--r-------------------

Summary Evaluationof theRentalRehabilitationProgram

Based on the Final ReportPrepared by:

Kathleen G. HeintzThomas G. KingsleyBarbara J. LipmanTed R. MillerAnn B. SchnareMargery A. Turner

The Urban InstituteWashington, D.C.

Contract HC-5747May 1987

The contents of this report do not necessarily reflect the views orpolicies of the Department of Housing and Urban Development or theU.S. Government.

TABLE OF CONTENTS

Introduction 1

Prnpose of the Evaluation 2

Overview of Program Perfo=ance 3

Admmistering Agencies 4

Subsidy Mechanism 5

NeIghborhood and Project Selection 7

Program Streamlining 8

Role of the PHA 9

Program Impact on Tenants 11

Private Sector Response 13

Nature and Cost of Repairs 14

Current and Long-Te= Perfo=ance 16

SUMMARY OF THEFINAL REPORT

Introduction

The Rental Rehabilitation Program (RRP) wasau~orized on November 30, 1983, by Section 301 of the198~ Housing and Urban-Rural Recovery Act. Congressiniti~y appropriated $300 million for the program ($150milliop each for FY 1984 and FY 1985) to be distnbuted ona formula entitlement basis to cities with a population of50,000 or more, urban counties, consortia of units ofgeneral local government, and States for the rehabilitationof privately owned rental housing.1 In addltion, funds wereappropriated to provide rental assistance, in the form ofSection 8 certificates and housing vouchers, forapproximately 60,000 households.

The primary objective of the Rental RehabilitationProgram is to increase the supply of safe, decent, andaffordable housing for lower income households throughthe renovation of the existing rental housing stock. Assuch, it reflects a general shift away from the moreexpensive new construction programs of the past. Theprogram is also targeted to a segment of the stock whichhas received less attention under previous rehabilitationprograms-- smaller rental properties with moderate repairneeds. Finally, the program breaks traditional patterns byadopting a "split subsidy" approach. Rehabilitationsubsidies are provided to property owners to help supportthe costs ofrepairs, but project rents are allowed to rise totheir market levels. At the same time, rental assistance ismade available to eligible lower income tenants, who caneither remain in the renovated units or move elsewhere.

The design of the Rental Rehabilitation Program isintended to allow maximum discretion at the local level.Grantees receive formula allocations of Rental Rehab fundswhich are then given to project owners to cover up to halfof the eligible rehabilitation costs or an average of $5,000per unit, wlnchever is less. Grantees choose the form inwhich the subsidy is offered, identify the neighborhoods inwhich the program will operate, determine the nature of

lOne nulhon dollars was allocated each year to tecbmcal assistance.

2 SummaI)' of the Fmal Report

repairs to be funded, and set the amount of subsidy to beprovided to individual property owners. The privatemarket onentation of the program is embodied in thehmitation of RRP funds to 50 percent of project costs andthe absence of any control on post-rehab rent levels.Significantly, the program relies on the market to set post­rehabilitation rents and assumes that proper selection ofneighborhoods and projects will result in rents that areaffordable to lower income renters while also yielding anadequate return to propeny owners.

In order to minimize displacement, the RRP alsoincludes a tenant assistance component, consisting ofspecial allocations of Section 8 certificates and housingvouchers. These may be used to help existing householdsremain in their units after rehab or to seek other housing oftheir choice. Until FY 1987, certificates and vouchers couldalso be offered to new households initially occupyingvacant units. However, rental assistance always followsthe tenant and is not tied to the rehabilitated unit. Thusoccupancy patterns in RRP projects are detennined bymarket forces, subjecting owners to both the rewards andrigors of the competitive process.

Purpose ofthe Evaluation

This evaluation examined the performance of theprogram approximately two years after its inception. Thestudy focused on the experience of cities and urbancounties, and excluded the State-administered componentof the program. It also excluded communines that had notcompleted a project by the sample selection date of March31, 1986. The analysis was based on a sample of 35representative sites and was designed to reflect the way mwhich the typical grant dollar has been administered.

One community in the sample -- New York City -­has been treated as a separate case study. Although itreceived the largest allocation of program funding (about 11percent of the national total), New York's program did notreflect the basic RRP model and thus was excluded fromthe core analYSiS. The fmdings presented in the body of thereport relate to program administranon and outcomes in theremaining 34 sites, which included 28 metropolitan citiesand 6 urban counties.

The purposes of this evaluation are twofold. First,it is intended to assess the extent to which the program ismeeting its primary objectives through an examination ofthe types of properties rehabilitated under the program, the

\

Rental Rehab Program

nature of the reparrs completed, and the characteristics ofthe tenants served. Second, it is intended to examine localapproaches to program operations.

The Rental Rehabilitation Program departs in manyrespects from previous rehabilitation approaches. WlnleFederal regulations establish the basic outline of the RRP,and HUD has developed a number of criteria against whichperformance could be judged, the success of the programultimately depends on the ways in which the RRP isdesigned and implemented at the local level. The study ISmtended to describe ImplementatIon approaches adoptedby local programs and, to the extent possible, identify thosefeatures ofprogram design and administration thatcontribute to better program performance.

The study employs two basic types of data: programdata, describing the RRP programs as they are operating III

a sample of 34 cities and counties nationwide, andproperty-level data, describing the characteristics of theprojects completed in the sample sites. Program data werecollected on site in each of the 34 sample grantees tlrroughadministrative interviews with RRP program staff,representatives of participating PHA's, and other actorsresponsible for the implementatIon of the program.IntervIews were conducted between June and September1986, after the program had been in operation forapproximately two years.

Property-level data include information on propertyand tenant charactenstics contained in HUD's RRPCash/Management Information System (COO) as well assupplemental data collected on site for a sample of 125completed properties. These data were collected tlrrough acombination of reviews of individual project fIles,interviews with property owners, an<;i inspections of thecompleted rental rehab properties. COO data reflectoutcomes in the 34 sites as of June 1, 1986.

Overview ofProgram Performance

At the time of the study (July 1986), the medIan sitein the sample had COmtultted 58 percent of its combinedFY 1984 and FY 1985 grant allocations and had expendedabout 10 percent of those allocations on completed projects.Based on the completions, the Rental Rehab Programappeared to be meeting the major performance objectivesestablished by HUD:

3

4

•

•

•

•

•

Summary ofthe Fmal Report.

lower income households were theprimary beneficianes of the program-- 93 percent had incomes below 80percent of the area median, and 79percent of all post-rehab tenants hadincomes below 50 percent of the areamedian;

post-rehab rents were generallyaffordable -- 90 percent of all unitshad rents that were at or below theapplicable Fair Market Rent (FMR);

the program was producing asubstantial share of larger units -- 80percent of all completed_units hadtwo or more bedrooms, and 20percent had three or more;

the RRP rehab subsidy was relativelylow -- $4,290 for the average unit($4,964, including all publicsubsidy); and

each dollar of RRP rehab subsidyhad been matched by $1.12 of privatefunds ($0.92 for all public subsidies).

However, sites within the study sample exhibitedtremendous variation on these and other measures ofprogram performance, and a locality's success (or failure)in meeting one criterion was generally unrelated to itsperformance with respect to others. This outcome wasneither surprising nor inappropriate, given the diversity oflocal conditions, priorities, and needs.

Administering Agencies

The communitIes selected for this evaluationinclude 28 cities and 6 urban counties, ranging in size fromabout 50,000 to over 3 million population. Typically, theprogram was designed and operated by a city or countyhousing rehab agency, working in conjunction with anindependent Public Housing Authority (PHA). Otherorganizational arrangements included six sites where therehab entity and the PHA were both departments of a singlecity or county agency, and three sites where a PHA wasresponsible for both the rehab and tenant assistancecomponents of the program. Agency type had little impacton performance.

Rental Rehab Program

Lead agencies in the 34 sites were primarily city orcounty community development (CD) departments, butincluded a few non-CD agencies and nonprofitorganizations, as well as the three PHA's. Notsurprisingly, there was substantial variation in their priorexperience in housing rehab. However, the relationshipbetween experience and RRP performance is notparticularly strong. Focusing on two key measures ofperformance -- production and leveraging -- it was foundthat sites with substantial output under prior CD programsachieve about the same level of commitments as less­experienced sites, but show less success in leveragingprivate funds than sites with either low or moderateexperience in housing rehab. By contrast, the sites with theleast amount of experience tend to be concentrated in thehighest performance category as measured by bothproduction and leveraging.

Similarly, participation in the RRP demonstrationdoes not appear to be associated with stronger performanceunder the program itself. Indeed, sites with demonstrationexperience were underrepresented in the highestperformance group, while nonparticipants tended to beconcentrated in this category. While difficult to explain,this pattern may in part reflect the greater use of repayablesubsidies (direct loans and repayable deferred-paymentloans) among demonstration participants. Sites that werenew to the RRP concept were more likely to adopt themodel promoted by HOD and offer subsidies in aforgivable form. As described below, the use offorgivablesubsidies (grants and forgivable deferred-payment loans)typically led to lugher commitment rates.

Subsidy Mechanism

The mechanism by which the RRP subsidy isdelivered proved to be one of the most important choicesopen to RRP grantees. Flexibihty in subsidy selectionpresumably allows Sites to gear the program to local marketcond1tions. This is done by making the subsidy attractiveenough to generate demand for the program while alsoholding public contributions to the minimum necessary tosupport needed repairs. However, program ad1ninistratorsand other local decision makers may not always assessmarket conditions correctly. Furthermore, local politicsappear to have made repayable forms of subsidy morefeasible than nonrepayable forms in a substantial number ofsites.

5

6 SummaI}' of the Fma! Report

The types of subsidIes offered by the sites includedgrants, forgivable deferred-payment loans, direct loans,repayable deferred-payment loans, and interest subsidies.Taken together, grants and forgivable deferred-paymentloans were the most popular approach to subsidy provlsion,with 17 sites choosing one or the other of thesenonrepayable forms. Fourteen sites delivered the subsidyas a repayable loan, and three sites offered both forgivableand repayable forms. In general, sites appear to haveplaced less emphasis on leveraging private funds thanmight have been expected. Other public funds, mostcommonly from the Community Development Block GrantProgram (CDBG), were frequently used to supplementRental Rehab subsidies, particularly in sites that offered thebasic subsidy as a loan. Altogether, 21 of the 34 sitesoffered additional public funding to at least some of theirRental Rehab projects. At the same time, very few sitesattempted to minimize subsidies through the use of gapfinancing techniques,2 variable loan terms, or theimposition of an RRP subsidy maximum lower than the 50percent of cost or $5,000 per unit cap established by theprogram. The sites' reluctance to use gap financingapproaches appears to reflect an lmportant tradeoffbetween the goal of minimizing subsidy amounts and thedesire to make the program attractive to potentialparticipants and to reduce the adIninistrative burden of theprogram.

As noted above, the type of subsidy offered by thesites appears to have a substantial impact on programperformance. Sltes offering a nourepayable subsidy showhigher rates in committing RRP funds overall, regardless ofthe markets in which they are operating. However, marketfactors also appear to playa role, with sites operating intight markets showing generally higher commitment ratesthan those operating in loose markets. Tight markets areassoclated with the lowest overall leveraging ratios,suggesting that programs in these areas have purchased atleast some of therr production (as measured by percentageof RRP funds committed) through the use of other funds tosupplement Rental Rehab dollars.

It is also important to note that, over the 2-yearperiod covered by the study, 9 of the 34 sites madeadjustments to their original subsidy approach in order tomcrease its attractiveness to property owners and maintainan adequate flow of applications. Adjustments mc1udedswitching from a repayable subsidy to a grant or forgivabledeferred-payment loan, reducmg interest rates and/ordeferring payments on direct loans, and providing

2Gap fmancmg analysis would be used to minmuze the rehab subsIdy on a case-by-case basIS tothe SpecIfIC amount needed to result m an appropnate rate of return for the property owner.

Rental Rehab Program

additional funding from CDBG allocanons. These changesmost clearly reflect the tradeoffs between producnon andother objectives, such as leveraging or obtaimng a paybackfrom RRP monies invested. They also demonstrate theflexIbility inherent in the program, which allows sites toadapt therr subsIdIes to changing market conditions or toredesIgn therr programs if necessary to achieve the desiredresults in a gIVen market.

Neighborhood andProject Selection

Another important aspect of program design is theselection of target neighborhoods. Despite the emphasis oncareful neighborhood selecnon, most programs opted todefme as broad a target area as possible. The maJonty ofthe sites designated eligible areas on the basis of HUDregulations, WIthout further attempts to focus the program.Five did not designate eligible areas at all, choosing insteadto qualITy projects as applications were received, based onproject- or block-level data By contrast, 13 sites selectedsome subset of the otherwise ehgible areas in therrjurisdictions, wIth about half selecting predesignatedcommunity development neighborhoods.

The principal rationale for less restrictive targetingwas a desire to cast a wide net in recrUlting project owners.As with subSIdy mechanisms, changes in target areadesignation were not uncommon. Over the 2-year penod,eight sites acted to expand or abandon originally deSIgnatedtarget areas in an attempt to broaden the base from whIchowners could be recruited. Nevertheless, based on data forthe areas in which RRP projects are actually located, itappears that selecnon of neIghborhoods has been consistentwith the intent of the program. Vanation in thecharacteristics of RRP neighborhoods has no consistentmfluence on program outcomes or overall performance.

The sites' approaches to project selection alsoreflected the need to achIeve or mamtain an adequate flowof commitments. Program descriptions and other earlydocumentation typically contamed explicit preferences forthe types of owners/projects sought, and occasionallycontained rather complex scoring systems for use in projectselection. In practice, however, projects were approved ona frrst-come, frrst-served basis, and in many cases the statedpreferences could not be implemented. The only factorsthat appeared to playa major role in project selecnon werethe need to avoid displacement and the size of the unitsproposed. Given the ratio of two- and three-bedroom umts

7

8 Sununary of the Fma! Report

specIfied by the program standards, projects containinglarger units generally received prionty.

While in concept both neighborhood and projectselection criteria are important features of program designand key elements of local dIscretion, in practtce the samplesites were unable to exercise much selectivity in thisregard. The vast majority of SItes indicated that pressuresto achieve target cOillillltment rates, combined in manysites with a dearth of applications, resulted in the selectionof virtually all projects that met basic eligibility criteria.Although most sites employed a variety of approaches toadvertise the program and recruit owners, about a thirddescnbed the level of applications as inadequate. Whilesome of these sites intended to intensify their marketingefforts in order to remedy the problem, at least some wereconsidering a redesign of the subsidy approach in order toincrease Its attractiveness to potential participants.

Program Streamlining

The study also examined the ways in which variousroles and responsibilities have been allocated under theprogram, and the extent to which a "streamlined" versus a"handholding" approach has been employed in the SItes. Ingeneral it was found that RRP staffs were small, with fewerthan three full-time eqUlvalents in about two-thirds of thesample sites. Except in the larger programs, RRP stafftended to be assigned only part ttme to the RRP, with otherprograms consuming a good proporuon of the director'stime. Responsibihty for various program activities tendedto be centralized in these staffs, with one or tworehabilitation specialists performing all necessaryfuncttons, occaSIOnally obtaining specialized help fromother city departments. Adlninistrattve costs were typicallyan mdIstingnishable part of CDBG adlninistration, with noseparate budget established for the RRP.

The RRP did exhibit an overall tendency towardsstreamlimng, particularly with respect to ownerresponsibility for obtaining secondary financing. In themajonty of sites, this was viewed as the sole responsibilityof the owner, and few sites made any formal attempt toinvolve private lenders in the program. In 14 of the sites,owners were also responsible for preparing the detailedspecifications of the work to be accomplished.3 Althoughthe specification writing function was retained by program

3TIus program deSIgn chOIce produced problems m the quahty of the reparrs See page 13 for amore complete dISCUSSIon

Rental Rehab Program

staff in the remaining areas, arguments for doing so werebased on efficiency and speed of processing, rather than onmaintaining control of rehab decisions.

Taking into consideration the allocation offunctions and other aspects of program processing, siteswere classified along a handholdmg/streamliningcontinuum. Both "handholdmg" and "streamlined" siteshad higher production and completIon rates than those inthe intermedlate group. Nevertheless, streamlinersachIeved slightly better results in both regards, at asubstantially lower cost. Based on rough calculations, it ISestimated that the average cost of adlninistration rangedfrom about 2 percent of the average per-unit subsidyamount m streamlined sites to about 9 percent in sites thatadopted a more "handholdmg" approach, with a mean of4.4 percent for the sample.

Finally, the study provides insights into HUD'sCash! Management Information System (COO), which isused both to control disbursements to committed projectsand to collect data on the progress and outcomes of theprogram. While the smdy did not explicitly focus on thefunds management aspect of the system, It appeared towork well in most sites. As a data collection/programmomtoring tool, the COO also appears to provide areasonably accurate picture of overall national programperformance. Although a sizable number of errors m thereported data for individual projects were detected, the netdifferences on individual items and the overall impact onthIS analysis were slight. Problems in COO reporting arealready being addressed by the Department and shouldresult in an improved data base for future program decisionmaking.

Role of the PHA

Because the Rental Rehab Program uses two typesof subsldles -- rehabilitation subsidies to property ownersand rental subsidies to tenants -- the program typicallyrequires iliat two separate entIties cooperate in ilieimplementation of ilie program. Normally, responsibilityfor rehab functions (including program design, propertyselection, specification of needed repairs, and projectprocessing) is assigned to a city/county rehab office.ResponsibIlity for issuing certificates or vouchers toeligIble tenan!s belongs to ilie PHA, typically an .independent local agency. In only three of the sample siteswas ilie entire program PHA adlffinistered.

9

10 Summary ofthe Fmal Report

Overall, it appears that PHA's have played a fairlymodest role in the Rental Rehab Program. City!countyagencies generally took the lead in program design, withonly six of the sample PHA's indicating that they hadplayed an "active" role. Simllarly, only five of the PHA'sindicated an ongoIng involvement at a policy level, andonly a few participated in the selection of properties ordecisions regarding property repairs. Although themajority conducted a final ~ection 8 Housing QualityStandards (HQS) inspection, PHA's rarely took the lead.Instead, their active role was typically limited to mattersdirectly related to the provision of tenant assistance.

Even within this category of activities, however,PHA functions were often fairly narrow. Specifically, inonly a minonty of the sites did PHA's appear to take a"managing" role with respect to overall tenant Issues.PHA's were typically not involved in such activIties ascollecting data on initial tenants or other program-relatedreporting, except for maintaining records on assistedhouseholds. Rather, reporting and monitoring functions -­to the extent they were performed -- were retamed by thegrantees. In a large number of sites the interaction betweengrantee and PHA staff was frequent and of a collaborativenature. In many others, however, city staff took the lead inall aspects of tenant assistance -- including initialscreening -- calling on the PHA only when specifichouseholds were ready for formal certification and theIssuance of a voucher or certificate.

Overall, both city and PHA actors appeared to besausfied with the allocation of functions between them andhad developed good working relationships in carrying outtheir respective responsibilitIes. Nevertheless, some areasof tension did arise. With respect to design of the program,the most common problem was the types of neighborhoodsin which RRP projects were located. About 23 percent ofthe PHA's expressed concern that the neighborhoods weretoo deteriorated or unattracuve to appeal to certificateholders. An equal proportion mentioned administrativeproblems related to HQS failures which requiredreinspection of units (23 percent) or to coordinaung thetiming of project completion (23 percent).

Despite general satisfaction with program roles onthe part of both grantee and PHA actors, it appears that theinteraction of two different actors can work to thedetriment of the program if responsibilities are notadequately speCified and understood by both parties.SpecIfIcally, In one of the sample sites it was found that amisunderstanding about which actor was responsible foroverall tenant issues -- IncludIng the Identification ofpotential displacees -- had resulted in a failure to performthese funcuons. In general, regardless of the diligence with

Rental Rehab Program

which grantees or PHA's appeared to perform theseactivities, lack of related record keeping made it virtnallyimpossible to detennine the exfent of displacement that hadoccnrred under the program or to collect data onhouseholds that had moved from Rental Rehab projects. Inaddinon, the overall impression was that sites needed todevote more attention to ensuring that tenants weremformed of their options under the program and that tenantmoves were monitored by one or the other of theparticipating actors.

This suggests a need WIthin the program as a wholeto clarITy responsibihties with regard to tenant management-- including such record-keeping requrrements as HUDdeems appropriate -- and to ensure that these funcnons areclearly assigned to one of the participating actors. HUD isnow m the process of clarifying tenant monitoring and datacollection responsibilities in Its communications to granteesand HUD field offices.

Program Impacton Tenants

The Rental Rehab Program is clearly serving thepopulation of households it was intended to serve: 79percent of all post-rehab tenants were very-Iow-income(below 50 percent of the area med1an) households, andanother 14 percent had incomes between 50 to 80 percent ofthe area median. Sixty-nine percent were lIDnorities, andabout half were female-headed families with children.4Thus households with the greatest overall incidence ofhousing needs appear to have been the primary recipientsof program benefits.

The great majority of RRP umts also meet theaffordab1lity standard adopted by HUD. While the averageunit experienced a rent increase as a result of itsrenovation, particularly III buildings where rehab costswere high, the overwhelmmg majority of RRP units hadpost-rehab rents that were at or below the apphcable fairmarket rent (FMR). The average unit in the sample rentedfor about 87 percent of the FMR, and only about 10 percentof all units had rents that were above the FMR. Thus mostof the units rehabilitated to date have been affordable tomany lower income households (i.e., those WIth incomes

4The C/MI data mdIcate that for the entlre program, the mmonty share was roughly 50 percent -­one of the few instances in wInch summary statLstIcs from the sample and the natLonal C/MI totalsdIffered

II

12 Summary of the Fmal Report

between 50 and 80 percent of the local median), even in theabsence of rental assistance.

While post-rehab rents were modest, very-Iow­income households generally did need Section 8certificates or housing vouchers in order to to afford RRPunits at 30 percent of their income. Two-thirds of all post­rehab tenants received some form of rental assistance.About 82 percent of all very-Iow-income households wereassisted, compared to only about 32 percent of allhouseholds with incomes between 50 and 80 percent of thelocal median. Forty-five percent of all rental assistanceassociated with the program went to initial occupants whoremained in RRP projects; 52 percent went to residentswho were new; and 3 percent went to previous residentswho used their certificates or vouchers to move.

In general, the overall pattern ofpost-rehab rentswas consistent with the market orientation of the RentalRehab Program. While the ratio of rents to the FMR variedwith market conditions, it did not depend on the level ofrehab costs, the size of the rent increase, or the amount ofpublic subsidy. Such outcomes reflect the basic philosophyof the RRP, which lets the market -- not the localadministrative agency -- detennine project rents.

The program achieved less, however, in meetingHUD's objective of serving lower income householdsalready in place. Only about half of all post-rehab tenantshad lived in the project pnor to rehab. In part, this factreflects'the rehabilitation of vacant properties, since about25 percent of all units were in buildings that wereunoccupied at the time of rehab. However, it also reflects arelatively high rate of tenant turnover during renovation.Twenty- eight percent of all pre-rehab tenants moved out oftheir dwellmg units before the renovation was completed.

The quality of the data do not pennit one to assessthe extent of displacement that has occurred to date. Ingeneral, households that moved out of RRP projects tendedto have higher incomes than those that stayed or those thatreplaced them. While mobility rates were higher inprojects with above-average rent increases, the fact thatmovers were replaced WIth lower income households tendsto temper a displacement argument. Information onreasons for moves, while incomplete, also does not suggestdisplacement.

Nevertheless, the potential for displacementcertainly exists. About 15 percent of all very-Iow-incomehouseholds initially living in RRP projects moved andapparently were not offered assistance. This figure couldunderstate the level of mobility that actually occurred ifmoves were made before the project was formally accepted

Rental Rehab Program

into the program.S Furthermore, many sites made little, ifany, effort to contact tenants during the initial stages ofrenovation or to monitor household turnover. Suchmonitoring needs to be strengthened in order to ensure thattenants are aware of their options under the program, andguidance for such monitoring is being prepared by theDepartment.

PrivateSector Response

The Rental Rehab Program has primarily attractedindividual owners of smaller properties with relativelymodest renovation needs. The average unit was in abuilding with four or five apartments and cost about$10,000 to rehabilitate. Sixty percent of all units wereowned by individuals, with partnerships and trusts the nextmost co=on ownership forms (accounting for 17 and 13percent of the sample, respectively).

While about 17 percent of all units were acquiredspecifically for the program, the typical investor used theRRP for renovation of a previously held property. Fifty­eight'percent of project owners ind1cated that the RRPgrant or loan was the primary reason for their participation.However, the remaining 42 percent viewed the Section 8subsidy as at least as important as the rehab subsidy,especially ill loose markets.

Forty-three percent of the cost of the average unitwas supported by the RRP grant or loan; another 9 percentwas funded by other public programs; and the remaining 48percent was covered by private funds. While only 25percent of all units receIved public funding in addition to 'the RRP contribution, such units had substantially hIgherrehab costs (over $14,000 per unit) and substantially lowerleveraging ratios (each dollar of public funds was matchedby less than $0.40 in private contributions). Privatecontribunons were more or less evenly divided betweenprivate loans and owner's cash. When private loans wereused, they were generally associated with higher rehabcosts.

Since conventional leveraging ratios do not reflectthe underlying value of the different subsidy types toproject owners, the net present value of the public subsidiesprovided under the program was also calculated. Theseaveraged $3,662 per unit for RRP grants and loans and

5By its nature, pre-rehab mObility would be extremely dIfficult to document; therefore, such datacollectlon was not attempted as apart of this study.

13

14 Sununary of the Fmal Report

$4,618 per unit for all public funds combined. Theyrepresent 40 and 46 percent, respectively, of the averageproject's rehab cost, and exhibit far less variation thanconventional leveraging ratios. The net present value offorgivable deferred-payment loans ($5,506 per unit) wasabout 45 percent higher than the value of repayable loans($3,804 per unit).

Based on survey responses of owners, three-fourthsof the units would have undergone some or all of the ):ehabwork perfo=ed under the RRP, even in the absence ofassistance. However, in only one-fourth of all units wouldthe owner have completed all of the work, and in mostcases renovations on these units would have beencompleted incrementally ifRRP assistance had not beenprovided.

Analysis of project fmancial conditions conf=edthat complete renovation would not have been financiallyfeasible for most units in the absence of RRP assistance.Almost half of all units would hiLVe generated negative cashflows if the total rehab effort had 'been supported withprivate fmancing, and two-thirds would have had cash­flow-to-revenue ratios of under 10 percent. Nevertheless,subsidy levels were not effectively tailored to individualproject financial conditions, and in many cases the citiesprovided larger subsidies than the lffinimum required tomake rehab financially feaSible. While greater use of gapfmancing techniques could have helped to minimizesubsidies, sites evidently viewed this as a trade-off,preferring a straight 50-percent-of-cost approach that wasattractive to potential participants and easy to admimster.

Nature andCost of Repairs

Roughly 20 percent of the units renovated under theprogram were uninhabitable prior to rehab, 40 percent weredilapidated, and 40 percent were in need of limited repairs.The average unit had major work perfo=ed on 3.4systems, with 5 or 6 systems typically overhauled orreplaced in dilapidated and uninhabitable projects. Theelectrical, the bathroom, and the heating, ventilation andair-conditioning systems were most frequently overhauled.In about 17 percent of the projects, units were added orenlarged, apparently in response to the program's emphasison larger individual units, i.e., more bedrooms to servelarger families.

After rehab, the great majority of projects providedsound, useful housing for people of modest income. The

Rental Rehab Program

study's rehab specialists rated the quality of workmanshIpand materials as "average" m 71 percent of the units and"high" in 20 percent. Only about 9 percent had "poor"quality ratings. In this respect, the RRP clearly hassucceeded in producing work of good quality. In about 25JJercent of the units, while the local code and HQSrequ_irements of the program were met, the study'smspectors felt that additional mmor repairs would havelIDproved the marketabJlity or-operating efficiency of theunits. However, about 9 percent of the units still neededmajor repairs at the time of the site visit and wouldprobably not meet HQS. In all but one case, these umtswere the same as those which receIved a poor quality ratingfrom the study's mspectors on the work that was done.

Significantly, 70 percent of the projects in whichboth the quality and scope of work were p-eOL(.seYen sites)were m localities where owners were responsible for-preparing the work specifications. However, makmg theowner responsible for drawing up the specificationsapparently worked quite well in other communities. Theother seven of the 14 sites that streanilined in this wayproduced umformly sound finished units, and only one hada problem with an owner-expanded scope.

Allowing local flexibility in determining eligiblerepairs generally did not result in the use of funds for workother than that necessary to create sound housing for lowerincome households. Expenditures to meet HQS and localcode requirements averaged $9,445 per unit and accountedfor 94 percent of total rehab costs ($9,978). Another 5percent was spent on other improvements consideredessential to marketability and sound management, and lessthan 1 percent went for general property improvements.Costs in excess of local code requIrements or HQS wereincurred in about half of all units but were relativelymodest in amount.

In addition to the RRP expenditures reported here,36 percent of the units involved owner expenditures onrenovation work above and beyond that required by theRRP. Such expenditures averaged about $500 in thesample as a whole and about $1,800 for units whichincurred such costs. Some of the expenditures wererequired to resolve unanticipated problems with HQS, andone owner enhanced two projects (containing 1 percent ofsample program units) with improvements inconsistentwith moderate-income occupancy. For the most part,however, the expendItures resulted from the fact that manyowners perceived the need for fmishing touches after theRRP work was completed.

Rehab cost per unit was largely detemIined by therehab work required. Significant explanatory factors were

15

\16 Summary of the Fmal Report

project size, average unit size, pre-rehab condition, and thenumber of major systems overhauled. Rehab costs werehigher in projects where the owner had some prior rehabexperience. They also were higher in projects wherelocalities supplemented RRP funds with other publicmonies, either because localities offered extra subsidies tomore costly projects or because availability of deepersubsiches encouraged owners to pursue more extensiverehab jobs.

Current andLong-Term Performance

Evaluating a program in its initIal stages requiressome speculation. Furthermore, restricting the sample tosites with completed projects may well overstate orunderstate the program's accomplishments. Nevertheless,based on the experience of the 34 sites, the Rent~ RehabProgram appears to be meeting expectations.

As noted throughout, specific program outcomes ina given community are closely related to the type ofsubsidy provided, the level of rehab funded, thecharacteristics of neighborhoods in which projects arelocated, and the strength of the local housing marketHowever, no one approach or market type is clearlysuperior to all the rest ifone considers the range of programobjectIves estabhshed by HUD. While relatively few siteshave been highly successful in all performance categories,most have done well on at least a subset of the RRP'smajor goals. This variation in outcomes reflects thediversity of local conditions, priorities, and needs.

Overall, the Rental Rehab Program appears to beworking well. While initIal production has been relatIvelylow, it has accelerated in recent months, and many siteshave made adjustlnents to their programs whIch shouldimprove performance in this regard. As noted above,tenant monitoring should be strengthened. Nevertheless,the types of households being served, the initialaffordability of the rehabilitated units, and the completionof appropriate repairs all conform to established nationalobjectives.

The performance of the program will ultimately bejudged on the extent to which it continues to providehousing of good quality at an affordable price for low­income households in the years to come. While the timingof the evaluation necessarily limits our ability to assess thelong-term affordability of RRP units, our findings doprovide some indirect evidence on this issue.

Rental Rehab Program

The majority of RRP projects developed thus farhave been in neighborhoods where incomes and rents wererelatively low. For example, median family income in theaverage RRP neighborhood was only about 66 percent ofthe local median in 1980. Ninety percent of the unitsstndied had rents at or below the applicable FMR.Furthermore, roughly half of all housing units rehabilitatedunder the program had rents that were less than 90 percentof the applicable FMR, and an6ther 10 percent had rentsbetween 90 and 95 percent of the FMR. Given a relativelylow rate of inflation, such units are likely to remainaffordable over the next few years. More problematic arethe units with rents which are currently close to or abovethe FMR. Nevertheless, given the nature of RRPneighborhoods, as well as the modest quality of the unitsdeveloped thus far, even these projects would seem likelyto retain their lower income character over the next fewyears.

*u S GOVERNMENT PRINTING OFFiCE 1987-181-397/74008

17