=Paper= {{Paper |id=Vol-2992/icaiw_aiesd_1 |storemode=property |title=Empirical approach to arable land and livestock using cointegration and causality techniques with panel data |pdfUrl=https://ceur-ws.org/Vol-2992/icaiw_aiesd_1.pdf |volume=Vol-2992 |authors=Micaela Calderon,Marcelo Leon,Sergio Núñez |dblpUrl=https://dblp.org/rec/conf/icai2/CalderonLN21 }} ==Empirical approach to arable land and livestock using cointegration and causality techniques with panel data== https://ceur-ws.org/Vol-2992/icaiw_aiesd_1.pdf

Empirical approach to arable land and livestock
using co-integration and causality techniques with
panel data
Micaela Calderon1 , Marcelo León2 and Sergio Núñez1
1
Universidad Tecnológica Empresarial de Guayaquil, Guayaquil, Ecuador
2
Universidad Nacional de Loja, Loja, Ecuador

Abstract
The availability of food and the right to food is linked to the concept of food security, poverty and
development of nations; and, therefore, to a focus on agricultural production and availability of land
suitable for cultivation. In this context, the main objective of this research is to evaluate the effect of
livestock production on the availability of arable land in Latin American and Caribbean countries, be-
tween 1961-2017, using cointegration and causality techniques to propose policies that contribute to a
smaller decrease in arable land. For the development of the research, control variables were added: pop-
ulation growth, average temperature variation and fertilizer use. Statistical information was collected
from the World Bank (2020) and the Food and Agriculture Organization of the United Nations (2020)
databases. The main results show a statistically positive relationship and the existence of a long-run
equilibrium relationship between the variables used in the model. The results indicate that a 1% increase
in head of cattle is related to an increase of 0.04 hectares in arable land. On the other hand, it was found
that the main causes of variations in arable land are livestock and fertilizer use. Policy implications
suggest some measures to ensure the availability of arable land considering the role of livestock and
fertilizer use.

Keywords
Arable land, Livestock, Food security, Population growth, Development

1. Introduction
Currently, there is widespread concern regarding the decline of arable land in the world.
According to data from the World Bank (2020) the amount of arable land per capita in the world
has decreased from 0.36 ha in 1961 to 0.18 ha in 2017. Modern agriculture has been successful
in increasing food production, as food production has even outpaced population growth [1].
However, the number of hungry people continues to rise, reaching 821 million in 2017, 1 in 9 of
the world’s population. United Nations Children’s Fund UNICEF, 99.7% of the food that humans
need to survive comes from the land; therefore, productive land and soils are vitally important
to our lives and economies.

ICAIW 2021: Workshops at the Fourth International Conference on Applied Informatics 2021, October 28–30, 2021,
Buenos Aires, Argentina
" mica.sol92@hotmail.com (M. Calderon); mleon@uteg.edu.ec (M. León); sejunuso@hotmail.com (S. Núñez)
0000-0003-2626-8921 (M. Calderon); 0000-0001-6303-6615 (M. León); 0000-0001-8804-3088 (S. Núñez)
© 2021 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073 CEUR Workshop Proceedings (CEUR-WS.org)

44
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

In regions that base their economies on agriculture, such as Latin America and the Caribbean
(LAC), the agriculture sector contributes significantly to economic growth, representing more
than 5% of GDP in approximately 20 countries. Given that the poor are concentrated in rural
areas, it also contributes significantly to the reduction of poverty and hunger, and contributes
substantially to employment, accounting for between 10% and 15% of total employment. Thus,
the availability of land suitable for cultivation is of fundamental importance as it becomes a
means to overcome malnutrition and hunger, and also a source of income for rural people.
The availability of food and the right to food is linked to the concept of food security and,
therefore, to a focus on agricultural production and availability of arable land [2]. Therefore, in
order for people to have permanent access to food both physically and economically, it requires
the capacity and resources to produce or obtain all the food needed for the household and its
members [3]. The supply of food is not the main issue, the key is whether people can buy
enough food to be able to enjoy an adequate diet, which translates into a lack of access. Lack of
access to food can be economic, due to high levels of poverty, high food prices, lack of credit;
and physical, due to poor road and market infrastructure [4].
On the other hand, there is evidence of an increase in the production of beef cattle worldwide,
as; it went from 27,684,560 tons in 1961 to 67,353,900 tons in 2018. The evolution of the livestock
sector in Latin America and the Caribbean, maintains a rapid pace of growth that is more the
result of increased inventories than the adoption of technologies to increase yields. According to
Graziano et al., [5] in recent decades there has been a significant increase in the global demand
for animal products; it is estimated that by 2050 there will be an increase of up to 70%. However,
Latin America and the Caribbean has responded favorably to this trend, becoming the main
global exporter of beef and poultry meat; thus, exports of beef meat recorded an increase of 7%,
equivalent to $737 million additional in 2020, compared to 2019.
This paper focuses mainly on what Malthus [6] said, admitting the possibility that the eating
patterns of the upper class could cause an increase in the land devoted to livestock and, therefore,
a decrease in the available food. This is where the importance of the research topic lies; it seeks
to examine the relationship between the availability of arable land and livestock. It is important
to emphasize that the literature on the subject is scarce; therefore, this research contributes
as new knowledge to the existing scientific field with respect to the analysis of the variations
of arable land caused by cattle production, population growth, average temperature variation
and the use of fertilizers in 21 countries that make up the Latin America and the Caribbean
region; 15 belonging to the Latin America sector and six to the Caribbean. Therefore, the main
contribution of this study is based on the econometric strategy: second generation cointegration
techniques, which control for the presence of cross-sectional dependence between countries
and are more effective in assessing the effects of long-run determinants of arable land.
The existing literature is quite scarce regarding the variables used in this research. However,
it has been possible to gather information from the few studies found on the subject in question.
Studies such as Alexander et al., [7] who find a negative relationship between cattle production
and arable land. While, Rabés et al., [8]; Chai et al., [9] and He et al., [10] find that cattle ranching
implies a greater grabbing of arable land. Another group of studies point out that future food
supply is constrained by the excessive use of arable land primarily intended to produce feed for
livestock [11] and [12]. Also, the literature assumes that climate change and excessive fertilizer
use have both positive and negative implications on land availability and crop yields. Finally,

45
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

there is evidence that unsustainable practices in the agriculture sector end up aggravating the
latent problems on climate change [13] and [14].
The results of this research contribute to fill the empirical gap regarding the analysis of arable
land use from the perspective of livestock farming. It could be evidenced that cattle production
has a statistically positive relationship with arable land; as well as a negative relationship
between temperature variation, fertilizer use and arable land. In the same way, it is proved that
the dependent variable and the control variables present an equilibrium in the long term. In
terms of causality, it was found that the main causes of variations in the availability of arable
land are livestock and fertilizer use.

2. Literature review
The theory underpinning the present research is that of Malthus [6] who admits the possibility
that the eating patterns of the upper class could cause an increase in land devoted to livestock,
and thus a decrease in available food, implying that the increased demand for meat is reflected
in an increase in livestock stocks, and this in turn leads to less land being devoted to growing
food for humans. However, it must be considered that there are other additional factors that can
determine variations in arable land. In this sense, this section is divided into four groups that
are written according to the explanatory variables used: cattle production, population growth,
climate change and fertilizer use.
In the first group are studies such as Alexander et al., [7] which by using panel data taken
from FAO; and a decomposition analysis, mention that the improvement of people’s income
implies a diet with the highest amount of animal products, which may cause a further reduction
of land used for agriculture. Also, a Greenpeace report (2019) points out that industrial livestock
farming is causing land grabbing, with more than two thirds of arable land already devoted to
growing food for livestock, while food production for people is losing ground. Along the same
lines, Yawson [15] through an estimation of future food balances and with the use of FAO data
determines that it is necessary to substantially increase the area of land currently allocated to
barley to meet the projected demand for feed use in the future, since the results indicate that
productivity gains must be complemented by increasing the harvested area. Likewise, Rodrigues
et al., [11] used a classification approach based on Landsat satellite image objects, whereby they
mention that future land use is determined by agro-pastoral expansion. Similarly, Soltani et
al., [12] concludes that the future food supply and efficiency of the country is at stake, due to
overexploitation of water and land resources, mainly for livestock expansion, deforestation
and mining. FAO indicate that the main driver of land use change is the livestock sector, as
large tracts of land are converted to pasture or animal feed crops, with serious implications for
agriculture and food systems; as one of the main challenges is to produce more with less.
In contrast, Rabès et al., [8] who through an analysis of covariance estimated the effect of
four food diets on three environmental indicators; found that, food systems constitute a burden
on the environment and resource use; with animal-based foods representing the highest land
occupation and GHG emissions. For their part, Chai et al., [9] through a systematic review of
34 studies indicate that reducing meat consumption can placate the impact of the meat industry
on the environment, and He et al., [10] indicate that eating habits imply a greater demand for

46
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

arable land, destined to produce livestock food.
The second group, relating population growth to arable land, used unique panel data for rural
households containing information on soil quality and show that population pressure reduces
soil quality and also induces agricultural intensification. Used unique panel data for rural
households containing information on soil quality and show that population pressure reduces
soil quality and also induces agricultural intensification. Likewise, Prabhakar [16] mentions that
the rapidly growing population and its needs are one of the main drivers of land use change. In
the third group, relating climate change and arable land, are Huang et al., [17] who, using a
RUSLE model, estimated the spatio-temporal variations in the rate of soil loss and followed by a
scenario design to decouple the effects of climate and land use changes, found that changes in
climate cause soil losses, but this is compensated by reforestation. Huang et al., [17] mentions
that crop yields are more sensitive to increased rainfall and temperature variability; thus, they
indicate that to ensure food security under a changing climate, best management practices that
improve soil structure and nutrient retention should be adopted. Severe soil losses has become
a very important environmental problem and is strongly affected by climate change and land
use [18]. On the other hand, the impact generated by temperature on land and crops, has much
to do with the area and type of crop [19]; in some cases in the long term significantly increases
production; but also reduces the efficiency of soils [20]. Likewise, Shakhawat et al., [21], using a
Ricardian model conducted a cross-sectional regression analysis by which they established that
as temperature increases, a reduction in the value of agricultural land is projected. Similarly,
Hossain et al., [22]; Arshad et al., [23] find that climate change is causing land degradation by
decreasing its value, agricultural productivity and income. In fact, by 2030, climate change will
exacerbate extreme poverty problems through impacts on agriculture and food security.
Regarding arable land as a driver of climate change, Yang et al., [24] and Bell et al., [25] indicate
that abandoned land helps to combat climate change as it undergoes a natural recovery of
vegetation and soil carbon; and helps to remove carbon dioxide from the atmosphere. Mekonnen
et al., [13] found that improved soil and water conservation practices positively influence
soil physico-chemical properties, which in turn leads to reduced pollutant emissions and
improved soil quality as natural sinks. According to Silveira et al., [26] greenhouse gas emissions
are directly associated with climate change problems. Part of these emissions are caused by
agriculture, by the burning of fossil fuels such as coal, natural gas and oil used as a source
of energy for the performance of agricultural machinery. On the other hand, Del Buono [27]
indicates that soil salinity due to the excessive use of fertilizers is one of the most problematic
causes that will increase anthropogenic climate change. Likewise, Shakoor et al., [14] points
out that croplands due to their large area and management practices emit harmful emissions to
the environment, which end up aggravating the problems of climate change.
Finally, the fourth group contains studies that relate fertilizer use to arable land, for example,
Hao et al., [28], conducting a field experiment in a moderate acid soil to quantify soil acidification
rates in response to fertilization with different types of fertilizers revealed that soil acidification
is induced by nitrogen fertilizers. Soil acidification causes changes in soil properties, hard soils
and susceptibility to pests; thus, the amount of chemical fertilizer applied to agricultural land
has resulted in nutrient losses from soils, which reduces water and nutrient holding capacity
and thus crop yields. Likewise, Zhang et al., [29] used a combination of scenario analysis and an
agricultural survey of 1500 farmers across China to explore the impacts of replacing fertilizer

47
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

with manure, and determined that there is a need for a transparent manure exchange market,
with advisors on manure use, accurate information on composition and price. Chai et al., [9]
investigated 12 types of biomass resources and calculated their nitrogen pools, where they
found that organic fertilizers have enormous potential to replace chemical fertilizers, mitigating
threats to the environment and human health.
Similarly, Habtemariam et al., [30] through a trade-off analysis for multidimensional impact
assessment model determines that the adoption of fertilizer and rainwater harvesting, reduces
the percentage of food insecure people and improves the yield and income of many farmers.
However, Adnan et al., [31] however, mentions that these resources alone do not generate
significant changes in terms of reducing poverty and food insecurity. In addition, Toledo [32]
indicates that the lack of knowledge about soil management has led farmers to use methods
based on the excessive use of fertilizers, which generates a better productive response of crops
and at the same time deterioration of productive soils.

3. Data and methodology
3.1. Data
The data used in this research were obtained from FAO and World Bank statistical sources. The
dependent variable used was the amount of arable land per capita in hectares, which measures

Table 1
Description of variables and data sources
Variable and Unit of Source Description
notation measure of data
Arable land per Hectares World These are those on which crops are grown that occupy the
capita (tcp) Bank land for extended periods of time. This category includes
land with flowering shrubs, fruit trees and nut trees, but
excludes land on which timber trees are grown.
Heads of cattle Stocks FAO Number of head of live cattle, animals that are used for
(lcgv) the production of meat and milk.
Temperature Degrees FAO Change in average surface temperature with respect to a
variation (vt) Celsius reference climatology. These average temperature varia-
tions are expressed in degrees Celsius.
Population Growth World The annual population growth rate is the mid-year popu-
growth (cpob) rate Bank lation growth from year t-1 to year t, expressed as a per-
centage. All residents are counted regardless of their legal
status or citizenship.
Fertilizer use Kilogram FAO They are those to which nitrogen or compounds derived
(lfn) from it are incorporated, they are used to promote plant
growth and improve their cellular structure. They cause
water and atmospheric pollution.

48
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Figure 1: Correlation between head of cattle, mean temperature variation, population growth, fertilizer
use and arable land per capita.

the availability of arable land for each person, and the independent variable was the number of
heads of cattle in units, which measures the presence of livestock in the region. Subsequently,
three control variables were added, as they were considered important for the estimation and,
above all, to make the model more robust: climate change, fertilizer use and population growth.
Due to the limited availability of data, only 21 countries that make up the Latin American and
Caribbean region could be covered; 15 belonging to the Latin American sector and six to the
Caribbean. In addition, fertilizer use and head of cattle are transformed into logarithms, while
mean temperature variation, arable land per capita and population growth are worked in their
original unit of measurement. Table 1 details the variables used in the econometric models.
In Figure 1 , the correlations between the variables cattle head, population growth, tempera-
ture variation, fertilizer use and arable land per capita show a positive and significant correlation,
which suggests that these variables play a very important role in the availability of arable land
in Latin American and Caribbean countries. However, mean temperature variation shows a
negative correlation with arable land; and, population growth a slight inclination towards a
positive correlation.
Subsequently, Table 2 presents the descriptive statistics of the dependent and independent
variable including the control variables, where it is observed that the data panel is balanced
because it consists of a total of 1197 observations, which in general involves 21 countries in a

49
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Table 2
Descriptive statistics
Variable Mean Thirst. Dev Min Max Observations
overall 0,263 0,213 0.018 1,101 N=1197
Arable land per capita between 0,202 0.039 0,896 n= 21
within 0,079 0,052 0,662 T= 57
overall 14,709 2,489 7,600 19,200 N=1197
Log Heads of cattle between 2,533 8,429 18,684 n= 21
within 0,288 13,759 15,386 T= 57
overall 1,641 0,938 -2,099 3,588 N=1197
Population growth between 0,752 0,029 2,687 n= 21
within 0,584 -2,150 3,081 T= 57
overall 0,393 0,447 -0,785 1,697 N=1197
Average temperature variatio between 0,095 0,177 0,551 n= 21
within 0,437 -0,713 1,649 T= 57
overall 17,259 2,285 2,625 22,367 N=1197
Log Fertilizer us between 2,094 11,818 20,525 n= 21
within 1,020 8,065 20,599 T= 57

period of time of 57 years. In addition, the mean, standard deviation, maximum and minimum
values and the number of observations are shown. The variation between countries is larger than
within countries, both for arable land per capita, cattle head, population growth and fertilizer
use. In contrast, the mean variation in temperature is greater within countries than between
countries, hence there is greater variability within countries. In conclusion, the variation of
the variables is mostly explained by the standard deviation between countries. It should be
emphasized that the maximum value of the average temperature variation indicates that the
temperature increase has been maintained at less than 2°C as proposed in the Paris Agreement
approved in 2015, as its value is 1.7°C, although this raises concern as it is a value very close to
the objective set out in the Paris Agreement.

3.2. Methodology
In order to evaluate the effect of livestock on arable land, a panel data analysis was carried out
using cointegration techniques for 21 countries in Latin America and the Caribbean, during
the period 1961-2017. The econometric strategy used was carried out according to the specific
objectives [33]; therefore, the research had several stages that are explained below. Firstly, a
baseline regression is based on the theoretical contribution of Malthus [6] which indicates that
the feeding patterns of the upper class could cause an increase in the land dedicated to livestock;
and, therefore, a decrease in the available food. Therefore, the present relationship is formalized
in Equation (1):

tcp 𝑖𝑡 = 𝛽0 + 𝛽1 lcgv 𝑖𝑡 + 𝜀𝑖𝑡 (1)

50
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

In order to determine what other variables affect the availability of arable land in Latin
American and Caribbean countries, the estimation of a Generalized Least Squares (GLS) model
is presented [34], which prior to its estimation it is very necessary to apply diagnostic tests such
as multicollinearity, autocorrelation and heteroscedasticity [35]. The model includes control
variables such as: population growth, average temperature variation and fertilizer use, whose
estimation is presented in Equation (2):

tcp 𝑖𝑡 = 𝛽0 + 𝛽1 lcgv 𝑖𝑡 + 𝛽2 cpob 𝑖𝑡 + 𝛽1 vt 𝑖𝑡 + 𝛽1 lfn 𝑖𝑡 + 𝜀𝑖𝑡 (2)
In Equation (1) and (2) tcp representing arable land per capita; lcgv representing the loga-
rithm of cattle production; cpob which represents population growth; vt indicates the average
temperature variation; lfn is the logarithm of fertilizer use; for the countries in the period
𝑖 = 1, 2, 3, ..., 21 in the period 𝑡 = 1961, 1962, ..., 2017 and, finally, the 𝜀𝑖𝑡 it is the error term.
Consecutively, to avoid biased and inconsistent results we test for cross-sectional dependence
using the Pesaran diagnostic test. (2015) which are recommended for balanced and unbalanced
panels. This Pesaran 𝐶𝐷𝑁 𝑇 of the Pesaran test (2015) has the following expression, described
in Equation (3):
⎡ ⎤
−1 ∑︁
√︃ 𝑁 𝑁 √
2 ∑︁
CD NT = ⎣ 𝑇 𝜌̂︀ij ⎦ 𝑁 (0, 1) (3)
𝑁 (𝑁 − 1)
𝑖=1 𝑗=𝑖+1

Where, 𝑁 denotes the number of cross sections (countries), 𝑇 indicates the period and 𝜌̂︀ij
shows the correlation by ordered pairs corresponding to the cross sections in each period, as
described in equation (4):
𝑇
∑︁
𝜌̂︀ij = 𝑇 −1 𝜀it 𝜀jt (4)
𝑖=1

Where, 𝜀𝑖𝑡 y 𝜀𝑗𝑡 denotes the scaled residuals of the specific Ordinary Least Squares (OLS)
regressions for each cross section (countries). 𝑖 = 1, 2, 3, ..., 𝑁 .
Therefore, for the panel with presence of cross-sectional dependence we estimate the second
generation unit root tests that are more robust and reliable in this case, for which we estimate
the CADF and CIPS* tests proposed by Pesaran (2007). Therefore, the first known cross-sectional
augmented Dickey-Fuller (CADF) test is specified in equation (5):

𝑌𝑖 = 𝛼𝑖 + 𝛽𝑖 𝑌𝑖,𝑡−1 + 𝜔0 𝑌̂︀𝑡−1 + 𝜙𝑖 ∆𝑌̂︀𝑡 + 𝜀it (5)

Where, 𝑌𝑖 = 𝛼𝑖 + 𝛽𝑖 𝑌𝑖,𝑡−1 + 𝜔0 𝑌̂︀𝑡−1 + 𝜙𝑖 ∆𝑌̂︀𝑡 + 𝜀it is regression error.
As for, the second test is calculated from the average of the individual ADF statistics aug-
mented in the cross section (CADF) is called CIPS* which analyzes the unit root properties of
the whole panel as shown in equation (6):
𝑁
1 ∑︁
CIPS * = CADF 𝑖 (6)
𝑁
𝑖=1

51
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Where, 𝐶𝐴𝐷𝐹𝑖 denotes the cross-sectional augmented Dickey-Fuller statistic for 𝑖 which
represents each cross-sectional unit.
In the presence of cross-sectional dependence, the Westerlund error correction test is applied
to verify the long-run relationship between the variables. Is applied to verify the long-run
relationship between the variables. The test allows us to conclude whether cointegration exists
for individual panels as well as for the whole panel as a whole, considering that the variables
analysed are stationary. Equation (7) expresses the error correction that defines the speed of
correction towards equilibrium:
𝑝𝑖
∑︁ 𝑝
∑︁
∆𝑦𝑖,𝑡 = 𝛿𝑑̇ 𝑡 + 𝜀𝑖 (𝑦𝑖,𝑡−1 − 𝛽𝑖 𝑥𝑖,𝑡−1
̇ )+ 𝜙𝑖,𝑗 𝑦𝑖,𝑡−1 + 𝜙𝑖,𝑗 𝑦𝑖,𝑡−1 + 𝜀𝑖𝑡 (7)
𝑗=1 𝑗=𝑞𝑖

Where, 𝑡 = 1, 2, 3, .., 𝑇 ; 𝑖 = 1, 2, 3, ..., 𝑁 y 𝑑𝑡 express the deterministic components; +𝜀𝑖
represents the constant term; 𝑝𝑖 y 𝑞𝑖 denote the orders and advancement of each of the countries.
The Westerlund [36] test yields four statistics where 𝐺𝜏 y 𝐺𝛼 indicate that at least one
cross section is cointegrated and the statistics 𝑃𝜏 y 𝑃𝛼 statistics show that the whole panel is
cointegrated, to evaluate the null hypothesis of no cointegration as shown in equations (8-11):
𝑁
1 ∑︁ 𝜀𝑖
𝐺𝜏 = (8)
𝑁 Se (̂︀
𝜀𝑖 )
𝑖=1

𝑁
1 ∑︁ 𝑇𝑖
𝐺𝛼 = (9)
𝑁 𝜀̇ 𝑖
𝑖=1

𝜀̂︀𝑖
𝑃𝜏 = (10)
Se (̂︀𝜀𝑖 )

𝑃𝛼 = 𝑇 𝜀̂︀ (11)
Finally, it was necessary to apply the Granger-type causality tests developed by Dumitrescu
and Hurlin [37] to check whether the results of a variable serve to predict another variable; that
is, whether these have a unidirectional or bidirectional behavior. More generally, if the behavior
of A causes the behavior of B in Granger’s sense, the relationship is said to be unidirectional. If,
on the other hand, the behavior of B predicts the behavior of A, a bidirectional relationship is
said to exist. The formal representation is presented in Equation (12):
𝐾
∑︁ 𝐾
∑︁
𝑦𝑖,𝑡 = 𝛼𝑖 + 𝛾𝑖𝑘 𝑦𝑖,𝑡−𝑘 + 𝛽𝑖𝑘 𝑋𝑖,𝑡−𝑘 + 𝜇𝑖,𝑡 (12)
𝑘=1 𝑘=1

Where, 𝛼𝑖 the intersection of the slope; 𝑘 shows the lag orders in all units assuming the
panel is balanced; 𝛾𝑖(𝑘) is the autoregressive parameter, 𝛽𝑖(𝑘) indicates the regression coefficient
differing between cross sections.

52
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

4. Discussion of results
Prior to the estimation of the GLS model, the existence of multicollinearity, heteroscedasticity,
autocorrelation was determined, the results of which ruled out multicollinearity; and, corrobo-
rated the presence of autocorrelation and heteroscedasticity. Consequently, the results of the
GLS model estimation are presented in Table 3, which indicate that the cattle head variable
shows a statistically significant and positive relationship with arable land; including the control
variables. That is, a 1% increase in cattle head is associated with a 0.04 hectare increase in arable
land. The variables mean temperature variation and fertilizer use also show significance and
a negative relationship with arable land, i.e., a 1°C increase in temperature causes a 0.004 ha
decrease in arable land area. Similarly, a 1% increase in fertilizer use decreases arable land by
0.012 hectares.
The positive relationship between livestock and arable land would not be favorable, since
this increase would be destined to livestock breeding and production, but not to feeding human
beings, which would endanger the food security of the inhabitants of the Latin American
and Caribbean region, worsening the scenarios of hunger and malnutrition. Likewise, this
increase in the use of arable land due to livestock farming indicates the lack of productivity
and performance of the livestock sector in the region, which implies higher costs for each unit
produced. These results do not support the contribution of Malthus [6] who points out that the
feeding patterns of the upper class would lead to an increase in land devoted to livestock and,
therefore, a decrease in the food available to the poorest. Likewise, Alexander et al., [7] also
note that a diet involving more animal products can lead to a further reduction in land devoted
to agriculture. Similarly, a Greenpeace report report points out that industrial livestock farming
is causing land grabbing, mostly for livestock feed, while food production for people is losing
ground.
On the contrary, He et al., [10] agrees with the present results; since, these authors indicate
that food habits and urbanization imply a greater demand for arable land, mainly destined to
produce food of animal origin. Similarly, Rabès et al., [8] mentions that food systems constitute
a burden for the environment and the use of resources; being food of animal origin the ones
that represent a greater occupation of land and emission of GHG. Therefore, Chai et al., [9]
indicates that reduction of meat consumption can placate the impact of the meat industry on
the environment.
Climate variability, when associated with events such as heavy rains, floods and droughts, are
factors that affect the production and distribution of crops, as these often lead to crop losses and
high costs. These losses also imply a reduction in income for the people who mostly live in rural
areas and depend mainly on agriculture, which can worsen the situation of poverty and affect
economic growth, since the agricultural sector contributes significantly to the total production
of the countries that make up the region. The present results are supported by Huang et al., [17]
and Huang et al., [38] who mention that increased rainfall and temperature variability cause soil
losses and soil sensitivity, but that this effect can be compensated by reforestation; therefore,
they indicate that to ensure food security under a changing climate, best management practices
that improve soil structure and nutrient retention should be adopted. Severe soil losses have
become a very important environmental problem and it is strongly affected by climate change
and land use [18].

53
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Table 3
Estimation of the GLS model including control variables
Basic Model Model with control variables
lcgv 0.00537* 0.0443***
(2.19) (25.94)
cpob -0.00543
(-1.38)
vt -0.00434*
(-2.37)
lfn -0.0117***
(-8.79)
Constant 0.0637 -0.217***
(1.88) (-11.08)
Observations 1197 1197
Adjusted R2
Note: * significance at 5%, **significance at 10%, ***significance at 1%.)

The use of fertilizers has its advantages and disadvantages. On the one hand, they increase
crop yields, thus allowing to feed the growing population; but, on the other hand, their excessive
use causes soil degradation, since their chemical composition causes damage to the soil, which
with excessive and prolonged use ends up worsening the health of the soil. In addition to this,
fertilizers are groundwater polluters that then end up being sources of irrigation for crops,
whose situation affects the normal growth of plants; and at the same time, contamination of
food, which translates into serious health problems for humans. These findings are similar to
those found by Hao et al., [28], who demonstrated that soil acidification is induced by nitrogen
fertilizers. Soil acidification causes changes in soil properties, hard soils and susceptibility to
pests; thus, the amount of chemical fertilizer applied to agricultural land has caused nutrient
losses from soils, which reduces the water and nutrient holding capacity and thus crop yields.
On the other hand, Zhang et al., [29] indicate that organic fertilizers have enormous potential
to replace chemical fertilizers, thus mitigating threats to the environment and human health.
Prior to estimating the long-term relationships, we applied the cross-sectional dependence
tests proposed by Pesaran which uses the correlation coefficients between the series of each
of the countries. Therefore, Table 4 reports the results of the cross-sectional dependence test
(CD); and, given that the p-value is less than 0.001, the null hypothesis of no cross-sectional
dependence between countries is strongly rejected and it is concluded that the variables present
cross-sectional dependence between countries at a significance level of 0.1%.
Since there is the presence of transversal dependence within the model, it is pertinent to
apply the second generation unit root tests, in order to verify the stationarity of the panel data
series, and the same happens with the long term estimations. In this sense, the CADF tests
and the CIPS of Pesaran which are more robust and reliable in the presence of cross-sectional
dependence. Table 5 shows the results obtained from the unit root tests in levels (constant and

54
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Table 4
Tests of cross-sectional dependence
Test Pesaran (2015)
Variables CD p-value
Arable land per capita 104.021*** 0.000
Log Heads of cattle 109.364*** 0.000
Population growth 92.356*** 0.000
Average temperature variation 90.863*** 0.000
Log Fertilizer use 109.072*** 0.000
Note: t denotes significance * p < 0.05, ** p < 0.01, *** p < 0.001

Table 5
Results of unit root tests on levels and second differences
CADF Test Statistics
Levels Second differences
Variables Constant Constant and trend Constant Constant and trend Order
Arable land per capita -1.873 -2.191 -3.782*** -4.139 *** I (1)
Log Heads of cattle -1.998 -2.440 -3.748 *** -3.729 *** I (1)
Population growth -1.401 -1.557 -4.877 *** -5.430 *** I (1)
Average temperature variation -3.727*** -3.599*** -5.989 *** -6.119*** I (1)
Log Fertilizer use -2.286 * -2.860*** -5.680 *** -5.793*** I (1)
CIPS Test Statistics
Arable land per capita -1.823 -2.237 -4.817*** -5.368*** I (1)
Log Heads of cattle -1.893 -2.236 -5.783*** -5.733*** I (1)
Population growth -1.469 -1.626 -3.176*** -3.595 *** I (1)
Average temperature variation -5.680*** -5.941*** -6.190*** -6.420*** I (1)
Log Fertilizer use -2.992 *** -3.783 *** -6.190*** -6.420*** I (1)
Note: t denotes significance * p < 0.05, ** p < 0.01, *** p < 0.001

constant-trend), in which it is observed that three series of five are non-stationary at level I (0).
Therefore, the first difference was performed to all variables to become stationary (constant and
constant-trend), determining that the series have an order of integration I (1) at a significance
level of 0.1%.
Once the order of integration of the variables is determined, we estimate the long-run
equilibrium relationship between the variables included in the model [39]. Considering that
the model presents transversal dependence, we apply the cointegration test developed by
Westerlund [36]. In order to interpret the results, it should be kept in mind that the statistics
Gt and Ga, test the alternative hypothesis that at least one unit is cointegrated, while Pt and
Pa test the alternative hypothesis that the panel is cointegrated [40]. The results for all the
four statistics that this test yields determine that there is cointegration between the variables;
since the p-values are less than 0.05, the results of which are shown in Table 6. Consequently, a
long-term relationship between the variables included in the model implies keeping in mind
that the availability of arable land is compromised in the future by variations in the number
of cattle, population growth, temperature variation and fertilizer use, which generates food,

55
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Table 6
Results of Westerlun’s cointegration test
Statistic Value Z-value VP-value
Gt -3,147*** -3,366 0,000
Ga -14,567*** -0,963 0,006
Pt -14,178** -4,000 0,029
Pa -17,480*** -4,941 0,000
Note: t denotes significance * p < 0.05, ** p < 0.01, *** p < 0.001

poverty and development problems for future generations.
Having clear that land is a limited resource, it is worrying the excessive use of it, as it can
reach the day when there is no more land to produce. Therefore, we should be concerned
about the available resources in the future, in the scenario that over time the demand for land
increases, and its use is not only to produce food, but for all kinds of activities that mostly
generate environmental damage and decrease of productive soils; in itself, over time the extent
of land will not disappear but if the amount of land suitable for cultivation will be reduced, due
to the decrease of nutrients and fertility necessary for food production needed by humans, not
only for their physical survival but also: economic and social. These results are validated by the
contribution of Yawson [15] who mentions that it is necessary to substantially increase the area
of land currently allocated to barley to meet the projected demand for feed use in the future,
since the results indicate that productivity gains must be complemented by increasing the
harvested area. Likewise, Rodrigues et al., [11] determines that future land use is determined by
agro-pastoral expansion. For their part, Soltani et al., [12] in Iran concludes that the future food
supply and efficiency of the country is at stake, due to the overexploitation of water resources
and land, mainly for the expansion of the livestock sector, deforestation and mining.
The long-run relationship between arable land and control variables is supported by Mugizi
et al., [41] who show that population pressure reduces soil quality and induces agricultural
intensification. This suggests that, although farmers are trying to mitigate the negative effects
of population, the rate of soil degradation is outpacing the rate of intensification. Likewise,
Prabhakar [16] mentions that the rapidly growing population and its needs are one of the main
drivers of land use change. On the other hand, the impact of temperature on land and crops has
a lot to do with the area and the type of crop; in some cases in the long term it significantly
increases production, but also reduces soil efficiency [20]. Likewise, Shakhawat et al., [21];
Hossain et al., [22]; Arshad et al., [23] find that climate change is causing land degradation
by decreasing its value, agricultural productivity and income. In fact, by 2030, climate change
will exacerbate extreme poverty problems through impacts on agriculture and food security.
Regarding the relationship with fertilizer use, those who agree with our findings are Fontana et
al., [42] who find that the proper use of fertilizers and the implementation of cover crops lead
in the long term to improved soil properties and higher soil yields.
Finally, in order to estimate causal relationships, the Granger causality test developed by
Dumitrescu and Hurlin [37] was used. In this context, it is established that the fact that two
variables are correlated with each other does not necessarily imply causality, i.e. the fact that

56
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

Table 7
Dumitrescu and Hurlin’s causality results
Address W-bar Z-bar P-value Conclusion
tch - lcgv 2,7994 5,8307 0,0000 Causality
lcgv - tch 3,1447 6,9497 0,0000 Causality
tch - cpob 10.1204 29.5535 0,0841 Non-causality
cpob - tch 1.7385 2.3930 0.0746 Non-causality
tch - vt 16.3029 49.5871 0,0000 Causality
vt - tch 0.6681 -1.0753 0.2822 Non-causality
tch - lfn 5.7041 15.2431 0,0000 Causality
lfn - tch 3.6789 8.6806 0,0000 Causality
Note. Adapted from FAO (2020) and World Bank (2020).

one variable correlates to another does not imply that this is the cause of the alterations in the
values of another. The results of the test show that there is a bidirectional causal relationship
between arable land and cattle production, and also between arable land and fertilizer use.
That is, variations in arable land cause variations in cattle production and variations in cattle
production cause variations in arable land, and the same is true for fertilizer. Similarly, arable
land shows unidirectional causality with mean temperature variation. The results are shown in
Table 7.
Specifically, it is determined that the main causes of variations in the availability of arable
land are livestock and fertilizer use, which can be said to generate distortions in both the extent
and degradation of land. On the one hand, livestock farming implies greater land use and at the
same time the sector emits emissions such as methane that are absorbed by the soil, since the soil
is a natural sink for emissions. These studies are find that the main driver of land use change is
the livestock sector, as large tracts of land are converted to pasture or feed crops. Between 1960
and 2011, animal food production has been responsible globally for 65% of land use change and
the expansion of cultivated land. Similarly, FAO mentions that increasing livestock production
has implications for agriculture and food systems; as one of the main challenges is to produce
more with less, while preserving and improving farmers’ livelihoods.
Fertilizer use on its part, as well as helps to improve soil yield and productivity; it equally
emits pollutant gases such as nitrous oxide, which in conjunction with methane are harmful to
the environment and human health. These findings are similar to those found by Habtemariam
et al., [30] and Adnan et al., [31] who mention that the adoption of fertilizers and rainwater
harvesting reduces the percentage of food insecure people and improves the yields and incomes
of many farmers. However, these resources alone do not generate significant changes in terms
of reducing poverty and food insecurity. In addition, Toledo [32] indicates that the lack of
knowledge about soil management has led farmers to use methods based on the excessive
use of fertilizers, which generates a better productive response of crops and at the same time
deterioration of productive soils.
Finally, a very important result within causality is that the availability of arable land is the
cause of the average temperature variation; whose relationship is explained by the emissions
that are thrown by the agricultural sector at the time of producing and distributing food; because,

57
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

these cause alterations in temperature. The production of land involves the use of chemicals that
release gases into the air and penetrate the soil, which act as natural carbon sinks, which alters
the climate system by increasing the amount of gases that are responsible for heat retention on
the planet, and thus aggravates the problems of climate change that are evident today. Thus, our
results are similar to the studies of Yang et al., [24] and Bell et al., [25] who state that abandoned
lands help to combat climate change as they experience a natural recovery of vegetation and
soil carbon; and help to remove carbon dioxide from the atmosphere. Similarly, Mekonnen
et al., [13] finds that improved soil and water conservation practices positively influence soil
physico-chemical properties, which in turn leads to reduced pollutant emissions and improved
soil quality as a natural sink.
According to Silveira et al., [26] greenhouse gas emissions are directly associated with climate
change problems. Part of these emissions are caused by agriculture, by the burning of fossil fuels
such as coal, natural gas and oil used as a source of energy for the performance of agricultural
machinery. On the other hand, Del Buono [27] indicates that soil salinity due to the excessive
use of fertilizers is one of the most problematic causes that will increase anthropogenic climate
change. Likewise, Shakoor et al., [14] points out that croplands due to their large area and
management practices emit harmful emissions to the environment, which end up aggravating
the problems of climate change. Finally, it is important to note that the discussion of causality
was not precisely conducted with studies using the same methodology.

5. Conclusions
The results of the present investigation do not fulfill the first hypothesis raised for the analysis
of the availability of arable land, seen from a Malthusian point of view; since, the results indicate
that the production of bovine cattle is increasing the arable land. Therefore, it is important
to emphasize that the increase of arable land is not destined precisely to feed the population,
but to the production of fodder, legumes and pastures for food and livestock breeding, which
leads to a smaller amount of land for the cultivation of food for human beings. These results
contribute to think that people include in their diet mostly meat, whose production generates
serious and irreversible environmental problems compared to crop production.
The cointegration tests applied to the model allowed us to appreciate that the variables have
a short and long term equilibrium relationship. Concretely, these relationships imply concern
for present and future generations and incite to propose and implement policies urgently, to
counteract the problems of food security and poverty that are seen to come to the lack of land
to cultivate. It should be considered that, not only are variations in arable lands evident in
extension, but also in the form of degradation and loss of fertility, which would be the result of
the abuse of chemicals and the excessive and unsustainable use of soils when cultivating food.
These findings allow for the acceptance of the second hypothesis raised above.
Bidirectional causality was found between cropland and livestock and fertilizer use. Likewise,
arable land is a cause of climate change. Livestock and fertilizer use, being polluting gas emitting
activities, generate a feedback effect, since their operation causes variations in extension and
degradation of arable land, which, in turn, generates impacts again on livestock and fertilizer
use, since it will be necessary to apply more fertilizers or extend the land for livestock. Climate

58
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

change, which is also related to GHG emissions, is affected by agriculture, due to unsustainable
agricultural practices. In this sense, the results allow us to accept the third hypothesis.
In general, it is found that the rapid growth of the livestock sector is mostly attributed to the
increase in livestock stocks, but not to the improvement in productivity, which implies greater
land use. On the other hand, the evident reduction in arable land is not primarily due to cattle
production, but to fertilizer use and climate change; therefore, further research is needed on the
determinants of this reduction. The short and long term relationship between the variables are
results that should draw the attention of governments to act immediately to reduce as much as
possible the reduction of arable land, which is the fundamental basis of the economies of Latin
America and the Caribbean.

References
[1] N. J. Smelser, P. B. Baltes, et al., International encyclopedia of the social & behavioral
sciences, volume 11, Elsevier Amsterdam, 2001.
[2] M. Fuster, E. Messer, P. Palma, H. Deman, O. Bermudez, ¿ se considera la alimentación
saludable parte de la seguridad alimentaria y nutricional?: perspectivas desde comunidades
pobres de el salvador, Perspectivas en Nutrición Humana 16 (2014) 11–24.
[3] M. C. Latham, Nutrición humana: en el mundo en desarrollo, volume 29, Fao Roma, Italia,
2002.
[4] D. F. Pedraza, Seguridad alimentaria y nutricional. determinantes y vias para su mejora,
Revista Salud Pública y Nutrición 6 (2005).
[5] J. Graziano da Silva, M. Jales, R. Rapallo, E. Díaz-Bonilla, G. Girardi, M. del Grossi, C. Luiselli,
O. Sotomayor, A. Rodríguez, M. Rodrigues, et al., Sistemas alimentarios en América Latina
y el Caribe: Desafíos en un escenario pospandemia, Food & Agriculture Org., 2021.
[6] T. R. Malthus, J. M. Keynes, P. de Azcárate Diz, J. Vergara, Primer ensayo sobre la población,
Alianza Editorial, 1966.
[7] P. Alexander, M. D. Rounsevell, C. Dislich, J. R. Dodson, K. Engström, D. Moran, Drivers
for global agricultural land use change: The nexus of diet, population, yield and bioenergy,
Global Environmental Change 35 (2015) 138–147.
[8] A. Rabès, L. Seconda, B. Langevin, B. Allès, M. Touvier, S. Hercberg, D. Lairon, J. Baudry,
P. Pointereau, E. Kesse-Guyot, Greenhouse gas emissions, energy demand and land use
associated with omnivorous, pesco-vegetarian, vegetarian, and vegan diets accounting for
farming practices, Sustainable Production and Consumption 22 (2020) 138–146.
[9] B. C. Chai, J. R. van der Voort, K. Grofelnik, H. G. Eliasdottir, I. Klöss, F. J. Perez-Cueto,
Which diet has the least environmental impact on our planet? a systematic review of
vegan, vegetarian and omnivorous diets, Sustainability 11 (2019) 4110.
[10] G. He, Y. Zhao, L. Wang, S. Jiang, Y. Zhu, China’s food security challenge: Effects of food
habit changes on requirements for arable land and water, Journal of Cleaner Production
229 (2019) 739–750.
[11] E. R. da Cunha, C. A. G. Santos, R. M. da Silva, V. M. Bacani, A. Pott, Future scenarios
based on a ca-markov land use and land cover simulation model for a tropical humid basin
in the cerrado/atlantic forest ecotone of brazil, Land Use Policy 101 (2021) 105141.

59
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

[12] A. Soltani, S. Alimagham, A. Nehbandani, B. Torabi, E. Zeinali, E. Zand, V. Vadez,
M. Van Loon, M. Van Ittersum, Future food self-sufficiency in iran: A model-based
analysis, Global Food Security 24 (2020) 100351.
[13] M. Mekonnen, T. Abeje, S. Addisu, Integrated watershed management on soil quality,
crop productivity and climate change adaptation, dry highland of northeast ethiopia,
Agricultural Systems 186 (2021) 102964.
[14] A. Shakoor, S. Shakoor, A. Rehman, F. Ashraf, M. Abdullah, S. M. Shahzad, T. H. Farooq,
M. Ashraf, M. A. Manzoor, M. M. Altaf, et al., Effect of animal manure, crop type, climate
zone, and soil attributes on greenhouse gas emissions from agricultural soils—a global
meta-analysis, Journal of Cleaner Production 278 (2021) 124019.
[15] D. O. Yawson, Estimating virtual land use under future conditions: Application of a food
balance approach using the uk, Land Use Policy 101 (2021) 105132.
[16] S. Prabhakar, A succinct review and analysis of drivers and impacts of agricultural land
transformations in asia, Land Use Policy 102 (2021) 105238.
[17] R. Huang, X. Gao, F. Wang, G. Xu, Y. Long, C. Wang, Z. Wang, M. Gao, Effects of biochar
incorporation and fertilizations on nitrogen and phosphorus losses through surface and
subsurface flows in a sloping farmland of entisol, Agriculture, Ecosystems & Environment
300 (2020) 106988.
[18] Y. Guo, C. Peng, Q. Zhu, M. Wang, H. Wang, S. Peng, H. He, Modelling the impacts of
climate and land use changes on soil water erosion: model applications, limitations and
future challenges, Journal of environmental management 250 (2019) 109403.
[19] K. Daza, J. Hernandez, H. Florez, Hardware and software system for hydric estimation
and crop irrigation scheduling, in: International Conference on Computational Science
and Its Applications, Springer, 2019, pp. 150–165.
[20] S. Rahman, A. R. Anik, Productivity and efficiency impact of climate change and agroecol-
ogy on bangladesh agriculture, Land Use Policy 94 (2020) 104507.
[21] M. S. Hossain, M. Arshad, L. Qian, H. Kächele, I. Khan, M. D. I. Islam, M. G. Mahboob,
Climate change impacts on farmland value in bangladesh, Ecological indicators 112 (2020)
106181.
[22] M. S. Hossain, M. Arshad, L. Qian, M. Zhao, Y. Mehmood, H. Kächele, Economic impact
of climate change on crop farming in bangladesh: An application of ricardian method,
Ecological Economics 164 (2019) 106354.
[23] M. Arshad, T. Amjath-Babu, S. Aravindakshan, T. J. Krupnik, V. Toussaint, H. Kächele,
K. Müller, Climatic variability and thermal stress in pakistan’s rice and wheat systems:
A stochastic frontier and quantile regression analysis of economic efficiency, Ecological
indicators 89 (2018) 496–506.
[24] Y. Yang, S. E. Hobbie, R. R. Hernandez, J. Fargione, S. M. Grodsky, D. Tilman, Y.-G. Zhu,
Y. Luo, T. M. Smith, J. M. Jungers, et al., Restoring abandoned farmland to mitigate climate
change on a full earth, One Earth 3 (2020) 176–186.
[25] S. M. Bell, C. Barriocanal, C. Terrer, A. Rosell-Melé, Management opportunities for soil
carbon sequestration following agricultural land abandonment, Environmental Science &
Policy 108 (2020) 104–111.
[26] F. da Silveira, J. E. Ruppenthal, F. H. Lermen, F. M. Machado, F. G. Amaral, Technologies
used in agricultural machinery engines that contribute to the reduction of atmospheric

60
Micaela Calderon et al. CEUR Workshop Proceedings 44–61

emissions: A patent analysis in brazil, World Patent Information 64 (2021) 102023.
[27] D. Del Buono, Can biostimulants be used to mitigate the effect of anthropogenic climate
change on agriculture? it is time to respond, Science of The Total Environment 751 (2021)
141763.
[28] T. Hao, Q. Zhu, M. Zeng, J. Shen, X. Shi, X. Liu, F. Zhang, W. de Vries, Impacts of nitrogen
fertilizer type and application rate on soil acidification rate under a wheat-maize double
cropping system, Journal of environmental management 270 (2020) 110888.
[29] T. Zhang, Y. Hou, T. Meng, Y. Ma, M. Tan, F. Zhang, O. Oenema, Replacing synthetic
fertilizer by manure requires adjusted technology and incentives: A farm survey across
china, Resources, Conservation and Recycling 168 (2021) 105301.
[30] L. T. Habtemariam, C. P. Mgeni, K. D. Mutabazi, S. Sieber, The farm income and food
security implications of adopting fertilizer micro-dosing and tied-ridge technologies under
semi-arid environments in central tanzania, Journal of Arid Environments 166 (2019)
60–67.
[31] N. Adnan, S. M. Nordin, I. Rahman, A. Noor, Adoption of green fertilizer technology
among paddy farmers: A possible solution for malaysian food security, Land use policy 63
(2017) 38–52.
[32] M. Toledo, Manejo de suelos ácidos de las zonas altas de honduras: conceptos y métodos,
2016.
[33] A. Morante, M. del Pilar Villamil, H. Florez, Framework for supporting the creation of
marketing strategies, International Information Institute (Tokyo). Information 20 (2017)
7371–7378.
[34] H. Florez, S. Singh, Online dashboard and data analysis approach for assessing covid-19
case and death data, F1000Research 9 (2020).
[35] C. Balsa, C. V. Rodrigues, I. Lopes, J. Rufino, Using analog ensembles with alternative
metrics for hindcasting with multistations, ParadigmPlus 1 (2020) 1–17.
[36] J. Westerlund, Testing for error correction in panel data, Oxford Bulletin of Economics
and statistics 69 (2007) 709–748.
[37] E.-I. Dumitrescu, C. Hurlin, Testing for granger non-causality in heterogeneous panels,
Economic modelling 29 (2012) 1450–1460.
[38] J. Huang, A. E. Hartemink, C. J. Kucharik, Soil-dependent responses of us crop yields to
climate variability and depth to groundwater, Agricultural Systems 190 (2021) 103085.
[39] H. Florez, M. E. Sánchez, J. Villalobos, Embracing imperfection in enterprise architecture
models., in: CEUR Workshop Proceedings, Citeseer, 2013, pp. 8–17.
[40] P. Gómez, M. E. Sánchez, H. Florez, J. Villalobos, An approach to the co-creation of models
and metamodels in enterprise architecture projects., Journal of Object Technology 13
(2014) 2–1.
[41] F. M. Mugizi, T. Matsumoto, Population pressure and soil quality in sub-saharan africa:
Panel evidence from kenya, Land Use Policy 94 (2020) 104499.
[42] M. B. Fontana, L. E. Novelli, M. A. Sterren, W. G. Uhrich, S. M. Benintende, P. A. Barbagelata,
Long-term fertilizer application and cover crops improve soil quality and soybean yield in
the northeastern pampas region of argentina, Geoderma 385 (2021) 114902.