The Impact of El Niño-Southern Oscillation Events on Price Volatility: The Case of South African Maize

Abstract

This study examines how ENSO episodes affect maize price volatility transmission between the United States and South Africa. Using daily price data, from 1997 to 2024, for U.S. corn and South African white and yellow maize futures, the study employs GARCH models augmented with ENSO phase indicators and the Southern Oscillation Index (SOI) to determine volatility spillovers. The results show that South African maize prices respond to lagged US corn prices and exchange rate fluctuations, with price volatility of both white and yellow maize prices being more persistent during El Niño and La Niña events. This study integrates climate variability indicators, specifically different ENSO phases and the SOI, to investigate climate-driven volatility transmission between developed and emerging markets. Significant results were obtained when the Southern Oscillation Index was added in the volatility equations. Not only does the inclusion of ENSO indicators and SOI enhance the explanatory power of GARCH models beyond existing studies, it also provides evidence of climate-driven volatility spillovers between a developed and developing market. These findings highlight the role of climate variability in agricultural market dynamics and stress the need for proactive risk management strategies such as buffer stocks and climate responsive financial instruments to ensure food security and market resilience in Southern Africa.

1. Introduction

The El Niño-Southern Oscillation (ENSO) is a complex climate pattern that involves changes in sea surface temperatures and atmospheric pressure across the equatorial Pacific Ocean, causing precipitation changes worldwide. ENSO manifests mainly in two opposing phases: El Niño, which is associated with warming sea surface temperatures, and La Niña, resulting in cooler sea surface temperatures. These changes in atmospheric and oceanic conditions have a direct impact on regional weather systems, which in turn, affect crop yields worldwide, particularly maize production, which is a staple crop in both South Africa and the United States [1]. Although ENSO events influence climate patterns worldwide, their effects on crop yields are not uniform across regions, resulting in either an increase in yields [2] or significant yield losses [3]. For instance, during the 2015/16 season in South Africa, a severe El Niño event resulted in a 50% reduction in maize yields compared to the five-year average [4]. Given the important contribution of maize to the gross domestic product (GDP) of both the US and South Africa [5,6], such drastic yield changes will inevitably result in increased price volatility [7]. Moreover, maize is not only an important crop for South Africa, but also for the countries within the Southern African Development Community (SADC) that depend on it for regional food security [8]. This dependency heightens the region’s vulnerability to climatic shocks affecting maize production [4].

Research suggests that ENSO-related supply shocks in the United States can trigger significant price volatility in other countries, especially for net-importing nations [9,10]. This volatility transmission is amplified by the interconnectedness between world commodity markets, which causes price changes in one location to spread quickly to others. Volatility transmission refers to the process by which fluctuations in price uncertainty in one market are passed on to another market. Previous research by Auret and Sayed [11], and subsequently, by Sayed and Auret [12], have confirmed the existence of significant volatility transmission from the US maize market to the South African maize market. However, no research has yet examined how ENSO-related events specifically impact this transmission between the US and South African markets. This study aims to fill that gap by providing novel evidence of how ENSO events and the Southern Oscillation Index (SOI) influence the transmission of maize price volatility between a developed and emerging market. Understanding this relationship is critical, as ENSO-driven climate variability can increase price fluctuations, influencing food security and risk management strategies.

Additionally, the study seeks to capture volatility spillovers beyond traditional GARCH approaches by incorporating ENSO event indicators and SOI directly into GARCH models. This integration links financial modelling and climate variability metrics, to explain the climate-driven component in price dynamics. This insight can help design successful hedging strategies, which are important for ensuring food security in the greater SADC region.

2. Literature Review

To effectively analyse the various effects of ENSO phases on agricultural productivity, it is essential to first understand how ENSO is classified and how these classifications have evolved over time. ENSO manifests in three distinct phases: El Niño (characterized by warmer sea surface temperatures (SST), La Niña (cooler SST), and Neutral conditions. In his seminal paper, Trenberth [13] offered a widely recognized definition of El Niño based on a sustained SST anomaly in the central Pacific that persists for at least six months. This established a scientific standard for identifying these events. However, the National Oceanic and Atmospheric Administration (NOAA) updated the definition to include atmospheric changes. El Niño is now defined as an event that occurs when Niño 3.4 (Niño 3.4 region is a central Pacific area between the International Date Line and the South American coast where temperature changes are crucial indicators of ENSO) SST anomalies are ≥+0.5 °C for at least three consecutive months, combined with weakened trade winds and a negative SOI [14]. The SOI monitors daily differences between air pressure at two locations, namely Darwin, in Australia and the island of Tahiti. Because the observed air pressure is related to water temperature, La Niña and El Niño events are classified based on the observed difference in air pressure between Darwin and Tahiti [15]. This revised definition improves real-time monitoring and forecasting, which is important for assessing ENSO’s impact on climate and agriculture.

The effects of El Niño events are not consistent worldwide; rather, the effects are spatially [16] and temporally heterogeneous. In the southern hemisphere, El Niño is associated with drought-like conditions, which can disrupt agricultural production, reduce crop yields, and threaten food availability [4]. These climate anomalies can leave farming communities more vulnerable, especially in areas where agriculture is the main source of income [17]. Southern Africa, with more than 95% arable land dependent on rain-fed agriculture, is especially prone to the negative consequences of El Niño. South Africa, as the largest producer of maize on the African continent and a key exporter to the SADC countries [8], plays a central role in regional food security. Given the SADC region’s reliance on maize, climate variability such as El Niño, poses a threat to food security [4]. A substantial decline in South Africa’s maize production can negatively impact food security in the SADC region [18]. For example, the 2015/16 El Niño event directly caused international aid efforts to feed about 40 million people [19].

In contrast to the yield reductions typically observed in Southern Africa during El Niño events, findings by Mourtzinis et al. [20] show that El Niño events in the US Corn belt (the major corn producing area in the US), are generally associated with increased maize yields. This is largely due to beneficial rainfall and cooler temperatures. The increased yields contribute to global food supply stability. Similar opposing patterns can be observed during La Niña events that typically bring droughts to the US Corn Belt, reducing yields by about 12% [20]. Conversely, during La Niña, the southern hemisphere normally experiences above-average rainfall with an increase in maize yields of about 20% [4]. This asymmetric effect of ENSO on crop yields causes fluctuations in commodity prices, prompting an examination of ENSO’s impact on commodity prices and commodity price volatility transmission.

Recent studies [7,21,22,23,24,25] investigated the effect of ENSO on agricultural commodity prices, and all found that although ENSO has a non-uniform [3] (that is regionally and economically different) impact, its effects on commodity prices are economically important and statistically significant. Non-uniformity in this context refers to the fact that the magnitude and nature of the impact can vary significantly between regions and countries, depending on factors such as income levels, adaptation capacities, and dependence on rainfed agriculture. Lower-income countries, in particular, tend to be more vulnerable in coping with climate variability [3].

The asymmetric effect in yields between the US and South Africa result in a unique global pricing dynamic. As the global price setter of maize [26], the lower US yield causes higher global maize prices. When South Africa simultaneously experiences higher domestic supply, it results in domestic prices trading closer to export parity levels. Export parity price refers to the domestic price level at which South African maize becomes competitive for exports to other countries after transport costs are taken into account. Trading near export parity indicates abundant local supply, often preventing South African producers from fully benefitting from higher international prices. The effect of this is the forfeiture of the opportunity to participate in higher international prices. This price paradox, where lower US yields push international prices higher, but ample supply in South Africa causes domestic prices to remain low, or drop, is confirmed by Auret and Sayed [11], who determined that the 2010/11 La Niña event caused South African maize prices to drop by 32%. These variations in price responses highlight how shocks in one part of the world can influence another, underscoring the interconnectedness underlying the transmission of volatility across markets [27].

In agricultural economics, commodity price volatility and its transmission across markets has long been of interest, particularly regarding global trade liberalization and market integration. According to the spatial price transmission model, when two free markets are connected through trade, supply shocks or excess demand in one market will lead to a proportionate change in price in the other markets [28] as exogenous supply shocks, such as weather, changes in input costs, policy changes, or geopolitical disruptions are primary sources of price volatility [29]. This is also visible in global grain markets, which have witnessed significant volatility, with a wide range of macroeconomic factors, including supply and demand imbalances, geopolitical conflicts, fluctuations in production costs, trade restrictions, and transport costs, which collectively impact grain price dynamics [30]. Building on the spatial price transmission concept, volatility transmission further explains how the ripple effects of shocks are seen beyond local markets due to the interconnected economic and financial relationships of these markets.

The principle of volatility transmission describes how trade and financial linkages cause economic or financial shocks in one market to affect another. This phenomenon is especially noticeable in agricultural markets, where interconnected supply chains and commodity exchanges intensify the effects of regional disturbances. Volatility transmission is facilitated by three mechanisms:

• Macroeconomic conditions, such as economic policy uncertainty, credit risks in financial markets, investor sentiment, and global political tension, influence market expectations and risk perceptions [31]. These factors amplify price fluctuations across commodities.
• Commodity derivative markets, including forwards, futures, and option contracts, serve as a tool for price discovery and price risk management. However, these markets can also transmit volatility due to the influence of climate variability, such as ENSO events, and economic and financial linkages across markets [7,12].
• Trade linkages, shaped by concentrated export dominance and import dependency [32] connect regional production shocks to international markets, making them more susceptible to external disruptions [32].

Weather uncertainty, particularly caused by ENSO events, has a complex impact on maize price across markets. Focusing on the US, Peri [7] found that ENSO-driven volatility depends on the timing of the event. He found that volatility significantly increases in the US if El Niño occurs during Spring-Summer time, the critical growing phase of maize. Ubilava [25] reached similar conclusions, showing that El Niño events early in the season tend to produce more asymmetric and amplified responses due to the nonlinear dynamics in maize markets. This is expected as El Niño events starting around planting and early growth stages of maize, can cause uncertainty about final yields. Although historical patterns suggest a higher yield, the extent of the increase remains uncertain early in the season. This uncertainty in supply outcomes contributes to increased price volatility. Asymmetric effects (different economic consequences of each phase) are also transmitted to financial markets where inflationary pressures are stronger during La Niña events [33]. During El Niño, Yahyaei et al. [34] found that implied US maize futures price volatility decreases due to decreased uncertainty in expected price fluctuations. This is expected as higher yields are associated with lower prices and the uncertainty of the direction of price movements diminishes.

On a global scale, Ceballos et al. [35] analysed maize price volatility transmission over 16 countries across Africa (excluding South Africa), Latin America, and South Asia. They found that maize price volatility transmission was only significant in a few cases. Specifically, Colombia, Benin, Nigeria, and Ethiopia showed statistically significant relationships. As noted by the authors, all these countries are relying on maize imports, which may explain their exposure to international price volatility.

When examining specific bilateral relationships between the US and South Africa, conflicting volatility transmission results were obtained. Van Wyk [36] aimed to quantify the amount of volatility spillover to the South African white and yellow maize market from the US corn contract traded on the Chicago Mercantile Exchange (CME). His results indicated that the volatility spillover from the US to SA was not statistically significant. The results further indicated that volatility in the South African market is driven locally and less sensitive to international volatility. On the other hand, Sayed and Auret [12] investigated the volatility spillover effects of South African white maize futures against other domestic grain and external grain market futures listed on the Johannesburg Stock Exchange. They found that South African white maize futures are significantly affected by volatility from other grain and currency futures. Neither study took ENSO into account.

Despite extensive evidence on the transmission of maize price volatility, questions remain unanswered regarding how different ENSO events affect volatility spillovers between developing and developed markets, specifically between South Africa and the US, and their implications for sustainability and food system resilience. This is the first study to specifically analyse volatility transmission between the US and South African markets during ENSO events, considering South Africa’s important role as a regional maize basket. Moreover, limited insight into phase-specific market dynamics exists, as few studies distinguish between volatility transmission patterns linked to different ENSO phases across countries, despite findings such as Peri [7], which highlighted increased US maize price volatility when El Nino occurs during the spring-summer phase. The need to investigate emerging versus developed markets under ENSO conditions is further highlighted by contradictory findings on US-SA spillovers, with Sayed and Auret [12] finding significant transmission and Van Wyk [36] suggesting limited transmission. Additionally, South Africa is often excluded from international studies; for example, Ceballos et al. [35] ignored South Africa’s status as a regional hub for Southern African maize, which affects import and export dependencies. While ENSO indicators are often used in climate studies, the integration of the SOI into volatility models remains underexplored. Furthermore, though volatility exacerbates economic disparities and vulnerabilities in the food system, the relationship between volatility transmission and sustainability outcomes, such as food security, poverty alleviation, and climate resilience, remains understudied. This study aims to address these gaps by quantifying ENSO-driven maize price volatility transmission between the US and SA markets, providing insights on market dynamics, phase-specific impacts, and policy measures to strengthen sustainable food systems. Understanding these ENSO-related volatility transmission mechanisms is important to improve the effectiveness of risk mitigation strategies.

El Niño not only has a significant effect on food security in the SADC region, but also impacts farmers, especially smallholder farmers, resilience against weather extremes and financial sustainability. To improve resilience and financial sustainability, integrated risk mitigation strategies that address not only the economic consequences, but also agricultural production and supply stability are needed [37]. Governments should invest in the establishment and strengthening of early warning systems for climatic shocks caused by ENSO events to enable timely action [38]. Through collaboration between governments and financial institutions, affordable climate-smart agricultural finance mechanisms can be developed that farmers can use to protect themselves against ENSO-induced risks [39].

Farmers can personally improve resilience by diversifying agricultural production by cultivating different types of crops and diversifying into other farming enterprises, such as livestock. They could also consider conservation tillage practices, reduce dependence on water resources, and plant more drought-resistant crop varieties [40]. Farmers can manage ENSO-related risks through market-based measures such as agricultural insurance and weather derivative contracts [41]. Strategies should also focus on stabilizing the market and prices. This can include subsidies and price controls [37]. Other strategies to be considered should focus on system resilience. Governments can create and manage strategic food reserves to buffer against climate-induced food shortages, or promote sustainable trade practices and reduce international trade barriers [37].

Understanding the effects of ENSO events on maize price volatility transmission can help farmers, traders, and policymakers manage risks more effectively and make informed decisions in maize markets. This research can also help in developing enhanced climate-related financial instruments, such as derivatives, insurance products, and climate bonds, to mitigate the risks posed by ENSO-related maize price volatility. The financial implications of ENSO-related events are substantial and effective risk mitigation strategies are needed to protect the agricultural economy and ensure sustainable development.

3. Materials and Methods

Various econometric methods can be used to capture the dynamic relationship between prices across regions, especially during ENSO-driven shocks. Various forms of Generalized Autoregressive Conditional Heteroskedasticity (GARCH) models are often used to understand the transmission mechanisms that underlie price fluctuations through volatility [7,12,30]. This includes specific multivariate specifications such as BEKK GARCH, which are often used to identify own-volatility spillovers and persistence, and to examine direct volatility transmission [42]. Sayed and Auret [12] used four multivariate models (Diagonal BEKK, Full BEKK, Constant Conditional Correlation (CCC) and Dynamic Conditional Correlation (DCC)) in their study to examine the transmission of maize price volatility. Peri [7] used the Volatility Impulse Response Function (VIRF) methodology to investigate the effects of ENSO variability on international maize and soybean price volatility. This method allows for the visual examination of VIRF for multivariate GARCH models.

Sayed and Auret [12] investigated potential volatility spillovers relating to South African futures of grain, currencies, equity and interest rates, employing multivariate GARCH models. In their conclusion, they mention that future studies could also consider shocks from international exchanges. Their study also did not consider the effect of climate variability. Peri [7] focused on price changes in yellow maize exported from ports of the Gulf of Mexico. The potential impact of weather events during different phases of the production process is considered, which subsequently influences yields and price volatility. Among the findings, is an increase in volatility during El Niño events compared to La Niña events. With an emphasis on the implications for resilience in Southern African maize markets, this study uses GARCH models to examine the transmission of volatility during ENSO events. The GARCH approach followed in this paper includes the effect of international maize prices as well as different ENSO events, measured by means of two indicators, namely ONI and SOI.

GARCH (Generalized Autoregressive Conditional Heteroskedasticity) models first estimate a mean value for a specific variable and then also model the conditional variance or volatility of a variable [43]. The GARCH variance equation is represented by:

$${\sigma }_t^2= \omega + \alpha {ϵ}_{t-1}^2+ \beta {\sigma }_{t-1}^2$$

where the ARCH term, ${ϵ}_{t-1}^2$, represents news or new shocks in the form of the squared residual from the mean equation in the previous period and the GARCH term, ${\sigma }_{t-1}^2$, the forecasted variance in the previous period. The mean equation includes exogenous variables and provides an error term, included in the variance equation above.

The empirical analysis employs daily maize price data, from 1997 to 2024, obtained from the Chicago Mercantile Exchange (CME) and the Johannesburg Stock Exchange (JSE). The CME data covers corn prices in US cent per bushel, while the JSE data indicates the price in South African Rand per ton for white and yellow maize. The CME data was converted in terms of exchange rate and weight to also reflect the daily trading price in rand per ton. The historic price data was not stationary, as could be expected, and was transformed into log first differences for the empirical analysis. The sample includes 7039 daily observations from 3 June 1997 to 24 May 2024 for the three price variables: DLNWMAZ (log first difference in South African white maize price), DLNYMAZ (log first difference in South African yellow maize price), DLNCBOTZAR (log first difference in US corn price).

Table 1. Variables included in the analysis.

Abbreviation	Description
DLNWMAZ	log first difference in South African white maize price—sourced from the JSE (in South African Rand per ton)
DLNYMAZ	log first difference in South African yellow maize price—sourced from the JSE (in South African Rand per ton)
DLNCBOTZAR	log first difference in US corn price; original format of US cent per bushel—sourced from the CME, converted to price in South African Rand per ton
DLNRUSD	log first difference in exchange rate (South African Rand per US dollar)—sourced from the South African Reserve Bank
DUMELN	El Niño dummy generated = 1 if sea temperature at least 0.5 °C > average
DUMLAN	La Niña dummy generated = 1 if sea temperature at least 0.5 °C < average
TAHITI − DARWIN	Difference in daily recorded air pressure at Tahiti and Darwin

provides a summary of the variables included in the analysis.

Two indicators of climate variability are considered and mentioned in Table 1: Oceanic Niño Index (ONI) published by the Climate Prediction Centre and the Southern Oscillation Index (SOI). Based on ONI, which is available in monthly frequency, ENSO events are divided into three distinct phases: El Niño, La Niña, and Neutral conditions. This resulted in the sample period including six sub samples that could be classified as La Niña and three classified as El Niño events.

4. Results

4.1. Descriptive Statistics

Table 2. Descriptive statistics for the three price variables (daily price changes).

	DLNCBOTZAR	DLNWMAZ	DLNYMAZ
	Whole sample period (7038 obs)
Mean	0.000275	0.000311	0.000277
Max	0.117220	0.704863	2.307918
Min	−0.279928	−0.694829	−2.297223
Std dev	0.019435	0.022945	0.045760
	El Niño (913 obs)
Mean	0.0000730	0.001556	0.000747
Max	0.071269	0.704863	0.078332
Min	−0.186497	−0.694829	−0.054808
Std dev	0.017928	0.037047	0.014510
	La Niña (2009 obs)
Mean	0.001251	0.000808	0.000846
Max	0.086650	0.120714	2.307918
Min	−0.146286	−0.101218	−2.297223
Std dev	0.018802	0.016839	0.081169

Source: Author’s own calculations.

summarizes the descriptive statistics for the three price variables across three sample periods: the whole sample period, El Niño events and La Niña events.

Across the whole sample period, the mean values of the three price variables do not differ much. The price indicator of South African white maize is distributed over a much bigger range than US corn, while South African yellow maize is spread over an even larger range. The minimum and maximum values for South African yellow maize are almost three times the value of white maize and the standard deviation is also at least double that of white maize. The El Niño subsample reports the opposite. Now South African white maize prices are more volatile with much higher minimum, maximum and standard deviation. It makes sense that white maize prices are more volatile during El Niño periods. El Niño typically results in below-average rainfall and lower yields in South Africa. Since white maize is the staple food for Southern Africa, prices react more dramatically to these production shortfalls. The last section of Table 2 suggests that the observed higher volatility in South African yellow maize prices is largely driven by the La Niña events, when the standard deviation reaches its highest value. For yellow maize, the higher volatility during La Niña is also consistent with expectations. La Niña usually results in above-average rainfall and higher yields, creating exportable surpluses. This makes yellow maize prices more sensitive to international market conditions and exchange rate fluctuations. To remain competitive in the international arena and successfully export surplus yellow maize, South Africa’s domestic price must react immediately to US price changes and the rand-dollar exchange rate. Any delay in price adjustment can reduce South Africa’s export competitiveness, especially in a market where buyers can source yellow maize from other countries. Overall, South African maize prices tend to exhibit a higher degree of variability compared to the US prices; however, during El Niño periods, yellow maize shows relatively lower volatility.

Figure 1 portrays the variability of the three price variables against the background of climate variability. The zero line in the climate graph indicates neutral events. Values above 0.5 and below −0.5 represent El Niño and La Niña events. The upper part of Figure 1 compares volatility in US corn prices with South African yellow maize. There is no obvious common trend in the volatility of the two series. The most volatile period for US corn prices was in 2013, and South African yellow maize in 2014—none of these coincide with a significant ENSO event. The bottom part of the figure portrays the variability of South African white maize prices and the climate indicator. Again, there is no obvious common trend. The highly volatile period of 2014 coincides with a relative neutral ENSO period and the same is observed during 2020. Increased volatility for US corn and South African white maize occurred in 2013, but again it coincides with a relative neutral ENSO period.

4.2. Regression Analysis

In order to answer the question regarding volatility in maize prices and potential volatility transmission, GARCH models are estimated and reported.

Table 3. GARCH estimation of maize prices over the whole period.

	Basic GARCH		With Dummies
	White	Yellow	White	Yellow
	Mean equation		Mean equation
C	** 0.000363 (0.0391)	** 0.000348 (0.0163)	0.000148 (0.5345)	0.000094 (0.6304)
DLNCBOTZAR(−1)	*** 0.273962 (0.0000)	*** 0.276835 (0.0000)	*** 0.273143 (0.0000)	*** 0.276535 (0.0000)
DLNRUSD	*** 0.163407 (0.0000)	*** 0.163156 (0.0.0000)	*** 0.163705 (0.0000)	*** 0.162888 (0.0000)
DUMELN			* 0.000901 (0.0912)	0.000644 (0.1366)
DUMLAN			0.000308 (0.4305)	0.000522 (0.1160)
	Variance equation		Variance equation
C	*** 0.00000946 (0.0000)	*** 0.00000811 (0.0000)	*** 0.0000091 (0.0000)	*** 0.00000842 (0.0000)
RESID(−1)2	*** 0.121122 (0.0000)	*** 0.123510 (0.0000)	*** 0.120950 (0.0000)	*** 0.124523 (0.0000)
GARCH(−1)	*** 0.855719 (0.0000)	*** 0.850682 (0.0000)	*** 0.856305 (0.0000)	*** 0.849374 (0.0000)
DUMELN			0.00000034 (0.7397)	−0.00000013 (0.8469)
DUMLAN			0.00000055 (0.4208)	−0.00000067 (0.3235)
Obs	7037	7037	7037	7037

Source: Author’s own calculations. Probabilities reported in brackets. Significance levels: * for p < 0.1, ** for p < 0.05; *** for p < 0.01.

reports the GARCH estimation results over the whole sample period from 1997 to 2024. The same specification is tested to account for potential volatility in South African white and yellow maize prices, respectively. The mean equation includes the lag of the US corn price (DLNCBOTZAR(−1)), which turned out to be more significant than the South African lags, as well as the exchange rate in terms of South African Rand per US dollar (DLNRUSD). The inclusion of the US corn price in the mean equation is based on a study by Geyser and Pretorius [44], in which they confirmed the long run cointegration relationship between US corn prices and South African maize prices. South Africa is in an earlier time zone compared to the US. It is, therefore, expected that prices in South Africa will not be influenced by US prices on the same day, but rather by US prices of the day before.

The first two results columns of Table 3 report on basic GARCH estimates without dummy variables. For both white and yellow maize, the US price and exchange rate is statistically significant in the mean equation. The ARCH (${ϵ}_{t-1}^2$) and GARCH (${\sigma }_{t-1}^2$) terms are also significant. Furthermore, the sum of the coefficients is less than one, confirming that volatility is stationary and the impact of shocks on volatility eventually dies out.

The basic GARCH models in Table 3 do not account for the potential effect of ENSO events on price volatility. Three approaches are considered in this regard. The first option is to generate dummy variables to act as explanatory variables. Three specific periods were classified as El Niño events: 28 April 1997–25 May 1998; 6 October 2014–25 April 2016; and 1 May 2023–22 April 2024. An El Niño dummy was consequently generated with a value of 1 for these periods and 0 otherwise. Five La Niña periods were identified: 4 June 2007–23 June 2008; 31 May 2010–23 May 2011; 1 August 2011–23 April 2012; 3 August 2020–24 May 2021; and 2 August 2021–30 January 2023. A La Niña dummy generated a value of 1 for these periods and 0 otherwise. The basic GARCH models were then replicated to include these dummy variables in both the mean and variance equations—see the last two columns of Table 3. None of the dummy variables were significant in the variance equations. In the mean equations the coefficients of both dummy variables were positive, potentially indicating higher South African prices during ENSO events. For white maize, the El Niño dummy was significant at 10% (probability of 0.0912). This is in line with the observation of higher white maize prices during El Niño as displayed in Table 2. For yellow maize, the probabilities were 0.1366 for El Niño and 0.1160 for La Niña, indicating slightly less significance.

In a second attempt to account for ENSO events, the models in Table 3 were estimated over shorter sample periods for each of the eight ENSO events identified. Not much changed in the estimated mean equations; however,

Table 4. Variance equations over selected ENSO periods.

	White Maize	Yellow Maize
El Niño: 1 May 2023 to 22 April 2024
C	0.0000259 (0.1490)	0.0000418 (0.3229)
RESID(−1)2	* 0.076589 (0.0771)	0.054713 (0.3351)
GARCH(−1)	*** 0.831314 (0.0000)	** 0.694696 (0.0183)
Obs	256	256
La Niña: 1 August 2011 to 23 April 2012
C	0.00000862 (0.2641)	0.00000138 (0.3565)
RESID(−1)2	* 0.140622 (0.0817)	** −0.051845 (0.0463)
GARCH(−1)	*** 0.833422 (0.0000)	*** 1.060675 (0.0000)
Obs	191	191
La Niña: 3 August 2020 to 24 May 2021
C	* 0.0000540 (0.0976)	0.00000497 (0.1411)
RESID(−1)2	** 0.196709 (0.0460)	** 0.201300 (0.0437)
GARCH(−1)	*** 0.594452 (0.0004)	** 0.530369 (0.0276)
Obs	211	211

Source: Author’s own calculations. Probabilities reported in brackets. Significance levels: * for p < 0.1, ** for p < 0.05; *** for p < 0.01.

reports on the variance equations with different results in some sample periods.

In the first El Niño period, the ARCH term is not significant for yellow maize and only significant at 10% for white maize. In the first La Niña period, the ARCH term again is only significant at 10% for white maize. For yellow maize, the ARCH term displays the wrong sign, while the GARCH coefficient is more than one. In the second La Niña period, the signs are correct, but the ARCH term is again only significant at 5%. These different results during certain ENSO events indicate changing patterns of price volatility during certain ENSO events. El Niño tends to have longer-lasting consequences compared to La Niña, particularly for South Africa, which is heavily dependent on rain-fed agriculture. A factor contributing to this persistence is the depletion of stock levels during drought conditions. The supply constraints increase uncertainty in the market, leading to prolonged price volatility. This helps explain why volatility during El Niño phases can exhibit greater persistence, as evidenced from Table 4. The differing responses between white and yellow maize stem from their primary market uses and price sensitivities. White maize responds differently to market shocks when compared to yellow maize, due to its role as a preferred staple for human consumption in Southern Africa [44]. This demand sensitivity can result in prices trading much higher during droughts as white maize is produced by only a few countries, which limits the availability of imports during domestic shortages. Yellow maize exhibits less pronounced reaction under similar conditions. This is due to yellow maize mainly being used for animal feed and industrial purposes. Its stock levels can be more easily supported through imports, making its price dynamics less sensitive to local supply disruptions.

While the ARCH term (squared past error term) accounts for the impact of immediate past shocks or errors, the GARCH term tests how the current variance is influenced by past levels of the variance; or longer lasting, persistent volatility. These results thus suggest that during certain ENSO events, there are more signs of persistent volatility compared to volatility caused by recent shocks or errors. An attempt was made to include recent shocks in the variance equation by replacing the ARCH term with changes in the US corn price and changes in the Rand US dollar exchange rate. However, none of these led to statistically significant results.

The third approach to account for ENSO events in GARCH modelling involves the daily SOI indicators. As discussed before, the difference in air pressure between two locations in Australia serves as an indicator of climate variability. In

Table 5. GARCH estimation with SOI.

	White Maize	Yellow Maize
	Mean equation
C	* 0.000427 (0.0784)	** 0.000420 (0.0350)
DLNCBOTZAR(−1)	*** 0.283359 (0.0000)	*** 0.286355 (0.0000)
DLNRUSD	*** 0.167892 (0.0000)	*** 0.167755 (0.0000)
(TAHITI − DARWIN)	−0.0000257 (0.6659)	−0.0000305 (0.5551)
	Variance equation
C	*** 0.000008 (0.0000)	*** 0.00000602 (0.0000)
RESID(−1)2	*** 0.124825 (0.0000)	*** 0.119117 (0.0000)
GARCH(−1)	*** 0.853702 (0.0000)	*** 0.858020 (0.0000)
(TAHITI − DARWIN)	** 0.00000045 (0.0134)	*** 0.000000475 (0.0019)
Obs	6626	6626

Source: Author’s own calculations. Probabilities reported in brackets. Significance levels: * for p < 0.1, ** for p < 0.05; *** for p < 0.01.

, this additional variable, TAHITI − DARWIN, representing the difference in their reported air pressures for each day, is added to the mean and variance equation. This SOI variable was not significant in any of the two mean equations, while it was significant in the variance equations.

As was reported in Table 3, the coefficients of lagged US corn price and the exchange rate were statistically highly significant in both mean equations. Both the ARCH and GARCH terms in the variance equations are statistically significant at 1%, as before, with roughly the same magnitude. The new variable is also significant at between 1% and 2%. This confirms that climate variability, specifically in terms of sea temperatures that affect rainfall, does increase price volatility for both South African white and yellow maize.

The increased price volatility transmission during ENSO events stresses the challenges faced by SADC countries to ensure food security. Smallholder farmers, whose livelihood depends on rainfed agriculture, experienced increased pressure on their financial sustainability as the price paradox can amplify their price risk exposure. This contributes to increased economic disparities, disproportionately impacting vulnerable households and increased food insecurity.

5. Conclusions and Discussion

The study examined how climate variability, captured through ENSO phase indicators and the SOI, affects maize price volatility transmission between the US and South Africa. Using GARCH-type models on daily price data from 1997 to 2024, the results show that:

• The lagged US corn price is statistically significant at 1% in the mean equations for white and yellow maize prices. This confirms the significant impact of US corn prices on South African maize prices. None of the climate variables are statistically significant in the mean equations, although there is some indication, with probabilities between 8% and 11%, that South African maize prices are higher during El Niño and La Niña periods compared to neutral ENSO periods.
• During certain El Niño and La Niña periods, the ARCH variable in the variance equation loses its significance. This points towards more persistent levels of volatility and less influence of more recent risks. While volatility patterns differ, this study did not attempt to identify factors associated with it. This may be due to factors like stock levels, the severity of the ENSO period, etc. However, it is highly unlikely that it is caused by price volatility transmission from US corn prices.
• The significance of the SOI indicator in the variance equations of both white and yellow maize underlines the significant effect of climate variability on South African maize prices.

Against the background of previous studies discussed in the literature review, these results both confirm and add new insights to previous findings. Van Wyk [36] found no volatility spillover from US prices to South Africa and concluded that volatility on the South African market is mainly caused by local factors. This study incorporates US corn prices in the mean equations of South African white and yellow maize and found it to be statistically significant. In this way, the influence of international conditions is confirmed. Sayed and Auret [12] highlighted the need to consider the potential effect of ENSO on South African price volatility. The results, especially when employing the SOI indicator, confirm that climate variability does have an influence on the volatility of South African maize prices. Peri [7] found increased volatility in the price of yellow maize during El Niño events. The results of this study confirm the influence of climate variability on the price volatility of both South African white and yellow maize.

The results showed heightened vulnerability of South Africa’s maize market during ENSO-driven events, which increases price volatility risks and threatens food security in the SADC-region. ENSO events appear to lengthen the duration of volatility for white maize especially. This implies that climate associated shocks can translate into longer price instability in an emerging market. The results underpin the importance of proactive procurement strategies, such as government buffer stocks, hedging strategies, improved early warning systems, weather-related derivative contracts, trade diversification and the development of climate-related agricultural finance products to enhance food security during El Niño events.

Looking further than pricing and market implications, ENSO-related volatility poses significant risks to the sustainability and resilience of Southern African agricultural systems. Although fluctuating maize prices challenge all market participants in the maize value chain, vulnerabilities among smallholder farmers and marginalized populations are amplified. To successfully address these issues, climate-smart agriculture, improved forecasting technologies and innovative financial tools are combined in an integrated approach. Multidisciplinary research is needed to understand the relationship of climatic, economic, and social drivers of volatility so that more effective adaptation and risk management frameworks can be designed.

The findings emphasize the need for targeted policy interventions to mitigate the risks associated with ENSO-driven volatility. Improving early warning systems is a first step. By improving forecasting capabilities, governments can support farmers and food programmes to respond proactively to anticipated ENSO shocks. By creating and managing strategic food reserves, governments can stabilize food supply chains and prices during droughts.

For farmers, improved early warning systems means having the opportunity to diversify their crop mix by adding more drought resistant varieties, thereby enhancing their financial resilience. Access to insurance products can also provide protection against drought-related losses. These products are normally expensive [41], and early warning will allow farmers to only buy insurance when yield risk is expected, making such instruments financially viable. Farmers can also consider weather derivatives to offset their price risk linked to climate variability. With improved early warning systems, governments can stock food reserves, making their countries less reliant on expensive imports and food aid.

The development of climate-smart financial products can support farmers to remain profitable and resilient in the face of climate risks. Farmers who engage in conservation tillage practices can benefit from lower interest rates charged by sustainability-linked loans. Famers who adopt carbon-sequestering practices, such as conservation tillage, can earn carbon credits that can generate an additional source of income when sold on carbon markets. Governments can lower financial barriers by offering tax credits or subsidies for farmers who practice conservation agriculture. Altogether, these strategies are important to ensure agricultural and economic sustainability amidst ENSO induced climatic challenges.

Future empirical studies may consider the potential effect of stock levels on local South African maize prices. Existing maize stock levels may alleviate, or worsen, the impact of climate variability on prices, as prices are not only reliant on recent yields. In this empirical study, neither of the dummy variables turned out to be significant in the variance equations. This may be due to the relative long periods of certain ENSO events, which could leave the potential impact insignificant. This may be addressed in two ways in future studies. The ENSO classification can be refined to involve dummies for different levels—weak, moderate or severe. Alternatively, empirical models can be run over rolling periods to account for the different phases of the growth season. The differences in phases of the growth season were particularly relevant in the Peri study.