Suproteem Sarkar

I’m excited to further study questions related to finance, technology+innovation, and behavioral economics, and to extend the scope and credibility of machine learning in empirical research

You can find the full paper at suproteem.cc/representati...

20.11.2024 20:06 — 👍 3    🔁 0    💬 0    📌 0

Paper

In summary:

I transform economic language into embedding vectors, and show these vectors are informative of perceptions and beliefs

I train LLMs that address credibility issues with ML in empirical research

I study economic mechanisms that drive valuation and misvaluation

20.11.2024 20:06 — 👍 0    🔁 0    💬 1    📌 0

Changes in attention

Valuations can fluctuate when perceptions change—is a customer service firm that adopts a language model really an “AI” firm, or just a “service” firm?

I find that these perception changes relate to selective attention, firm communication, and technology transformations

20.11.2024 20:06 — 👍 2    🔁 0    💬 1    📌 0

Summary of results

Main results:

1️⃣ Embeddings explain valuations + outperform traditional characteristics
2️⃣ Returns reflect changes in how businesses are valued + changes in the perceived business model itself
3️⃣ Some changes in embeddings reflect misperceptions, which generate misvaluation

20.11.2024 20:06 — 👍 1    🔁 0    💬 1    📌 0

Similarity between embeddings

Embeddings also encode similarity between firms—geometric distance relates to established measures of perceived similarity

Taken together, these results demonstrate that a firm’s embedding is informative of its perceived business model

20.11.2024 20:06 — 👍 1    🔁 0    💬 1    📌 0

First, I train new language models on historical data to avoid lookahead bias. I’ve released these LLMs to the research community

Second, I use contrastive representation learning to construct embeddings of firms. The geometry of these vectors relates to economic features of firms

20.11.2024 20:06 — 👍 0    🔁 0    💬 1    📌 0

Summary of procedures

I transform financial news language into embedding vectors

Embeddings put quantitative structure on unstructured data, and have contributed to the success of machine learning over the past decade

20.11.2024 20:06 — 👍 0    🔁 0    💬 1    📌 0

Abstract of paper

Economic valuations fluctuate in ways empirical research cannot fully explain

What information are we missing? Economic theories emphasize the role of hard-to-quantify beliefs and perceptions

My job market paper develops algorithms + measurement to quantify perceptions of firms

20.11.2024 20:06 — 👍 18    🔁 8    💬 1    📌 1