📈 economics

Climate-Pressure Pass-Through to Food Inflation in ASEAN: Mixed-Frequency Quantile Evidence

This paper employs a mixed-frequency quantile regression framework to demonstrate that while climate-pressure indicators like heat stress show limited association with ASEAN food inflation on average, they serve as significant supplementary signals for monitoring elevated inflation risks in the upper tail of the distribution, whereas global food prices and currency depreciation remain the primary drivers.

Original authors: Nuttaphat Sukchitt, Pasin Marupanthorn, Waewwan La-ongsri

Published 2026-08-03

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Nuttaphat Sukchitt, Pasin Marupanthorn, Waewwan La-ongsri

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the economy as a giant, bustling kitchen where millions of people are trying to cook dinner every day. In this kitchen, the price of food is like the temperature of the stove: it can be steady and predictable, or it can suddenly flare up, burning your toast and making everyone hungry. Economists, who are like the head chefs of the world, spend a lot of time trying to figure out why the stove gets too hot. They know that sometimes the fire gets big because the price of ingredients from other countries goes up, or because the money people use to buy things loses its value. But there's another invisible force at play: the weather. Just like a sudden heatwave can wilt a garden or a flood can wash away a harvest, the weather can make food prices jump. The big question for scientists is: Can we actually see the weather's fingerprints on the price of food, or is it just a messy coincidence? This is the puzzle researchers in Southeast Asia set out to solve, looking for a hidden signal in the noise of daily life.

This paper is like a detective story where the investigators are trying to find out if the weather is secretly controlling the price of food in eight Southeast Asian countries (ASEAN). Instead of just looking at the average price of food over a whole year, the authors decided to look at the "worst-case scenarios"—the times when food prices were already skyrocketing. They used a clever trick to turn daily weather reports (like "it was 35°C today" or "it rained 50mm") into a single score that represents "heat pressure" or "drought pressure" for the month. They then ran a special kind of math test called "quantile regression," which is like checking the weather forecast not just for a normal day, but specifically for the hottest, most chaotic days of the year.

Here is what they found, and it's a bit surprising. If you look at the average month, the weather doesn't seem to have a strong, clear connection to food prices. The math says the link is weak and fuzzy, almost like trying to hear a whisper in a noisy room. However, when the researchers zoomed in on the top of the price list—those scary months when food inflation is already high—the story changes. They discovered that "heat pressure" (when it's unusually hot) does seem to push prices up even higher. Specifically, when the heat index goes up by one standard deviation (a big jump in heat), the food inflation rate at the very top of the distribution (the 90th percentile) jumps by about 0.069 percentage points. It's as if the stove is already hot, and a little extra heat from the sun makes it boil over just a bit more.

But here is the twist: the paper explicitly rules out a few things. First, they found that "excess rain" (too much water) doesn't have a stable, predictable effect on prices; the data was too messy to tell a clear story. Second, they tested a "time travel" idea by looking at weather that happens three months in the future. They found that future weather has no connection to current prices, which proves they aren't just seeing random patterns or seasonal tricks. Finally, they are very careful to say they haven't "proved" that heat causes prices to rise in a strict, unchangeable way. Instead, they suggest that heat is a useful "warning signal" that helps us understand why prices get so high during bad months, even if it doesn't explain the average price of food.

So, what's the takeaway? Think of the economy as a car. Global food prices and the value of money are like the engine and the fuel—they are the main things driving the car. The weather, specifically the heat, isn't the engine, but it's like a sudden gust of wind hitting the car. On a calm day, you barely notice the wind. But if the car is already speeding (high inflation), that gust of wind can push it over the edge. The authors suggest that while we can't use the weather alone to predict the future, keeping an eye on "heat pressure" is a smart way to spot when food prices might get out of control, especially in those already hot and expensive months. It's a new tool for the chefs in the kitchen to help them keep the stove from burning the toast.

Technical Summary: Climate-Pressure Pass-Through to Food Inflation in ASEAN

Problem Statement
Food inflation in ASEAN is a critical macroeconomic and welfare issue, particularly for lower-income households. While the drivers of food prices are well-understood to include domestic supply, trade, global commodity prices, and exchange rates, the specific role of weather variation remains complex. Weather affects food supply through multiple channels (yields, livestock, spoilage, logistics), yet existing literature often relies on conditional-mean regressions, annual data, or global panels. These approaches may obscure distributional heterogeneity, failing to detect whether high-frequency weather pressure is associated specifically with the upper tail of monthly food inflation. Furthermore, there is a gap in combining high-frequency weather predictors with a comprehensive set of domestic and open-economy price controls, and a lack of distinction between in-sample correlations and real-time operational forecasting capabilities.

Methodology
The study utilizes an unbalanced monthly panel of eight ASEAN economies (Cambodia, Indonesia, Lao PDR, Malaysia, the Philippines, Singapore, Thailand, and Vietnam) spanning January 2010 to December 2024.

Data Construction:
- Meteorological Data: Daily mean temperature and precipitation data are sourced from NASA POWER for capital-city reference points.
- Climate-Pressure Indicators: The authors construct three one-sided anomaly indicators relative to a country-specific 2000–2019 seasonal reference:
  - Heat: Positive temperature deviations.
  - Rainfall Deficit: Negative precipitation deviations.
  - Excess Rainfall: Positive precipitation deviations.
- Mixed-Frequency Aggregation: These daily anomalies are aggregated into monthly indices using a fixed, recency-weighted 60-day window with an exponential decay parameter ( $\kappa = 0.045$ ). This preserves the timing of daily observations rather than collapsing them into unweighted monthly averages. The resulting indices are standardized within each country using the full 2001–2024 climate history.
- Control Variables: The models control for lagged food inflation, contemporaneous headline and energy inflation, global food and energy inflation, exchange-rate depreciation, and a post-2020 pandemic indicator.
Econometric Strategy:
- Conditional-Mean Model: A pooled dynamic regression with country and calendar-month fixed effects is estimated via Ordinary Least Squares (OLS) with HC1 heteroskedasticity-robust standard errors.
- Quantile Regression: The study employs pooled quantile regression (Koenker–Bassett estimator) at the 25th, 50th, 75th, and 90th conditional quantiles. This shifts the focus from average inflation to the conditional upper tail, interpreting the 90th quantile as a measure of elevated inflation pressure.
- Placebo Test: A timing placebo is estimated using heat and rainfall-deficit indicators from three months ahead of the inflation observation to test for spurious seasonal artifacts.

Key Contributions

Mixed-Frequency Framework: The paper introduces a method to transform daily weather observations into asymmetric, recency-weighted monthly climate-pressure indicators for use in monthly inflation models without estimating a joint MIDAS lag polynomial.
Distributional Focus: Moving beyond conditional means, the analysis explicitly targets the conditional upper tail (90th quantile) of food inflation, aligning with the macro-at-risk logic that vulnerability often resides in the tail even when the mean is stable.
Incremental Association Interpretation: The study positions climate indicators as incremental monitoring signals within an open-economy conditioning set, rather than as structural causal estimates, acknowledging the limitations of observational data and contemporaneous price controls.

Results

Conditional-Mean Estimates: In the average model, the climate-pressure indicators (heat, rainfall deficit, excess rainfall) show small, statistically imprecise coefficients. In contrast, global food-price inflation and exchange-rate depreciation are positively and significantly associated with average food inflation.
Distributional Estimates: A distinct pattern emerges in the quantile regressions:
- Heat Pressure: The coefficient for heat increases monotonically across the distribution, rising from 0.021 at the median to 0.069 at the 90th quantile. The coefficients at the 75th and 90th quantiles are statistically significant (5% level), suggesting heat pressure contains incremental information about elevated inflation states that is obscured in average regressions.
- Rainfall Deficit: Shows a weaker, less stable pattern, peaking at the 75th quantile (0.047, significant at 10%) but losing precision at the 90th quantile.
- Excess Rainfall: Exhibits no stable aggregate relationship, with coefficients fluctuating around zero and changing signs across quantiles.
Placebo Results: The three-month-ahead heat and rainfall-deficit indicators yield small, statistically insignificant coefficients, providing no evidence that future weather systematically explains current average inflation.

Significance and Claims
The paper claims that representative-point climate measures (specifically heat pressure) should be interpreted as supplementary upper-tail monitoring signals rather than as causal estimates or stand-alone forecasts. The results indicate that while average regressions may miss the link between climate and food inflation, the heat signal is concentrated in elevated inflation states.

The authors explicitly state that the study does not establish causality and that the available inference does not formally prove that coefficients differ across quantiles due to limitations in handling panel dependence with a small number of countries. The paper identifies specific requirements for operational use, including:

Spatially relevant exposure measures (e.g., crop-weighted gridded data rather than capital-city points).
Disaggregated food price data.
Non-food domestic controls.
Dependence-robust inference procedures.
Genuine out-of-sample evaluation.

Ultimately, the paper contributes a framework for assessing distributional food-inflation risk, suggesting that heat pressure adds distinct information for monitoring high-inflation scenarios in ASEAN, provided it is used alongside traditional indicators like global commodity prices and exchange rates.

Technical Summary: Climate-Pressure Pass-Through to Food Inflation in ASEAN

More like this