I am an applied economics undergraduate at Beijing Normal-Hong Kong Baptist University (BNBU), Class of 2027. My work runs on two tracks that eventually meet — environmental and public policy on one side, finance on the other — with cities as the ground where both become visible. Across all three, I care less about the elegance of a model than about the mechanism inside it: what actually changes a firm's behavior, a market's price, or a household's choice, and whether the claim holds up once real data is allowed to argue back.
Cities & space城市与空间
Rosen–Roback · hedonic valuation · agglomeration
My most rigorous coursework to date: the core structural models of urban economics — spatial equilibrium and the wage–rent trade-off, hedonic price functions, agglomeration economies — including the quantitative bid-rent and valuation problems they generate. The same fascination runs through Block Detective, which turns any city block into a bilingual analytical case file.
Environmental policy & the firm环境与政策
Regulation · abatement decisions · the price of clean air
I grew up in Xingtai, an industrial city in southern Hebei that spent years on the lists of China's most polluted cities — and the years after getting off them. That turnaround is why I read environmental policy as a lived system rather than a textbook chapter: what a tightening rule actually does inside firms, who absorbs the compliance costs, and why the same regulation reads differently in a mayor's office and at a factory gate. From 2025-2026 my internship in the environmental business brings me first-hand data on this agenda — see Beyond.
Markets & finance金融与市场
Factor investing · macro-financial linkages · how markets digest policy
The other half of my training treats markets as the fastest recording device of expectations ever built. I am most interested in the junction where policy and macro information become prices: how a credit impulse, a regulatory shock, or a five-year plan shows up in factor returns. This agenda has a working instrument — a personal quant research system that replicates published factor research end-to-end before extending it — described over on Projects.
Crowding In: How China's Affordable Rental Housing Program Affected Private Rents保障性租赁住房与私人租金市场
Independent research · under review at The Journal of Real Estate Finance and Economics (Springer)
Hand-collected programme targets for 40 key cities under China's latest Five-Year Plan — each verified against official documents, planned intensity varying thirty-fold — assembled into a 74-city × 98-month private-rent panel (7,209 observations), matched with house-price indices and demographic and GDP controls. A continuous-dose difference-in-differences design — 28 pre-treatment coefficients at zero, randomization inference, a specification curve across 30 configurations — finds that a one-standard-deviation rise in programme intensity lifts private rents by 4.0–4.3%, concentrated in high-rent, high-inflow cities: demand-side crowding-in, the opposite sign of the American subsidy literature. Read the preprint ↗
Wavelet-Driven Multi-Model Ensemble: A Synthesis Box for Time Series Forecasting小波驱动的多模型集成预测
ACAI 2024 (IEEE) · second author · EI-indexed conference paper
A wavelet-decomposition ensemble framework combining FFT, ARIMA and AdaBoost for multi-city time series forecasting — SMAPE down to 0.8%, goodness of fit 0.818. My part was the data: multi-city series preparation and feature construction, with smoothing-window and lag-structure comparisons that set up the model evaluation pipeline. IEEE Xplore ↗
Air Quality Analysis and Prediction Based on the Adaboost-VAR ModelAdaboost-VAR 空气质量分析与预测
ICDA 2024 (IEEE) · second author · EI-indexed conference paper
A hybrid of AdaBoost and vector autoregression that outperforms single VAR/ARIMA baselines in forecasting PM2.5, PM10, NO2 and SO2 across multiple cities. I handled air-quality data cleaning, smoothing-window design and VAR lag-order selection — my first full pass through multivariate time-series modeling, benchmark comparison and robustness checks. IEEE Xplore ↗
Underneath all three terrains sits the same plumbing — applied econometrics, causal inference, reproducible pipelines in Stata and Python. I treat methods as instruments, not identity: they matter exactly as much as the question they serve.
A full CV (education, coursework, skills, activities) is available as PDF — placeholder until the final version is exported.
→ assets/cv.pdf — TODO(上线前)：导出正式 CV PDF 放入 assets/