Yifan (Freda) Fang

Research

Do Market Forces Reduce Carbon Leakage? Evidence from the U.S. Electric Utility Sector

Presentations: Rotman School of Management (2023), MFA Regular Session (2024), NFA PhD Session (2024), Rotman Sustainable Finance Research Roundtable (2025), EsFA (2025), Inaugural HKU Governance and Sustainability PhD Workshop (2025)

Abstract: I use the setting of U.S. power plant divestitures to test whether customer demand for green products affects a firm’s environmental performance. Leveraging divestitures from 2004 to 2020, I show that firms, on average, divest more carbon-intensive plants, and divested plants experience a 17% rise in total emissions post-transaction. However, this leakage effect is mitigated when plants face both heightened demand for low-carbon energy and competition. The demand channel is also reflected in prices: power producers operating in deregulated markets gain pricing power by selling renewable energy, with higher prices associated with stronger demand. Overall, my results suggest a potential role for green energy demand in improving power producers’ environmental performance.

Venture Fraud, with Alexander Dyck, Camille Hebert, and Ting Xu

Presentations: IPC Spring Research Symposium (2026), Bretton Woods Accounting and Finance Ski Conference (2026),UD Weinberg-ECGI Symposium (2026), ASU Sonoran Winter Finance Conference (2026), HEC Paris Entrepreneurship Workshop (2025), Rotman School of Management (2025)

Abstract: We assemble the first comprehensive sample of venture fraud cases involving 614 U.S. venture capital (VC)-backed startups founded since 2000. We find that VC-backed firms are 54% more likely to face fraud charges than comparable non-VC-backed firms within a subsample of newly public firms where detection likelihood is high and homogeneous. We then examine the role of governance in explaining venture fraud, focusing on two features that have risen in recent years: founder-friendly structures and cap table complexity. In a panel prediction model examining all venture fraud cases, we find that fraud is more likely in startups with stronger founder control rights, more convex founder cash flow rights, more investors, and greater participation of non-traditional investors. Founder-controlled boards are 88% more likely to commit fraud than VC-controlled or shared-control boards, even within the same firm. Governance variables matter much more than founder characteristics in predicting fraud. Hot funding conditions at the initial round, which weaken governance incentives, predict future fraud. Fraudulent entrepreneurs continue to found new VC-backed startups unharmed relative to non-fraudulent entrepreneurs, suggesting a lack of market discipline. Overall, our results highlight rising agency costs in VC-backed firms that could lead to misallocation and broader social costs. 

Silicon Valley Everywhere? Startup Accelerators and the Geography of Entrepreneurship, with Mohaddeseh Heydari Nejad 

Presentations: RCFS Winter Conference (2026),  Rotman School of Management (2026),  Kelley School of Business (2025)

Abstract: Entrepreneurial entry reflects a fundamental tradeoff between strategic complementarities within local ecosystems and competition for scarce resources. This paper examines how this tradeoff shapes regional entrepreneurs’ entry decisions. We exploit the staggered rollout of the Creative Destruction Lab (CDL), a global mentoring-based accelerator that introduced 27 streams across 10 cities between 2012 and 2022. To address endogenous program placement, we construct a novel instrument based on academic collaboration networks between business schools and the CDL’s founding institution, the University of Toronto. We show that sector-specific launches reshape the composition of local entrepreneurship by reallocating startup formation toward treated sectors. Entry responses are weakened when local spillover potential is low, consistent with a complementarity-driven mechanism. Investor capital does not respond, resulting in intensified competition for financing. Motivated by these patterns, we develop a structural model that organizes the underlying mechanisms and provides suggestive policy counterfactuals. We document substantial heterogeneity in responses across sectors, with some targeted sectors experiencing costly crowding. Overall, the model suggests that generic policies may coordinate a broader base of participants more effectively than narrow targeting.