{ "title": "Forecasting the Present in Developing Economies", "sourceUrl": "https://www.imf.org/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns", "canonical": "https://www.imf.org/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns", "overlayPath": "/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/index.md", "summary": "Nowcasting and new data sources can provide timelier and more frequent indicators", "sections": [ { "heading": "Nowcasting: concept and evolution", "content": "- Nowcasting uses real-time data to predict the present, very recent past, and very near future of economic activity; term adapted from meteorology to economics in work beginning in 2002 and formalized in a 2008 paper by Giannone, Reichlin, and Small.\n- Nowcasting put informal judgment-based reads on real-time releases into a formal statistical framework and has been expanded to weekly estimates (example: weekly estimates of quarterly US economic growth).\n- Recent advances in large language models and artificial intelligence expand possibilities for using text as data and integrating data with metadata." }, { "heading": "Data gaps and urgency in developing economies", "content": "- About a third of countries publish only annual GDP; many developing economies publish official indicators infrequently or with a delay.\n- First estimates of GDP in many advanced economies come out about a month after the end of the quarter; two months in some major emerging market economies; in developing economies it can take more than three months.\n- The COVID-19 pandemic exposed “statistical darkness,” motivating central banks and agencies to mobilize alternative high-frequency data sources and enhance nowcasting (Irving Fisher Committee/Bank for International Settlements, 2023).\n- Transactional and commercial high-frequency data can provide earlier signals than official releases (examples: Kenya, South Africa)." }, { "heading": "Practical applications and country examples", "content": "- Rwanda\n - During COVID-19, a weekly economic activity index combined central bank measures with tax authority electronic billing machine consumer spending, exports, and imports to detect rapid deterioration and inform policy; nowcasting is included in staff briefings before quarterly Monetary Policy Committee meetings.\n - National Bank of Rwanda expanded its staff of economists, statisticians, and data scientists; chief economist Thierry Kalisa joined the central bank as chief economist in 2021.\n- Kenya\n - National Bureau of Statistics releases GDP about three months after a quarter ends.\n - The central bank can start gauging the quarter after only a week using private consumer spending data; remittance data available after two weeks; trade, money supply, tourism, and electricity data available in about 40 days to refine the picture.\n - IMF-supported nowcasting work shows it is possible to approximate Kenya’s economic growth ahead of official releases and to apply nowcasting where quarterly GDP estimates are unavailable.\n- South Africa\n - Commercial agricultural data provided early reads on meat and other crops to gauge food inflation; commercial indicators lead CPI by one to three months across components.\n - Statistics South Africa released CPI for September three weeks after month-end; corresponding commercial data were available two weeks earlier.\n- Samoa\n - Population about 220,000; central bank has fewer than 90 staff members.\n - Two years into a formal nowcasting system (supported by IMF technical assistance); economics department expected to expand from 8 people to 10.\n- Democratic Republic of the Congo\n - Official GDP figures are published only annually and with long delays.\n - Nowcasting combines copper and cobalt production and prices, money supply, satellite night-light intensity, and Google search trends to feed a quarterly projection model linking short-term assessments with medium-term forecasts and policy scenarios.\n - Mining sector contributes about a third of GDP.\n- Guinea-Bissau\n - In May 2024, deployed blockchain technology to manage public wage bill, in collaboration with IMF, EY, and donors.\n - Wage bill fell to half of tax revenue in 2024 from about four-fifths of tax revenues previously.\n - Platform to be expanded to cover all 26,600 public officials and 8,100 pensioners nationwide.\n- Madagascar\n - In October (year not restated in source), customs introduced agentic AI to detect inconsistencies in customs declarations by cross-analyzing declarations, invoices, manifests, and databases, automating many manual inspection tasks and enabling experts to focus on complex cases." }, { "heading": "Methodological considerations and limitations", "content": "- Practitioners caution that nowcasting and new data sources do not replace the need to expand and improve official statistics; rigorous data collection remains essential.\n- Reichlin: “First start with the data, then you get to the technique.” Simple models often perform best; nowcasting exploits different data releases and combines series at different frequencies.\n- New data sources can be noisy or nonrepresentative; models that perform well can fail when the world changes.\n- Broader concerns include privacy, transparency, legitimacy, and governance when using novel data and AI (Joshua Blumenstock).\n- Capacity constraints in many developing economies: limited budgets, scarce skilled staff, and need for computing tools." }, { "heading": "Policy implications and operational recommendations", "content": "- Use nowcasting to provide timelier, higher-frequency indicators to complement official releases and inform policy decisions in real time.\n- Prioritize identification and use of high-quality domestic data or reliable proxies before deploying complex techniques.\n- Invest in capacity development: expand rosters of economists, statisticians, and data scientists and equip them with necessary computing resources.\n- Combine traditional high-frequency indicators (trade, money supply, production, electricity use) with nontraditional inputs (transaction data, satellite night-lights, Google search trends, commercial market data) to detect turning points and calibrate policy responses.\n- Apply nowcasting regionally and through global linkages: interpolate for low-information countries using neighbors, trade partners, and global GDP comovement.\n- Leverage AI and LLMs cautiously to expand data access and text-as-data applications while addressing privacy, transparency, and governance issues.\n\nForecasting the Present in Developing Economies — Jeff Kearns, F&D Magazine, December 2025.\n\n---\n\n Content in this bundle\n\n- Forecasting the Present in Developing Economies\n - Forecasting the Present in Developing Economies (Markdown version){rel=\"alternate\" type=\"text/markdown\"}\n - Forecasting the Present in Developing Economies (PDF){rel=\"external\" type=\"application/pdf\"}\n\n---\n\nSource: https://www.imf.org/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns" } ], "bullets": [ "[Markdown version](/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/index.md)", "[Structured JSON version](/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/index.json)", "[Bundle manifest](/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/bundle-manifest.json)", "Authors: Jeff Kearns", "Published: November 17, 2025", "Nowcasting uses real-time data to predict the present, very recent past, and very near future of economic activity; term adapted from meteorology to economics in work beginning in 2002 and formalized in a 2008 paper by Giannone, Reichlin, and Small.", "Nowcasting put informal judgment-based reads on real-time releases into a formal statistical framework and has been expanded to weekly estimates (example: weekly estimates of quarterly US economic growth).", "Recent advances in large language models and artificial intelligence expand possibilities for using text as data and integrating data with metadata.", "About a third of countries publish only annual GDP; many developing economies publish official indicators infrequently or with a delay.", "First estimates of GDP in many advanced economies come out about a month after the end of the quarter; two months in some major emerging market economies; in developing economies it can take more than three months.", "The COVID-19 pandemic exposed “statistical darkness,” motivating central banks and agencies to mobilize alternative high-frequency data sources and enhance nowcasting (Irving Fisher Committee/Bank for International Settlements, 2023).", "Transactional and commercial high-frequency data can provide earlier signals than official releases (examples: Kenya, South Africa).", "Rwanda", "Kenya", "South Africa", "Samoa", "Democratic Republic of the Congo", "Guinea-Bissau", "Madagascar", "Practitioners caution that nowcasting and new data sources do not replace the need to expand and improve official statistics; rigorous data collection remains essential.", "Reichlin: “First start with the data, then you get to the technique.” Simple models often perform best; nowcasting exploits different data releases and combines series at different frequencies.", "New data sources can be noisy or nonrepresentative; models that perform well can fail when the world changes.", "Broader concerns include privacy, transparency, legitimacy, and governance when using novel data and AI (Joshua Blumenstock).", "Capacity constraints in many developing economies: limited budgets, scarce skilled staff, and need for computing tools.", "Use nowcasting to provide timelier, higher-frequency indicators to complement official releases and inform policy decisions in real time.", "Prioritize identification and use of high-quality domestic data or reliable proxies before deploying complex techniques.", "Invest in capacity development: expand rosters of economists, statisticians, and data scientists and equip them with necessary computing resources.", "Combine traditional high-frequency indicators (trade, money supply, production, electricity use) with nontraditional inputs (transaction data, satellite night-lights, Google search trends, commercial market data) to detect turning points and calibrate policy responses.", "Apply nowcasting regionally and through global linkages: interpolate for low-information countries using neighbors, trade partners, and global GDP comovement.", "Leverage AI and LLMs cautiously to expand data access and text-as-data applications while addressing privacy, transparency, and governance issues.", "**Forecasting the Present in Developing Economies**" ], "related": [ { "title": "Forecasting the Present in Developing Economies", "role": "document", "sourceUrl": "https://www.imf.org/-/media/files/publications/fandd/article/2025/12/kearns.pdf", "summary": { "path": "/-/media/files/publications/fandd/article/2025/12/kearns.pdf.md", "mime": "text/markdown" }, "binary": { "path": "/-/media/files/publications/fandd/article/2025/12/kearns.pdf", "mime": "application/pdf" } } ], "alternates": { "markdown": "/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/index.md", "json": "/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/index.json", "bundleManifest": "/en/publications/fandd/issues/2025/12/forecasting-the-present-in-developing-economies-jeff-kearns/bundle-manifest.json" }, "generatedAtUtc": "2026-09-15T20:11:10.616Z" }