Descriptive Statistics

Flood Factor scores, recent residential sales, and FEMA individual assistance claims

Purpose

This page documents several descriptive statistics used in the dissertation text. The calculations summarize:

  1. the number and share of residential properties with elevated First Street Foundation Flood Factor scores;
  2. the number of high-risk residential properties that sold after Flood Factor information became publicly available;
  3. the number of FEMA individual assistance claims associated with selected Cook County flood disasters, by flood insurance status.

This page is designed as a transparent calculation record. The code is shown so readers can see how each statistic was produced.

Data sources

The calculations below use the following files.

File Description Source note
data/raw/parcel_universe_currentyear.csv Parcel universe file used to define the broader parcel population. Downloaded from Cook County Data website
data/processed/floodfactor_scores.csv First Street Foundation Flood Factor scores joined to parcel identifiers. The acquisition script uses the CCAO parcel API and describes a 2019 First Street snapshot
data/raw/residential_pins_ever.csv Residential parcel identifiers and property class information. Pulled residential PINs within ptaxsim that were residential for at least one year. (Check class used for pull)
data/processed/delete_probably/res_sales.rds Processed residential sales file. Existing processed sales input
data/processed/City_Name_Mapping.csv Crosswalk from raw municipality names to cleaned municipality names. Existing municipality-name crosswalk
data/raw/indiv_assistance_V2_CookCounty.csv FEMA Individual Assistance records for Cook County. Existing FEMA Individual Assistance export

Dissertation page 2: coverage calculation and provenance

Page 2 compares the residential share in FEMA SFHAs with the share assigned major flood risk by First Street. The existing calculation below counts class-200 records with Flood Factor scores of at least 5 and divides by the number of distinct (pin, class) records in the 2023 residential file. It is a population calculation, not a share of sales or of the Q1 regression sample. Missing Flood Factor scores do not enter the numerator but remain in the denominator.

scripts/pull_distinct_resPINs.R identifies the residential file as a PTAXSIM extraction covering 2006-2023, using the local ptaxsim-2023.0.0.db database. scripts/pull_floodfactor_pin_list.R identifies the scores as a 2019 Flood Factor snapshot retrieved through the Cook County Assessor parcel API (nj4t-kc8j) and writes the processed score file after duplicate handling. Those script descriptions establish the intended sources; they do not independently verify the vintage of every current CSV or record its actual download date.

Open the Flood Factor calculation. Source-file notes.

Residential parcels and Flood Factor scores

The first step loads the parcel universe, First Street Foundation Flood Factor scores, and residential parcel identifiers. Residential parcels are restricted to the 2023 residential PIN file and then joined to the Flood Factor score data.

Code
puniverse <- read_csv("data/raw/parcel_universe_currentyear.csv")

ff_scores <- read_csv("data/processed/floodfactor_scores.csv")

res_pins <- read_csv("data/raw/residential_pins_ever.csv") |>
  filter(year == 2023) |>
  distinct(pin, class)


res_sales <- read_rds("data/processed/delete_probably/res_sales.rds") |>
  group_by(pin) |>
  summarize(year = max(year), .groups = "drop")

recent_res_sales <- res_sales |>
  filter(year > 2019)

pins <- left_join(res_pins, ff_scores)

Residential parcels with high Flood Factor scores

This calculation counts residential class 200 properties with a Flood Factor score greater than 4. In this file, env_flood_fs_factor > 4 corresponds to Flood Factor scores of 5 or higher.

Code
high_ff_residential_count <- pins |>
  filter(env_flood_fs_factor > 4, class > 199, class < 300) |>
  summarize(n = n()) |>
  pull(n)

high_ff_residential_count 
[1] 198309

The share of residential parcels with high Flood Factor scores is calculated as follows.

Code
share_high_ff_residential <- high_ff_residential_count / nrow(res_pins)

share_high_ff_residential
[1] 0.1236108
Code
percent(share_high_ff_residential, accuracy = 0.1)
[1] "12.4%"

Numerator, denominator, and matching checks

These diagnostics expose the units used by the existing calculation. A difference between records and unique PINs indicates that a parcel can contribute more than once. Duplicate score matches or multiple classes per PIN should be resolved against the intended historical definition before changing the reported percentage.

Code
coverage_counts <- tibble::tibble(
  Quantity = c("Residential denominator records (distinct PIN/class, year 2023)",
               "Unique PINs in the residential denominator", "Joined score records",
               "Joined records with missing Flood Factor score",
               "High-risk class-200 numerator records", "Existing high-risk share"),
  Value = c(as.character(nrow(res_pins)),
            as.character(dplyr::n_distinct(res_pins$pin)),
            as.character(nrow(pins)),
            as.character(sum(is.na(pins$env_flood_fs_factor))),
            as.character(high_ff_residential_count),
            scales::percent(share_high_ff_residential, accuracy = 0.1))
)
knitr::kable(coverage_counts)
Quantity Value
Residential denominator records (distinct PIN/class, year 2023) 1604301
Unique PINs in the residential denominator 1604301
Joined score records 1604301
Joined records with missing Flood Factor score 13360
High-risk class-200 numerator records 198309
Existing high-risk share 12.4%

Recent sales among high Flood Factor properties

This section identifies high Flood Factor properties that sold after 2019. This helps describe the number of properties for which flood risk information may have been capitalized into transaction prices after the public release of Flood Factor information.

Code
high_ff_recent_sales <- pins |>
  filter(env_flood_fs_factor > 4, pin %in% recent_res_sales$pin) |>
  distinct(pin)

n_high_ff_recent_sales <- nrow(high_ff_recent_sales)

n_high_ff_recent_sales
[1] 48232

The share of high Flood Factor residential properties with a recent sale is calculated below.

Code
share_high_ff_with_recent_sale <- n_high_ff_recent_sales / high_ff_residential_count

share_high_ff_with_recent_sale
[1] 0.2432164
Code
percent(share_high_ff_with_recent_sale, accuracy = 0.1)
[1] "24.3%"

Flood Factor thresholds

The next table summarizes the number of parcels above several Flood Factor thresholds. The table separates residential class 200 parcels from other parcels.

Code
puni_pins_coded <- pins |>
  mutate(
    FFS_4plus = env_flood_fs_factor >= 4,
    FFS_5plus = env_flood_fs_factor >= 5,
    FFS_6plus = env_flood_fs_factor >= 6,
    Res_C2 = class > 199 & class < 300
  )

ff_threshold_summary <- puni_pins_coded |>
  group_by(Res_C2) |>
  summarize(
    FFS_4plus = sum(FFS_4plus, na.rm = TRUE),
    FFS_5plus = sum(FFS_5plus, na.rm = TRUE),
    FFS_6plus = sum(FFS_6plus, na.rm = TRUE),
    n = n(),
    .groups = "drop"
  ) |>
  mutate(
    share_FFS_4plus = FFS_4plus / n,
    share_FFS_5plus = FFS_5plus / n,
    share_FFS_6plus = FFS_6plus / n
  )

ff_threshold_summary |>
  mutate(
    across(starts_with("share_"), ~ percent(.x, accuracy = 0.1))
  )
Table 1: Number of parcels above selected First Street Foundation Flood Factor thresholds, by residential class 200 status.
# A tibble: 2 × 8
  Res_C2 FFS_4plus FFS_5plus FFS_6plus       n share_FFS_4plus share_FFS_5plus
  <lgl>      <int>     <int>     <int>   <int> <chr>           <chr>          
1 FALSE       3639      2604      2248   17408 20.9%           15.0%          
2 TRUE      274782    198309    166660 1586893 17.3%           12.5%          
# ℹ 1 more variable: share_FFS_6plus <chr>

FEMA Individual Assistance claims

This section summarizes FEMA Individual Assistance claims for selected Cook County flood disaster declarations.

The FEMA Individual Assistance file was downloaded from FILL IN FEMA DATA SOURCE. The file is filtered to Cook County, Illinois, and to selected flood-related disaster numbers.

Code
city_names <- read_csv("./data/processed/City_Name_Mapping.csv")

ind_assist_raw <- read_csv("./data/raw/indiv_assistance_V2_CookCounty.csv")

ind_assist <- ind_assist_raw |>
  filter(incidentTypeCode != "B", disasterNumber != "1763") |>
  mutate(
    year = str_sub(declarationDate, 1, 4),
    year = as.numeric(year),
    joinyear = case_when(
      year <= 2010 ~ 2000,
      year < 2020 & year > 2010 ~ 2010,
      year > 2020 ~ 2020,
      TRUE ~ year
    ),
    tract = str_sub(censusGeoid, 6, 9),
    tract = str_pad(tract, width = 6, pad = "0", side = "right")
  ) |>
  left_join(city_names)

Cook County disaster records

The object below keeps all Cook County records, excluding disaster number 4819.

Code
cook_assist <- ind_assist |>
  filter(county == "Cook (County)", disasterNumber != "4819")

cook_assist |>
  count(disasterNumber, declarationDate, incidentTypeCode, sort = TRUE) |> nice_dt()

Selected 2023 flood disaster declarations

The dissertation text focuses on selected 2023 Cook County flood events. This section filters to disaster numbers 4728 and 4749.

Code
selected_flood_assist <- ind_assist |>
  filter(
    county == "Cook (County)",
    disasterNumber %in% c("4728", "4749", 4728, 4749)
  )

selected_flood_assist |>
  count(disasterNumber, declarationDate, sort = TRUE) |> 
nice_dt()

The table below counts Individual Assistance claims with positive ihpAmount, grouped by disaster number and flood insurance status.

Code
claims_by_insurance <- selected_flood_assist |>
  filter(ihpAmount > 0) |>
  group_by(disasterNumber, floodInsurance) |>
  summarize(
    claimcount = n(),
    .groups = "drop"
  ) |>
  arrange(desc(claimcount)) |>
  pivot_wider(
    names_from = floodInsurance,
    values_from = claimcount,
    values_fill = 0
  )

nice_dt(claims_by_insurance)
Table 2: FEMA Individual Assistance claims with positive IHP amounts for selected Cook County flood disasters, by flood insurance status.

Notes for revision

Before publishing this page, update the source notes above by filling in:

  • the exact source for each raw data file;
  • the download date or API access date;
  • whether the file is publicly redistributable and whether size limits require a separate archive;
  • whether the website hosts the raw file, a processed file, or only summary statistics;