Skip to content

ebracht/Class-Project

BranchesTags

Repository files navigation

Class Project

Checkpoint 1: Progress on Eight Major Tasks

1) Propose a research topic

Each team member proposed a thoughtful research topic along with relevant data sources. After deliberation (see meeting minutes below), our team chose to study how mortgage rates influence home ownership rates across the U.S. To study this topic, our team is constructing a panel dataset comprised of several reliable data sources including data from the U.S. Census Bureau, FRED, and BLS. We will organize our findings to Scott Turner, Secretary of Housing and Urban Development (HUD), to inform U.S. housing policy decisions.

2) Create a GitHub repository and establish best practices for team collaboration

We have created a GitHub repository. Each member has made substantial contributions. Thus far, we have included the following in this repository: the project topic, meeting minutes, information on the packages and API keys needed for analysis, code to import relevant data, summary statistics for key variables, and several helpful visualizations.

3) Demonstrate merging of multiple data sources

To construct our panel dataset, we have imported, cleaned, and merged several data sources: population and income income data from the Census, home ownership data from the Census, the housing price index, unemployment data from BLS, and mortgage data from FRED.

4) Visualize data using Tableau, R, Python, or a combination

We have used R to create several figures, including line graphs and histograms to accompany the summary statistics.

5) Generate meaningful summary statistic (KPIs) of the data

We have generated summary statistics for the key variables we intend to analyze.

6) Submit draft of progress at Checkpoint 1 and Checkpoint 2

This branch is our Checkpoint 1 Submission.

7) Summarize your findings in a short video presentation

We will make progress on this in the coming weeks.

8) Publish a detailed, well formatted markdown report of your analytical story to your GitHub repository

We have posted this markdown file to our landing page. This will become the basis of our report.

Research Topic

Our proposed research question is: How have changes in mortgage interest rates and housing prices affected home ownership rates across U.S. states over time?

To study this, we plan to build a panel data set that examines each state-year on home ownership rate, housing affordability, mortgage rates, and other relevant variables like median income or unemployment rate. Our primary data sources will be the U.S. Census Bureau and the Federal Reserve Economic Data (FRED), which includes state home ownership rates and the 30-year fixed mortgage rate. Additionally, the Federal Housing Finance Agency’s House Price Index, which provides state-level data on house price changes, and the Bureau of Labor Statistics, which has state-level data on unemployment, will be pulled. Together, these sources will provide panel data that can be used to examine how rising mortgage rates and increasing prices affect home ownership, while allowing room for controls like unemployment, median income, or population, to understand this effect better.

Once we have established the relationship between mortgage interest rates, housing affordability, and home ownership rates, we will consider what policy options or levers are available to improve home ownership rates. We may do this by conducting additional primary analysis to test the efficacy of certain policies where data are available, or by consulting the relevant literature to form reasonable conclusions about the linkages between various housing policies and mortgage rates/affordability. Once we have identified policy recommendations, we will present these to the U.S. Secretary of Housing and Urban Development, Scott Turner.

The Secretary of Housing and Urban Development (HUD) manages the Department of Housing and Urban Development, with responsibilities including the creation and oversight of affordable housing policies and programs and advising the President on housing issues. As our ultimate aim is to contribute to improving home ownership rates, we will explore the various economic indicators that may positively or negatively impact this, strategically analyzing these relationships in order to make predictions and recommendations for affecting future policy change. Secretary Turner has the power to act on these recommendations, with his actions guided by a goal of creating affordable housing policy. Our work exploring home ownership rates is well-aligned not only with this goal, but also Turner’s previous efforts on opportunity zones. Thus, Scott Turner is the key figure whom we will focus the implications of our findings toward. We plan to create these visualizations using a mix of software programs (including R and Tableau) to create easily and quickly digestible visualizations that best inform and support our recommendations.

Sources:

Meeting Minutes and Personal Contributions

Meeting 1 (02/25):

Initial meeting of Liz, Ryan, and Levi; we decided on a weekly meeting time, went through the requirements for the project, and established a shared document. We planned to each pitch an idea (i.e., provide a brief description and data source links) in advance of our next meeting, at which we hoped to narrow in on a topic.

Meeting 2 (03/04):

Ideas

  • Levi: Minimum Wage/Food Security/Poverty Programs
    • Description: The Center for Poverty Research compiled an impressive and easy-to-use dataset with information on state policies such as the minimum wage, SNAP, EITC, and so on. I propose using this dataset to measure whether/how these policies have influenced variables of interest such as poverty, employment, or food insecurity. The data set is fairly comprehensive (part of the appeal), but I think it would be easy to find opportunities to supplement it with other datasets. For example, the food insecurity data in the Center for Poverty Research dataset only goes up to 2001. If we wanted to look at how state economic policies have influenced food insecurity over the past two decades, we would need to pull in another dataset. I just filled out an online form offered by Feeding America to request access to their food insecurity data (available at the state level). They agreed to share it with us, so that’s one option. I can think of plenty of other options, too. For example, we could pull in Census data (I have experience scraping Census data with R using an API key) to analyze whether these policies influenced the ratio of people that own vs rent their homes. Generally speaking, these analyses would require rigorous controls, so there would be ample opportunity to pull in multiple datasets to try and control for other factors that may influence these endpoints.
    • Potential data sources:
      • National Welfare Data: “The Center for Poverty Research annually updates our state-level panel data series covering population, employment, unemployment, welfare, poverty, and politics. Our current update includes information for the majority of the 2024 calendar year. We will update the remainder when available. These data are publicly available to all users.”
      • Link: https://ukcpr.uky.edu/resources
      • Feeding America Food Insecurity Data: “Since 2011, Feeding America has produced the Map the Meal Gap study, providing estimates of local food insecurity and food costs on an annual basis to better understand people and places facing hunger and to inform decisions and actions that will help us achieve our mission. We do this by generating national and local data about food insecurity, translating those data into insights and tools like the interactive map below, and engaging partners to help them use and improve our data and research in the future.”
      • Link: https://map.feedingamerica.org/
  • Liz: Accessibility to alcohol’s relationship to levels of binge drinking in a given (Iowa) county
    • Description: The Iowa Department of Health and Human Services has already launched a new campaign called Say “Yes” to Drinking Less Alcohol, to combat high rates of binge drinking. Aiming to investigate whether factors like accessibility to alcohol are correlated to higher levels of binge drinking could inform where to best focus future resources and interventions. Targeted campaigns and/or subsequent action by Iowa Department of Health and Human Services, in conjunction with community partners, in areas that show higher levels of average binge drinking could aid in the ultimate goal of improving health, safety, and alcohol responsibility in Iowa. We could compare the rate across states, explore whether/how the severity has changed over time, and look into the local data from class to compare behavioral data to accessibility (e.g., looking at number of stores in a given area) and purchase rates at a county level.
    • Potential data sources:
  • Ryan: Housing Costs, Interest Rates, and home ownership in the U.S.
    • Description: My idea is to study how mortgage interest rates and housing prices affect home ownership rates across states. We could combine Census housing data, FRED mortgage rates, and FHFA house price indexes to build a panel dataset and estimate how changes in borrowing costs and housing prices influence home ownership or rent burden over time. One potential research question would be: How do changes in mortgage interest rates and housing prices affect home ownership rates across U.S. states over time? Data should be relatively easy to access and merge. Housing outcomes are influenced by many economic factors, so we could add controls like median income, unemployment rates, or population growth to isolate the effect of interest rates and housing prices. We could explore regional differences, like if changes in interest rates affect home ownership differently in high-cost housing markets compared to more affordable states. We’d probably create a panel dataset that observes states over time, so each state-year would be an observation. The panel would track how home ownership rates change as mortgage rates, housing prices, and other economic variables change.
    • Potential data sources:

Discussion: After talking through the ideas, Levi pointed out that Ryan’s idea had overlap with the Federal Reserve data that we had to use for R HW 3. We agreed to proceed with that topic and aimed to think about it in parallel with the homework. We each tried to merge data sets to make sure we had the skill to do so moving forward.

Meeting 3 (03/11)

We finalized our decision to focus on Ryan’s proposed topic, and prepared for our topic submission due 03/13. In preparation, we split up the work as such:

  • Ryan will submit the topic on Canvas course page
  • All will work on respective sections
    • Levi: Process of choosing a topic, opportunities to expand.
    • Liz: Policy implications Scott Turner HUD
    • Ryan: Topic (research question, data sources)
  • Levi will inform Bangjun (former group member) of the plan to log notes in this document rather than sending weekly emails
  • Liz will create a GitHub page, will add everyone as collaborators & email Prof Chale with a status update about a potential issue
  • Ryan will upload a file joining Census and FRED data, including a loop to bring in multiple years of Census data
  • Levi will investigate data availability for controlling for state housing policy

Meeting 4 (03/18)

We talked through the Checkpoint 1 criteria and made the following plan in advance of Meeting 5:

  • Work on README/update on Github (update on personal contributions, add progress updates for each checkpoint, etc.)
  • Try to run code that someone else posted on github and become more familiar with platform/collaborative repositories
  • If possible, try to create simple visualizations
  • Date to keep in mind: April 14th (project draft is due)

Meeting 5 (03/25)

We discussed our progress on using our collaborative github repo. We also brainstormed ideas about additional potential directions in response to the feedback we received on our topic submission earlier in the week. We then divided up remaining tasks necessary for the Checkpoint 1 submission as follows:

  • Liz: Add remaining (Meeting 5) updates; fix formatting (adjust headers, resolve spacing issues, add bullets, etc.); clean up/formalize wording of notes where necessary
  • Levi: Fill in progress on 8 major tasks and submission
  • Ryan: Create histogram plots of variable and move to README

In advance of Meeting 6, we plan to look into:

  • Potential variables to control for like state housing policies (explore law atlas data sets)
  • How home ownership impacts wealth for households, per topic submission feedback
  • Lags in mortgage and home ownership rates

Load Packages, Setup API Keys, Import Data

Load Required Packages:

  • tidyverse
  • janitor
  • readr
  • readxl
  • tidycensus
  • fredr
  • lubridate
  • tsibble
  • fpp3

Set API Keys:

Import online data:

#Set years to include the last two decades, excluding 2020 due to the COVID-19 pandemic.
years <- c(2005:2019, 2021:2024)

#Extract home ownership data from the Census
homeownership <- map_dfr(
  years,
  function(yr) {
    get_acs(
      geography = "state",
      variables = c(
        total_occupied = "B25003_001",
        owner_occupied = "B25003_002"
      ),
      year = yr,
      survey = "acs1"
    ) %>%
      select(NAME, variable, estimate) %>%
      pivot_wider(names_from = variable, values_from = estimate) %>%
      mutate(
        year = yr,
        homeownership_rate = 100 * owner_occupied / total_occupied
      ) %>%
      transmute(
        state = NAME,
        year,
        homeownership_rate
      )
  }
)

#Extract income data from the Census
income_list <- map(
  years,
  ~ get_acs(
      geography = "state",
      variables = "B19013_001",
      year = .x,
      survey = "acs1"
    ) %>%
      transmute(
        state = NAME,
        year = .x,
        median_income = estimate
      )
)

#Bind the income data into a DF
income <- bind_rows(income_list)

#Extract population data from the Census
population_list <- map(
  years,
  ~ get_acs(
      geography = "state",
      variables = "B01003_001",
      year = .x,
      survey = "acs1"
    ) %>%
      transmute(
        state = NAME,
        year = .x,
        population = estimate
      )
)

#Bind the population data into a DF
population <- bind_rows(population_list)

#Extract mortgage data from FRED
mortgage_raw <- fredr(
  series_id = "MORTGAGE30US",
  observation_start = as.Date("2005-01-01")
)

#Estimate annual mortgage rates
mortgage_annual <- mortgage_raw %>%
  mutate(year = year(date)) %>%
  group_by(year) %>%
  summarise(
    mortgage_rate = mean(value, na.rm = TRUE),
    .groups = "drop"
  )

Import local data:

# Import Local Data
fhfa_raw <- read_csv("DaAn Midterm/hpi_at_state (1).csv", col_names = FALSE) %>%
  setNames(c("state", "year", "quarter", "hpi"))

# Set HPI data by state/year
fhfa_annual <- fhfa_raw %>%
  clean_names() %>%
  filter(year >= 2005) %>%
  group_by(state, year) %>%
  summarise(
    hpi = mean(hpi, na.rm = TRUE),
    .groups = "drop"
  )

hpi_annual <- fhfa_annual %>%
  mutate(state = state.name[match(state, state.abb)])

# Import local unemployment data
bls_data <- read_delim("DaAn Midterm/la.data.2.AllStatesU.txt", delim = "\t") %>%
  clean_names()

bls_series <- read_delim("DaAn Midterm/la.series.txt", delim = "\t") %>%
  clean_names()

bls_area <- read_delim("DaAn Midterm/la.area", delim = "\t") %>%
  clean_names()

# Join the actual data to the series
bls_merged <- bls_data %>%
  left_join(bls_series, by = "series_id")

# Use key to select code
bls_merged <- bls_merged %>%
  left_join(
    bls_area %>% select(area_code, area_text),
    by = "area_code"
  )

# Keep annual-average unemployment rate
state_unemployment <- bls_merged %>%
  filter(period == "M13") %>%          # annual average
  filter(measure_code == "03") %>%     # unemployment rate
  transmute(
    state = area_text,
    year = as.integer(year),
    unemployment_rate = as.numeric(value)
  ) %>%
  filter(state %in% c(state.name, "District of Columbia")) %>%
  arrange(state, year)

Merge Data

#Merge the data sets by year and state
merged <- left_join(
  population, 
  income, 
  by=c("state", "year")) %>%
  left_join(
    homeownership,
    by=c("state", "year")) %>%
      left_join(
        hpi_annual,
        by=c("state", "year")) %>%
          left_join(
            state_unemployment, 
            by = c("state", "year")) %>%
              left_join(
                mortgage_annual,
                by="year")

Compute Summary Statistics

Compute population summary statistics:

## Summary statistics for population:

##     Min.  1st Qu.   Median     Mean  3rd Qu.     Max. 
##   495226  1773866  4236748  6177740  7053887 39557045

## Standard Deviation: 6964177

## Variance: 4.849976e+13

Compute median income summary statistics:

## Summary statistics for median income:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   17184   47153   55609   57792   66969  109707

## Standard Deviation: 14977.34

## Variance: 224320711

Compute home ownership rate summary statistics:

## Summary statistics for homeownership rate:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   39.12   64.78   67.47   66.54   69.94   76.30

## Standard Deviation: 5.580628

## Variance: 31.14341

Compute mortgage rate summary statistics:

## Summary statistics for mortgage rate:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   2.958   3.936   4.545   4.864   6.027   6.807

## Standard Deviation: 1.152178

## Variance: 1.327514

Compute HPI summary statistics

## Summary statistics for HPI:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##   183.5   286.2   365.8   406.6   473.6  1229.7      38

## Standard Deviation: NA

## Variance: NA

Compute Unemployment summary statistics

## Summary statistics for Unemployment:

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##   1.800   3.700   4.800   5.294   6.500  13.500      19

## Standard Deviation: NA

## Variance: NA

Create Visualizations

Plot histograms of key variables:

Plot mortgage rate and national average home ownership rate:

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors