Lishu Zhang 张郦姝
Financial Economist · PhD candidate, Tilburg University · Postdoctoral Researcher, CREST (Paris)

I am a sixth-year PhD candidate in Economics at Tilburg University, supervised by Harry Huizinga, Harald Benink, and Louis Raes. I passed my pre-defense in April 2026 and am awaiting the final ceremonial defense, expected in April 2027. I am also a postdoctoral researcher in the Finance group at CREST in Paris, working with Peter Tankov and Olivier David Zerbib.
I am on the job market 2026-2027.
My research interests lie within Trade, Environmental Economics, and Finance. I draw on trade and macroeconomic theory to motivate and structure my empirical analysis. I study how global financial markets respond to climate change, and how climate-related risks are redistributed across countries, firms, and investors. My dissertation examines this through cross-border equity investment, M&A, and bank lending. At CREST, I extend the agenda to critical raw material (CRM) risk, combining production networks, trade, climate transition, and AI-assisted measurement.
Job Market Paper
Measuring Critical Raw Material Dependence
Who ultimately bears a raw-material shock?
Follow the material to the firm.
INDUSTRY INPUT–OUTPUT TABLES
Links pooled · Industry peers indistinguishable
SUPPLIER–CUSTOMER DATA
Direct links only · Not material-specific
THIS PAPER
Host firms on material-specific production networks
Illustrative production paths and firms
COVERAGE
55 materials · 12,481 firms
Half the variation within industries
BOTTLENECKS
One product, every route
For 25 of the 55 materials
VALIDATION
Costs match predictions one for one
(Industry measures cannot!)
What does that exposure mean financially?
A / FIRM ADJUSTMENT
A liquidity shock
Operating cash flow
Net external financing
% of initial assets per predicted cost increase of 1% of identified revenue · 95% intervals · “Before”: pre-window
B / EVENT REPLAY
Scenario · historical price changes applied to 2024–2025 exposures
How much financing could a disruption require?
Cash flow at risk $20.0bn across all 27 priced-event replays
External financing $35.4bn across all 27 priced-event replays
Mostly undocumented uses · 96% of cash flow at risk for the 2025 bismuth, indium and tungsten controls
C / EQUITY RISK
What firms use matters for equity risk.
RISK AT THE FIRM
The material mix matters.
Higher dependence median → 90th percentile
More volatile materials same dependence, +1 SD price risk
Material effect stronger at higher dependence · downside risk is expected shortfall
RISK ACROSS FIRMS
Material connections cross industry boundaries.
More shared materials +1 SD material connection
About a quarter of the effect of industry closeness
Evidence behind these findings
Material mix. At a given dependence rank, a one-standard-deviation higher material-price-risk basket is associated with 0.74 percentage points higher annualized stock volatility and 0.063 points higher expected shortfall. The price-risk association with stock volatility strengthens by 2.19 points from the bottom to the top of the dependence ranking. The paper uses exposure-robust uncertainty for these estimates.
Material connections. A one-standard-deviation higher material-connection rank is associated with 0.041 correlation points more weekly market-adjusted return comovement, and 0.049 among pairs in different three-digit industries. One correlation point is 0.01 in raw correlation. These are incremental associations; they do not establish a return premium or the share of portfolio risk caused by materials.
D / VISIBILITY
Markets follow salience.
Documented share of dependence
More for materials that are larger, harder to replace, or recently restricted
Stock prices at supply events
Sources, units, and interpretation
Measurement · From physical requirements to cost exposure. Technical evidence establishes material-specific production links; input cost shares turn those links into product cost elasticities; product revenues aggregate them to firm dependence. The network drawing is schematic. In the cost-validation regressions, predicted cost pressure is measured relative to identified revenue, while realized changes in cost of goods sold are scaled by initial sales. Their estimated magnitudes closely match; the comparison does not equate these two denominators. Source: Dollar Responses per Percentage Point of Predicted Cost Pressure.
A · How firms adjust. We trace cash flow and financing after material-price changes, comparing the component associated with identified supply-tightening events with the remaining price variation. For a predicted cost increase of 1% of identified revenue, the event-driven estimates at quarter 4 imply cumulative cash flow lower by 0.391% of initial assets and net external financing higher by 0.692%. This pattern is consistent with firms meeting liquidity needs through external finance. Shading and error bars show 95% confidence intervals; “Before” shows pre-event responses. The comparison series labeled “Predictable price changes” also includes pre-announcement price movements.
B · What past events imply for firms today. We apply historical event-driven price changes to firms’ 2024–2025 material exposures, using the estimated quarter-4 cash-flow and financing responses. Across all 27 priced-event replays, the implied amounts are $20.0 billion of cash flow at risk and $35.4 billion of external financing. These are scenarios, not recorded losses; totals include offsetting price declines. The brown segments show relationships where specialist reports miss uses accounting for more than half of a firm’s material dependence. Their size shows how relying on those reports alone can miss financially important exposure.
C · Individual and shared equity risk. The material-mix results compare users with positive dependence on priced materials, controlling for dependence, firm characteristics, country-by-year and six-digit industry-by-year effects. Material-price risk is the dependence-weighted volatility of the firm’s materials; downside risk is expected shortfall. The basket and interaction estimates use exposure-robust standard errors. The return-comovement test uses quarterly Fama–MacBeth regressions over 2019–2025, controlling for industry, products, country, firm characteristics and past return correlation, with Newey–West standard errors. Material connection weights the covariance of material prices measured before the quarter by the two firms’ dependence. Both the full-sample and different-industry coefficients are significant at the 5% level. Sources: Material Price Risk and Market Outcomes, Panels A–B, and Material Connection and Return Comovement, columns 3–4. The results identify associations; they do not isolate material prices as the cause of observed return comovement.
Where the exposure sits and what is visible. About half of the variation in dependence ranks lies within six-digit industry–year groups. In 2019, 3,165 of 8,073 firms (39.2%) have positive measured dependence on at least one material. For 25 of the 55 materials, a single product blocks every mapped route to downstream products that together account for at least 90% of measured dependence. Documented relationships carry 38.3% of measured dependence in firms’ filings and 41.0% in specialist reports, each in its matched sample. Announcement returns are three-day returns around 27 supply events, 2017–2025; firms that mine or refine raw materials earn about 1.1 percentage points more than firms without measured dependence.