Replication package for "Persistent dependent variables and specification fragility in ARDL studies of development outcomes"

Published: 11 September 2026| Version 1 | DOI: 10.17632/wh3g9v76tx.1
Contributor:
Le Thi Giang

Description

data/ bang-ma-hoa-29-bai.xlsx Coding spreadsheet for the twenty-nine reviewed articles. Sheet "Coding" holds one row per article and one column per coded field. Sheet "Summary" recomputes the counts reported in Section 2 directly from that sheet. Sheet "Huong dan" holds the coding protocol, reproduced in English as Appendix B of the paper. double-coding-coder2_FILLED.xlsx Independent second coding of ten randomly selected articles. Sheet "Instructions" holds the rules supplied to the second coder; sheet "Coder2_Form" holds that coder's entries; sheet "Comparison" computes the agreement rates reported in Section 2.1. code/ nardl_fdi_hdi_vietnam.do Empirical analysis. Enters the Vietnamese data, retrieves the World Bank series through wbopendata, and produces the unit root tests, bounds tests, long-run estimates, diagnostics and robustness checks reported in Sections 3 and 5. montecarlo-nardl.do Monte Carlo simulations reported in Section 4. output/ montecarlo.log Full Stata log of the simulation runs, including all rejection rates reported in Section 4. fig1_hdi_fdi.gph / fig1_hdi_fdi.png Figure 1. Vietnamese Human Development Index and FDI as a share of GDP, 1991 to 2022, plotted on two vertical axes. fig2_mc_size.gph / fig2_mc_size.png Figure 2. Empirical rejection rate of the long-run symmetry test against the signal-to-noise ratio of the dependent variable, with the nominal five per cent level marked. fig3_cusum.gph / fig3_cusum.png Figure 3. Recursive CUSUM statistic for the baseline HDI model with ninety-five per cent confidence bands. Files with the .gph extension are native Stata graph files and can be reopened and edited with the graph use command. The .png files are the same figures exported for viewing without Stata. README.txt This file.

Files

Steps to reproduce

SOFTWARE Stata 17 or later. Three user-written commands must be installed from SSC before running anything: ssc install ardl, replace ssc install zandrews, replace ssc install wbopendata, replace An internet connection is required, both for the SSC installation and because the macroeconomic series are retrieved live from the World Bank API. STEP 1 — EMPIRICAL RESULTS Open code/nardl_fdi_hdi_vietnam.do. Change the path on the "global path" line to a folder on your own machine. Run the whole file from top to bottom. This produces the unit root tests, bounds tests, long-run estimates, diagnostics and robustness checks reported in Sections 3 and 5 of the paper, together with Figures 1 and 3. Human Development Index values are entered directly in the do-file, so no download is needed. All macroeconomic series are retrieved automatically through wbopendata; the series codes are listed in section 3 of README.txt. STEP 2 — MONTE CARLO SIMULATIONS Open code/montecarlo-nardl.do. Change the "cd" line in the same way. Run the whole file. Expect ten to fifteen minutes. The seed is set to 20260830. Leaving it unchanged reproduces the figures in the paper exactly. The full output is also supplied in output/montecarlo.log for comparison. This produces the rejection rates reported in Tables 9 to 14 and Figure 2. STEP 3 — CODED REVIEW No code is needed. Open data/bang-ma-hoa-29-bai.xlsx. Sheet "Coding" holds one row per reviewed article; sheet "Summary" recomputes the counts reported in Tables 1 to 8 directly from it. Agreement rates reported in Section 2.1 are on the "Comparison" sheet of data/double-coding-coder2_FILLED.xlsx. NOTE ON REPRODUCIBILITY Because the World Bank series are retrieved live, figures may differ marginally from those in the paper if the World Bank revises the underlying national accounts after the retrieval date stated in the manuscript. The simulation results do not depend on any external data and reproduce exactly.

Categories

Economics

Funders

  • University of Finance – Marketing

Licence