Publications

You can also find my latest articles on my Google Scholar profile.

1. Cox, S. C., Ettema, M. H., Chambers, L. A., Stephens, S. A., Bodeker, G. E., Nguyen, Q., Diaz-Rainey, I., & Moore, A. B. (2025). Empirical models of shallow groundwater and multi-hazard flood forecasts as sea-levels rise. Earth’s Future, 13(2), e2024EF004977.

   Knowledge of coastal hydrogeology and hazards as groundwater responds to sea-level rise (SLR) can be improved through installation of shallow groundwater monitoring piezometers and continuous observations. Interpolation of site data enables mapping of the present-day state of groundwater elevation, depth to groundwater (DTW), their temporal statistical variation, and differing spatial responses to tides and rainfall. Future DTW and its variability can be projected under increments of SLR, with assumptions and caveats, to show where and when episodic and/or permanent inundation can be expected. This methodology is outlined in a case study of Dunedin, New Zealand, which enabled comparison of rising groundwater's contribution to pluvial flooding and groundwater emergence with coastal inundation. Changes in relative land exposure with SLR shows evolution in flood hazard from current pluvial-dominated events, into “flooding from below” and groundwater emergence, in advance of any overland coastal inundation. Dunedin exemplifies how groundwater transfers effects of SLR surprisingly far inland, but the lowest-lying or shoreline-proximal suburbs are not necessarily the most vulnerable. Unlike coastal inundation, rising groundwater is unconstrained by protective topography and presents as a creeping hazard, or contributor to hazards such as pluvial flooding, which can be widespread, occurring already and difficult to defend against. The empirical models contain assumptions and uncertainties important to the veracity of results and application. While conservative (“risk averse”) and a compromise from computationally expensive numerical solutions, their value is in providing the spatial and temporal precision needed for multi-source hazard assessment and holistic adaptive planning.

2. Nguyen, Q., Diaz-Rainey, I., Kitto, A., McNeil, B., Pittman, N., & Zhang, R. (2022). Scope 3 Emissions: Data Quality and Machine Learning Prediction Accuracy. PLOS Climate, 2(11), e0000208.

   This paper explores the quality of Scope 3 emission data in terms of divergence and composition and the performance of machine learning models in predicting Scope 3 emissions. We do so using the Scope 3 emission datasets of three of the largest data providers (Refinitiv Eikon, and ISS). We find considerable divergence between third party providers, making it difficult for investors to know their ‘ exposure to Scope 3 emissions. Surprisingly, divergence exists between the datasets for emissions values that have been reported by firms (identical data points between Bloomberg and Refinitiv Eikon). The divergence is even larger for ISS when it adjusts reported values using its proprietary models ( identical data points). With respect to the composition of Scope 3 emissions, firms generally report incomplete compositions, yet they are reporting more categories over time. There is a persistent contrast between relevance and completeness in the composition of Scope 3 emissions across sectors, as irrelevant categories such as travel emissions are reported more frequently than relevant ones, such as the use of products and processing of sold products We also find that the application of machine learning algorithms can improve the prediction accuracy of the aggregated Scope 3 emissions (up to 6%) and its components, especially when each category is estimated individually and aggregated into the total Scope 3 emissions values (up to 25%). It is easier to predict upstream emissions than downstream e missions. Prediction performance is primarily limited by low observations in particular categories, and predictor importance varies by category. We conclude that users of the Scope 3 emission datasets should consider data source, quality and prediction errors when using data from third party providers in their risk analyses.

3. Nguyen, Q., Diaz-Rainey, I., & Kuruppuarachchi, D. (2023).In search of climate distress risk. International Review of Financial Analysis, 85, 102444.

   Using the Merton distance to default model we investigate whether a firm's climate risk affects its default (distress) risk. S&P 500 non-financial firms during 2010–2019 are analysed and we employ both corporate carbon footprints and climate risk disclosures in annual filings to measure climate risk. Our results show that climate risk has a negative impact on firms' distance to default. This impact is limited to the disclosure of transition risk in annual filings. In contrast, disclosures of physical or non-specific risk do not affect firm-level default risk, while the impact of corporate carbon footprints is inconsistent but insignificant in most models. We also find that the negative effect of climate transition risk on firms' distance to default is stronger for firms headquartered in states with carbon pricing (California and states covered by the Regional Greenhouse Gas Initiative) and temporarily strengthens because of the Paris Agreement in 2015. However, this ‘Paris’ effect is short-lived and fades away in subsequent years.

4. Nguyen, Q., Diaz-Rainey, I., Kuruppuarachchi, D., McCarten, M., & Tan, E. K. (2023).Climate transition risk in US loan portfolios: Are all banks the same?. International Review of Financial Analysis, 85, 102401.

   We examine banks' exposure to climate transition risk using a bottom-up, loan-level methodology incorporating climate stress test based on the Merton probability of default model and transition pathways from the Intergovernmental Panel on Climate Change (IPCC). Specifically, we match machine learning predictions of corporate carbon footprints to syndicated loans initiated in 2010–2018 and aggregate these to loan portfolios of the twenty largest banks in the United States. Banks vary in their climate transition risk not only due to their exposure to the energy sectors but also due to borrowers' carbon emission profiles from other sectors. Banks generally lend a minimal amount to coal (0.4%) but hold a considerable exposure in oil and gas (8.6%) and electricity firms (4.6%) and thus have a large exposure to the energy sectors (13.5%). We observe that climate transition risk profile was stable over time, save for a temporary (in some cases) and permanent (in others), reduction in their fossil-fuel exposure after the Paris Agreement. From the stress testing, the median loss is 0.5% of US syndicated loans, representing a decrease in CET1 capital of 4.1% when extrapolated to the whole balance sheet. The loss is twice as large in the 1.5°C scenarios (1.4%–2.1% of loan value, 12%–16% of CET1 capital) compared to the 2°C target (0.6%–1.1% of loan value, 5%–9% of CET1 capital) with significant tail-end risk (7.7% of loan value, 62% of CET1 capital). Banks' vulnerabilities are also driven by the ex-ante financial risk of their borrowers more generally, highlighting that climate risk is not independent from conventional risks.

5. Nguyen, Q., Thorsnes, P., Diaz‐Rainey, I., Moore, A., Cox, S., & Stirk‐Wang, L. (2022).Price recovery after the flood: Risk to residential property values from climate change‐related flooding. Australian Journal of Agricultural and Resource Economics, 66(3), 532–560.

   We take advantage of a combination of a severe weather event from 3 to 4 June 2015 and a local policy, to investigate the housing market response to climate change-related flooding hazard. The study focuses on a residential area in a low-lying coastal suburb of Dunedin, New Zealand, where the groundwater level is shallow and close to sea level. An unusually heavy rain event in June 2015 resulted in flooding of a significant portion of land in especially low-lying areas. The city council responded by reviewing processes for storm-water management and by imposing minimum-floor-level [MFL] requirements on new construction in the low-lying areas previously identified as at risk of flooding. Applying a ‘diff-in-diff-in-diff’ strategy in hedonic regression analyses, we find that houses in the MFL zone sell for a discount of about 5 per cent prior to the flood. This discount briefly tripled in the area that flooded, but disappeared within 15 months, indicating either very short memory among homebuyers or no long-run change in perception of hazard.

6. Nguyen, Q., Diaz-Rainey, I., & Kuruppuarachchi, D. (2021).Predicting corporate carbon footprints for climate finance risk analyses: A machine learning approach. Energy Economics, 95, 105129.

   Corporations have come under pressure from investors and other stakeholders to disclose and reduce their greenhouse gas emissions (GHG). Corporate GHG footprints, proxying for transition risk, are dominated by carbon emissions from energy use. Thus the growing attention on the carbon emissions of corporations from, principally, their energy use, motivates firms to invest in energy efficiency and renewable energy. However, only a subset of corporations disclose their GHG/carbon footprints. This paper uses machine learning to improve the prediction of corporate carbon emissions for risk analyses by investors. We introduce a two-step framework that applies a Meta-Elastic Net learner to combine predictions from multiple base-learners as the best emission prediction approach. It results in an accuracy gain based on mean absolute error of up to 30% as compared with the existing models. We also find that prediction accuracy can be further improved by incorporating additional predictors (energy production/consumption data) and additional firm disclosures in particular sectors and regions. This provides an indication of where policymakers should concentrate their efforts for greater level of disclosure.