Session 5: Climate Finance and Banking

The analysis of policy impacts in a dynamic and uncertain reality is vital to supporting informed economic policy design and implementation. Dynamic, stochastic economic models used in policy evaluation necessarily simplify the world as we know it. This motivates us to explore, refine, and extend tools aimed at producing marginal valuations that shed light on why some policies are optimal and how others, though suboptimal, can be improved. We present novel representations of these marginal valuations that embrace uncertainty and support robust implementation—even in environments characterized by “deep uncertainties.” These representations offer a more complete understanding of how interactions among multiple state variables, concerns about model misspecification, and uncertainties surrounding potentially long-term implications contribute to the cogent assessment of policies. We argue that these methods are particularly salient for evaluating the global cost of climate change and the global value of research and development with long-term prospects for success.

We study competition and pass-through in the market for retail deposits. Federal funds rate increases are associated with only mini-mal increases in deposit rates. Leveraging new data on offers mailed by banks to households, we show that federal funds rate increases are strongly associated with changes in mail volumes, sign-up bonuses, and offers targeting new customers. These margins and not interest rates are the primary ways in which interest rate changes affect the deposit market. We rationalize the use of these margins and not interest rates in a simple model with active and sleepy depositors. Depositor heterogeneity and adverse selection, as opposed to market power, are the primary reasons why banks offer far less than the full present value of future rate spreads to depositors.

How do large, national credit card lenders affect interest rates, the size of the credit card market, and household welfare? We answer this question by developing and estimating a novel theory in which national lenders strategi-cally influence market-level and economy-wide interest rates. Our theory pre-dicts that pass-through rates from bank funding costs to credit card spreads depend on the bank’s footprint within a market (market-level shares) as well as the bank’s national footprint (national-level shares). To estimate and test the model’s predictions, we first define U.S.-wide credit card markets using expert and regulatory reports on market segmentation. We then build a daily panel of millions of credit card originations from OCC and Y-14M data cover-ing the 2008-2019 time period, and we measure how banks pass through high-frequency monetary rate changes depending on their market- and national-level shares. We find pass-through rates are 15% lower when banks have large market-level shares and 20% lower for large national-level shares. Lastly, we use these moments to discipline market- and national-level rate setting power in our model and measure its macroeconomic effects. Moving from status quo to perfect competition reduces credit card interest rates by 1.6pp, increases credit-to-GDP by 2.9pp, and yields a consumption equivalent gain between 0.03% and 0.11%.

We build a general equilibrium model to study how climate transition risks affect energy prices and the valuations of different firms in the energy sector. We consider two types of fossil fuel firms: incumbents that have developed oil reserves they can extract today or tomorrow, and new entrants that must invest in exploration and drilling today to have reserves to potentially extract tomorrow. There are also renewable energy firms that produce emission-free energy but cannot currently serve non-electrifiable sectors of the economy. We analyze three sources of climate transition risk: (i) changes in the probability of a technological breakthrough that allows renewable energy firms to serve all economic sectors; (ii) changes in expected future taxes on carbon emissions; and (iii) restrictions on today's development of additional fossil fuel production capacity. We show that different transition risks—and, importantly, uncertainty about their realizations—have distinct effects on firms' decisions, on their valuations, and on equilibrium energy prices. We provide empirical support for the heterogeneous effects of different transition risks on energy prices and stock returns of firms in different energy sub-sectors.

This paper uses spatiotemporal convolutional neural networks (ST-CNNs) to forecast wildfire risk across the state of California. ST-CNNs capture spatial and temporal dependencies interactively in spatiotemporal panel data like wildfires. We find that ST-CNN significantly outperforms both logistic regression and LSTM, a baseline ML model for time-series forecasting. ST-CNN successfully recalls the largest wildfires and corresponds well to the aggregate zip-code risks of California wildfires. Our data include 6,999 wildfires in California from 2001 through 2021 and their effects on the universe of over 7 million single-family residences. Out-of-sample estimates of the probability of wildfire and aggregate losses to the residential tax basis provide disturbing evidence concerning wildfire risks to California homeowners, lenders, and California's largest Property and Casualty (P&C) insurer, State Farm General. Finally, we discuss the urgent need for continued research on improving forward-looking wildfire-occurrence models since the California Department of Insurance approved their use in December 2024.

Using advanced machine learning methods, we construct a comprehensive, nearly-error-free database of the universe of approximately 12 million U.S. patents from 1836 to 2024. We analyze the resilience of innovation to disaster shocks using hurricane landfall data spanning two centuries. Major hurricanes destroy local innovative capacity for up to a decade and lead to permanent losses relative to the counterfactual. Relative to individual inventors, firm-based innovation is more resilient to disasters. Innovative capacity is not perfectly substituted across regions, leading to aggregate innovation loss. Our findings reveal that the capital destruction and capital constraints stemming from natural disasters induce large, long-lasting innovation losses and can explain regional variation in economic outcomes.

In a world with rising risk, how much are U.S. households willing to pay for homeowners insurance, and what does their demand imply for the future of insurance markets? We provide the first estimates of household willingness to pay for homeowners insurance and the drivers of household insurance demand elasticities by exploiting quasi-exogenous regulatory shocks to insurance pricing. We utilize newly available individual-level data on homeowners insurance contracts covering the entire United States for over a decade, with rich information on mortgage contracts, property characteristics, and climate exposures. We document pervasive underinsurance, particularly among the most financially vulnerable households. We find that even at actuarially fair premiums, households’ willingness to pay is below expected losses, and demand remains elastic—results that are inconsistent with the textbook models of insurance demand. Financial constraints are a key force: constrained households reduce coverage as premiums rise, while unconstrained households borrow more to maintain insurance coverage. Exogenous increases in the cost of credit also reduce coverage demand. These results raise the concern that f inancial constraints reduce willingness to pay for insurance even below the actuarially fair price required for insurers to remain solvent, suggesting that the market may disappear for the most constrained, financially vulnerable households. If prices were to continue growing at historical rates moving forward, our estimates imply that between 17% to 31% of households would hit binding LTV constraints and be forced to reduce coverage substantially, meaning insurance markets may shrink even as losses from natural disasters rise.

We quantify ambiguity aversion by studying how real estate markets respond to long-run sea-level rise (SLR) risk. Using a novel high-resolution dataset covering over two million coastal properties, we show that housing prices reflect not only expected SLR risk but also deep uncertainty across climate scenarios. Leveraging a canonical asset-pricing framework, we recover one of the first market-based estimates of ambiguity aversion, finding substantial tilting of probability weights toward worst-case outcomes. We also estimate a relatively low long-run discount rate, suggesting strong valuation of distant climate risks. As an application, we quantify how ambiguity aversion amplifies the willingness to invest in SLR adaptation.