Since the 2008 global financial crisis and the advent of the Zero Lower Bound (ZLB) environment, central bank communication has evolved into a critical policy tool, extending beyond conventional interest rate adjustments. The policy signals conveyed b...
Since the 2008 global financial crisis and the advent of the Zero Lower Bound (ZLB) environment, central bank communication has evolved into a critical policy tool, extending beyond conventional interest rate adjustments. The policy signals conveyed by central banks play a decisive role in mitigating financial market uncertainty and shaping the expectations of market participants. However, discrepancies in tone can arise between communication channels—such as official written statements and the verbal remarks of the governor—and the market impact of these signals can be heterogeneous, varying with firm-specific characteristics like financial health.
This study investigates the multifaceted impact of the Bank of Korea’s monetary policy communication on the stock prices of 606 non-financial firms listed on the KOSPI from 2013 to 2025. We aim to distinctly identify the market effects of the tones conveyed through three channels: the official Monetary Policy Decision Statement (text), the press conference Q&A (text), and the governor’s remarks during the press conference (voice). Furthermore, we examine how these effects differ based on corporate debt levels and the prevailing macroeconomic conditions.
To quantify the tone of communication across these dimensions, we employ state-of-the-art methodologies. For the textual analysis of the decision statements and press conference Q&A, we utilize KR-FinBERT, a Korean language model specialized for finance, which allows for a deep contextual understanding beyond simple word frequency. This enables us to classify the text into hawkish, neutral, or dovish categories, thereby generating a text-based tone index. To capture the non-verbal cues in the governor’s speech, we construct a voice-based tone index using a Deep Neural Network (DNN) model that analyzes the physical properties of the audio, such as frequency and amplitude. Using these three tone indices as independent variables and firm-level excess stock returns as the dependent variable, we conduct a panel regression analysis.
The empirical findings reveal that the market effects of monetary policy are complex and contingent on the policy direction, the communication channel, and the financial health of the firm. First, on average, an interest rate cut led to a decline in stock prices, suggesting that the negative “information effect”—signaling concerns about a future economic downturn—outweighed the positive liquidity effect. Second, this average response conceals significant heterogeneity based on corporate leverage. For high-debt firms, stock prices rose following a rate cut, driven by the traditional “financial channel” via expectations of reduced interest burdens. Conversely, for low-debt firms, stock prices fell as the “information channel” dominated, with these firms reacting more sensitively to signals of an impending recession. Third, the impact of communication tone was conditional. During rate-cutting phases, high-debt firms responded positively only to a dovish vocal tone from the governor, indicating their reliance on the non-verbal signals of confidence from the policymaker. In contrast, low-debt firms reacted negatively to the dovish text in the official statement but positively to the dovish tone in the press conference Q&A, showing conflicting responses depending on the information channel.
In conclusion, this study establishes that the effectiveness of monetary policy communication is a complex phenomenon determined by the interplay of three key axes: the communication channel (text vs. voice), corporate financial health, and the policy context. Notably, it is the first study in the Korean context to empirically demonstrate the independent informational value of the central bank governor’s vocal tone.