This Selected Issues paper delves into few applications of machine learning (ML), with a particular application to economic forecasts in Lesotho. Amid delayed and often revised gross domestic product data, this paper explores the potential of ML to provide real-time insights into growth and inflation trends, crucial for informed policymaking. By leveraging nontraditional data and employing a variety of ML models, the paper presents a comprehensive analysis of current economic activity, evaluates the accuracy of standard statistical measures, and forecasts future inflation trends. The findings underscore the efficacy of ML in reducing prediction errors and highlight the significant role of alternative data in circumventing the limitations posed by traditional economic indicators. This paper contributes to the broader debate on the application of advanced computational techniques in economic forecasting, offering valuable insights for policymakers in Lesotho and similar countries grappling with data constraints and the need for timely economic analysis.
International Monetary Fund. Monetary and Capital Markets Department
This Technical Assistance (TA) report analyzes expanding the nowcasting toolbox at the National Bank of Rwanda (NBR). The mission built on the progress made during the March 2022 mission, which focused on improving the nowcasting framework for the key domestic variables and building tools for analyzing new data releases and assessing the nowcasting systems. The TA should continue to focus on the developments of the nowcasting framework for inflation and gross domestic product (GDP). Specifically, the new consumer price index (CPI) and GDP Near-Term Forecast (NTF) tools should be used on a monthly basis as part of the forecasting process and Forecasting and Policy Analysis System work at the NBR. The new CPI NTF system now includes the monthly forecasts of ten subgroups of the core CPI inflation as well as two subgroups of food inflation, thus enabling the assessment of the key drivers of inflation as well as the nature of inflation shocks. It also allows for the ‘real time’ monitoring of monthly inflation outcomes relative to the forecast.
International Monetary Fund. Monetary and Capital Markets Department
The currency in circulation forecasting model presently used by the Central Bank of Jordan is aligned with international practices and provides a solid basis for liquidity management. The central bank uses an Auto Regressive Integrated Moving Average (ARIMA) model with many indicator variables to model binary seasonality and to capture special events. The ARIMA model is fitted on daily currency in circulation data using a standard maximum likelihood estimator. This ARIMA approach is aligned with the models traditionally used by central banks in emerging and middle-income countries.
This Article IV Consultation highlights that the economic expansion continues, driven primarily by private consumption and exports of goods and services. Discussions primarily focused on increasing the economy’s flexibility and resilience. Fiscal performance has been strong, however, the materialization of contingent liabilities from government guarantees is likely to reduce the overall surplus. Low public and private investment, and continued emigration appear to weigh on medium-term growth prospects. Downside risks in the near-term stem could be due to possible changes in regional or global economic and financial conditions, and the further realization of contingent liabilities. The IMF staff advocated for a moderately faster fiscal adjustment. The report recommends accelerating the pace of debt reduction that would build fiscal space and help reduce downside risks. The Central Bank may need to address potentially tighter external conditions while continuing with strong bank supervision and macroprudential policies. Additional measures to prevent excessive household borrowing could be considered if needed.