Worked example: FDI and economic growth in Vietnam (ARDL)

This page illustrates EcoLab's entire 5-step research workflow through a topic familiar to graduate students and PhD candidates in economics: the impact of foreign direct investment (FDI) on Vietnam's economic growth. The goal is to show how to go from a research question to a report with replication code — not to publish official empirical findings. The figures in the results section are format illustrations.

Summary: with a national time series for 1990–2023 and variables of mixed I(0)/I(1) order of integration, the ARDL model is a suitable choice to estimate both the long-run and short-run relationships between FDI and growth.

Step 1 — Ideation

Research question: Does FDI promote Vietnam's economic growth in the long run, and how fast does the economy adjust to equilibrium?
Context: Vietnam has attracted FDI strongly since the 1990s; the FDI–growth relationship is still debated in terms of sign and magnitude.
Intended contribution: test FDI–growth cointegration for an updated period, controlling for trade openness and inflation.

In the Ideation module, enter keywords ("FDI", "economic growth", "Vietnam", "ARDL") so the system suggests questions, gaps and preliminary variables.

Step 2 — Literature Review

In the Literature Review module, EcoLab finds and synthesizes related papers, standardizes citations (APA/Harvard…) and clarifies:

Theoretical framework: endogenous growth theory, the technology-spillover role of FDI.
Gap: a lack of updated ARDL studies using recent Vietnamese data with full controls.
Proposed variables: the dependent and explanatory/control variables (see Step 3).

Step 3 — Data Collection

Variable	Symbol	Measurement	Source
GDP growth	`growth`	Annual % change in real GDP	World Bank WDI; GSO Vietnam
FDI intensity	`fdi`	Net FDI / GDP (%)	World Bank WDI; EcoData
Trade openness	`open`	(Exports + Imports)/GDP (%)	World Bank WDI; Vietnam Customs
Inflation	`inf`	Annual % change in CPI	World Bank WDI; GSO

Period: 1990–2023 (annual data).
In EcoLab, connect EcoData or public sources to extract, merge by year, clean, and view descriptive statistics before estimation.

Step 4 — Modeling

Unit-root tests (ADF/PP/KPSS) for each variable to confirm no variable is I(2) — the condition for using ARDL.
Choose the Time Series group → ARDL; set growth as the dependent variable and fdi, open, inf as regressors.
Let the system pick lags by AIC/BIC.
Run the bounds test to conclude on cointegration; review the long-run coefficients and the ECM.

Illustrative results (format — not real results):

Component	Coefficient	Std. error	p-value
Long run: fdi	0.42	0.15	0.012
Long run: open	0.08	0.04	0.061
Long run: inf	−0.05	0.03	0.089
ECM ( $\alpha$ )	−0.55	0.13	0.001
Bounds F-stat	6.10	above upper I(1) bound	cointegrated

Sample interpretation: a negative, significant ECM coefficient confirms a long-run relationship; FDI has a positive long-run effect on growth in this illustrative specification.

Stata
R
Python

* --- FDI & Growth: ARDL ---
tsset year

* Estimate ARDL with automatic lag selection
ardl growth fdi open inf, lags(. 2 2 2)

* Bounds test for cointegration
estat btest

* Error Correction Model
ardl growth fdi open inf, lags(. 2 2 2) ec

* Diagnostics
estat bgodfrey
estat hettest

# --- FDI & Growth: ARDL ---
library(ARDL)

# Estimate ARDL
model <- ardl(growth ~ fdi + open + inf, data = df,
              order = c(2, 2, 2, 2))
summary(model)

# Bounds test
bounds <- bounds_f_test(model, case = 3)
print(bounds)

# Error Correction Model (long-run & ECM)
ecm <- recm(model, case = 3)
summary(ecm)

from statsmodels.tsa.ardl import ARDL

# Estimate ARDL(2, 2, 2, 2)
y = df['growth']
X = df[['fdi', 'open', 'inf']]

model = ARDL(endog=y, lags=2, exog=X, order=2)
result = model.fit()
print(result.summary())

# Review AIC/BIC for lag selection
print("AIC:", result.aic, " BIC:", result.bic)

Step 5 — Reporting

In the Reporting module, EcoLab produces an academic draft (APA 7th, Chicago, Harvard, IEEE or MLA) including: introduction, literature review, data & methodology, results, discussion, and a replication-code appendix (Stata/R/Python). You download it and continue refining it for your journal or committee.

Reproducibility and verification

Every estimation step is generated by EcoLab as complete source code in Stata, R or Python under the Replication Code tab. You can re-run it independently on your local machine to verify, ensuring the results are transparent and reproducible — a core requirement for academic publishing.

Video tutorial

Video Tutorial: Running ARDL in EcoLab

Step 1 — Ideation​

Step 2 — Literature Review​

Step 3 — Data Collection​

Step 4 — Modeling​

Step 5 — Reporting​

Reproducibility and verification​

Video tutorial​

See also​