Research

Investing in energy efficiency is crucial for a low-carbon economy, particularly in the building sector. Despite various subsidy programs, meeting energy targets is challenging because households do not invest sufficiently. Here we study homeowners' low levels of energy efficiency retrofits. We use narratives, an emerging method based on open-ended survey responses, to identify the barriers and determinants behind renovation decisions. Using Natural Language Processing, we transform narratives into quantifiable metrics. While financial considerations are a major barrier, homeowners' main reasons for renovating are unrelated to energy savings. Most homeowners delay energy-saving investments until their buildings require renovations. Co-benefits like environmental concerns and comfort gains are equally or more important than financial motivations. Many homeowners are unaware of existing policies and would favor reducing the bureaucracy of retrofits. Subsidies, while popular, are likely to be mistargeted. Effective policies should also consider institutional factors such as bureaucratic burden and accessibility of information.

We provide an empirical analysis on the relation between culture and revealed environmental preferences. Switzerland's citizens share the same set of institutions but belong to multiple population groups, which differ by culture and language across distinct geographical locations. This unique setting allows us to disentangle the effect of culture on individual consumer preferences from institutional characteristics. We analyze the effect of culture on energy efficient vehicle registration, using municipality level data and applying a spatial fuzzy Regression Discontinuity Design at the internal French/German language border. Our results indicate that French-speaking municipalities have a 3 to 6 percentage points higher share of energy efficient vehicles, compared to their German-speaking counterparts. These findings suggest that French-speakers place a higher value on the environment, which may be due to their higher sense of collectivism and altruism. 

This paper introduces the concept of sustainable finance literacy. We survey a large sample of Swiss households and measure financial, sustainability, and sustainable finance literacy using two complementary approaches. First, we use traditional multiple-choice questions, and second, a novel approach based on open-ended questions that ask respondents to write a text response. We find that Swiss households, which are generally highly financially literate by international standards, exhibit low levels of sustainable financial literacy. Interestingly, multiple-choice questions lead to a gender gap, with women performing worse than men. However, this difference disappears when open-ended questions are used. Moreover, despite its low level, knowledge about sustainable finance turns out to be a highly significant factor for the ownership of sustainable products. Therefore, our results show that there is an urgent need to create transparent regulatory standards and to strengthen information campaigns about sustainable financial products.

This paper develops a novel topic model for text data by conditioning on observables, named the "Conditional Topic Allocation" (CTA). This data-driven method allows identification of latent topics that explain other observable variables. It is particularly suited for small-scale text data, such as open-ended survey responses. First, CTA is used to extract topics from open-ended text answers that explain single observable variables. Then, in empirical models where text is a control variable for unobservable characteristics, CTA's flexible scope of applications allows uncovering latent variables from the text. As a proof-of-concept, this approach is used to analyze the sentimental value homeowners place on their homes and how this relates to energy-efficiency valuation. Specifically, responses from open-ended survey questions are used as control variables in a hedonic regression, and CTA serves to identify latent preferences associated with nearly 50% of the valuation of energy efficiency for single-family houses.

Large Language Models (LLMs) have made significant progress in recent years, achieving remarkable results in question-answering tasks (QA). However, they still face two major challenges at inference time: hallucination and outdated information. These challenges take center stage in critical domains like climate change, where obtaining accurate and up-to-date information from reliable sources in a timely fashion is essential. To overcome these barriers, one potential solution is to provide LLMs with access to external, scientifically accurate, and robust sources (long-term memory). This access assists LLMs in continuously updating their knowledge and preventing the propagation of inaccurate, incorrect, or outdated information. In this study, we enhanced GPT-4 by providing it access to the Sixth Assessment Report of the Intergovernmental (IPCC-AR6), the most comprehensive, up-to-date, and reliable source in this domain. We present our conversational AI prototype, available at www. chatclimate.ai. The product answers challenging questions accurately in three different QA scenarios: asking from 1) GPT-4, 2) ChatClimate, and 3) hybrid ChatClimate. The answers and their sources were evaluated by a team of IPCC authors, who used their expert knowledge to score the accuracy of the answers from 1 (very-low) to 5 (very-high). The evaluation showed that the Standalone assistant provided more accurate answers.