Jiannan (Jay) Jiang
I am a second-year Ph.D. student in Economics at The University of Texas at Austin (UT-Austin).
My research interests are in macroeconomics, as well as the use of textual data and machine learning in economics.
You can reach me by jiannanjiang@utexas.edu. Here is my CV.
Working Paper
Inflation Drivers in Firms' Words: Bridging Micro Narratives and Macro Dynamics
[Working Paper] [SSRN]Abstract: This paper develops a framework that links microeconomic narratives to structural macro models of firm pricing. Using the Federal Reserve Bank's Beige Book data, we employ an LLM to extract qualitative measures of firms' price adjustments and factor attributions at the micro level. These measures are then used to estimate a state-space model of aggregate inflation dynamics implied by a menu-cost model featuring lumpy price adjustments and reporting frictions. We demonstrate how micro price narratives provide useful information for correctly decomposing the macro drivers of inflation.
Presentations: BSE Summer Forum (2025 cancelled; 2026); CEF 2025; IAAE 2025, 2026; ifo 2025; Midwest Macro Spring 2025; SED 2026.
Working Paper (Computer Science)
Long-Horizon Forecasting of Complete Financial Statements with Forma
† These authors contributed equally.
[Working Paper] [SSRN] [ProForma-20Q Benchmark] [Forma Code]Abstract: Specialist training beats generalist scale when forecasting financial statements. To our knowledge, no prior work jointly forecasts complete financial statements beyond one year, yet in a discounted-cash-flow valuation most firm value sits past that window. We release ProForma-20Q, a reproducible benchmark for forecasting 78 statement line items 1–20 quarters ahead, for anonymized firms, from past statements and an industry code, scored by change-space R2. On it, Forma, a transformer that reads statements as sets of (account, quarter, value) tuples and maximizes a masked-tuple Gaussian likelihood, beats every competitor we field: classical machine learning, chained gradient boosting, a zero-shot time-series foundation model, and frontier large language models. Its lead widens with horizon, where valuation needs accuracy most, and its Gaussian predictive intervals never under-cover. Forma's forecasts nearly satisfy accounting identities; exact coherence is recoverable at no statistically significant accuracy cost. Its tuple interface supports scenario analysis without retraining, and we show that pinning future revenue paths sharpens the rest of the statement.
Projects
- We develop ProForma-20Q, a benchmark for long-horizon forecasting of complete financial statements, and release the model architecture, training algorithm, and weights in the Forma code repository.
- I develop dynare-copilot with Zhenkun Lu, an open-source AI skill for DSGE modeling.
Teaching
- Teaching Assistant at UT-Austin: Economic Statistics, Mathematical Microeconomic Theory with Advanced Applications, Introduction to Econometrics, Introduction to Microeconomics.
- Teaching Assistant at CUHK(SZ) for master’s-level courses: Natural Language Processing and Learning with Text, Advanced Econometrics, Advanced Microeconomics, Mathematics in Economics.
- Undergraduate Student Teaching Fellow at CUHK(SZ): Advanced Macroeconomics.
Awards
- Presidential Award for Outstanding Graduate Students, CUHK(SZ), 2024.
- First Class Academic Excellence Scholarship (Ranked 1st for GPA=4.0), CUHK(SZ), 2024.
The meaning of a word is its use in the language. — Wittgenstein