REALPRICE
CRITICAL ANALYSIS
03/28/2019
1.
REALPRICE
Price your real estate right
2
3
Table of contents
→ Market Analysis
→ Regression Analysis
→ Excel
→ R
→ Final Model
→ Limitations
→ Business Model
→ Opportunity
→ Value Proposition
Iowa State University, Ames
4
Executive Summary
→ Consider different house features to develop a model that predicts a fair price,
based on market and regression analysis
→ Create a transparent model to be implemented in the real estate asset
management market
→ Use this as a basis for comprehensive calculations and expand the offering to
other cities and regions across the US
Reiman Gardens, Ames
2.
Market Analysis
United States vs. Ames, Iowa
5
Developments since recession
→ Relatively relaxed lending conditions
and rising employment levels have
resulted in further investment in real
estate sector.
→ Housing prices have seen an increase
over recent years (see graph)
The real estate market:
United States
6* Compared to 4.7% between 2014 and 2019
Outlook
→ With anticipated economic growth, Federal Reserve will be forced to raise interest
over the next years, which will increase borrowing costs that in turn influence
demand for homeownership.
→ Sets potential limits to industry growth over the next five years (expected to grow
a lower pace: 2.3% between 2019 and 2024*)
Developments in 2018 (Iowa)*
→ Increase in median sales price
to USD 161,000 (↑3.9%)
→ Decrease in number of homes
sold to 41,387 (↓0.9%)
→ On average, houses were 64
days on the market (↓16.9%)
→ Decrease in available inventory
of single-family detached
homes, however, offset by
townhouse-condo properties
The real estate market:
Ames, Iowa
Differentiating factors for Ames
→ Demand for housing remained
steady even after recession in
2008, due to stable job market
(especially through ISU)
→ Decrease in construction labor
force (left the area after the
recession)
→ Today: Lack of available
building space increases prices
7* Compared to 2017
3.
Regression Model
Excel & R
8
Allocation of variables
→ 72 out of 80 variables were
grouped into 11 categories:
→ Location (7)
→ House construct (8)
→ Utilities (1)
→ House Style (6)
→ Date (2)
→ Exterior Detail (4)
→ Basement (11)
→ Floor Measurement (7)
→ Garage (8)
→ House Amenities (12)
→ House Features (6)
Regressions in Excel
→ Variables with high p-values nor
signs that did not make sense
(variables of scales with negative
sign) were deleted
→ Several regressions produced a
very low R² with less than 50%
(some were still kept due to low p-
values and correct signs of
coefficients).
→ Mixing the variables from
different categories in further
regressions resulted in a final
outcome of 26 variables.
9
10
Considering this final set of
26 variables, further
regressions were run in “R”:
→ Primary regression
resulted in 0.80 R² but
contained further bad p-
values
→ Continuous reduction of
further variables down
to 16 final variables for
the regression model
Regressions in R
11
Remaining 16 variables and
forecasting model
Close Downtown
OverallQual
YearRemodel
MasVnrArea
ExternalQuality
Exposure
BsmtType1
TotalBsmtSF
1stFlrSF
2ndFlrSF
BsmtFullBath
KitchenQualNum
TotRmsAbvGrd
Fireplaces
GarageArea
WoodDeckSF
House Price (y) = -137,424 + 4,186xCDowntown + 14,221xOverallQ + 3,576xYearremodel +
30xMasVnrArea + 14,986xExternalQuality + 6,876xExposure + 6,552xBsmt +
12xTBsmtSF + 45x1stFlSF + 39x2stFlSF + 7,674xBsmFullBath + 11,493xKitchenQualN+
2,024xTRoomsAbvG + 8,510xFireplaces + 34xGarageArea + 26xWoodDeckSF
12
→ 0.81 R² (i.e. 81% of change in the price
can be explained by changes in the
variables)
→ All p-values are very low (i.e. there is
a statistically significant relationship
between these variables and the
price)
→ Residual plots showed no patterns
→ The highest coefficients are derived
for quality-related variables, showing
their importance as determining
factors for individuals that consider
buying a house.
→ Checking the available data for all
1,460 houses, the model missed the
price by about 2% on average with a
standard deviation of 21%.
Regression Analysis
→ Two test approaches
→ 80% bottom vs. 20% top and 80% / 20%
randomly divided through “R” into train
and test
→ for all regressions:
R² = 0.79, low p-values
→ for all forecasts:
average differences of 0.2% and
standard deviation close to 19%
→ Both approaches delivered very close
results to determine whether the model
is overfitted.
→ It can thus be inferred that the model is
useful to be applied in the prediction of
house price.
13* Compared to 2017
Overfitting Test
→ Number of bathrooms vs.
differentiation of bathroom types
→ Model does not consider exterior
amenities like pool or barbecue
place (not enough data)
→ Sample size of 1,460 is very small
14
Limitations
Example 1
→ Listed price: USD 185,000
→ Difference: -0.88%
15
Examples
Example 2
→ Listed price: USD 215,000
→ Difference: 0.49%
3.
Business Model
Real Estate Asset Management
16
17
→ Mainly targeted to real estate asset management market (mainly individuals and
businesses that invest in, develop or carry out transactions in real estate).
→ Recent increases in housing prices has driven a rising demand for real estate
services.
→ Low interest rates on the US market as well as company’s and consumers’ rising
confidence has created further growth in the residential and non-residential
markets.
Business opportunity
18
Value proposition
→ Offer new feature for Bloomberg Terminal
→ help asset managers in pricing houses they
consider for investments
→ determine additional value created through
remodeling or expansion and its influence on a
property’s selling price.
→ Major client will be Bloomberg L.P., which owns
the data driven platform, providing 24-hours
financial information of pricing data, financials and
news, including similar pricing developments in
relation to the stock market.
→ A new equation of the model will be released
every week, since the real estate market faces
slower fluctuations than other markets.
→ Estimated Revenue: USD 500,000
19
→ Real estate brokers currently rely mostly on average square feet and their personal
instinct on demand and supply
→ New feature will provide another screen to showcase historic data, graphs, percentage
changes, cities comparisons and the possibility to combine variables according to
individual needs within a specific time period
→ Real estate asset manager will gather necessary information to determine the optimum
selling price and receive a recommendation, based on individual factors.
→ Provide a valuable opportunity to the industry to create a common standard of pricing
evaluation and generate fair treatment on the market.
Changing the industry…
Thank you!
TEAM 6
Joao Victor Maciel
Lucy Hernandez
Monika Berber
Shresth Sethi
Ursula Jansen
20

Real Estate Pricing Analysis – Ames, Iowa

  • 1.
  • 2.
  • 3.
    3 Table of contents →Market Analysis → Regression Analysis → Excel → R → Final Model → Limitations → Business Model → Opportunity → Value Proposition Iowa State University, Ames
  • 4.
    4 Executive Summary → Considerdifferent house features to develop a model that predicts a fair price, based on market and regression analysis → Create a transparent model to be implemented in the real estate asset management market → Use this as a basis for comprehensive calculations and expand the offering to other cities and regions across the US Reiman Gardens, Ames
  • 5.
  • 6.
    Developments since recession →Relatively relaxed lending conditions and rising employment levels have resulted in further investment in real estate sector. → Housing prices have seen an increase over recent years (see graph) The real estate market: United States 6* Compared to 4.7% between 2014 and 2019 Outlook → With anticipated economic growth, Federal Reserve will be forced to raise interest over the next years, which will increase borrowing costs that in turn influence demand for homeownership. → Sets potential limits to industry growth over the next five years (expected to grow a lower pace: 2.3% between 2019 and 2024*)
  • 7.
    Developments in 2018(Iowa)* → Increase in median sales price to USD 161,000 (↑3.9%) → Decrease in number of homes sold to 41,387 (↓0.9%) → On average, houses were 64 days on the market (↓16.9%) → Decrease in available inventory of single-family detached homes, however, offset by townhouse-condo properties The real estate market: Ames, Iowa Differentiating factors for Ames → Demand for housing remained steady even after recession in 2008, due to stable job market (especially through ISU) → Decrease in construction labor force (left the area after the recession) → Today: Lack of available building space increases prices 7* Compared to 2017
  • 8.
  • 9.
    Allocation of variables →72 out of 80 variables were grouped into 11 categories: → Location (7) → House construct (8) → Utilities (1) → House Style (6) → Date (2) → Exterior Detail (4) → Basement (11) → Floor Measurement (7) → Garage (8) → House Amenities (12) → House Features (6) Regressions in Excel → Variables with high p-values nor signs that did not make sense (variables of scales with negative sign) were deleted → Several regressions produced a very low R² with less than 50% (some were still kept due to low p- values and correct signs of coefficients). → Mixing the variables from different categories in further regressions resulted in a final outcome of 26 variables. 9
  • 10.
    10 Considering this finalset of 26 variables, further regressions were run in “R”: → Primary regression resulted in 0.80 R² but contained further bad p- values → Continuous reduction of further variables down to 16 final variables for the regression model Regressions in R
  • 11.
    11 Remaining 16 variablesand forecasting model Close Downtown OverallQual YearRemodel MasVnrArea ExternalQuality Exposure BsmtType1 TotalBsmtSF 1stFlrSF 2ndFlrSF BsmtFullBath KitchenQualNum TotRmsAbvGrd Fireplaces GarageArea WoodDeckSF House Price (y) = -137,424 + 4,186xCDowntown + 14,221xOverallQ + 3,576xYearremodel + 30xMasVnrArea + 14,986xExternalQuality + 6,876xExposure + 6,552xBsmt + 12xTBsmtSF + 45x1stFlSF + 39x2stFlSF + 7,674xBsmFullBath + 11,493xKitchenQualN+ 2,024xTRoomsAbvG + 8,510xFireplaces + 34xGarageArea + 26xWoodDeckSF
  • 12.
    12 → 0.81 R²(i.e. 81% of change in the price can be explained by changes in the variables) → All p-values are very low (i.e. there is a statistically significant relationship between these variables and the price) → Residual plots showed no patterns → The highest coefficients are derived for quality-related variables, showing their importance as determining factors for individuals that consider buying a house. → Checking the available data for all 1,460 houses, the model missed the price by about 2% on average with a standard deviation of 21%. Regression Analysis
  • 13.
    → Two testapproaches → 80% bottom vs. 20% top and 80% / 20% randomly divided through “R” into train and test → for all regressions: R² = 0.79, low p-values → for all forecasts: average differences of 0.2% and standard deviation close to 19% → Both approaches delivered very close results to determine whether the model is overfitted. → It can thus be inferred that the model is useful to be applied in the prediction of house price. 13* Compared to 2017 Overfitting Test
  • 14.
    → Number ofbathrooms vs. differentiation of bathroom types → Model does not consider exterior amenities like pool or barbecue place (not enough data) → Sample size of 1,460 is very small 14 Limitations
  • 15.
    Example 1 → Listedprice: USD 185,000 → Difference: -0.88% 15 Examples Example 2 → Listed price: USD 215,000 → Difference: 0.49%
  • 16.
    3. Business Model Real EstateAsset Management 16
  • 17.
    17 → Mainly targetedto real estate asset management market (mainly individuals and businesses that invest in, develop or carry out transactions in real estate). → Recent increases in housing prices has driven a rising demand for real estate services. → Low interest rates on the US market as well as company’s and consumers’ rising confidence has created further growth in the residential and non-residential markets. Business opportunity
  • 18.
    18 Value proposition → Offernew feature for Bloomberg Terminal → help asset managers in pricing houses they consider for investments → determine additional value created through remodeling or expansion and its influence on a property’s selling price. → Major client will be Bloomberg L.P., which owns the data driven platform, providing 24-hours financial information of pricing data, financials and news, including similar pricing developments in relation to the stock market. → A new equation of the model will be released every week, since the real estate market faces slower fluctuations than other markets. → Estimated Revenue: USD 500,000
  • 19.
    19 → Real estatebrokers currently rely mostly on average square feet and their personal instinct on demand and supply → New feature will provide another screen to showcase historic data, graphs, percentage changes, cities comparisons and the possibility to combine variables according to individual needs within a specific time period → Real estate asset manager will gather necessary information to determine the optimum selling price and receive a recommendation, based on individual factors. → Provide a valuable opportunity to the industry to create a common standard of pricing evaluation and generate fair treatment on the market. Changing the industry…
  • 20.
    Thank you! TEAM 6 JoaoVictor Maciel Lucy Hernandez Monika Berber Shresth Sethi Ursula Jansen 20