Information, Response Delays, and the Effects of Monetary Policy

这个项目有助于理解货币政策的传导机制，也就是说，货币政策的变化影响实际活动和通货膨胀的方式。建立一个更准确的这些效应模型的主要目的是，允许对采用一种或另一种货币政策规则的后果进行更精确的分析，并计算出最优规则。分析的基础是优化私营部门的行为，以便根据私营部门目标实现的程度来分析备选政策的福利后果。现有的货币传导机制优化模型的一个缺陷是，它们过于“前瞻性”：通胀和实际经济活动都被预测为完全由当前和预期的未来利率政策决定，而不受过去条件的影响。相反，关于货币政策影响的经验证据表明存在反应延迟，这意味着政策变化的影响不会立即被观察到，而且它们比对利率的影响持续的时间更长。该项目考虑了标准模型的两种修改的后果，这两种修改允许响应延迟，同时仍然假设私营部门的优化行为。第一类模型假设私人决策者不会根据当前条件不断地重新优化，而是在一定时间间隔后才重新考虑他们的行为。这可以延迟货币政策变化对总支出的影响，因为不是每个人都立即修改他们以前的支出计划。当这种间歇性优化模型应用于消费者支出时，它对总消费的影响类似于最近流行的“习惯持久性”假设，但对个别家庭行为没有一定的影响，这与面板数据证据不相符。第二类模型假设私人决策者不会为了节省有限的信息处理能力而持续地完全准确地监测当前的情况。假设决策者使用信息论意义上的“噪声信道”来监测环境。这不仅仅是一个隐喻，而是提供了决策者对环境感知中的错误来源的定量模型，特别是提供了一种有用的方法来量化对信息处理能力的假设约束。这使得货币政策变化的影响得以延迟，因为决策者无法立即确定地察觉到变化已经发生。由此产生的理论与20世纪70年代的不对称信息模型有一些相似之处，但能够回答对这些理论提出的重要异议，因为不对称信息是基于个人关注环境中一切事物的能力的限制，而不是基于相关数据不公开的任何主张。这类模型的一个有趣特征是，即使个别价格制定者关于当前总体状况的信息相对准确，也可能导致价格调整的显著延迟。这就足够了（因为由于信道容量不足导致的观察误差），对于其他人可能相信的其他人可能相信的其他人可能相信的其他人可能相信的……关于聚合状态。换句话说，与其说是对经济状况的不确定性，不如说是缺乏常识导致了反应迟缓。这种解释反应延迟的方法有助于解释具有粘性价格的优化模型无法解释的观察到的货币政策影响的各个方面，包括价格对不同类型冲击的不同反应速度，以及在某些条件下通货膨胀率（相对于价格水平）的明显粘性。这类模型的一个更复杂的版本试图解释决策者监控单个变量的精确程度，通过假设一个全局通道容量约束（对每段信息可以处理的总比特数的约束），决策者根据这个约束选择最佳的信息收集策略。这提供了一种理论的前景，它可以产生非常精确的预测，同时允许信息结构随着经济状况的变化而发生重大变化，包括货币政策的剧烈变化。该项目考虑了最优货币政策的设计，既考虑了信息结构是给定的，也考虑了信息结构对政策制度的内生反应。

This project contributes to the understanding of the transmission mechanism of monetary policy, which is to say, of the way in which changes in monetary policy affect real activity and inflation. The primary aim of developing a more accurate model of these effects is to allow a more precise analysis of the consequences of adopting one or another monetary policy rule, and the calculation of an optimal rule. The analysis is grounded in optimizing private-sector behavior, in order to allow an analysis of the welfare consequences of alternative policies in terms of the extent to which private objectives are fulfilled.A defect of existing optimizing models of the monetary transmission mechanism is that they are too exclusively "forward-looking": both inflation and real activity are predicted to be determined purely by current and expected future interest-rate policy, with no effect of past conditions. Empirical evidence on the effects of monetary policy instead indicate the existence of response delays, that imply both that the effects of policy changes are not observed immediately and that they persist longer than do the effects upon interest rates. The project considers the consequences of two kinds of modifications of standard models that allow for response delays while still assuming optimizing behavior on the part of the private sector.The first class of models assumes that private decisionmakers do not continuously reoptimize in the light of current conditions, but instead only reconsider their actions after a certain interval. This allows a delay in the effects of a monetary policy change upon aggregate expenditure, on the ground that not everyone immediately revises their previous spending plans. When this model of intermittent optimization is applied to consumer expenditure, it has consequences for aggregate consumption similar to those of the recently popular hypothesis of "habit persistence", but without certain implications for individual household behavior that do not fit with panel data evidence.The second class of models assumes that private decisionmakers do not continually monitor current conditions with complete accuracy, in order to economize on limited information-processing capacity. It is assumed that decisionmakers monitor their environment using a "noisy channel" in the information-theoretic sense. This is not simply a metaphor, but provides a quantitative model of the sources of error in decisionmakers' perception of their environment, and in particular provides a useful way of quantifying the assumed constraint on information-processing capacity. This allows a delay in the effects of a monetary policy change, on the ground that decisionmakers are not immediately able to discern with certainty that the change has occurred. The resulting theory has some similarities to asymmetric-information models of the 1970s, but is able to answer important objections that were raised to those theories because the asymmetric information is based upon limitations upon individuals' ability to pay attention to everything in their environment, rather than upon any claim that the relevant data are not publicly available.An interesting feature of this type of model is that significant delays in price adjustment can result even when the information of individual price-setters about current aggregate conditions is relatively accurate. It suffices that (because the errors in observation resulting from inadequate channel capacity) there be significant uncertainty about what others may believe that others may believe that others may believe ... about the aggregate state. In other words, it is not uncertainty about the state of the economy so much as the absence of common knowledge that results in response delays. This approach to explaining response delays helps to explain aspects of the observed effects of monetary policy that optimizing models with sticky prices cannot, including differential speeds of response of prices to different types of shocks, and apparent stickiness of the rate of inflation (as opposed to the price level) under certain conditions.A more sophisticated version of this class of models seeks to explain the degree of precision with which individual variables are monitored by decisionmakers, by assuming a global channel-capacity constraint (a constraint upon the overall number of bits per period of information that can be pro-cessed), subject to which decisionmakers choose an optimal information-collection strategy. This offers the prospect of a theory that yields very tight predictions, and at the same time allows for important changes in information structure in response to changes in economic conditions, including sufficiently dramatic changes in monetary policy. The project considers the design of optimal monetary policies, both when the information structure is taken as given, and when endogenous response of the information structure to the policy regime is taken into account.