Market Based Climate Stress Tests

气候的变化也将改变投资者面临的风险和机遇。如果银行面临气候变化的风险，那么这些风险可能会影响金融市场的运作。该项目将评估银行和其他金融机构对气候变化风险的敞口。在仔细测试了这些方法的准确性和可靠性后，结果将每周更新一次，并免费发布在互联网上，供监管机构、投资者和政策制定者查看。这些估计将帮助监管机构评估金融体系在发生气候相关冲击时的韧性。这个项目采用了一种全新的方法来对气候变化进行压力测试。它关注的是银行贷款等资产的价格，而不是实际损害。这样的银行压力测试可以显示出气候变化对损害的风险敞口，这种损害在20多年内不会发生，但可能会在短期内让一家银行破产，因为资产价格今天可能会下跌，以应对有关遥远未来的坏消息。这些敞口指标本质上是一种方式，通过检查银行股票回报与气候风险因素之间的相关性，来了解银行应对气候相关风险的弹性有多大。事实上，这一观察结果通过考察投资回报与一系列气候风险因素之间的相关性，来衡量任何投资的“绿色性”。该项目将应用最先进的统计程序来衡量金融机构对气候风险的敞口。该程序首先制定一个反映气候变化风险的投资资产组合。如果气候变化变得更加严重，这一投资组合的价值将会下降。这一投资组合的几个候选者基于过渡风险或实物风险或其他版本的气候风险。为了评估银行的风险敞口，银行股票市场估值的变化与气候风险投资组合进行了回归，以确定它们在决定银行价值变化方面的重要性。这种回归允许系数按照Engle(2016)中的动态条件Beta(DCB)框架变化。为了确保风险敞口的准确衡量，该项目将a)开发计量经济学来估计动态条件贝塔模型中的可信区间，b)使用气候效率因子模拟可持续销售的公开可用基金的投资组合来制定额外的气候风险衡量标准，c)确定哪个投资组合是过渡风险的最佳衡量标准，d)考虑金融机构的其他风险因素，e)纳入细粒度监管数据以确定市场衡量标准反映实际银行持有量的程度，f)比较VRI模型和主要反映NGFS方法的监管措施的风险敞口排名。这项研究的一个关键特点是使用来自世界不同地区央行的机密监管数据。这些独特的数据来源加在一起，将有助于证实和改进计量经济学的估计。只需通过等级相关性就可以将调查结果与主管的调查结果进行比较。然而，更有趣的比较可能涉及理解哪里存在差异。例如，使用一组曝光来预测另一组，并包括特征数据将显示差异最重要的地方。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

A changing climate will also create changes in the risks and opportunities faced by investors. If banks are exposed to the risks of climate change, then these risks could potentially affect how financial markets function. This project will evaluate the exposure of banks and other financial institutions to climate change risks. After these methods are carefully tested for accuracy and reliability, the results will be updated weekly and posted freely on the internet for regulators, investors and policy makers to see. These estimates will help regulators assess financial system resilience in the event of climate related shocks. This project takes a completely new approach to stress testing for climate change. It focusses on prices of assets such as bank loans, rather than actual damages. Such bank stress tests can show exposure to climate change from damages that will not occur for more than twenty years but which could bankrupt a bank over a short period of time since asset prices can fall today in response to bad news about the distant future. These exposure measures are essentially a way to see how resilient banks are to climate related risks by examining the correlation between its stock return and climate risk factors. In fact, this observation leads to the measurement of the “greenness” of any investment by looking at the correlation between its return and a suite of climate risk factors. This project will apply state of the art statistical procedures to measure the exposure of financial institutions to climate risks. The procedure first formulates a portfolio of investment assets that reflects the risks of climate change. If climate change becomes more severe, this portfolio will fall in value. Several candidates for this portfolio are based on transition risks or physical risks or other versions of climate risk. To assess the exposure of banks, the change in a bank’s stock market valuation is regressed on the climate risk portfolios to see how important they are in determining the change in value of the bank. This regression allows the coefficients to change following the Dynamic Conditional Beta (DCB) framework as in Engle (2016). To be certain that this accurately measures the exposure, the project will a) develop the econometrics to estimate the confidence intervals in the dynamic conditional beta model, b) develop additional climate risk measures using climate efficient factor mimicking portfolios of publicly available funds marketed as sustainable, c) determine which portfolio is the best measure of transition risk, d) consider other risk factors for financial institutions, e) incorporate granular supervisory data to determine the extent to which market measures reflect actual bank holdings, f) compare the exposure rankings from the VRI model and from supervisory measures which mostly reflect NGFS methodology. A key feature of this research involves the use of confidential supervisory data from central banks in different parts of the world. Together these unique data sources will help to corroborate and improve the estimates from the econometrics. Comparison of the findings with those from supervisors can be carried out simply by rank correlations. However more interesting comparisons may involve understanding where there are differences. For example, using one set of exposures to predict another and including characteristics data will show where the differences are most important.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.