









































Pa
ge

 
1



Pa
ge

 
1

American Journal of  Applied 
Statistics and Economics (AJASE)

Forecasting Global Wheat Price in the Context of  Changing Climate and Market
Dynamics: An Application of  SARIMA Modeling Technique

Sahil Ojha1*, Lila B. Karki1

Volume 4 Issue 1, Year 2025
ISSN: 2992-927X (Online)

DOI: https://doi.org/10.54536/ajase.v4i1.4116
https://journals.e-palli.com/home/index.php/ajase

Article Information ABSTRACT

Received: November 30, 2024

Accepted: January 02, 2025

Published: February 07, 2025

The global wheat market is influenced by several factors, such as climate change, inflation, 
market fluctuations, geopolitical situations, and government policies, leading to continuous 
fluctuations in wheat prices. There is a need for a robust forecasting model that accurately 
captures seasonal variations and trends in global wheat prices to support informed decision-
making. The objective of  this study was to forecast the monthly global price of  wheat by 
adopting the SARIMA model using historical data from January 1990 to October 2024. The 
original monthly global wheat price data was log-transformed to stabilize the variance in the 
data and improve forecast precision. The seasonal variations in the data were adjusted by 
applying decomposition and differencing before modeling. Using the ‘auto.arima’ function 
from the ‘forecast’ package in R 4.3.3 for Windows, SARIMA(0,1,1)(0,0,1)12 was identified 
as the best-fitted model for forecasting global wheat prices. Residual analysis validated the 
model’s accuracy by visualizing the ACF and PACF plots and applying the Ljung-Box test, 
confirming that the residuals were white noise. The model forecasted a steady rise in the 
monthly global price of  wheat from $198.51 per metric ton in November 2024 to $201.36 
per metric ton in October 2025, peaking at $203.62 per metric ton in August 2025. These 
projections could help farmers, policymakers, and other relevant stakeholders to anticipate 
global price fluctuations and make informed decisions amidst global uncertainties. Future 
research could integrate external factors, such as climate change and geopolitical events, for 
enhanced predictive accuracy.

Keywords

Informed Decisions, Price 
Forecasting, SARIMA Model, 
Wheat

1 Department of  Agriculture, Food, & Resource Sciences, School of  Agricultural and Natural Sciences, University of  Maryland 
  Eastern Shore (UMES), Princess Anne, MD 21853, USA
* Corresponding author’s e-mail: sojha@umes.edu

INTRODUCTION
Wheat is the staple food for billions of  people worldwide 
and is considered one of  the most important crops in the 
world. Wheat belongs to the genus Triticum and is primarily 
cultivated for its seeds, which are processed into flour 
and used in various food products, including bread, pasta, 
and pastries. Triticum aestivum, commonly known as bread 
wheat, is the most widely grown species worldwide, while 
Triticum durum is primarily used for pasta production. The 
domestication and use of  wheat are closely linked with 
human efforts to ensure food security and gain control 
over their food supply. Wheat is grown in all geographical 
regions due to its high yield potential and adaptability to 
a wide range of  climates (Geren, 2021). As of  2023-24, 
China, the European Union, and India currently lead the 
global wheat production (USDA, Foreign Agricultural 
Service., n.d.). 
Agricultural production has been considered risky 
since yields are affected by extraneous factors beyond 
the producers’ control, such as weather patterns, pest 
infestations, and disease outbreaks. Furthermore, price 
volatility and market fluctuations during harvest are 
unknown when farmers make production decisions. 
Greater price volatility makes it harder to predict 
future prices and creates uncertainty regarding future 
price expectations and the profitability of  production 
(Drugova et al., 2019). Traditional forecasting methods 
fail to capture seasonal variations and trends in price, 
limiting their application in decision-making amidst 

dynamic market conditions. While several studies have 
focused on regional or national wheat price forecasting, 
there is a need to develop a forecasting model capable 
of  capturing complexities in global wheat prices and 
providing reliable price predictions that could support 
the decision-making process. By leveraging historical 
price data and applying rigorous model selection 
criteria, this study aimed to forecast global wheat prices 
using the Seasonal Autoregressive Integrated Moving 
Average (SARIMA) model to provide short-term price 
projections. This study will contribute to the existing 
literature on agricultural price forecasting by delivering 
valuable insights for stakeholders to make informed 
production decisions and manage risks associated with 
wheat price volatility. The findings are expected to have 
significant implications for global food security planning 
and formulation of  agricultural policies.

LITERATURE REVIEW
Global Wheat Market Dynamics
Wheat is the second-largest grain produced worldwide in 
terms of  cultivated area and production volume. In 2022-
23, global wheat production reached just over 789 million 
metric tons, marking an increase of  approximately nine 
million metric tons from the 2021-22 production level 
(Figure 1). In 2023-24, China produced 136.59 million 
metric tons, the European Union 134.94 million metric 
tons, and India 110.55 million metric tons, collectively 
accounting for 48% of  global wheat production (USDA, 



Pa
ge

 
2

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

Foreign Agricultural Service., n.d.). In the United States, 
wheat is produced in almost every state, with North 
Dakota and Kansas leading the output. Many developing 
nations heavily rely on wheat imports from Ukraine and 
Russia. As of  2022, Armenia and Mongolia imported 
wheat from Russia. Similarly, Laos exhibited the highest 

dependence on wheat from Ukraine, with 98% of  its 
wheat coming from Ukraine. The global wheat market 
has experienced significant growth over the past decade. 
Since 2014-15, global wheat export volume has expanded 
by about 33.6%, reaching more than 216 million metric 
tons in 2022-23 (Shahbandeh, 2024b).

Figure 1: Global wheat production and export volume (in million metric tons) from 2014/15 to 2022/23 
Source: Shahbandeh (2024a) and Shahbandeh (2024b)

Impact of  Climate Change on Wheat Production 
and Price
The effects of  climate change, including rising global 
temperatures and continuous changes in precipitation, 
have impacted global food production (Howard et al., 
2016). Several studies have examined the effects of  climate 
change on agricultural production. Rosenzweig and Parry 
(1994) conducted a global assessment to study the impact 
of  climate on world food supply, which concluded that 
doubling atmospheric carbon dioxide concentration 
would likely result in a modest decline in global crop 
production, with developing nations bearing the greater 
impacts. Tol (2002) evaluated the broader impacts of  
climate change on agriculture as well as other sectors and 
found that a 1°C rise in global mean surface temperature 
may yield net positive effects for China, the Middle East, 
and the Organization for Economic Co-operation and 
Development (OECD) member countries, but negatively 
impact other countries. Dhakhwa and Campbell (1998) 
analyzed the influence of  fluctuations in the differential 
warming of  day and night temperatures on crop yields. 
Their findings suggested that potential crop damage may 
be less severe under asymmetric day-night warming than 
uniform warming. 
Kang et al. (2009) conducted a thorough literature 
review on climate change impacts on crop yield, water 
productivity, and food and water security, concluding that 
climate change could increase water availability in some 
areas, enhancing water efficiency and crop production 
but potentially causing environmental degradation from 
expanded irrigation. They highlighted that the effects of  
climate change on crop yields vary by location, with some 
regions seeing an increase in output, while others may see 
a decrease. Expanding irrigated farmland could increase 
crop production, though it might degrade food and 
environmental quality. Tack et al. (2015) analyzed the effect 

of  weather on Kansas wheat yield using data from 1985 
to 2013, finding that the main drivers of  yield reduction 
were fall freezes and spring heat waves. Lobell et al. (2011) 
reviewed climate trends and global crop production 
since 1980, noting declines in global maize and wheat 
production by 3.8% and 5.5%, respectively, relative to a 
scenario with no climate change. However, the United 
States was an exception in their findings. Deschênes 
and Greenstone (2011) developed a new approach to 
studying climate change impacts on the U.S. agricultural 
sector, leveraging annual variations in temperature and 
precipitation to assess effects on agricultural profits using 
county-level panel data. Their findings suggested that the 
overall impact of  climate change on agricultural profits 
was minimal. However, effects varied across states, and 
predicted temperature and precipitation increases were 
unlikely to affect major crop yields such as corn and 
soybeans.
The anticipated impacts of  climate change on wheat yields 
are expected to influence wheat prices and global markets 
significantly. By 2050, global wheat prices may rise by 
7-18%, with the potential for even larger increases under 
extreme climate scenarios. Shifts in international trade 
patterns are also likely as production capacities change 
across different regions (Steen et al., 2023). Increased 
price volatility could be anticipated as more frequent yield 
disruptions occur due to extreme weather events (Song et 
al., 2022). Net wheat imports were projected to grow for 
many developing countries, exacerbating food security 
concerns, particularly in import-dependent nations across 
Africa and Asia (Habib-ur-Rahman et al., 2022).

Price Volatility in the Global Wheat Market
A dynamic interplay of  supply and demand factors 
influences wheat prices. Enghiad et al. (2017) reported 
that wheat price was significantly affected by weather 



Pa
ge

 
3

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

conditions, pest outbreaks, oil prices, and previous wheat 
prices. The study noted that the global wheat market 
was sensitive to supply shocks, with relatively inelastic 
demand. They indicated that wheat was susceptible to 
temperature fluctuations, with potential yield reductions 
due to global warming. For instance, drought in the major 
wheat-producing regions can lead to supply shortages 
and price spikes. Rising global wheat prices have led to 
higher import costs, which in turn have driven up the 
prices of  foods heavily reliant on wheat as an ingredient. 
In numerous developing nations, wheat-based foods 
comprise a significant portion of  household diets, so 
increases in wheat prices could significantly impact food 
costs and security.
Algieri (2016), Jebabli et al. (2014), and Sadorsky (2014) 
argued that oil price was one of  the key factors impacting 
agricultural commodity prices. The oil market influences 
wheat prices directly through production-related costs 
and indirectly through the demand for biofuels, which 
can lead to substitution effects. Prices for fertilizer, farm 
machinery, and transportation are all affected by crude 
oil prices, which influence wheat production costs. Baffes 
and Haniotis (2016) reported that when oil prices were 
high, farmers diverted their agricultural resources, like 
land, to energy crops, such as corn, instead of  wheat, due 
to higher demand for biofuels. This competition for land 
reduced wheat production, contributing to both volatility 
and upward pressure on wheat prices (Chen et al., 2010).
Government policies, such as export restrictions or 
subsidies, can affect the wheat supply chain. Anderson 
and Nelgen (2012) concluded that price insulation policies 
in domestic markets had spillover effects on global prices, 
potentially intensifying price volatility. Price volatility 
in wheat prices was also associated with demand-side 
factors. Population growth and changing dietary patterns 
in developing countries can increase wheat demand. This 
rising demand, if  not met by corresponding increases 
in supply, can contribute to price volatility (Godfray 
et al., 2010). Economic growth and urbanization in 
emerging economies can also increase wheat demand 
as consumers shift towards wheat-based products. This 
trend can pressure global wheat supplies and influence 
prices (Alexandratos & Bruinsma, 2012). Using wheat 
for non-food purposes, such as biofuel production, can 

create additional demand pressure. Changes in biofuel 
policies or oil prices can indirectly affect wheat demand 
and prices (Headey & Fan, 2008).

Application of  ARIMA and SARIMA Models in 
Agricultural Price Forecasting
The Autoregressive Integrated Moving Average (ARIMA) 
model has emerged as an effective tool for forecasting 
time series data, including agricultural commodities 
prices (Jadhav et al., 2017). This model incorporates 
three key components: autoregression (AR), differencing 
to achieve stationarity (I), and moving averages (MA) 
terms to forecast prices. ARIMA models have been 
popularly applied in various agricultural contexts due to 
their flexibility and ability to capture complex patterns, 
seasonality, trends, and irregular fluctuations in time series 
data (Iqbal et al., 2005). If  the data exhibited seasonal 
patterns, the Seasonal Autoregressive Integrated Moving 
Average (SARIMA) model, an extension of  ARIMA 
modeling, is applied for precise forecasting.
Several studies have demonstrated the effectiveness of  
ARIMA  and SARIMA models in forecasting agricultural 
prices. For instance, Jadhav et al. (2017) applied ARIMA 
models to forecast paddy, ragi, and maize prices in 
Karnataka, India, demonstrating the model’s power for 
price forecasting. Applying ARIMA models extends 
beyond price forecasting to other areas of  agricultural 
production. Ahmadzai and Eliw (2020) used the ARIMA 
model to forecast various economic variables related to 
wheat production in Afghanistan, including area under 
cultivation, productivity, and consumption. This broader 
application highlighted the versatility of  the ARIMA 
model in addressing various aspects of  agricultural 
economics and food security.
The ARIMA and SARIMA models are widely used time 
series forecasting methods due to their effectiveness in 
capturing temporal dependencies in data. Applying the 
ARIMA and SARIMA models to wheat price forecasting 
is particularly useful for predicting future trends, as it can 
handle non-stationary data, which is common in agricultural 
prices influenced by seasonal and external market factors. 
Table 1 presents various ARIMA and SARIMA models 
identified by previous literature to forecast wheat prices 
across different geographical regions.

Table 1: Wheat price forecasting using ARIMA and SARIMA models
Commodity Model identified Contributors
Wheat SARIMA(0,1,1)(0,1,1)12 (Darekar & Reddy, 2018)
Wheat ARIMA(1,1,1) (Sharma, 2015), (Novković et al., 2019)
Wheat ARIMA(1,2,1) (Du, 2014)
Wheat ARIMA(1,1,0) (Kumar, 2019)

In wheat price forecasting, ARIMA and SARIMA models 
have shown reliability in short-term forecasting, especially 
when underlying price patterns are stable. However, 
ARIMA and SARIMA models might face limitations 
when predicting prices under highly volatile or unexpected 

conditions. In that condition, other forecasting methods, 
like machine learning models or hybrid approaches (e.g., 
ARIMA combined with neural networks), could improve 
accuracy by capturing non-linear relationships in the data 
(Mahapatra & Dash, 2019).



Pa
ge

 
4

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

Policy Implications of  Wheat Price Forecasting
Sharma (2015) stated that price forecasts were essential for 
market participants to make production and marketing-
related decisions and for policymakers managing 
commodity programs and evaluating market impacts of  
domestic and global events. The author forecasted wheat 
prices in Rajasthan, India, using the ARIMA(1,1,1) model. 
The study’s policy implication highlighted that accurate 
price forecasting enhanced planning and development, 
empowering policymakers to anticipate commodity price 
trends and make informed decisions. This forecasting 
capability aided in formulating effective policies related 
to price structures, production levels, and consumption 
patterns. Furthermore, it supported strategic decisions 
in international relations, allowing governments to adjust 
trade policies, maintain market stability, and respond 
proactively to global market fluctuations. 
Forecasting the global price of  wheat allows governments 
to proactively address food security concerns by managing 
supply and demand fluctuations, particularly in wheat-
import-dependent nations. Accurate price forecasts enable 
effective trade policy adjustments, helping countries plan 
for imports or exports to maintain stable domestic markets. 
Forecasting also informs subsidy and support programs for 
wheat producers, enhancing income stability and agricultural 
productivity. Additionally, it aids in understanding the 

economic impacts of  global events, such as climate change 
or geopolitical disruptions, on wheat prices, enabling better 
resource allocation and risk management to stabilize price 
variations (Bentley et al., 2022).

MATERIALS AND METHODS
The monthly wheat prices (not seasonally adjusted, 
measured in U.S. dollars per metric ton) from January 
1990 to October 2024 are made publicly available by 
the International Monetary Fund [PWHEAMTUSDM]. 
For this study on forecasting the global price of  wheat, 
secondary data from January 1990 to October 2024 
were retrieved from the Federal Reserve Economic 
Data (FRED) (https://fred.stlouisfed.org/series/
PWHEAMTUSDM) on November 15, 2024. The data 
analysis was conducted using R 4.3.3 for Windows, open-
source software for statistical computing and graphics, 
provided by the R Foundation for Statistical Computing, 
with the ‘tseries’ and ‘forecast’ packages to estimate 
model parameters and fit the SARIMA model. The 
average monthly price of  wheat from January 1990 to 
October 2024 was $183.75 per metric ton, with a standard 
deviation of  $68.22. The dataset exhibited considerable 
dispersion, which affected the model’s fit. Therefore, 
the original data were log-transformed to stabilize the 
variance and enhance forecasting accuracy.

Figure 2: (a). Histogram of  original data and (b). Histogram of  log-transformed data
Source: Author’s computation based on FRED data, 2024

            a               b

Model Descriptions
Autoregressive (AR) Models
An autoregressive (AR) model uses previous time steps 
of  a variable to predict its future values. The general form 
of  an AR model is expressed as:
xt= σ + ϕ1 x(t-1 )+ et               (1)
where, xt is the value of  the time series at time t (for t = 1, 
2,.., n), σ is a constant centered around the mean, ϕ values 
are coefficients that represent the influence of  past values 
on xt, xt-1 is the lagged price by one time period, and et 
is the random error term that is uncorrelated. When this 
error term has a zero mean and constant variance σ2 (white 
noise), xt follows a first-order autoregressive process, or 
AR(1). In this model, the value of  x at time t depends on 
its previous value plus a random shock at time t.

For a second-order autoregressive process, or AR(2), the 
model is:
xt= δ + ϕ1 x(t-1 )+ϕ2 x(t-2)+et               (2)
where, xt now depends on its values from the previous 
two periods, centered around the mean δ. Generally, for 
an AR process of  order p, the current values of  the series 
depend linearly on the past p values, expressed as:
xt= δ + ϕ1 x(t-1 )+ϕ2 x(t-2)+…+ϕp x(t-p)+et                (3)
Here, xt represents an AR(p) process, where each ϕ term 
represents the influence of  previous values on the current 
value (Gujarati, 2003). 

Moving Average (MA) Models
The value of  xt can also be generated using a Moving 
Average (MA) process, which is formulated as:



Pa
ge

 
5

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

xt= δ +et-θ1 e(t-1)                (4)
where, δ and θ1 are constants, and et is a white noise 
error term. In this case, xt is defined as a constant plus 
a moving average of  the current and prior error terms, 
indicating that xt follows a first-order moving average, or 
MA(1) process.
If  x follows the expression:
xt= δ +et-θ1  e(t-1)-θ2 e(t-2)                  (5)
It is the second-order moving average or MA(2) process. 
More generally, for any positive integer q, the MA process 
is represented as:
xt= δ +et-θ1 e(t-1)-θ2 e(t-2)-…-θq e(t-q)                 (6)
Where, xt is an MA(q) process, a linear combination of  
white noise error terms (Gujarati, 2003).

Autoregressive Moving Average (ARMA) Models
The ARMA model combines autoregressive (AR) and 
moving average (MA) components, allowing it to capture 
the behavior of  a time series and predict future values 
using historical data. 
The most general ARMA model has an order of  p and q 
and is created by combining the equations for AR(p) and 
MA(q) processes (Gujarati, 2003). It is expressed as:
xt= δ+ϕ1 x(t-1)+ϕ2 x(t-2)+…+ϕp x(t-p)+et-θ1 e(t-1)-θ2 e(t-2)-…-
θq e(t-q)                 (7)
where, δ, ϕ1 … ϕp and θ1 … θq are fixed parameters. This 
model is called a mixed autoregressive moving average 
model of  order (p, q).

Autoregressive Integrated Moving Average (ARIMA) 
Models
ARIMA model, also known as the Box-Jenkins 
methodology (Montgomery et al., 2015), combines the 
Autoregression and Moving Average components with 
differencing to transform non-stationary time series 
data into a stationary form. Time series models generally 
assume that the series involved are weakly stationary, 
meaning that the series has a constant mean and variance, 
and its covariance is time-invariant. If  the time series, such 
as a price series, is already stationary (with constant mean 
and variance), then an ARMA(p, q) model can be applied. 
However, if  the series is not stationary, it can be made 
stationary by differencing it d times, and an ARIMA(p, d, 
q) model is then used. In this context:

• p represents the order of  the autoregressive (AR) 
process,

• d is the number of  differencing (I) needed to achieve 
stationarity,

• q indicates the order of  the moving average (MA) 
process.
According to the theoretical framework provided by Box 
and Jenkins, both AR and MA processes can be utilized 
in time series analysis. The Box-Jenkins method fits an 
ARIMA model to a given dataset for accurate forecasting.

MATERIALS AND METHODS
The Box-Jenkins approach to time series analysis and 
forecasting consists of  three main steps: identification, 
estimation and diagnostic checking, and forecasting. In 
the identification stage, initial values are chosen for the 
parameters p, d, and q. Initial estimates for the coefficients 
(ϕ1, ϕ2, …, ϕp) and (θ1, θ2, …, θq) are then obtained. 
Next, diagnostic checks are performed to assess how well 
the model fits the data. If  these checks indicate that a 
different model might be more appropriate, the process 
is repeated until a satisfactory model is identified. Finally, 
forecasts are generated based on the final model selected 
during the estimation process and confirmed through 
model selection criteria. 
For this study, the ‘auto.arima’ function in R was used 
to select the most appropriate model and forecast the 
price of  wheat by using that model. The ‘auto.arima’ 
function in R is a powerful and versatile tool widely used 
for time series forecasting. It is part of  the ‘forecast’ 
package in R, which is used to automatically identify the 
best-fitting ARIMA model for a given time series data 
using advanced algorithms. The ‘auto.arima’ function 
first checks for the stationarity in the time series data. If  
the data is non-stationary, differencing (d) is applied to 
make it stationary.
Statistical tests, such as the ADF test, determine the 
number of  differencing required. Seasonal differencing 
is applied if  the algorithm detects seasonal components 
in the data. After differencing, the function selects the 
value for autoregressive (p) and moving average (q) terms. 
For seasonal data, the function also determines the values 
of  autoregressive (P), differencing (D), moving average 
(Q), and seasonal period (m) terms in addition to the 
ARIMA terms. By evaluating different combinations 
of  these parameters using information criteria, such as 
AIC and BIC, the algorithm provided the best-fitting 
model for the given data. The model was selected so 
that the AIC value was minimal. Once the best ARIMA 
model was identified, it was fitted to the data to estimate 
its parameters. The best-fitted model was then used to 
forecast global wheat prices from November 2024 to 
October 2025.

RESULTS AND DISCUSSION
The global monthly price of  wheat extracted from FRED 
was not a time series data. So, the data was converted 
to make them time series and used for further analysis. 
The time series plot of  the log-transformed data (Figure 
3) and the lag plots (Figure 4) showed some fluctuations 
in the data over the period of  January 1990 to October 
2024. Seasonal variations were apparent in the data, 
possibly due to weather fluctuations, harvest patterns, 
and market fluctuations (changes in demand and supply 
forces) during those time periods.



Pa
ge

 
6

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

The data were then tested for stationarity using the ADF 
test, and it was found that the data was not stationary 
(Dickey-Fuller statistic = -2.8432, p-value=0.2212). 
The data is considered non-stationary when the p-value 
of  the ADF test is greater than 0.05 (null hypotheses 

accepted). The ACF and the PACF plots were also 
visualized. The ACF plot (Figure 5a) also showed a 
gradual decay over time lags but never cut off  to zero, 
suggesting that the data must be made stationary for 
further analysis.

Figure 3: Time series plot of  the monthly global price of  wheat (logged) from January 1990 to October 2024
Source: Author’s computation based on FRED data, 2024

Figure 4: Lagged plots of  the monthly global price of  wheat (logged) from January 1990 to October 2024
Source: Author’s computation based on FRED data, 2024

Figure 5: (a). ACF plot of  the global price of  wheat (logged) before differencing, and (b). PACF plot of  the global 
price of  wheat (logged) before differencing
Source: Author’s computation based on FRED data, 2024

      a                  b



Pa
ge

 
7

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

The data were then decomposed to separate the seasonal 
effects, trends, and random variation in the data (Figure 
6). The data exhibited similar seasonal trends over the 
periods, which were adjusted in the data to improve 

forecast accuracy. The ADF test of  the seasonally adjusted 
data yielded a value of  -2.8726 (p-value=0.2088); thus, we 
failed to reject the null hypothesis, and it was concluded 
that the data was still not stationary.

Figure 6: Decomposition of  the global price of  wheat (logged) data
Source: Author’s computation based on FRED data, 2024

The seasonally adjusted data were then differenced to make 
the data stationary. After each differencing, the p-value from 
the ADF was analyzed along with their corresponding ACF 
and PACF plots until the data became stationary. After 
the first-order differencing, the Dickey-Fuller test statistic 
value was -7.5652 (p-value=0.01). The null hypothesis was 
rejected and confirmed that the data was stationary after (d 
=1). However, in the ACF and PACF plots of  first-order 

differencing [Figure 7 (a and b)], there were some significant 
spikes, which suggested that there might be some structures 
or patterns that needed to be accounted for before modeling. 
Seasonal differencing was attempted to capture seasonal 
patterns in the data. The ACF and PACF plots of  the 
seasonal differencing [Figure 8 (a and b)] showed numerous 
spikes, suggesting that the SARIMA model (p,d,q)(P,D,Q)
[m] could be more appropriate for this data.

Figure 7: (a). ACF plot of  the global price of  wheat (logged) after first-order differencing, and (b). PACF plot of  the 
global price of  wheat (logged) after first-order differencing
Source: Author’s computation based on FRED data, 2024

      a                b



Pa
ge

 
8

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

Figure 8: (a). ACF plot of  the global price of  wheat (logged) after seasonal differencing, and (b). PACF plot of  the 
global price of  wheat (logged) after seasonal differencing
Source: Author’s computation based on FRED data, 2024

           a              b

The manual choice of  the model through empirical 
observation was somewhat arbitrary. Therefore, the 
‘auto.arima’ function from the forecast package in R was 
employed to estimate the best-fitted model and generate 
the forecast. The ‘auto.arima’ function accounts for the 
stationarity and seasonality in the data and sets the model 

parameters to generate the forecast based on the AIC and 
BIC values. The log-transformed data was modeled using 
the ‘auto.arima’ function, resulting in SARIMA(0,1,1)
(0,0,1)12 as the best-fitted model for our time series data. 
The parameter estimates of  the best-fitted model are 
presented in Table 2.

Table 2: Parameter estimates of  the fitted model
Estimates from SARIMA(0,1,1)(0,0,1)12

Coefficients MA1 SMA1
0.2523 -0.1042

S.E. 0.0487 0.0514
Log Likelihood 548.66
AIC -1091.32
RMSE: 0.0648 MAE: 0.0487 ME: 0.0004

Note: MA = Moving Average, SMA = Seasonal Moving Average, S.E. = Standard Error, AIC = Akaike Information Criterion, 
RMSE = Root Mean Square Error, MAE = Mean Absolute Error, and ME = Mean Error
Source: Author’s computation based on FRED data, 2024

The monthly global price of  wheat for the next year 
(November 2024-October 2025) was attempted using 
the SARIMA(0,1,1)(0,0,1)12 model. The forecasted price, 
generated on a logarithmic scale, was converted back to the 
original price scale by applying the exponential function 

to the logged values for convenient interpretation. The 
forecasted global price of  wheat obtained from the study 
data with a 95% confidence interval is presented in Table 
3. A visual representation of  these forecasts is presented 
in Figure 9.

Table 3: The forecasted monthly global price of  wheat (November 2024 - October 2025)
Year Month Forecasted price (US$ / metric ton) Lower 95% CI Upper 95% CI

2024 November 198.51 174.74 225.50
December 197.37 160.89 242.12

2025

January 197.71 152.55 256.25
February 198.25 146.20 268.83
March 199.09 141.16 280.80
April 199.40 136.48 291.33
May 197.66 130.99 298.25
June 199.95 128.61 310.86
July 202.40 126.59 323.61



Pa
ge

 
9

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

August 203.62 124.04 334.28
September 202.21 120.12 340.39
October 201.36 116.79 347.17

Source: Author’s computation based on FRED data, 2024

Figure 9: Forecasted monthly global price (logged) of  wheat from November 2024 to October 2025 with a 95% 
confidence interval
Source: Author’s computation based on FRED data, 2024

After forecasting the monthly global price of  wheat, 
residual diagnostics were performed to assess the model’s 
accuracy. The residual plot of  the best-fitted model (Figure 
10a) illustrated that the residuals were random, with a mean 
and variance of  0.0004 and 0.0042, respectively. Residuals 
with a mean and variance close to zero confirmed a 

well-fitted model. Also, the ACF and PACF plots of  the 
residuals showed no significant autocorrelation among the 
residuals [Figure 10 (b and c)]. Hence, the residuals were 
white noise. Moreover, the Ljung-Box test also yielded a 
p-value of  0.3918, confirming that the residuals exhibited 
no significant autocorrelation.

              a

      b                c
Figure 10: (a). Residual plot of  the fitted model, (b). ACF plot of  the residuals, and (c). PACF plot of  the residuals
Source: Author’s computation based on FRED data, 2024



Pa
ge

 
10

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

CONCLUSION
This study used historical data on global wheat prices 
from January 1990 to October 2024 to model and 
forecast prices from November 2024 to October 2025. 
The study identified SARIMA(0,1,1)(0,0,1)12 as the best-
fitted model by using the ‘auto.arima’ function in R. The 
model predicted the global price of  wheat solely based 
on historical prices, without explicitly incorporating 
other factors such as climate change, market fluctuations, 
inflation, and government policies. The forecasted price 
increased steadily, with some fluctuations, from $198.51 
per metric ton in November 2024 to a peak of  $203.62 
per metric ton in August 2025. These projections could 
be useful in allowing wheat farmers to make production 
decisions beforehand. Policymakers can also use this 
forecast to anticipate global wheat prices and make 
informed decisions at the national and international 
levels. Furthermore, wheat-importing and exporting 
countries could use these forecasts to adjust their trade 
policies to deal with price fluctuations.
This study has several limitations. First, the forecast 
presented in this study is based on the assumption 
of  linearity, which may not represent the non-linear 
characteristics of  global wheat prices. Second, this 
forecast is based on historical patterns in price, which may 
not hold in dynamic market conditions. Third, the global 
price of  wheat could be affected by several other factors, 
such as climate change and change in market equilibrium, 
which this model did not capture. Fourth, results from a 
single study like this may not be fully generalized to other 
agricultural contexts or time periods. Fifth, time series 
forecasting like this is only appropriate for short-term 
forecasting and might result in inaccurate results if  used 
for long-term forecasting. Future studies should consider 
developing more robust forecasting models, such as 
hybrid models capable of  capturing complex market 
phenomena affecting wheat prices globally. Moreover, 
researchers should cross-validate this model’s findings 
across different geographical locations and time spans to 
improve the model’s external validity.

REFERENCES
Alexandratos, N., & Bruinsma, J. (2012, June). World 

agriculture towards 2030/2050: The 2012 revision 
(ESA Working Paper No. 12-03). https://doi.
org/10.22004/ag.econ.288998

Algieri, B. (2016). A roller coaster ride: An empirical 
investigation of  the main drivers of  wheat price. In 
Food price volatility and its implications for food security and 
policy (pp. 207–237). Springer International Publishing. 
https://doi.org/10.1007/978-3-319-28201-5_10

Anderson, K., & Nelgen, S. (2012). Trade barrier 
volatility and agricultural price stabilization. World 
Development, 40(1), 36–48. https://doi.org/10.1016/j.
worlddev.2011.05.018

Baffes, J., & Haniotis, T. (2016). What explains agricultural 
price movements? Journal of  Agricultural Economics, 67(3), 
706–721. https://doi.org/10.1111/1477-9552.12172

Bentley, A. R., Donovan, J., Sonder, K., Baudron, F., 
Lewis, J. M., Voss, R., Rutsaert, P., Poole, N., Kamoun, 
S., Saunders, D. G. O., Hodson, D., Hughes, D. P., 
Negra, C., Ibba, M. I., Snapp, S., Sida, T. S., Jaleta, M., 
Tesfaye, K., Becker-Reshef, I., & Govaerts, B. (2022). 
Near- to long-term measures to stabilize global wheat 
supplies and food security. Nature Food, 3(7), 483–486. 
https://doi.org/10.1038/s43016-022-00559-y

Chen, S. T., Kuo, H. I., & Chen, C. C. (2010). Modeling 
the relationship between the oil price and global food 
prices. Applied Energy, 87(8), 2517–2525. https://doi.
org/10.1016/j.apenergy.2010.02.020

Deschênes, O., & Greenstone, M. (2011). Climate change, 
mortality, and adaptation: Evidence from annual 
fluctuations in weather in the US. American Economic 
Journal: Applied Economics, 3(4), 152–185. https://doi.
org/10.1257/app.3.4.152

Dhakhwa, G. B., & Campbell, C. L. (1998). Potential 
effects of  differential day-night warming in 
global climate change on crop production. 
Climatic Change, 40(3/4), 647–667. https://doi.
org/10.1023/A:1005339800665

Drugova, T., Pozo, V. F., Curtis, K. R., & Fortenbery, 
T. R. (2019). Organic wheat prices and premium 
uncertainty. Journal of  Agricultural and Resource Economics, 
44(3), 551–570. https://doi.org/10.2307/26797575

Du, S. (2014). The wheat prices prediction based on 
ARIMA model. Comprehensive Evaluation System of  
College Foreign Language Cloud, 2(4), 239.

Enghiad, A., Ufer, D., Countryman, A. M., & Thilmany, 
D. D. (2017). An overview of  global wheat market 
fundamentals in an era of  climate concerns. 
International Journal of  Agronomy, 2017, 1–15. https://
doi.org/10.1155/2017/3931897

Geren, H. (2021). Wheat importance, history, and 
adaptation. Theoretical and Practical New Approach in 
Cereal Science and Technology, 3.

Godfray, H. C. J., Beddington, J. R., Crute, I. R., Haddad, 
L., Lawrence, D., Muir, J. F., Pretty, J., Robinson, S., 
Thomas, S. M., & Toulmin, C. (2010). Food security: 
The challenge of  feeding 9 billion people. Science, 
327(5967), 812–818. https://doi.org/10.1126/
science.1185383

Gujarati, N. D. (2003). Basic Econometrics. McGraw-Hill.
Habib-ur-Rahman, M., Ahmad, A., Raza, A., Hasnain, 

M. U., Alharby, H. F., Alzahrani, Y. M., Bamagoos, 
A. A., Hakeem, K. R., Ahmad, S., Nasim, W., Ali, 
S., Mansour, F., & EL Sabagh, A. (2022). Impact of  
climate change on agricultural production; Issues, 
challenges, and opportunities in Asia. Frontiers in Plant 
Science, 13. https://doi.org/10.3389/fpls.2022.925548

Headey, D., & Fan, S. (2008). Anatomy of  a crisis: The 
causes and consequences of  surging food prices. 
Agricultural Economics, 39(s1), 375–391. https://doi.
org/10.1111/j.1574-0862.2008.00345.x

Howard, J., Cakan, E., Upadhyaya, K. P. (2016). Climate 
change and its impact on wheat production in Kansas. 
International Journal of  Food and Agricultural Economics, 



Pa
ge

 
11

https://journals.e-palli.com/home/index.php/ajase

Am. J. Appl. Stat. Econ. 4(1) 1-11, 2025

4(2), 1-10. www.foodandagriculturejournal.com/vol4.
no2.pp1.pdf

Iqbal, N., Bakhsh, K., Maqbool, A., & Ahmad, A. S. 
(2005). Use of  the ARIMA model for forecasting 
wheat area and production in Pakistan. Journal of  
Agriculture and Social Sciences, 1(2), 120–122.

Jadhav, V., Chinnappa, R. B., & Gaddi, G. M. (2017). 
Application of  ARIMA model for forecasting 
agricultural prices. Journal of  Agricultural Science and 
Technology, 19(5), 981-992. 

Jebabli, I., Arouri, M., & Teulon, F. (2014). On the effects 
of  world stock market and oil price shocks on food 
prices: An empirical investigation based on TVP-VAR 
models with stochastic volatility. Energy Economics, 45, 
66–98. https://doi.org/10.1016/j.eneco.2014.06.008

Kang, Y., Khan, S., & Ma, X. (2009). Climate change 
impacts on crop yield, crop water productivity and 
food security – A review. Progress in Natural Science, 
19(12), 1665–1674. https://doi.org/10.1016/j.
pnsc.2009.08.001

Karim Ahmadzai, M., & Eliw, M. (2020). Using ARIMA 
models to forecasting of  economic variables of  wheat 
crop in Afghanistan. Asian Journal of  Economics, Business 
and Accounting, 1–21. https://doi.org/10.9734/
ajeba/2019/v13i430180

Lobell, D. B., Schlenker, W., & Costa-Roberts, J. (2011). 
Climate trends and global crop production since 
1980. Science, 333(6042), 616–620.

Mahapatra, S. K., & Dash, Dr. A. (2019). Application 
of  time series modelling in price forecasting of  
agricultural commodities. International Journal of  
Agriculture and Food Science, 1(3), 27–29. https://doi.
org/10.33545/2664844X.2019.v1.i3a.23

Montgomery, D. C., Jennings, C. L., & Kulahci, M. (2015). 
Introduction to time series analysis and forecasting. John 
Wiley & Sons.

Rosenzweig, C., & Parry, M. L. (1994). Potential impact of  

climate change on world food supply. Nature, 367(6459), 
133–138. https://doi.org/10.1038/367133a0

Sadorsky, P. (2014). Modeling volatility and correlations 
between emerging market stock prices and the prices 
of  copper, oil and wheat. Energy Economics, 43, 72–81. 
https://doi.org/10.1016/j.eneco.2014.02.014

Sharma, H. (2015). Applicability of  ARIMA models 
in wholesale wheat market of  Rajasthan: An 
investigation. Economic Affairs, 60(4), 687–691.

Song, C., Huang, X., Les, O., Ma, H., & Liu, R. (2022). 
The economic impact of  climate change on wheat 
and maize yields in the North China Plain. International 
Journal of  Environmental Research and Public Health, 19(9), 
5707. https://doi.org/10.3390/ijerph19095707

Shahbandeh, M. (2024a, June 18). Global wheat production 
from 1990/1991 to 2023/2024. Statista. https://www.
statista.com/statistics/267268/production-of-wheat-
worldwide-since-1990/

Shahbandeh, M. (2024b, May 22). Wheat export volume 
worldwide from 2014/15-2023/24. Statista. https://
www.statista.com/statistics/737926/wheat-export-
volume-worldwide/

Steen, M., Bergland, O., & Gjølberg, O. (2023). Climate 
change and grain price volatility: Empirical evidence 
for corn and wheat. 1971–2019. Commodities, 2(1), 
1–12. https://doi.org/10.3390/commodities2010001

Tack, J., Barkley, A., & Nalley, L. L. (2015). Effect of  
warming temperatures on US wheat yields. Proceedings 
of  the National Academy of  Sciences, 112(22), 6931–6936. 
https://doi.org/10.1073/pnas.1415181112

Tol, R. S. J. (2002). Estimates of  the damage costs 
of  climate change, Part II. Dynamic estimates. 
Environmental and Resource Economics, 21(2), 135–160. 
https://doi.org/10.1023/A:1014539414591

USDA, Foreign Agricultural Service. (n.d.). Production - 
Wheat. U.S. Department of  Agriculture. https://fas.
usda.gov/data/production/commodity/0410000


