Mitsui Finance Seminar Series - Winter 2025

These seminars are sponsored by the Mitsui Life Financial Research Center.

If you would like to be added to the email distribution list, please contact Gabriella Ring at [email protected].


march 28

Savitar Sundaresan, Imperial College London

Title: Extensive Attention, Intensive Attention and the Origins of Random Choice

Abstract: Using repeated choices and eye-tracking data across 180 menu instances from 50 subjects we link the randomness of choice to two notions of attention: extensive attention (what options are looked at) and intensive attention (how long options are looked at). We show that models that seek to explain randomness through attention should capture four key facts.  First, extensive attention and intensive attention are both related to randomness in choice, although intensive attention, on average, is a better predictor of choice.  Second, the mechanism of choice, and the degree of randomness, is very different for large compared to small choice sets.  Third, the relative importance of attention in generating randomness in choice is smaller between-person than within-person. Fourth, greater attentional capacity is not associated with reduced randomness at the individual level.  Fifth, increased attention is not strongly correlated with better choices, indicating additional attention may be deployed in situations where higher randomness is already likely.

Time: 10:30-11:50 a.m.        
Location: R2230


april 4

Stijn van Nieuwerburgh, Columbia

Title:  The Commercial Real Estate Ecosystem

Abstract: We develop a new approach to understand the joint dynamics of transaction prices and trading volume in the market for commercial real estate. We start from a micro-founded model in which buyers and sellers differ in their private valuation of building characteristics, such as size, location, and quality. Consistent with the decentralized nature of the commercial real estate market, we model the probability that a seller meets a particular buyer, where the meeting probability depends on the characteristics of the buyer, the seller, and the building. In equilibrium, the mapping from building characteristics to observed transaction prices depends on the identity of the buyer and the seller, an important property missed by traditional hedonic valuation models. We estimate the model using granular data on commercial real estate transactions, which contain detailed information on the identity of buyers and sellers. Our central finding is that the identity of buyers and sellers has a first-order effect on both property valuation and the likelihood of trade. The importance of investor characteristics for valuations remains true, in fact is amplified, in a rich machine learning model that allows for non-linearities and interactions. We show how the model can be used for out-of-sample predictability and for counterfactual analyses on investment flows and prices. As a concrete example, we find that the Manhattan office market would have seen 7% lower valuations if it had not been for a large inflow of foreign buyers in 2013–2021. Our methodology extends to other private markets, including private equity, private credit, and infrastructure.

Time: 10:30-11:50 a.m.        
Location: R2230


april 11

Ralph Koijen, Chicago Booth

Title:  Asset Embeddings

Abstract: Firm characteristics, based on accounting and financial market data, are commonly used to represent firms in economics and finance. However, investors collectively use a much richer information set beyond firm characteristics, including sources of information that are not readily available to researchers. We show theoretically that portfolio holdings contain all relevant information for asset pricing, which can be recovered under empirically realistic conditions. Such guarantees do not exist for other data sources, such as accounting or text data. We build on recent advances in artificial intelligence (AI) and machine learning (ML) that represent unstructured data (e.g., text, audio, and images) by high-dimensional latent vectors called embeddings. Just as word embeddings leverage the document structure to represent words, asset embeddings leverage portfolio holdings to represent firms. Thus, this paper is a bridge from recent advances in AI and ML to economics and finance. We explore various methods to estimate asset embeddings, including recommender systems, shallow neural network models such as Word2Vec, and transformer models such as BERT. We evaluate the performance of these models on three benchmarks that can be evaluated using a single quarter of data: predicting relative valuations, explaining the comovement of stock returns, and predicting institutional portfolio decisions. We also estimate investor embeddings (i.e., representations of investors and their strategies), which are useful for investor classification, performance evaluation, and detecting crowded trades. We discuss other applications of asset embeddings, including generative portfolios, risk management, and stress testing. Finally, we develop a framework to give an economic narrative to a group of similar firms, by applying large language models to firm-level text data.

Time: 10:30-11:50 a.m.        
Location: R2230


april 25

Michael Ewens, Columbia

Title: Corporate Hierarchy

Abstract: We introduce a novel measure of corporate hierarchies for over 2,600 U.S. public firms. This measure is obtained from online resumes of 16 million employees and a network estimation technique that allows us to identify hierarchical layers. Equipped with this measure, we document several facts about corporate hierarchies. Firms have on average ten hierarchical layers and a pyramidal organizational structure. More hierarchical firms have a more educated workforce, higher internal promotion rates, and longer employee tenure. Their operating performance is higher, but they face higher administrative costs. They are more active acquirers and produce more patents, but not higher-quality patents. They exhibit lower stock return volatility and more stable cash flows. We also examine how companies adjust their hierarchies in response to demand and knowledge shocks. We find that biotech companies increased their number of layers following the Covid-19 pandemic, while companies flatten their hierarchies following the adoption of artificial intelligence (AI) technologies. These findings are consistent with the theoretical predictions of existing models of corporate hierarchies.

Time: 10:30-11:50 a.m.        
Location: B1590 Corner Commons