Loading...
HomeMy WebLinkAboutUnderstanding the MarketUC Berkeley IURD Working Paper Series Title Understanding the Market for Secondary Units in the East Bay Permalink https://escholarship.org/uc/item/9932417c Authors Wegmann, Jake Chapple, Karen Publication Date 2012-10-01 eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA Understanding the Market for Secondary Units in the East Bay Working Paper 2012-03 Jake Wegmann and Karen Chapple October 2012 IURD  Working  Paper  WP-­‐2012-­‐03:     Understanding  the  Market  for  Secondary  Units  in  the  East  Bay   Jake  Wegmann  and  Karen  Chapple             INTRODUCTION  AND  SUMMARY   In  order  to  understand  the  potential  market  for  secondary  units,  it  is  important  to  gain  some  insight   into  the  functioning  of  the  existing  market  for  such  dwellings.  Examining  this  market  in  detail  is   made  challenging,  as  compared  to  other  residential  submarkets,  by  the  widespread  lack  of  building   and/or  zoning  permits  for  secondary  units.  Because  of  these  informal  characteristics,  market   information  that  is  routinely  published  for  other  types  of  residential  dwellings  by  third-­‐party   private  entities,  scholars,  cities,  the  US  Census  and  other  sources  is  essentially  non-­‐existent  for   secondary  units.  We  therefore  must  find  ways  to  study  what  essentially  amounts  to  a  black  market   for  secondary  unit  housing  without  relying  on  the  data  that  is  typically  available  for  housing  in   general.         This  working  paper  describes  in  detail  the  methods  and  findings  from  two  separate  techniques  that   we  used  to  investigate  the  market  for  secondary  unit  housing.  One  is  a  survey  of  homeowners   residing  within  the  rail  transit  station  areas  of  the  study  region.  The  other  is  an  analysis  of  rental   Internet  advertisement  data  for  units  for  lease  throughout  the  study  area.  Each  technique  should  be   viewed  as  a  separate  approximation,  or  “cut”  through  the  market  for  secondary  unit  housing  in  the   East  Bay.  Each  technique  has  a  differing  set  of  strengths  and  shortcomings.  The  summary  of  our   findings  from  the  two  methods  is  presented  below,  while  much  lengthier  descriptions  of  each   method  are  given  afterwards.  Together,  our  two  methods  begin  to  paint  a  picture  of  the  operation   of  a  market  that  functions  largely  hidden  in  plain  sight.     Summary  of  rental  market  analysis  results   Reviewing  the  descriptive  statistics  and  hedonic  modeling  results  presented  below,  evidence  for   several  important  characteristics  of  secondary  units  as  a  rental  submarket  distinct  from  the  overall,   mainstream  rental  market  emerges.  The  most  notable  results  from  the  two  analyses  are  as  follows:   • Approximately  16  percent%  of  single-­‐family  residential  properties  in  the  study  area  have  at   least  one  secondary  unit.  This  result  provides  evidence  that  secondary  units,  despite  being   frequently  unpermitted,  are  not  a  marginal  or  aberrational  phenomenon,  but  rather  are   widespread  throughout  East  Bay  station  areas  and  almost  certainly  throughout  the  East  Bay   flatlands.   • Secondary  units  most  commonly  take  the  physical  form  of  a  detached  rear  yard  structure,  a   converted  garage,  or  converted  rooms  in  the  main  house.  A  large  majority  were  installed  five   or  more  years  ago.  The  vast  majority,  though  not  all,  are  used  as  housing  units.     • While  not,  on  average,  cheaper  in  terms  of  price  per  square  foot,  secondary  units  are  more   likely  to  represent  smaller,  and  therefore  cheaper,  housing  packages  than  the  typical  rental   unit  in  the  East  Bay.  This  result  suggests  that  secondary  units  are  particularly  important  to   small  households  of  one  to  two  people  seeking  affordable  housing.  Secondary  units  rented   to  strangers  are  at  least  6  percent%  cheaper  than  non-­‐secondary  units  when  their   affordability  is  expressed  in  terms  of  a  proportion  of  Area  Median  Income  (AMI),  as  is   common  practice  in  the  affordable  housing  industry.   • Ownership  and  operation  by  amateurs  is  much  more  common  (though  far  from  ubiquitous)  in   secondary  units  than  in  the  general  rental  market.  This  has  important  consequences  in  the   functioning  of  the  secondary  unit  submarket.  For  instance,  there  is  at  least  modest  evidence   that  on-­‐site  secondary  unit  landlords  appear  to  offer  rents  at  a  discount  to  their  tenants  vis-­‐ à-­‐vis  their  absentee  landlord  peers.  In  addition,  secondary  unit  rents,  controlled  for   amenities,  appear  to  exhibit  a  wider  degree  of  variation  in  rents  than  typical  rental  units,   suggesting  the  importance  to  secondary  unit  tenants  of  engaging  in  extensive  searches  to   find  an  optimal  housing  package  when  navigating  this  idiosyncratic  market  if  they  do  not   happen  to  have  a  personal  relationship  with  a  secondary  unit  landlord  in  a  desired  location.   Indeed,  our  results  show  that  the  most  common  methods  by  which  secondary  unit  landlords   find  tenants  are  by  already  knowing  the  person,  Craigslist,  and  personal  referrals  (in   descending  order  of  frequency).  Finally,  we  observe  that  26%  of  secondary  units  are   occupied  by  the  homeowning  household  itself,  while  an  additional  29%  are  occupied  by   friends,  family  or  acquaintances  of  the  homeowner  household,  some  of  whom  are  living   rent-­‐free  and  others  of  whom  are  likely  paying  reduced  rents.   • Secondary  units  are  much  less  likely  to  offer  off-­‐street  parking  than  standard  rental  units.   Indeed,  single-­‐family  residential  (SFR)  properties  with  secondary  units  do  not  appear  to   have  significantly  more  off-­‐street  parking  than  those  that  lack  secondary  units.  This   suggests  that,  regardless  of  off-­‐street  parking  requirements  imposed  by  cities,  secondary   units  are  often  offered  to  tenants  who  either  do  not  have  a  car  or  else  park  their  car(s)  on   the  street.  Furthermore,  in  the  relatively  infrequent  cases  in  which  secondary  units  do  offer   off-­‐street  parking,  landlords  very  seldom  charge  for  this  amenity.  This  implies  that  off-­‐ street  parking  requirements,  if  followed,  would  often  impose  a  cost  on  secondary  unit   landlords  that  they  would  not  or  could  not  pass  on  to  tenants.     • While  most  homeowners  are  aware  of  secondary  units  on  their  streets,  and  while  most  of  these   homeowners  do  not  perceive  there  to  be  a  negative  impact  from  secondary  units,  those  that  do   perceive  a  negative  impact  are  by  far  most  likely  to  cite  increased  pressure  on  on-­‐street   parking  as  the  reason.  Indeed,  our  results  suggest  that  SFR  properties  with  one  secondary   unit  generate,  on  average,  0.9  more  total  cars  and  consume  0.7  more  on-­‐street  parking   spaces  than  SFR  properties  lacking  secondary  units.  At  the  same  time,  SFR  properties  with   secondary  units  exhibit  wider  variability  in  car  generation  than  those  that  do  not.     • Secondary  units  appear  to  be  much  likelier  than  other  types  of  rental  units  to  offer   substandard  cooking  facilities  (though  it  should  be  noted  that  the  vast  majority  do  appear  to   offer  full  kitchens).  This  is  likely  a  consequence  of  both  the  amateur  operation  of  secondary   units  and  the  widespread  existence  of  unpermitted  secondary  units.  Substandard  cooking   facilities  are  likely  associated  with  other  substandard  conditions,  although  this  cannot  be   directly  gleaned  from  the  data  presented  herein.   • While  we  found  no  overwhelming  demographic  differences  between  households  that  have   secondary  units  and  those  that  do  not,  the  former  have  a  somewhat  smaller  number  of   people,  slightly  lower  incomes,  and  somewhat  fewer  white  adults.  Adults  living  in   secondary  units  are  considerably  younger  than  adults  in  homeowner  households,  though   relatively  few  are  under  25,  and  very  few  are  elderly.  There  are  few  children  residing  in   secondary  units.   • Secondary  units  are  much  more  likely  to  share  utility  costs  with  other  units  or  to  have  them   included  in  the  rent.  This  is  likely  associated  with  widespread  non-­‐professional  management   as  well  as  a  dearth  of  fully  independent  gas,  electric,  water  and  sewer  connections  to  utility   main  lines  for  secondary  units.     • The  potential  market  for  secondary  units  amongst  homeowners  that  do  not  already  have  them   is  about  30%  of  the  total  universe  of  such  homeowners.   HOMEOWNER  SURVEY   The  homeowner  survey  was  intended  to  sample  the  existing  population  of  homeowners  living   within  the  study  corridor  in  order  to  ascertain  i)  the  prevalence  rate  of  secondary  units  for  SFR   properties  within  the  corridor;  ii)  the  attitudes  towards  secondary  units  of  homeowners  residing   on  properties  that  lack  them;  iii)  physical  characteristics  of  extant  secondary  units;  iv)  demographic   characteristics  of  homeowners  (with  and  without  secondary  units)  and  secondary  unit  occupants;   v)  details  concerning  the  economic  relationship  between  secondary  unit  homeowners  and  their   tenants;  and  vi)  the  parking  habits  of  homeowner  households  (with  and  without  secondary  units)   and  secondary  unit  occupant  households.  Because  none  of  the  cities  lying  within  the  study  corridor   had  reliable  records  of  secondary  units  within  their  boundaries  –  whether  permitted  or  otherwise  –   it  was  necessary  to  survey  a  sample  of  randomly-­‐selected  SFR  properties  in  order  to  find  ones  with   secondary  units,  even  though  (as  expected)  a  large  majority  of  such  properties  lacked  secondary   units.  However,  receiving  survey  data  from  a  large  number  of  SFR  properties  lacking  secondary   units,  in  addition  to  making  it  possible  to  derive  a  secondary  unit  prevalence  estimate,  also  made  it   possible  to  contrast  non-­‐secondary  unit  homeowner  households  with  households  that  have  such   units,  and  to  gauge  attitudes  towards  secondary  units  amongst  the  households  that  lack  them.   The  homeowner  survey  instrument  (Appendix  1),  in  addition  to  asking  respondent  homeowners  to   answer  specific  questions,  also  allowed  them  to  express  any  opinions  they  chose  to  share  with  the   researchers  about  secondary  units,  whether  or  not  they  had  any  on  their  own  properties.  Some  of   the  resulting  responses  (edited  for  spelling  and  length)  are  quoted  throughout  this  Working  Paper.   While  each  of  these  quotes  only  represents  the  viewpoint  of  one  particular  homeowner  that   responded  to  the  survey,  it  nevertheless  offers  a  glimpse  into  the  perspective  of  one  among  the   individuals  whose  responses  to  specific  questions  are  quantified  and  reported  in  this  document.         Design  of  homeowner  survey   Selection  of  sample:  The  properties  selected  for  inclusion  in  the  survey  sample  were  restricted  to   those  lying  within  the  half-­‐mile  station  areas  described  in  Yes,  but  will  they  let  us  build?  The   feasibility  of  secondary  units  in  the  east  bay  (IURD  WP-­‐2012-­‐02).  Only  properties  classified  as  SFR   by  the  Alameda  and  Contra  Costa  tax  assessors  were  included  in  the  pool  of  properties  that  could  be   selected.1  Properties  were  then  selected  for  inclusion  in  the  sample  on  the  basis  of  random   numerical  rankings.  We  dropped  properties  that  could  be  readily  seen  to  be  non-­‐owner  occupied   (because  of  either  a  discrepancy  between  the  property  address  and  the  mailing  address  for  the   owner,  or  because  the  owner  was  listed  as  a  firm  or  trust  rather  than  a  person)  from  the  sample.2   This  was  because  i)  we  presumed  that  we  were  much  less  likely  to  receive  a  response  from  non-­‐ owner  occupied  properties  and  ii)  in  this  study  we  are  primarily  concerned  with  the  actions  of   owner-­‐occupants  and  their  decisions  regarding  whether  or  not  to  install  secondary  units  on  their   properties.  Finally,  properties  were  selected  in  the  order  of  the  random  rankings  until  a  sample  of   the  desired  size  had  been  assembled.   Survey  modes:  We  deployed  a  mixed-­‐mode  approach  that  used  three  methodologies  for  contacting   and  gathering  survey  responses  from  homeowners  whose  properties  had  been  selected  to  be  in  the   sample.  The  methodologies  were  as  follows:   1) Every  homeowner  in  the  sample  selected  from  parcel  data  received  a  postcard  (as  shown  in   Appendix  2),  sent  via  U.S.  Mail,  with  a  very  brief  description  of  this  study,  an  invitation  to   participate,  and  a  web  link  to  an  online  survey  instrument.  Every  postcard  was  printed  in   English  on  one  side  and  in  Spanish  on  the  other  side  (with  the  web  link  on  the  Spanish  side   pointing  to  a  Spanish-­‐language  version  of  the  survey  instrument).  In  rough  concordance   with  the  Dillman  method,3  all  homeowners  who  had  not  replied  to  the  first  postcard  within   one  week  of  its  mailing  were  mailed  a  second  postcard.  Some  homeowners  who  had  not                                                                                                                             1  Discussions  with  staff  from  the  various  cities  revealed  that  city  and  county  tax  assessor  records  of  current   land  use  for  residential  (and  other  properties)  can  sometimes  conflict.  We  relied  on  the  county  tax  assessor   records  for  building  the  samples,  since  these  were  the  comprehensive  data  sets  to  which  we  had  access.  It  is   also  worth  mentioning  that  the  classification  of  a  property  as  SFR  in  no  way  precluded  it  from  having  a   secondary  unit,  as  the  survey  results  presented  in  this  Working  Paper  indicate;  indeed,  it  is  likely  that  a   majority  of  secondary  units  are  not  included  in  the  assessments  of  their  host  properties  for  the  purposes  of   property  taxation.  The  implications  of  this  are  discussed  further  in  Scaling  up  secondary  unit  production  in  the   East  Bay:  Impacts  and  policy  implications  (WP-­‐2012-­‐5).             2  Because  California  estate  laws  encourage  homeowners  to  place  their  property  in  a  trust,  the  assessor’s   databases  include  many  resident  homeowners  listed  as  trusts.  We  conservatively  exclude  these  properties   because  of  the  possibility  that  the  homeowner  is  non-­‐resident.  However,  this  could  introduce  a  bias  by  also   excluding  some  resident  homeowners,  who  are  likely  to  be  longer-­‐term  and  more  educated  residents.   3  See  Dillman,  Don,  Jolene  Smyth  and  Leah  Christian.  2009.  Internet,  Mail,  and  Mixed-­‐Mode  Surveys:  the   Tailored  Design  Method.  New  York:  Wiley  &  Sons.   replied  to  the  survey  one  week  after  being  mailed  the  second  postcard  were  mailed  a  third   postcard.4       2) Each  postcard  gave  a  phone  number  where  a  homeowner  could  leave  a  voice  mail  message   (whether  in  English  or  Spanish)  requesting  that  a  paper  version  of  the  survey  (in  the  caller’s   chosen  language),  along  with  a  pre-­‐addressed  and  pre-­‐stamped  return  envelope,  be  mailed   to  her  home.  We  included  this  option  in  order  to  be  sure  that  potential  survey  respondents   lacking  the  knowledge  or  the  means  to  access  the  Internet  would  have  a  way  to  respond  to   the  survey  other  than  via  the  online  version.   3) We  contacted  the  moderators  of  listservs  for  ten  organizations  representing  the  interests  of   neighborhoods  surrounding  the  station  areas  in  Oakland,  Berkeley,  and  El  Cerrito,  and   asked  them  to  send  an  e-­‐mail  to  their  lists  requesting  that  single  family  house  owner-­‐ occupants  respond  to  the  online  version  of  the  survey  (as  in  #1  above)  or  telephone  to   request  a  hard  copy  version  (as  in  #2).  This  e-­‐mail  was  written  in  both  English  and  Spanish,   and  largely  mirrored  the  wording  of  the  postcard.  As  a  result  of  this  effort,  six  of  the   organizations  contacted  their  members  requesting  that  they  respond  to  the  survey.  In   addition,  an  online  newspaper  in  El  Cerrito  published  a  story  urging  those  of  its  readers   eligible  to  respond  to  the  survey  to  do  so.   As  an  inducement  for  their  participation,  survey  respondents  were  offered  the  opportunity  to  enter   into  a  raffle  drawing  for  three  prizes:  a  $200  gift  certificate  for  a  popular  local  electronics  store,  a   $150  BART  pass,  and  a  $100  BART  pass.   Sample  size,  response  rates  and  bias:  We  mailed  postcards  to  2,529  households.  A  first   subsample  of  1,551  households  was  split  almost  exactly  evenly  amongst  the  five  station  areas.   Following  lower  level  of  responses  from  the  Del  Norte  and  Plaza  station  areas,  we  decided  to  have   each  of  them  comprise  30%  of  the  properties  in  the  second  subsample  of  1,001  households,  with   the  other  40%  split  equally  amongst  the  three  other  station  areas.  We  received  and  acted  upon  31   separate  intelligible  requests,  received  via  telephone  voice  messages,  to  mail  written  versions  of  the   survey  instrument  (all  of  them  in  English)  to  potential  respondents,  18  of  whom  ended  up  filling   out  and  returning  the  survey.   We  used  the  minimum  sample  size  formula  nmin  =  p(1-­‐p)(Zc/E)2  to  calculate  our  targets  for  the   desired  number  of  survey  responses.  For  an  estimate  of  the  prevalence  of  secondary  units  amongst   owner-­‐occupied  SFR  properties  in  the  study  area,  assuming  that  p  =  0.2  (corresponding  to  the   anticipated  probability  of  a  given  SFR  having  a  secondary  unit  being  20%),  Zc  =  1.645  (the  Z  score   corresponding  to  a  10%  confidence  interval,  typically  used  in  planning  studies)  and  E  =  10%  (a   sampling  error  of  10%,  also  commonly  used  in  planning  studies)  yields  a  minimum  sample  size  of   43.3.  Therefore,  a  defensible  secondary  unit  prevalence  estimate  would  require  44  survey   responses  (i.e.  43.3  rounded  upwards  to  the  nearest  whole  number).                                                                                                                             4  As  described  further  in  the  following  section,  the  sample  was  broken  into  two  groups.  Non-­‐respondents  in   the  first,  larger  sub-­‐sample  received  three  postcards,  while  non-­‐respondents  in  the  second,  smaller  sub-­‐ sample  received  only  two.  We  decided  to  forego  sending  the  third  postcard  to  the  second  sub-­‐sample  after  we   were  satisfied  that  we  had  received  enough  responses  for  statistical  validity,  as  is  discussed  in  the  following   section.   However,  relying  upon  survey  results  to  draw  statistically  meaningful  inferences  about  the  nature   of  the  secondary  unit  population  requires  many  more  than  44  responses.  Using  the  most   conservative  value  of  p,  0.5,  yields  a  minimum  sample  size  of  68  secondary  units.  Because  most   responses  (perhaps  four  out  of  five,  depending  on  the  actual  secondary  unit  prevalence  rate)  would   be  from  properties  lacking  secondary  units,  we  would  need  on  the  order  of  350  survey  responses  to   be  able  to  make  supportable  generalizations  about  various  characteristics  of  the  secondary  units  in   our  study  area.   As  it  happened,  we  obtained  enough  survey  responses  to  meet  both  the  lower  threshold  for  the   secondary  unit  prevalence  rate  and  the  higher  threshold  for  secondary  unit  characteristics.  We   received  a  grand  total  of  515  survey  responses.  Of  these,  334  responses  originated  from  the   postcard  mailings  to  2,529  households,  corresponding  to  a  response  rate  of  13.2%.  These  334   responses  easily  exceed  the  threshold  of  44  survey  responses  needed  for  a  secondary  unit   prevalence  ratio  amongst  owner-­‐occupied  properties  in  the  study  area.5  Of  the  334  responses   originating  from  the  postcards,  316  were  online  and  18  were  mailed.  The  remainder  of  the  515   survey  responses,  181,  originated  as  responses  to  the  e-­‐mails  sent  out  to  neighborhood  group   listservs.  None  of  the  515  total  survey  responses  were  in  Spanish;  all  were  in  English.   Of  the  515  responses,  87  were  reported  by  respondents  to  be  from  properties  with  secondary  units.   Using  a  stricter  definition  for  secondary  unit  that  we  employed  (discussed  below),  we  determined   only  81  of  the  87  responses  to  actually  be  from  properties  with  secondary  units.  Still,  81  survey   responses  from  properties  with  secondary  units  comfortably  exceeds  the  minimum  threshold  of  68   responses  needed  for  statistically  defensible  generalizations  about  secondary  unit  characteristics,   as  discussed  two  paragraphs  above.   Leaving  aside  the  thresholds  for  statistical  significance,  we  are  still  left  with  the  question  of   whether  the  response  set  that  we  obtained  is  generally  representative  of  the  population  of  SFR   properties  in  the  study  region.  From  a  geographical  standpoint,  at  least,  the  responses  are   reasonably  well-­‐balanced,  with  El  Cerrito  somewhat  over-­‐represented  and  Richmond  significantly   under-­‐represented  in  the  response  set  as  compared  to  the  cities’  share  of  station-­‐area  SFR  lots,  but   with  no  glaring  discrepancies  otherwise  observable  (Table  1).  As  discussed  in  the  results  section,   responding  homeowners  reported  being  from  households  that  were,  on  average,   disproportionately  high-­‐income,  well-­‐educated  and  comprised  of  white  non-­‐Hispanic  people.   Finally,  the  lack  of  a  single  Spanish  language  response  out  of  the  515  received  suggests  that   immigrant  homeowners,  at  least  Spanish-­‐speaking  ones,  are  not  well-­‐represented  in  the  sample.     In  summary,  while  the  responses  are  sufficiently  numerous  as  to  be  likely  to  be  broadly   representative  of  the  population  of  SFR  properties  within  the  station  areas,  the  foregoing  cautions   should  be  kept  in  mind  when  interpreting  the  results.                                                                                                                               5  We  consider  responses  sourced  from  postcards,  rather  than  the  entire  set  of  responses,  to  be  the   appropriate  data  set  to  use  for  calculating  the  secondary  unit  prevalence  estimate,  since  the  postcards  were   mailed  to  a  truly  random  sample  of  owner-­‐occupied  SFR  properties  lying  within  the  station  areas  within  the   study  corridor.  For  all  other  estimates,  we  rely  on  the  full  response  set,  not  just  the  portion  that  originated   from  postcards.   Table  1.  Geographic  distribution  of  survey  responses.     Homeowner  survey  results   While  some  of  the  returned  surveys  included  non-­‐responses  to  certain  questions,  seven  out  of  the   515  returned  surveys  had  so  many  missing  responses  as  to  be  non-­‐usable.  In  the  end,  therefore,  we   used  508  out  of  the  515  returned  surveys  in  the  response  set.  The  following  sections  report  survey   results  according  to  various  themes.  Note  that  while  results  for  particular  questions  for  the  entire   data  set  are,  in  many  cases,  numerous  enough  to  be  statistically  significant,  results  for  individual   cities  in  most  cases  are  not,  and  are  therefore  not  presented  here.     Secondary  unit  prevalence.  As  discussed  in  the  previous  section,  we  considered  only  81  of  the  87   reported  secondary  units  to  meet  our  minimum  standard  for  actually  qualifying  as  secondary  units:   namely,  having  at  least  one  bathroom.6  Of  the  60  survey  responses  reporting  secondary  units  that   were  sourced  from  postcard  mailings  (as  opposed  to  solicitation  from  neighborhood  listservs),  55   of  the  secondary  units  were  reported  to  have  bathrooms.  We  therefore  regard  16%,  or  55   responses  indicating  secondary  units  sourced  from  postcards  units  out  of  a  total  of  334  responses   sourced  from  postcards,  as  our  best  estimate  of  the  proportion  of  SFR  lots  in  station  areas  within   the  study  corridor  that  have  secondary  units.   Of  the  81  properties  in  the  total  response  set  (sourced  both  from  postcards  and  from  neighborhood   listservs)  reporting  at  least  one  secondary  unit,  69  properties  reported  having  one  secondary  unit,   11  properties  reported  having  two  extra  units,  and  one  reported  having  three  extra  units.   Physical  characteristics  and  construction  history.  As  can  be  seen  in  Figure  1,  the  secondary   units  in  the  sample  set  come  in  a  variety  of  physical  configurations,  with  the  most  common  being,  in   descending  order  of  frequency,  a  freestanding  structure  in  the  rear  yard  (approximately  one  third   of  the  total),  a  converted  first  floor  or  basement,  and  a  converted  garage,  with  these  three   categories  collectively  accounting  for  more  than  two  thirds  of  all  85  cases.7                                                                                                                             6  Secondary  units  lacking  full  kitchen  facilities,  including  a  multi-­‐burner  stove  and  an  oven,  can  potentially   overcome  these  deficiencies  for  the  purpose  of  functioning  as  fully  independent  units  by  relying  on  hot  plates   and/or  microwave  ovens.  By  contrast,  living  quarters  that  lack  indoor  plumbing  cannot  be  viewed  as  meeting   even  a  minimal  standard  for  an  independent  dwelling  unit.   7  Some  of  the  reported  results  have  a  number  of  responses  that  exceeds  81,  the  number  of  properties  in  the   overall  data  set  with  secondary  units  that  have  bathrooms.  This  is  because  some  properties  have  more  than   City  Postcards  mailed  (%)  Survey  responses  (%)  Station  area  SFR  lots  (%)   Albany  4.5  2.5  3.5   Berkeley  30.2  42.3  46.8   El  Cerrito  34.4  31.7  21.1   Richmond  9.0  1.6  8.6   Oakland  21.9  21.9  20.0   Total  100  100  100   Secondary  unit  homeowners  were  asked  whether  they  or  a  previous  owner  had  instigated  the   installation  of  their  unit:  23%  said  that  they  had  done  it,  while  the  remaining  77%  said  that  a   previous  homeowner  had  undertaken  the  work  (out  of  77  responses).  All  18  homeowners  that   reported  having  overseen  the  installation  of  their  secondary  unit  were  asked,  and  answered,  the   question  of  who  had  actually  done  the  physical  work  of  installing  the  unit.  The  most  common   responses  were  a  hired  contractor  (44%),  the  respondent’s  household  (22%),  and  a  combination  of   the  two  (22%).  A  friend  or  relative  had  done  the  work  in  one  case,  and  the  other  two  responses   were  categorized  as  “other.”   All  secondary  unit  homeowners  were  asked  if  their  unit  has  a  complete  kitchen  (defined  as  a   rangetop  stove,  oven  and  refrigerator):  88%  of  the  74  respondents  answered  that  they  did,  while   the  remaining  12%  answered  “no.”   Respondents  were  asked  the  number  of  bedrooms  in  their  secondary  units.  As  might  be  expected,  a   majority,  63%,  of  secondary  units  were  reported  to  be  either  studios  or  one-­‐bedrooms,  but  there   were  also  significant  numbers  of  units  that  have  two  bedrooms  or  even  three  bedrooms  or  more   (Figure  2).     Figure  1.  The  physical  configuration  of  reported  secondary  units.                                                                                                                                                                                                                                                                                                                                                                                                                   one  secondary  unit,  each  of  which  is  counted  separately  unless  otherwise  stated.  Thus,  the  data  in  Figure  2  is   derived  from  a  response  set  of  84  secondary  units,  but  these  units  are  associated  with  a  subset  of  the  81   properties  that  contain  at  least  one  secondary  unit.     0%  10%  20%  30%  40%   Apt  over  garage   A1ached  apt  behind  main  house   Detached  apt  behind  main  house   Freestanding  garage  w/  apt   Converted  garage   Built  as  mulBunit  structure   Converted  basement/1st  floor   Converted  aDc   Converted  rooms  in  main  house   Two  side-­‐by-­‐side  houses  on  one  lot   Other       Figure  2.  The  number  of  bedrooms  reported  for  the  secondary  units.         As  can  be  seen  in  Figure  3,  close  to  a  majority  of  homeowners  with  secondary  units  do  not  know   when  their  units  were  installed,  which  suggests  that  they  were  installed  at  least  prior  to  when  the   respondents  purchased  their  home.  Homeowners  representing  15  percent  of  the  84  units  for  which   responses  to  this  question  were  received  reported  that  their  secondary  units  had  been  installed   within  the  last  five  years.  For  secondary  units  with  particular  reported  ages  (i.e.  where  the   respondents  did  not  answer  “I  don’t  know”  to  this  question),  the  average  age  of  installation  for  the   unit  was  38  years.                 Studio   27%   1  BR   36%   2  BRs   27%   3+  BRs   10%   Figure  3.  The  reported  age  of  responding  homeowners’  secondary  units.   Details  of  secondary  use,  occupancy,  and  rental.  The  majority  (85%  of  84  units)  of  secondary   units  are  reported  to  be  occupied  by  at  least  one  person.  Twelve  of  the  13  unoccupied  secondary   units  (92%)  are  reported  to  be  used  for  a  non-­‐residential  purpose,  such  as  a  home  office,  workshop   or  studio  space,  while  the  owner  of  the  remaining  unit  is  seeking  a  tenant.  Of  the  71  secondary  units   that  are  occupied,  the  relationship  of  the  occupant(s)  to  the  homeowner  household  was  reported   for  70  of  them.  The  largest  number  (46%)  of  secondary  unit  occupant  households  consist  of  people   that  the  homeowner  did  not  previously  know,  26%  of  the  units  are  occupied  by  all  or  part  of  the   homeowner  household,  and  29%  are  occupied  by  relatives,  friends,  or  acquaintances  of  the   homeowner  household.   By  far  the  most  common  methods  that  homeowners  used  to  find  occupants,  besides  having  a  part   or  all  of  their  own  household  occupying  the  unit  (26%  of  69  responses),  are  by  already  knowing  the   occupants  (25%)8,  Craigslist  (25%),  and  referrals  from  a  known  person  (7%).  Other  methods   include  the  tenants  already  being  in  place  at  the  time  the  property  was  purchased  (4%),  the   University  of  California-­‐Berkeley  Housing  Office  (3%),  and  a  combination  of  Craigslist  and  the  UC-­‐ Berkeley  Housing  Office  (6%).  In  4%  of  the  responses,  the  secondary  unit  is  separately  owned  by  its   occupant  as  a  condominium  or  tenancy-­‐in-­‐common  (TIC).                                                                                                                             8  There  is  an  inconsistency  between  the  reported  figures  of  29%  of  units  being  occupied  by  friends,  family  or   acquaintances  that  are  not  part  of  the  homeowner  household,  and  25%  of  unit  occupant  households  having   been  sourced  via  the  homeowner  household  having  previously  known  the  occupants.  This  discrepancy  arises   because  there  are  two  cases  in  which  the  respondent  reports  that  the  unit  is  occupied  by  one  or  more   acquaintances,  but  that  the  tenants  were  sourced  via  a  personal  referral.  In  these  cases,  presumably,  the   tenants  have  become  acquaintances  after  moving  in,  or  else  the  homeowner  knew  them  by  reputation   beforehand  but  had  not  actually  met  them.  Thus,  in  short,  the  inconsistency  between  the  two  figures  arises   from  an  ambiguity  in  the  precise  nature  of  the  term  “acquaintance,”  which  was  not  rigorously  defined  in  the   survey  instrument.       0%   5%   10%   15%   20%   25%   30%   35%   40%   45%   50%   As  can  be  seen  in  Table  2,  34  out  of  41  respondents  reported  that  the  occupants  of  their  secondary   units  pay  rent,  while  the  remaining  seven  stay  for  free  or  perform  in-­‐kind  work  in  exchange  for   living  in  the  secondary  unit.  Rents  charged  vary  widely.  (Please  refer  to  the  discussion  of  the  results   from  a  separate  analytical  technique,  the  rental  advertisement  data  study  later  in  this  document,  for   a  much  more  detailed  examination  of  market  rents  in  secondary  dwelling  units.)  Of  the  tenants   paying  rent,  75%  are  people  that  the  homeowner  did  not  previously  know,  while  the  rest  are   friends,  acquaintances,  or  relatives.  All  seven  of  the  tenants  staying  for  free  or  in  exchange  for  in-­‐ kind  rent  are,  not  surprisingly,  friends  or  relatives  of  the  homeowner.     Table  2.  Rent  paid  by  unit  size  (number  of  bedrooms).    Studio  1  BR  2  BRs  3+  BRs  Total   Free  or  in-­‐kind  rent  2  3  1  1  7   Paid  rent  7  12  10  5  34   Minimum  paid  rent  ($)  650    550    650    1,250      -­‐-­‐       Maximum  paid  rent($)  1,750    1,500    2,975    2,200      -­‐-­‐       Average  paid  rent($)  948    931    1,486    1,483      -­‐-­‐         Demographics  of  homeowner  households.  As  is  evident  in  Table  3,  households  with  or  without   secondary  units  differ  little  demographically,  though  they  are  quite  different  from  residents  of  the   five  East  Bay  cities  as  a  whole  (including  renter  households).  The  results  indicate  that,  as  might  be   expected,  households  with  secondary  units  have  somewhat  smaller  household  sizes  (both  with   respect  to  adults  and  children)  than  non-­‐secondary  unit-­‐owning  homeowner  households.  In   addition,  adults  in  secondary  unit-­‐owning  households  are  slightly  older,  earn  slightly  less,  are   slightly  more  likely  to  be  college  educated,  and  are  somewhat  less  likely  to  be  non-­‐Hispanic  whites   than  their  non-­‐secondary  unit-­‐owning  peers.  The  differences,  however,  are  quite  small.  Secondary   unit-­‐owning  households  are  notable  more  than  anything  for  their  broad  demographic  similarity  to   their  non-­‐secondary  unit-­‐owning  peers.   Note  that  these  results  indicate,  as  mentioned  in  an  earlier  section,  that  the  respondent  households   are,  in  general,  highly  affluent,  comprised  of  highly  educated  adults,  and  comprised  of  relatively  few   adults  of  color.  While  the  income  and  education  results  can  be  regarded  with  some  skepticism,  due   to  a  possible  tendency  for  people  to  exaggerate  their  income  and  education  status,  we  should   nevertheless  bear  in  mind  that  the  survey  results  herein  appear  to  be  skewed  towards  a  well-­‐off   group  of  homeowners,  and  calibrate  our  interpretation  of  the  results  accordingly.               Table  3.  Demographics  of  homeowner  households  with  and  without  secondary  units,   relative  to  East  Bay  study  area.      Households    East  Bay*  With  Secondary  Units  Without  Secondary  Units   Demographic  characteristic    Average  Response   set  size   Average  Response   set  size  Adults  in  household  1.77     1.73  77  HHs  1.95  404  HHs   Children  in  household  0.71     0.37  68  HHs  0.43  405  HHs   Avg.  age  of  adults  in  household  35.4  49.6  117  adults  48.9  771  adults   Household  income  ($)  $77,775  $105,000    61  HHs  $109,000    382  HHs   White  non-­‐Latino  adults  (%)  34.2  70.8  120  adults  79.7  784  adults   African  American  adults  (%)  22.9  4.2  120  adults  5.2  784  adults   Latino  adults  (%)  20.7  6.7  120  adults  4.2  784  adults   Asian/Pacific  Islander  adults  (%)  18.8  12.5  120  adults  8.0  784  adults   Adults  with  high  school  or  less  (%)  36.7  5.8  120  adults  4.7  783  adults   College  graduates  (%)  37.1  82.5  120  adults  78.4  783  adults   Male  adults  (%)  47.9  42.5  120  adults  46.2  783  adults   *  Includes  data  from  Albany,  Berkeley,  El  Cerrito,  Oakland,  and  Richmond  from  either  Census  2010  or  the  2005-­‐2009   American  Community  Survey.    Income  is  adjusted  to  2010  dollars.     Demographics  of  secondary  unit  occupant  households.  Survey  respondents  reported  that  73   adults  were  living  in  48  secondary  units,  for  an  average  of  1.52  adults  per  secondary  unit.9  The   average  age  for  the  68  adults  residing  in  the  secondary  units  with  reported  ages  is  38.6  years,  with   17%  of  the  adults  ranging  in  age  from  18  to  25  and  another  4%  with  ages  65  and  greater.  One  or   more  children  were  reported  to  be  living  in  only  four  out  of  the  50  secondary  units  (8%)  for  which   this  question  was  answered.  The  racial/ethnic  breakdown  of  the  62  adult  secondary  unit  occupants   for  whom  this  information  was  provided  is  as  follows:  80%  white  non-­‐Latino,  8%  Latino,  8%   Asian/Pacific  Islander,  and  5%  mixed  race.  Of  the  63  adult  secondary  unit  occupants  with  reported   gender,  37%  of  them  are  male.   Perceptions  of  secondary  units  by  non-­‐secondary  unit  homeowners.  The  421  survey   respondents  that  reported  not  having  at  least  one  secondary  unit  were  asked  a  series  of  questions   regarding  their  attitudes  towards  secondary  units.  One  question,  which  drew  404  responses,  asked   these  homeowners  why  they  do  not  already  have  a  secondary  unit.  As  can  be  seen  in  Figure  4,  the   potential  market  for  future  secondary  unit  installations  –  that  subset  of  homeowners  that  either  has   already  attempted,  is  actively  planning  to  or  might  consider  installing  a  secondary  unit  –  is  about   30%  of  homeowners  that  do  not  already  have  such  an  extra  unit  on  their  properties.                                                                                                                             9  Note  that  homeowner  households  were  asked  to  answer  questions  about  the  occupants,  if  any,  of  the   secondary  units  on  their  properties,  if  any.  While  we  considered  attempting  to  ask  homeowners  to  pass  along   a  separate  survey  to  the  occupants  of  their  secondary  units  for  them  to  fill  out  directly,  we  concluded  that  this   was  unlikely  to  result  in  a  high  response  rate,  as  well  as  being  cost-­‐prohibitive.  Consequently,  figures  on   secondary  unit  occupants  should  be  interpreted  with  the  caution  that  they  originate  from  information   provided  by  a  member  of  the  homeowner  household  rather  than  directly  from  the  occupants  themselves.  Also   note  that  figures  for  secondary  unit  occupant  demographics  exclude  instances  in  which  the  entire   homeowner  household  occupies  the  secondary  unit  –  those  are  included  in  the  previous  section  summarizing   homeowner  household  demographics.   Figure  4.  Stated  reasons  that  homeowners  lacking  a  secondary  unit  do  not  already  have  one.     Homeowners  that  reported  having  tried  but  failed  to  install  secondary  units  were  asked  for  the   reason  for  the  failure.  Their  responses  can  be  seen  in  Figure  5.  The  leading  reason  is  an  inability  to   fit  the  required  amount  of  off-­‐street  parking,  but  other  prominent  answers  are  excessive  cost  and   difficulty  of  the  city’s  regulatory  process,  among  others.  One  Berkeley  homeowner  exemplified  all   three  of  these  leading  reasons  for  not  having  installed  a  secondary  unit:  “To  add  a  unit,  one  must   have  noncontiguous  parking  spaces.  I  would  have  to  put  a  car  in  my  front  yard,  which  is  probably   illegal,  or  give  up  the  back  yard.  I  have  an  old  garage/studio  I  would  love  to  turn  into  an  additional   unit  but  [cost]  and  permits  and  parking  make  it  difficult  [or]  impossible.”   Homeowners  lacking  secondary  units  on  their  properties  were  asked  if  they  knew  of  secondary   units  on  other  properties  on  their  block.  Out  of  407  responses,  61.9%  were  in  the  affirmative,  8.6%   of  homeowners  stated  that  there  were  no  secondary  units  on  their  street,  and  29.5%  were  not  sure.   Homeowners  indicating  that  they  knew  of  secondary  units  on  their  street  were  then  asked  whether   or  not  they  felt  that  these  units  had  a  negative  impact  on  their  quality  of  life,  and  if  so,  what  the   leading  negative  impact  was.  Of  the  255  responses  to  this  question,  by  far  the  most  common   (62.0%)  was  that  the  nearby  secondary  units  have  no  negative  impact  on  the  respondent’s  quality   of  life.  The  leading  negative  impact  reported  (24.7%)  was  an  increased  number  of  cars  competing   for  on-­‐street  parking.  In  the  words  of  an  El  Cerrito  homeowner:  “Neighbors  who  have  built  extra   units  and  rented  their  property  have  created  tensions  and  problems  because  now  there  are  six  cars   vying  for  one  off-­‐street  shared  parking  spot.”       Don't  want   one   38%   Never   considered   it   31%   Tried  but   failed   7%   Might  want   to   21%   Planning  on   it   3%       Figure  5.  Reasons  for  failure  of  attempted  secondary  unit  installations.   Other  negative  impacts  were  reported  in  much  smaller  numbers,  including  the  presence  of  renters   on  a  street  mostly  occupied  by  homeowner  households  (5.1%),  density-­‐related  impacts  (3.1%),   unruly  tenants  (2.0%),  aesthetic  impacts  (1.2%),  and  reduced  property  values  (0.8%).  (Figure  6.)   An  Oakland  homeowner  disturbed  by  the  prevalence  of  secondary  units  on  the  local  block   expressed  some  of  the  concerns  about  increased  density  that  arose  with  a  minority  of  homeowners:   “The  extra  units,  which  are  rentals,  make  it  difficult  to  know  all  the  neighbors.  People  seem  to  come   and  go.  It  makes  the  block  seem  less  friendly  [than]  when  there  was  one  family  per  property.  I  used   to  know  all  my  neighbors,  now  I  know  only  a  few.  I've  also  seen  a  big  increase  in  the  number  of  cars   on  my  block  since  these  additional  units  were  added.  With  all  the  additional  cars,  if  I  didn't  have  my   one  parking  spot  on  my  property,  I  would  often  have  to  park  a  block  or  so  away.  That  would  make   me  think  twice  about  going  out  at  night.  It  also  makes  it  difficult  when  friends  visit.”                         0  2  4  6  8  10  12   City  process   DisrupBon   Too  expensive   Couldn't  fit  parking   Lot  too  small   Other   Figure  6.  Perceived  impact  of  secondary  units  on  quality  of  life.     Parking.  All  homeowner  households  were  asked  to  report  the  number  of  off-­‐street  parking  spaces   on  their  property  (whether  in  the  form  of  driveway  spaces,  spaces  in  a  garage,  or  others),  the   number  of  cars  their  household  parks  off-­‐street  on  their  property,  and  the  number  of  cars  their   household  parks  on-­‐street.  Households  reporting  having  secondary  units  with  occupants  were   asked  how  many  cars  those  occupants  park  off-­‐street  on  the  parcel  and  how  many  cars  they  park   on-­‐street.   While  the  majority  of  households  occupying  secondary  units  have  at  least  one  car,  22%  (out  of   responses  from  37  properties)  have  no  car  at  all  (Figure  7).                           0  50  100  150  200   No  negaBve  impact   Yes  -­‐-­‐  aestheBc  impacts   Yes  -­‐-­‐  presence  of  renters   Yes  -­‐-­‐  reduced  property  values   Yes  -­‐-­‐  unruly  tenants,  safety  and/or  noise   Yes  –  parking   Yes  -­‐-­‐  density-­‐related  impacts   Figure  7.  Number  of  cars  parked  by  households  in  secondary  units.     Note:  If  a  given  property  has  more  than  one  secondary  unit,  all  cars  reported  as  having  been  generated  by  those  units  are   included  in  the  total  for  that  property.         Meanwhile,  households  with  at  least  one  secondary  unit  provide  off-­‐street  parking  spaces  at  a  rate   that  is  statistically  indistinguishable  from  houses  lacking  secondary  units  (Table  4).       Table  4.  Average  number  of  parking  spaces  reported.    With  secondary  unit  Without  secondary  unit   mean  #  of  off-­‐street  spaces  1.86  1.67   standard  deviation  1.23  1.16   #  of  observations  66  406   significantly  different  @  95%  level?  No   Note:  Statistical  significance  for  the  difference  of  means  is  determined  via  a  standard  two-­‐tailed  T-­‐test  with  a  95%   significance  threshold.   Properties  with  secondary  units  generate  a  greater  number  of  cars  from  the  combined  homeowner   household  and  secondary  unit  occupant  household(s),  as  can  be  seen  in  Table  5.  Properties  with   secondary  units  generate,  on  average,  3.35  cars,  while  properties  without  them  generate  an  average   of  two  cars,  a  difference  of  almost  1.4  cars  in  the  average  case.  However,  when  properties  with  no   more  than  one  secondary  unit  are  excluded,  the  average  number  of  cars  generated  drops  to  2.91,  or   a  (statistically  significant)  increase  of  only  about  0.9  cars  above  properties  that  lack  secondary   units.10  As  seen  in  the  standard  deviation  figures,  properties  with  secondary  units  exhibit  more                                                                                                                             10  In  our  opinion,  properties  with  only  one  secondary  unit  are  the  relevant  analytic  category.  While  many   cities  might  be  expected  to  ease  the  regulatory  burdens  that  currently  prevent  SFR  properties  from  legally   adding  one  secondary  unit,  it  is  difficult  to  imagine  many  cases  in  which  cities  would  be  willing  to  consider   allowing  two  secondary  units  in  areas  zoned  for  single-­‐family  houses.  For  that  reason,  examining  data  sets   that  exclude  properties  with  more  than  one  secondary  unit  allows  us  to  gain  insight  into  the  parking  impacts   no  cars,  22%   1  car,  40%   2  cars,  19%   3  cars,  8%   3+  cars,  8%   5  cars,  3%   variability  than  those  without  them,  implying  that  a  wider  range  of  parking  behaviors  is  seen  on   such  properties  than  in  properties  that  lack  secondary  units  (something  that  is  presumably  related   to  the  large  variation  in  car  ownership  habits  of  secondary  unit  households  seen  previously  in   Figure  7).11       Table  5.  Number  of  cars  generated  by  units.    Secondary  Units   1  or  more  0  1   mean  #  of  total  cars  3.35  2.00  2.91   standard  deviation  1.92  1.05  1.60   #  of  observations  43*  405  35*   significantly  different  @  95%  level?  Yes       significantly  different  @  95%  level?      Yes   Note:  The  difference  of  means  test  is  done  in  the  same  way  described  in  Table  4.                   *8  were  imputed.  The  imputation  technique  is  described  in  footnote  #9.   Table  6  shows  the  number  of  cars  parked  on-­‐street  reported  to  be  generated  by  properties  with  and   without  secondary  units,  since  on-­‐street  parking  is  of  concern  to  some  neighbors  of  secondary  unit   properties,  as  evidenced  earlier  by  Figure  6.  Properties  with  secondary  units  generate  an  extra  1.1   cars  parked  on-­‐street,  on  average,  vis-­‐à-­‐vis  properties  without  secondary  units.  This  discrepancy,   however,  drops  to  only  0.7  cars  when  properties  with  more  than  one  secondary  unit  are  excluded   from  the  analysis.     While  interpreting  these  figures  in  tandem  with  the  results  shown  in  Table  6  should  be  done  with   caution,  particularly  since  we  did  not  impute  on-­‐street  car  generation  for  some  of  the  incomplete   records  as  we  did  with  the  number  of  total  cars,12  nevertheless  an  intuitive  picture  begins  to   emerge.  It  would  appear  that  people  residing  on  properties  with  secondary  units  “pick  up  more  of   the  slack”  of  their  available  off-­‐street  parking  spaces  by  putting  more  of  them  to  use  than  their   peers  living  on  properties  lacking  secondary  units.  This  might  explain  why  properties  with  one   secondary  unit  generate  0.9  more  cars  than  properties  without  secondary  units,  but  result  in  only                                                                                                                                                                                                                                                                                                                                                                                                           of  the  house-­‐plus-­‐secondary  unit  configuration  that  would  become  much  more  common  under  the  policy   outcome  that  cities  would  be  most  likely  to  seek.   11  Note  that  some  of  the  data  records  for  total  and  overspill  cars  used  for  Tables  5  and  7,  respectively,  are   imputed,  as  is  indicated  in  the  charts.  We  undertook  the  imputation  in  an  attempt  to  boost  the  sample  size  for   the  difference  of  means  computations,  since  many  of  the  records  for  secondary  unit  properties  lack  all  of  the   data  needed  to  compute  total  and  overspill  cars  generated.  Where  enough  data  was  available,  imputations   were  computed  by  selecting  a  random  value  from  an  assumed  normal  distribution  of  the  number  of  cars  per   person  calculated  from  the  set  of  non-­‐secondary  unit  homeowner  households,  and  using  this  to  impute  the   number  of  cars  generated  by  a  secondary  unit  households  from  the  number  of  adults  reported  to  comprise   the  household.       12  Intuitively,  one  would  expect  that  on-­‐street  parking  behavior  for  secondary  unit  occupants  might  differ   markedly  from  that  of  homeowner  households.  Thus  we  decided  that  it  would  not  be  helpful  to  attempt  to   impute  the  former  from  the  latter.     0.7  additional  cars  parked  on-­‐street,  despite  having  on  average  essentially  no  extra  off-­‐street   parking  available  (as  seen  in  Table  4).13     Table  6.  The  number  of  cars  parked  on-­‐street  generated  by  properties  with  and  without   secondary  units.    Number  of  Secondary  Units    1  or  more  0  1   mean  #  of  cars  parked  on-­‐street  1.97  0.90  1.61   standard  deviation  1.61  0.85  1.32   #  of  observations  35  406  27   significantly  different  @  95%  level?  Yes       significantly  different  @  95%  level?      Yes     If  all  of  “the  slack”  of  off-­‐street  parking  spaces  were  to  be  taken  up  (which  one  might  envision   occurring  under  conditions  in  which  all  available  on-­‐street  parking  spaces  are  in  use),  how  many   cars  would  households  living  on  properties  with  secondary  units  push  out  onto  the  street  as   compared  to  their  peers  living  on  parcels  without  such  units?  One  metric  that  can  shed  some  light   on  the  matter  is  “overspill  cars,”  which  we  define  to  be  the  difference  between  the  number  of  cars   generated  by  the  household(s)  residing  on  the  parcel  (as  summarized  in  Table  5)  and  the  number  of   off-­‐street  parking  spaces  available  on  the  property  (as  summarized  in  Table  4).  Intuitively,  a   measure  of  “overspill  cars”  is  equivalent  to  the  number  of  cars  that  are  pushed  into  on-­‐street   parking  once  all  available  off-­‐street  parking  spaces  on  the  parcel  have  been  used.  (A  negative  value   for  the  overspill  cars  metric  indicates  that  the  property  has  more  than  enough  off-­‐street  parking   spaces  for  the  cars  owned  by  the  household  or  households  living  on  the  parcel.)           Table  7  shows  the  average  number  of  overspill  cars  generated  by  properties  with  and  without   secondary  units.  Again,  while  the  discrepancy  between  the  two  categories  is  large  (1.1  cars)  and   statistically  significant,  it  drops  to  a  much  lower  level  (0.7  cars),  albeit  one  that  is  still  statistically   significant,  when  properties  with  more  than  one  secondary  unit  are  excluded.  Properties  with  only   one  secondary  unit  do,  however,  exhibit  an  appreciably  higher  standard  deviation  in  the  number  of   overspill  cars  that  they  generate  compared  to  properties  without  secondary  units  (2.23  versus   1.40).  This  suggests  the  need  for  flexible  policy  tools  that  can  span  the  wide  variety  of  car   ownership  propensities  exhibited  by  households  living  on  properties  with  secondary  units.                                                                                                                                 13  In  the  words  of  a  Berkeley  homeowner,  “No  one  around  here  parks  in  their  garage,  for  various  reasons  I   guess,  thus  the  garage  parking  spaces  are  only  available  in  ‘theory’.“  There  are  several  reasons  why  this  might   be  so.  For  instance,  an  off-­‐street  space  may  be  arranged  in  tandem  with  another  off-­‐street  space  that  is  used   frequently,  requiring  the  owner  of  one  of  the  cars  to  move  her  automobile  to  make  way  for  the  other.  A   driveway  or  garage  entrance  may  be  narrow  and  require  that  a  driver  back  into  it  slowly  and  carefully.  In   many  neighborhoods,  on-­‐street  parking  directly  in  front  of  the  main  house  is  frequently  available  at  least   some  of  the  time,  and  could  be  perceived  as  the  most  convenient  option.       Table  7.  Overspill  cars  per  unit.    Number  of  Secondary  Units    1  or  more  0  1   mean  #  of  cars  parked  on-­‐street  1.47  0.35  1.09   standard  deviation  2.32  1.40  2.23   #  of  observations  43*  406  35*   significantly  different  @  95%  level?  Yes       significantly  different  @  95%  level?      Yes   Note:  We  define  the  measure  of  “overspill  cars”  as  the  difference  between  the  number  of  cars  generated  by  a  property  (as   summarized  in  Table  5)  and  the  number  of  off-­‐street  parking  spaces  available  on  that  property  (as  summarized  in  Table   4).  Intuitively,  it  measures  the  parking  impact  of  a  parcel  on  its  street  and  (possibly  on  nearby  streets)  if  all  of  its  off-­‐street   parking  spaces  are  being  fully  used.                                    *8  were  imputed.  The  imputation  technique  is  described  in  footnote  #9.       RENTAL  INTERNET  ADVERTISEMENT  STUDY   Purpose  and  data  collection  methodology   While  the  survey  of  homeowners  described  above  provides  many  insights  into  the  characteristics,   behaviors  and  attitudes  of  homeowners,  the  survey  is  of  limited  usefulness  for  systematically   analyzing  the  open  rental  market  for  secondary  units,  and  the  ways  in  which  it  differs  from  the   corresponding  market  for  rental  dwellings  that  are  not  secondary  units.  The  survey  did  not  collect  a   sufficient  quantity  of  rental  market  data  on  secondary  units  to  make  it  possible  for  us  to  relate  these   units’  characteristics  to  the  level  of  rent  charged  via  a  hedonic  model  or  other  quantitative   technique.  In  addition,  the  questions  that  we  would  have  needed  to  add  to  the  survey  instrument  in   order  to  capture  some  of  the  information  important  to  a  hedonic  analysis  might  have  risked  making   the  survey  instrument  too  long,  thus  reducing  response  rates.     We  therefore  pursued  a  separate  analysis  in  which  we  collected  data  about  rental  units  –  both   secondary  units  and  other  types  of  dwellings  –  from  advertisements  placed  on  Craigslist.com,  a   website  widely  used  by  San  Francisco  Bay  Area  tenants  and  landlords.  Craigslist  ads  offered  several   advantages.  First,  we  could  assume  that  landlords  placing  advertisements  seeking  tenants  on   Craigslist  were  not  seeking  to  rent  to  people  they  already  knew,  and  thus  were  not  likely  planning   to  offer  “friends  and  family”  discounts.  By  contrast,  as  previously  described,  about  26%  of  the   secondary  units  in  the  survey  response  set  are  occupied  by  the  homeowner  household,  while  a   further  25%  are  occupied  by  at  least  one  person  known  to  the  homeowning  household  prior  to   move-­‐in,  whether  as  an  acquaintance,  friend,  or  relative.  Second,  we  could  readily  collect  data  on   secondary  and  non-­‐secondary  rental  units  simultaneously,  thus  making  it  easy  to  compare  the  two.   Finally,  landlords  advertising  on  Craigslist  presumably  have  a  direct  financial  incentive  (one  not   present  for  survey  respondents)  to  list  all  pertinent  features  of  the  rental  unit  that  they  are  offering   in  order  to  attract  prospective  tenants.       During  the  three-­‐month  period  of  May,  June,  and  July  of  2011,  we  collected  unit  characteristics  from   all  Craigslist  advertisements  for  secondary  units14  listed  for  rent  and  located  in  the  general   corridor15  under  study.  We  also  collected  information  from  a  similar  number  of  ads  for  non-­‐ secondary  rental  units.16  After  discarding  records  that  were  unusable  for  lack  of  essential   information  (such  as  rent  charged,  number  of  bedrooms,  and  number  of  bathrooms),  we  ended  up   with  a  sample  of  174  secondary  units  and  164  non-­‐secondary  units.     In  usable  ads,  amenities  such  as  off-­‐street  parking,  dishwashers  and  on-­‐site  laundry  were  assumed   to  not  be  in  place  unless  mentioned,  due  to  the  direct  incentive  for  landlords  to  trumpet  all   attractive  features  about  the  apartments  they  were  seeking  to  rent.  One  exception  to  this  general   rule  was  for  cooking  facilities,  which  were  assumed  to  be  fully  provided  (i.e.,  to  include  the  typical   minimal  configuration  of  a  four-­‐burner  range  top  stove  and  oven)  unless  otherwise  mentioned,  due   to  typical  market  expectations  concerning  kitchens.  Locational  characteristics  were  captured  via   three  dimensions:  a  measure  of  the  unit’s  walkability  (from  the  publicly  available  website   www.walkscore.com),  which  we  took  to  serve  as  a  proxy  for  the  general  level  of  neighborhood   amenity,  an  index  of  the  crime  rate  for  the  local  area  (from  the  proprietary  website   www.neighborhoodscout.com),  and  the  unit’s  proximity  to  a  freeway,  which  we  assumed  to   coincide  with  environmental  blight  due  to  noise,  fumes,  visual  impacts  and  so  forth.17  In  a  few  cases,   we  needed  to  impute  the  values  for  certain  variables  that  could  not  be  recorded  for  every  record  in   the  data  set.  Some  independent  variables  that  we  originally  intended  to  collect,  and  that  indeed   might  have  proved  to  be  useful,  such  as  the  number  of  units  in  the  apartment’s  building,  had  to  be   dropped  from  our  model,  because  they  could  not  be  inferred  or  plausibly  imputed  from  every  single   advertisement.  Details  about  all  of  the  variables  included  in  the  model  are  described  in  Appendix  3.                                                                                                                             14  A  unit  was  deemed  to  be  a  “secondary  unit”  for  these  purposes  if  this  characteristic  could  be  ascertained   from  the  advertisement,  such  as  a  picture  depicting  a  secondary  unit,  or  text  giving  some  tell-­‐tale  description   such  as  “a  cottage  behind  the  main  house.”  Note  that  for  these  purposes  the  analytic  distinction  between  a   “secondary  unit”  and  a  unit  in  a  duplex  or  triplex  is  bound  to  be  blurry.  The  inherent  characteristic  of  a   secondary  unit  of  interest  for  this  analysis  is  that  it  is  located  on  a  property  with  only  two  or  three  units,   possibly  (though  not  necessarily)  with  an  owner-­‐operator  living  on-­‐site  or  in  the  immediate  vicinity.   15  This  corridor  is  roughly  defined  as  that  portion  of  the  urban  East  Bay  that  extends  from  the  bay  shore  on   the  west  to  the  topographic  break  in  the  hills  to  the  east  (corresponding  more  or  less  to  the  “Flatlands,”  as   described  in  WP-­‐2012-­‐01),  from  approximately  one  mile  north  of  El  Cerrito  Del  Norte  BART  on  the  north  end   to  approximately  one  mile  south  of  MacArthur  BART  on  the  south  end.  While  this  region  generally  coincides   with  the  areas  considered  in  the  rest  of  this  study,  it  is  not  limited  to  only  the  BART  station  areas  within  this   corridor.   16  When  collecting  data  for  non-­‐secondary  units,  we  kept  gathering  these  advertisements  until  their  numbers   roughly  matched  those  of  the  secondary  unit  advertisements.  Once  the  latter  exceeded  the  former,  we   resumed  gathering  the  latter.  Thus,  while  the  non-­‐secondary  unit  advertisements  are  a  subset  of  the  entire   universe  of  ads  placed  for  units  lying  within  the  corridor,  one  can  assume  that  the  sample  was  randomly   drawn.   17  Because  of  the  high  density  of  freeway  coverage  in  the  study  area  and  the  consequent  ease  of  accessing  a   freeway  from  anywhere  within  it  (no  location  in  the  study  area  lies  more  than  three  miles  from  an  on-­‐ramp),   we  follow  the  lead  of  Cervero  and  Landis  in  treating  close  freeway  proximity  in  the  urban  East  Bay  as  a   disamenity  with  no  countervailing  proximity-­‐based  advantages  for  slightly  greater  distances.  See  Cervero,  R.   and  Landis,  J.  1997.  “Twenty  Years  of  the  Bay  Area  Rapid  Transit  System:  Land  Use  and  Development   Impacts.”  Transportation  Research  A,  Vol  31,  No.  4,  pp.  309-­‐333.   The  dependent  variable  used  for  the  hedonic  analysis  is  the  adjusted  monthly  rent.  To  facilitate  a   true  “apples  to  apples”  comparison,  the  rent  for  each  unit  was  adjusted  to  account  for  the   availability  or  lack  thereof  of  off-­‐street  parking  (whether  free  or  paid)  and  the  partial  or  total   inclusion  of  three  utility  costs  (gas/electric,  water/sewer,  and  cable/Internet)  with  monthly  rent.   While  four  of  the  secondary  units  offered  opportunities  for  tenants  to  offset  their  rent  through  on-­‐ site  work  (such  as  childcare,  yard  maintenance,  etc.),  this  type  of  in-­‐kind  work  was  considered  to  be   above  and  beyond  the  rental  transaction  (since  it  required  labor  to  be  performed)  and  thus  was  not   factored  into  adjusted  rent.  Note  that  for  the  full  data  set,  we  chose  to  not  normalize  rent  by   dwelling  unit  floor  area,  since  square  footages  were  only  directly  provided  in  24%  of  the  real  estate   ads  that  were  recorded  in  the  data  set  (the  rest  were  imputed).  (We  conducted  hedonic  model  runs   using  both  the  full  data  set  and  the  subset  that  includes  only  records  with  directly  observable   square  footages,  as  is  described  below  in  the  discussion  of  the  hedonic  model  results.)       With  the  variables  constructed  as  described  above,  we  were  able  to  conduct  two  types  of  analysis   on  the  data  set:  1)  a  series  of  difference-­‐of-­‐means  tests  among  the  various  variables,  in  order  to   ascertain  the  ways  in  which  secondary  unit  characteristics  differ  from  those  of  non-­‐secondary  units,   and  to  gather  summary  statistics  on  these  two  submarkets;  and  2)  a  hedonic  model,  which  allowed   us  to  compare  how  various  unit  amenities  and  locational  characteristics  are  differentially  priced   into  rents  in  the  secondary  unit  and  non-­‐secondary  unit  rental  markets.  The  results  of  these   analyses  are  treated  in  turn  in  the  following  two  sections.     Table  8  shows  the  distribution  of  dwellings  in  the  data  set  by  municipal  jurisdiction.  This   distribution  suggests  that  a  cautionary  note  be  kept  in  mind  when  interpreting  these  results:  60%   (secondary  units)  to  65%  (non-­‐secondary  units)  of  the  apartments  from  each  category  are  located   in  Berkeley  and  the  adjacent  unincorporated  community  of  Kensington.  By  comparison,  the   Berkeley  flatlands  and  Kensington  account  for  only  38%  of  the  SFR  parcels  of  the  study  area  (Table   8).  The  concentration  of  rental  units  in  and  around  Berkeley  may  reflect  the  heavy  influence  that   the  local  University  of  California  campus  exerts  on  the  rental  market  of  the  corridor  under  study,   and  perhaps  of  the  particular  ubiquity  of  Craigslist  as  a  means  of  locating  rental  housing   opportunities  among  the  highly  educated  population  of  Berkeley.  The  first  concern  is  perhaps   tempered  by  the  timing  of  the  data  collection,  which  occurred  in  the  late  spring  and  early  summer,  a   period  during  which  university  students  are  unlikely  to  seek  housing  (although  it  may  be  a  time   during  which  students  are  vacating  housing  and  leaving  the  region,  and  landlords  are  seeking  to  fill   the  resulting  vacancies).  As  can  be  seen,  Oakland  and  Emeryville  are  also  somewhat  over-­‐ represented  (though  less  so  than  Berkeley/Kensington).  El  Cerrito  and  Albany  are  mildly  under-­‐ represented,  while  Piedmont  and  the  relevant  portion  of  Richmond  are  highly  under-­‐represented.             Table  8.  Distribution  of  units  and  SFR  parcels  by  jurisdiction.   Location  of  non-­‐secondary  and  secondary  units  in  data  set  (left  and  central  columns,  respectively)  compared   against  proportion  of  total  study  region  SFR  parcels  (right  column).  Results  in  the  right  column  are  estimated   from  the  following  sources:  Center  for  Community  Innovation,  and  2009  Housing  Elements  for  the  cities  of   Albany,  Emeryville,  and  Piedmont  and  for  the  County  of  Contra  Costa  (which  contains  unincorporated   Kensington).     Table  9  gives  summary  statistics  for  secondary  and  non-­‐secondary  units  on  a  variety  of  variables,   along  with  the  results  of  a  difference-­‐of-­‐means  test  (T  test)  performed  on  each  variable.  The   statistics  reveal  a  number  of  intuitive,  yet  notable,  differences  between  rental  units  that  are   secondary  units  versus  rental  units  that  are  not,  suggesting  that  secondary  units  occupy  a   distinctive  niche  within  the  overall  rental  market.     Jurisdiction  Non-­‐secondary  units  (%)  Secondary  units  (%)  SFR  parcels  in  jurisdiction   (%)  Berkeley  flatlands  65  60  38   Oakland  flatlands  21  24  19   City  of  Albany  4  6  13   El  Cerrito  flatlands    8  9  13   S  Richmond  flatlands    2  1  7   City  of  Piedmont  <  1  <  1  12   Total  100%  100%  100%    non-­‐secondary  units*  secondary  units**  significantly  different***   physical  characteristics               #  of  bedrooms  1.52  0.99  yes   #  of  bathrooms  1.11  1.00  yes   floor  area  (sf)  703  559  yes   secure  entrance  (%)  12.8  2.9  yes   economic  characteristics             adjusted  monthly  gross  rent  ($)  1,358  1,106  yes   adjusted  monthly  gross  rent  per  ($/sf)  2.17  2.30  no   adjusted  monthly  gross  rent  as  %  of  AMI  68.4  62.4  yes   owner  on  site  (%)  2.4  52.9  yes   Parking             no  off-­‐street  parking  (%)  45.7  87.4  yes   free  off-­‐street  parking  (%)  31.7  10.9  yes   paid  off-­‐street  parking  available  (%)  22.6  1.7  yes   Amenities             coin-­‐operated  laundry  (%)  46.3  0.0  yes   free  laundry  (%)  16.5  45.1  yes   full  kitchen  facilities  (%)  99.4  89.1  yes   Dishwasher  (%)  14.0  9.2  no   Microwave  (%)  15.1  26.8  yes   Utilities             included  gas/electric  (worst  =  0,  best  =  1)  0.052  0.454  yes   included  water/sewer  fees  (worst  =  0,  best  =   1)   0.377  0.583  yes   included  cable/Internet  (worst  =  0,  best  =  1)  0.040  0.419  yes   locational  attributes             Walkscore  (worst  =  0,  best  =  100)  84.5  72.9  yes   crime  index  (most  dangerous  =  0,  safest  =   100)   18.3  40.6  yes   freeway  proximity  (furthest  =  0,  closest  =  1)  0.241  0.1865  no   Table  9.  The  mean  value  of  various  characteristics  of  units,  divided  into  the  categories  of  non-­‐secondary  units  (left  column,   N=164)  and  secondary  units  (central  column;  N=174).  The  right-­‐hand  column  indicates  whether  each  variable  is  or  is  not   significantly  different,  between  the  non-­‐secondary  and  secondary  unit  subsets,  at  the  95%  level  of  confidence.   First,  secondary  units  are  smaller,  with  on  average  one  half  of  a  bedroom  less  than  non-­‐secondary   units,  one  tenth  of  a  bathroom  less,  and  about  140  sf  less  floor  area.  Not  surprisingly,  the  adjusted   rent  of  the  average  secondary  units  is  considerably  less  than  that  of  the  average  non-­‐secondary   unit,  by  about  $250  per  month.  When  expressed  in  terms  of  Area  Median  Income,  in  the  manner   typical  of  subsidized  affordable  housing,  the  adjusted  rents  of  secondary  units  are  more  affordable   to  a  modest  though  significant  degree:  the  average  secondary  unit  is  affordable  to  a  household   earning  62%  of  AMI,  versus  about  68%  of  AMI  for  non-­‐secondary  units.18  While  adjusted  rent   normalized  by  floor  area  is  slightly  higher  for  secondary  units  (albeit  not  to  a  statistically  significant   degree),  we  are  nevertheless  left  with  a  distinct  impression  that  secondary  units  are  more  likely   than  units  in  the  general  rental  market  to  provide  smaller,  more  affordable  housing  packages  for   the  households  that  are  seeking  them.  If  anything,  the  estimated  6%  difference  in  affordability  as  a   minimum  AMI  percentage  is  a  lower-­‐bound  estimate,  since  non-­‐secondary  units  advertised  on   Craisglist  are  likely  to  be  less  expensive,  on  average,  than  apartments  rented  via  other,  more  costly   means  (such  as  freeway  billboards,  posters  in  train  stations,  listings  in  glossy  apartment  magazines,   etc.).  In  addition,  the  secondary  units  advertised  on  Craigslist  were  being  rented  to  strangers,  and   therefore  excluded  the  share  of  secondary  units  (which  we  estimated  in  the  homeowner  survey  to   be  51%)  that  are  rented  to  family,  friends  or  acquaintances  for  free  or  reduced  rents  in  many  cases.   If  it  were  possible  to  include  the  secondary  units  occupied  by  people  already  known  to  the   homeowner  household  in  the  analysis,  doubtless  the  disparity  in  affordability  between  secondary   and  non-­‐secondary  units  would  be  even  greater.     It  is  also  worth  noting  that  secondary  units  in  the  dataset  are  far  more  likely  (53%  versus  2%)  to   have  owners  residing  on-­‐site  than  their  non-­‐secondary  counterparts.  This  is  not  surprising  when   one  considers  that  owner-­‐occupancy  of  one  of  the  two  units  on  a  property  consisting  of  a  main   house  and  a  secondary  unit  is  mandated  by  law  in  the  cities  of  Albany,  Berkeley,  El  Cerrito,   Piedmont,  and  Oakland.19  The  strong  contrast  in  owner-­‐occupancy  between  non-­‐secondary  and   secondary  units  likely  accounts  for  many  of  the  differences  between  these  two  housing  submarkets.   It  also  provides  indirect  but  strong  evidence  that  secondary  units  are  much  more  likely  to  be  owned   and  operated  by  amateur  landlords  who  earn  a  portion  or  a  majority  of  their  living  from  activities   other  than  real  estate,  whereas  non-­‐secondary  units  are  much  more  likely  to  be  owned  and   operated  by  real  estate  professionals  or  firms.                                                                                                                                     18  Area  Median  Income  is  taken  from  the  2011  income  limit  data  published  by  the  US  Department  of  Housing   and  Urban  Development  for  the  Oakland-­‐Fremont  HMFA,  which  includes  all  of  Alameda  and  Contra  Costa   Counties,  which  in  turn  include  the  entirety  of  this  report’s  study  area.  “Affordability”  is  taken  here  to   correspond  to  the  HUD  standard  wherein  a  household  expends  no  more  than  30%  of  its  gross  income  on   housing  rent  plus  utility  costs.  The  Area  Median  Income,  again  following  standard  practice,  is  computed  for  a   household  size  that  equates  to  1.5  persons  per  bedroom  in  the  unit  (or  1  person  per  studio  apartment).  Our   method  of  adjusting  rent  to  account  for  utility  costs  roughly  mirrors  standard  practice  in  affordable  housing,   since  utility  costs,  when  not  covered  by  the  landlord,  are  effectively  deducted  from  rent  for  affordability  ratio   computation  purposes.   19  The  reason  that  owner-­‐occupancy  amongst  what  we  are  classifying  as  “secondary  units”  for  the  purposes  of   the  rental  Internet  advertisement  study  falls  short  of  100%  is  some  combination  of  1)  non-­‐compliance  with   the  law  by  property  owners  and  2)  the  inclusion  of  units  in  the  “secondary  unit”  data  set  that,  legally   speaking,  are  actually  duplexes  or  triplexes  and  thus  not  subject  to  the  owner-­‐occupancy  requirement.   Possibly  as  a  consequence  of  much  higher  levels  of  non-­‐professional  management,  secondary  units   appear  to  handle  parking  for  their  tenants  in  a  markedly  different  manner  from  non-­‐secondary   rental  units.  This  can  be  seen  from  the  distribution  of  the  three  mutually  exclusive  categories  of   parking  leasing  for  these  two  submarkets:  no  parking  provided,  free  off-­‐street  parking,  and  off-­‐ street  parking  offered  for  a  fee.  Secondary  units  are  far  more  likely  (87%)  to  offer  no  off-­‐street   parking  whatsoever  (whether  free  or  otherwise)  than  non-­‐secondary  units  (46%).  In  the  relatively   rare  cases  where  off-­‐street  parking  is  offered  for  secondary  units,  it  is  much  likelier  to  be  offered   for  free  (11%)  than  for  a  fee  (2%).  A  comment  from  a  Berkeley  homeowner  typifies  the  most   common  scenario  with  secondary  units:  “There  are  three  households  on  our  block  with  extra   housing  units  …  No  one  on  our  block  has  a  usable  garage  and  most  do  not  or  cannot  use  their   driveways  for  parking.”       Differences  in  amenities  between  secondary  and  non-­‐secondary  units  are  less  dramatic  than  is  the   case  with  parking,  but  a  couple  of  comments  concerning  these  variables  are  nonetheless  in  order.   First,  the  likelihood  of  secondary  units  offering  a  fully-­‐equipped  kitchen  is  lower  than  for  their  non-­‐ secondary  peer  units  to  a  modest  but  statistically  significant  degree.  This  could  well  be  reflective  of   the  prevalence  of  unpermitted  secondary  units  (as  discussed  in  WP-­‐2012-­‐02).  This  could  also   explain  the  somewhat  higher  incidence  of  microwave  oven  provision  in  secondary  units,  i.e.,   microwave  ovens  could  be  viewed  as  a  means  by  which  landlords  attempt  to  partially  compensate   tenants  for  offering  substandard  cooking  facilities.  Indeed,  13  of  the  18  secondary  units  lacking   cooking  facilities  (72%)  offer  microwave  ovens,  whereas  the  one  non-­‐secondary  unit  that  does  not   have  full  cooking  appliances  does  have  a  microwave  oven.       Some  portion  or  all  of  three  utility  bills  (gas/electric,  water/sewer,  and  cable/Internet)  are   markedly  and  statistically  significantly  more  likely  to  be  included  as  part  of  rent  in  secondary  units   than  in  non-­‐secondary  units.  Note  that  this  should  not  be  taken  to  imply  that  secondary  units  are   therefore  a  better  deal  for  their  tenants  than  non-­‐secondary  units,  because  these  utility  bills  have   been  factored  into  the  adjusted  rent  computed  for  each  unit  (as  described  in  the  previous  section).   It  does,  however,  mean  that  housing  packages  with  utility  payments  included  are  more  common  for   secondary  unit  tenants  than  for  their  peers  not  living  in  secondary  units.  This  discrepancy  is  likely   attributable  to  1)  the  higher  likelihood  of  sharing  utility  costs  with  an  on-­‐site,  amateur  landlord  and   2)  the  greater  likelihood  of  single  connections  to  electric  lines  and  gas,  water  and  sewer  mains   being  shared  between  a  secondary  unit  and  other  units  on  the  property.       That  locational  attributes  are  not  dramatically  dissimilar  between  secondary  and  non-­‐secondary   units  in  the  dataset  is  not  terribly  surprising,  given  that  the  corridor  from  which  the  units  were   selected  is  a  reasonably  homogeneous  environment  in  terms  of  its  urban  development  patterns.   Nonetheless,  we  see  that  secondary  units  are,  to  a  modest  degree,  located  in  less  walkable  and   lower-­‐crime  neighborhoods,  and  are  less  likely  to  be  located  within  close  proximity  to  a  freeway,   than  is  the  case  for  non-­‐secondary  units.  This  is  to  be  expected,  as  secondary  units  are  presumably   more  likely  to  be  located  in  predominantly  residential  and  lower-­‐density  neighborhoods.       Hedonic  model  methodology  and  results   A  hedonic  model,  by  regressing  the  dependent  variable  (adjusted  monthly  rent)  against  the   characteristics  described  in  the  previous  section,  can  quantify  how  the  average  actor  in  the  housing   market  values  certain  unit  and  locational  attributes.  By  modeling  the  entire  dataset,  as  well  as   separately  modeling  the  roughly  equal  portions  comprised  of  secondary  and  non-­‐secondary  units,   we  can  infer  whether  participants  in  these  two  submarkets  value  attributes  differently.   The  hedonic  model  is  specified  as  follows:   ln(adjusted _rent)=num_BRs +num _BAs +sqfootage +SECURE +OWNONSITE +ON _STREET +COINOP _LAUND +FREE _LAUND +MICROW +DISHWASHER +FULL _KITCHEN +Walkscore +crime_index +freeway +ε   Here  adjusted_rent  (rent  adjusted  according  to  the  method  described  in  the  first  section  of  this   chapter)  is  the  dependent  variable;  num_BRs  and  num_BAs  indicate  the  number  of  bedrooms  and   bathrooms,  respectively;  sqfootage  is  the  unit’s  interior  square  footage;  SECURE  is  a  dummy   variable  indicating  whether  or  not  the  unit  has  a  secure  entry  (i.e.  an  entry  that  lies  behind  a  locked   gate  or  door  with  limited  access);  OWNONSITE  is  a  dummy  variable  indicating  whether  or  not  the   owner  of  the  rental  unit  resides  on-­‐site  in  another  unit;  ON_STREET  is  a  dummy  variable  indicating   whether  or  not  the  tenant  must  find  parking  on  the  street  if  she  has  a  car;  COINOP_LAUND  is  a   dummy  indicating  whether  or  not  the  unit  has  coin-­‐operated  laundry  machines  on  the  property;   FREE_LAUND  is  a  dummy  set  to  1  only  if  the  unit  has  access  to  laundry  machines  on  the  property   for  no  charge;  MICROW  is    a  dummy  variable  indicating  whether  or  not  a  microwave  oven  is   included  in  the  unit;  DISHWASHER  is  a  dummy  variable  indicating  whether  or  not  the  unit  includes   a  dishwasher;  FULL_KITCHEN  is  a  dummy  variable  indicating  whether  or  not  the  unit  has   appliances  befitting  a  full  kitchen  (four-­‐burner  stove  and  full-­‐sized  oven);  Walkscore  is  a  freely   available  index  of  the  unit’s  address’s  “walkability”  from  0  (worst)  to  100  (best);  crime_index  is  a   proprietary  index  of  crime  for  a  given  location  based  on  published  crime  statistics  ranging  from  0   (worst)  to  100  (best);  and  freeway  is  an  index  of  proximity  to  the  nearest  freeway  that  ranges  from   0  (more  than  1000’  from  a  freeway)  to  2  (within  500’  of  a  freeway).20  (As  previously  mentioned,  a   fuller  description  of  the  construction  of  the  variables,  along  with  the  imputation  procedures  used  to   populate  variables  with  sparse  coverage,  is  provided  in  Appendix  3.)   The  semi-­‐log  form  of  the  model  allows,  in  the  case  of  a  coefficient    computed  by  the  model  for  a   particular  independent  variable    ,  for    to  be  interpreted  as  the  proportional  change  in  the   dependent  variable  resulting  from  a  unit  change  in  X  .  Thus,  if    were  computed  by  the  model   to  be  0.2,  or  20%,  for  example,  then  that  would  signify  a  20%  modeled  increase  in  adjusted_rent                                                                                                                             20  In  the  previous  section,  this  variable  was  normalized  to  run  from  a  minimum  of  0  to  a  maximum  of  1,  rather   than  from  0  to  2,  for  ease  of  interpretation.   k X ek −1 ek −1 with  an  increase  in  X  by  one  unit.  This  form  therefore  allows  for  an  intuitive  interpretation  of  the   monetary  contribution  to  rents  from  the  various  independent  variables  that  are  modeled.21     Results  from  full  data  set.  Table  10  allows  a  comparison  of  the  results  from  three  model  runs:  one   performed  on  the  full  data  set,  and  the  other  two  performed  on  two  subsets,  one  including  only  the   non-­‐secondary  units  and  the  other  including  only  the  secondary  units.  As  might  be  expected,  the   increase  to  adjusted  rent  resulting  from  the  addition  of  a  bedroom  is  significant  and  large  in  each   case.  The  same  is  largely  true  for  bathrooms  (although  the  magnitude  of  the  increase  is  smaller),   except  that  this  variable  is  omitted  from  the  secondary  unit-­‐only  model  run  as  a  result  of  every   single  secondary  unit  in  the  data  set  (all  174  of  them)  having  one  bathroom.  These  results  are   intuitive,  as  is  the  positive  and  significant  effect  of  square  footage  (above  and  beyond  what  is   captured  in  the  bedroom  and  bathroom  variables)  in  the  full  data  set.   It  is  interesting  to  note  that  the  presence  of  an  owner  on-­‐site,  all  else  being  equal,  results  in  a   statistically  significant  reduction  in  adjusted  rent  of  almost  9%  in  the  case  of  secondary  units,  and   none  at  all  in  non-­‐secondary  units.  This  result  gives  support  to  the  supposition  that  amateur,  on-­‐ site  property  owners  are  less  concerned  about  maximizing  rental  revenues  from  tenants,  and  are   perhaps  more  concerned  with  maintaining  long-­‐term  amity  with  their  renters.  From  the  tenant’s   point  of  view,  this  result  suggests  that  secondary  units  residing  on  owner-­‐occupied  properties  are   often  a  housing  bargain.                                                                                                                           21  See,  for  example,  Sirmans,  G.  Stacy  and  David  A.  Macpherson.  2003.  “The  Value  of  Housing  Characteristics.”   Washington,  DC:  National  Association  of  Realtors,  December.   Table  10.  Hedonic  model  results  (excluding  imputed  square  footages).  Results  are  shown  for  hedonic  model  runs  performed   with  the  full  data  set,  non-­‐secondary  units  only,  and  secondary  units  only.  Note  that  because  the  model  was  run  in  the  semi-­‐log   form,  and  because  what  is  displayed  here  represents  the  exponentiation  of  the  model  coefficients,  each  independent  variable’s   value  shown  here  can  be  interpreted  as  the  percentage  change  in  the  dependent  variable  (adjusted  rent)  that  would  result  from   an  increase  of  one  unit  of  that  independent  variable.        Units   All  Non-­‐secondary  Secondary  units   regression  statistics         #  of  data  points  336  164  172   r-­‐squared  of  model  run  0.51  0.64  0.41               modeled  variable  coefficients  (%)         #  of  bedrooms  21.0  21.5  20.0   #  of  bathrooms  13.7  15.9  -­‐-­‐ floor  area  (per  100  sf)  1.6  1.5  1.7   secure  entry  -­‐0.8  4.6  -­‐12.0   owner  on-­‐site  -­‐6.4  -­‐1.2  -­‐8.7   no  off-­‐street  parking  -­‐4.8  -­‐10.0  2.6   free  laundry  11.9  24.1  6.5   coin-­‐operated  laundry  3.1  6.9  -­‐-­‐   microwave  oven  5.0  16.8  0.6   dishwasher  11.0  1.1  20.8   full  kitchen  facilities  43.2  24.2  43.1   Walkscore  (per  10  points)  0.7  2.6  0.2   crime  index  (per  10  points)  0.4  -­‐0.3  0.0   adjacent  to  freeway  -­‐10.0  -­‐12.0  -­‐3.3   Coefficients  that  are  in  boldface  are  statistically  significant  at  the  95%  level.     Failing  to  offer  on-­‐street  parking  depresses  rents  in  non-­‐secondary  units  by  about  9%,  while  the   comparable  coefficient  for  secondary  units  is  not  statistically  significant.  The  9%  decrease  in  rent   for  non-­‐secondary  units  compares  to  similar  decreases,  found  in  a  hedonic  study  of  for-­‐sale  housing   San  Francisco  in  the  mid-­‐1990s,  of  12%  for  SFR  and  13%  for  condominium  properties.22  The  9%   reduction  found  for  East  Bay  non-­‐secondary  rental  units  suggests  that  while  provision  of  off-­‐street   parking,  or  lack  thereof,  is  priced  into  transactions  in  mainstream  rental  units,  this  is  not  the  case  in   the  market  for  East  Bay  secondary  units.  Because  the  vast  majority  of  secondary  units  in  this  region   appear  to  offer  no  off-­‐street  parking  at  all  (as  shown  in  the  previous  section),  whether  paid  or  free,   it  could  be  that  tenants  seeking  secondary  units  have  little  to  no  expectation  of  off-­‐street  parking   being  offered,  and  therefore  do  not  take  this  factor  into  account  when  formulating  their  bids  for   housing.         Amenities  are  reflected  in  rents  to  varying  degrees.  For  instance,  the  availability  of  free  laundry   machines  on  the  premises  shows  up  as  a  strong  influence  on  rent  in  the  overall  data  set  and  for   non-­‐secondary  units,  but  not  in  secondary  units.  Perhaps  this  is  because  secondary  unit  tenants  are   so  often  able  to  share  the  use  of  laundry  machines  with  the  main  house  that  on-­‐site  laundry   availability  is  less  of  a  prized  feature  than  in  the  general  rental  market,  where  a  lack  of  laundry   machines  usually  means  relying  on  a  commercial  laundromat.  Meanwhile,  it  is  difficult  to   understand  the  large  magnitudes  of  the  coefficients  in  favor  of  microwave  ovens  (for  non-­‐ secondary  units)  and  dishwashers  (for  secondary  units),  particularly  the  former.  Perhaps  the   offering  of  microwave  ovens  in  mainstream  apartments,  which  are  much  less  likely  to  have   substandard  kitchens  (as  discussed  in  the  previous  section),  is  associated  with  other  attractive   kitchen  amenities,  while  the  presence  of  microwaves  in  secondary  units  could  more  often  be  a   partial  recompense  for  the  substandard  kitchens  more  likely  to  be  seen  in  that  submarket   (evidence  for  which  was  also  presented  in  the  previous  section).  That  a  large  and  statistically   significant  effect  of  kitchen  appliances  on  rents  is  only  seen  for  secondary  units  can  be  interpreted   as  the  consequence  of  market  participants  taking  into  account  the  wide  variation  of  kitchen  quality   in  secondary  units  (whereas  mainstream  rental  units,  as  previously  discussed,  tend  to  have  more   uniformly  up-­‐to-­‐standard  kitchens).   It  is  interesting  that  none  of  the  three  locational  effects  have  significant  effects  on  rent,  with  the   exception  of  freeway  proximity,  which  depresses  the  rents  offered  for  non-­‐secondary  units.   Perhaps  the  perceived  benefit  of  proximity  to  commercial  and  other  neighborhood  amenities,   prized  by  some  tenants,  that  is  reflected  in  a  high  Walkscore  is  negated  by  decreased  tranquility  for   others.  Finally,  East  Bay  renters  in  general  appear  to  not  have  a  great  deal  of  price  sensitivity  to   crime  rates.                                                                                                                                     22  Jia,  W.  and  M.  Wachs.  1998.  Parking  and  Affordable  Housing.  Access  13.  See  also  Litman,  T.  2009.    Parking   Requirement  Impacts  on  Housing  Affordability.  Victoria,  CA:  Victoria  Transportation  Policy  Institute,  in   which  it  was  argued  that  subsidized  rental  housing  developments  could  reduce  development  costs  –  and   thus,  presumably,  rents  –  by  10%  if  off-­‐street  parking  did  not  have  to  be  provided.        Lastly,  it  is  worth  noting  that  the  model  run  for  the  non-­‐secondary  units  has  a  considerably  higher    value  than  that  for  the  secondary  units  (0.64  versus  0.41).  This  can  be  interpreted  as  evidence   that  the  rental  market  for  mainstream  rental  units  is  more  transparent  (given  the  usage  of  real   estate  professionals  of  rental  brokers,  apartment  magazines,  etc.)  and  thickly  traded.  On  the  flip   side,  one  could  surmise  that  the  secondary  unit  rental  market,  in  its  much  heavier  reliance  on   amateur  actors,  exhibits  more  unpredictability  in  rent  levels  as  a  result  of  varying  experience  levels   amongst  its  landlords,  informal  renting  practices,  less  sophisticated  tenant  sourcing  (such  as  via   Craigslist),  and  the  like.       Results  using  only  non-­‐imputed  square  footages   Because  conventional  hedonic  studies  almost  always  include  actual,  rather  than  imputed,  unit  floor   areas,  we  re-­‐ran  the  hedonic  model  described  above  for  a  restricted  data  set  that  only  included   units  for  which  actual  unit  square  footage  amounts  had  been  specified  as  a  check  on  the  validity  of   the  results  from  the  full  data  set.  The  results  from  the  reduced  data  set  are  shown  in  Table  11.     The  results  highlight  the  tension  between  using  data  points  (i.e.,  properties)  that  are  complete   versus  using  a  larger  number  of  data  points.  They  also  illustrate  the  difficulties  of  collecting  data  on   a  black  market  phenomenon  –  in  a  typical  hedonic  analysis  of  rental  apartments,  having  an  actual   floor  area  value  for  each  dwelling  unit  in  the  model  would  not  be  an  issue.     Restricting  the  data  set  to  only  those  units  with  known  square  footages  reduces  the  size  of  the  data   set  by  more  than  three-­‐quarters  overall  (73%  for  the  non-­‐secondary  units  and  79%  for  the   secondary  units).  On  the  other  hand,  the  values  increase  appreciably:  from  0.51  to  0.64  for  the   full  data  set,  from  0.64  to  0.88  for  the  non-­‐secondary  units,  and  from  0.41  to  0.60  for  the  secondary   units.  It  would  appear  that  having  actual,  as  opposed  to  imputed,  square  footages  for  the  units  has  a   major  impact  on  the  predictive  power  of  the  hedonic  model. r2 r2 Table  11.  Hedonic  model  results  (excluding  imputed  square  footages).  Hedonic  model  results  are  shown  for  a  restricted  data  set   in  which  only  units  with  known  floor  areas  are  included.  (By  contrast,  Table  10  shows  the  results  from  model  runs  in  which   units  with  imputed  floor  areas  are  also  included.)  As  in  Table  10,  each  independent  variable’s  value  shown  here  can  be   interpreted  as  the  percentage  change  in  the  dependent  variable  (adjusted  rent)  that  would  result  from  an  increase  of  one  unit  of   that  independent  variable.          Units   All  Non-­‐secondary  Secondary  units   regression  statistics         #  of  data  points  80  44  36   r-­‐squared  of  model  run  0.64  0.88  0.60               modeled  variable  coefficients  (%)         #  of  bedrooms  13.0  37.1  1.5   #  of  bathrooms  16.7  3.0  -­‐-­‐   floor  area  (per  100  sf)  5.3  2.1  5.2   secure  entry  5.3  22.5  -­‐26.3   owner  on-­‐site  -­‐2.9  -­‐6.0  -­‐3.6   no  off-­‐street  parking  -­‐4.2  -­‐1.5  -­‐0.3   free  laundry  -­‐5.4  13.6  -­‐9.3   coin-­‐operated  laundry  -­‐1.9  -­‐0.1  -­‐-­‐   microwave  oven  2.3  10.8  -­‐1.3   dishwasher  18.5  28.5  31.0   full  kitchen  facilities  51.5  5.6  70.4   Walkscore  (per  10  points)  -­‐1.1  10.3  -­‐3.2   crime  index  (per  10  points)  0.9  -­‐1.4  0.9   adjacent  to  freeway  -­‐11.8  -­‐12.9  -­‐8.6   Coefficients  that  are  in  boldface  are  statistically  significant  at  the  95%  level.     In  the  model  runs  on  the  restricted  data  set,  the  number  of  bedrooms  has  a  large  and  significant   effect  in  the  expected  direction  on  adjusted  rents  for  non-­‐secondary  units,  although  none  for   secondary  units.  Number  of  bathrooms  ceases  to  have  a  significant  effect  –  presumably  this  variable   has  been  absorbed  by  square  footage,  whose  coefficient  becomes  large  and  significant  in  the  case  of   secondary  units.  In  the  restricted  data  set,  a  secure  entry  emerges  as  a  highly  valued  and  significant   amenity.  Meanwhile,  off-­‐street  parking  recedes  as  a  significant  amenity,  as  do  free  laundry  and   microwave  ovens.  Dishwashers  and  full  kitchens  remain  very  important  and  significant  for   secondary  units.  Owner-­‐occupancy  ceases  to  be  significant  for  secondary  units.  Finally,  the   locational  factors  remain  virtually  unchanged  in  the  reduced  data  set.   Thus,  in  interpreting  the  results  of  the  hedonic  data  study,  we  must  exercise  some  caution  with  the   results  presented  from  the  full  data  set  with  respect  to  the  importance  of  owner-­‐occupancy  in   secondary  units,  and  the  importance  of  off-­‐street  parking  provision,  free  laundry,  and  microwave   ovens  in  non-­‐secondary  units,  which  cease  to  be  significant  in  the  reduced  data  set.  Secondary  units   in  the  reduced  data  set  remain  highly  sensitive  to  the  quality  of  kitchen  facilities  (as  measured  by   the  presence  of  a  dishwasher  and  of  a  full  kitchen).  Adjusted  rents  for  secondary  units  also  remain   much  more  unpredictable  (as  measured  by    value),  thereby  confirming  that  the  rental  market  for   secondary  units  is  more  idiosyncratic  than  that  for  non-­‐secondary  units.           CONCLUSION   Using  both  the  homeowner  survey  and  the  rental  Internet  advertisement  study,  we  are  able  to   overcome  at  least  some  of  the  limitations  posed  by  the  opacity  of  the  market  for  secondary  unit   housing  and  to  gain  some  insights  into  the  nature  of  this  market.  From  the  homeowner  survey,  we   learn,  first  of  all,  that  secondary  units  are  a  widespread  and  prevalent,  rather  than  an  aberrational   and  exceptional,  phenomenon.  This  suggests  that  scholars  interested  in  the  functioning  of  the  rental   housing  market  in  the  San  Francisco  Bay  Area  (and  probably  other  high-­‐cost  regions  like  it)  would   be  well-­‐advised  to  take  account  of  the  existing  stock  of  secondary  units.   From  the  homeowner  survey  we  learn  about  the  physical  nature  of  secondary  units  in  the  urban   core  of  the  East  Bay.  For  instance,  most  units  are  freestanding  structures  or  converted  garages.  In   addition,  we  have  survey  evidence  that  suggests  that  homeowners  with  secondary  units  are   mainstream  insofar  as  they  appear  to  differ  little,  from  a  demographic  standpoint,  from   homeowners  who  lack  such  units.   The  survey  also  gives  us  an  indication  of  what  homeowners  who  lack  secondary  units  at  present   think  about  them.  Awareness  of  the  existence  of  secondary  units  is  high,  and  in  most  cases  the  units   themselves  are  seen  as  relatively  benign.  Where  secondary  units  are  viewed  negatively,  impacts  on   on-­‐street  parking  resulting  from  their  tenants  are  the  most  common  reason.  Nonetheless,  the   potential  market  for  secondary  unit  installation  is  quite  sizable,  with  30%  of  homeowners   expressing  at  least  a  potential  interest  in  installing  them.  This  highlights  an  unusual  aspect  of   r2 secondary  units  in  relation  to  other  types  of  rental  housing:  while  they  are  sometimes  seen  to  pose   impacts  to  local  homeowners’  quality  of  life,  as  is  typical  with  rental  housing,  they  also,  quite   unusually,  are  often  seen  as  something  from  which  homeowners  could  personally  benefit.   From  the  rental  Internet  advertisement  study  we  are  able  to  infer  that  even  most  of  the  secondary   units  rented  to  strangers  are  owned  and  operated  by  homeowners  who  live  on-­‐site.  In  addition,  the   survey  results  tell  us  that  a  sizable  fraction  of  secondary  units  are  rented  to  acquaintances,  friends   or  family,  in  some  cases  for  free  and  in  other  cases,  we  presume,  for  reduced  rents.  Thus,  the  two   methods  combine  to  give  us  a  picture  of  secondary  units  as  frequently  (though  not  always)  forming   part  of  an  amateur-­‐operated  rental  market,  with  some  informal  characteristics.  These  features  are   consistent  with  our  view  of  this  type  of  housing  as  a  black  market,  or  one  that  operates  at  least   partially  outside  of  governmental  taxation  and  regulation.  An  Albany  homeowner  provides  a   perfect  example  of  this  type  of  informality:  “We  are  on  a  corner  and  have  five  street  parking  places   as  well  as  a  driveway  and  a  garage  (three  places),  yet  we  were  forced  to  add  two  parking  spaces   when  we  added  our  [secondary]  unit.    We  are  now  in  the  process  of  illegally  removing  those  two   [newly-­‐added  parking  spaces]  as  they  take  up  virtually  the  entire  back  yard.”  Whereas  one  would   generally  assume  compliance  with  off-­‐street  parking  or  other  regulations  imposed  on   professionally-­‐managed  rental  housing,  the  picture  differs  considerably  for  secondary  unit  housing.             The  rental  Internet  advertisement  study,  finally,  provides  fairly  strong  evidence  that  secondary   units  comprise  cheaper  housing  packages  to  small  households  of  one  or  two  people  than  what  is   commonly  available  in  the  unsubsidized  overall  rental  housing  market.  We  calculate,  as  a  lower-­‐ bound  estimate,  that  this  discrepancy  corresponds  to  a  savings,  for  tenants  in  secondary  units,  of   6%  of  AMI  on  average.  Secondary  units  are  cheaper  for  a  variety  of  reasons:  they  tend  to  be  smaller,   they  are  less  likely  to  offer  off-­‐street  parking,  they  are  less  likely  to  provide  interior  amenities  such   as  dishwashers,  and  they  are  more  likely  to  share  utility  costs  with  the  homeowner  or  with  other   units,  for  example.  The  hedonic  model,  using  data  collected  in  the  rental  Internet  advertisement   study,  provides  at  least  some  evidence  that  amateur  ownership  and  operation  has  a  positive   influence  on  secondary  unit  affordability,  although  more  research  will  be  needed  to  fully  test  this   claim.               We  are  therefore  left  with  a  striking  picture  of  an  important,  and  often  overlooked,  segment  of  the   existing  market  for  rental  housing  in  the  East  Bay,  and  one  that  appears  to  be  particularly   important  for  low-­‐moderate  to  moderate  income  rental  housing.  While  our  two  means  of  probing   the  rental  market  for  secondary  units,  the  homeowner  survey  and  the  rental  Internet   advertisement  study,  have  provided  what  we  believe  to  be  important  insights  into  this  market,   much  more  work  will  be  needed  –  both  using  other  methods  and  also  studying  other  high-­‐cost   housing  markets  –  to  paint  a  full  picture  of  the  elusive  phenomenon  of  existing  secondary  units.             Appendix  1:  Homeowner  survey  instrument  (hard  copy,  English-­‐ language  version)   Thank you for participating in our study on housing in the East Bay. This survey should be filled out by the owner (or by one of the owners if there is more than one) of the property to which this survey was mailed. It should take about 15 minutes of your time to fill it out. PLEASE NOTE THAT WE WILL MAINTAIN STRICT ANONYMITY FOR ALL ANSWERS THAT YOU GIVE. YOUR IDENTITY AND INDIVIDUAL ANSWERS WILL NOT BE SHARED WITH ANYONE OUTSIDE THE RESEARCH TEAM. Extra Housing Units question Many residential properties in this area include one or more Extra Housing Units in addition to the main dwelling. Extra Housing Units are often rented out to tenants. Extra Housing Units can take different forms, including a first floor or basement that has been converted to a unit, a freestanding backyard cottage, a garage that has been turned into an apartment, and others. (Please see images of several examples of Extra Housing Units below.) 1. How many Extra Housing Units (see above for definition) are on your property? (Circle one.) a) I don’t have any extra housing units on my property. b) 1 c) 2 d) 3 e) 4 or more If your answer to question 1 was a (i.e., that you have no Extra Housing Units on your property), then skip ahead to question 56. Otherwise, please answer the following questions about each Extra Housing Unit that is on your property. WE WILL ASK YOU A SERIES OF QUESTIONS ABOUT EACH EXTRA HOUSING UNIT ON YOUR PROPERTY. THE FOLLOWING QUESTIONS ARE ABOUT THE FIRST ONE, WHICH WE WILL CALL EXTRA HOUSING UNIT #1 2. When was the decision made to install Extra Housing Unit #1 on your property? (Circle one.) a) It happened while my household owned the property b) It happened before my household owned the property If you answered “It happened before my household owned the property” to question 2, then skip to question 4. Otherwise, proceed to question 3. 3. Who did the actual construction work of installing Extra Housing Unit #1 on your property? (Circle all that apply.) a) I did it. b) A hired contractor did the work. c) A friend or relative did the work. d) Other __________________________________________. 4. How would you best describe the physical layout of Extra Housing Unit #1? (Circle one.) a) Part or all of basement or first floor converted to an apartment b) Garage converted to an apartment c) Apartment above a garage d) Attic converted to an apartment e) Rooms inside main part of house converted to an apartment f) Apartment behind main house and attached to it g) Apartment behind main house and in its own separate structure h) Other: Please describe _________________________ 5. Does Extra Housing Unit #1 have its own complete kitchen (sink, range, and refrigerator)? Yes No (circle one) 6. Does Extra Housing Unit #1 have at least one complete bathroom (toilet, sink, and shower/bath)? Yes No (circle one) 7. How many bedrooms does Extra Housing Unit #1 have? (Circle one.) a) None (it is a studio or efficiency unit) b) 1 c) 2 d) 3 or more 8. About how many years ago was Extra Housing Unit #1 installed? (If less than one year ago, just mark “0.” If you don’t know, check “I don’t know.”) ________ years ago _____ I don’t know 9. Is Extra Housing Unit #1 occupied by at least one person? Yes No (circle one) If you answered Yes to question 9, then skip ahead to question 11. Otherwise, proceed to question 10. 10. Why is Extra Housing Unit #1 currently unoccupied? (Circle one.) a) It needs physical work to be rentable. b) It is vacant, but I am looking for a tenant. c) It is being used as something other than an apartment (home office, workshop, studio, etc). d) Other reason: __________________________________________ If you answered question 10, then skip ahead to question 20 (if you have a second Extra Housing Unit) or proceed to question 60 (if you don’t). 11. Which of the following best describes the relationship of the occupant(s) in Extra Housing Unit #1 to you at the time of move-in? (Circle one) a) My household lives in Extra Housing Unit #1. b) Relative(s) c) Friend(s) d) Acquaintance(s) e) I did not know the occupant(s) before move-in. If your answer to question 11 was a), then skip to question 20 (if you have a second Extra Housing Unit) or question 60 (if you don’t). Otherwise, please proceed to question 12. 12. How did you find the occupant(s) in Extra Housing Unit #1? (Circle all that apply.) a) I already knew the occupant(s) b) Craigslist c) Other Internet source d) Newspaper classified ad e) Referred by someone I know f) Other: ____________________________ 13. How much is the rent paid to your household by the occupant(s) in Extra Housing Unit #1? If the occupant(s) are staying in Extra Housing Unit #1 for free, then mark “$0.” $__________ per month 14. Please circle all of the following utilities that are included in the rent, if any, paid to your household by the occupant(s) in Extra Housing Unit #1. Water/Sewer Gas/Electricity Trash/Recycling Pickup Telephone Cable TV Internet 15. How long has the person who has been living in Extra Housing Unit #1 for the longest time lived there? _________ years _____ less than one year (check if applicable) _________ not sure 16. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #1 (if any) normally park on your property (i.e. not on the street)? (Please choose one.) a) none b) 1 c) 2 d) 3 or more 17. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #1 (if any) normally park on the street within a five-minute walk of where you live? (Circle one.) a) none b) 1 c) 2 d) 3 or more 18. How many children (i.e. people under age 18) live in Extra Housing Unit #1? a) none b) 1 c) 2 d) 3 or more 19. For each adult (i.e. person of age 18 or more) living in Extra Housing Unit #1, please answer the following questions. Question Adult #1 Adult #2 Adult #3 Adult #4 What is your best estimate of this person’s age? _____ years _____ years _____ years _____ years How would you best describe the ethnicity or race of this person? (Circle all that apply.) White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race What is this person’s gender? (Circle one.) Male Female Male Female Male Female Male Female 20. THE FOLLOWING QUESTIONS ARE ABOUT EXTRA HOUSING UNIT #2. IF YOU DON’T HAVE A SECOND EXTRA HOUSING UNIT, PLEASE SKIP TO QUESTION #60. OTHERWISE, PROCEED. When was the decision made to install Extra Housing Unit #2 on your property? (Circle one.) a) It happened while my household owned the property b) It happened before my household owned the property If you answered “It happened before my household owned the property” to question 20, then skip to question 22. Otherwise, proceed to question 21. 21. Who did the actual construction work of installing Extra Housing Unit #2 on your property? (Circle all that apply.) a) I did it. b) A hired contractor did the work. c) A friend or relative did the work. d) Other __________________________________________. 22. How would you best describe the physical layout of Extra Housing Unit #2? (Circle one.) a) Part or all of basement or first floor converted to an apartment b) Garage converted to an apartment c) Apartment above a garage d) Attic converted to an apartment e) Rooms inside main part of house converted to an apartment f) Apartment behind main house and attached to it g) Apartment behind main house and in its own separate structure h) Other: Please describe _________________________ 23. Does Extra Housing Unit #2 have its own complete kitchen (sink, range, and refrigerator)? Yes No (circle one) 24. Does Extra Housing Unit #2 have at least one complete bathroom (toilet, sink, and shower/bath)? Yes No (circle one) 25. How many bedrooms does Extra Housing Unit #2 have? (Circle one.) a) None (it is a studio or efficiency unit) b) 1 c) 2 d) 3 or more 26. About how many years ago was Extra Housing Unit #2 installed? (If less than one year ago, just mark “0.” If you don’t know, check “I don’t know.”) ________ years ago _____ I don’t know 27. Is Extra Housing Unit #2 occupied by at least one person? Yes No (circle one) If you answered Yes to question 27, then skip ahead to question 29. Otherwise, proceed to question 28. 28. Why is Extra Housing Unit #2 currently unoccupied? (Circle one.) a) It needs physical work to be rentable. b) It is vacant, but I am looking for a tenant. c) It is being used as something other than an apartment (home office, workshop, studio, etc). d) Other reason: __________________________________________ If you just answered question 28, then skip ahead to question 34 (if you have a third Extra Housing Unit) or proceed to question 60 (if you don’t). 29. Which of the following best describes the relationship of the occupants(s) in Extra Housing Unit #2 to you at the time of move-in? (Circle one. a) My household lives in Extra Housing Unit #2. b) Relative(s) c) Friend(s) d) Acquaintance(s) e) I did not know the tenant(s) before move-in. If your answer to question 29 was a), then skip to question 38 (if you have a third Extra Housing Unit) or question 60 (if you don’t). Otherwise, please proceed to question 30. 30. How did you find the occupant(s) in Extra Housing Unit #2? (Circle all that apply.) a) I already knew the occupant(s) b) Craigslist c) Other Internet source d) Newspaper classified ad e) Referred by someone I know f) Other: ____________________________ 31. How much is the rent paid to your household by the occupant(s) in Extra Housing Unit #2? If the occupant(s) are staying in Extra Housing Unit #2 for free, then mark “$0.” $__________ per month 32. Please circle all of the following utilities that are included in the rent, if any, paid to your household by the occupant(s) in Extra Housing Unit #2. Water/Sewer Gas/Electricity Trash/Recycling Pickup Telephone Cable TV Internet 33. How long has the person who has been living in Extra Housing Unit #2 for the longest time lived there? _________ years _____ less than one year (check if applicable) _________ not sure 34. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #2 (if any) normally park on your property (i.e. not on the street)? (Please choose one.) a) none b) 1 c) 2 d) 3 or more 35. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #2 (if any) normally park on the street within a five-minute walk of where you live? (Circle one.) a) none b) 1 c) 2 d) 3 or more 36. How many children (i.e. people under age 18) live in Extra Housing Unit #2? a) none b) 1 c) 2 d) 3 or more 37. For each adult (i.e. person of age 18 or more) living in Extra Housing Unit #2, please answer the following questions. Question Adult #1 Adult #2 Adult #3 Adult #4 What is your best estimate of this person’s age? _____ years _____ years _____ years _____ years How would you best describe the ethnicity or race of this person? (Circle all that apply.) White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race What is this person’s gender? (Circle one.) Male Female Male Female Male Female Male Female 38. THE FOLLOWING QUESTIONS ARE ABOUT EXTRA HOUSING UNIT #3. IF YOU DON’T HAVE A THIRD EXTRA HOUSING UNIT, PLEASE SKIP TO QUESTION #60. OTHERWISE, PROCEED. When was the decision made to install Extra Housing Unit #3 on your property? (Circle one.) a) It happened while my household owned the property b) It happened before my household owned the property If you answered “It happened before my household owned the property” to question 38, then skip to question 40. Otherwise, proceed to question 39. 39. Who did the actual construction work of installing Extra Housing Unit #3 on your property? (Circle all that apply.) a) I did it. b) A hired contractor did the work. c) A friend or relative did the work. d) Other __________________________________________. 40. How would you best describe the physical layout of Extra Housing Unit #3? (Circle one.) a) Part or all of basement or first floor converted to an apartment b) Garage converted to an apartment c) Apartment above a garage d) Attic converted to an apartment e) Rooms inside main part of house converted to an apartment f) Apartment behind main house and attached to it g) Apartment behind main house and in its own separate structure h) Other: Please describe _________________________ 41. Does Extra Housing Unit #3 have its own complete kitchen (sink, range, and refrigerator)? Yes No (circle one) 42. Does Extra Housing Unit #3 have at least one complete bathroom (toilet, sink, and shower/bath)? Yes No (circle one) 43. How many bedrooms does Extra Housing Unit #3 have? (Circle one.) a) None (it is a studio or efficiency unit) b) 1 c) 2 d) 3 or more 44. About how many years ago was Extra Housing Unit #3 installed? (If less than one year ago, just mark “0.” If you don’t know, check “I don’t know.”) ________ years ago _____ I don’t know 45. Is Extra Housing Unit #3 occupied by at least one person? Yes No (circle one) If you answered Yes to question 45, then skip ahead to question 47. Otherwise, proceed to question 46. 46. Why is Extra Housing Unit #3 currently unoccupied? (Circle one.) a) It needs physical work to be rentable. b) It is vacant, but I am looking for a tenant. c) It is being used as something other than an apartment (home office, workshop, studio, etc). d) Other reason: __________________________________________ Please skip ahead to question 60. 47. Which of the following best describes the relationship of the occupant(s) in Extra Housing Unit #3 to you at the time of move-in? (Circle one.) a) My household lives in Extra Housing Unit #1. b) Relative(s) c) Friend(s) d) Acquaintance(s) e) I did not know the occupant(s) before move-in. If your answer to question 47 was a), then skip to question 60. Otherwise, please proceed to question 48. 48. How did you find the occupant(s) in Extra Housing Unit #3? (Circle all that apply.) a) I already knew the occupant(s) b) Craigslist c) Other Internet source d) Newspaper classified ad e) Referred by someone I know f) Other: ____________________________ 49. How much is the rent paid to your household by the occupant(s) in Extra Housing Unit #3? If the occupant(s) are staying in Extra Housing Unit #3 for free, then mark “$0.” $__________ per month 50. Please circle all of the following utilities that are included in the rent, if any, paid to your household by the occupant(s) in Extra Housing Unit #3. Water/Sewer Gas/Electricity Trash/Recycling Pickup Telephone Cable TV Internet 51. How long has the person who has been living in Extra Housing Unit #3 for the longest time lived there? _________ years _____ less than one year (check if applicable) _________ not sure 52. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #3 (if any) normally park on your property (i.e. not on the street)? (Please choose one.) a) none b) 1 c) 2 d) 3 or more 53. How many total cars do the people (if any) who are living on your property in Extra Housing Unit #3 (if any) normally park on the street within a five-minute walk of where you live? (Circle one.) a) none b) 1 c) 2 d) 3 or more 54. How many children (i.e. people under age 18) live in Extra Housing Unit #3? a) none b) 1 c) 2 d) 3 or more 55. For each adult (i.e. person of age 18 or more) living in Extra Housing Unit #3, please answer the following questions. Question Adult #1 Adult #2 Adult #3 Adult #4 What is your best estimate of this person’s age? _____ years _____ years _____ years _____ years How would you best describe the ethnicity or race of this person? (Circle all that apply.) White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race Indian/Alaska Native Other: _______________ Mixed Race Indian/Alaska Native Other: _______________ Mixed Race Indian/Alaska Native Other: _______________ Mixed Race What is this person’s gender? (Circle one.) Male Female Male Female Male Female Male Female Please skip ahead to question 60. Extra Housing Unit questions 56. Which of the following best describes why you do not have any Extra Housing Units on your property? a) I have never given any thought to installing Extra Housing Units on my property. b) I don’t want any Extra Housing Units on my property. c) I might like to have one or more Extra Housing Units on my property, but I haven’t gotten around to installing any. d) I looked into installing one or more Extra Housing Units on my property, but it didn’t work out. e) I am planning on installing one or more Extra Housing Units on my property. 57. If your answer to the previous question was d (“I looked into installing one or more Extra Housing Units on my property, but it didn’t work out,”) please answer this question: what was the most important reason that you did not end up installing one or more Extra Housing Units on your property? a) It would have been too expensive. b) I couldn’t get financing. c) One or more of my neighbors would have been unhappy with it. d) My property couldn’t fit the amount of off-street parking required by the city. e) Other city requirement: ________________________ f) Other reason: _______________________________ 58. Is there at least one residential property on your street that has one or more Extra Housing Units on it? a) Yes b) No c) I’m not sure 59. If you answered a (“Yes”) to the previous question, do you think that the Extra Housing Units in one or more residential properties on your street have a negative impact on the neighborhood? (Please circle the best choice.) a) Yes – too many cars trying to park on my street. b) Yes – too much noise. c) Yes – unruly tenants. d) Yes – other impact: _________________________________________ e) No – there is no negative impact. Transportation questions 60. How many cars (if any) do you and the other people living in your unit, and NOT including any people living in any other units on your property (if any), normally park on your property (i.e. not on the street)? a) none b) 1 c) 2 d) 3 or more 61. How many cars (if any) do you and the people living in your unit, and NOT including any other people living in your property's other units (if any), normally park on the street within a five-minute walk of where you live? (Please choose one.) a) none b) 1 c) 2 d) 3 or more 62. How many currently usable parking spaces are on your property, including garage and driveway spaces? a) none b) 1 c) 2 d) 3 e) 4 f) 5 g) more than 5 Questions about you and your household 63. How many children (people under the age of 18) live on your property in your household (not including children who are living on your property but are members of other households)? a) 0 b) 1 c) 2 d) 3 or more 64. For each adult (over the age of 18) living in your household (but NOT including people living in Extra Housing Units, if any), please answer the following questions. Question You (person filling out this survey) Adult #2 Adult #3 Adult #4 What is your best estimate of this person’s age? _____ years _____ years _____ years _____ years How would you best describe this person’s race or ethnicity? (Circle all that apply.) White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race White/Caucasian Latino/Hispanic Black/African American Asian/Pacific Islander American Indian/Alaska Native Other: _______________ Mixed Race What is this person’s gender? (Circle one.) Male Female Male Female Male Female Male Female Which of the following best describes this person’s educational attainment? (Circle one.) Didn’t finish high school High school grad Some college College grad Graduate degree Didn’t finish high school High school grad Some college College grad Graduate degree Didn’t finish high school High school grad Some college College grad Graduate degree Didn’t finish high school High school grad Some college College grad Graduate degree 65. What was your household’s before-tax income in the last 12 months? Please include all income, including salaries, wages, investments, government benefits, etc. Please do not include people living in units on your property (if any) other than the one you live in as members of your household for the purposes of this question. (Please choose one.) a) Less than $10,000 b) $10,000 to $14,999 c) $15,000 to 24,999 d) $25,000 to $34,999 e) $35,000 to $49,999 f) $50,000 to $74,999 g) $75,000 to $99,999 h) $100,000 to $149,999 i) $150,000 to $199,999 j) $200,000 or more Final questions In the next two questions, we will ask you if you would be willing to be interviewed, and if you would like to be part of a raffle. If you do not want to be either interviewed at a later time or included in the raffle, we will replace your name in our records with a number so that your anonymity will be completely protected. If you agree to be interviewed, we will not publish your name. If you elect to be part of our raffle prize drawing but do not agree to be interviewed, we will replace your name with an identifying number once the prizes have been mailed out. We are committed to protecting your anonymity. 66. While we expect that we will learn a lot from the results of this survey, there is also a lot that we can only learn by doing in-person interviews. Would you be willing to have us contact you to schedule an interview at a later date? As is the case with this survey, strict anonymity would be maintained. Yes No (Circle one) 67. As a thank-you to people who have taken the time to fill out this survey, we are conducting a raffle drawing. The prizes will be 1) a $200 Apple Store certificate; 2) a $150 BART card; and 3) a $100 BART card. Would you like to take part in the raffle drawing? Yes No (Circle one) 68. While we tried to be comprehensive in selecting the questions to include in this survey, there surely are issues that we have not considered. If there is anything else about your experience with housing on your property or in your neighborhood that you would like to tell us about, please write it below. ___________________________________________________________________ ___________________________________________________________________ ___________________________________________________________________ ___________________________________________________________________ ___________________________________________________________________ THANK YOU FOR TAKING THE TIME TO FILL OUT THIS SURVEY. PLEASE PUT THE COMPLETED SURVEY INTO THE SELF-ADDRESSED, PRE-STAMPED ENVELOPE, SEAL IT, AND PUT IT IN THE MAIL TO RETURN IT TO THE RESEARCHERS. WE APPRECIATE YOUR TIME. Appendix  2:  Homeowner  respondent  solicitation  postcard  (English-­‐ language  version)     Appendix  3:  Summary  of  variables  and  imputation  methods  used  in   rental  market  analysis  of  online  apartment  advertisements   This  section  gives  a  fuller  description  of  how  each  variable  harvested  from  Craigslist   advertisements  for  rental  apartments  and  used  in  the  hedonic  model  and  for  descriptive  statistics   was  computed.  In  cases  where  variables  were  partially  populated  via  imputation,  the  methodology   by  which  this  was  done  is  described.   adjusted_rent  is  the  dependent  variable  in  the  hedonic  analysis.  The  baseline  rent  was  ascertained   from  the  advertisement  (all  advertisements  that  failed  to  report  a  monthly  rent  were  discarded),   and  then  adjusted  according  to  whether  or  not  off-­‐street  parking  was  offered,  and  whether  or  not   all  of  part  of  gas/electric,  water/sewer,  and  cable/Internet  utility  costs  were  included  in  the  rent.   Where  an  off-­‐street  parking  space  was  offered  for  a  given  monthly  cost,  this  cost  was  added  to  the   rent.  Where  an  off-­‐street  parking  space  was  offered  for  an  unspecified  price,  a  price  randomly   selected  from  the  units  with  specified  parking  costs  was  used.  Where  an  off-­‐street  parking  space   was  offered  for  free,  the  rent  was  decreased  by  a  price  randomly  selected  from  the  set  of  units  with   specified  parking  costs.  (In  cases  where  no  off-­‐street  parking  was  offered  at  all,  and  where  any   tenant  parking  therefore  occurred  on  the  public  street,  adjusted_rent  remained  unchanged  but  the   ON_STREET  dummy,  described  below,  was  flagged.)  Upward  or  downward  adjustments  were  done   on  the  basis  of  imputation  in  76  records  (out  of  a  total  of  338),  while  upward  adjustments  on  the   basis  of  directly  reported  off-­‐street  parking  charges  were  done  in  33  cases.  In  cases  where   gas/electric  and  water/sewer  charges  were  fully  included  with  rent,  monthly  rent  was  adjusted   downward  by  an  amount  selected  from  the  table  of  utility  allowances  (as  of  12/1/2010)  published   by  the  Oakland  Housing  Authority  for  use  in  subsidized  housing  rent  calculations  (electric  heating   and  cooking  were  assumed,  although  the  amounts  were  little  different  than  had  gas  usage  been   assumed).  These  allowances  vary  with  the  number  of  persons  assumed  to  be  in  the  household,   which  in  turn  vary  with  the  number  of  bedrooms  reported  for  the  unit  (1  person  is  assumed  to   occupy  a  studio,  and  1.5  people  are  assumed  to  occupy  each  bedroom  in  a  unit  that  has  one  or  more   bedrooms).  In  cases  where  one  or  both  of  these  charges  were  shared  with  the  landlord,  one  or  both   of  the  amounts  of  the  downward  revision(s)  were  halved.  Cable  and  Internet  packages  were   assumed  to  be  worth  $60  per  month  based  on  a  perusal  of  offerings  by  local  companies  serving  the   East  Bay.  In  cases  where  only  electric  but  not  gas,  or  only  cable  but  not  Internet,  or  vice  versa,  were   offered,  the  adjustment  was  also  cut  in  half.    num_BRs  and  num_BAs  were  straightforward:  the  number  of  bedrooms  and  bathrooms  indicated  in   the  ad  were  recorded  in  the  database.  Studio  units  were  recorded  as  having  zero  bedrooms.  Every   advertisement  that  failed  to  indicate  a  number  of  bedrooms  and  bathrooms  for  the  unit  was   excluded  from  the  database.   sq_footage  was  recorded  from  the  advertisement  in  cases  where  it  was  specified.  Since  square   footage  was  specified  in  only  80  out  of  the  338  records  that  were  recorded,  it  was  necessary  to   impute  square  footage  for  the  majority  of  units.  This  was  done  by  randomly  selecting  a  square   footage  amount  from  the  set  of  units  with  the  same  bedroom/bathroom  configuration  with  known   square  footages.  In  cases  where  there  were  not  enough  units  for  this  to  be  possible  (as  was  the  case   generally  with  units  with  larger  numbers  of  bedrooms),  a  smaller  unit  configuration  was  randomly   selected  and  then  square  footages  were  adjusted  upward  by  an  assumed  100  sf  per  bedroom  and   50  sf  per  bathroom  (or  25  sf  per  half  bathroom).   SECONDUNIT  was  recorded  as  a  dummy  variable  to  indicate  whether  or  not  a  unit  is  a  secondary   unit.  Note  that  this  variable  was  not  actually  included  in  the  hedonic  model  runs,  but  rather  was   used  to  split  the  data  set  into  portions  of  roughly  equivalent  size  exclusively  composed  of  either   secondary  or  non-­‐secondary  units.   SECURE  is  a  dummy  variable  that  indicates  whether  or  not  a  unit  has  a  secure  entry  (i.e.  an  entry   that  lies  behind  a  locked  gate  or  door  with  access  limited  to  those  residing  on  the  property).   ON_STREET  is  a  dummy  variable  that  takes  on  the  value  of  0  if  off-­‐street  parking,  whether  free  or   paid,  is  provided  on  the  property  (in  which  case  adjusted_rent  is  adjusted  either  upward  or   downward,  as  described  above),  and  1  if  the  only  parking  to  be  found  is  on-­‐street  (in  which  case   there  is  no  adjustment  to  adjusted_rent).  Free  off-­‐street  parking,  paid  off-­‐street  parking,  and  on-­‐ street  parking  are  considered  to  be  mutually  exclusive  categories  for  each  unit.     OWNONSITE  is  a  dummy  variable  that  indicates  whether  or  not  the  owner  of  the  rental  unit  resides   on  the  same  property  in  another  unit.  It  was  only  set  in  the  affirmative  if  the  text  of  the  ad  made  it   clear  that  the  owner  was  present.  Because  absentee  landlordism  is  often  seen  as  a  negative  by   renters,  it  was  assumed  that  this  characteristic  was  not  in  place  unless  it  was  specifically  mentioned   (i.e.,  since  landlords  would  have  an  incentive  to  mention  the  presence  of  a  landlord  in  an  ad  if  on-­‐ site  landlord  presence  were  indeed  seen  as  a  good  thing).   COINOP_LAUND  is  a  dummy  variable  that  was  set  to  1  if  the  ad  mentioned  the  presence  of  coin-­‐ operated  laundry  machines  (washer  and  dryer)  on  the  same  property  as  the  unit.   FREE_LAUND  is  a  (mostly)  dummy  variable  that  was  set  to  1  if  the  ad  mentioned  the  presence  of   laundry  machines  (washer  and  dryer)  on  the  same  property  as  the  unit  (whether  in  the  unit  or   elsewhere  on  the  premises)  that  the  tenant  could  use  for  free.  In  one  instance,  the  ad  mentioned   that  laundry  machines  could  be  used  only  on  a  Friday  or  Saturday;  in  this  case,  this  variable  was   assigned  a  value  of  0.5.   MICROW  is  a  dummy  variable  indicating  whether  or  not  the  unit  was  equipped  with  a  microwave   oven.  A  microwave  oven  was  assumed  to  not  be  provided  unless  affirmatively  specified.   DISHWASHER  is  a  dummy  variable  indicating  whether  or  not  the  unit  included  a  dishwasher   (assumed  to  not  be  provided  unless  specifically  mentioned).   FULL_KITCHEN  is  a  dummy  variable  that  was  set  to  1  if  the  unit  had  up-­‐to-­‐standard  kitchen   appliances,  including  a  four-­‐burner  stove  (whether  gas  or  electric)  and  a  full-­‐sized  oven.  An  up-­‐to-­‐ standard  kitchen  was  assumed  unless  the  advertisement  mentioned  otherwise  (due  to  our   presumptions  about  tenants’  expectations  of  rental  properties).  If  an  ad  made  any  mention  of   elements  of  a  substandard  kitchen  –  for  example,  the  presence  of  a  hotplate,  two-­‐burner  stove,   convection  oven,  or  half-­‐sized  refrigerator  –  then  a  value  of  0  was  assigned  to  this  variable.   Walkscore  is  a  percentile  index  ranging  from  0  to  100  indicating  the  “walkability”  of  any  address  in   the  United  States.  It  is  measured  via  a  free  web  tool  at  www.walkscore.com.    Walkability  is   construed  to  correspond  to  the  number  and  variety  of  businesses  and  convenience  services  that  can   be  reached  within  a  short  walk  of  a  given  location.  A  value  of  0  means  that  a  location  is  equivalent   in  walkability  to  the  least  walkable  locations  in  the  United  States  and  a  value  of  100  means  that  it  is   equivalent  to  the  most  walkable  locations.   crime_index  is  a  percentile  index  ranging  from  0  to  100  indicating  the  blended  level  of  violent  and   property  crime  rates  (from  publicly  available  data)  for  a  given  address  in  the  United  States.  It  can   be  determined  via  the  use  of  a  proprietary  web-­‐based  tool  at  www.neighborhoodscout.com.  An   index  of  0  indicates  that  a  location  is  as  dangerous  as  the  least  safe  neighborhoods  in  the  United   States,  whereas  an  index  of  100  means  that  the  location  is  as  safe  as  the  safest  locations  in  the   nation.    freeway  is  an  index  of  proximity  to  a  freeway.  It  takes  on  a  value  of  0  if  a  location  is  at  least  1,000’   from  the  nearest  freeway,  1  if  it  is  between  500’  and  1,000’  from  the  nearest  freeway,  and  2  if  it  is   within  500’  of  the  nearest  freeway.