Multifamily Predictive Analytics System
August
10, 2020
Webster Hughes, PhD, Managing Principal, Multifamily Comps
Our company Multifamily
Comps LLC (“MFC”) co-authored a recent White Paper introducing our new Multifamily
Predictive Analytics System (“MPAS”). The MPAS White Paper and can be downloaded from www.multifamilycomps.com. This article provides a brief summary
of MPAS methodology and applications.
The MFC database is created from
origination files and line-item operating statements for over 30,000 loans and
26,000 properties backing Freddie Mac K-Series Multifamily CMBS collected
monthly starting in 2009. In addition to financial data pulled from the Freddie
CMBS files, MPAS uses US Census data collected from a 1-mile radius of each
location to create a customized demographic profile for each property. The database
includes, on average, 4 years of serialized financial and demographic data for
each loan and property. This data configuration allows MFC to statistically
analyze property-level operating performance, line-item expenses, valuation
metrics, and predictive relationships across the 30,000+ loan underwritings and
over 120,000 observations of year-on-year property-level performance.
Competitive Property Set Selection
The most basic MPAS application is Competitive
Set Selection and Line-Item Financial and Demographic Comparisons to a Client
Acquisition Candidate or Portfolio Property. MPAS Comp Set Selection Algorithms
filter the MFC Database on Subtype, Distance to Client Property, Year
Built/Renovated, Median Rent, Median Household Income, and Population Per
Square Mile sourced from a 1-mile radius of each Property. Selecting Comps is
time-consuming and tedious; and poor choice of Comps provides misleading
results and invalidates analyses. Properties even within the same zip code can
have very different demographic profiles (download MPAS White Paper for
examples). MPAS Comp Set Selection Algorithms using localized Census data filters
are key system components. Clients are additionally provided the ability to
hand-select Comps from the database based on the full range of financial and
demographic data.
Line-Item Financial and Demographic
Comparisons
Once a Comp Set has been selected for a
Client Property, MPAS Algorithms perform Line-Item Financial and Demographic
Comparisons used for Acquisition Screening and Underwriting. MPAS Algorithms
also identify financial line-items for which pro-active Asset Management
provided the best opportunities to improve NOI. An application of Acquisition
Screening would be to identify that Repairs, Utilities, and Cap Ex provided in
a broker sales memorandum are too optimistic in comparison to the Comps. An application
for Asset Management would be to identify those line-items as too high versus
the Comps and therefore present an opportunity to reduce expenses. This
functionality is fully automated with results available for both Excel download
and in Tableau visualization. Example reports are available on www.multifamilycomps.com.
Property-Level Predictive Indicators
The most technologically advanced MPAS
capabilities are based on proprietary statistical analyses across the MFC
database. The mathematical objective of these statistical analyses is to
identify persistent Predictive Indicators across subtypes, metros, broader
geographic regions, historical time-periods, and 100+ independent variables. As
an example, MPAS Predictive Indicators for forward NOI growth in Charlotte, NC
Garden Apartment Market are: 1) below mean Revenue Per Unit; 2) above mean
prior-year Expense Growth; 3) above mean Repair Cost; 4) below mean Year Built;
5) above mean Percent of Population with Cash Rent over 30% of Median Household
income. The first three indicators guide an investor toward properties with
revenue upside and opportunity to reduce expenses. The last two indicators
guide investors to older properties in lower income neighborhoods. These
criteria define a Value-Add Strategy focused on older apartments in lower
income areas. MPAS Comp Set Algorithms discussed in the paragraph above automate
the data collection and analyses required to determine the degree to which an
Acquisition Candidate satisfies these criteria.
Automated Valuation and Operating
Performance Forecasts
MPAS Automated Valuation and Performance
Forecasts are generated by integrating the Property-Level Predictive Indicators
discussed in the paragraph above with National Baseline Scenarios for Revenue,
Expenses, Cap Ex and Cap Rates. Our National Baseline Scenarios assume a deep
multi-year recession with substantial variation across demographic profiles due
to uneven expected job loss, renter financial condition, and exposure to the
COVID-19 virus. In the MPAS White-Paper we describe our Demographic Risk Score
which we use to model the uneven economic impact across demographic profiles.
We regard this as one of the most important elements of forecasting in the
COVID-19 environment.
Distressed Debt
The MFC database includes origination
loan term and monthly updated payment and balance information on all loans
backing each Freddie CMBS deal. This granular Loan-Level data is integrated
with the Property-Level Valuation and Operating Performance Forecasts to
provide a
current and forward analysis of debt
coverage, loan-to-value, free cash flow, and refinancing risks. The MFC dataset
shows that as of the 6/25/2020 data release date, 1971 loans were either in
forbearance or delinquent. This up from 297 as of the 2/25/2020 data release.
MPAS Forecast Models project that 30% of loans backing Freddie Mac CMBS will
require some level of capital infusion to service debt and/or refinance over
the next 3 years. MPAS
provides comprehensive capital structure analysis for all database properties
and any Client property with available financial and loan information.
Conclusion
Log into www.multifamilycomps.com to download the MPAS White Paper,
review our Comp Set Analytics, and contact our team for more information.
No comments:
Post a Comment