:  1. Any pawnshop or precious metal dealer shall be located not less than 750 feet from a  pawnshop, precious metal dealer or currency exchange, as measured at the lot line. In  the case of a multi‐use building, distances from the use shall be measured from the  portion of the structure occupied by the pawnshop or precious metal dealer.  2. Any pawnshop or precious metal dealer shall be located not less than 350 feet from a  school, place of assembly, library, public park, or any parcel zoned R‐1 or R‐2, as  measured at the lot line. In the case of a multi‐use building, distances from the use shall  be measured from the portion of the structure occupied by the pawnshop or precious  metal dealer.  3. Such uses shall be contained within a completely enclosed building, and no outside  storage, display, or sale of merchandise shall be permitted.  4. Exterior loudspeakers or public address systems are prohibited.  5. Visibility into the store shall be maintained by utilizing clear, transparent glass on all  windows and doors, and by keeping all windows free of obstructions for at least three  feet into the store. Product may be displayed for sale in the window as long as the display,  including signage, does not occupy more than 30 percent of the window area.  6. All entrances to the business, with the exception of emergency fire exits which are not  usable by patrons, must be visible from the public right‐of‐way.  When such businesses  are located within an enclosed commercial complex, all patron entrances must open onto  the common concourse.    Payday Lenders/Currency Exchanges  Payday Lender: any person or business that has as its primary activity the providing of short‐term  consumer loans for the borrower’s own personal, family, or household purpose which are usually  for a period of forty‐five (45) days or less. Payday lenders do not include banks.    Currency Exchange: Any person, except a bank, trust company, savings bank, savings association,  credit union, or industrial loan and thrift company, engaged in the business of cashing checks,  drafts, money orders, or travelers' checks for a fee. The term does not include a person who  provides these services incidental to the person's primary business if the charge for cashing a  check or draft does not exceed $1.00 or one percent of the value of the check or draft, whichever  is greater.    4    Include these uses as restricted uses in the Commercial (C) zoning district, subject to the  following restrictions:  1. Any currency exchange shall be located not less than 750 feet from a pawnshop,  precious metal dealer or currency exchange, as measured at the lot line. In the case of a  multi‐use building, distances from the use shall be measured from the portion of the  structure occupied by the currency exchange.  2. Any currency exchange shall be located not less than 350 feet from any parcel zoned R‐1  or R‐2, as measured at the lot line. In the case of a multi‐use building, distances from  the use shall be measured from the portion of the structure occupied by the pawnshop  or precious metal dealer.  3. All entrances to the business, with the exception of emergency fire exits which are not  usable by patrons, must be visible from the public right‐of‐way.  When such businesses  are located within an enclosed commercial complex, all patron entrances must open  onto the common concourse.    Commissioner Questions/Comments  Why are these uses considered detrimental to public safety or health? What research exists on the  topic?    In establishing a moratorium on these uses, the City Council’s interim ordinance noted the  potential for these land uses to have a negative impact on public health safety welfare, however  additional research was required to corroborate this analysis. Staff has had opportunity to review  a number of papers on the topic, although finding more recent studies on the topic was  somewhat challenging. Staff has tried to include a handful of contemporary academic articles  with this memo addressing both the potential for these uses to impact crime and public safety,  and in their impact upon the health and economic welfare of community members. Summaries  for these articles are provided below:    Do Fringe Banks Create Fringe Neighborhoods?: Examining the Spatial Relationship between  Fringe Banking and Neighborhood Crime Rates   Charis E. Kubrin & John R. Hipp, 2016    This article gets at some of the reasons why people use payday lenders and other fringe banking  options, but is more focused on the impacts of these uses on their surroundings. Identifies that in  certain instances, payday lenders and pawnshops can be considered “crime attractors” due to  their functions making them “well suited for motivated offenders to find attractive and weakly  guarded victims or targets.” (Kurbin, p. 5) Given that these uses involve largely cash transactions,  consumers are attractive targets for robberies. Similarly, pawnshops can facilitate traffic in stolen  goods and are often convenient to access for offenders in this regard. Of note, the authors  believe that the exact impact on crime attraction from fringe banking establishments is also  highly dependent on other neighborhood factors, such as concentrated poverty or population  5    density. (Kurbin, p. 9) In a study of fringe banking establishments of Los Angeles, the following  takeaways were arrived at (p. 21‐22):   Fringe banks consistently related to higher crime levels on the block they were located on,  even controlling for correlated factors. This is especially true for robbery crimes.   Fringe banks often also impacted crime statistics on adjacent blocks   Payday lenders and check cashing businesses had a stronger association with elevated  crime levels compared to pawnshops   The association between fringe banks and crime is moderated by local contexts, especially  being located in busier or high population density areas    From Payday Loans to Pawnshops: Fringe Banking, The Unbanked, and Health  By Jerzy Eisenberg‐Guyot, Caislin Firth, Marieka Klawitter, and Anjum Hajat, 2018    Contrasting with the previous article’s perspective on the impact of fringe banks on criminal  activity, this article from 2018 focuses on how fringe banks and lenders can be harmful to those  who rely on them for their financial services. While advertised as one‐off loans for emergency  expenses, typical users of these lenders often take out multiple loans per year. Interest rates for  short‐term fringe loans can exceed the value of the loan itself. Referring to a Pew Institute study  from 2012, the average payday borrower is indebted for five months, and will pay $520 in fees  and interest on a loan averaging $375. While there are obvious material impacts to debt and  financial burden, this study tried to examine a potential link between a person’s use of fringe  banks and their health.     The study principally used the Current Population Survey (CPS) administered by the Census  Bureau, along with supplementary data sets from the Federal Deposit Insurance Corporation and  the Annual Social and Economic Supplement to the CPS. To summarize their findings:   Persons who had used a fringe banking service in the past year had a 38% higher  prevalence of reporting their health as “poor or fair”    Persons who were unbanked (no traditional banking account) had a 17% higher  prevalence   Use of check‐cashing services or tax refund anticipation loans in the past year had little to  no effect on health  o Researchers hypothesized this was because check‐cashing services have a one‐ time fee, but do not lead to debt creation and trapping like payday loans. (p. 433)    What impact do the suggested distance buffers have on available land?    Staff has included a number of supplemental maps illustrating the scope of the buffer restrictions  proposed by staff. The buffer restrictions proposed fall into two categories:  1. Proximity to other uses – a minimum distance from uses such as residential and schools,  given the potential  for these uses to be detrimental neighbors  2. Density – a minimum distance between two or more pawnshops/PMD/currency  exchanges, to mitigate an overabundance of these uses in a neighborhood  6    Maps included provide a visual for the proposed minimum distances (350’ from other uses, 750’  from other pawnshops/PMD/currency exchanges). For the density restrictions, given the lack of  an active business in the City, staff tried to provide a few different examples, both for whole  parcels and one example for a facility located in a multi‐use building. As shown on the maps, all  of these restrictions still allow for these types of businesses to locate in the City, although not  every commercially‐zoned property will be considered a viable location. Given that staff’s initial  language for currency exchanges only restricted proximity to R‐1 and R‐2 uses, there is slightly  more land available for this use compared to pawnshops and precious metal dealers.    In addition, these two restrictions would work in concert with one another. Staff provided in one  of the mapping examples a case where both the pawnshop proximity restrictions and a sample  density restriction are overlaid on the same map. The two buffers combined result in limited  opportunity for another pawnshop or payday lender use to collocate on the same block or area.  In this way, the buffer restrictions provide a means of decentralizing these types of uses to  different commercial areas of the City.     What concerns/insight does the police department have on pawnshops/PMD/payday lenders?    Similar to what staff’s research showed, Golden Valley Police noted there is some potential with a  pawnshop use to function as a means for criminals to sell stolen goods. This is made more  difficult they noted, due to the licensing requirements for a pawnshop which requires cataloging  of all pawned items to be reported to the PD, and for items like cameras, jewelry or guns, these  reports need to be filed the same day the item was received.      Staff did not receive comments on the suggested zoning changes relating to pawnshops, although  the police department noted that some sections of the licensing code are in need of updating.  They noted the licensing ordinance refers to an older automated records system (APS versus the  current LeadsOnline used by the department) and it also does not account for other businesses  that buy used goods and work with the police department to determine whether or not they are  stolen goods.     Payday Lenders/Currency Exchange definition modifications    Payday Lender: any person or business that has as its primary activity the providing of short‐term  consumer loans, for the borrower’s own personal, family, or household purpose which are  usually for a period of forty‐five (45) days or less. Payday lenders do not include banks.    Currency Exchange: Any person or business, except a bank, trust company, savings bank, savings  association, credit union, or industrial loan and thrift company, engaged in the business of  cashing checks, drafts, money orders, or travelers' checks for a fee. The term does not include a  person who provides these services incidental to the person's primary business if the charge for  cashing a check or draft does not exceed $1.00 or one percent of the value of the check or draft,  whichever is greater.    7    Commissioners also discussed whether or not it made sense to expand the title of “payday  lender” to “short‐term lender.” If it were kept to only referring to payday lenders, should the  definition refer in some way to the person’s employment/paycheck/etc.? If broadened,  Commissioners suggested pulling from the language in state statute for Consumer Small Loans,  which is the statute that mandates the state level licensing of businesses like payday or title  lenders. Statute language is as follows:    Subdivision 1.Definitions.  For purposes of this section, the terms defined have the meanings given them:  (a) "Consumer small loan" is a loan transaction in which cash is advanced to a borrower for  the borrower's own personal, family, or household purpose. A consumer small loan is a short‐ term, unsecured loan to be repaid in a single installment. The cash advance of a consumer small  loan is equal to or less than $350. A consumer small loan includes an indebtedness evidenced by  but not limited to a promissory note or agreement to defer the presentation of a personal check  for a fee.  (b) "Consumer small loan lender" is a financial institution as defined in section 47.59 or a  business entity registered with the commissioner and engaged in the business of making  consumer small loans.    From the definition by statute we could elect to modify our definition of payday lenders to  instead refer to all consumer small loan lenders.    Consumer Small Loan Lender: Any person or business that has as its primary activity the providing  of consumer small loans for the borrower’s own personal, family, or household purpose. A  consumer small loan is a short‐term, unsecured loan to be repaid in a single installment.  Consumer small loan lenders do not include banks.    Conclusion  Given the potential for pawnshops, precious metal dealers, or payday lenders to present negative  impacts on the health, safety, and welfare of surrounding properties or community members,  staff feels that the suggestion to treat these as a restricted use makes sense. Given that these  uses can and have been operated in a legal manner within the state, staff feels that an outright  prohibition of these uses is not warranted, and rather that these restrictions would help to  prevent some of the potential negative externalities that may arise from having a concentration  of these uses in a single area.     As shown in the attached maps, both buffer restrictions would still allow for adequate  commercial land should such a user look to locate within the City.  The remaining regulations on  loudspeakers, window sightlines, and visible entrances further mitigates the potential for these  uses to impact surrounding properties, either via nuisance or any increase in criminal activity.         8    Recommended Action  This meeting serves only as a discussion of the topic and does not require a vote from  Commissioners. Staff will be bringing this topic back as part of a public hearing on August 9,  incorporating any comments or feedback provided at tonight’s meeting.    Attachments   Minutes from 3/22/21 Planning Commission Meeting (4 pages)   Do Fringe Banks Create Fringe Neighborhoods?: Examining the Spatial Relationship  between Fringe Banking and Neighborhood Crime Rates (42 pages)   From Payday Loans to Pawnshops: Fringe Banking, The Unbanked, and Health (9 pages)   Maps of Proposed Buffer Restrictions (5 pages)         REGULAR MEETING MINUTES    This meeting was held via Webex in accordance with the local emergency declaration made by  the City under Minn. Stat. § 12.37. In accordance with that declaration, beginning on March 16,  2020, all Planning Commission meetings held during the emergency were conducted  electronically. The City used Webex to conduct this meeting and members of the public were  able to monitor the meetings by watching it on Comcast cable channel 16, by streaming it on  CCXmedia.org, or by dialing in to the public call‐in line.     1. Call to Order  The meeting was called to order at 7:00 by Chair Blum.    Roll Call  Commissioners present: Rich Baker, Ron Blum, Adam Brookins Andy Johnson, Noah Orloff, Lauren  Pockl, Ryan Sadeghi, Chuck Segelbaum   Staff present:    Jason Zimmerman – Planning Manager, Myles Campbell – Planner   Council Liaison present:  Gillian Rosenquist    2. Approval of Agenda  Chair Blum asked for a motion to approve the agenda.  MOTION made by Commissioner Baker, seconded by Commissioner Johnson, to approve the agenda  of March 22, 2021. Staff called a roll call vote and the motion carried unanimously.    3. Approval of Minutes  Chair Blum asked for a motion to approve the minutes from March 8, 2021.  Commissioner Johnson asked for the expanded conversation on above or below stormwater  retention to be addressed as well as edit his comment to accurately reflect the irony of cutting  down trees to then sell electric vehicles.   MOTION made by Commissioner Segelbaum, seconded by Commissioner Brookins, to approve  minutes. Staff called a roll call vote and the motion carried unanimously.    4. Discussion – RLUIPA   Jason Zimmerman, Planning Manager, started his presentation by defining RLUIPA: Religious Land Use  an Institutionalized Persons Act. This act was passed in 2000 to protect religious organizations from  discrimination in land use regulation. The goal was to state no substantial burdens could be places on  religion unless it can be shown there is a compelling interest. Zimmerman went on to explain ways  RLUIPA regulations can come up and reminded Commissioners that the Planning Commission  discussed RLUIPA as it relates to the Golden Valley zoning chapter in 2017. This conversation led to  recommended updates. Some of the recommendations became part of the City Code recodification. A  March 22, 2021 – 7 pm    City of Golden Valley    Planning Commission Regular Meeting  March 22, 2021 – 7 pm       2  recommendation related to parking will be revisited in conjunction with a review of all City parking  requirements later in 2021.   The final item to discuss is if the City should expand areas for religious institutions to locate. A list was  displayed to show zoning districts where there are existing and proposed permissions for places of  worship.   Commissioners discussed the definition of “a place of worship” as well as other language choices in  the memo.  Differentiating between places of worship and other facilities with a religious affiliation  was discussed, such as a community center or daycare.    Commissioners and staff discussed zoning districts, uses, locations, definitions, and inclusivity.      5. Discussion – Pawn Shops, Precious Metal Dealers, and Payday Lenders  Myles Campbell, Planner, reminded Commissioners that at the request of City Council a moratorium  on new pawnshops, precious metal dealers and payday lenders put in place in August 2020. Staff was  directed to examine how these establishments are handled in the code. The concerns mainly revolved  around potential issues surrounding public safety, risks regarding stolen goods, impact on surrounding  properties, and consumer protections.   The examination of uses within the context of the current zoning code aligns with the current revisions  to the land use tables.   Campbell went on to explain the current code regarding Pawnshops and Precious Metal Dealers.  Pawnshops are currently classified as “general retail” and are a licensed use. Payday lenders are  considered a financial institution; no licensing is required with the City but a state license is required.     Staff reviewed zoning code in adjacent cities while assessing changes to the Golden Valley Zoning Code.  Staff found most communities use the definition of “Currency Exchanges” rather than payday lenders  to differentiate banks from other financial businesses.    Campbell provided a list of zoning control examples for both pawn shops and currency exchanges,  many akin to those in place for tobacco sales.     Staff recommendations:   Introduce “Pawnshops and Precious Metal Dealers” and “Currency Exchange” to the Zoning Code   Pawnshop and Precious Metal Dealers are defined in Sec. 16‐389. of City Code, no new definition  required  o Added as a restricted use to the Commercial Zoning District only   Restrictions should at a minimum include density restriction, proximity to  residential zoned properties, and visibility requirements   Currency Exchanges will need a new definition in the zoning code  o Commissioners can consider also adding a definition for “Payday Lenders”  o Currency Exchanges are added as a restricted use in the Commercial Zoning District   Restrictions should again focus on a density restriction between exchanges, and a  proximity restriction to residential zoned properties.  City of Golden Valley    Planning Commission Regular Meeting  March 22, 2021 – 7 pm       3  Campbell wrapped up with providing definitions of currency exchanges and payday lenders, as well as  provide a list of potential restrictions for pawnshops and currency exchanges.     Commissioner Baker supported staff recommendations but wanted input from staff’s Equity &  Diversity Manager to ensure the list isn’t excluding services vital to some populations. Commissioner  Segelbaum mostly echoed Baker’s statements while adding he too doesn’t want to put so many  restrictions on currency exchanges that it limits access to anyone who cannot access a bank. Chair  Blum cited case studies from Florida that show pawnshops enhance property crimes and facilitate  criminal activity. He added that property crimes related to pawnshops tend to be underreported and  often lack evidence to be further investigated.   Commissioner Johnson stated some of the proximity restrictions in light industrial, may interfere with  places of worship, per the previous conversation. Johnson added he was surprised to see the definition  of a payday lender included to provide a loan. He added a question asking why defining a person’s use  for an establishment is listed in a city code. Campbell responded that the particular language  referenced is from another city and state licensing references “short‐term consumer loans”. He added  that purpose language could probably be removed aside from the business use description.   Campbell stated that City Council direction wasn’t related to crime and safety but to consumer  protections. There are studies that show payday lenders utilize predatory practices to garnish wages.    Campbell asked Commissioners if they are comfortable with removing some language in the definition  as suggested by Commissioner Johnson. Chair Blum and Commissioner Pockl suggested edits.  Segelbaum pointed out that payday loans are very specific and some of the suggested edits sound like  a general short‐term lender. Campbell said staff would review statute language and state licensing  requirements to make sure city code is as in line with them as possible.     Commissioner Pockl asked if Golden Valley Police are included in the conversations related to pawn  shops. Campbell responded that it’s part of the next step in this process.          6. Discussion – PC Annual Report/Work Plan  Jason Zimmerman, Planning Manager, presented the annual report, he summarized the work the  Planning Commission did, the application types, code amendments, and recommendations of approval  or denial. The end of the report illustrates the work plan for 2021.     Commissioner Baker asked if corrections need to be made to adjust for the light rail no longer coming  through Golden Valley. Zimmerman said there are a few areas to be rezoned and the final decision will  occur when the City knows the final status of the light rail. Commissioners and staff discussed a few  more specifics regarding the 2021 work plan.     Chair Blum pointed to the bullets on the Planning Commission purpose statement. He suggested  developers not be first but mention the residents of Golden Valley first. Blum felt that this will show  commitment to the people and not developers. Zimmerman responded that this was in discussion the  previous year and the process to change this is a bylaw change. City Council needs to approve that  change.    City of Golden Valley    Planning Commission Regular Meeting  March 22, 2021 – 7 pm       4       7. Discussion – BZA Annual Report  Myles Campbell, Planner, presented the annual BZA report, he summarized the work done, variance  application types, hearings, practical difficulty standards, and recommendations of approval or denial.  BZA members will be attending a similar diversity and equity training as the Planning Commission.     Televised portion of the meeting concluded at 8:56 pm    8. Council Liaison Report   Council Member Rosenquist informed the Commission that the rowhouse changes to the zoning  chapter were approved by the City Council. She updated the Commission on talks with the  Minneapolis Park and Recreation Board around a Memorandum of Understanding for operations  at Twin Lake, and reported on public engagement around the Police Task Force. She also noted  that concepts were being developed for the Facilities Study.    9. Reports on Board of Zoning Appeals and other Meetings  None.    10. Other Business  None.    11. Adjournment  MOTION by Commissioner Segelbaum to adjourn, seconded by Commissioner Pockl, and  approved unanimously. Meeting adjourned at 9:08 pm.                                                                                                            ________________________________                                                                                                  Adam Brookins, Secretary  ________________________________  Amie Kolesar, Planning Assistant          UC Irvine UC Irvine Previously Published Works Title Do fringe banks create fringe neighborhoods? Examining the spatial relationship between fringe banking and neighborhood crime rates Permalink https://escholarship.org/uc/item/00z4t1p8 Journal Justice Quarterly, 33(5) ISSN 0741-8825 Authors Kubrin, CE Hipp, JR Publication Date 2016-07-28 DOI 10.1080/07418825.2014.959036 Peer reviewed eScholarship.org Powered by the California Digital Library University of California Do Fringe Banks Create Fringe Neighborhoods?: Examining the Spatial Relationship between Fringe Banking and Neighborhood Crime Rates Charis E. Kubrin John R. Hipp Post-print. Published in Justice Quarterly 2016 33(5): 755-784 Do Fringe Banks Create Fringe Neighborhoods?: Examining the Spatial Relationship between Fringe Banking and Neighborhood Crime Rates In the aftermath of one of the worst recessions in U.S. history, high unemployment and underemployment have placed millions of Americans in precarious financial positions. At the same time, more Americans than ever are opting out of traditional financial services (Damar 2009; FDIC 2012). Traditional financial services are offered by regulated financial institutions, such as banks and credit unions, and include checking and savings accounts as well as home mortgage and auto loans. According to a recent national survey conducted by the Federal Deposit Insurance Corporation (FDIC), over 8% of U.S. households are “unbanked” (i.e., they lack any kind of deposit account at an insured depository institution) and over 20% are “underbanked” (i.e., they hold a bank account but also rely on alternative financial service providers). As these statistics suggest, individuals and households are turning elsewhere to take care of their banking needs. More and more, they are relying on “fringe lenders” to manage their finances. Fringe lenders offer different financial services provided by non-traditional financial institutions, such as check cashers, payday lenders, and pawnshops. The services they offer include small loans, check cashing, payday loans, auto title loans, pawnshop transactions, rent-to-own financing, and income tax refund anticipation loans, among others. Common to all fringe lenders, however, are the exorbitantly high interest rates and fees associated with the business transactions. Given the concern that consumers who can least afford to pay for high-cost, high-risk financial products are the most likely to use them (Fox 2007), fringe lenders have attracted scrutiny by state regulators and legislators, and have generated substantial media attention. Well-documented are the detrimental effects of fringe banking on minorities (Stegman and Faris 2003) and the working poor (Fox 2007; Melzer 2011), as well as the growing web of fringe banking largely concentrated in low-income and disproportionately minority communities (Cover and Kleit 2011; Damar 2009; Gallmeyer and Roberts 2009; Graves 2003; Li et al. 2009; Prager 2009; Stegman 2007). Much less well-documented, 2 however, are the effects of fringe lenders on the communities where they are located. As just noted, fringe lenders are highly spatially concentrated in certain areas within cities, typically low to moderate income neighborhoods with greater concentrations of racial and ethnic minorities—areas that typically have higher crime rates. These patterns raise the question of whether fringe lenders themselves may be criminogenic; that is, whether the presence of fringe banking establishments is associated with crime rates in neighborhoods. To date, we are aware of only one published study that has examined this issue. In a study of neighborhoods in Seattle, Washington, Kubrin et al. (2011) find that a concentration of payday lending establishments is associated with higher crime rates, controlling on a range of factors traditionally linked to neighborhood crime. Kubrin and colleagues conclude there are broader community costs—namely higher crime rates—that all residents incur in those neighborhoods where payday lenders are concentrated. We build on this foundational study in several ways. First, we broaden the focus to incorporate several types of fringe banking establishments beyond payday lenders, considering the impact of payday lenders, check cashers, and pawnshops on neighborhood crime rates. In particular, we assess fringe lenders’ collective impact as well as determine which lenders are most associated with crime. Second, we capture what is arguably a micro spatial process by measuring crime and fringe banks at the block (not census tract) level. Third, we take into account the possible spatial effects of fringe banks located on adjacent blocks and up to three blocks away. And fourth, we assess whether the socio-demographic context of the block or block group moderates the fringe bank-crime relationship. Our study examines the association between fringe lending and violent and property crime rates in neighborhoods in Los Angeles—a city that has witnessed typical growth in fringe lenders. Fringe Lenders: Check Cashers, Payday Lenders, and Pawnshops Check cashing outlets are the most commonly used fringe financial service (Fox 2007:143). These outlets cash government benefit checks and payroll checks, and provide immediate cash without 3 waiting for the check to clear. Consumers pay a percentage of the check’s face value as a fee. Check cashing outlets were originally designed to serve consumers who did not have a traditional bank account however a growing number of Americans are turning to check cashers to access their funds more quickly. Today, check-cashing outlets cash more than 180 million checks worth more than $55 billion annually (Fox 2007:138). A fee of 2.52% is typically charged to cash computer-generated paychecks but fees can reach as high as 5% (Fox 2007:138). Some consumers who find themselves short of funds turn to payday lenders to secure a loan. Payday loans are small cash advances extended to borrowers in exchange for a postdated check (or automatic withdrawal authorization) for the amount of the advance plus a lender’s fee, typically around $15 per $100 borrowed. The lender holds the check or authorization for an agreed upon period of time, usually a few weeks. At that point, the borrower may pay off the loan, allow the lender to deposit the check or debit the borrower’s account, or renew the loan (referred to as a “roll over”), resulting in another lender’s fee (Burt et al. 2006:12). When annualized, interest rates on payday loans are well in excess of the rates for conventional use. The cost of borrowing, expressed as an annual percentage rate (APR), can range from 300 to 1,000%, according to the FDIC. To obtain a payday loan, a borrower must have a bank account, identification, and a source of income but is not required to demonstrate an ability to repay or possession of sound credit. According to the Consumer Federation of America, at the end of 2010, payday lenders extended $29.2 billion in loan volume annually, with $4.7 billion in revenue for loans made by payday lenders (Consumer Federation of America). At pawnshops, consumers obtain a loan for a stated term and in return pledge some collateral with the pawn broker, typically tangible personal property such as jewelry, consumer electronics, tools, musical instruments, or firearms. If the customer does not repay the loan by the specified date, s/he forfeits the collateral and extinguishes the debt. The average size of a pawn loan is quite small—around $100—and its term is typically one month. Pawnbrokers do not assess the creditworthiness of their customers. Rather, they rely upon the estimated value of the collateral in making their loan decisions (Prager 2009). Fees charged for pawn loans are typically stated as a percentage of the loan amount, and 4 can vary from as low as 12% to as high as 300% annually, depending, to a large degree, on legal limits imposed by the state in which the loan is made. The National Pawnbrokers Association (NPA) estimates there are over 13,000 pawnshops, or one for every 23,750 residents (Cover et al. 2011), constituting a $14.5 billion dollar industry. Fringe lenders are controversial for several reasons—they are concentrated in distressed communities and there is evidence of adverse economic consequences for those who rely on these institutions for financial services, in particular, borrowers risk becoming mired in a “debt trap” (Fox 2007:139; Stegman and Faris 2003). A counter-argument is that the industry is growing so rapidly precisely because it provides services that consumers want. Moving beyond this debate is the less-well studied question regarding the effects of fringe lenders on the communities where they are concentrated, including increased crime and victimization. Below we offer a theoretical argument for why fringe lending is likely to be associated with neighborhood crime rates. Theoretical Perspectives on Fringe Banking and Neighborhood Crime Rates Our theoretical approach incorporates a land use perspective grounded in environmental criminology, and which also draws from social disorganization theory. This focus directs attention to specific types of land uses and suggests that they shape the quality of life for residents and contribute to neighborhood local reputation, housing market values, perceived incivilities and disorder, informal social control, and of course, local crime and victimization rates (McCord et al. 2007; Taylor and Gottfredson 1986; Stucky and Ottensmann 2009; Wilcox et al. 2004). Of particular interest is criminogenic land use, or nonresidential land uses thought to facilitate criminal victimization. We consider fringe lenders to be one type of criminogenic land use. We theorize two paths through which fringe lenders may impact crime: direct and indirect. Unfortunately, due to data limitations, in the current study we only measure associations in our examination of direct effects, and are even more limited in our examination of indirect effects. Still, given the importance of fully theorizing fringe lending’s potential impact, we also theorize potential indirect 5 effects below. We return to this issue in the Discussion and Conclusion section, where we acknowledge this as a weakness of the study and encourage future researchers to empirically examine both direct and indirect paths. Concerning the direct path, it is argued that land use activity itself generates variations in risks for criminal victimization across areas. This argument considers two types of criminogenic land uses: crime generators and crime attractors (Brantingham and Brantingham 1995; Bernasco and Block 2011). As the names suggest, these types of land use are expected to attract criminals and generate crime in an area. Crime generators are “businesses, institutions, and facilities that bring large numbers of different kinds of people into a locale. Among those brought to the locale are some potential offenders and some potential victims” (McCord et al. 1997:299). Crime generators may become crime hot spots because the presence of large groups of people creates occasions for crime (Bernasco and Block 2011:35). Examples of crime generators are shopping precincts, high schools, and subway stops. Crime attractors, like generators, draw in users but given the purposes of these land uses and the composition of those drawn there for those purposes, a higher fraction of potential offenders or victims is likely with attractors (McCord et al. 1997). In other words, crime attractors are places that do not necessarily bring together large groups of people at the same time, but their function makes them well suited for motivated offenders to find attractive and weakly guarded victims or targets (Bernasco and Block 2011:35). Examples of crime attractors are bars, homeless shelters, halfway houses, and drug- treatment centers. In line with others (Bernasco and Block 2011; McCord et al. 1997), we suggest that pawnshops, check cashing stores, and payday lending outlets are also crime attractors. A defining aspect of crime attractors is that they almost always involve cash economies; that is, they represent places where numerous transactions occur and where the majority of these transactions involve cash, as opposed to payments by credit card or electronic payment systems (Bernasco and Block 2011:35). In their study of robbers’ motives and methods, Wright and Decker (1997:76-78) provide examples of how robbers are attracted to places where cash flows. And in a similar study using data on roughly 13,000 robberies, Bernasco, Block, and Ruiter (2013) find that robbers “attack near their own 6 homes, on easily accessible blocks, where legal and illegal cash economies are present…” (pg. 199). Cash economies are characteristic of fringe lending establishments; consumers routinely exit check cashing, payday lending, and pawnshop outlets with fairly large sums of cash in their pockets, drawing potential criminals who find these targets attractive to the area. Thus, victims and commercial premises that carry large amounts of cash are attractive targets, and places where such targets are abundant are likely to be attractive places for robbery. This is characteristic of fringe lending outlets, which Bernasco and Block (2009:101) describe as “good hunting grounds for robbers” (pg. 101). Another example of how fringe lenders serve as crime attractors can be seen with respect to pawnshops, which are linked with facilitation of traffic in stolen goods (Fass and Francis 2009:158; McCord et al. 1997:299). As criminals frequently use pawn brokers to exchange stolen goods for money, pawnshops have been referred to as the modern thief’s automatic cash machine (Glover and Larubbia 1996). Cromwell’s (1991) interviews of apprehended burglars in Texas showed that 18% used pawnshops as a primary method of disposal. Fass and Francis’ (2009) analysis of transcripts from interviews by Wright and Decker (1993) of 100 burglars in St. Louis, Missouri suggests that 42% used pawnshops for goods disposal, half of them regularly (pg. 159). They conclude, “…pawnbrokers, as omnipresent today as McDonald’s restaurants, offer thieves a potentially convenient method of disposing of merchandise, especially items with no obvious markings. Another fact…is that the population of prolific pawners contains a large segment of people with robust arrest records. …this strongly intimates that the population contains a substantial corps of habitual thieves who actually do rely on pawnbrokers for their recurrent service needs.” (Fass and Francis 2009:170).1 Note an interesting distinction is that there may be a different mix of persons who patronize pawnshops compared to those who patronize check cashers and payday lenders. Whereas the former may be more likely to be patronized by those who are at some times offenders, the latter would not disproportionately attract offenders. This may impact crime patterns around such establishments, as 1 Research finds that criminals generally prefer to sell stolen goods locally (Sutton 2010:7; Wellsmith and Burrell 2005:743). 7 patrons of pawnshops may be seen as “harder” targets than those patronizing the other fringe establishments. The above discussion becomes especially relevant when considering the economic forces behind criminal activity. For example, given the increased likelihood of debt traps associated with fringe lenders, desperate individuals may turn to crime as a way to pay off their loans. It is also reasonable to assume that the ready supply of cash and the illicit drug trade are happy partners in neighborhoods with large concentrations of fringe lenders. In particular, some increase in crime could be attributable to the manner in which fringe lenders might lubricate the cash-only drug trade. In places where cash is available on a moment’s notice to anyone with a job or government check, those wanting to fuel an addiction need not wait until payday with ample loan opportunities (Kubrin et al. 2011:441). In essence, we argue criminal target choice is spatially structured and that part of that structuring involves the presence of crime attractors in an area, which could include the presence of fringe lenders. We also argue that the effect of fringe lenders is likely to be strongest for robbery, burglary, and drug-related crimes. In line with this argument, research has found that specific land uses, and crime attractors in particular, are associated with heightened crime rates in areas, although this research examines the presence of retail establishments and commercial land uses more generally (Bernasco and Block 2009; Bernasco and Block 2011; Stucky and Ottensmann 2009; Wilcox et al. 2004). Still, as Bernasco and Block (2011) show, “…it was empirically demonstrated that blocks that host crime attractors and generators not only have elevated numbers of robbery themselves but also radiate their elevated crime risk to adjacent blocks. Thus, they do not function as lightning rods that reduce the risk of damage in their immediate environment but instead infect their immediate environment with increased risk” (pg. 51; see also Bernasco, Block and Ruiter 2013). In sum, to the extent that nonresidential land uses provide criminal opportunity, the effects should be direct such that we argue a concentration of fringe lending establishments should be associated with heightened crime rates. The second path through which criminogenic land use may impact crime is indirect, and draws upon social disorganization theory, which suggests that crime rates are higher in socially disorganized 8 neighborhoods with weakened social ties and decreased informal social control. According to the theory, weakened social ties and decreased informal control are presumed to stem from the social-structural qualities of a community, such as its socioeconomic status, ethnic and racial heterogeneity, and residential mobility patterns. Yet land use can also influence community disorganization, and thus, crime. Wilcox et al. (2004:186) point out that ineffective neighbor networks are related to physical structural qualities of an area, including the use of land. And criminogenic land uses can impede the ability of residents to maintain social control by generating street traffic, which increases the number of strangers in an area, reducing residents’ ability to tell locals from outsiders (Stucky and Ottensmann 2009:1226). Informal social control in the form of surveillance, communication, supervision, and intervention is thus thought to be one of the mechanisms intervening between land use and crime (Wilcox et al. 2004:188). Although their focus is not on fringe lenders per se, Wilcox et al. (2004) find that non-residential land use (e.g., stores or gas stations, bars or nightclubs, fast food restaurants, shopping centers or malls) contributes to heightened crime rates in Seattle, Washington neighborhoods. It is also possible that criminogenic land uses can influence crime by increasing actual or perceived neighborhood deterioration, disorder, or incivilities (Stucky and Ottensmann 2009:1226). Residents make inferences and assumptions and gather information from others about the places through which they move and nearby locations; from these, they generate more general ideas about crime and disorder locally (McCord et al. 2007:298). The closer someone lives to criminogenic land uses, the more likely his or her awareness space would be affected by those land uses and the activities and events surrounding them and, thus, the more crime and disorder he or she might perceive and report. Research finds there is more extensive physical deterioration, in the form of litter, vandalism, dilapidated properties, and abandoned properties, on street blocks with more nonresidential land use (Taylor et al. 1995). Indeed, in their study on the link between commercial land use and perceived incivilities and perceived crime, McCord et al. (2007) found that, controlling for neighborhood context, residents living closer to more crime attractors than their neighbors perceived more disorder and crime in their neighborhoods. This effect remained largely unchanged after controlling for residents’ characteristics, 9 stability, racial composition, and local crime rates. As this discussion hints, the effects of fringe lenders on crime rates may be at least partially indirect. We theorize these indirect effects would be strongest for property crime.2 A final theoretical point is warranted. As Stucky and Ottensmann (2009:1224) argue, studies generally assume that land use impacts crime irrespective of social context. As such, the effect of criminogenic land use is expected to be the same in all types of communities. We agree with Stucky and Ottensmann (2009) that such an assumption seems unwarranted, because the potential for land uses to create opportunities for crime likely depends on, among other factors, the willingness and/or capacity of occupants in an area to exercise informal social control, something also likely to vary based on neighborhood context. We theorize, therefore, that any impact of fringe banks on crime should differ depending on the neighborhood context. In particular, we suggest four characteristics that are likely to condition the fringe lending-crime relationship based on the arguments discussed above: concentrated disadvantage, residential (in)stability, racial heterogeneity, and population density. One expectation is that any positive effect of fringe lending on crime is likely to be amplified in more socially disorganized neighborhoods given such areas have relatively fewer social ties and less informal social control, and relatively higher levels of disorder and incivilities. We thus theorize a stronger relationship between fringe lenders and crime in neighborhoods with more concentrated disadvantage, residential instability, and racial heterogeneity—all markers of disorganization. In line with this expectation, while only a handful of studies have examined whether the effects of land uses on crime may be conditioned by socioeconomic characteristics of the area, these studies report that, in varying ways, disadvantaged context magnifies the criminogenic impact (cf Wilcox et al. 2004). For example, in their study of face blocks in a mid-sized southeastern U.S. city, Smith, Frazee, and Davison (2000) find that the influence of some commercial land uses (e.g., hotels, motels, bars, restaurants, gas stations) on 2 These arguments help explain why the impact of fringe lenders on crime is likely to be qualitatively different compared to, say, conventional banks or ATMS, both of which provide immediate cash to customers, thus creating criminal opportunity. Although like fringe lenders conventional banks and ATMs perform similar functions, unlike fringe lenders, banks and ATMS in a community are not associated with reduced informal social control, deterioration, disorder, and incivilities. 10 robbery is greater as the number of single-parent households in a face-block increased. And in their analysis of 1,000 X 1,000-feet square grid cells in Indianapolis, Indiana, Stucky and Ottensmann (2009) not only show that certain land uses are associated with crime but that the effect of some are dependent on the level of disadvantage (and vice versa). Thus, whereas high-density residential units enhanced the impact of disadvantage on violent crime, commerce, industry, and busy roads dampened the effect of disadvantage. Our expectations regarding population density are more mixed. On the one hand, our discussion of crime generators and attractors suggests that the effect of fringe lenders on crime should be amplified in highly populated or dense areas. Stated alternatively, if the function of fringe lenders makes them well- suited for motivated offenders to find attractive targets, then the fringe lending-crime relationship should be stronger in high population neighborhoods where there are presumably more attractive targets. On the other hand, if the presence of more people in highly populated or dense areas also serves to increase the number of capable guardians in the area, as routine activities theory would suggest, the effect of fringe lenders on crime could be alleviated—rather than amplified—in highly populated areas. This is due to the associated social control benefits through encouraging a steady stream of “eyes on the street” (Browning et al. 2010:5). In conclusion, we theorize that fringe lending establishments will be associated with heightened crime rates both because the presence of these establishments increases opportunities for crime and because they contribute to social disorganization in areas where they are located. We expect the effects to be conditioned by the socioeconomic characteristics of the area.3 Data and Methods To investigate the fringe banking-crime relationship across communities, we compiled data on fringe lenders (check cashers, payday lenders, and pawnshops), crime rates, and socio-economic and demographic characteristics of neighborhoods for the city of Los Angeles. We examine Los Angeles 3 We are careful to avoid claims of causality in the current study given our data are not longitudinal. 11 because it is a representative U.S. city (with a population of more than 3.7 million, of which non-whites, including Latinos, account for roughly 50%) and is located in a state where fringe banking has grown substantially since the early 1990s. We use census blocks as the unit of analysis to capture what we believe is a process that likely occurs at a small spatial scale. In geographic criminology, it is increasingly recognized that the appropriate spatial unit of analysis must be carefully chosen, matching the theoretical perspective that guides the analysis (Bernasco and Block 2011:36; Hipp 2007). Given that crime attractors and generators are, without exception, smaller than a neighborhood (Bernasco and Block 2011:36), we argue it is necessary to utilize such small-scale units in the analyses. Whereas an advantage of small units is that they better allow researchers to unpack spatial processes, a well-known consequence is that small units are more strongly influenced by their spatial environment than are larger ones (Bernasco and Block 2011:35). We therefore include in our analyses measures constructed at three geographic units of analysis: blocks; block groups (neighborhoods); and the five mile area surrounding the block group with an inverse distance decay function. This strategy allows us to account for socio-demographic processes at various scales that might impact the fringe banking- crime relationship. Variables Our outcome measures are counts of the number of events of six Part 1 crime types in the block: homicide, robbery, aggravated assault, burglary, larceny, and motor vehicle theft.4 The crime data were obtained from the Los Angeles Police Department and the incidents were geocoded to the block in which they occurred. Our key independent variable is the number of fringe lenders in each block. To identify fringe lenders, we used an on-line version of the Yellow Pages (see McCord et al. 2007:303). After the initial “scrape” of fringe outlets in Los Angeles and after eliminating duplicates (based on the same address), we verified the accuracy of the data using Google Street View, which provides up-to-date photos of the exact location (and surrounding areas) for each fringe lender. If the Google Street View 4 We don’t analyze rape given well-known reporting issues with this crime type (e.g., Jensen and Karpos 1993). 12 image did not match what was listed in the Yellow Pages (e.g., the photo identified a restaurant rather than fringe lender), we excluded the case from the sample (n = 33). In cases where it was unclear (e.g., the image of the establishment on Google was blocked by a tree or other obstruction, n = 34), we called the phone number provided in the Yellow Pages to verify the establishment was indeed a fringe lender. Our final sample included 340 lenders located on blocks with nonzero population, which were aggregated to blocks and block groups to capture the number of lenders in each area. We also computed the number of lenders within a series of spatial buffers: 0 to 399 feet; 400 to 799 feet; and 800 to 1200 feet (approximately ¼ mile). Finally, we constructed measures which disaggregate lenders into the three types: payday lenders, check cashers, and pawn shops. To minimize the possibility of obtaining spurious effects, we included control variables at three levels of analysis: blocks, block groups, and a 5-mile buffer around the block group. Block data were obtained from the U.S. Census FTP website (http://www2.census.gov/census_2010/04Summary_File_1/); block group data from the American Community Survey (2007-11 5-year summary) were obtained from the U.S. Census FTP website (http://www2.census.gov/acs2011_5yr/summaryfile/20072011_ACSSF_By_State_All_Tables/). We constructed a measure of concentrated disadvantage, which is a factor score computed after a factor analysis of four measures: 1) % at or below 125% of the poverty level; 2) % single-parent households; 3) average household income; and 4) % with at least a bachelor’s degree.5 The last two measures had reversed loadings in the factor score.6 We accounted for the presence of racial/ethnic minorities in the area with two measures, the % African American and the % Latino. To capture the possible effect of racial mixing we constructed a measure of racial/ethnic heterogeneity—the Herfindahl Index based on 5 We also constructed an index of concentrated disadvantage using other variables sometimes employed in such indices (e.g., percent divorced, per capita income, the poverty rate, the unemployment rate). The correlations among the various factor scores ranged from .94 to .99. Thus, the choice of variables is not crucial. 6 Given that only the percent single-parent households variable is available for blocks, we used an aerial interpolation technique utilizing ancillary data based on the technique of Flowerdew, Green, and Kehris (1991). The other variables used in the imputation model were: percent owners, racial composition, percent divorced households, percent households with children, percent vacant units, population density, and age structure (percent aged: 0-4, 5-14, 20-24, 25-29, 30-44, 45-64, 65 and up). 13 five racial/ethnic groupings (white, African-American, Latino, Asian, and other races) (Gibbs and Martin 1962), which has been used in prior neighborhoods and crime research (Hipp 2007). We constructed a measure of residential stability by computing the mean of the standardized values of % homeowners and average length of residence. We used the % homeowners as a proxy for residential stability in blocks given that the length of residence measure is not available at the block level. Given their possible criminogenic effect, we computed the % vacant units. Given that the presence of residents can affect crime by increasing the number of potential offenders, targets, and even guardians, we also included a measure of logged population at the block level, as well as the population density of the larger units. (As blocks are typically similarly sized, calculating density for these small units is not necessary). We account for the age/crime curve by including a measure of the % residents aged 16 to 29. We also accounted for other land use in the area to minimize the possibility that any detected fringe banking effects are not reducible to the physical characteristics of the area. Moreover, because zoning and other restrictions limit where fringe lenders can locate, it is important to account for the more general presence of retail establishments in the area. In particular, we constructed measures of the % of land use in the following categories: 1) industrial (both light and heavy industry); 2) office space (both low rise and high rise buildings); 3) residential (e.g., single family, multi-family); 4) retail (e.g., shopping centers, strip malls, restaurants, hair stylists, conventional banks); and 5) other (the reference category, including for example, parks, parking lots, open space, roads, etc.). We also calculated the socio-demographic measures in the 5-mile area around the block group.7 We accomplished this by constructing a spatial weights matrix with an inverse distance decay capped at five miles (beyond this point block groups are assumed to have no additional effect on the focal block), 7 We also constructed measures based on smaller (2.5 mile) buffers, and the results for our fringe bank measures were unchanged. The use of larger buffers makes little difference given the inverse distance decay function used to weight the data. 14 and multiplied this matrix by the values in the block groups for our variables of interest.8 Summary statistics for the variables used in the analyses are displayed in Table 1. <<<Table 1 about here>>> We assessed and found no evidence that multicollinearity adversely impacted our estimates. As Goldberger (1991) observed, the true concern of multicollinearity is the imprecision of estimates as captured in the standard errors, and O’Brien (2007) noted that VIF’s are only one of four elements when computing standard errors (the others are the sample size, the variance of the variable of interest, and the variance of the residuals, i.e., the r-square). Thus, although the largest VIF observed was 24.7 for the spatial lag of concentrated disadvantage, using O’Brien’s techniques and adjusting for the model R-square (.2 in an ancillary OLS model) and sample size (22,000), the standard error for this coefficient was approximately the same as one from a simple regression with a single predictor variable (thus no collinearity, by definition), a sample size of 925, and an R-square of .2 (a model that would not normally be considered problematic). Thus, our very large sample size mitigates what might otherwise be considered large VIF values; indeed, this highlights why Goldberger (1991) somewhat tongue in cheek proposed renaming multicollinearity as an problem of micronumerosity given that the imprecision of estimates can be overcome by a large sample size. Using O’Brien’s techniques make this clear. Methods Given the nature of our outcome variables (i.e., crime counts) and the exhibited overdispersion in these counts, we estimated negative binomial regression models. In all models we included the control variables described earlier. The first set of models for the six crime types includes the measures of the number of fringe banks in the block and the block group. The second set of models adds the three small- scale spatial measures of nearby fringe banks (within 0-399 feet, 400-799 feet, and 800-1200 feet). The third, fourth, and fifth sets of models substitute the measures of the three types of fringe banks for the 8 By using these buffers we avoid the boundary problem. These buffers include information on all blocks within 5 miles given that we have Census data for all blocks (both inside and outside the city). If we had instead constructed spatial buffers of crime we would encounter the boundary problem given that we do not have crime data outside the city boundaries. 15 total fringe bank measures included in the second set of models. The final set of models includes interactions between the presence of fringe banks on the block or block group and the neighborhood measures of residential stability, concentrated disadvantage, racial/ethnic composition, and population density.9 Before discussing the results, two analytical issues merit attention. First, we assessed whether there was any additional spatial autocorrelation in the residuals from our models after including our spatial measures and found no such evidence. Note that whereas the Moran’s I values for the crime types ranged from .02 for homicide and larceny to .19 for aggravated assault, suggesting some spatial clustering of crime events, the Moran’s I values for the residuals of our models were all less than .04, implying that our models effectively account for the spatial clustering in the data. That is, the residuals assess how much spatial clustering remains after accounting for our explicit spatial modeling approach. A second issue relates to identifying the causal direction of the fringe banking-crime relationship. One concern is that the level of crime in an area might impact the location of fringe lenders, which would bias our results. Although this is possible, we argue it is less of a concern for our analyses for two reasons. First, given that our analysis focuses on the spatially precise location of fringe banks in blocks, it is implausible to presume that fringe banks would choose to locate on the precise block in a larger area with the highest crime rate. Given that fringe lenders do not want their customers to be victimized, they would arguably avoid higher crime blocks to the greatest extent possible. Moreover, whereas rents are lower in higher crime areas, we are aware of no evidence that rental rates fluctuate considerably from block to block based on the local crime level. And, whereas fringe bankers arguably prefer the lowest rents (indeed, all businesses do), it is nonetheless the case for them (as for all businesses) that there is a tradeoff between lower rents and locating in a high crime area. For these reasons, we argue endogeneity is unlikely in our study (see also Stucky and Ottensmann 2009:1229). 9 We tested ancillary models in which we also included interactions between fringe banks and block or block group concentrated disadvantage. These interactions were never statistically significant, suggesting that the economic context does not constitute an important moderating effect of these relationships. 16 Second, the previous fringe lending-neighborhood crime study by Kubrin et al. (2011), which used larger units of analysis (tracts), conducted an instrumental variables analysis to test for reciprocal effects and found that the results for violent crime were unchanged compared to a model ignoring endogeneity, and found that the property crime results were even stronger for payday lenders when accounting for endogeneity. Due to the spatial complexity of the models in our analyses (e.g., fringe banks in blocks, nearby blocks, and block groups), it is nearly impossible to identify instrumental variables for each of the fringe bank measures in the models. We, therefore, cannot fully estimate instrumental variable models. However, as an approximate approach to assess the robustness of our models, we followed the strategy of Kubrin et al. (2011) and used conventional banks as an instrumental variable; in our case we used the block-level measure of conventional banks to instrument block-level fringe banks. The results for these instrumental variable models remained robust.10 Results We begin our discussion with the models that include counts of the number of fringe lenders in the block and the broader block group, along with our set of control variables. As seen in Table 2, and consistent with our predictions, the presence of more fringe banks on the block is positively associated with all crime types. This relationship exhibits statistical significance for all crime types excluding homicide (for which we have limited statistical power to detect differences for this rare crime). Our models show that each additional fringe bank on a block is associated with 50% more robberies, even after controlling for other types of land use in the area as well as for other socio-demographic characteristics in the block, block group, and surrounding 5 mile area (exp(.403)=1.496). Our results also suggest that each additional fringe bank is associated with 22% more aggravated assaults, 14% more burglaries, 27% more larcenies, and 9% more motor vehicle thefts. <<<Table 2 about here>>> 10 The coefficients for the instrumented block-level fringe banks were: 8.46 (t-value = 6.76) for robbery; 4.44 (t- value = 3.28) for aggravated assault; 6.18 (t-value =8.23) for burglary; 9.71 (t-value = 14.67) for larceny; 6.89 (t- value = 9.58) for motor vehicle theft; 6.78 (t-value = 1.37) for homicide. 17 It is notable that in these models there is minimal additional effect from the number of fringe banks in the entire block group. Each additional fringe bank in the block group is associated with only 4.5% more robberies on a focal block—a finding that underscores the micro-spatial effect of fringe lenders. As further evidence of this, we estimated ancillary models in which we aggregated crime to block groups and estimated more common “neighborhood level” models (only including the block group and spatially lagged measures). These results suggest that more fringe banks in a block group are consistently associated with higher crime rates in block groups. Our more spatially precise models, however, demonstrate that this is in fact a micro-spatial process. We next assessed whether the spatial effect of fringe banks extends beyond the local block to adjacent blocks. For these models we included measures of fringe banks within 400 feet of the focal block (generally within one block), within 401-800 feet, and within 801-1200 feet (about ¼ mile). The results are displayed in Table 3. There are several noteworthy results. First, the effect of fringe banks on crime in the focal block generally remains robust in these models. Second, the effect of fringe banks in the block group is not statistically significant in any of these models when we account for the presence of fringe banks in nearby blocks. In fact, we now see that whereas blocks with a fringe bank are associated with more larcenies, blocks that do not have a fringe bank but are within a block group which does have a fringe bank are associated with somewhat fewer larcenies. This may indicate a displacement effect in which such crimes are being pulled towards blocks with fringe banks. <<<Table 3 about here>>> We find even stronger spatial effects for robberies and aggravated assaults. A fringe bank on a neighboring block increases the level of aggravated assault in the focal block by 14%. The spatial effect extends two blocks for robberies: a lender within 400 feet increases the robbery rate by 19%, whereas a lender from 400 to 800 feet increases it by yet another 7%. Note that this is in addition to the effect of a 18 fringe bank in the focal block, which increases the robbery rate by 56%. Thus, we see that fringe banks are associated with crime not only on their local block, but also in adjacent blocks as well.11 Types of Fringe Lenders We next examined whether the fringe lender-crime association differs across the three types of lenders in our study. Turning first to the results for pawnshops (see 2nd panel of Table 3) we find that this type of fringe lender has the weakest effect on crime. The presence of a pawnshop on a block is associated with 23% more larcenies but is not significantly related to the other crime types. Moreover, there is no evidence that more pawnshops in the broader block group are associated with more crime of any type. We do, however, find a strong association between the presence of check cashers and crime rates (see 3rd Panel of Table 3). A check casher on a block is associated with 70% more robberies, roughly 40% more aggravated assaults and larcenies, and about 15% more burglaries and motor vehicle thefts. There are also micro-spatial effects, especially for robbery and larceny. For both these crime types, there is a distinct spatial decay process; a check casher within 400 feet increases the robbery rate by 16%, one from 400 to 799 feet increases it by 5%, and one from 800-1200 feet increases it by 3%. For larcenies, these values are 8%, 4%, and 2%, respectively. There is also evidence that check cashers on the adjacent block are associated with 11% more aggravated assaults. Note that whereas we observe strong micro-spatial effects, there is not an additional effect of check cashers in the block group on block crime rates. Thus, check cashers are associated not just with crime in the block in which they are located, but also with crime in nearby blocks. Turning to the 4th Panel of Table 3, we find a strong association between the presence of payday lenders and crime rates, consistent with the results reported in Kubrin et al. (2011). A payday lender on a 11 Because of the completeness of our model, any one individual variable’s share of the variance explained will typically not be very large. Furthermore, pseudo r-squares are not an exact analogue to variance explained. Nonetheless, we assessed that, on average, the inclusion of the fringe bank measure increased the pseudo r- square 0.24%. With the exception of the population measures which had a larger impact on the pseudo r-square, the inclusion of the other variables in the model one at a time only increased the pseudo r-square between 0.12% and 3.06%. 19 block is associated with 94% more robberies, 53% more aggravated assaults, 41% more larcenies, and 22% more burglaries. Similar to the results for the other fringe lenders, there is no evidence that the presence of payday lenders in the block group is associated with crime in a focal block. There is, however, evidence that payday lenders impact nearby blocks: a payday lender within 400 feet is associated with about 40% more robberies and aggravated assaults, and over 20% more motor vehicle thefts. Moderating Effect of Social Context on Fringe Bank and Crime Relationship In our final set of analyses, we examined whether the fringe bank-crime association differs depending on the neighborhood context. We assessed this by including interactions between fringe banks and the block and block group measures of concentrated disadvantage, residential stability, racial composition, and population density. We only present the significant interactions in the final models, which are displayed in Table 4. It is notable that both the racial composition and the economic context of the block did not moderate the fringe lending-crime relationship. Instead, as indicated in Table 4, we found significant interactions for residential stability (measured at the block level as percentage owners), and population density (measured at both the block and the block group level). The population density effects were the most consistently present, and of the greatest magnitude, suggesting that the presence of residents nearby is a salient context for moderating the fringe lending-crime relationship. That is, the positive association of fringe banks with crime rates is strongest in areas with low population density, but weakest in areas with high population density, indicating that the presence of nearby persons may reduce the impact of fringe banks. <<<Table 4 about here>>> We graphed these effects to visually display their magnitude. For all graphs we plotted the predicted crime rate from the model for a block with or without a fringe bank from one standard deviation below the mean to one standard deviation above the mean for the moderating variable (and at the mean on all other variables in the model). Figure 1 plots the relationship between the percentage of homeowners in the block and the robbery rate for blocks both with and without a fringe bank. This figure reveals that 20 first, blocks with a fringe bank have higher robbery rates regardless of the percentage homeowners present (given that the lines do not cross), and second, although there is a negative association between the percent owners on the block and the robbery rate when there is not a fringe bank (the bottom line), this is actually a positive association for blocks with a fringe bank (the top line). Thus, the usual protective effect of homeowners on crime rates is not evident for blocks with a fringe lender. <<<Figure 1 about here>>> We see in Figure 2 that denser population in the surrounding block group ameliorates, to some degree, the impact of fringe banks on robbery. Whereas there is little relationship between block-group population density and the block-level robbery rate in blocks without a fringe bank (the bottom line in the Figure), the fringe lender-robbery association is noticeably reduced as the population density of the block group increases (the top line in the Figure). The gap in the robbery rate between blocks with and without fringe lenders is more than cut in half when the block is in a high population density block group rather than a low one. <<<Figure 2 about here>>> The effect of block group population density is similar for aggravated assaults. Although not shown, the plotted figure is very similar to Figure 2. Thus, whereas increasing population density has no effect on the aggravated assault rate for a block without a fringe bank, the deleterious impact of a fringe bank on the aggravated assault rate is greatly diminished as block group population density increases. Although not presented here, we see a similar interaction patterns for the property crimes of larceny and motor vehicle theft; in particular, the effect of fringe banks for each crime type is increased as the percentage owners on the focal block increases, but it is diminished as the population density of the block group increases. The one additional pattern we find with the property crime interactions is that population density in the focal block also diminishes the fringe lending-crime association, though this is a less dramatic relationship than the others graphed. When plotted, the percentage gap in the motor vehicle theft rate between a block with and without a fringe bank narrows as the block population increases. 21 Discussion and Conclusions More Americans than ever are leaving conventional banking behind. From 2009-2011, 821,000 households opted out; during this same time period, 17 million adults were without a checking or savings account and 51 million adults had a bank account but continued to patronize payday lenders, check cashers, pawnshops and other fringe outlets (Douglas 2012). This raises the question of whether this has consequences for neighborhood crime, and the results of our study suggest this is a legitimate concern. One key takeaway point of the study is that the presence of fringe banks on a block is consistently related to higher levels of crime, even after controlling for concentrated disadvantage, different types of land use, and other correlates. Consistent with expectations, this relationship was strongest for the crime of robbery, which implies that the customers of fringe banks may be at higher risk of robbery victimization. Although we cannot say for sure who is being robbed, it is nonetheless the case that such establishments often appear to be located near a robbery hot spot. There are also elevated levels of other types of crimes near fringe banks, such as larcenies and aggravated assaults, indicating a relatively high crime environment. A second important takeaway point is the spatial impact of fringe lenders. On the one hand, we found no evidence that fringe banks within a block group are associated with crime, after taking into account their more micro spatial presence. On the other hand, not only did the presence of a fringe bank impact crime on the local block, but it also often impacted crime on adjacent blocks. This effect was most pronounced for robberies, which conforms to our expectations. Elevated robbery levels are consistent with the notion that customers of fringe banks are carrying large sums of cash and therefore are suitable targets both on the block on which the fringe bank is located, as well as on nearby blocks where customers may pass through. There was also some evidence of elevated aggravated assault rates on blocks adjacent to those with fringe banks. A third key takeaway point is that the relationship with crime varies by type of fringe lender. We found that pawnshops are not associated with crime. In contrast, payday lenders and check cashers appear to be more strongly associated with crime rates. We suggest this difference may be due to the mix of 22 persons patronizing pawnshops compared to payday lenders and check cashers, if offenders indeed more frequently utilize pawnshops. This is clearly speculative, but suggests an avenue for future research. The presence of a check casher or a payday lender is associated with elevated levels of robbery and larceny on the focal block, as well as in blocks up to 800 feet away. There are also more aggravated assaults on the block where a check casher or payday lender is located, again suggesting a crime hot spot. These results demonstrate the importance of measuring these processes at a precise spatial scale. Our fourth important takeaway point is that the fringe lender-crime association is moderated by the local context. Most notably, the presence of more residents nearby—as measured by population density—moderates this relationship. As a consequence, the positive relationship between fringe lenders and crime is somewhat ameliorated if the lender is located in a neighborhood with high population density, perhaps because there are more “eyes and ears” on the street. Likewise, the strongest increase in crime occurs when a fringe bank is located in a low population density neighborhood. These contextual effects highlight that whereas the direct effect of fringe banks is a spatially micro one, the larger context of the neighborhood nonetheless has important consequences. In sum, our collective findings suggest that communities with fringe banks are at an “ecological disadvantage” (St. Jean 2007) relative to their counterparts. This disadvantage equates with significantly higher crime rates in communities where payday lenders, check cashers, and pawnshops are located. Our findings should be interpreted in the context of the study’s weaknesses. We examine neighborhoods in only one city, Los Angeles. Although it is unlikely that the fringe lending establishments examined here would have a different relationship with crime in other cities (indeed our findings are consistent with those reported in Kubrin et al. 2011, the only other published study on fringe lending and neighborhood crime rates of which we are aware), such comparisons with other cities in future research would help clarify the generality of our findings. Perhaps more importantly, however, our cross-sectional approach does not allow us to establish causal ordering. Although our theoretical framework emphasizes the criminogenic impact of fringe lenders, our cross-sectional analysis constrained us to examining the association between fringe lenders 23 and crime. While one strategy with cross-sectional data is to use instrumental variables to tease apart such effects, it is exceedingly difficult to detect instruments for models with measures at so many spatial aggregations. Given the difficulty of locating instruments for our spatially-complex model, we suggest caution in interpreting causal effects in the absence of longitudinal data. We did, however, estimate ancillary models including an instrument for just the block-level fringe banks, and the results were robust; but these are only suggestive given the need to instrument the other spatial fringe bank measures as well. Although we have suggested it is unlikely that fringe lenders would choose to locate on the specific high crime block in a neighborhood (which is what would be necessary to explain our findings), it is nonetheless an empirical question that should be addressed in the future with longitudinal data. Finally, as is the case with most research on land use and crime more generally, we were unable to directly measure the mechanisms hypothesized to be associated with heightened crime rates in areas with concentrations of fringe lenders: opportunity, informal social control, or incivilities. As such, while our results indicate that the presence of fringe lenders is associated with neighborhood crime rates, we cannot ascertain the mechanisms that account for this association. Fully explicating the role of fringe banking in generating crime awaits a study that can incorporate measurements of these key intervening variables (see also Stucky and Ottensmann 2009:1252). We offer several additional important avenues for future research beyond empirically examining both the direct and indirect mechanisms behind the fringe lending-crime association. Certainly expanding the investigation beyond single cities is an important next step as is examining the relationship between fringe lenders and neighborhood crime rates in a longitudinal framework. Concerning the latter, this is critical both for fully determining causality as well as recognizing the extreme growth in fringe lending that has occurred over the decades. For example, while check cashing outlets first came into existence in the 1930s in Chicago and New York City, and the industry did not expand beyond the five or six largest urban areas of the U.S. until the 1990s, the number of check-cashing outlets grew quite rapidly from the early 1980s through the mid-1990s (Prager 2009). Likewise, while payday lenders were virtually nonexistent in 1990, by 2006 more than 15,000 outlets extended $25 billion in credit to consumers 24 (Lawrence and Elliehausen 2008:299) and by 2009, more than 22,000 locations originated more than $27 billion in loan volume annually (Parrish and King 2009:11). Given these statistics, it is important to determine how changes in the industry may be associated with changes in crime rates over time— although the nature of this relationship is not so straightforward. Given that fringe banking rapidly increased over a period of time when crime, particularly violent crime, rapidly decreased, an interesting puzzle for researchers involves determining how the two trends co-vary in light of other changes that occurred during the crime drop. Related to this, also of interest is the linkage between fringe lending and broader banking-related dynamics like foreclosures, conventional home mortgage lending, and subprime lending, which have increasingly become the focus of researchers, and their impacts on crime. One last future direction involves investigating the impact of the most recent trend in fringe lending—the migration from storefront fringe banking stores to online lenders. This migration has occurred most extensively with payday lenders who, in an attempt to avoid unfavorable state regulation, are turning to Internet-based payday loans. Online payday lenders incorporate in states with less restrictive payday lending laws, operate from states that do not require licensing, or locate themselves outside the U.S. and purport to make loans subject to the laws of their “home country” (Burt et al. 2006). With 15 states currently banning payday loans, a growing number of lenders have set up online operations in more hospitable states or in off-shore locales such as Belize, Malta, and the West Indies. Unbeknownst to many, major banks like JPMorgan, Chase, Bank of America, and Wells Fargo have quickly become behind-the-scenes allies of Internet-based payday lenders that offer short-term loans with interest rates sometimes exceeding 500 percent (Silver-Greenberg 2013). While the loans are not made by these banks per se, they serve as a critical link for the lenders, enabling them to withdraw payments automatically from borrowers’ bank accounts, even in states where the loans are banned entirely. While there are no exact measures of how many lenders have migrated online, a report by the Pew Charitable Trusts estimates that roughly three million Americans obtained an Internet payday loan in 2010. This disturbing trend calls for more research on fringe lenders generally, as well as the implications of this trend for the 25 customers who use them. In particular, we would be interested in determining how this may change the nature of the spatial relationship between fringe lenders and crime. References Bernasco, Wim and Richard Block. 2009. “Where Offenders Choose to Attack: A Discrete Choice Model of Robberies in Chicago.” Criminology 47:93-130. Bernasco, Wim and Richard Block. 2011. “Robberies in Chicago: A Block-Level Analysis of the Influence of Crime Generators, Crime Attractors, and Offender Anchor Points.” Journal of Research in Crime and Delinquency 48:33-57. Bernasco, Wim, Richard Block, and Stijn Ruiter. 2013. “Go Where the Money Is: Modeling Street Robbers’ Location Choices.” Journal of Economic Geography 13:119-143. Brantingham, P. J. and P. L. Brantingham. 1993. “Nodes, Paths, and Edges: Considerations on the Complexity of Crime and the Physical Environment.” Journal of Environmental Psychology 13:3-28. Browning, Christopher R., Reginald A. Byron, Catherine A. Calder, Lauren J. Krivo, Mei-Po Kwan, Jae-Yong Lee, and Ruth D. Peterson. 2010. “Commercial Density, Residential Concentration, and Crime: Land Use Patterns and Violence in Neighborhood Context.” Journal of Research in Crime and Delinquency 47:329-357. Burt, Frank, Farrokh Jhabvala, Jason Kairalla, and Ari Gerstin. 2006. “Refund Anticipation, Payday, and Auto Title Loans: A Survey of Select Fringe Lending Products.” http://www.jordenburt.com/industry-updates-978.html Caskey, John P. 1994. Fringe Banking: Check-Cashing Outlets, Pawnshops, and the Poor. New York: Russell Sage Foundation. Consumer Federation of America. Payday Loan Consumer Information. http://www.paydayloaninfo.org/facts#4. Last accessed on June 4, 2013. Cover, Jane, Amy F. Spring, and Rachel G. Kleit. 2011. “Minorities on the Margins? The Spatial 26 Organization of Fringe Banking Services.” Journal of Urban Affairs 33:317-344. Cover, Jane, Christopher S. Fowler, and Rachel Garshick Kleit. 2011. “The Geography of Fringe Banking.” Working paper accessible at: http://csfowler.com/drupal/sites/default/files/sites/default/files/documents/The_Geography_of_Fri nge_Banking_submit.pdf Damar, Evren H. 2009. “Why Do Payday Lenders Enter Local Markets? Evidence from Oregon.” Review of Industrial Organization 34:173-191. Douglas, Danielle. 2012. “More Americans opting out of banking system.” September 12. Washington Post. http://www.washingtonpost.com/business/economy/more-americans-opting- out-of-banking-system/2012/09/12/6380b986-fcf1-11e1-a31e-804fccb658f9_story.html. FDIC. 2012. 2011 FDIC National Survey of Unbanked and Underbanked Households. Washington, D.C.: Federal Deposit Insurance Corporation. Fass, Simon and Janice Francis. 2004. “Where Have all the Hot Goods Gone? The Role of Pawnshops.” Journal of Research in Crime and Delinquency 41:156-179. Flowerdew, Robin, Mick Green, and Evangelos Kehris. 1991. “Using Areal Interpolation Methods in Geographic Information Systems.” Papers in Regional Science 70:303-315. Fox, Jean A. 2007. “Fringe Bankers: Economic Predators or a New Financial Services Model?” Western New England Law Review 30:135-149. Gallmeyer Alice and Wade T. Roberts. 2009. “Payday Lenders and Economically Distressed Communities: A Spatial Analysis of Financial Predation.” The Social Science Journal 46:521- 538. Gibbs, Jack and Walter Martin. 1962. “Urbanization, Technology, and the Division of Labour: International Patterns.” American Sociological Review 27:667–677. Graves, Steven M. 2003. “Landscapes of Predation, Landscapes of Neglect: A Location Analysis of Payday Lenders and Banks.” The Professional Geographer 55:303-317. Hipp, John R. 2007. “Income Inequality, Race, and Place: Does the Distribution of Race and Class within 27 Neighborhoods Affect Crime Rates?” Criminology 45:665-697. Jensen, Gary F. and Maryaltani Karpos. 1993. “Managing Rape: Exploratory Research on the Behavior of Rape Statistics.” Criminology 31:363-385. King, Uriah, Leslie Parrish, and Ozlem Tanik. 2006. Financial Quicksand: Payday Lending Sinks Borrowers in Debt with $4.2 Billion in Predatory Fees Every Year. Washington, D.C.: Center for Responsible Lending. http://www.responsiblelending.org/payday-lending/research- analysis/rr012-Financial_Quicksand-1106.pdf Kubrin, Charis E., Gregory D. Squires, Steven M. Graves, and Graham C. Ousey. 2011. “Does Fringe Banking Exacerbate Neighborhood Crime Rates? Investigating the Social Ecology of Payday Lending.” Criminology and Public Policy 10: 437-466. Lawrence, Edward C. and Gregory Elliehausen. 2008. “A Comparative Analysis of Payday Loan Customers.” Contemporary Economic Policy 26:299-316. Levine, Samuel. 1913. The Business of Pawnbroking: A Guide and a Defense. New York: D. Halpern Company. McCleary, Richard and Alan C. Weinstein. 2009. “Do ‘Off-Site’ Adult Businesses Have Secondary Effects? Legal Doctrine, Social Theory, and Empirical Evidence.” Law and Policy 31:217-235. McCord, Eric S., Jerry H. Ratcliffe, R. Marie Garcia, and Ralph B. Taylor. 2007. “Nonresidential Crime Attractors and Generators Elevate Perceived Neighborhood Crime and Incivilities.” Journal of Research in Crime and Delinquency 44:295-320. Melzer, Brian T. 2011. “The Real Costs of Credit Access: Evidence from the Payday Lending Market.” Quarterly Journal of Economics 126:517-555. National Pawnbreakers Assocation. 2012 Trend Survey. http://pawnshopstoday.com/News.html. Last accessed June 4, 2013. Peterson, Christopher Lewis and Steven M. Graves. 2005. “Predatory Lending and the Military: 28 The Law and Geography of 'Payday' Loans in Military Towns.” Ohio State Law Journal 66:653- 832. Prager, Robin A 2009. Determinants of the Locations of Payday Lenders, Pawnshops and Check- Cashing Outlets. Washington, D.C.: Federal Reserve. Accessed at: http://www.federalreserve.gov/pubs/feds/2009/200933/200933pap.pdf Smith, William R., Sharon Glave Frazee, and Elizabeth L. Davison. 2000. “Furthering the Integration of Routine Activity and Social Disorganization Theories: Small Units of Analysis and the Study of Street Robbery as a Diffusion Process.” Criminology 38:489-523. Stegman, Michael A. 2007. “Payday Lending: A Business Model that Encourages Chronic Borrowing.” Journal of Economic Perspectives 21:169-190. St. Jean, Peter K. 2007. Pockets of Crime, Broken Windows, Collective Efficacy, and the Criminal Point of View. Chicago: University of Chicago Press. Stegman, Michael A. and Robert Faris. 2003. Payday Lending: A Business Model that Encourages Chronic Borrowing.” Economic Development Quarterly 17:8-32. Stucky, Thomas D. and John R. Ottensmann. 2009. “Land Use and Violent Crime.” Criminology 47:1223-1264. Sutton, Michael. 2010. “Stolen Goods Markets.” Office of Community Oriented Policing Services, U.S. Department of Justice. Report can be accessed here: http://www.popcenter.org/problems/stolen_goods. Taylor, Ralph B. and Denise Gottfredson. 1986. “Environmental Design, Crime and Prevention: An Examination of Community Dynamics.” Pp. 387-416 in Communities and Crime, edited by A. J. Reiss and Michael Tonry. Chicago: University of Chicago Press. Taylor, Ralph B., Barbara Koons, E. Kurtz, J. Greene, and D. Perkins. 1995. “Streetblocks with More Nonresidential Landuse Have More Physical Deterioration: Evidence from Baltimore and Philadelphia.” Urban Affairs Review 30:120-136. Temkin, K. and N. Sawyer. 2004. “Alternative Financial Services Providers.” Fannie Mae 29 Foundation. Report can be accessed here: http://www.urban.org/UploadedPDF/410935_AltFinServProviders.pdf Wellsmith, Melanie and Amy Burrell. 2005. “The Influence of Purchase Price and Ownership Levels on Theft Targets: The Example of Domestic Burglary.” British Journal of Criminology 45:741-764. Wilcox, Pamela, Neil Quisenberry, Debra T. Cabrera, and Shayne Jones. 2004. “Busy Places and Broken Windows? Toward Defining the Role of Physical Structure and Process in Community Crime Models.” Sociological Quarterly 45:185-207. Wright, Richard T. and Scott H. Decker. 1993. “Exploring the House Burglar’s Perspective: Observing and Interviewing Offenders in St. Louis, 1989-1990” [Computer file]. St. Louis, MO: Richard Wright and Scott H. Decker [producers], 1992. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor]. Wright, Richard T. and Scott H. Decker. 1997. Armed Robbers in Action. Boston: Northeastern University Press. 30 Blocks Mean SD Mean SD Concentrated disadvantage 0.0 13.0 10.4 9.4 % owners 54.8 33.2 23.7 26.6 Racial/ethnic heterogeneity 41.2 19.9 40.8 22.9 % black 10.3 19.6 14.8 23.6 % Latino 40.1 32.1 52.8 31.6 % vacant units 5.8 7.4 7.1 8.6 % aged 16 to 29 20.6 10.9 24.3 12.8 Fringe banks 0.0 0.0 1.1 0.4 % industrial 4.7 15.1 5.1 16.8 % offices 4.7 12.0 7.0 15.3 % residential 64.2 35.8 52.5 37.4 % retail 10.6 19.4 22.0 27.2 Block groups Concentrated disadvantage 1.2 12.2 6.1 9.4 Residential stability -0.1 0.9 -0.5 0.7 Racial/ethnic heterogeneity 44.5 18.1 45.4 18.6 % black 10.3 17.4 10.7 16.6 % Latino 44.4 30.5 54.9 27.6 Population density 16.4 14.3 17.5 13.9 % vacant units 6.6 6.6 6.9 6.9 % aged 16 to 29 22.8 10.5 24.8 7.8 Fringe banks 0.0 0.0 1.4 0.7 Surrounding 5 miles Concentrated disadvantage 0.1 1.0 0.3 0.8 Residential stability -0.1 0.9 -0.3 0.9 % black 9.6 9.3 9.4 8.9 % Latino 47.2 18.1 52.2 15.8 % vacant units 6.9 1.2 6.8 1.2 No Fringe banks Has fringe banks Table 1. Summary statistics of variables used in analyses, split by blocks with and without fringe banks 31 Block measures Fringe banks 0.1460 0.4029 **0.1951 *0.1290 *0.2417 **0.0886 † (0.64) (5.12) (2.38)(2.34) (5.24)(1.68) % industrial land use -0.3834 -0.3400 **-0.2314 †0.1400 0.0651 0.5159 ** -(0.79) -(2.61) -(1.67)(1.57) (0.91)(6.19) % office space land use -0.7383 -0.2266 †-0.4864 **-0.2628 **-0.1721 *-0.0408 -(1.41) -(1.70) -(3.35) -(3.21) -(2.49) -(0.49) % residential land use -0.4119 -1.2479 **-1.0316 **-0.5337 **-1.1309 **-0.4255 ** -(1.54) -(17.19) -(13.15) -(11.93) -(30.68) -(8.98) % retail land use 0.6517 †1.9049 **1.0164 **0.4552 **1.3386 **0.6101 ** (1.83) (18.65) (9.44)(6.71) (24.58)(9.01) Concentrated disadvantage 0.0046 0.0041 †0.0060 *0.0051 **0.0016 -0.0023 (0.51) (1.69) (2.32)(3.48) (1.29) -(1.53) Racial/ethnic heterogeneity 0.0052 0.0029 **0.0044 **0.0014 *0.0017 **0.0027 ** (1.21) (2.79) (3.93)(2.34) (3.35)(4.10) % black 0.0100 0.0019 0.0096 **0.0003 0.0002 0.0024 * (1.59) (1.19) (5.59)(0.30) (0.20)(2.15) % Latino 0.0062 0.0030 **0.0062 **-0.0041 **-0.0051 **0.0032 ** (1.19) (2.70) (4.90) -(6.23) -(9.63)(4.56) Table 2. Predicting levels of crime in blocks using measures in blocks, block groups, and the surrounding area. Homicide Robbery Aggravated assault Burglary Larceny Motor vehicle theft 32 % owners -0.0050 †-0.0025 **-0.0048 **0.0034 **-0.0025 **-0.0047 ** -(1.70) -(3.48) -(5.97)(7.83) -(6.94) -(10.35) % vacant units 0.0166 *0.0102 **0.0178 **0.0090 **0.0064 **0.0057 ** (2.42) (5.12) (8.42)(6.89) (6.17)(4.32) Population (logged)0.5600 **0.5152 **0.6086 **0.6423 **0.6297 **0.6425 ** (8.46) (30.21) (33.03) (61.94) (75.65) (59.72) % aged 16 to 29 0.0123 †0.0006 -0.0017 -0.0022 †0.0029 **0.0026 * (1.72) (0.32) -(0.87) -(1.95) (3.23)(2.19) Block group measures Fringe banks 0.1079 0.0436 †0.0158 0.0035 -0.0154 -0.0409 * (1.49) (1.79) (0.62)(0.22) -(1.12) -(2.57) Concentrated disadvantage 0.0010 0.0053 0.0128 **-0.0012 -0.0021 0.0044 * (0.07) (1.52) (3.46) -(0.62) -(1.25)(2.09) Racial/ethnic heterogeneity -0.0002 0.0011 0.0004 0.0010 -0.0001 0.0014 * -(0.05) (1.02) (0.32)(1.54) -(0.20)(2.02) % black 0.0184 **0.0017 0.0055 **-0.0001 -0.0002 -0.0002 (2.62) (0.93) (2.89) -(0.06) -(0.25) -(0.21) % Latino 0.0134 *0.0027 *0.0044 **-0.0015 *-0.0020 **0.0010 (2.26) (2.06) (3.07) -(2.07) -(3.22)(1.24) Residential stability -0.2723 *-0.0242 0.0172 -0.0074 -0.0546 **-0.0236 -(2.50) -(0.85) (0.57) -(0.47) -(4.00) -(1.38) 33 % vacant units 0.0018 -0.0045 *0.0012 -0.0022 †-0.0018 †-0.0020 (0.24) -(2.04) (0.53) -(1.71) -(1.67) -(1.52) Population density -0.0017 0.0029 *-0.0009 -0.0043 **-0.0064 **-0.0040 ** -(0.37) (2.01) -(0.65) -(4.71) -(7.86) -(4.35) % aged 16 to 29 -0.0138 †0.0035 †0.0003 0.0036 **0.0010 0.0009 -(1.94) (1.93) (0.13)(3.45) (1.18)(0.79) Area surrounding block group (2 miles) Concentrated disadvantage 0.1758 0.3363 **0.4350 **0.2834 **0.1540 **-0.0910 * (0.69) (4.75) (5.94)(7.12) (4.49) -(2.20) % black -0.0044 0.0169 **-0.0003 0.0088 **-0.0065 **0.0167 ** -(0.25) (3.56) -(0.06)(3.37) -(2.79)(5.84) % Latino 0.0117 -0.0021 -0.0076 *-0.0074 **-0.0029 †0.0148 ** (0.99) -(0.66) -(2.28) -(4.19) -(1.87)(8.02) Residential stability 0.1731 -0.2284 **0.1257 **-0.0238 -0.0691 **-0.1321 ** (1.56) -(8.26) (4.26) -(1.62) -(5.42) -(8.09) % vacant units 0.0292 -0.0822 **0.0101 -0.1198 **-0.0557 **-0.0883 ** (0.37) -(4.11) (0.47) -(10.97) -(6.12) -(7.42) Intercept -4.0225 -10.1162 **-0.9077 -12.7686 **-5.3673 **-11.6641 ** -(0.56) -(5.68) -(0.47) -(13.16) -(6.65) -(10.96) Note: ** p < .01; * p < .05; † p < .10. T-values in parentheses. N= 22,151 blocks. Negative binomial regression models 34 Fringe banks Block 0.1323 0.4422 **0.2034 *0.1249 *0.2526 **0.0787 (0.57) (5.54) (2.46) (2.25) (5.42) (1.48) Block group 0.1172 -0.0050 0.0048 0.0083 -0.0275 †-0.0299 (1.37) -(0.18) (0.16) (0.45) -(1.73) -(1.63) Within 400 feet 0.0806 0.1698 *0.1333 *-0.0177 0.0548 0.0607 (0.44) (2.53) (2.02) -(0.38) (1.38) (1.42) Within 800 feet -0.0819 0.0693 *-0.0064 -0.0202 0.0250 -0.0382 † -(0.84) (2.31) -(0.21) -(1.01) (1.43) -(1.96) Within 1200 feet 0.0326 0.0210 -0.0040 0.0114 -0.0059 -0.0165 (0.48) (0.93) -(0.17) (0.79) -(0.46) -(1.18) Pawnshops Block -0.0678 0.1977 -0.0784 0.0686 0.2093 *0.0245 -(0.14) (1.10) -(0.42) (0.57) (2.02) (0.21) Block group 0.2722 -0.1145 †-0.0573 -0.0012 -0.0315 0.0204 (1.48) -(1.68) -(0.82) -(0.03) -(0.87) (0.48) Within 400 feet -0.0101 0.1212 0.2154 0.0207 -0.1006 -0.1476 -(0.03) (0.88) (1.53) (0.21) -(1.21) -(1.52) Within 800 feet -0.0502 0.1877 **-0.0084 -0.0121 -0.0212 -0.1266 ** -(0.27) (3.22) -(0.14) -(0.30) -(0.61) -(3.11) Within 1200 feet 0.0293 -0.0105 -0.0077 0.0138 -0.0701 **-0.0780 ** (0.22) -(0.24) -(0.17) (0.48) -(2.76) -(2.70) Aggravated assaultRobberyHomicide Burglary Larceny Motor vehicle theft Table 3. Testing effect of fringe banks in nearby blocks on crime in focal block Homicide Robbery Aggravated assault Burglary Larceny Motor vehicle theft 35 Check cashers Block 0.1868 0.5290 **0.3120 **0.1515 *0.3140 **0.1198 † (0.63) (5.65) (3.20) (2.23) (5.46) (1.84) Block group 0.0877 0.0263 0.0143 0.0124 -0.0303 -0.0460 * (0.80) (0.79) (0.41) (0.56) -(1.60) -(2.09) Within 400 feet 0.0128 0.1489 *0.1004 †-0.0283 0.0745 *0.0514 (0.07) (2.55) (1.71) -(0.67) (2.21) (1.35) Within 800 feet -0.0691 0.0506 †0.0278 -0.0131 0.0430 **-0.0005 -(0.74) (1.85) (1.00) -(0.68) (2.69) -(0.03) Within 1200 feet -0.0095 0.0336 0.0097 0.0177 0.0192 -0.0064 -(0.14) (1.62) (0.46) (1.29) (1.63) -(0.49) Payday lenders Block 0.2625 0.6601 **0.4269 **0.2015 †0.3468 **-0.0313 (0.60) (4.29) (2.68) (1.92) (3.81) -(0.30) Block group 0.3482 †0.0515 -0.0266 0.0203 -0.0157 0.0121 (1.77) (0.81) -(0.39) (0.52) -(0.47) (0.32) Within 400 feet -0.2452 0.3526 **0.3167 *-0.1070 0.1337 †0.1933 * -(0.55) (2.93) (2.57) -(1.13) (1.84) (2.45) Within 800 feet 0.0043 0.0696 -0.0326 0.0176 0.0959 **-0.0156 (0.02) (1.20) -(0.52) (0.45) (3.04) -(0.43) Within 1200 feet -0.0597 0.0544 -0.0121 0.0101 -0.0051 -0.0083 -(0.40) (1.23) -(0.26) (0.36) -(0.21) -(0.31) Note: ** p < .01; * p < .05; † p < .10. T-values in parentheses. N= 22,151 blocks. Negative binomial regression models. All models include all control variables listed in Table 2. Homicide Robbery Aggravated assault Burglary Larceny Motor vehicle theft Homicide Robbery Aggravated assault Burglary Larceny Motor vehicle theft 36 Block measures Fringe banks 0.5409 **0.3172 †0.6309 **0.6637 **0.8465 ** (3.24)(1.78)(2.58)(3.05)(3.13) % industrial land use -0.3382 **-0.2342 †0.1418 0.0693 0.5187 ** -(2.59) -(1.69)(1.59)(0.97)(6.23) % office space land use -0.2315 †-0.4872 **-0.2627 **-0.1748 *-0.0466 -(1.73) -(3.36) -(3.21) -(2.53)-(0.56) % residential land use -1.2476 **-1.0326 **-0.5323 **-1.1309 **-0.4246 ** -(17.20) -(13.17) -(11.90) -(30.70)-(8.97) % retail land use 1.9021 **1.0101 **0.4531 **1.3265 **0.5977 ** (18.63)(9.39)(6.68) (24.37)(8.83) Concentrated disadvantage 0.0041 †0.0061 *0.0051 **0.0016 -0.0022 (1.67)(2.34)(3.49)(1.34)-(1.48) % owners -0.0026 **-0.0048 **0.0034 **-0.0026 **-0.0048 ** -(3.63) -(6.00)(7.82) -(7.16) -(10.49) Racial/ethnic heterogeneity 0.0029 **0.0045 **0.0014 *0.0016 **0.0027 ** (2.80)(3.97)(2.33)(3.28)(4.10) % black 0.0018 0.0096 **0.0003 0.0001 0.0022 * (1.13)(5.59)(0.26)(0.16)(2.05) % Latino 0.0030 **0.0063 **-0.0041 **-0.0051 **0.0031 ** (2.72)(4.96) -(6.28) -(9.76)(4.46) % vacant units 0.0101 **0.0178 **0.0090 **0.0064 **0.0058 ** (5.10)(8.46)(6.89)(6.20)(4.36) Table 4. Testing effect of fringe banks on crime in focal block based on context of block and block group Aggravated assaultRobbery Burglary Motor vehicle theftLarceny 37 % aged 16 to 29 0.0005 -0.0018 -0.0022 *0.0028 **0.0025 * (0.29) -(0.89) -(1.99)(3.15)(2.11) Population (logged)0.5156 **0.6087 **0.6451 **0.6323 **0.6470 ** (30.25) (33.05) (61.65) (75.33) (59.54) Block group measures Fringe banks 0.0427 †0.0151 0.0036 -0.0153 -0.0409 ** (1.75)(0.60)(0.23) -(1.12)-(2.58) Concentrated disadvantage 0.0052 0.0128 **-0.0012 -0.0022 0.0044 * (1.48)(3.46) -(0.61) -(1.28)(2.05) Residential stability -0.0220 0.0188 -0.0072 -0.0516 **-0.0208 -(0.77)(0.62) -(0.46) -(3.78)-(1.22) Racial/ethnic heterogeneity 0.0010 0.0003 0.0010 -0.0001 0.0013 * (0.90)(0.22)(1.54) -(0.24)(1.97) % black 0.0019 0.0056 **0.0000 -0.0002 -0.0001 (1.04)(2.92) -(0.03) -(0.21)-(0.11) % Latino 0.0027 *0.0044 **-0.0015 *-0.0019 **0.0011 (2.11)(3.04) -(2.05) -(3.09)(1.34) Population density 0.0036 *-0.0005 -0.0043 **-0.0060 **-0.0037 ** (2.39) -(0.34) -(4.71) -(7.37)-(3.97) % vacant units -0.0042 †0.0014 -0.0022 †-0.0017 -0.0019 -(1.94)(0.61) -(1.70) -(1.54)-(1.42) % aged 16 to 29 0.0034 †0.0003 0.0036 **0.0010 0.0009 (1.91)(0.18)(3.46)(1.18)(0.81) 38 Area surrounding block group (2 miles) Concentrated disadvantage 0.3375 **0.4345 **0.2822 **0.1556 **-0.0920 * (4.77)(5.94)(7.09)(4.54)-(2.23) Residential stability -0.2296 **0.1257 **-0.0240 -0.0689 **-0.1316 ** -(8.31)(4.26) -(1.63) -(5.42)-(8.07) % black 0.0166 **-0.0005 0.0089 **-0.0067 **0.0167 ** (3.52) -(0.09)(3.38) -(2.86)(5.83) % Latino -0.0023 -0.0077 *-0.0074 **-0.0030 *0.0147 ** -(0.73) -(2.32) -(4.16) -(1.98)(8.00) % vacant units -0.0820 **0.0115 -0.1198 **-0.0556 **-0.0874 ** -(4.11)(0.53) -(10.98) -(6.11)-(7.36) Fringe banks X logged population (block)-0.0934 *-0.0767 †-0.1314 ** -(2.11) -(1.95)-(2.81) Fringe banks X owners (block)0.0082 *0.0054 0.0051 **0.0047 * (2.55)(1.55)(2.82)(2.27) Fringe banks X population density (block group) -0.0172 **-0.0112 *-0.0083 *-0.0079 * -(3.27) -(2.03)-(2.53)-(2.04) Intercept -3.2401 **-4.1488 **-2.8069 **-1.8259 **-3.0660 ** -(39.14) -(45.45) -(54.34) -(45.29) -(57.02) Note: ** p < .01; * p < .05; † p < .10. T-values in parentheses. N= 22,151 blocks. Negative binomial regression models 39 0.03 0.04 0.05 0.06 0.07 0.08 0.09 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87Robbery Percent owners in block Figure 1. Robbery: interaction of fringe banks and block percent owners No Fringe bank Fringe bank 40 0.03 0.04 0.05 0.06 0.07 0.08 0.09 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25Robbery Population density in block group Figure 2. Robbery: interaction of fringe banks and block group population density No Fringe bank Fringe bank By Jerzy Eisenberg-Guyot, Caislin Firth, Marieka Klawitter, and Anjum Hajat From Payday Loans To Pawnshops: Fringe Banking, The Unbanked, And Health ABSTRACT The fringe banking industry, including payday lenders and check cashers, was nearly nonexistent three decades ago. Today it generates tens of billions of dollars in annual revenue. The industry’s growth accelerated in the 1980s with financial deregulation and the working class’s declining resources. With Current Population Survey data, we used propensity score matching to investigate the relationship between fringe loan use, unbanked status, and self-rated health, hypothesizing that the material and stress effects of exposure to these financial services would be harmful to health. We found that fringe loan use was associated with 38 percent higher prevalence of poor or fair health, while being unbanked (not having one’s own bank account) was associated with 17 percent higher prevalence. Although a variety of policies could mitigate the health consequences of these exposures, expanding social welfare programs and labor protections would address the root causes of the use of fringe services and advance health equity. T hefringebankingindustryincludes paydaylenders,whichgivecustom- ers short-term loans pending their next paychecks; pawnbrokers, which buy customers’property and allowthemtorepurchaseitlateratahighercost; car-title lenders, which hold customers’titles as collateral for short-term loans; and check cash- ers, which cash checks for a fee. 1 In the US, the industry has burgeoned in recent decades. The paydaylendingindustry,whichbeganintheear- ly 1990s, 2 extended $10 billion in credit in 2001 and $48 billion in 2011. 3 The check cashing industry, which was nearly nonexistent before the mid-1970s, 4 had $58 billion in transactions in 2010. 3 Similar growth has occurred in the pawnbroker4 and car-title lending 5 industries. This growth parallels the expansion of lending through credit cards, student loans, and mort- gages.6 On the eve of the Great Recession in 2007, average US household debt peaked at 125 percent of annual disposable personal in- come, up from 60 percent in 1980. 7 Fringe borrowing is costly, and credit checks are generally not required. 5 Short-term fringe loans can carry annual percentage interest rates (APRs) of 400–600 percent. 5 Although the loans are marketed as one-time emergency loans, bor- rowersoftentakeoutmultipleloansperyearand rarelydischargethedebtsquickly.8,9 Theaverage paydayborrowerisindebtedforfivemonths and pays$520infeesandinterestforloansaveraging $375.8 One in five car-title borrowers have their vehicle seized due to default. 9 Background Growth in the fringe banking industry resulted from several factors. 10 Beginning in the 1970s, political, economic, and regulatory forces put pressure on states to loosen interest-rate caps. Federal monetary policy to control inflation in- creased long-term commercial interest rates, andthehighcostsoffundsmadeoperatingwith- doi:10.1377/hlthaff.2017.1219 HEALTH AFFAIRS 37, NO. 3 (2018): 429–437 ©2018 Project HOPE— The People-to-People Health Foundation, Inc. Jerzy Eisenberg-Guyot (jerzy@uw.edu) is a PhD student in the Department of Epidemiology, School of Public Health, at the University of Washington, in Seattle. Caislin Firth is a PhD student in the Department of Epidemiology, School of Public Health, at the University of Washington. Marieka Klawitter is a professor at the Daniel J. Evans School of Public Policy and Governance, University of Washington. Anjum Hajat is an assistant professor in the Department of Epidemiology, School of Public Health, at the University of Washington. March 2018 37:3 Health Affairs 429 Determinants Of Health Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. in state interest-rate caps difficult for banks and other lenders. Many states altered their caps or granted exemptions for certain lenders. In addi- tion, a 1978 Supreme Court decision weakened state control over lending by allowing federally chartered banks to charge customers in other states their home-state interest rates. Subse- quently, state-chartered banks successfully lob- bied Congress for the same export rights, and states weakened rate caps to attract business. Throughout the 1980s and 1990s, other regu- latory changes allowed banks to diversify their investment activities and expand across state lines, contributing to growth and consolidation inthefinancialsector.3 Historically,thebanking and credit needs of the working poor had been met by local, commonly ownedinstitutions such as credit unions and savings and loan associa- tions.4 Aslocalinstitutionsmergedwithnational banks,however,theyreducedlessprofitableser- vices, such as small loans, that catered to the needs of the working poor. 11 Moreover, banks’ heightened needs for revenue contributed to rising fees on deposit accounts, which rendered the accounts prohibitive for many low-income consumers.12 From 1977 to 1989, the prevalence of unbanked households (those without a bank account) increased from 9.5 percent to 13.5 percent. 13 Among low-income households, the prevalence increased from 29.7 percent to 40.8 percent, 13 and many unbanked households turned to fringe banks. 14 Though the prevalence of unbanked households has decreased since the 1990s, 15 cuts in social services, 16 rising costs of necessities such as health care, 17 stagnating wages,6 and concomitant declines in personal savings rates 6 have left Americans increasingly dependent on fringe loans for survival. 2 Inequities In Fringe Borrowing And The Unbanked Fringe borrowing is most common among people with low or volatile incomes, 18 and borrowers use the proceeds primarily for recurring living expenses such as rent or un expectedexpensessuchasmedicalbills.8Mirror- ing patterns in income and wealth inequity, na- tionally representative data show that past-year fringeborrowingismorecommonamongblacks (12.9 percent), Hispanics (9.7 percent), and “other”racial/ethnic groups (16.1 percent) than amongwhites (6.2 percent) and Asians(4.6 per- cent).18 It is also more common among families headed by females (14.5 percent) than those headed by males (9.7 percent) or married cou- ples (6.2 percent), and more common among peoplewithdisabilitiesthanothers(14.6percent versus 7.8 percent). 18 Discriminatory practices have contributed to these inequities by preventing people of color and women from accumulating wealth and ac- cessing certain financial programs, such as the cheap credit available to white men that fueled the post–World War II boom. 19 For example, the Federal Housing Administration encouraged redlining,wherebybanksrefusedtolendincom- munities of color. 19 Moreover, lenders often re- quired single, divorced, or widowed women to secure their mortgages with a man’s signature. 19 Although marginalized groups gained credit access in the 1960s and 1970s, today, under “reverse redlining,”accessible loans are often high-cost and risky. 20 Indeed, people of color, particularly women, were disproportionately dispossessed of wealth during the 2007–08 sub- primelendingcrisis.19Fringebanksarefrequent- ly located in poor neighborhoods with few mainstream banks and large African American populations, thereby exploiting financial dis- tress for profit. 4 The 7 percent of US households that are un- banked are especially likely to use fringe ser- vices.18Thesehouseholdsgounbankedprimarily because they lack enough money for an account, wantprivacyanddistrustbanks,orcannotafford fees.18Overdraftfees,rarebeforederegulationin the 1980s, 12 generated $32.5 billion for banks in 201521—which often sequence withdrawals from largest to smallest to maximize revenue. 3 Over- draftfeesdisproportionatelyburdenlow-income groups,andtheydosoatahighcost.Iftheywere construed as loans to account holders, typical overdraftswouldcarryAPRsofabout17,000per- cent.21 Thecostsofbeingunbankedarealsohigh, however. According to one estimate, the average unbanked family earning $25,000 per year spends $2,400 annually on check-cashing ser- vices, money orders, and bill-paying services— more than it spends on food. 22 Fringe Borrowing, The Unbanked, And Health The costs of fringe banking may exacer- bate the well-known deleterious effects of finan- cial hardship on health. 23 However, while fringe lendersclearlychargeonerousinterestrates,the financial harms of fringe borrowing relative to the alternatives are controversial. 21 Using fringe loans for recurring expenses can be especially harmful,leadingtospiralingdebtandbankrupt- cy.24Moreover,fringelendersoftenprovidemis- leading information about loan contract terms, causing borrowers to underestimate the true costs of the loan and overestimate their ability to repay the debt. 10 Nonetheless, the poor often lack options, 8 and forcertain borrowers—partic- ularly those borrowing sparingly in states with APR limits—fringe loans may be the least costly option.24 The material consequences of fringe loans aside, borrowers’health may be harmed by the stressofexcessivedebtandaccompanyingfinan- Determinants Of Health 430 Health Affairs March 2018 37:3 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. cial instability. Indebtedness is often a source of shame,7 and fringe debt may be especially stig- matized.25 Social isolation, looming default, and harassment from debt collectors also contribute to debt-induced anxiety, depression, and sui- cide.23Chronic stress puts people at risk for met- abolicandcardiovasculardiseasesbydysregulat- ing the systems that respond to stress, such as the hypothalamic-pituitary-adrenal axis and the immune and inflammatory systems, and by con- tributing to behaviors such as substance use. 26 People who use fringe services frequently face other chronic stressors, such as discrimination, thatamplifythehealtheffectsoffinancialstrain. The net stress from fringe debt, however, must bebalancedagainstthestressofthealternatives, which may include forgoing necessities or de- faulting on other loans. 3 Meanwhile, being un- banked in a largely noncash economy generates its own stress. Bills must be paid in person, at certain locations, and within certain hours, irre- spective of transportation costs, wait times, and conflicting obligations. 22 In this study we used data from the Current Population Survey (CPS) to test the relationship betweenfringeborrowing,unbankedstatus,and self-rated health. We hypothesized that fringe borrowingand being unbankedwould beassoci- ated with worse self-rated health as a result of their material and stress effects. Study Data And Methods Data The CPS is an annual survey conducted by theCensusBureautocollectworkforcedata.The Federal Deposit Insurance Corporation (FDIC) funds a biennial June supplement that focuses on fringe services and the unbanked. Questions on self-rated health are asked annually in the MarchAnnualSocialandEconomic(ASEC)Sup- plement. Households sampled for the CPS are interviewed eight times: monthly for two four- month periods, separated by an eight-month break.Inthisstudyweusedanalgorithmcreated byBrigitteMadrian27 andChristopherNekarda28 to create a person-level identifier to merge data fromtheJune2011,2013,and2015FDICsupple- mentswithdatafromtheMarch2012,2014,and 2016ASECSupplements.Weconductedanalyses onadatasetconsistingofrespondentswhowere both nonproxy respondents and household fi- nancial decision makers, to avoid misclassifica- tion of self-rated health by proxy response and because we hypothesized that stress would be most pronounced among those who bore house- hold financial responsibilities. Respondents in our sample were interviewed once for the ASEC Supplement and once for the FDIC supplement nine months later. We excluded respondents younger than age eighteen, the minimum fringe borrowing age in many states. We did not use survey weights, since merging data across sup- plements complicates weighting. The Census Bureau cleans CPS data and imputes missing values. Exposure And Outcome Variables We de- fined fringeborrowing aspast-yearuseofahouse- hold payday, pawn, or car-title loan and being unbanked aslivinginahouseholdwithoutabank account. Self-rated health was measured using a standard question (“Would you say your health ingeneralis…?”) anddichotomizedaspoor/fair versus good/very good/excellent. Confounders For the relationship between fringe borrowing and self-rated health, we iden- tified the following confounders: demographic and socioeconomic variables (age, income, edu- cation, gender, employment status, race/ethnic- ity, foreign-born status, veteran status, health insurance, and food stamp receipt), indicators of financial marginalization (unbanked status and past-year household use of check-cashing services,rent-to-ownpurchasing,andtaxrefund anticipationloans),andcorrelatesofbothfringe service access and health (metro/non-metro residence, state of residence, and year). For the relationship between unbanked status and self-rated health, we identified the same con- founders except for use of check-cashing ser- vices, rent-to-own purchasing, and tax refund anticipation loans, which we hypothesized were mediators of the relationship. All covariates aside from health insurance and food stamp receipt were measured contemporaneously with the exposures. Variable specification is dis- cussed in more detail below. Primary Analyses To disentangle the health effects of fringe borrowing and being unbanked from the health effects of confounding factors, such as having low socioeconomic status, we used a propensity score–matching approach. 29,30 Matching subjects on the propensity score, Borrowers’health may be harmed by the stress of excessive debt and accompanying financial instability. March 2018 37:3 Health Affairs 431 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. which is the probability of exposure (fringe bor- rowing or being unbanked), allows one to con- struct comparable groups for whom exposure is independentofobservedconfounders.30Because of the matching procedure, which matched un- exposed respondents (for example, those in banked households) to exposed respondents (those in unbanked households) on the propen- sity score and discarded unmatched respon- dents, propensity score–matched analyses pro- vide an estimate of the average treatment effect on the treated rather than the average treatmenteffect—assumingnounmeasuredcon- founding.29 Identifying the health effects of fringe borrowing or being unbanked on fringe borrowers or the unbanked (the “treated”) was prioritized over identifying the health effects of fringe borrowing or being unbanked on all respondents—some of whom had high or very low socioeconomic status and thus had a low probability of exposure. For the propensity score–matched analyses, we calculated each respondent’s propensity score by predicting fringe borrowing and un- banked status via logistic models that used the confounders, including squared age and income terms. Next, using the R MatchIt package, we performed nearest-neighbor matching without replacement to match each exposed respondent to up to two unexposed respondents within 0.05 propensity score standard deviations. 31 To test the relationship between fringe borrowing or unbanked statusand health in the matched sam- ples, we calculated prevalence ratios for poor or fair health via Poisson regression. 32 For each ex- posure, we calculated crude and, to address re- sidual covariate imbalance, covariate-adjusted models.31 Because of concerns about model con- vergence and positivity, in the outcome model we adjusted only for the variables that we hy- pothesized were strong confounders and might be unbalanced after matching. 33 For fringe bor- rowing, that included income; education; race/ ethnicity; unbanked status; and use of check- cashing services, rent-to-own purchasing, and tax refund anticipation loans. For unbanked sta- tus, that included income, education, and race/ ethnicity (more details on variable specification are available below). To correctly estimate the varianceresultingfrompropensityscoreestima- tion and matching, we calculated bootstrapped estimates of the coefficients and standard errors (normal approximation) by reestimating the matching and regression 1,000 times. 29,30 We as- sessed postmatching covariate balance across exposure groups by calculating the median stan- dardized mean difference 34 in each covariate over the 1,000 matched samples (see online ap- pendix A1 for details). 35 Sensitivity Analyses To assess potential un- measured confounding by factors such as wealth, other sources of debt, and baseline health, we implemented the same propensity score–matching procedure used in our primary analyses but replaced fringe borrowing with the use of check-cashing services and refund antici- pation loans—which we treated as control expo- sures. These services are used by populations similar to those that use fringe loans but are transactionalratherthandebt-creatingandthus, we hypothesized, not comparably harmful for health. If unmeasured confounding were mini- mal, we expected these exposures to have smaller health effects than fringe borrowing. We did not run sensitivity analyses for the use of rent-to-own purchasing because that service resemblesfringeloans,requiringrepeatedcostly payments. Since consumers sometimes use fringe loans to cover fallout from illness, such as medical expenses or missed work, and since our expo- sure and outcome were measured only once, we were also concerned about reverse causation— that is, poor health precipitating fringe borrow- ing. Similarly, respondents may have become unbanked as a result of financial fallout from illness. To address reverse causation, we merged the March 2011, 2013, and 2015 ASEC Supple- ments, conducted three months prior to expo- sure ascertainment, with our primary data set and excluded respondents in the ASEC Supple- ments who reported poor or fair health. Alterna- tively, we excluded those who received disability benefit income or those who were uninsured, since fringe borrowing among these respon- dents may also have resulted from poor health. Not all respondents included in our main anal- yses were interviewed in the ASEC Supplements three months before baseline, and excluding thosewhoreportedpoororfairhealth,disability benefit income, or being uninsured further reduced the sample sizes. Thus, we conducted Poisson regression on the entire samples rather than on propensity score–matched samples to ensure adequate sample sizes. These models Thecoreofthefringe banking problem is financial instability and scarce resources. Determinants Of Health 432 Health Affairs March 2018 37:3 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. were adjusted for the same confounders that we identified above, and confidence intervals were calculatedwithrobuststandarderrors.Ifreverse causation were minimal, we expected the exclu- sions not to decrease the prevalence ratio es- timates. We also tested for reverse causation by con- ducting two-stage least squares analyses, pre- dicting fringe borrowing with indicators of state-level regulations of payday loans, pawn loans, and check-cashing services. 36 See appen- dix A3 for details. 35 Limitations Our analyses had limitations. First, there may be unmeasured confounding byfactorssuchashouseholdwealth,othersourc- es of debt, or baseline health. Moreover, self-rated health may be influenced by negative affect (which was unmeasured), particularly for respondents facing other hardships. 37 Nonethe- less, we adjusted for a variety of household char- acteristics,includinguseofotherfringeservices, that may serve as proxies for the unmeasured confounders,andthesensitivityanalysesprovid- ed evidence about unmeasured confounding. Second, in our primary analyses, the expo- sures and outcome were measured only once, making reverse causation possible. However, the sensitivity analyses addressed potential re- verse causation. Third, although self-rated health is predictive of morbidity and mortality, it is less predictive among blacks and Hispanics and people of low socioeconomic status. 37,38 However, dichotomiz- ing self-rated health improves reliability. 38 Fourth,wedidnothavedataonfringeborrow- ingfrequencyoramounts,onlythatrespondents had any past-year borrowing—which prevented us from analyzing whether more frequent bor- rowing or larger loans were more harmful than less frequent borrowing or smaller loans. To our knowledge, no data sets contain more detailed information about fringe services and health. Finally, we did not use survey weights. This limited our ability to obtain estimates that were representative of the US population and did not accountforthesurveydesign,whichaffectedthe standarderrorsofourestimates.Ouruseofboot- strapped and robust standard errors might miti- gate concern about this. Study Results The fringe borrowing data set included informa- tion about 14,473 respondents, among whom 589 (4.1 percent) reported past-year fringe bor- rowing, while the unbanked data set included 15,039 respondents, among whom 603 (4.0 per- cent) reported being unbanked. 39 Both fringe borrowers and the unbanked tended to have lower socioeconomic status than nonfringe bor- rowers and the banked, reporting lower in- comes, education, and probability of health in- surance and employment. Fringe borrowers and the unbanked were also more likely to report a race/ethnicity other than non-Hispanic white. The unbanked tended to have lower socioeco- nomic status than fringe borrowers. Thematchingprocedurecreatedmatcheddata sets with a median of 1,472 respondents for fringeborrowingand 1,437for unbankedstatus. Descriptive statistics for a matched data set are showninexhibit1.Aftermatching,allcovariates aside from rent-to-own purchasing use in the fringe borrowing analysis had median standard- ized mean differences less than 0.10 (see appen- dix A1), 35 which indicates that the procedure successfully matched exposed respondents to unexposed respondents who were comparable on observed confounders. In adjusted propensity score–matched anal- yses, past-year fringe borrowing was associated with38percenthigherprevalenceof poororfair health, while being unbanked was associated with 17 percent higher prevalence (exhibit 2). Sensitivity analyses supported these findings. Past-year use of check-cashing services and tax refund anticipation loans had negligible health effects (exhibit 3). Given minimal unmeasured confounding, this is what we hypothesized, since check cashing services and tax refund an- ticipation loans are transactional rather than debt creating and thus unlikely to substantially harm health. Excluding respondents who re- ported poor or fair health before baseline did not change the fringe borrowing prevalence ra- tio and increased the unbanked status preva- lence ratio, though both estimates had poor pre- cision. Excluding respondents who reported disability income or being uninsured before baseline did not change the prevalence ratios (appendix A2). 35 Finally, two-stage least squares analyses also suggested that fringe borrowing was associated with higher prevalence of poor or fair self-rated health (appendix A3). 35 Discussion Inthisstudywefoundthatfringeborrowingand beingunbankedwereassociatedwithworseself- ratedhealth.Ouranalyseshadseveralstrengths. First,toourknowledge,thisisthefirstempirical analysis of the association between fringe bor- rowing, unbanked status, and health. Second, few public health studies have leveraged the CPS’s panel structure to follow respondents longitudinally. Third, we matched on an array of confounding factors, and after matching, all covariates were well balanced across exposure March 2018 37:3 Health Affairs 433 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. groups. Finally, sensitivity analyses indicated that reverse causation and unmeasured con- founding were unlikely explanations for the ob- served results. Nonetheless, given the limita- tions of our data, we could not rule out the influence of these factors. Policy Implications Addressing the health effects of fringe borrowing and being unbanked can be approached from three angles: regula- tions, alternative banking institutions, and so- cial welfare programs and labor protections. ▸REGULATIONS:Regulations alone are un- likely to suffice. Many states have APR limits on fringe loans—typically 36 percent, 21 which is less than a tenth of APRs charged in states with no limit. 40 Borrowing decreases after such regulations are implemented because fringe lending becomes unprofitable. 36 However, basic Exhibit 1 Descriptive statistics for a propensity score–matched sample stratified by household fringe borrowing and unbanked status Fringe borrowing Unbanked status No Yes No Yes N % N % SMD N % N % SMD N 1,006 65.6 526 34.4 986 64.6 540 35.4 Education 0.02 0.02 At least a bachelor’s degree 164 16.3 87 16.5 36 3.7 19 3.5 Some college 333 33.1 171 32.5 201 20.4 110 20.4 High school 361 35.9 193 36.7 396 40.2 215 39.8 Less than high school 148 14.7 75 14.3 353 35.8 196 36.3 US-born 902 89.7 475 90.3 0.02 731 74.1 392 72.6 0.04 Race/ethnicity 0.03 0.03 Hispanic 106 10.5 59 11.2 265 26.9 149 27.6 Non-Hispanic black 195 19.4 98 18.6 255 25.9 141 26.1 Non-Hispanic white 616 61.2 324 61.6 352 35.7 193 35.7 Other 89 8.8 45 8.6 114 11.6 57 10.6 Male 425 42.2 236 44.9 0.05 424 43.0 219 40.6 0.05 Employment status 0.03 0.05 Employed 620 61.6 218 60.5 446 45.2 234 43.3 Not in the labor force 297 29.5 159 30.2 452 45.8 252 46.7 Unemployed 89 8.8 49 9.3 88 8.9 54 10.0 Had health insurance 799 79.4 427 81.2 0.04 684 69.4 372 68.9 0.01 Year 0.03 0.04 2012 382 38.0 203 38.6 358 36.3 206 38.1 2014 351 34.9 177 33.7 339 34.4 178 33.0 2016 273 27.1 146 27.8 289 29.3 156 28.9 Residence in a metro area 0.03 0.05 Yes 765 76.0 397 75.5 748 75.9 399 73.9 No 222 22.1 117 22.2 230 23.3 136 25.2 Unknown 19 1.9 12 2.3 8 0.8 5 0.9 Veteran 95 9.4 47 8.9 0.02 37 3.8 19 3.5 0.01 Food stamps receipt 262 26.0 142 27.0 0.02 366 37.1 232 43.0 0.12 Unbanked 125 12.4 69 13.1 0.02 —a —a —a —a —a Rent-to-own purchasing use b 78 7.8 55 10.5 0.09 —a —a —a —a —a Check cash use b 213 21.2 120 22.8 0.04 —a —a —a —a —a RAL use b 66 6.6 49 9.3 0.10 —a —a —a —a —a Mean SD Mean SD SMD Mean SD Mean SD SMD Age (years) 44.3 13.2 44.1 13.5 0.01 44.3 14.4 44.1 14.1 0.02 Equivalized income ($) c 2.4 2.3 2.5 3.4 0.03 1.6 3.2 1.4 1.8 0.07 SOURCE Authors’analysis of data merged across consecutive June Federal Deposit Insurance Corporation supplements and March Annual Social and Economic Supplements of the Current Population Survey, 2011–16.NOTES The propensity score–matched sample consisted of people randomly sampled from the bootstrapped matching procedure described in the text. SMD is standardized mean difference. SD is standard deviation. RAL is refund anticipation loan.aThese variables were not matched on in the analyses of the relationship between unbanked status and health because we hypothesized they were mediators of the relationship, not confounders. bPast-year household use of service. cEquivalized income is income adjusted to household size using the following formula, used by the Organization for Economic Cooperation and Development: (household income/10000) / (1+(0.7*number of non–head of household adults +0.5*number of children). See Organization for Economic Cooperation and Development. What are equivalence scales? [Internet]. Paris: OECD; [cited 2018 Feb 5]. Available from: http://www.oecd.org/eco/growth/OECD-Note-EquivalenceScales.pdf Determinants Of Health 434 Health Affairs March 2018 37:3 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. needsmaybeleftunmetorbesatisfiedatgreater cost. Other potentially beneficial regulations, some of which may become federal, include limiting borrowing frequency and capping pay- ments based on borrowers’income.40 Some states have reported positive effects from these measures. For example, after North Carolina banned payday lending, over 90 percent of low- and middle-income households reported that the ban had neutral or positive effects on them.41 However, strict regulations may force consumerswholackotheroptionsintohigh-cost alternatives such as paying late fees. 21 Conse- quently, some researchers, pointing to states such as Colorado, have argued for moderate reg- ulations that cheapen credit without restricting supply. Nonetheless, Colorado’s 120 percent payday loan APR limit is higher than the limit supported by consumer groups. 40 Moreover, lenders often skirt regulations by disguising their services and moving online. 21,36 Concerning mainstream banks, some re- searchers have argued that giving banks and creditunionsclearerguidanceaboutpermissible underwriting practices, loan terms, and pricing and allowing them to charge realistic APRs wouldfacilitatesmall-dollarlending.40 However, providing financial services to low-income con- sumers is expensive: They often hold low depos- its, borrow small amounts, and frequently default.4 More regulation is unlikely to enable banksandcreditunionstooffersufficientafford- able services to substantially reduce the need for fringe banking. 21 Moreover, recent scandals concerning discriminatory lending, fraudulent accounts, and overdraft fees raise concerns about the role of commercial banks in low- income lending. 21 Thus, while certain regula- tions(suchaslimitsonAPRsandfeecaps)might be beneficial, in isolation they cannot be relied upontoimprovefinancialwell-beingandhealth. ▸ALTERNATIVE BANKING INSTITUTIONS:Re- cent government initiatives to provide the poor with financial services have relied on main- stream banks and credit unions. However, ini- tiatives such as the FDIC’s Small-Dollar Loan PilotProgramandtheCommunityReinvestment Act of 1977 reveal tensions between low-income communities’need for affordable services and the banks’need for profit.While the Community Reinvestment Act has encouraged banks to lend in underserved communities, those loans are often subprime. 4 Meanwhile, the Community Development Banking Act of 1994, which aimed to create community-oriented banks in low- income communities (called community devel- opment financial institutions), was premised on the proposition that these institutions could serve the poor and maintain their profitability with minimal government assistance. However, most Community Development Banking Act fundshavebeenusedforrealestateandbusiness development, not banking for the poor, and many community development financial institu- tions have struggled to survive. 4 Reconciling the needs of low-income commu- nities and mainstream commercial banks re- mainsproblematic.Inthepast,bankingservices for these communities were often provided by credit unions and savings and loan associations Exhibit 2 Association between past-year fringe borrowing or unbanked status and poor or fair health Prevalence ratio 95% CI Na Fringe borrowing Unadjusted 1.40 1.14, 1.72 1,473 Adjustedb 1.38 1.14, 1.68 1,472 Unbanked status Unadjusted 1.21 1.02, 1.43 1,434 Adjustedc 1.17 0.99, 1.39 1,437 SOURCE Authors’analysis of data merged across consecutive June Federal Deposit Insurance Corporation supplements and March Annual Social and Economic Supplements of the Current Population Survey, 2011–16.NOTES The exhibit shows prevalence ratios from Poisson models calculated on propensity score–matched samples: specifically, the ratio of prevalences of poor/ fair health among those reporting (versus not reporting) fringe borrowing or unbanked status. See the text for more explanation. CI is confidence interval. aMedian number of respondents in matched samples across bootstrap repetitions. bAdjusted for use of check cashing, rent-to-own purchasing, and refund anticipation loan services, unbanked status, income quartiles, high school education, and non-Hispanic white. cAdjusted for income quartiles, education (all categories), and race/ethnicity (all categories). Exhibit 3 Sensitivity analyses to assess potential unmeasured confounding and reverse causation in the relationship between fringe borrowing or unbanked status and self-rated health Prevalence ratio 95% CI Na Control exposures b Check cashing use in past year 1.14 0.95, 1.37 1,473 Tax refund anticipation loan use 1.01 0.72, 1.41 698 Excluding people in poor or fair health before baseline c Fringe borrowing 1.37 0.93, 2.01 7,534 Unbanked status 1.40 1.01, 1.92 7,843 SOURCE Authors’analysis of data merged across consecutive June Federal Deposit Insurance Corporation supplements and March Annual Social and Economic Supplements of the Current Population Survey, 2011–16.NOTES The exhibit shows prevalence ratios from Poisson models calculated on propensity score–matched samples for the control exposure analyses and calculated on the full sample for the reverse causation analyses: specifically, the ratio of prevalences of poor/fair health among those reporting (versus not reporting) check cashing and tax refund anticipation loan use or fringe borrowing and unbanked status. See the text for more explanation. CI is confidence interval. aMedian number of respondents in matched samples across bootstrap repetitions. bPropensity score–matched analyses were matched on the variables described in the text and adjusted for the use of fringe loans, other fringe banking services, unbanked status, income quartiles, high school education, and non-Hispanic white. If unmeasured confounding were minimal, we expected to find null or small prevalence ratio estimates. cAnalyses (not propensity score matched) adjusted for the variables described in the text. If reverse causation were minimal, we expected to find estimated prevalence ratios similar to those identified in the main analyses (see exhibit 2). March 2018 37:3 Health Affairs 435 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. that were outside the mainstream banking sec- tor. Similarly, the government could now foster appropriate services by providing community developmentfinancialinstitutionswithstronger regulatory oversight and more financial sup- port.4 Government-supported and community- led lending circles, which pool community resources to provide low-cost credit, are another option.4 Resurrecting a US Postal Service bank- ing system, which existed from 1910 to 1967 and has analogues in other countries, could address geographic barriers to banking in low-income communities (because of the ubiquity of post offices) and the costs of low-income banking (given a nonprofit mission). 4 Municipal banks could serve similar functions. 42 Finally, mobile banking, a growing industry in the US and else- where, offers inexpensive and easy-to-use ser- vices attractive to the underbanked. 3 However, they require customers to have internet access and digital literacy, which could pose a barrier for the poor and elderly, and their services are difficult to regulate. 8 ▸SOCIAL WELFARE PROGRAMS AND LABOR PROTECTIONS:With nearly half of Americans reporting that they would be unable to produce $400 cash for an emergency 22 and the common use of fringe loans for necessities, 18 the core of the fringe banking problem is financial instabil- ityandscarceresources.Robustpublicprovision of necessities such as public health programs, health care, housing, and disability assistance— coupledwithinitiativestoraiseincomes,suchas minimum wage increases and support for labor protections—would address the root causes of demand for fringe services. 18 One study found that California’s early Medicaid expansion was associated with an 11 percent reduction in pay- day borrowing, 43 while another found that each $1 increase in the state-level minimum wage was associatedwitha 40percentreduction inpayday borrowing.44 These programs also have salutary effects on other social determinants of health. 45 Addressing broader structural factors that deep- enfinancialinstabilityandpovertyformarginal- izedgroups,suchassegregationandmassincar- ceration, might also reduce fringe borrowing and improve health equity. 46 ConclusionThisresearchaddstothegrowing evidence that connects specific kinds of house- holddebtandfinancialexclusiontopoorhealth. Effectively addressing the health consequences of fringe borrowing and being unbanked will likelyrequireexpandingsocialwelfareprograms and labor protections. Future research should exploreinmoredepthhowthetwo-tierUSfinan- cial system—one for the wealthy and one for the poor—affects health and worsens health in- equities.▪ An abbreviated version of this article was presented in a poster session at the Annual Interdisciplinary Population Health Research Conference, in Austin, Texas, October 2, 2017. Anjum Hajat’s work was supported by strategic hire funds provided by the University of Washington School of Public Health and the Bill & Melinda Gates Foundation. NOTES 1 Drysdale L, Keest KE. The two-tiered consumer financial services market- place: the fringe banking system and its challenge to current thinking about the role of usury laws in today’s society. S C Law Rev. 2000; 51:589–669. 2 Rivlin G. Broke, USA: From pawn- shops to Poverty, Inc.—how the working poor became big business New York (NY): Harper Business; 2010. 3 Servon L. The unbanking of America: how the new middle class survives. Boston (MA): Houghton Mifflin Harcourt; 2017. 4 Baradaran M. How the poor got cut out of banking. Emory Law J. 2013; 62:483–548. 5 Martin N, Adams O. Grand theftauto loans: repossession and demo- graphic realities in title lending. Miss Law Rev. 2012;77(1):41–94. 6 Barba A, Pivetti M. Rising household debt: its causes and macroeconomic implications—a long-period analy- sis. Cambridge J Econ. 2009;33(1): 113–37. 7 Zimmermann C. On household debt. FRED Blog [blog on the Internet]. 2015 Jan 12 [cited 2018 Jan 10]. Available from: https://fredblog .stlouisfed.org/2015/01/on- household-debt/ 8 Bourke N, Horowitz A, Roche T. Payday lending in America: who borrows, where they borrow, and why [Internet]. Washington (DC): Pew Charitable Trusts; 2012 Jul [cited 2018 Jan 10]. Available from: http://www.pewtrusts.org/~/ media/legacy/uploadedfiles/pcs_ assets/2012/pewpaydaylending reportpdf.pdf 9 Consumer Financial Protection Bu- reau. Single-payment vehicle title lending [Internet]. Washington (DC): The Bureau; 2016 May [cited 2018 Jan 10]. Available from: http:// files.consumerfinance.gov/f/ documents/201605_cfpb_single- payment-vehicle-title-lending.pdf 10 Graves SM, Peterson CL. Predatory lending and the military: the law and geography of “payday”loans in mil- itary towns. Ohio State Law J. 2005; 66(4):653–832. 11 Minnich AA. Rational regulation of payday lending. Kans J Law Public Policy. 2006;84:84–107. 12 Pew Charitable Trusts. Consumers need protection from excessive overdraft costs [Internet]. Washing- ton (DC): Pew Charitable Trusts; 2016 Dec [cited 2018 Jan 10]. (Brief). Available from: http:// www.pewtrusts.org/~/media/ assets/2016/12/consumers_need_ protection_from_excessive_ overdraft_costs.pdf 13 Caskey JP, Peterson A. Who has a bank account and who doesn’t: 1977 and 1989. East Econ J. 1994;20(1): 61–73. 14 Peterson CL. Taming the sharks: towards a cure for the high-cost credit market. Akron (OH): Univer- sity of Akron Press; 2004. 15 Federal Reserve Board. 2016 SCF chartbook [Internet]. Washington (DC): Federal Reserve Board; 2017 Determinants Of Health 436 Health Affairs March 2018 37:3 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org. Oct 16 [cited 2018 Jan 10]. Available from: https://www.federalreserve .gov/econres/files/BulletinCharts .pdf 16 Karger H. Shortchanged: life and debt in the fringe economy. San Francisco (CA): Berrett-Koehler Publishers, Inc.; 2005. 17 Dickman SL, Woolhandler S, Bor J, McCormick D, Bor DH, Himmelstein DU. Health spending for low-, middle-, and high-income Ameri- cans, 1963–2012. Health Aff (Mill- wood). 2016;35(7):1189–96. 18 Federal Deposit Insurance Corpora- tion. 2015 FDIC National Survey of Unbanked and Underbanked Households: appendix tables [Inter- net]. Washington (DC): FDIC; 2016 Oct 20 [cited 2018 Jan 10]. Available from: https://www.fdic.gov/ householdsurvey/2015/2015 appendix.pdf 19 Baker AC. Eroding the wealth of women: gender and the subprime foreclosure crisis. Soc Serv Rev. 2014;88(1):59–91. 20 Schmitz AJ. Females on the fringe: considering gender in payday lend- ing policy. Chic Kent Law Rev. 2014;89(1):65–112. 21 Servon L. Are payday loans harmful to consumers? Response to Horowitz. J Policy Anal Manage. 2017;36(1):255–8. 22 Baradaran M. How the other half banks: exclusion, exploitation, and the threat to democracy. Cambridge (MA): Harvard University Press; 2015. 23 Richardson T, Elliott P, Roberts R. The relationship between personal unsecured debt and mental and physical health: a systematic review and meta-analysis. Clin Psychol Rev. 2013;33(8):1148–62. 24 Lim Y, Bickham T, Dinecola CM, Broussard J, Weber BE, Gregory A. Payday loan use and consumer well- being: what consumers and social workers need to know about payday loans. J Poverty. 2014;18(4):379–98. 25 Williams B. Debt for sale: a social history of the credit trap. Philadel- phia (PA): University of Pennsylva- nia Press; 2004. 26 McEwen BS, Seeman T. Protective and damaging effects of mediators of stress. Elaborating and testing the concepts of allostasis and allostatic load. Ann N Y Acad Sci. 1999;896: 30–47. 27 Madrian BC, Lefgren LJ. An ap- proach to longitudinally matching Current Population Survey (CPS) respondents. J Econ Soc Meas. 2000;26:31–62. 28 Nekarda CJ. A longitudinal analysis of the Current Population Survey: assessing the cyclical bias of geo- graphic mobility [Internet]. Unpub- lished paper [last revised 2009 May 27; cited 2018 Jan 10]. Available from: https://chrisnekarda.com/ papers/lpd-geomob_20090527.pdf 29 Oakes J, Johnson P. Propensity score matching for social epidemiology. In: Oakes JM, Kaufman JS, editors. Methods in social epidemiology. San Francisco (CA): Jossey-Bass; 2006. p. 370–93. 30 Caliendo M, Kopeinig S. Some practical guidance for the imple- mentation of propensity score matching. J Econ Surv. 2008; 22(1):31–72. 31 Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1–21. 32 Spiegelman D, Hertzmark E. Easy SAS calculations for risk or preva- lence ratios and differences. Am J Epidemiol. 2005;162(3):199–200. 33 Nguyen T-L, Collins GS, Spence J, Daurès J-P, Devereaux PJ, Landais P, et al. Double-adjustment in propen- sity score matching analysis: choos- ing a threshold for considering re- sidual imbalance. BMC Med Res Methodol. 2017;17(1):78. 34 Austin PC. An introduction to pro- pensity score methods for reducing the effects of confounding in obser- vational studies. Multivariate Behav Res. 2011;46(3):399–424. 35 To access the appendix, click on the Details tab of the article online. 36 McKernan S-M, Ratcliffe C, Kuehn D. Prohibitions, price caps, and disclosures: a look at state policies and alternative financial product use. Washington (DC): Urban Insti- tute; 2010 Nov 10. 37 Dowd JB, Zajacova A. Does the pre- dictive power of self-rated health for subsequent mortality risk vary by socioeconomic status in the US? Int J Epidemiol. 2007;36(6):1214–21. 38 Zajacova A, Dowd JB. Reliability of self-rated health in US adults. Am J Epidemiol. 2011;174(8):977–83. 39 Sample sizes differed because exposure variables had different universes. 40 Horowitz A. Payday loans harm consumers, but reform is possible. J Policy Anal Manage. 2017;36(1): 248–55. 41 Ratcliffe J, Manturuk K. North Car- olina consumers after payday lend- ing: attitudes and experiences with credit options. Chapel Hill (NC): University of North Carolina at Chapel Hill, Center for Community Capital; 2007 Nov. 42 Beitel K. Municipal banking: an overview: financing socially just and sustainable forms of urban develop- ment [Internet]. New York (NY): Roosevelt Institute; 2016 Apr [cited 2018 Jan 10]. Available from: http:// rooseveltinstitute.org/wp-content/ uploads/2016/04/Municipal- Banking-An-Overview.pdf 43 Allen H, Swanson A, Wang J, Gross T. Early Medicaid expansion associ- ated with reduced payday borrowing in California. Health Aff (Millwood). 2017;36(10):1769–76. 44 Dettling LJ, Hsu JW. Minimum wages and consumer credit: impacts on access to credit and traditional and high-cost borrowing [Internet]. Washington (DC): Board of Gover- nors of the Federal Reserve System; 2017 [cited 2018 Jan10]. (Finance and Economics Discussion Series No. 2017-010). Available from: https://www.federalreserve.gov/ econresdata/feds/2017/files/ 2017010pap.pdf 45 Navarro V.What is a national health policy? Int J Health Serv. 2007; 37(1):1–14. 46 Bailey ZD, Krieger N, Agénor M, Graves J, Linos N, Bassett MT. Structural racism and health in- equities in the USA: evidence and interventions. Lancet. 2017; 389(10077):1453–63. March 2018 37:3 Health Affairs 437 Downloaded from HealthAffairs.org on March 09, 2018. Copyright Project HOPE—The People-to-People Health Foundation, Inc. For personal use only. All rights reserved. Reuse permissions at HealthAffairs.org.     Date:  July 26, 2021  To:  Golden Valley Planning Commission  From:  Valerie Quarles, Community Development Intern  Subject:  Accessory Dwelling Unit Discussion #2    Summary   This memo is split into two parts. The first addresses the questions that the Planning Commission  had during the July 12 discussion on ADUs. Commissioners’ questions fit into two categories:  impacts not discussed during the initial presentation and measures of affordability, as they relate  to rental and construction costs.   The second portion considers the geographic extent to which ADUs might be built in Golden  Valley. While there are fewer regulations that can affect which parcels would best support an  ADU, understanding the potential distribution of ADUs across the city will give much‐needed  context as discussions continue.   Impacts  What are the potential negative impacts of ADUs and how might they be addressed?   Parking can be a contentious topic regarding ADUs, but policymakers need to be careful about  their assumptions. Portland, for instance, has an average of 0.93 extra cars for each ADU, and  data shows that half of those cars are parked on the street regularly. With fewer than 2 percent  of homes having ADUs (the highest percentage in the nation), this amount to one extra car  parked on the street for every six city blocks. A potential regulation could include requiring new  parking areas only if the ADU displaces the primary dwelling’s parking (not enough room in the  driveway). Though many places alleviate this issue by allowing on‐street parking, this is a more  salient point in Golden Valley due to the restriction on street parking in the winter.   Rental use of ADUs was described in the July 12 presentation as a tertiary use, secondary to  aging‐in‐place and multigenerational living. All of these uses should be considered equal  possibilities. Many municipalities in the Twin Cities with ADU regulations allow ADU rentals, and  they are subject to the same licensing process as any other rental property. In that vein,  discussion at the July 12 meeting landed on what can be called the “life‐cycle” of an ADU. Owners  who initially construct an ADU will often do it for a family member. When that family member no  longer needs to use the unit, the owner may choose to rent to a tenant. But this doesn’t indicate  that the ADU will always be a rental from then on. As properties with ADUs sell, and as the  population of the Twin Cities continues to age, ADU properties will be popular with  multigenerational buyers who, again, hope to house a family member. This is encouraged in part  by an owner‐occupancy requirement, which every local municipality has (with the exception of  Crystal and Stillwater). This requirement eliminates the possibility that an off‐site landlord could  purchase the entire property and rent out both units.   As both a pro and a con, the visual impact of an ADU is limited to immediate neighbors. While an  internal ADU (like a basement conversion) has very little visual impact, an addition or detached  structure can be seen. Regulations on sizing, setbacks, and minimum lot size will alleviate this  issue, as well as the degree to which neighbors are able to comment on the process pre‐ construction.   Affordability  How do rental costs of ADUs compare to similar apartments?   A 2018 study from Portland State University showed that Portland’s long‐term rental ADUs  generally rent for about the same or lower than similarly‐sized apartments. Short‐term rentals,  which make up about a quarter of the city’s ADU stock, are more lucrative. Golden Valley’s City  Council previously discussed short‐term rentals in the period leading up to Minneapolis’ Super  Bowl, and has stated that the issue of short‐term rentals will be revisited in the future if problems  arise. Any new regulation of short‐term rentals would likely fall under the City’s licensing code,  and therefore be reviewed by the City Council rather than the Planning Commission.   According to the most recent Comprehensive Plan, approximately 55 percent of rental units in  Golden Valley are affordable (to a family making 80% AMI or less) without any public subsidy.  Given that all ADUs in the coming years would be new construction, the equivalent percentage  for ADUs will likely be lower.   What factors impact the construction cost of an ADU?  The construction cost of an ADU is impacted by factors in two categories: factors that  traditionally impact home construction (excluding the cost of land) and city regulations. While  Golden Valley’s regulations will have little impact on things like material costs, other  requirements can make large differences in the cost of construction. Every additional regulation,  while often valuable, will cause timelines to stretch and prices to rise. The biggest factors include:    Whether separate utilities are required. While City staff is continuing to discuss this  interdepartmentally, most cities in the metro allow utilities to connect to the main home.    Excessive minimum parking requirements. Having to add additional pavement runs afoul  of existing regulations (extending the process) and adds material cost.    A long approval process. Again, the longer the timeline, the greater the cost, and the less  likely homeowners will follow through from beginning to end.   How do homeowners fund the construction of an ADU?   Funding an ADU is generally easier for homeowners that have established significant equity in  their home. However, there are multiple ways homeowners may acquire funds, with or without  equity.    Home equity line of credit (HELOC).    Cash‐out refinance of existing mortgage.    Personal line of credit.    Construction or renovation loan.    Cash savings.    Community development financial institution (CDFI). This is the least likely option – very  few local institutions might lend for this type of use.  What does affordability mean?    The final portion of the discussion on July 12 attempted to establish what was meant by  “affordability”. Affordability measures vary based on the stakeholder in question.    A homeowner is considering the cost of construction, and depending on who they hope to  house in the new unit:   o An aging family member will represent a cost savings by avoiding the significant  expense of assisted living.   o An adult child commuting to work or school will not require help with rent or dorm  costs.   o An unrelated renter will be a source of income, allowing the homeowner to more  easily afford their mortgage or other housing expenses.    An occupant is considering the cost of rent, depending on their situation:   o A 1‐2 person household is more affordably accessing a single‐family neighborhood,  since rental costs for an ADU are likely lower than renting a full single‐family  property.   o An aging family member or adult child is able to live with family for free or  reduced rent.    A second owner, having sought out a property with an ADU, can go straight to housing a  family member more affordably or renting for additional income without having to  consider the cost of construction.   Geography  Geographic context provides a much‐needed lens through which to view the potential extent of  ADUs in Golden Valley. Three measures so far can begin to give an idea of where it’s most likely  that homeowners will take advantage of ADU regulations: lot size, housing age, and property  value. The City’s final ordinance will likely include a minimum lot size. Housing age is not a direct  correlation – an older or newer home may not be more or less likely to support an ADU.  However, homebuilding in the 1950s often only included single‐story homes. Detached  structures, then, are probably less likely in these neighborhoods. Property values point to where  homeowners may have the most financial leverage to afford the cost of construction.   Potential cutoffs for minimum lot size could be 10,000, 12,000 or 15,000 square feet. (In acres,  that’s 0.23, 0.28, and 0.34 acres respectively). These are examples of cutoffs but certainly not the  only options.   Minimum lot size Number of R‐1 properties Location  10,000 sq ft / 0.23 acres 5700 Most R‐1 lots are at least  10,000 square feet.   12,000 sq ft / 0.28 acres 4921 Many R‐1 lots are at least  12,000 square feet, but less in  the western and northern  portions of the city.   15,000 sq ft / 0.34 acres 2405 While there are large lots in  most parts of the city, they’re  concentrated in the southern  and eastern portions of the  city.     Homes constructed from 1950‐1969  make up a significant portion of the city’s  housing stock. When considering height  regulations, policymakers should  consider the difference between a  universal height cutoff and regulating  height as it relates to the primary  structure. Requiring that and ADU be the  same height or shorter than the primary  structure may exclude much of this  portion of the city’s housing.                 Property value is again, not the only indicator of ADU likelihood but can be an important factor  when it comes to financing construction. The map below comes from the most recent  Comprehensive Plan. Properties with higher values may indicate greater equity to be leveraged.     Next Steps  Over the next two weeks, staff will continue to discuss interdepartmental impacts of ADUs, like  rental licensing, building code, and other physical considerations. A sample ordinance will be  introduced at the next Planning Commission meeting to provide a basis for discussion around  how all potential regulations, separate so far, function as a complete ordinance and impact one  another.   Resources   Home + Home: Twin Cities ADU Guidebook (Family Housing Fund)  ADU’s: Housing Options for a Growing Region (Family Housing Fund)  The ABC’s of ADUs (AARP)  KnowledgeBase Collection: Accessory Dwelling Units (American Planning Association)