Climate Change and Residential Mortgage Lenders

Kivonat, rövid leírás

The dissertation studies the linkage between residential mortgage lending and local climate projections by directly linking future climate projections to current lender behaviour, using climate data new to the finance literature. The topic has hitherto received little attention in the scientific climate finance discourse, with the existing research focussing on sea level rise risk only or reactions to natural disasters or extreme weather. Yet there are a number of reasons why the topic is important. First, US mortgages are crucial to financial cycles, and issues in the mortgage market can quickly affect other credit markets and cross borders. Evidence shows that the US real estate market has not fully incorporated climate change. Sudden risk reassessments could, therefore, lead to financial instability and reduced credit availability. Second, local climate prospects coupled with lenders' practices also affect social inequality. Third, climate risk mitigation through securitisation to government-sponsored enterprises (GSEs) has a number of policy implications. Fourth, the financial sector has an indirect influence on climate change, highlighting the need for lenders to consider environmental impacts in their decisions. The two datasets that form the cornerstone of the study are: i) climate change projections data from a downscaled version of global climate models that are also used in the UN IPCC reports and ii) HMDA mortgage data, created by Congress which is the most comprehensive publicly available database for US mortgages. The dissertation presents evidence in three case studies. Two have already been published in journals (one of which is a leading Q1 journal in the area of climate change). Case study 1: Loan amounts and rejection rates – extreme heat Research questions: How does the volume of mortgages originated in the US counties that are most vulnerable to future heat waves, compare to such counties' share of land area, economic importance and population? What do we know about supply and demand effects in lending patterns? Methodology: To answer the second question, I construct simple denials rates (Duan and Li, 2019) and the sophisticated denials index (Key and Mulder, 2020). The latter is a measure of how application denials have evolved across counties and years, beyond the known loan-level characteristics. Results: the mortgage share of areas exposed to future heatwaves is higher than their land share, and this appears to be linked to their greater economic activity and higher population. In fact, lenders deny slightly more loan applications in these areas, which appears to suggest that it is not a greater lending appetite that is behind the higher lending volumes. Case study 2: Interest rates and loan terms – extreme heat Research questions: Are interest rates higher and loan terms shorter in areas that are more exposed to climate change, controlling for other variables? Do we see additional concerns reflected in mortgage characteristics at the extremes of projected levels of hot days? Do climate change concerns appear more pronounced in the mortgage rates of certain lenders? Methodology: Following the literature, in the baseline specification I use a linear regression equation with OLS estimation for the rate spread, and a probit model for the probability of a sub-standard loan term. As a robustness check, I follow the IV/2SLS approach (Ambrose et al., 2018). Results: Considering a range of controls and potential sources of bias, I find that larger projected increases during the coming decades in the number of hot days are associated with higher rate spreads and an increased probability that loan terms are shorter than the standard 30 years. In counties projected to experience an extreme number of hot days, both the rate spread and the probability of a short loan term are higher still. It is lending from non-banks, rather than banks, that appears sensitive to the changing climate. Case study 3: Interest rates and loan terms – extreme heat, drought and flooding Research questions: Are GSE onselling rates (the proportion of originated mortgages under GSEs’ conforming limit sold on to GSEs, controlling for other factors) higher in areas most exposed to climate change? Has this relationship changed in the past few years? Is there evidence of firm heterogeneities in GSE on-selling activity with respect to climate change exposure? Methodology: The main methodology applies a difference-in-difference style estimator with the synthetic control method for matching. For robustness I also apply nearest neighbour matching, pooled OLS and FE panel regression. Results: Both banks and independent mortgage companies have sold proportionately more loans to GSEs in areas that are most exposed to the changing climate – based on my climate change indicator encompassing risks of extreme heat, drought and flood. The observed relationship can be traced back to 2013 but is more marked since 2016 when granular climate change projections became public. Some implications of the dissertation/ further research avenues:  Some risk incorporation has happened, it would be useful to ascertain whether the extent of it is sufficient. Banks, in particular, are worth focussing on from a regulatory standpoint.  It would be useful to investigate who is ultimately backing the climate risk transferred to GSEs – the taxpayer, households in areas less exposed to the risk or others.  Incorporate the the financial stability implications of residential mortgage lenders' climate change exposure in economic models.  Policy implications of how to manage the social impact of changing conditions in housing financing in the most climate-exposed areas.