Excessive food price variability (volatility) affects farmers, traders, processors and consumers – and it threatens food security. In the aftermath of the 2007-2008 food price crisis and post-crisis commodity price volatility, the Food Security Portal developed the Excessive Food Price Variability Early Warning System1. This system aims to identify unusual periods of excessive price variability, defined as price variability that exceeds a pre-established threshold1.
Initially, the system monitored the volatility of various agricultural products such as hard and soft wheat, rice, maize, soybeans, coffee, cocoa, sugar, and cotton. It also included energy products like crude oil and natural gas. Price information for these commodities was available on a daily basis, providing a rich dataset for analysis.
More recently, the Early Warning System has been expanded to track fertilizer price volatility. This expansion concentrates on urea, DAP, and potash prices from the US Gulf. A key difference for fertilizers is that their prices are updated weekly rather than daily, presenting specific challenges due to smaller sample sizes.
The tool identifies excessive volatility periods using nonparametric estimators for conditional value-at-risk (CVaR). Unusual periods of excess price variability are identified when the return of the underlying price series exceeds the estimated CVaR. This instrument serves as an early warning system for unusual periods of excessive price variability, which is crucial for providing timely warnings in food systems during crises or shortages, enabling appropriate and timely responses to potential negative effects on production decisions and rural incomes.
About the tool and its methdology
About the tool
The Excessive Food Price Volatility tool provides you with a visual representation of historical periods of excessive global price volatility from 2000-present, as well as a daily volatility status. The price volatility tool is updated on a daily basis to identify periods of high, moderate, and low price variability. It is based on a statistical model that measures the degree of fluctuation in commodity prices in futures markets. Specifically, it estimates the returns to commodity trading in those markets as reflected in day-to-day percentage changes of futures market prices closest to maturity. When the number of days with extreme positive price changes is high compared to what is expected by the model, the tool alerts that we are in a period of high price volatility.
What the tool identifies
Periods of excessive price variability. This occurs when we observe a large number of extreme positive returns in the three months preceding the day in question. An extreme positive return is defined as a return that exceeds a certain pre-established threshold. This threshold is normally taken to be a high order (95% or above) conditional quantile, (i.e. a value of return that is exceeded with low probability: 5% or less).
Days that are within periods of (high, moderate or low) price variability. This reflects the number of days in the current level of variability based on the number of extreme daily price increases seen over the last three months. For example, 20 days of low variability means that in each of the last 20 reported days, the number of days in the preceding three months with extreme daily price increases was within normal levels.
How the model works
The model works in two steps. First, the threshold to identify extreme positive price returns is calculated based on the 95th percentile of the last 4000 observations. The tool uses the ‘conditional value at risk’ (CVaR) of returns as the threshold, which is a standard measure of risk in financial economics. When a daily return is higher than that threshold, it counts as one day where there was an unusual high price return.
The second step of the model involves counting the number of days in the past three months that experienced such a high daily price increase. When this total number is higher than what is expected by the model, a period of high price variability is indicated. Specifically, the probability that we will observe k days of extreme price returns (returns above the 95% quantile as explained in the definition of excessive price variability) in a period of D consecutive days is defined as:
We implement a one-sided test based on a normal approximation for the binomial distribution, using a period of 60 consecutive days that precede any date (i.e. D=60).
The decision rule embedded in the color system
- RED or excessive volatility: If the probability value is less than or equal to 2.5%, the null that violations (i.e. days of extreme price returns) are consistent with expected violations is highly questionable, meaning that we are in a period of an excessive number of days of extreme price returns relative to that expected by the model. Therefore, we characterize that date as belonging to a period of excessive volatility.
- ORANGE or moderate volatility: If the probability value is bigger than 2.5% or less than or equal to 5%, the null that violations are consistent with expectations is questionable at a low level, meaning that we are in a period of moderate number of days of extreme price returns relative to that expected. Therefore, we characterize that date as belonging to a period of moderate volatility.
- GREEN or low volatility: If the probability value is bigger than 5%, we accept the null that violations are consistent with expectations, meaning that the number of extreme price returns is consistent to what is expected from the model. Therefore, we characterize that date as belonging to a period of low volatility.
For more information on the tool methodology, visit the individual commodity pages through the links below.
The decision rule embedded in the color system
- RED or excessive volatility: If the probability value is less than or equal to 2.5%, the null that violations (i.e. weeks of extreme price returns) are consistent with expected violations is highly questionable, meaning that we are in a period of an excessive number of weeks of extreme price returns relative to that expected by the model. Therefore, we characterize that date as belonging to a period of excessive volatility.
- ORANGE or moderate volatility: If the probability value is bigger than 2.5% or less than or equal to 5%, the null that violations are consistent with expectations is questionable at a low level, meaning that we are in a period of moderate number of weeks of extreme price returns relative to that expected. Therefore, we characterize that date as belonging to a period of moderate volatility.
- GREEN or low volatility: If the probability value is bigger than 5%, we accept the null that violations are consistent with expectations, meaning that the number of extreme price returns is consistent to what is expected from the model. Therefore, we characterize that date as belonging to a period of low volatility.
Price Volatility Models: Daily Prices vs. Weekly Prices
The price volatility model used for commodities with daily data (like agricultural and energy prices) is often based on the Martins-Filho et al. two-step nonparametric method for estimating high conditional quantiles. This approach requires large, clean datasets and is described as computationally heavier. In contrast, the price volatility model used for weekly fertilizer prices adapts this methodology for smaller, limited datasets. This Yao (2025) Hill-based approach uses a simpler, faster Hill estimator for tail estimation, making it easier and quicker to implement. It was found to be more robust and effective for the weekly fertilizer data, demonstrating better out-of-sample performance.
