36

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

Research on World Agricultural Economy
https://journals.nasspublishing.com/index.php/rwae

Copyright © 2023 by the author(s). Published by NanYang Academy of Sciences Pte. Ltd. This is an open access article under the Creative 
Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) License. (https://creativecommons.org/licenses/by-nc/4.0/).

1. Introduction
The U.S. cheese sector has been steadily growing 

since the 1990s and has become among the most impor-
tant commodities in the U.S. dairy agricultural economy. 

U.S. consumers are consuming twice as much cheese per 
capita as they did in 1980 [1]. Moreover, annual cheese 
production in the U.S. has been steadily growing from 6.94 
billion lbs. in 1995 [2] to 13.1 billion lbs. in 2019 [3]; with 

DOI: http://dx.doi.org/10.36956/rwae.v4i3.870

Received: 1 June 2023; Received in revised form: 8 July 2023; Accepted: 13 July 2023; Published: 20 July 2023

Citation: Wang, Z., Tejeda, H., Kim, M.K., et al., 2023. Cheese Price Softening in the U.S.: Determining Effects from 
Excessive Cheese in the Market. Research on World Agricultural Economy. 4(3), 870. http://dx.doi.org/10.36956/
rwae.v4i3.870

*Corresponding Author:
Man-Keun Kim, 
Department of Applied Economics, Utah State University, Logan, Utah, 84321, United States; 
Email: mk.kim@usu.edu

RESEARCH ARTICLE
Cheese Price Softening in the U.S.: Determining Effects from Excessive 
Cheese in the Market

Zuyi Wang1    Hernan Tejeda2   Man-Keun Kim1*    Wai Yan Siu3

1. Department of Applied Economics, Utah State University, Logan, Utah, 84321, United States
2. Department of Agricultural Economics and Rural Sociology, Twin Falls Research and Extension Center, Twin Falls, 
University of Idaho, Idaho, 83844, United States
3. Haub School of Environment and Natural Resources, University of Wyoming, Laramie, WY 82072, United States

Abstract: The United States (U.S.) cheese sector has experienced continuous production and consumption growth since 
the 1990’s with its market characterized as having oligopolistic behavior, signaling that prices respond to supply. Despite 
the steady industry growth, it experienced a recent multi-year period of declining prices. This paper addresses how 
growth towards an all-time record surplus of cheese, in response to excess milk production and export drops, weakened 
U.S. cheese prices. This study finds there is a significant short-run effect on price from overly expanded cheese supply, 
i.e., specifically taking a 2019 monthly average supply of 2.48 billion pounds, a 10% rise in cheese supply results in an 
immediate price decrease of 8.7%, translating to an average decline of USD 0.15/pound. The cumulative effect on its 
price from this 10% change results in a cumulative drop in cheese prices of 18.9%, approximately equal to a decrease 
of USD 0.34/pound. Findings provide relevant information to cheese, dairy producers and stakeholders, for milk 
production schedules, risk management and dairy policy analysis.

Keywords: Autoregressive distributed lag model; Cheese prices; Price softening; Supply growth

http://dx.doi.org/10.36956/rwae.v4i3.870
http://dx.doi.org/10.36956/rwae.v4i3.870
http://dx.doi.org/10.36956/rwae.v4i3.870
mailto:mk.kim@usu.edu
https://orcid.org/0000-0001-6796-5307
https://orcid.org/0000-0002-0205-9618


37

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

supply growth increasing more rapidly in the recent period 
from 2015 to early 2019 versus the previous five years, i.e. 
4.3% annum versus 2.2% annum, respectively. Despite 
the consumption growth, U.S. cheese markets experienced 
a steady decline in prices between 2015 and early 2019a. 
Industry analysts have described the drop in cheese prices 
as the result from an abundance in the supply of cheese, 
accompanied by increases in imports and drops in exports. 
This paper seeks to quantify market effects from an overly 
abundant growth in cheese supply that weakened its pric-
es, and in turn, affect milk production prices due to milk 
production pricing mechanisms [4]. The findings of this 
study enhance our understanding of the relationship be-
tween cheese price and its supply, derived from (perisha-
ble) milk production, with significant implications for risk 
management, investment decisions, and potential policy 
analysis. As a result, agribusiness sectors and supply chain 
actors involved in cheese markets stand to benefit from 
this research paper, as it quantifies the impact of excessive 
growth of cheese supply in the market on both short-run 
and long-run prices. Moreover, these insights may provide 
valuable information for strategic decision-making within 
the cheese industry. Studying the price of cheese holds 
significant importance, primarily due to the dairy sector’s 
prominent role as the main agricultural industry in several 
U.S. states, including California, Wisconsin, New York, 
Idaho, Michigan, and New Mexico, among others. Many 
of these states are also major cheese producers. In 2021, 
more than 42% of US milk fat was used for cheese pro-
duction [5], and the daily consumption of cheese per person 
increased to 0.74 cup-equivalents (1 cup-equivalent = 1 
cup milk) in 2021 from 0.36 cup-equivalents per person in 
1981 [6]. Moreover, cheese exports have witnessed growth 
in recent years, with increased shipments to countries 
such as Mexico, the Middle East, Japan, Central America, 
the Caribbean, Korea, Australia, and Colombia in 2022. 
Hence, as previously mentioned, quantifying the impact of 
unprecedented growth in cheese supply may assist in pro-
viding valuable information for strategic decision-making 
within the cheese industry.

From U.S. Department of Agriculture (USDA) data, 
the amount of cheese surplus (beginning stock) in the 
U.S. grew from about 1 to 1.4 billion pounds between 
2016 and 2019 [5], reaching record numbers, as shown in 
Panel A ( highlighted within the blue oval) in Figure 1. 
The substantial increase in cheese storage began around 
2008 in response to milk production exceeding its rates of 
use/consumption, and driving the milk surplus to produce  

a This study does not include price shifts during subsequent period of 
COVID-19 which is under a different study.

cheese [7].b Cheese exports have constituted an average of 
5.6% of yearly production since 2010. However recently, 
China and Mexico imposed retaliatory tariffs on U.S. dairy 
exports in the summer of 2018 as a consequence of a trade 
war; which resulted in annual cheese shipments dropping 
by 63% and 10% to China and Mexico; respectively [9]. 
Between 2010 and 2019 U.S. milk production increased 
by a total of 13.3% [3], and cheese production grew as 
well, but at a higher rate of 29.8% (3.3% per annum)—as 
observed in Panel B in Figure 1. Total cheese supply (sum 
of stocks, production, and imports) has likewise steadily 
increased, as shown in Panel C Figure 1, even more in the 
period 2016 to 2019 (blue oval) as mentioned previously. 
It is important to note that in 2019, cheese supply growth 
decreased dramatically as seen in Panel A and Panel C 
(gold oval).

The increased supply over time has had a notable im-
pact on cheese prices, particularly during the period from 
2015 to 2018 as depicted within the blue oval in Figure 2. 
However, in 2019, cheese prices increased and inventory 
remained rather steady given the minor cheese production 
increase of 0.8%, in comparison to 3.1% and 3.8% of the 
previous two years (Figure 2, yellow oval)c accompanied 
by a rise in cheese exports of about 3%. As aforemen-
tioned, lower cheese prices also affected dairy producers’ 
milk prices regulated through Federal Milk Marketing Or-
ders (FMMOs), which govern about 75% of the US milk 
supply [11,12]. The price of cheese plummeted again in the 
spring of 2020 as the COVID-19 pandemic hit the U.S. 
(not shown in Figure 2). COVID-19 produced a signifi-
cant and unexpected shift in cheese demand since a large 
portion of cheese consumption occurs through restaurants 
and school cafeterias, and these shut down during the 2nd 
quarter of 2020 [13]. At the same time, cheese consumption 
grew in family household cooking settings, though at a 
lower rate. The shift in consumption outlets as a conse-
quence of the pandemic had varying effects on different 
agents along the dairy supply chain, as described by Wolf 
et al. [14].

Cheese prices may be sensitive to supply due to the 
oligopolistic behavior observed in the structure of the U.S. 
cheese market, as supported by previous studies by Mu-
eller and Marion [15], Arnade et al. [16], and Bolotova and 
Novakovic [4]. That is in the classical oligopolistic market 
model, e.g., the Cournot model, market prices tend to be 
a function of supply [17]. The empirical strategy employed 

b Milk surplus (oversupply) is noted by average milk production costs 
being lower than prices; i.e. below perfect competition equilibrium lev-
els [7].
c In 2019, there was a significant slowdown in year-over-year milk pro-
duction growth [10]. 



38

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

in this study is grounded in this theoretical foundation. 
To provide an intuitive representation of the oligopolistic 
market structure Figure 3 is presented. In this illustration, 
the horizontal axis represents the quantity of cheese, q, 
and the vertical axis represents the cheese price, p. Each 
point in Figure 3 indicates an equilibrium where supply 
(S) and demand (D) curves intersect at a specific period 
in time. Over time, demand and supply curves shift up or 
down based on supply and demand shifters; for example, 
milk production affects cheese supply curves and house-
hold income influences the cheese demand curves. As a 

result, multiple supply-demand equilibrium points exist, 
reflecting different supply-demand interactions over time. 
The fitted line connecting these equilibrium points repre-
sents the long-run relationship (LR) depicted in Figure 3 
represents, denoted as p = f (q). Panel A in Figure 3 dem-
onstrates a negative long-run relationship between price 
and quantity, while Panel B shows a positive relationship. 
This long-run relationship depends on how supply and 
demand shift over time. For the period in this study, the 
growth in cheese supply has outpaced changes in demand, 
resulting in a negative long-run relationship referred to 
as “price softening”. The theoretical underpinnings and 
graphical representation provided in Figure 3 help to un-
derstand the relationship between cheese price and supply 
within the context of an oligopolistic market structure.

Market analysts interviewed by financial press suggest 
that the surge of inventory has depressed prices, despite 
recent increases in cheese consumption [18]. The hypoth-
esis for this study is taken from these previous assertions, 
with the primary objective of estimating the market im-
pact from an overly abundance of cheese supply affect-
ing its market price, including the long-run price effect. 
To date, there has been a research gap in examining the 
short and long-term implications on cheese price and its 
variability resulting from the excessive growth of cheese 
supply. Persistent declines in cheese prices may adversely 
affect cheese and dairy industry players in the long run, 
potentially bringing about more industry consolidation [19]. 
To address this matter, and assume an oligopolistic market 

Figure 1. Cheese storage, production, and supply.

Source: Monthly cheese beginning stock (Panel A), production (Panel B) and supply (Panel C ) from USDA-ERS [5]. Note that 
supply in Panel C is the sum of beginning stock, production, and import (not shown here). 

Figure 2. Cheese prices.

Source: Monthly average market prices for cheddar cheese from 
USDA-AMS (USDA, AMS, 2020b).



39

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

structure, this study employs an autoregressive distributed 
lag model (ARDL) to estimate price as a function of sup-
ply. Taking 2019 as a reference year, empirical results 
indicate that a 1% increase in cheese supply (equivalent to 
approximately 24.9 million pounds/month) immediately 
leads to a 0.87% decrease in cheese prices (approximately 
1.5 cents/pound). Moreover, this 1% surge in cheese sup-
ply resulted in a 1.89% decrease in prices (roughly 3.4 
cents/pound) after a period of six to seven months.

The theoretical implications of this study can be sum-
marized in two key aspects. Firstly, by examining the 
relationship between cheese supply and prices within an 
oligopolistic market structure, this study significantly con-
tributes to the theoretical understanding of market dynam-
ics within the cheese industry. As previously mentioned, 
it shed light on the intricacies of pricing mechanisms and 
the influence of supply on market outcomes. This analy-
sis augments knowledge of how market forces operate 
in this specific industry context. Secondly, the findings 
of this study hold practical significance for policymak-
ers, industry stakeholders, and market participants. The 
insights gained regarding the impact of cheese oversup-
ply on prices can serve as valuable guidance for better 
decision-making processes related to supply management, 
risk mitigation, and investment strategies. By providing 
empirical evidence and an increased understanding of the 
dynamics at play, this study supports the formulation of 
effective policies and aids in informed strategic decision-
making within the cheese industry.

The remainder of this study is organized as follows. 
The next section reviews prior studies that have re-
searched the U.S. cheese market. Section 3 introduces 
data and examines the (estimated) ARDL model and its 
pertinence. Results, discussions, and implication remarks 
follow in Sections 4 and 5.

2. Literature Review

Numerous studies investigating the U.S. cheese sec-
tor have addressed matters of cheese markets. Recently 
Bolotova [20] investigated the spot cheese market and its 
behavior along different periods of FMMOs finding that 
this relation has intensified over the years, with increasing 
effects on price volatility. Tejeda and Kim [21] investigated 
price dynamics among different cheese varieties find-
ing periods where prices of American and Other cheese 
types were decoupled. Studies addressing cheese market 
structure and its oligopolistic nature include Mueller and 
Marion [15] who examined the trade behavior of leading 
cheese companies on the National Cheese Exchange, 
which despite trading less than 0.2% of all cheese sold in 
the U.S. provided market signals for formula-pricing of 
90-95% of all bulk cheese, and found evidence of market 
manipulation from oligopolistic cheese producers. Arnade 
et al. [16] investigated the level of retail competition in the 
U.S. cheese market, finding that the existence of price 
markups suggested the presence of imperfect competitive 
behavior. Kim and Cotterill [22] investigated pass-through 
rates for processed cheese under market conditions and 
found significant differences in these rates for processed 
cheese under different market conditions regime in com-
parison to being under a Nash-Bertrand price competitive 
regime. For this latter, the pass-through rate was at least 
three times that of under collusion.

More recently, Bolotova and Novakovic [4] investigated 
the farm-to-wholesale price transmission process affect-
ing the pricing practices used by Chicago Mercantile 
Exchange (CME) cheese wholesalers. Findings revealed 
that pricing strategies used by cheese sellers are consistent 
with oligopolistic behavior. Lopez et al. [23] determined the 
level of oligopoly markups above that of being perfectly 
competitive markets for several U.S. food processing in-

Figure 3. Long-run impact of cheese supply on prices.

Source: Created by authors.



40

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

dustries by estimating the associated Lerner index, which 
measures the amount of market power (0 = perfect com-
petition, 1 = monopoly), and found a moderate degree of 
market power for the cheese manufacturing industry, i.e. 
corroborating prior results of the presence of oligopolistic 
behavior. The softening of cheese prices from sizable in-
creases in supply has not been investigated and this study 
attempts to fill this gap in the literature.

3. Materials and Methods

Monthly average market prices of cheese for the period 
of January 2000 to December 2019 from the Agriculture 
Marketing Service (AMS) of USDA for the dairy program 
are used USDA AMS [24]. Monthly U.S. cheese supply 
data include production, beginning stock, and imports for 
the same period and are collected from the Economics Re-
search Service (ERS) of USDA ERS [5]d. Table 1 provides 
a summary of the data. Cheese supply is the sum of begin-
ning stock, production, and imports (imports not shown in 
Figure 1). From Table 1, cheese supply is on average 1.87 
billion pounds per month (beginning stock 0.98 billion 
pounds plus production 0.87 billion pounds plus import 
0.03 billion pounds) during the sample period, having a 
consistent upward trend, as shown in Panel C in Figure 
1. The average monthly cheese supply during 2018 and 
2019 surpassed 2.47 billion pounds after considering the 
average monthly beginning stock of 1.36 billion pounds. 
Conversely, the price of cheese displays a relative oppo-
site trend especially from 2015 to early 2019, a period of 
steady increase in cheese stock. As noted, many market 
analysts quoted in the business media have expressed con-
cern about cheese prices falling during this period [7,25,26].

The empirical strategy to quantify the impact of un-
precedented growth in cheese supply on both short-run 
and long-run market prices is based on the oligopolistic 
cheese market structure [4,15]. From prior findings of the 
U.S. cheese sector depicting an oligopolistic market be-
havior, it is anticipated that some degree of market power 
is exerted. In the classical oligopoly market model, e.g., 
the Cournot model, the market price is a function of total 
supply [17]. As such, we estimate cheese price as a func-
tion of total supply. In addition, the price from a previ-
ous period may affect the current price since cheese is a 
storable commodity, i.e., affecting the adjustment of the 
supply schedule. To reflect the unique characteristics of 
the cheese market in the U.S., an innovative approach 
utilizing an autoregressive distributed lag model (ARDL) 
considering cointegration [27] is adopted in this study. The 

d This study aggregates American and Other-than-American cheese 
types.

ARDL model provides a robust framework for investigat-
ing this matter. By employing the ARDL model, we are 
able to examine both the short-run and long-run relation-
ships between cheese prices and cheese supply, capturing 
the dynamics of the market over time. An advantage of 
using the ARDL model with a cointegration approach is 
that the dependent variable, in this case, cheese prices, is 
allowed to be non-stationary [28]. Another benefit of using 
the ARDL model is that through reparameterization we 
can construct an error correction model, which enables us 
to find the short-run and long-run effects.

Table 1. Descriptive s tatistics (Jan 2000-Dec 2019, 240 
observations).

Variable Mean Std. Dev. Max Min

Beginning stock (million lbs.) 976 209 1,413 621

Production (million lbs.) 868 133 1,154 636

Import (million lbs.) 27.2 6.7 46.0 1.4

Supply (million lbs.) 1,871 335 2,567 1,335

Price ($ per lbs.) 1.58 0.98 2.35 1.02

Sources: USDA AMS [24] for cheese prices and USDA ERS [5] 
for beginning stock, production, and import. Supply is the sum 
of beginning stock, production, and import.

Applying Kripfganz and Schneider [29], the ARDL (p, q) 
model is expressed by:

7

The empirical strategy to quantify the impact of unprecedented growth in cheese supply
on both short-run and long-run market prices is based on the oligopolistic cheese market
structure [4,15]. From prior findings of the U.S. cheese sector depicting an oligopolistic market
behavior, it is anticipated that some degree of market power is exerted. In the classical oligopoly
market model, e.g., the Cournot model, the market price is a function of total supply [17]. As such,
we estimate cheese price as a function of total supply. In addition, the price from a previous
period may affect the current price since cheese is a storable commodity, i.e., affecting the
adjustment of the supply schedule. To reflect the unique characteristics of the cheese market in
the U.S., an innovative approach utilizing an autoregressive distributed lag model (ARDL)
considering cointegration [27] is adopted in this study. The ARDL model provides a robust
framework for investigating this matter. By employing the ARDL model, we are able to examine
both the short-run and long-run relationships between cheese prices and cheese supply, capturing
the dynamics of the market over time. An advantage of using the ARDL model with a
cointegration approach is that the dependent variable, in this case, cheese prices, is allowed to be
non-stationary [28]. Another benefit of using the ARDL model is that through reparameterization
we can construct an error correction model, which enables us to find the short-run and long-run
effects.

Table 1. Descriptive statistics (Jan 2000-Dec 2019, 240 observations).

Variable Mean Std. Dev. Max Min

Beginning stock (million lbs.) 976 209 1,413 621
Production (million lbs.) 868 133 1,154 636
Import (million lbs.) 27.2 6.7 46.0 1.4
Supply (million lbs.) 1,871 335 2,567 1,335
Price ($ per lbs.) 1.58 0.98 2.35 1.02

Sources: USDA AMS [24] for cheese prices and USDA ERS [5] for beginning stock, production, and import. Supply
is the sum of beginning stock, production, and import.

Applying Kripfganz and Schneider [29], the ARDL (p, q) model is expressed by:

 =  +
=1



− +
=0



− +  (1)

where  is cheese prices at time , and represented as the sum of lagged cheese prices and the
sum of lagged cheese supply,  . Both  and  may be stationary, non-stationary or
cointegrated [27,28]. The optimal lags of p and q are determined by minimizing information criteria
such as the Akaike information criterion (AIC) or the Bayesian information criterion (BIC). The
error term  is assumed to be serially uncorrelated. Note that ordinary least squares (OLS) is
used to estimate model in Equation (1) even though there are lagged dependent variables present
on the right-hand side [30]. Pesaran and Shin [28] showed that the OLS estimators of the
parameters in Equation (1) are  -consistent and asymptotically normal, in other words,
 , 

d

d

d
 N(, 2) where  = , 

'
.

 (1)

where Pt is cheese prices at time t, and represented as the 
sum of lagged cheese prices and the sum of lagged cheese 
supply, St. Both Pt and St may be stationary, non-stationary 
or cointegrated [27,28]. The optimal lags of p and q are de-
termined by minimizing information criteria such as the 
Akaike information criterion (AIC) or the Bayesian in-
formation criterion (BIC). The error term εt is assumed to 
be serially uncorrelated. Note that ordinary least squares 
(OLS) is used to estimate model in Equation (1) even 
though there are lagged dependent variables present on 
the right-hand side [30]. Pesaran and Shin [28] showed that 
the OLS estimators of the parameters in Equation (1) are 

7

The empirical strategy to quantify the impact of unprecedented growth in cheese supply
on both short-run and long-run market prices is based on the oligopolistic cheese market
structure [4,15]. From prior findings of the U.S. cheese sector depicting an oligopolistic market
behavior, it is anticipated that some degree of market power is exerted. In the classical oligopoly
market model, e.g., the Cournot model, the market price is a function of total supply [17]. As such,
we estimate cheese price as a function of total supply. In addition, the price from a previous
period may affect the current price since cheese is a storable commodity, i.e., affecting the
adjustment of the supply schedule. To reflect the unique characteristics of the cheese market in
the U.S., an innovative approach utilizing an autoregressive distributed lag model (ARDL)
considering cointegration [27] is adopted in this study. The ARDL model provides a robust
framework for investigating this matter. By employing the ARDL model, we are able to examine
both the short-run and long-run relationships between cheese prices and cheese supply, capturing
the dynamics of the market over time. An advantage of using the ARDL model with a
cointegration approach is that the dependent variable, in this case, cheese prices, is allowed to be
non-stationary [28]. Another benefit of using the ARDL model is that through reparameterization
we can construct an error correction model, which enables us to find the short-run and long-run
effects.

Table 1. Descriptive statistics (Jan 2000-Dec 2019, 240 observations).

Variable Mean Std. Dev. Max Min

Beginning stock (million lbs.) 976 209 1,413 621
Production (million lbs.) 868 133 1,154 636
Import (million lbs.) 27.2 6.7 46.0 1.4
Supply (million lbs.) 1,871 335 2,567 1,335
Price ($ per lbs.) 1.58 0.98 2.35 1.02

Sources: USDA AMS [24] for cheese prices and USDA ERS [5] for beginning stock, production, and import. Supply
is the sum of beginning stock, production, and import.

Applying Kripfganz and Schneider [29], the ARDL (p, q) model is expressed by:

 =  +
=1



− +
=0



− +  (1)

where  is cheese prices at time , and represented as the sum of lagged cheese prices and the
sum of lagged cheese supply,  . Both  and  may be stationary, non-stationary or
cointegrated [27,28]. The optimal lags of p and q are determined by minimizing information criteria
such as the Akaike information criterion (AIC) or the Bayesian information criterion (BIC). The
error term  is assumed to be serially uncorrelated. Note that ordinary least squares (OLS) is
used to estimate model in Equation (1) even though there are lagged dependent variables present
on the right-hand side [30]. Pesaran and Shin [28] showed that the OLS estimators of the
parameters in Equation (1) are  -consistent and asymptotically normal, in other words,
 , 

d

d

d
 N(, 2) where  = , 

'
.

-consistent and asymptotically normal, in other words, 

7

The empirical strategy to quantify the impact of unprecedented growth in cheese supply
on both short-run and long-run market prices is based on the oligopolistic cheese market
structure [4,15]. From prior findings of the U.S. cheese sector depicting an oligopolistic market
behavior, it is anticipated that some degree of market power is exerted. In the classical oligopoly
market model, e.g., the Cournot model, the market price is a function of total supply [17]. As such,
we estimate cheese price as a function of total supply. In addition, the price from a previous
period may affect the current price since cheese is a storable commodity, i.e., affecting the
adjustment of the supply schedule. To reflect the unique characteristics of the cheese market in
the U.S., an innovative approach utilizing an autoregressive distributed lag model (ARDL)
considering cointegration [27] is adopted in this study. The ARDL model provides a robust
framework for investigating this matter. By employing the ARDL model, we are able to examine
both the short-run and long-run relationships between cheese prices and cheese supply, capturing
the dynamics of the market over time. An advantage of using the ARDL model with a
cointegration approach is that the dependent variable, in this case, cheese prices, is allowed to be
non-stationary [28]. Another benefit of using the ARDL model is that through reparameterization
we can construct an error correction model, which enables us to find the short-run and long-run
effects.

Table 1. Descriptive statistics (Jan 2000-Dec 2019, 240 observations).

Variable Mean Std. Dev. Max Min

Beginning stock (million lbs.) 976 209 1,413 621
Production (million lbs.) 868 133 1,154 636
Import (million lbs.) 27.2 6.7 46.0 1.4
Supply (million lbs.) 1,871 335 2,567 1,335
Price ($ per lbs.) 1.58 0.98 2.35 1.02

Sources: USDA AMS [24] for cheese prices and USDA ERS [5] for beginning stock, production, and import. Supply
is the sum of beginning stock, production, and import.

Applying Kripfganz and Schneider [29], the ARDL (p, q) model is expressed by:

 =  +
=1



− +
=0



− +  (1)

where  is cheese prices at time , and represented as the sum of lagged cheese prices and the
sum of lagged cheese supply,  . Both  and  may be stationary, non-stationary or
cointegrated [27,28]. The optimal lags of p and q are determined by minimizing information criteria
such as the Akaike information criterion (AIC) or the Bayesian information criterion (BIC). The
error term  is assumed to be serially uncorrelated. Note that ordinary least squares (OLS) is
used to estimate model in Equation (1) even though there are lagged dependent variables present
on the right-hand side [30]. Pesaran and Shin [28] showed that the OLS estimators of the
parameters in Equation (1) are  -consistent and asymptotically normal, in other words,
 , 

d

d

d
 N(, 2) where  = , 

'
..

A trend variable is added to Equation (1) since visual 
inspection indicates a possible positive trend in cheese 
prices, as observed in Figure 2. To control for seasonality, 
11 monthly dummies are also included in Equation (1). 
Cheese prices and supply are log-transformed and thus the 
relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be 



41

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

formulated by reparameterization in terms of the lagged 
levels and the first differences of Pt and St 

[29], and is ex-
pressed by:

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 (2)

where α is the speed of adjustment (or error correcting) 
coefficient, defined by:

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 (3)

The expression Pt–1 – θSt–1 from Equation (2) can be 
interpreted as the long-run equilibrium denoted by LR, 
the fitted line, in Figure 3. When cheese prices and/or 
cheese supply deviates from the long-run equilibrium 
(LR), cheese prices and/or cheese supply adjust to restore 
the equilibrium relation at the speed of α. A value of α ≈ 0 
implies a very slow adjustment back to the long-run equi-
librium. Thus the value of α measures how fast cheese 
prices react to a deviation from the long-run relationship, 
LR. The long-run effects parameter, θ, is computed by:

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 (4)

Coefficient θ describes the equilibrium effect of sup-
ply on cheese prices in the long run. The short-run coef-
ficients, ψpi and ψsj in Equation (2) explain the short-
run fluctuations not due to deviations from the long-run 
equilibrium. A value of ψsj being close to one in absolute 
terms indicates very rapid price adjustment after a change 
in cheese supply. 

To check for the existence of the long-run relationship 
in Equation (2), a bounds test is suggested by Pesaran  
et al. [27]. The null hypothesis is that there is no long-run 
relationship between cheese prices and cheese supply.

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 (5)

If 

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 is rejected with F-test, a t-test is required to test 
whether α is zero or not.

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 (6)
A long-run relationship between the price of cheese 

and cheese supply may be confirmed by rejecting both 
previous F-test and t-test. Lastly, the proportion of 

8

A trend variable is added to Equation (1) since visual inspection indicates a possible
positive trend in cheese prices, as observed in Figure 2. To control for seasonality, 11 monthly
dummies are also included in Equation (1). Cheese prices and supply are log-transformed and
thus the relationship is in proportional or percentage terms.

The error correction form of the ARDL model can be formulated by reparameterization
in terms of the lagged levels and the first differences of  and  [29], and is expressed by:

∆ =  −  −1 − −1
=1

−1

∆− +
=0

−1

∆− +  (2)

where α is the speed of adjustment (or error correcting) coefficient, defined by:

 = 1 −
=1



 (3)

The expression −1 − −1 from Equation (2) can be interpreted as the long-run
equilibrium denoted by LR, the fitted line, in Figure 3. When cheese prices and/or cheese supply
deviates from the long-run equilibrium (LR), cheese prices and/or cheese supply adjust to restore
the equilibrium relation at the speed of . A value of  ≈ 0 implies a very slow adjustment back
to the long-run equilibrium. Thus the value of  measures how fast cheese prices react to a
deviation from the long-run relationship, LR. The long-run effects parameter, , is computed by:

 =
=0
 ∑


=
=0
 ∑

1 − =1
 ∑

(4)

Coefficient  describes the equilibrium effect of supply on cheese prices in the long run.
The short-run coefficients,  and  in Equation (2) explain the short-run fluctuations not due
to deviations from the long-run equilibrium. A value of  being close to one in absolute terms
indicates very rapid price adjustment after a change in cheese supply.

To check for the existence of the long-run relationship in Equation (2), a bounds test is
suggested by Pesaran et al. [27]. The null hypothesis is that there is no long-run relationship
between cheese prices and cheese supply.

0
:  = 0 and

=0



 = 0 (5)

If 0
 is rejected with F-test, a t-test is required to test whether  is zero or not.

0
 :  = 0 (6)

A long-run relationship between the price of cheese and cheese supply may be confirmed
by rejecting both previous F-test and t-test. Lastly, the proportion of 1 (contemporaneous
adjustment) to  (long-run adjustment) enables to measure the degree of relationship between
cheese price and supply, i.e., how much of the price change occurs immediately.

4. Results

 
(contemporaneous adjustment) to θ (long-run adjustment) 
enables to measure the degree of relationship between 
cheese price and supply, i.e., how much of the price 
change occurs immediately.

4. Results

Empirical results from estimating Equation (1) for 
the price of cheese and its supply, between January 2000 
and December 2019, are presented. The optimal lags for 
cheese supply and cheese prices were selected by mini-
mizing BIC resulting in p = 2 and q = 1. Table 2 presents 
the top three ARDL models with the lowest BIC for com-
parison; as mentioned ARDL (2,1) is found to be the pre-
ferred resulting model. Note that the bounds test confirms 
that there is a long-run relationship between the price of 
cheese and cheese supply (F-stat = 11.79 and t-stat = 7.30; 
both H F and H t from Equations (5) and (6) are rejected at 
1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is 
expressed by:

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 (7)

where β0 is interpreted as the concurrent (instantaneous) 
supply effect, t is a trend and Dm is a monthly dummy 
(results are not reported in Table 2 to save space and are 
available upon request).

The coefficients from Table 2 are not easy to interpret 
because there are lagged dependent and independent 
variables. As discussed in the previous method section, 
a reparameterization of the estimated parameters permits 
converting the model into an error correction form as in 
Equation (2), from which estimated results are in Table 3. 
The long-run coefficient, θ, from Equations (2) and (4) is 
in the LR row of Table 3, representing the long-run effects 
of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten 
in error correction form as follows:

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 (8)

As mentioned, the coefficient of 

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 from Equation 
(8), 

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 measures the contemporaneous effect on the 
price of cheese due to changes in cheese supply. Table 3,  

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

= –0.866 with p-value of 0.02, implies that 1% increase 
in supply leads to an immediate 0.87% decrease in cheese 
prices. The estimated value of α, the speed of adjustment 
coefficient, is –0.164. This suggests that it takes about  

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 ≈ 6.3 months to correct an equilibrium disturbance. 
The estimated parameter, θ, which indicates the long-
run effect of an increase in cheese supply on its prices,  



42

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

is –1.891. In other words, a 1% increase in cheese supply 
results in a 1.89% decrease in cheese prices in the long 
run. Moreover, the proportion of 

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 (contemporaneous 

adjustment) to θ (long run adjustment), 

9

Empirical results from estimating Equation (1) for the price of cheese and its supply,
between January 2000 and December 2019, are presented. The optimal lags for cheese supply
and cheese prices were selected by minimizing BIC resulting in  = 2 and  = 1 . Table 2
presents the top three ARDL models with the lowest BIC for comparison; as mentioned ARDL
(2,1) is found to be the preferred resulting model. Note that the bounds test confirms that there is
a long-run relationship between the price of cheese and cheese supply (F-stat = 11.79 and t-stat =
7.30; both  and  from Equations (5) and (6) are rejected at 1% significant level).

The optimal model ARDL (2, 1) from Equation (1) is expressed by:

ln  =  + 1 ln −1 + 2 ln −2 +0 ln  + 1 ln −1 +  +
=1

11

 +  (7)

where 0 is interpreted as the concurrent (instantaneous) supply effect,  is a trend and  is a
monthly dummy (results are not reported in Table 2 to save space and are available upon request).

The coefficients from Table 2 are not easy to interpret because there are lagged
dependent and independent variables. As discussed in the previous method section, a
reparameterization of the estimated parameters permits converting the model into an error
correction form as in Equation (2), from which estimated results are in Table 3. The long-run
coefficient, , from Equations (2) and (4) is in the LR row of Table 3, representing the long-run
effects of cheese supply on the price of cheese.

The model ARDL (2, 1) from Equation (7) is rewritten in error correction form as follows:

Δ ln  =  −  ln −1 −  ln −1 + 1∆ ln −1 + 1∆ ln  + 

+
=1

11

 +  (8)

As mentioned, the coefficient of Δ ln  from Equation (8), 1 measures the
contemporaneous effect on the price of cheese due to changes in cheese supply. Table 3, 1� =
− 0.866 with p-value of 0.02, implies that 1% increase in supply leads to an immediate 0.87%
decrease in cheese prices. The estimated value of , the speed of adjustment coefficient, is –
0.164. This suggests that it takes about

1
0.16

≈ 6.3 months to correct an equilibrium disturbance.
The estimated parameter, , which indicates the long-run effect of an increase in cheese supply
on its prices, is –1.891. In other words, a 1% increase in cheese supply results in a 1.89%
decrease in cheese prices in the long run. Moreover, the proportion of 1 (contemporaneous
adjustment) to  (long run adjustment), 

�1
�

= 0.46, which indicates that the immediate change in
cheese prices is 46% of the long run change.

5. Conclusions and Implications
The present study seeks to determine the extent that effects from a substantial growth in

cheese supply have had on the decline of its prices. This situation not only affects the cheese

 = 0.46, which 
indicates that the immediate change in cheese prices is 
46% of the long run change.

5. Conclusions and Implications

The present study seeks to determine the extent that ef-
fects from a substantial growth in cheese supply have had 
on the decline of its prices. This situation not only affects 
the cheese sector stakeholders but also the milk prices of 
dairy producers through FMMOs. Previous studies have 
characterized the cheese market as having an oligopolistic 
nature, with its prices being a function of supply. Under 
an oligopolistic market scenario, we make use of monthly 
cheese prices and supply data from 2000 to 2019 to esti-
mate an ARDL model and quantify the market effects. 

Table 2. ARDL model results.

ARDL (2,0) ARDL (2,1) ARDL (3,1)

ln Pt−1 1.1357
∗∗∗ 1.1224∗∗∗ 1.1686∗∗∗

(0.064) (0.064) (0.067)

ln Pt−2 –0.3086
∗∗∗ –0.2860∗∗∗ –0.4494∗∗∗

(0.065) (0.065) (0.099)

ln Pt−3 0.1451
∗∗

(0.067)

ln St 0.3974
∗∗ 1.1754∗∗∗ –1.0991∗∗∗

(0.178) (0.369) (0.370)

ln St-1 0.8661
∗∗

(0.362)
0.8733∗∗

(0.359)

Trend 0.0012∗∗

(0.0005)
0.0010∗∗

(0.0005)
0.0008
(0.0005)

Constant 0.1554∗∗ 0.1287∗∗ 0.0951

(0.062) (0.063) (0.065)

Obs 238 236 237

Adj R2 0.878 0.877 0.881

F-stat 114.4[0.00] 105.6[0.00] 103.4[0.00]

BIC –533.79 –534.56 –534.09

Note: *, **, and *** indicate the significance at 10%, 5% and 1%, 
respectively.
Results for monthly dummies are not reported to save space, 
which are available upon request.
Pt represents cheese prices at time t and St is cheese supply at 
time t. 
Numbers in brackets in F-stat row are p-values associated with 
the F-stats.

Empirical results indicate that the recent substantial 
growth in cheese supply indeed significantly decreased 
cheese prices. Results from an estimated optimal ARDL 
(2,1) model find that a 1% increase in cheese supply leads 
to a 0.87% decrease in cheese prices in the short run and 
a 1.89% decrease in the long run. That is, for a monthly 
average supply in 2019 of 2,486 million pounds, a 1% 
increase in supply is roughly equivalent to 25 million 
pounds, which would depress cheese prices by 1.53 cents/
pound in the short run and 3.32 cents/pound in the long 
run. Implications from the findings are compelling. Con-
sidering 2019 as the reference year, and assuming all other 
things being equal, if cheese supply had decreased by 7.5% 
(roughly, 186 million pounds), 2019 cheese prices would 
have on average been around $2/pound; however, the ac-
tual average price was about $1.75/pound.

In conclusion, this study carries important theoretical 
implications that contribute to our understanding of mar-
ket dynamics within the cheese industry. By examining 
the relationship between cheese supply and prices within 
an oligopolistic market structure, we have shed light on 
the intricacies of pricing mechanisms and the influence 
of excessive supply growth on market outcomes. This 
analysis enriches our knowledge of how market forces 
operate within this specific industry context. Moreover, 
the practical implications of our findings hold significant 
relevance for policymakers, industry stakeholders, and 
market participants. The insights gained from this study 
regarding the impact of the unprecedented growth of 
cheese supply on prices provide valuable guidance for 
decision-making processes related to supply management, 
risk mitigation, and investment strategies. Empirical evi-
dence and a deeper understanding of the dynamics at play 
can inform the formulation of effective policies and facili-
tate informed strategic decision-making within the cheese 
industry. It is crucial to note that discussions surrounding 
cheese prices should consider the unusual and significant 
growth in cheese supply, as revealed in this study. Such 
considerations have implications for risk management 
and trade policy analysis. By incorporating these insights, 
stakeholders can navigate challenges posed by the men-
tioned cheese supply phenomena studied and make well-
informed decisions to foster a viable and sustainable 
cheese sector.

There are several potential avenues for future research 
based on the findings and implications of this study. First-
ly, while this study primarily focused on the impact of the 
unprecedented growth of cheese supply on prices, future 
research could expand on this by examining the role of 
demand factors in shaping cheese prices. Investigating the 
relationship between consumer preferences, demographic 



43

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

changes, and market demand could provide a more com-
prehensive understanding of the dynamics of price for-
mation in the cheese industry. Secondly, further research 
could delve into the broader implications of supply chain 
efficiency and coordination on market outcomes. Factors 
such as transportation costs, inventory management, and 
distribution strategies can significantly impact the overall 
functioning of the cheese market. Exploring these aspects 
can provide insights into how supply chain dynamics 
affect price dynamics and market outcomes. Thirdly, 
considering the growing export market for cheese, future 
research could explore the impact of international trade on 
domestic cheese prices and market integration. Analyzing 
trade patterns, tariffs, and trade agreements can offer valu-
able insights into the relationship between global market 
dynamics and domestic cheese prices.

Lastly, it is important to address a caveat. This study 
assumed an oligopolistic market structure in the cheese in-
dustry based on previous studies [4,15,16]. However, it is es-
sential to acknowledge that market structures may evolve 

over time, and the assumption of an oligopoly in the U.S. 
cheese market may no longer be the case. Future research 
could investigate the current market structure and its im-
plications for price dynamics to provide updated insights.

Authors Contributions

The authors confirm contribution to the paper as fol-
lows: study conception and design: Z. Wang, H. Tejeda, 
and M-K. Kim; data collection: Z. Wang and W. Siu; anal-
ysis and interpretation of results: Z. Wang, H. Tejeda, M-K. 
Kim and W. Siu; draft manuscript preparation: Z. Wang, H. 
Tejeda, M-K. Kim and W. Siu. All authors reviewed the 
results and approved the final version of the manuscript.

Funding 

This research received no specific grant from any fund-
ing agency in the public, commercial, or not-for-profit 
sectors.

Data Availability

Data is available upon request (all the data used in the 
manuscript are available from the public domain). Please 
contact the corresponding author.

Conflict of Interest

Authors certify that we have no affiliation with or in-
volvement in any organization or entity with any financial 
interest in the subject matter or materials discussed in this 
manuscript. 

References

[1] Durisin, M.. Americans are drinking less milk, but 
they can’t get enough cheese. Bloomberg, 2017 Jan 
11. Available from: https://www.bloomberg.com/
news/articles/2017-01-11/cheesier-pizzas-rescue-
dairy-prices-as-americans-drink-less-milk

[2] Dairy Products 1995 Summary [Internet]. National 
Agricultural Statistics Service (NASS); 1996 [cited 
2021 Jan 17]. Available from: https://downloads.
usda.library.cornell.edu/usda-esmis/files/jm214p131/
pn89d8896/9k41zh017/DairProdSu-05-02-1996.pdf 

[3] Dairy Products 2019 Summary [Internet]. National 
Agricultural Statistics Service (NASS); 2020 [cited 
2021 Jan 17]. Available from: https://downloads.
usda.library.cornell.edu/usda-esmis/files/jm214p131/
sb397v22f/mg74r618b/daryan20.pdf 

[4] Bolotova, Y.V., Novakovic, A.M., 2015. An empirical 
analysis of wholesale cheese pricing practices on the 
Chicago Mercantile Exchange (CME) spot cheese 

Table 3. ARDL in error correction form.

ARDL (2,0) ARDL (2,1) ARDL (3,1)

ADJ ln Pt−1 –0.1730
∗∗∗ –0.1636∗∗∗ –0.1357∗∗∗

(0.035) (0.035) (0.037)

LR ln St –2.2980
∗∗∗ –1.8910∗∗ –1.6636

(0.839) (0.914) (1.141)

SR ∆lnPt−1 0.3086
∗∗∗ 0.2860∗∗∗ 0.3043∗∗∗

(0.065) (0.065) (0.065)

∆lnPt−2 –0.1451
∗∗

(0.067)

∆lnSt –0.8661
∗∗ –0.8733∗∗

(0.362) (0.359)

Trend 0.0012∗∗ 0.0010∗ 0.0008

(0.0005) (0.0005) (0.0005)

Constant 0.1554∗∗ 0.1266∗ 0.0951

(0.062) (0.064) (0.065)

Note: *, **, and *** indicate the significance at 10%, 5% and 1%, 
respectively.
Results for monthly dummies are not reported to save space, 
which are available upon request. 
ARDL (p, q) represents autoregressive distributed lag model; p 
is number of lags of autoregressive terms, Pt−i, and q is number 
of distributed lag terms, St−j. See Equation (1).
ADJ represents α, the speed of adjustment, in Equation (3). 
LR represents θ, the long run coefficient, in Equation (4).
SR represents ψpi and ψsj, the short run coefficients, in Equation 
(2).

https://www.bloomberg.com/news/articles/2017-01-11/cheesier-pizzas-rescue-dairy-prices-as-americans-drink-less-milk
https://www.bloomberg.com/news/articles/2017-01-11/cheesier-pizzas-rescue-dairy-prices-as-americans-drink-less-milk
https://www.bloomberg.com/news/articles/2017-01-11/cheesier-pizzas-rescue-dairy-prices-as-americans-drink-less-milk
https://www.bloomberg.com/news/articles/2017-01-11/cheesier-pizzas-rescue-dairy-prices-as-americans-drink-less-milk
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/pn89d8896/9k41zh017/DairProdSu-05-02-1996.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/pn89d8896/9k41zh017/DairProdSu-05-02-1996.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/pn89d8896/9k41zh017/DairProdSu-05-02-1996.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/pn89d8896/9k41zh017/DairProdSu-05-02-1996.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/sb397v22f/mg74r618b/daryan20.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/sb397v22f/mg74r618b/daryan20.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/sb397v22f/mg74r618b/daryan20.pdf
https://downloads.usda.library.cornell.edu/usda-esmis/files/jm214p131/sb397v22f/mg74r618b/daryan20.pdf


44

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

market. International Food and Agribusiness Man-
agement Review. 18(3), 49-66.

 DOI: https://doi.org/10.22004/ag.econ.208403
[5] Dairy Data [Internet]. U.S. Department of Agricul-

ture (USDA), Economic Research Service (ERS); 
2021 [cited 2023 Jun 30]. Available from: https://
www.ers.usda.gov/data-products/dairy-data/

[6] Cheese and Yogurt Popularity Grew over Last Four 
Decades [Internet]. U.S. Department of Agriculture 
(USDA), Economic Research Service (ERS); 2023 
[cited 2023 Jul 4]. Available from: https://www.
ers.usda.gov/data-products/chart-gallery/gallery/
chart-detail/?chartId=105777 

[7] Raphelson, S. Nobody is moving our cheese: Amer-
ican surplus reaches record high. National Public 
Radio, 2019 Jan 9. Available from: https://www.npr.
org/2019/01/09/683339929/nobody-is-moving-our-
cheese-american-surplus-reaches-record-high 

[8] Milk Cost of Production Estimates [Internet]. U.S. 
Department of Agriculture (USDA), Economic Re-
search Service (ERS); 2021 [cited 2023 Jul 3]. Avail-
able from: https://www.ers.usda.gov/data-products/
milk-cost-of-production-estimates/ 

[9] Haddon, H. American, cheddar and other cheese pil-
ing up in the U.S. as trade slows and tastes change. 
The Wall Street Journal, 2018 Dec 17. Available 
from: https://www.marketwatch.com/story/american-
cheddar-and-other-cheese-piling-up-in-the-us-as-
trade-slows-and-tastes-change-2018-12-17

[10] Largest Decline in U.S. Dairy Farms in 15-Plus Years 
in 2019 [Internet]. The American Farm Bureau Fed-
eration; 2020. [cited 2021 Jan 17]. Available from: 
https://www.fb.org/market-intel/largest-decline-in-u.
s.-dairy-farms-in15-plus-years-in-2019

[11] Federal Milk Marketing Orders (FMMO) [Internet]. 
International Dairy Foods Association (IDFA); 2020 
[cited 2020 May 13]. Available from: https://www.
idfa.org/federal-milk-marketing-orders 

[12] Federal Milk Marketing Orders [Internet]. U.S. De-
partment of Agriculture (USDA), Agricultural Mar-
keting Service (AMS); 2020 [cited 2020 May 13]. 
Available from: https://www.ams.usda.gov/rules-reg-
ulations/moa/dairy 

[13] Newman, J., Bunge, J. Farmers dump milk, break 
eggs as coronavirus restaurant closings destroy de-
mand. The Wall Street Journal, 2020 Apr 9. Available 
from: https://www.wsj.com/articles/farmers-deal-
with-glut-of-food-as-coronavirus-closes-restau-
rants-11586439722 

[14] Wolf, C.A., Novakovic, A.M., Stephenson, M.W., 
2021. COVID-19 and the U.S. dairy supply chain. 

Choices. 36(3), 1-8. Available from: https://www.
choicesmagazine.org/choices-magazine/theme-arti-
cles/agricultural-market-response-to-covid-19/covid-
19-and-the-us-dairy-supply-chain 

[15] Mueller, W.F., Marion, B.W., 2000. Market power 
in the cheese industry: Further evidence. Review of 
Industrial Organization. 17, 177-191.

[16] Arnade, C., Gopinath, M., Pick, D., 2007. Measuring 
the degree of retail competition in US cheese mar-
kets. Journal of Agricultural & Food Industrial Orga-
nization. 5(1).

 DOI: https://doi.org/10.2202/1542-0485.1151
[17] Church, J., Ware, R., 2000. Industrial organization: A 

strategic approach. McGraw-Hill: New York.
[18] Haddon, H. America can’t move its cheese. The 

Wall Street Journal, 2018 Dec 17. Available from: 
https://www.wsj.com/articles/america-cant-move-its-
cheese-11545042600 

[19] MacDonald, J.M., Law, J., Mosheim, R., 2020. Con-
solidation in U.S. Dairy Farming [Internet]. Available 
from: https://www.ers.usda.gov/webdocs/publica-
tions/98901/err-274.pdf

[20] Bolotova, Y.V., 2020. Price discovery in the United 
States dairy industry. Journal of Agricultural & Food 
Industrial Organization. 20(2), 83-97.

 DOI: https://doi.org/10.1515/jafio-2019-0002
[21] Tejeda, H.A., Kim, M.K., 2021. Dynamic price rela-

tionships and price discovery among cheese markets. 
International Food and Agribusiness Management 
Review. 24(1), 1-13.

 DOI: https://doi:org/10.22434/IFAMR2019.0206
[22] Kim, D., Cotterill, R.W., 2008. Cost pass‐through in 

differentiated product markets: The case of us pro-
cessed cheese. The Journal of Industrial Economics. 
56(1), 32-48.

 DOI: https://doi.org/10.1111/j.1467-6451.2008.00331.x
[23] Lopez, R.A., He, X., Azzam, A., 2018. Stochastic 

frontier estimation of market power in the food in-
dustries. Journal of Agricultural Economics. 69(1), 
3-17.

 DOI: https://doi.org/10.1111/1477-9552.12219
[24] Dairy Mandatory Market Reporting [Internet]. U.S. 

Department of Agriculture (USDA), Agricultural 
Marketing Service (AMS); 2020 [cited 2020 May 
13]. Available from: https://www.ams.usda.gov/
rules-regulations/mmr/dmr 

[25] Dewey, C. America’s cheese stockpile just hit an all-
time high. The Washington Post, 2018 Jun 28. Avail-
able from: https://www.washingtonpost.com/news/
wonk/wp/2018/06/28/americas-cheese-stockpile-
just-hit-an-all-time-high 

https://doi.org/10.22004/ag.econ.208403
https://www.ers.usda.gov/data-products/dairy-data/
https://www.ers.usda.gov/data-products/dairy-data/
https://www.ers.usda.gov/data-products/chart-gallery/gallery/chart-detail/?chartId=105777
https://www.ers.usda.gov/data-products/chart-gallery/gallery/chart-detail/?chartId=105777
https://www.ers.usda.gov/data-products/chart-gallery/gallery/chart-detail/?chartId=105777
https://www.npr.org/2019/01/09/683339929/nobody-is-moving-our-cheese-american-surplus-reaches-record-high
https://www.npr.org/2019/01/09/683339929/nobody-is-moving-our-cheese-american-surplus-reaches-record-high
https://www.npr.org/2019/01/09/683339929/nobody-is-moving-our-cheese-american-surplus-reaches-record-high
https://www.npr.org/2019/01/09/683339929/nobody-is-moving-our-cheese-american-surplus-reaches-record-high
https://www.ers.usda.gov/data-products/milk-cost-of-production-estimates/
https://www.ers.usda.gov/data-products/milk-cost-of-production-estimates/
https://www.marketwatch.com/story/american-cheddar-and-other-cheese-piling-up-in-the-us-as-trade-slows-and-tastes-change-2018-12-17
https://www.marketwatch.com/story/american-cheddar-and-other-cheese-piling-up-in-the-us-as-trade-slows-and-tastes-change-2018-12-17
https://www.marketwatch.com/story/american-cheddar-and-other-cheese-piling-up-in-the-us-as-trade-slows-and-tastes-change-2018-12-17
https://www.fb.org/market-intel/largest-decline-in-u.s.-dairy-farms-in15-plus-years-in-2019
https://www.fb.org/market-intel/largest-decline-in-u.s.-dairy-farms-in15-plus-years-in-2019
https://www.idfa.org/federal-milk-marketing-orders
https://www.idfa.org/federal-milk-marketing-orders
https://www.ams.usda.gov/rules-regulations/moa/dairy
https://www.ams.usda.gov/rules-regulations/moa/dairy
https://www.wsj.com/articles/farmers-deal-with-glut-of-food-as-coronavirus-closes-restaurants-11586439722
https://www.wsj.com/articles/farmers-deal-with-glut-of-food-as-coronavirus-closes-restaurants-11586439722
https://www.wsj.com/articles/farmers-deal-with-glut-of-food-as-coronavirus-closes-restaurants-11586439722
https://www.choicesmagazine.org/choices-magazine/theme-articles/agricultural-market-response-to-covid-19/covid-19-and-the-us-dairy-supply-chain
https://www.choicesmagazine.org/choices-magazine/theme-articles/agricultural-market-response-to-covid-19/covid-19-and-the-us-dairy-supply-chain
https://www.choicesmagazine.org/choices-magazine/theme-articles/agricultural-market-response-to-covid-19/covid-19-and-the-us-dairy-supply-chain
https://www.choicesmagazine.org/choices-magazine/theme-articles/agricultural-market-response-to-covid-19/covid-19-and-the-us-dairy-supply-chain
https://doi.org/10.2202/1542-0485.1151
https://www.wsj.com/articles/america-cant-move-its-cheese-11545042600
https://www.wsj.com/articles/america-cant-move-its-cheese-11545042600
https://www.ers.usda.gov/webdocs/publications/98901/err-274.pdf
https://www.ers.usda.gov/webdocs/publications/98901/err-274.pdf
https://doi.org/10.1515/jafio-2019-0002
https://doi:org/10.22434/IFAMR2019.0206
https://doi.org/10.1111/j.1467-6451.2008.00331.x
https://doi.org/10.1111/1477-9552.12219
https://www.ams.usda.gov/rules-regulations/mmr/dmr
https://www.ams.usda.gov/rules-regulations/mmr/dmr
https://www.washingtonpost.com/news/wonk/wp/2018/06/28/americas-cheese-stockpile-just-hit-an-all-time-high
https://www.washingtonpost.com/news/wonk/wp/2018/06/28/americas-cheese-stockpile-just-hit-an-all-time-high
https://www.washingtonpost.com/news/wonk/wp/2018/06/28/americas-cheese-stockpile-just-hit-an-all-time-high
https://www.washingtonpost.com/news/wonk/wp/2018/06/28/americas-cheese-stockpile-just-hit-an-all-time-high


45

Research on World Agricultural Economy | Volume 04 | Issue 03 | September 2023

[26] Hobson, J. Make America grate again? There’s 1.4 
billion pounds of surplus cheese in the U.S. Boston’s 
NPR News Station, 2019 Jan 8. Available from: 
https://www.wbur.org/hereandnow/2019/01/08/
cheese-surplus-united-states 

[27] Pesaran, M.H., Shin, Y., Smith, R.J., 2001. Bounds 
testing approaches to the analysis of level relation-
ships. Journal of Applied Econometrics. 16(3), 289-
326.

[28] Pesaran, H., Shin, Y., 1999. An autoregressive dis-

tributed lag modeling approach to cointegration anal-
ysis. Econometrics and economic theory in the 20th 
century: The Ragnar Frisch centennial symposium. 
Cambridge University Press: Cambridge, UK.

[29] Kripfganz, S., Schneider, D.C. (editors), 2016. ardl: 
stata module to estimate autoregressive distributed 
lag models. Stata Conference; 2016 Jul 28-29; Chica-
go. Stata Users Group. 

[30] Greene, W., 2000. Econometric analysis, 4th Edition. 
Prentice Hall: Hoboken.

https://www.wbur.org/hereandnow/2019/01/08/cheese-surplus-united-states
https://www.wbur.org/hereandnow/2019/01/08/cheese-surplus-united-states