Skip to contents

Deprecated residual preprocessing utility

Source: R/utils_residuals.R
prepare_cd_input.Rd

prepare_cd_input() is deprecated and is not used by cd_test(). Transforming residuals before a dependence test can change the tested hypothesis. Pass the original residual matrix to cd_test() and use its documented missing-data policy instead.

For compatibility, this function still performs:

  1. Dropping time periods with fewer than min_per_time finite observations.

  2. Optional row-wise standardization to unit variance over available times.

  3. Optional demeaning across units at each time (changes the tested residuals).

Usage

prepare_cd_input(
  E,
  standardize = c("row", "none"),
  demean_time = FALSE,
  min_per_time = 2L
)

Arguments

E

A numeric matrix of residuals (\(N x T\)); rows are units, columns are time; may be unbalanced (contain NA).

standardize

One of "row", "none". If "row", scale each row by its observed standard deviation.

demean_time

Logical; if TRUE, subtract the cross-sectional mean at each time from available residuals in that column.

min_per_time

Integer; drop time columns with fewer than this many finite observations.

Value

A list with:

Z

Processed residual matrix (\(N x T^*\)) after filtering/standardizing/demeaning.

kept_t

Integer indices of kept time columns (relative to the original E).

m_t

Integer vector of cross-sectional counts per kept time (number of finite rows).

row_sds

Numeric vector of row standard deviations used (invisibly NA if standardize="none").

col_means

Numeric vector of time means subtracted when demean_time=TRUE.

Details

This helper is retained temporarily for compatibility. Row scaling changes the weighted covariance underlying CDw, while cross-sectional time demeaning can mechanically induce dependence. Neither transformation is applied by cd_test().

Transformation steps

  1. Time periods with fewer than min_per_time finite observations are removed.

  2. If standardize = "row", each unit is scaled by its observed standard deviation.

  3. If demean_time = TRUE, each time slice is demeaned across available units.

Why this preprocessing matters

CD-type tests are sensitive to scale heterogeneity and sparse columns in unbalanced panels. This helper creates a better-conditioned input matrix while preserving as much usable information as possible.

Examples

data(PWT_60_07, package = "csdm")
df <- PWT_60_07
ids <- unique(df$id)[1:10]
df_small <- df[df$id %in% ids & df$year >= 1970, ]
fit <- csdm(
  log_rgdpo ~ log_hc + log_ck + log_ngd,
  data = df_small,
  id = "id",
  time = "year",
  model = "cce",
  csa = csdm_csa(vars = c("log_rgdpo", "log_hc", "log_ck", "log_ngd"))
)
E <- get_residuals(fit)
#> Warning: 'get_residuals()' is deprecated; use residuals() for csdm fits or pass a residual matrix directly to cd_test().
prep <- prepare_cd_input(E, standardize = "row", demean_time = TRUE, min_per_time = 3)
#> Warning: 'prepare_cd_input()' is deprecated and is not used by cd_test(); pass original residuals to cd_test() instead.
dim(prep$Z)
#> [1] 10 38