This is an empirical study of whether emerging-market currencies react to climate disasters. It uses only public data and reports both what was found and what could not be tested. A short glossary of the recurring technical terms (fixed effects, winsorization, p-value, dose-response, exchange-market pressure) appears in the box in Section 3; each is also defined at first use, and re-explained briefly wherever it recurs far from that box.
Contents
Executive summary
1. Research question and contribution
1.1 Relation to the literature
1.2 Contrast-bloc test: where does the missing effect go?
2. Data and sample
2.1 Construction and the damage-missingness issue
2.2 Damage-missingness table
3. Methodology
3.1 Depreciation return and sign convention
3.2 Within-month volatility and the log ratio
3.3 Panel regression with fixed effects
3.4 Treatment intensity: letting the data decide whether severity matters
3.5 Two outcomes: NEER and bilateral USD
3.6 Exchange-market pressure (EMP) index
3.7 Inference: two-way clustered errors and a small-cluster caution
3.8 Robustness rules and the pre-trend check
3.9 Daily event study
4. Descriptive event-study results and the pre-trend check
4.1 Pre-trends: the raw pattern is composition
5. Main results
5.1 Severity and any-event: no depreciation effect
5.2 The consistent negative short-run sign
5.3 Exchange-market pressure eases; regime mechanism unresolved
5.4 Reserves: three distinct objects, kept separate
5.5 Global-risk conditioning
5.6 Volatility: no event-specific spike; raw daily fall mostly seasonal
6. Channels: exposure conditioners, not transmission mechanisms
6.1 Leg (i): do climate events move the channels?
6.2 Leg (ii): what conditions depreciation
6.3 What the channel results establish
7. ND-GAIN vulnerability and readiness
8. Heterogeneity: does the pooled null hide category-specific effects?
8.1 Hazard-type raw means (descriptive)
9. Interpretation: why the null — and why the easing
9.1 Why these particular results: a proportionality account
9.2 Root causes of the other empirical patterns
10. What the public data could and could not test
11. Code and reproducibility
12. Bottom line
Appendix B. Plain-language guide to the methods
B.1 – B.12 Methods explained for non-quantitative readers
References and data sources