Published Online:July 2026
Product Name:The IUP Journal of Accounting Research & Audit Practices
Product Type:Article
Product Code:IJARAP040726
DOI:10.71329/IUPJARAP/2026.25.3.60-85
Author Name:Bishal Routh and Joy Sarkar
Availability:YES
Subject/Domain:Finance
Download Format:PDF
Pages:60-85
Exchange rate fluctuations are a significant concern for multinational companies (MNCs). Although financial reports provide adequate provisions to mitigate this issue under IAS 21, the actual effects become apparent through cash inflows and outflows. A firm that is involved in any kind of foreign transaction may suffer this fate. It can be negated using financial derivatives; however, sole dependence on forward/future market might not be the optimal strategy. Several economic, univariate econometric, and machine learning (ML) models have been used to provide accurate forecasting; however, they do not have high accuracy in out-of-sample. Hence, the paper introduces a new framework that provides high-level accuracy; it was tested across 6 alternative periods that encompass almost entire floating. The proposed model was used to forecast the rupee-dollar exchange, and the forecasting was compared to the benchmark models, ARIMA and ANN. The proposed model completely outperformed ARIMA, while ANN was able to topple the proposed model in one period. Overall, ARIMA generates lower error compared to ANN. The proposed model provides accuracy above 99% for 3 periods and 98% for 2. The success and high accuracy of the model open many ways for participant-optimized positioning in spot and forward markets.
There are two main types of exchange rate exposure: accounting (or translation) exposure and economic exposure. The theoretical background of exchange rate exposure shows how it affects both foreign and domestic companies.