Professional Documents
Culture Documents
4.1 Introduction
This chapter will deliberate and describe the results obtained from testing the variables. The
chapter analyses inferential as well as descriptive statistics. For this study, data was retrieved
from the World Bank, United Nations Departments of Economic and Social Affairs (UNDESA),
and the Central Bank of Nigeria. The data employs annual data series and spans a period of 39
years, from 1981-2020. The results that shall be interpreted and presented consist of: descriptive,
correlation matrix, unit root test, cointegration test, ARDL long run and short run,
Table 4.1
Dev.
Observations 40 40 40 40 40 40
Before regressing the model, it is necessary to examine the data features. Descriptive statistics
define the various characteristics of the data set in this study. The table is divided into three
sections: measures of central tendency, measures of dispersion, and measures of normality. The
mean, median, maximum, and minimum are the constituents of the central tendency measure.
The mean of a data set is also known as the average; it is calculated by adding the values of the
observations and dividing by the number of observations. The median is then determined by
obtaining the middle value after arranging the observations in ascending or descending order.
Furthermore, the highest value in the series per variable is referred to as the maximum, while the
The standard deviation is the best measure of dispersion because it provides a comprehensive
and accurate result. The standard deviation measures the distribution's deviation from the mean.
When the standard deviation is low, data tend to cluster around the mean, whereas when the
Skewness and kurtosis are classified as measures of normality and are commonly referred to as
shape statistics. Going further, skewness measures the level of symmetry in distribution.
Skewness is calculated as the mean minus the mode divided by the data set's standard deviation.
In the data mentioned above set, agricultural GDP (AGDP), emigration (EM), immigration (IM)
and remittances (REM) with values of 0.507403, 4.982129, 2.125078 and 0.476097 are
positively skewed. This means that the right tail of a graph is longer and has a higher proportion
of values than the mean. Contrarily, the agricultural labour force (ALF) and trade openness (TO)
has values of -0.752848 and -0.330826, respectively, indicating that they are negatively skewed,
the left tail of the graph is longer, and the majority of their values are lower than the mean.
The sharpness of the distribution is referred to as kurtosis. It depicts the distribution's peakedness
mesocratic. Any value (distribution) greater than three is referred to as leptokurtic, while any
value (distribution) less than three is referred to as platykurtic. ALF (2.238122), REM
(1.332404), AGDP(1.675889) and TO (2.254798) are platykurtic and have values less than three;
however, EM (28.90082) and IM (6.504654) have values greater than three and are leptokurtic
The Jarque-Bera test is a goodness-of-fit test that determines if a given data sample has the same
skewness and kurtosis as a normal distribution. It also has a probability value that indicates the
test's significance. Using 5% as the significance level, any value less than 0.05 indicates
significance, and thus the null hypothesis is rejected. Any probability value greater than 0.05 is
considered insignificant, and thus the null hypothesis is not rejected. A probability value of less
than 5% (0.05) indicates that the data set is not normally distributed. As a result, trade openness
(TO), Emigration (EM), immigration (IM) and are not normally distributed, with probability
Agricultural labour force (ALF) and remittances (REM) are normally distributed with the value
Pearson's correlation coefficient is used to measure and assess the magnitude of a relationship
between two or more variables. The closer the correlation coefficient value is to one, the stronger
the intensity and relationship between the variables, according to this correlation method. As a
result, the correlation matrix indicates the degree and orientation of the interaction between the
variables under consideration. "A positive correlation coefficient indicates a direct relationship
between the two variables, whereas a negative correlation coefficient indicates an inverse
relationship".
AGDP 1.000
EM -0.0116374 1.000
assesses the magnitude of the association. The coefficient of the correlation runs between 1 and -
1. This range is further subdivided into three groups. In absolute terms, a coefficient of 0
suggests that there is no correlation between the variables and a value of 1 implies a perfect -
correlation between the variables. A coefficient that ranges between 0 and 0.25 denotes a weak
correlation; one that runs between 0.25 and 0.75 reveals an intermediate correlation. One that
ranges between 0.75 and 1 implies a high correlation between the variables. A coefficient with a
Furthermore, the table reveals a weak negative relationship between agricultural GDP and
emigration, trade openness and emigration, remittances and agricultural GDP with values of -
between immigration and agricultural GDP, trade openness and agricultural GDP, and
agricultural labour force and agricultural GDP with the correlation coefficient value of 0.60,
0.60 and 0.46, respectively. In addition, there is a weak negative correlation between trade
openness and emigration, with correlation values of -0.13. An intermediate positive correlation
exists between remittances and immigration with a coefficient value of 0.36. Thus, from the
values above, as agricultural GDP increases, emigration reduces, immigration reduces, and
remittances fall.
This study will conduct an empirical analysis to meet the study's objectives and ensure that the
impact, relationship, and importance of the impact are fully established to make informed
judgments and policies. This section presents and interprets the results of the unit root tests,
A unit root test determines whether a variable in a time series is stationary or non-stationary. The
null hypothesis states that the variable has a unit root problem and is non-stationary. In contrast,
the alternative hypothesis states that the variable does not have a unit root and is stationary.
Given that most time series is not stationary, at least at varying levels, the use of these variables
leads to bias and thus spurious regression (a regression run with non-stationary variables), which
renders the results almost insignificant and eventually affects decision making, forecasting, and
prediction (Gujarati, 2003). Therefore, in this study, the Philips-Perron method of unit root test
will be used, and the appropriate judgment regarding the use of the variable in the model will be
made. It is sufficient to say that a variable is considered stationary if the Philips-Perron test
statistic value is less than any of the test critical values at various levels of significance.
Variables PP Result
Levels First
Difference
As can be seen in Table 4.3 above, the outcome of the Philips-Perron test states that remittances
(REM), emigration(EM), immigration (IM), agricultural GDP(AGDP) and trade openness (TO)
are not stationary at levels I (0). This indicates that none of the values was significant at 1%, 5%
and 10 % at levels; however, the probability of the t-statistic was insignificant. This shows that
we fail to reject the null hypothesis; therefore, there is a unit root problem.
Agricultural labour forces (ALF) is the only variable stationary at level, i.e. I (0). Remittances
(REM), emigration (EM), immigration (IM), agricultural GDP, and trade openness (TO), on the
other hand, are all stationary at the first difference, i.e. I (1). Because the stationarity levels are a
combination of I (0) s and I (1) s, the ARDL approach can be taken to estimate the model.
The Autoregressive Distributed Lag (ARDL) model incorporates lagged forms of the variables to
generate short-run and long-run dynamics. On the other hand, a cointegration test is required to
demonstrate that the variables have a long-term relationship. In this study, the ARDL F-bound
test is used to determine whether the independent and dependent variables in the model have a
1% 2.7 3.73
At all levels of significance, the F-statistic of 21.25595 is greater than the values of the lower and
upper bounds (i.e. 1 per cent, 5 per cent and 10 per cent). As a result, we conclude that the F-
statistic is significant at 1%, and thus the null hypothesis, which states that there is no long-run
relationship, is rejected. As a result, in this study, there is a long-run relationship between the
Since cointegration has been established, an error correction form of the model can be formed.
For the result to be reliable, the error correction coefficient (CointEq (-1)) should be negative and
significant at 1%. The results of the error correction form are shown in Table 4.6 below.
The short-run results of the independent variable coefficients used in the model are shown in
Table 4.6 above. Some of the variables in their unlagged forms do not completely match the a
priori expectations. Although remittances (REM) were expected to be positive, it is negative.
Furthermore, immigration (IM) is positive, per the presumption that it would be positive. The
agricultural labour force (ALF) is positive, which corresponds to the a priori expectation.
Furthermore, Trade openness (TO) is negative, which is inconsistent with the a priori expectation
of being positive. Furthermore, emigration (EM) is negative, consistent with the a priori
expectation. Furthermore, in the short run, LEM, LTO and LIM can be significant, while the
It can also be seen that some of the lagged forms of the variables do not align completely with
the a priori expectations. The second and third lags of trade openness are negative, which is not
in accordance with the a priori expectation. The lagged form of REM is negative, which goes
against the a priori expectation that it would be positive. The lags of ALF are negative, which is
not in accordance with the a priori expectation, while the unlagged form of ALF is positive,
which is in accordance with the a priori expectation. Furthermore, the lagged form of emigration,
which is negative, is in line with the a priori expectation. The First lag of AGDP is positive,
In the short run, it can be seen that emigration (EM) is negative, as a 1% increase in emigration
results in a 0.82% decrease in agricultural output. The relationship is in line with the a priori
expectation as a rise in emigration leads to brain drain. This further implies that labour supply in
the agricultural sector will fall, and the production of agricultural commodities will fall, leading
to a fall in agricultural GDP. However, emigration significantly impacts agricultural GDP in the
short run.
In the short run, it can be seen that immigration (IM) is positive, as a 1% increase in immigration
results in a 1% increase in agricultural output. The relationship is in line with the a priori
expectation as a rise in emigration leads to an increasing labour supply and capital into the
agricultural sector, production of agricultural commodities will increase, and this will lead to an
In the short run, it can be observed that remittances (REM) are negative, as a 1% increase in
remittances results in a 0.06% decrease in agricultural output. The relationship is in line with the
serve as capital received from households in rural areas. As remittances increase, people don't
invest the remittances into their agricultural businesses. Thus this reduces investment in the
agricultural sector causing agricultural GDP to fall. However, remittances significantly impact
The error correction form is identified in the short-run ARDL, and the speed with which the
variables change from deviations to steady-state equilibrium is tested. The projected coefficient
for error correction form (CointEq (-1) should be negative and statistically significant, and the
absolute value should be less than one. The error correction form has a coefficient of -0.433949,
We established that the dependent and explanatory variables had a long-run connection. As a
result, a long-run analysis was carried out, with the following results:
*= 1% Significance ** 5% Significance
The ARDL model below displays the estimated long-run coefficients (elasticities) shown in the
table below.
The agricultural labour force is significant, which implies that it positively affects agricultural
GDP. An increase in the agricultural labour force leads to an increase in the production of
agricultural commodities; when the commodities are sold to satisfy the demand of the
will lead to a fall in agricultural labour; this will further lead to a fall in agricultural output and
production. There is an inverse long-run relationship between emigration and agricultural GDP;
thus, the effect is statistically significant as the p-value of the coefficients is significant at 1 per
cent. Hence the impact is significant to the changes in agricultural GDP in Nigeria.
Immigration is significant, which implies that it positively affects agricultural GDP in Nigeria.
When immigration increases, the number of labour and capital rises because immigrants bring
skills, labour and capital into the economy. This resource helps in the development of the
agricultural sector. The immigrants invest their resources into the agricultural sector. This
facilitates the production of agricultural goods to meet demand. This further increases
agricultural GDP. There exist a positive relationship between IM and AGDP. This states that an
increase in immigration will lead to an increase in agricultural GDP. There is a direct long-run
relationship between immigration and agricultural GDP; thus, the effect is statistically significant
as the p-value of the coefficients is significant at 1 per cent. Hence, the impact is significant to
Remittances also have a significant value, which negatively affects agricultural GDP. When the
number of remittances received by Nigerian households increases, households will not use the
remittances received to purchase farm implements and buy arable land to support their
agricultural businesses. The resultant effect of this is that it will lead to a decrease in agricultural
output (AGDP) as investment in the agricultural sector decreases. There exist a positive
relationship between LREM and AGDP. This states that an increase in the number of remittances
will lead to an increase in agricultural GDP. There exists a direct long-run relationship between
remittances and agricultural GDP. However, it is statistically significant as the p-value of the
coefficients is significant at 1 per cent. Hence the impact is significant to the changes in
Furthermore, there exists a negative relationship between IM, TO and GDP. This thus states
increase in trade openness and immigration will lead to a decrease in AGDP in the long run.
However, the effect of TO (Trade openness) does not significantly impact agricultural GDP in
A 1 per cent increase in LOGALF leads to a 0.03 per cent increase in agricultural GDP in the
long run. A 1 per cent increase in LOGEM will lead to a 1.44 per cent increase in agricultural
GDP in the long run. A per cent increase in LOGIM will cause a 2.89 per cent decrease in
agricultural GDP in the long run. Furthermore, a percentage increase in LOGREM will lead to a
0.065 per cent decrease in agricultural GDP in the long run. Lastly, a per cent increase in
LOGTO will cause a 0.05 per cent decrease in agricultural GDP in the long run.
Serial correlation occurs when there is a positive correlation between the disturbance terms at
Multicollinearity occurs when an econometric model has a strong correlation between two or
more decision variables. (Hanck et al., 2019). Multicollinearity makes it difficult for explanatory
The normality test will be conducted to know if the distribution is normal, and this will be done
The null hypothesis (H0) of the autocorrelation test run states that there is no autocorrelation,
while the null hypothesis (H0) of the heteroscedasticity test states that there is no
heteroscedasticity and that the distribution is normal. According to the table above, none of the
tests rejects the null hypothesis. As a result, we conclude that the model lacks autocorrelation
and heteroscedasticity, has a normal distribution, and thus has relevant econometric properties.
C 18.36457 53.73029 NA
The centred VIF values must be less than 10 for multicollinearity to exist. If the value is greater
than or equal to 10, it indicates that the model contains multicollinearity. The values of the
centred VIF for all variables in table 4.9 are less than 10. As a result, the model does not exhibit
multicollinearity.
The results of our long-run analysis in the previous section will be discussed in this subsection in
The effect of immigration on Nigeria’s agricultural GDP will be analyzed in the short and long
run.
The null hypothesis states that immigration does not significantly impact Nigeria's agricultural
GDP. The long-run and short-run analysis results show that immigration has a significant
positive impact on agricultural GDP in Nigeria. Immigration is significant at 1%. Hence, the null
hypothesis is completely rejected, and the alternate hypothesis, which states that immigration has
As the number of immigrants increases, they bring in their skills, labour and capital. This
resource helps in the development of the agricultural sector. The immigrants invest their
resources (skills, labour and capital) into the agricultural sector. This facilitates the production of
agricultural goods to meet demand. This further increases agricultural GDP. These findings go in
accordance with the findings of Mansur and Mansur (1987) in that their findings showed that
Test of Hypothesis 2
The null hypothesis states that emigration does not significantly impact Nigeria's agricultural
GDP. This would be analyzed both in the short run and long run. The result from the long run
and short run analyses indicate that emigration significantly negatively impacts agricultural GDP
Emigration causes brain drain, and a loss of manpower, an increase in emigration will lead to a
fall in agricultural labour, which will further lead to a fall in agricultural output and production
The short-run analysis results indicate that emigration has a negative and significant impact on
agricultural GDP. Hence the null hypothesis is rejected, and the alternate hypothesis, which
states that emigration has a significant impact on Nigeria's agricultural GDP, is accepted.
Test of hypothesis 3
The effect of remittances on Nigeria’s agricultural GDP will be analyzed in the short and long
run.
The null hypothesis states that remittances do not have a significant impact on Nigeria's
agricultural GDP. The results from the long run and short run analysis show that remittances
have a significant negative impact on agricultural GDP in Nigeria. This finding is in accordance
with the findings of Raju, Chandan, and Naveen (2014). However, this is because migrant
remittances received by a household may not be used to invest in their agribusiness and purchase
farm implements to boost agricultural output. Thus the null hypothesis that remittances do not
have a significant impact on Nigeria's agricultural GDP should be rejected, and the alternate
hypothesis that remittances have a significant impact on Nigeria's agricultural GDP should be
accepted.
CHAPTER FIVE
This research examined migration and remittances' impact on Nigeria's agricultural GDP. This
chapter focuses on summarizing the findings of this research based on which conclusions are
5.1 Summary
Agricultural GDP was used as the dependent variable in this research to evaluate the impact of
migration and remittances on agricultural GDP in Nigeria. Agricultural GDP was expressed as a
function of immigration, emigration, agricultural labour force, remittances and trade openness in
this study which used yearly data from 1981- 2020. This study's model was built on the Lewis
model. The Lewis model illustrates the effect of migration on the agricultural sector, which is the
The Autoregressive Distributed Lag Model (ARDL) was used to examine the effect of migration
and remittances on agricultural GDP, and the results obtained using empirical estimate
I. The stationarity of the variables utilized in the study was determined using the Philips-Perron
unit root tests. Except for the agricultural labour force (ALF), the data showed that all variables
II. The F-bounds test was then performed, revealing a long-run relationship between migration,
remittances, and agricultural GDP in Nigeria, which validated the adoption of the ARDL model
approach.
III. Autoregressive Distributive Lag Model shows that in the long run, there is a positive
relationship between emigration and agricultural GDP, agricultural labour force and agricultural
GDP. There is a negative relationship between IM and AGDP, TO and AGDP, and REM and
GDP.
IV. Long-run analysis shows that emigration significantly negatively impacts agricultural GDP.
This indicates that as emigration increases, agricultural GDP decreases. This goes in accordance
with the a priori expectation. The analyses also showed that immigration has a significant
positive relationship with agricultural GDP. This goes in line with the a priori expectation of
being positive. It was also discovered that remittances significantly negatively impacted
agricultural GDP.
V. The ECM test revealed that the agricultural labour force had a positive and significant impact
on agricultural GDP in the short run, and immigration had a significant positive impact on
agricultural GDP in the short run. Furthermore, emigration significantly negatively impacted
agricultural GDP in the short run. Remittances negatively and significantly impacted agricultural
5.2 Conclusion
The purpose of this paper was to determine the effect of migration and remittances on Nigeria's
agricultural GDP between 1981-2020. The first null hypothesis, which stated that immigration
has no significant impact on agricultural GDP, was rejected, and the alternative hypothesis,
which states that immigration has a significant impact on agricultural GDP, was accepted.
Additionally, the second hypothesis, which states that emigration has no significant impact on
agricultural GDP, was rejected as the study found a significant effect on agricultural GDP.
In conclusion, migration is still a global phenomenon and a big bone of contention for many
a glance. This is because the impact of migration on agricultural GDP is multifaceted. For
example, the study found that international emigration causes brain drain and a loss of
manpower, which negatively impact agricultural growth and GDP. On the other hand,
remittances from Nigerians in Diaspora were found to account for a reasonable amount of our
GDP.
Furthermore, the study mentioned earlier showed that immigration contributed positively to
industrialization and increasing national income through tourism. The question today is what the
net effect of migration is and how the government can optimize it to benefit agricultural GDP.
The government must pay attention to migration (immigration, emigration, and rural-urban
migration) and remittances received to improve agriculture sector performance and increase
agricultural GDP.
5.3 Recommendations
The following recommendations have been made based on the findings of the analysis conducted
in this study to increase agricultural GDP in Nigeria, consequently promoting the growth and
I. The government should provide technical assistance and capacity development to support
II. The government should intensify its effort in developing the agricultural sector to prevent
the brain drain of labour in the agricultural sector. The government should allocate for the
development of the agricultural sector. The government can do this by giving farmers
access to credit facilities (agro-based loans), making land available to farmers and
III. Sustainable agricultural practices should be adopted to limit the impact of climate
IV. Rural residents should be educated on ways to use their remittances for investment in
The study's major limitation was data availability, mainly due to poor data collection and
documentation in Nigeria. There is limited literature on the impact of migration and agriculture
in Nigeria.
This research concentrated on some aspect of migration because of the difficulty in obtaining
data, as seen in the literature review in chapter 2, migration can be analyzed from different
perspective. Therefore, using other forms of migration (internal migration such as rural-urban
migration urban-urban-urban, rural- urban etc.) can be used for further studies.