Skip to contents

Function to retrieve customized U.S. military construction spending data.

Source: R/get_builddata.R
get_builddata.Rd

get_builddata() subsets build_data_20260918 to a customized data frame containing location-project-year observations of U.S. military construction and housing spending, as published in the U.S. Department of Defense Comptroller Annual Report C-1 exhibits. Users can filter by host country/location, organization, transaction type, facility title, budget activity, project title, spending type, and amount range.

Usage

get_builddata(
  host = NA,
  location = NA,
  organization = NA,
  transaction_type = NA,
  facility_category = NA,
  budget_activity = NA,
  project = NA,
  spend_type = "all",
  min_amount = NA,
  max_amount = NA,
  include_requests = TRUE,
  startyear,
  endyear
)

Arguments

host

Character or numeric. If character, the three-letter ISO3C country code (iso3c field), or a vector of such codes (e.g., c("KWT", "JPN", "DEU") for Kuwait, Japan, and Germany). Matching is case-insensitive. If numeric, the value is treated as a Gleditsch and Ward (G&W) country code, or vector of codes (gwcode field), matched exactly. Domestic (US) construction is coded as "USA" / 2. Use NA (the default) to return all locations.

The G&W system does not assign codes to dependencies. Those locations carry the custom codes that the troop deployment data use for the same places, so one code selects a place in both data sets: Puerto Rico (6, "PRI"), Greenland (1002, "GRL"), Diego Garcia (1004, "IOT"), Guam (1008, "GUM"), the Northern Mariana Islands (1011, "MNP"), the U.S. Virgin Islands (1013, "VIR"), Wake Island (1014, "UMI"), American Samoa (1041, "ASM") and Ascension Island (1042, "SHN"). Roughly a quarter of rows report an amount but no usable location ("Unspecified Worldwide Locations" and similar) and have neither code.

location

Character. A string or regular expression to match against location_name and location_full_name. Matching is case-insensitive. A vector of patterns is collapsed into a single alternation. Use NA (the default) to return all locations.

organization

Character. One or more organizations to filter by, given as full names: "Army", "Navy", "Air Force", "Defense-Wide", "Special Operations Command", "Defense Logistics Agency", "Defense Health Agency", "Department of Defense Education Activity", "National Security Agency" and other defense agencies. Matching is case-insensitive and unmatched values raise an error. Use NA (the default) to include all organizations. See sort(unique(troopdata::build_data_20260918$organization)) for the full list, and organization_reported for the abbreviation each source sheet used.

There is no "Marine Corps" value: the C-1 exhibits carry Marine Corps construction in the Navy accounts. Guard and reserve components appear in appn_title rather than here. For FY2004-FY2007 the service of an Army, Navy or Air Force line is read from the "Treasury Agency" column of the report, which is where those workbooks print it.

transaction_type

Character. One or more transaction types to include: "BUDGET", "REPROGRAM", "REDUCTION", "CONGRESSIONAL RESCISSION", or "PUBLIC LAW RESCISSION". Use NA (the default) to include all transaction types. NOTE: this field is reported for FY2001 and FY2002 only, so filtering on it restricts the result to those years.

facility_category

Character. One or more facility titles to filter by, at either level of the DoD facility taxonomy: the coarse group ("Operational facilities", "Family housing operations", "Training facilities") or the finer three-digit category ("Training buildings", "Airfield operational buildings", "Aircraft maintenance facilities"). Matching is case-insensitive and unmatched values raise an error. Use NA (the default) to include all facility titles.

A value is matched against facility_group_title and facility_category_title together, and a row is kept if either column holds it. The two columns are levels of one vocabulary and the reports do not keep them apart: the coarse labels arrive under the header "Facility Group Title" in some years, "Fiscal Category Title" in others, and in the FY2022-forward workbooks in the facility category column itself. Separating them in the data is therefore a reconstruction, and searching both is the only way to filter without having to know which header a given report year used. Three labels ("Energy conservation", "Family housing P&D", "NATO security investment program") are genuinely present at both levels and return rows from each.

See sort(unique(c(troopdata::build_data_20260918$facility_group_title, troopdata::build_data_20260918$facility_category_title))) for the 190 values present in the data. Not every row carries a facility title, so filtering on this argument drops the rest.

budget_activity

Character. One or more budget activity titles to filter by (e.g., "MAJOR CONSTRUCTION", "PLANNING AND DESIGN", "LEASING"). These labels are stored in upper case, but matching is case-insensitive. Use NA (the default) to include all budget activities. Reported for FY2001 onward.

project

Character. A string or regular expression to match against project_title. Matching is case-insensitive. A vector of patterns is collapsed into a single alternation. Use NA (the default) to return all projects.

spend_type

Character. The amount column used for dropping missing values and for applying min_amount / max_amount. One of:

  • "all" (default): no amount column is required to be present; amount filters are applied to toa_amount

  • "appn": appropriations (appn_amount), FY2000 onward

  • "auth": authorizations (auth_amount), FY2000 onward

  • "auth_appn": authorized for appropriation (auth_appn_amount), FY2001 onward

  • "toa": total obligational authority (toa_amount), FY2001 onward

When spend_type is not "all", rows where the selected amount is NA are dropped. All amount columns are retained in the output in every case. Note that toa_amount is not reported for FY2000, so an amount filter with spend_type = "all" or "toa" excludes that year.

min_amount

Numeric. Minimum spending amount in thousands of current US dollars. Rows below this threshold are excluded. Use NA (the default) for no lower bound.

max_amount

Numeric. Maximum spending amount in thousands of current US dollars. Rows above this threshold are excluded. Use NA (the default) for no upper bound.

include_requests

Logical. Each C-1 report covers two or three fiscal years, the last of them as a budget request, and the data holds every fiscal year from the most recent report that covers it. That is an enacted or actual figure for every year except FY2016, whose only report among the source files is the one that requests it. FY2010 also has 26 lines that come from an overseas contingency operations request sheet and are on no enacted sheet. TRUE (the default) returns everything, with those rows flagged by the is_request column. FALSE drops FY2016 and the 26 FY2010 lines.

startyear

Numeric. The first fiscal year for the series.

endyear

Numeric. The last fiscal year for the series.

Value

A data frame of location-project-year observations of U.S. military construction spending. Amount fields are in thousands of current US dollars.

Details

Each fiscal year is taken whole from the most recent C-1 report that covers it, so a project-year appears once. A row is one line of that report; source_file, source_sheet and source_row identify it.

Several fields appear only in a subset of years, because the C-1 exhibit layout changed over time and a later report does not always print what an earlier one did. transaction_type and appn_title are FY2001-FY2002 only; pe FY2003-FY2004; dollar_type and existing_mission FY2003-FY2005; location_code FY2000-FY2006; facility_category_code FY2000 and FY2003-FY2006; project_number FY2000-FY2006 and FY2013 onward; state_country_sort FY2006; classification FY2011 onward; facility_category_title FY2003-FY2005, FY2007-FY2021 and FY2026. FY2006 reports appropriations only: toa_amount, auth_amount, auth_appn_amount and budget_activity_title are missing for that year, and toa_amount for FY2000. percent_recapitalization has no values. Filtering on one of these implicitly restricts the result to the years that report it.

The *_reported columns (organization_reported, budget_activity_title_reported, facility_category_title_reported, location_name_reported, project_title_reported) preserve the label exactly as printed in the source workbook, before harmonization. source_file, source_sheet, source_row, report_year, and sheet_type record provenance and can be used to trace any row back to the originating exhibit.

latitude and longitude are present only where the location was found inside the country or state the row is filed under. Rows filed under no country (unspecified, worldwide and classified locations) and rows whose location is a label rather than a place ("Various Locations") have none. geo_source says where a point came from: a geocoding service, or "manual" for the installations the services placed at the centre of the country or at another place of the same name, which are entered by hand.

References

Michael A. Allen, Michael E. Flynn, and Carla Martinez Machain. 2020. "Outside the wire: US military deployments and public opinion in host states." American Political Science Review. 114(2): 326-341.

Author

Michael E. Flynn

Examples


if (FALSE) { # \dontrun{
library(troopdata)

# All observations from 2008 through 2019
example <- get_builddata(startyear = 2008, endyear = 2019)

# Kuwait and Japan, budget transactions only. transaction_type is reported
# for FY2001 and FY2002 only.
example2 <- get_builddata(
  host = c("KWT", "JPN"),
  transaction_type = "BUDGET",
  startyear = 2001,
  endyear = 2002
)

# Army and Navy major construction, TOA >= $10 million
example3 <- get_builddata(
  organization = c("Army", "Navy"),
  budget_activity = "MAJOR CONSTRUCTION",
  spend_type = "toa",
  min_amount = 10000,
  startyear = 2015,
  endyear = 2023
)

# Projects matching "hospital" in title
example4 <- get_builddata(
  project = "hospital",
  spend_type = "appn",
  startyear = 2010,
  endyear = 2023
)

# Enacted aircraft maintenance construction, excluding budget requests
example5 <- get_builddata(
  facility_category = "Aircraft maintenance facilities",
  include_requests = FALSE,
  startyear = 2010,
  endyear = 2024
)
} # }