Disagreement about Monetary Policy

货币政策是争论的焦点。市场参与者和预测者对未来的通胀和利率意见不一。然而，政策制定者本身对未来政策走向也存在分歧，这应该会影响市场参与者对政策的看法。这个项目关注的是欧元区决策者之间的后一种分歧。在前瞻性指引等非常规货币政策的背景下，分歧尤其重要，因为前瞻性指引有赖于央行提供有关未来政策信息的能力。如果分歧混淆了对央行信号的解读，那么政策可能就不那么有力了。我们研究欧洲央行(ECB)管理委员会中政策制定者的分歧。与其他央行不同的是，欧洲央行的决策过程仍然不透明。我们通过基于文本数据，即政策制定者发表的讲话和定期新闻发布会的文字记录来衡量分歧，从而克服了这一问题。我们计划分三步解决分歧：首先，我们使用新闻发布会上的信息来构建异议的叙事性索引。我们将根据会议中的异议或一致意见来评估市场在会议日的反应是否不同。我们还会评估会议中的不同意见是否会影响私营部门意见的分散。其次，我们将扩大我们现有的数据库，以包括欧洲央行管理委员会成员的所有演讲。基于计算语言模型，我们将演讲语料库分成不同的主题，这样我们就可以构建特定于主题的分歧度量。在时间序列模型中，我们评估了不一致对政策传递的影响。第三，我们研究了关于通胀和实体经济活动的不同偏好，将其作为理事会成员之间分歧的一个重要来源。在最优货币政策的理论模型中，偏好是通过损失函数来表达的，央行基于经济模型将损失函数最小化。与文献相反，我们根据演讲的语气，而不是完整的经济模型的解决方案，直接估计理事会和个别成员的损失函数。这也让我们能够评估非二次损失函数，在这种函数中，决策者对通胀或失业率与目标的正向和负向偏离给予不同的权重。该项目的所有三个部分都将为货币政策和金融市场之间的关系以及货币政策脉冲的传递提供新的见解。

Monetary policy is subject to debate. Market participants and forecasters disagree about future inflation and interest rates. However, there is also disagreement among policy-makers themselves about the future course of policy, which should affect the perception of policy by market participants. This project focuses on the latter type of disagreement among policy-makers in the euro area. Disagreement should be particularly important against the backdrop of unconventional monetary policy such as forward guidance, which relies on the ability of the central bank to provide information about future policy. If disagreement obfuscates the interpretation of signals from the central bank, policy might be less powerful.We study disagreement of policy-makers in the Governing Council of the European Central bank (ECB). In contrast to other central banks, the decision-making process at the ECB remains opaque. We overcome this problem by measuring disagreement based on text data, i.e. speeches delivered by policy-makers and transcripts of the regular press conferences. We plan to address disagreement in three steps: First, we use information from the press conference to construct a narrative index of dissent. We will assess whether the market reaction on meeting days is different depending on dissent or unanimity in the meeting. We will also evaluate whether dissent in the meeting affects the dispersion of views of the private sector. Second, we will extend our existing database to include all speeches of members of the ECB’s Governing Council. Based on a computational linguistic model, we slice the corpus of speeches into distinct topics, such that we can construct a topic-specific measure of disagreement. In a time-series model we evaluate the consequences of disagreement for policy transmission. Third, we study different preferences about inflation and real economic activity as one important source of disagreement among members of the council. In theoretical models of optimal monetary policy, preferences are expressed by a loss function, which the central bank minimizes based on a model of the economy. In contrast to the literature, we estimate loss functions directly, both for the council and for individual members, based on the tone of speeches rather than the solution of a full model of the economy. This also allows us to evaluate non-quadratic loss functions, where policy-makers put different weights on positive and negative deviations of inflation or unemployment from target. All three parts of the project will provide new insights into the relationship between monetary policy and financial markets and the transmission of monetary policy impulses.