Abstract

Following the 2008 global financial crisis and the collapse of U.S. and European market, increasingly sought safer assets. Milesi-Ferretti and Tille (2011) show that the surge in global risk amplified foreign capital outflows from emerging markets, including those with relatively strong fundamentals. Capital flows in emerging markets are highly sensitive to global monetary conditions, particularly when the interest rate differential between domestic policy rates and the Federal Funds Rate narrows. While the conventional interest rate channel is well documented, empirical evidence remains limited on how central bank communications shapes capital flow dynamics through the formation of market narratives and expectations.
We explore the possibility of utilizing generative AI for constructing a prompt indicator based on news data. By providing the generative AI with news data, a novel dataset that integrates textual analysis of Bank Indonesia's official communication outputs, including press releases, monetary policy statements, and official policy narratives, with high-frequency media-based sentiment indicators derived from financial and economic news coverage following policy announcements, including how policy messages are framed and interpreted by the media in terms of credibility, tone, and perceived risk. We leverage publicly available large language models (LLMs), a subset of generative AI specialized in natural language processing, to enhance the accuracy of sentiment analysis.
These sentiment indicators are then combined with daily portfolio flows and exchange rate data within an event-study regression framework to assess whether media sentiment serves as a behavioral transmission channel of monetary policy communication. The analysis is designed to test three main hypotheses. First, a narrowing interest rate differential between the BI Rate and the Federal Funds Rate is associated with stronger capital outflow pressures. Second, more positive media sentiment surrounding Bank Indonesia's policy communication is expected to be associated with relatively more stable capital flows. Third, media sentiment is hypothesized to play a moderating role, where favorable media narratives potentially mitigate the adverse effects of global monetary tightening, while unfavorable narratives may amplify capital outflows and exchange rate volatility.
By examining these relationships, the study seeks to examine the potential effectiveness of Bank Indonesia's monetary policy communication in mitigating capital outflow pressures during periods of global monetary tightening especially in emerging markets.