Research paper

United States Housing Price Analysis

Dataset Resource: Empirical Evidence from Housing Price in

Student Name - 199807

ANLY 502, Spring 2017

Instructor: Ali Motamedi

March 4, 2017

2

Instruction

“The United States housing bubble was a real estate bubble affecting over half of the U.S.

states. Housing prices peaked in early 2006, started to decline in 2006 and 2007, and reached

new lows in 2012. On December 30, 2008, the Case-Shiller home price index reported its largest

price drop in its history. The credit crisis resulting from the bursting of the housing bubble is—

according to general consensus - the primary cause of the credit default swap bubble of the

2007–2009 recessions in the United States” (Wikipedia). At that time, this may negatively affect

housing buyers investing options and discourage home owners. In real world, real estate industry

reached its downturn and wasn’t able to recover for a long time.

This study is based on the historical dataset on empirical evidence from housing price in

Boston. Even though this dataset was collected and evaluated in 1970s, it will be a good tool to

start with. Housing price can be affected by a lot of factors. As time goes by, these factors that

drive housing price will also change due to technology, economy, and people’s purchasing

power and consumption habits change. However, by studying historical data using regression

model(s), it will be useful to apply in the real world. Most importantly, it will be useful for

irrational buyers. Here is a typical example to define irrational buyers. The reason is the same as

why people buy more at the grocery when they're hungry, the researchers say. It's impossible for

the brain to turn off what it wants in that moment. So don't go to look at swimming pools on a

hot day--chances are you'll cave. Here's a graph showing the value of houses with swimming

pools by season: (Lutz, 2012).

3

From above graph, it is obvious that if the housing searching happens in summer time,

housing hunters are more likely to buy houses with swimming pools. Therefore, price of houses

with swimming pools dramatically increases in summer time. Vice versa, in winter time, housing

price drops due to fewer buyers take swimming pools into consideration. Given supply and

demand theory, increasing demand results in price increase while decreasing demand results in

price decrease. Here, season is considered an irrational housing price driven factor. Season is

changeable instead of stable or effective factor. Housing price is driven by people’s sentiment

due to season.

In the dataset chosen in Empirical Evidence from Housing Price in Boston, thirteen factors

are chosen as effective factors to apply regression model for further analysis in theory. By the

end of this study, homeowners or homebuyers, even for investment institutions may take this

interpretation as a useful reference.

4

Factors/ Variables (in Theory) Drive Housing Price

• Crime Rate

The major reason related to crime rate could be wealth inequality in the United States.

Wealth inequality can be defined as the unequal distribution of assets among residents of the

United States. Once wealth gap gets larger and larger, people in lower class may or may not put

themselves in an unfair situation. From psychology perspective, these group people could be a

reason for crime rate increase. People in middle or upper class are intended to choose a safe

place to call it home. These group people are also a representative of strong purchasing power.

They are more willing to spend more money to buy safety, which drives the housing price in this

kind of community get higher. From this stand of point, crime rate is considered a factor

affecting house price.

• Proportion of Residential Land

Starting from ancient age, human beings have been living in groups, which enable

communication and socialization easier. Given this aspect, urban area makes residential center

and business center separate. As economy develops, business center is developed as well. It gets

shopping center, grocery stores, banks and even employment center together. As a result, human

traffic is much larger than other area, especially on weekends. However, for employees who are

busy with work during weekdays, it could be a bad idea living in area with large human traffic.

Alternatively, living in higher proportion of residential land may get popular among house

buyers or renters. Hence, higher proportion of residential land may have positive correlation with

housing price.

5

• Proportion of Non-retail Business

Similar concept to proportion of residential land, proportion of non-retail business is

another reason worthy to be considered as one of the factors. In first consideration, proportion of

non-retail business may have negative correlation with housing price. The major reason is due to

inconvenient life experience. For employees, driving a long distance to shop for grocery won’t

be favorable.

• Landscape: Charles River

Charles River here plays an important role as a factor related to environment. From personal

experience, I would put environment as a significant aspect in housing searching. In the past year,

I had been living along the riverside. Even though, it was a rental house, price there is placed

very high among other landlords. And facing river becomes an outstanding selling privilege.

During the year when I lived there, life was enjoyable. Facing and staring at computer screen all

day long, eye sight and health may be negatively affected. I would spend half hour standing in

my balcony and enjoying beautiful river landscape. My eyes may feel relaxed. Enjoy landscape

can be another factor having positive correlation to housing price.

• Nitric Oxides Concentration

Nitrogen dioxide is an irritant gas, which at high concentrations causes inflammation of the

airways. NOx mainly impacts on respiratory conditions causing inflammation of the airways at

high levels. Long term exposure can decrease lung function, increase the risk of respiratory

conditions and increases the response to allergens. NOx also contributes to the formation of fine

particles (PM) and ground level ozone, both of which are associated with adverse health effects

(2017). People nowadays are paying more and more attention to heathy problems.

6

Based on this fact of Nitric Oxides, high concentration of Nitric Oxides may have negative

correlation to housing price.

• Number of Rooms per Dwelling

With the number of rooms increasing at home, it is necessary for total areas to increase. It is

obvious that more room can accommodate more people. First, from family perspective,

American family has more kids than residents in other country, for example, Chinese family. For

kids, having their own room when they grow up may be helpful for their future development. So

when choosing house for the entire family, number of rooms per dwelling is important for

households. Second, from renter perspective, one-bedroom apartment costs less than two-

bedroom apartment.

• Age of House

Age of house could be a factor which may have medium correlation to housing price, which

could depend on house hunters’ taste. Some people like old house may because old house can

tell a story or it has its own history. Some people may like new house because it has modern

structure. People may concern that old house could not be as strong as new house. One defensive

example could be White House, which has a history of 200 years. As a result, age of house could

be a considerable factor to be validated.

• Distance to Employment Centers

Distance to employment centers is another obvious factor may have negative correlation to

housing price. Houses have farther distance to employment center might make their price

relatively cheaper than houses close to employment centers. Young people today get used to stay

up late at night instead of being morning person. It may be hard for them to get up early in the

7

morning. If they could live close to work, they can save more time in the morning for sleeping

longer. Considering people’s psychology, housing price is asked higher.

• Access to Radial Highways

Driving at highways to work is normal life of American people. Especially for large

company, they tend to locate their work site at large and open area. Therefore, driving highways

is necessary. Houses that have easy access to radial highways may price higher compared to

houses that are far away from highways.

• Property Tax Rate

Property tax rate may have direct impact on housing price. Houses have higher tax rate tend

to have higher price while houses have lower tax rate may have lower price.

• School

School is a factor that more depends on different groups of people. First, for homebuyers,

their purposes are possibly because people have decided to be a family and what’s next? It could

be education for kids. Future parents will start to think about distances to schools. Therefore,

houses closer to schools may price higher than houses far away from any schools.

• Race

Based historical data, communities where minority group lives in may or may not have

lower price than majority group lives. In the United States, distribution by race reveals as below:

(2016) White people are considered major group in the United States, which is up to 61%.

8

Houses where white people groups may price highest, but it could be higher than where black

lives.

• Status of Population

Back to first point about crime rate analyzed, status of population is a relatively important

factor. Thinking of purchasing power, people have same or similar purchasing power tend to live

in the same community because they can afford the housing price. Reversely, houses where

people in upper classes live have higher price than houses where people in lower class live.

• Number of Owner-occupied Homes

Last factor is about number of owner-occupied homes. This factor will be compared to

renters. For houses which are built for rental purposes, they are priced higher than owner-

occupied homes. On the other hand, for homes that are owner-occupied, their purposes are

different from business purposes as rental landlords.

9

Data Analysis

To deep analyze the dataset, following steps are used to generate data analysis. First, start

from correlation is the first step to prove hypothesis. Second, use ANOVA analysis to interpret

the relationship between each variable and housing price and also generate plots to visualize their

relationship. Third, validate linear model by generating plots for Residual vs. Fitted, Normal Q-

Q, Scale-Location, and Residual vs. Leverage.

Above correlation table is generated in R studio and sorted in Excel based on the dataset in

attached R codes. The above table clearly analyzes how each variables correlates to housing

price. Combining each plots in the attached R codes with above two-way table. Three variables

are considered to have relatively strong relationship with housing price. They are low-status

population, number of rooms and crime rate.

10

• Low-status Population

From above linear regression plot, we can conclude lower-status class has strong negative

correlation with housing price. The less lower-status class, the higher housing price. It verifies

the assumption at the beginning. People in upper class tend to live in the same community with

their neighbors within same social class. Bring purchasing power theory back to here again, it

results to housing price has strong relationship with social class.

• Number of Rooms per Dwelling

11

From above linear regression between number of rooms and housing price, it indicates they

have relatively strong positive relationship with each other. With number of rooms increasing,

housing price increases as well. Those outliers could be affected by other factors, such as age of

house, or long distance from workplace. This plot reflects both rentals and houses available for

sale in the market. It also shows that number of rooms around six is most popular in the market,

and it also indicates this type of houses with six rooms may be more common and have higher

availability. If number of rooms reaches over eight, the price drops down. Here we can apply

supply and demand theory to explain. Demand for houses with eight rooms can be very low in

the market. Therefore, suppliers may lower the house price to attract more buyers.

• Crime Rate

Even though, the coefficient between crime rate and housing price is only -0.39, from above

graph, we can still conclude crime rate may have medium negative correlation with housing

price. With crime rate going up, housing price drops. The graph shows area where crime rates at

zero has various housing price, which shows other factors may have impacts on housing price as

well. For example, people who have no roommate may consider lowest crime rate in the first

12

place. Under this case, one-bedroom or two-bedroom apartment will be the first option for this

group of people. Therefore, this could be one of the noises that results in low coefficient between

crime rate and housing price.

13

Model Validation

From first graph, we can tell residuals of majority data are not all around y=0. Data points

369, 372 and 373 can be treated as outliers. This indicates the dataset may not be fitted in linear

model perfectly. For the second Normal Q-Q plot, the data is in ascending order along the

straight line. We can say the residuals are normally standardized. For the third Scale-Location,

the line is not horizontal and data is not equally spread. For the last Residual vs. Leverage plot,

data points 369, 372 and 373 are the only three data between 0.5 and 1. From Cook’s distance

theory, we can conclude data points 369, 372 and 373 are influential data or noise.

14

Conclusion

From statistics perspective, no model could be perfect to interpret a dataset. However,

through this study, major factors influencing housing price are effectively interpreted. Number of

rooms, landscape, and proportion of residential land may have positive correlation with housing

price. Proportion of low status class, crime rate, distance to employment centers and air quality

(NO concentration) may have negative correlation with housing price. Furthermore, these factors

can be effective factors for rational homebuyers or even renters to take into consideration for

house searching. For homeowners, this analysis could be a good reference in evaluating their

own property and making better decision.

15

Reference

Lutz, A. (2012, July 13). Irrational Homebuyers Pay Way Too Much Attention To The Season.

Retrieved April 15, 2017, from http://www.businessinsider.com/irrational-consumers-pay-way-

too-much-attention-to-the-season-2012-7

Nitrogen Oxide (NOx) Pollution. (n.d.). Retrieved April 15, 2017, from http://www.icopal-

noxite.co.uk/nox-problem/nox-pollution.aspx

Population Distribution by Race/Ethnicity. (2016, November 18). Retrieved April 15, 2017, from

http://kff.org/other/state-indicator/distribution-by-

raceethnicity/?currentTimeframe=0&selectedRows=%7B%22wrapups%22%3A%7B%22united-

states%22%3A%7B%7D%7D%7D&sortModel=%7B%22colId%22%3A%22Location%22%2C

%22sort%22%3A%22asc%22%7D