Skip to contents

Are spending gaps within a peer group closing or widening? This is the classic public-finance convergence test, pointed at school dollars: it regresses each district's growth in per-pupil spend on its starting level, within each peer group. A negative slope (beta) means low-spenders grew faster and the group is converging; a positive slope means the high-spenders pulled further ahead (divergence).

Usage

tges_convergence(
  tges,
  metric_col = "Per Pupil costs",
  table = "CSG1",
  peer = c("tges_group", "dfg", "county", "statewide"),
  start_year = NULL,
  end_year = NULL,
  calc_type = "Budgeted",
  dfg_revision = 2000
)

Arguments

tges

Output of fetch_many_tges() (multi-year).

metric_col

Character. Numeric column to track. Default "Per Pupil costs".

table

Character. TGES table code carrying metric_col. Default "CSG1" (budgetary per-pupil cost).

peer

Character. Peer group. One of "tges_group" (default), "dfg", "county", "statewide".

start_year, end_year

Numeric. Endpoints. Default: the min and max end_year present.

calc_type

Character. Calc type to keep. Default "Budgeted".

dfg_revision

Numeric. DFG revision when peer = "dfg". Default 2000.

Value

A per-district tibble: entity columns, peer_group, start_year, end_year, start_value, end_value, log_start_value, growth, and the broadcast group statistics beta, beta_pvalue, r_squared, n_districts, and converging (beta < 0 and beta_pvalue < 0.05).

Details

Needs a multi-year input (fetch_many_tges()). For each district the function takes the metric at start_year and end_year, drops to one value per district-year, and computes annualized log growth \((\ln v_{end} - \ln v_{start}) / (end - start)\). Within each peer group it fits growth ~ log(start_value) by OLS (requires >= 4 districts with both endpoints and starting-level variation). The slope and its stats are broadcast onto every district row in the group, so the same frame both plots the convergence scatter (log_start_value vs growth) and reports the group beta. For one peer group's headline, take distinct(peer_group, beta, beta_pvalue, r_squared, n_districts, converging).

Examples

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

conv <- tges_convergence(fetch_many_tges(2015:2024), peer = "dfg")

# one row per peer group: is each DFG converging on spend?
conv %>% distinct(peer_group, beta, beta_pvalue, converging, n_districts)
} # }