Essays on household finance

概述。家庭的财务决策对于理解经济动态至关重要。然而，考虑到决策问题的复杂性和所需的动态选择，我们对家庭实际决策与最佳理论基准的接近程度的经验理解是有限的。迅速发展的家庭金融领域背后的一个重要驱动力是使用新获得的微观层面数据来更好地识别和衡量这些差异(Campbell，2006)。为了配合这一充满希望的研究议程，我提议的研究项目使用来自英国的微观数据集和回归不连续设计等准实验方法，研究家庭抵押贷款选择和住房。在发达经济体的家庭资产负债表中，住房占家庭财富的大部分，而抵押贷款占家庭负债的最大份额，使这成为一个重要的研究领域(见Badarinza，Campbell和Ramadorai，2016a)。抵押贷款市场具有复杂的合同特征，如当前和未来的利率、费用、抵押贷款期限和再融资成本。这意味着家庭的最佳抵押贷款选择需要在多个维度和时间上进行权衡(参见Stanton and Wallace，1998和Badarinza，Campbell and Ramadorai，2016b)。鉴于这种复杂的选择环境，行为经济学和心理学的最新进展提供了另一种预测，这些预测建立在人类决策本质上是比较的和依赖于背景的发现(例如，Kahnemann和Tversky，2000)。在我的主要项目中，我们与帝国理工学院的Tarun Ramadorai和牛津大学的Pedro Bordalo共同开发了一个抵押贷款选择的简化基准模型，该模型捕捉了家庭面临的关键权衡，并用它来预测家庭对监管干预或竞争带来的抵押贷款市场变化的反应。我们将其与一个模型进行比较，该模型捕获了给定特定选择环境(在本例中为可用抵押贷款菜单)的上下文相关选择。这是基于Bordalo、Gennaioli和Shleifer(2013、2015)的有影响力的工作，他们提出了一个选择模型，在该模型中，个人将注意力集中在最显著的属性上，如质量或价格。他们的模型预测，人们过度关注某些方面，而忽视其他方面，其方式随着选择环境的不同而有系统地变化。在抵押贷款市场，可供选择的成本构成方式可能会让人们更多地关注一些成本(例如费用)，而不是其他成本，这对消费者的选择具有重要影响。数据和方法。我们有关于英国抵押贷款起源的详细微观数据(FCA产品销售数据(PSD))，以及MoneyFaces提供的一段时间内抵押贷款提供的数据(通过我作为英格兰银行学术访问者的从属关系访问)。这使我们能够比较不同市场条件下个人选择的抵押贷款，这些个人在其他方面是相似的，并测试基准模型和基于显著程度的模型的预测。为了补充基于模型的分析，我们将寻求找出抵押贷款菜单的变化(如合并、利率和监管的变化)，这些变化似乎是家庭偏好的外生性变化，并估计这些变化对选择背景对实际消费者选择的影响。了解家庭抵押贷款选择的关键决策变量和家庭在不同宏观经济情景下面临的权衡具有高度的政策相关性：它可以帮助我们更好地了解货币传导渠道的有效性，并帮助校准住房和抵押贷款市场使用的新的宏观审慎措施，这些措施取决于预测家庭反应的能力和各种选择变量之间的关系的稳定性。

Overview. Households' financial decisions are critical for understanding economic dynamics. Yet given the complexity of the decision problems and dynamic choices required, our empirical understanding is limited in terms of how close households' actual decisions come to optimal theoretical benchmarks. An important driver behind the burgeoning field of household finance is the use of newly available micro-level data to better identify and measure these discrepancies (Campbell, 2006). To link up with this promising research agenda, my proposed research project studies household mortgage choice and housing, using micro-level datasets from the UK and quasi-experimental methods such as regression discontinuity designs. Within household balance sheets across advanced economies, housing accounts for the majority of household wealth, while mortgages make up the largest share of household liabilities, making this an important area of study (see Badarinza, Campbell and Ramadorai, 2016a).Mortgage choice and salience. Mortgage markets have complex contract characteristics such as current and future interest rates, fees, term of the mortgage and refinancing costs. This means that households' optimal mortgage choice requires trade-offs over multiple dimensions and time (see Stanton and Wallace, 1998 and Badarinza, Campbell and Ramadorai 2016b for a review). Given this complex choice environment, recent advances in behavioural economics and psychology provide alternative predictions that build on findings that human decision making is inherently comparative and context-dependent (e.g. Kahnemann and Tversky, 2000). In my main project, which is joint work with Tarun Ramadorai (Imperial College) and Pedro Bordalo (University of Oxford), we develop a simplified benchmark model of mortgage choice that captures key trade-offs households face and which we use to predict how households react to changes in the mortgage market, brought about either by regulatory intervention or competition. We compare this to a model that captures context-dependent choice given a specific choice environment (in this case, the available menu of mortgages). This is based on influential work by Bordalo, Gennaioli and Shleifer (2013, 2015) who propose a model of choice in which individuals focus attention on the most salient attributes, such as quality, or price. Their model predicts that people focus excessively on some aspects and neglect others, in ways that vary systematically with the choice environment. In the mortgage market, the way costs are structured across available options is likely to draw more attention to some costs (e.g. fees) rather than others, with important implications for consumer choice.Data and methods. We have detailed micro-level data on the universe of mortgage originations in the UK (FCA Product Sales Data (PSD)), and data by Moneyfacts on mortgage offers available over time (accessed through my affiliation as academic visitor to the Bank of England). This allows us to compare mortgages chosen by individuals in different market conditions, who are otherwise similar, and test predictions from the benchmark model against those of the salience-based model. To complement the model-based analysis, we will seek to identify changes to mortgage menus (as arising from mergers, changes in interest rates and regulation) that are plausibly exogenous to the households' preferences, and estimate the impact of these changes to the choice context on actual consumer choice. Understanding the key decision variables for household mortgage choice and trade-offs that households face under different macroeconomic scenarios is highly policy-relevant: it can help us better understand the effectiveness of the monetary transmission channel, and aid calibration of novel macro-prudential measures used in the housing and mortgage market, which depend on the ability to predict household reactions and the stability of relationships across choice variables.