shiftNA: NA-Padded Lag and Lead Operator

In Westerlund: Panel Cointegration Tests Based on Westerlund (2007)

NA-Padded Lag and Lead Operator

Description

Shifts a vector forward or backward by a specified number of positions and fills out-of-range values with NA. This helper is designed for position-based transformations where missing values should be explicitly propagated as NA rather than assumed to be zero.

Usage

shiftNA(v, k)

Arguments

v	A vector (numeric, character, etc.).
k	Integer. Shift order. Positive values correspond to lags (shifting the series downward, i.e.\ t-k), while negative values correspond to leads (shifting the series upward, i.e.\ t+k).

Details

This function performs a position-based shift of the input vector v. Unlike strict time-indexed helpers (such as get_lag), shiftNA() does not rely on an explicit time index and does not propagate gaps based on timestamps. Instead, it performs a simple index shift, padding the resulting empty slots with NA.

Let n denote the length of v. The behaviour is:

• k = 0: return v unchanged,

• k > 0: lag by k positions; the first k elements are NA,

• k < 0: lead by |k| positions; the last |k| elements are NA.

Formally, for k > 0,

\text{out}_t = \begin{cases} \text{NA}, & t \le k, \\ v_{t-k}, & t > k, \end{cases}

and for k < 0,

\text{out}_t = \begin{cases} v_{t+|k|}, & t \le n-|k|, \\ \text{NA}, & t > n-|k|. \end{cases}

Value

A vector of the same length as v, containing the shifted values with NA inserted where the shift would otherwise go out of range.

Usage and Propagation

This section explains the implications of using NA padding in recursive or difference-based calculations.

Why NA-padding?

Using NA is the standard behavior in R for out-of-bounds operations. It ensures that subsequent calculations (like v - shiftNA(v, 1)) correctly result in NA for boundary cases, preventing the accidental use of arbitrary values (like zero) in statistical estimations unless explicitly intended.

Difference from get_lag()

Unlike get_lag, shiftNA() is position-based and ignores time indices. Use shiftNA() when you need a fast, simple shift on vectors that are already correctly ordered.

See Also

WesterlundBootstrap, get_lag, get_diff

Examples

v <- c(1, 2, 3, 4, 5)

## Lag by one (k = 1)
shiftNA(v, 1)
# [1] NA 1 2 3 4

## Lead by one (k = -1)
shiftNA(v, -1)
# [1] 2 3 4 5 NA

## Larger shifts
shiftNA(v, 2)
# [1] NA NA 1 2 3

shiftNA(v, -2)
# [1] 3 4 5 NA NA

Westerlund documentation built on Feb. 7, 2026, 5:07 p.m.