This is a Mock-Up Article
Short-run vs long-run dynamics using ARDL bounds testing
Background
Understanding how exports respond to exchange rate movements is central to evaluating expenditure-switching policies and the effectiveness of currency depreciation as a trade tool. For Indonesia, a commodity-dependent open economy, this question carries particular weight: the rupiah has experienced several sharp depreciations since the 1997 Asian financial crisis, yet the export response has been uneven across sectors and time.
This analysis estimates Indonesia’s aggregate export elasticity using an Autoregressive Distributed Lag (ARDL) bounds testing framework, augmented with Fourier terms to capture smooth structural breaks without imposing a known break date. The approach follows Enders and Jones (2016) and is estimated over quarterly data from 1990Q1 to 2023Q4.
Data and Variables
The model uses four core variables. Real exports are measured in constant 2015 USD from the World Bank national accounts data. The real effective exchange rate (REER) is sourced from the BIS, indexed to 2020. Foreign income is proxied by a trade-weighted average of GDP growth among Indonesia’s top ten export destinations. Domestic supply capacity is controlled for using gross fixed capital formation as a share of GDP.
All variables are log-transformed. Unit root properties are confirmed using ADF and KPSS tests, with mixed orders of integration — I(0) and I(1) — making the ARDL framework appropriate over the Johansen cointegration approach.
Estimation Results
The bounds test rejects the null of no long-run relationship at the 1% significance level (F-statistic = 6.84, upper bound critical value = 5.06 at 1%). The long-run coefficients are reported below.
| Variable | Coefficient | Std. Error | t-statistic | p-value |
|---|---|---|---|---|
| ln(REER) | −0.842 | 0.134 | −6.28 | 0.000 |
| ln(Foreign Income) | 1.203 | 0.211 | 5.70 | 0.000 |
| ln(GFCF/GDP) | 0.461 | 0.188 | 2.45 | 0.016 |
| Fourier sin(1) | −0.073 | 0.029 | −2.51 | 0.013 |
| Fourier cos(1) | 0.041 | 0.027 | 1.52 | 0.131 |
The long-run exchange rate elasticity of −0.842 implies that a 10% real depreciation of the rupiah is associated with an 8.4% increase in real exports in the long run, consistent with the Marshall-Lerner condition being satisfied. Foreign income elasticity exceeds unity, suggesting exports are income-elastic with respect to trading partners — a result that underlines Indonesia’s sensitivity to demand conditions in China, the United States, and Japan.
Short-run Dynamics
The error correction term (ECT) is estimated at −0.213, significant at the 1% level, implying that roughly 21% of any deviation from the long-run equilibrium is corrected each quarter. The half-life of a shock is approximately 3.3 quarters.
Short-run exchange rate effects are smaller and less precisely estimated. The contemporaneous coefficient on ΔREER is −0.31 (p = 0.04), rising to −0.58 when accumulated over four quarters. This J-curve pattern — where the trade balance initially worsens before improving following a depreciation — is consistent with prior findings for Indonesian trade (Bahmani-Oskooee and Kovyryalova, 2008).
Diagnostic Tests
Breusch-Godfrey LM test (serial correlation): F = 1.24, p = 0.29 ✓
ARCH test (heteroskedasticity): F = 0.88, p = 0.42 ✓
Ramsey RESET (functional form): F = 1.61, p = 0.21 ✓
CUSUM test: Stable at 5% level ✓All diagnostics pass. The CUSUM and CUSUM-of-squares tests indicate parameter stability across the sample period, with the Fourier terms absorbing the smooth transition associated with the 1997–98 crisis and the 2008 global financial crisis.
Discussion
The results suggest Indonesia’s export sector does respond to real exchange rate movements in a direction consistent with standard trade theory, but the magnitude is moderate rather than large. The long-run elasticity of −0.84 is below what expenditure-switching models would predict for a fully competitive export sector, likely reflecting the commodity composition of Indonesia’s export basket — palm oil, coal, and natural gas are priced in USD and less sensitive to bilateral exchange rate movements than manufactured goods.
The high income elasticity (1.20) implies that sustaining export growth depends more on conditions in partner economies than on domestic exchange rate policy. This has direct implications for fiscal and monetary coordination: currency depreciation alone is an insufficient lever for export promotion if global demand is weak.
Conclusion
This analysis confirms a stable long-run cointegrating relationship between Indonesian real exports, the REER, and foreign income. The exchange rate elasticity satisfies the Marshall-Lerner condition but remains below unity, suggesting moderate expenditure-switching effects. Short-run adjustment is slow, with a 21% quarterly error correction rate and evidence of J-curve dynamics. Future work should disaggregate by sector to identify whether commodity versus manufactured exports drive these aggregate results differently.
References
Bahmani-Oskooee, M., & Kovyryalova, M. (2008). Impact of exchange rate uncertainty on trade flows. The World Economy, 31(8), 1097–1128.
Enders, W., & Jones, P. (2016). Grain prices, oil prices, and multiple smooth breaks in a VAR. Studies in Nonlinear Dynamics & Econometrics, 20(4), 399–419.
Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326.