A study of Banda Aceh City's annual economic data found no statistically significant short-term relationship between investment and either inflation or the interest-rate spread. Inflation's estimated short-term coefficient was negative, at -0.002432, but its reported probability was 0.3331. The two spread terms were also non-significant: the lagged term was 0.009680 with a probability of 0.8130, while the current-period change was -0.053912 with a probability of 0.1757. In practical terms, the analysis did not produce a clear enough signal to say that either measure tracked investment over the short run.
The result did not disappear into a model with no persistence. Previous-period investment was positively associated with current investment, with a coefficient of 1.034371 and a probability of 0.0000. The selected model reported an R-squared of 0.9864 and an adjusted R-squared of 0.9804, measures of how closely it fit the observed sample. Those figures describe the model's in-sample performance; they do not turn the associations into proof that one variable caused another.
How the analysis was built
The researchers used an autoregressive distributed lag, or ARDL, model, a time-series approach that includes selected earlier values alongside current terms. The annual secondary dataset contained 15 observations from 2010 through 2024; after accounting for the lag structure, 14 observations were used for estimation. Model selection chose ARDL(1,0,1), with one investment lag, no additional inflation lag, and current plus one prior spread lag.
A long-term link without clear adjustment
At the longer horizon, the F-Bounds test produced a statistic of 9.899769, above the upper bound at the reported 10%, 5% and 1% levels. That result indicated a long-term level relationship among investment, inflation and the interest-rate spread. In plain language, the variables showed a pattern consistent with moving together in levels over time, but the test alone does not establish that one caused another or that the relationship was stable after a shock.
The detailed long-term estimates were less decisive. Inflation had a positive coefficient of 0.070749 with a probability of 0.6082, while the spread had a negative coefficient of -0.281639 with a probability of 0.7829. Neither was statistically significant. The adjustment coefficient was positive at 0.034371 with a probability of 0.5017, so the analysis did not confirm convergence toward long-term equilibrium.
Why the result needs a narrow reading
The paper's authors judged the estimated model stable and reliable for interpretation, although they noted a brief period of residual-variance pressure in 2020. Here, the evidence supports a reported level relationship, but not a statistically confirmed inflation or spread effect on investment. The study's design describes associations in a time series, not proof of causation.
The findings also come with a narrow evidence base. The effective sample had only 14 observations, which the authors identify as a limitation that may reduce precision. The interest-rate spread was measured at national level and may not represent financing costs in Banda Aceh. The result should therefore be read as context-specific evidence from one city's annual series, not as a broad rule about how investment responds to inflation or the spread.
Taken together, the analysis presents a mixed picture: previous investment was positively associated with current investment, and the bounds test pointed to a long-term relationship among the variables, but neither inflation nor the spread produced a statistically significant short- or long-term coefficient. The paper reports funding from Politeknik Kutaraja.
Paper data and sources
Original title: The Effect of Inflation and Interest Rate Spread on Real Sector Investment in Banda Aceh City: An ARDL Approach
Authors: Ulin Nur Arifah, Lidyana Dinda
Journal/Repository: Indonesian Journal Economic Review (IJER)
Status: Peer-reviewed
First online: 2026-09-28
DOI: 10.59431/ijer.v6i3.921
Original paper · Full text