Skip to main content

The Housing Crisis Through Building Permits

DS 350 · Data Wrangling & Visualization

Mapped the change in single-family building permits from 2006 to 2010 across the U.S. and California counties.
R
Maps
sf
Author

Spencer Payne

Overview

How hard did the 2008 housing crisis hit new construction? Using U.S. Census building permit data, I calculated the percent change in single-family permits from 2006 to 2010. I mapped it by state with sf and then zoomed into California counties, using a diverging color scale so that declines and increases stand out.

Show the code 
library(tidyverse)
library(sf)
library(USAboundaries)
library(geofacet)
library(scales)
library(viridis)
library(buildings)

data("permits")
Show the code 
state_permits <- permits %>%
  filter(variable == "Single Family")
Show the code 
california_county_permits <- permits %>%
  filter(variable == "Single Family") %>%
  filter(StateAbbr == "CA")
Show the code 
us_states <- us_states() %>%
  filter(stusps != "PR" & stusps != "HI" & stusps != "AK")
ca_counties <- us_counties(states = "California")
Show the code 
state_permits <- state_permits %>%
  mutate(state_fips = str_pad(state, width = 2, pad = "0"))

california_county_permits <- california_county_permits %>%
  mutate(state_fips = str_pad(state, width = 2, pad = "0")) %>% 
  mutate(county_fips = str_pad(county, width = 3, pad = "0")) %>% 
  mutate(full_county_fips = paste0(state_fips, county_fips))
Show the code 
state_permits_map <- us_states %>%
  mutate(state_fips = statefp) %>%
  left_join(state_permits, by = "state_fips")

ca_county_permits_map <- ca_counties %>%
  mutate(full_county_fips = geoid) %>%
  left_join(california_county_permits, by = "full_county_fips")
Show the code 
state_change <- state_permits %>%
  group_by(state_fips, StateAbbr, year) %>% 
  summarize(total_permits = sum(value, na.rm = TRUE), .groups = "drop") %>%
  filter(year %in% c(2006, 2010)) %>%
  pivot_wider(names_from = year, values_from = total_permits, 
    names_prefix = "permits_") %>%
  mutate(percent_change = (permits_2010 - permits_2006) / permits_2006)
Show the code 
state_change_map <- us_states %>%
  mutate(state_fips = statefp) %>%
  left_join(state_change, by = "state_fips")
Show the code 
ggplot(state_change_map) +
  geom_sf(aes(fill = percent_change), color = "white", linewidth = 0.2) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue", midpoint = 0,
    labels = percent_format()) +
  labs(title = "Single Family Building Permits Nationwide",
    subtitle = "% change from '06 to '10",
    fill = "% change") +  
  theme_classic()

Show the code 
ca_county_change <- california_county_permits %>%
  group_by(full_county_fips, year) %>% 
  summarize(total_permits = sum(value, na.rm = TRUE), .groups = "drop") %>%
  filter(year %in% c(2006, 2010)) %>%
  pivot_wider(names_from = year, values_from = total_permits, 
    names_prefix = "permits_") %>%
  mutate(percent_change = (permits_2010 - permits_2006) / permits_2006)
Show the code 
ca_county_change_map <- ca_counties %>%
  mutate(full_county_fips = geoid) %>%
  left_join(ca_county_change, by = "full_county_fips")
Show the code 
ggplot(ca_county_change_map) +
  geom_sf(aes(fill = percent_change), color = "white", linewidth = 0.2) +
  scale_fill_gradient2(low = "red", mid = "white", high = "blue", midpoint = 0, 
    labels = percent_format()) +
  labs(title = "Single Family Building Permits by County in CA",
    subtitle = "% change from '06 to '10",
    fill = "% change") +
  theme_classic()

These graphics show the trend of single family building permits in the US and California specifically from 2006-2010 following and after the mortgage crisis. In each graph we can observe that the largest majority of geographical areas at both the state and national level fell by a large margin, with only very specific and select places not falling too much or increasing in one particular area.

In organizing this data I chose to focus on California because it is my home state and because it has San Jose, California which is a huge place for Silicon Valley which seemed to do alright despite the crisis. In filtering the data I only included the continental united states. I also wanted to emphasize the the negative and positive trends over everything else so that’s why no change is a neutral white.

Back to top