CORRECTION OF HETEROSCADASTICITY USING BOX-COX TRANSFORMATION OF VARIABLE IN THE STUDY OF GOVERNMENT EXPENDITURE AND ECONOMIC DEVELOPMENT IN NIGERIA.
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Statistics, Nasarawa State University Keffi
DOI
Abstract
This research focused on the correction ofheteroscedaslicity using box-cox transformation of
'variables with study on government’s expenditures and economic development in Nigeria.
The objectives of this study i$ to find heteroscedasticity existence in the fitted econometrics
model of government expenditure and gross domestic product (GDP), to correct the violated -
homoscedasticity in the model and to determine the decline/increase ^in government
expenditure and contribution to gross domestic product (GDP). The Data was secondary
data extracted from CBN statistical bulletins from 1962 to 2019. Breusch pagan. Goldfield
Quandt and white test were used in R package to check the presence of heteroscedast
and box-cox transformation of variables to remove the presence of heteroscedasticity in me
data. The ordinary least square (OLS) before Box-Cox transformation gives an R-square of
0.7679 implies that, the economic development (GDP) is explained by Government
expenditures with 76.79% and found to be significant with p-value of 2.2e-16, that's all
tests show the presence of heteroscedasticity. The ordinaiy least square (OLS) after box-cox
transformation gives an R-square of 0.406 implies that, the economic development (GDP) is
explained by Government expenditures with 40.6% and found to be significant with p-value
of 9.149e-13 and the response variable indicates an increase in economic development by
every increase in Government expenditures. The entire Breusch pagan, goldfield Quandt and
white test showed the presence of heteroscedasticity. The Table: 4 show the results of cd the
tests after Box-Cox transformation have been applied, all the Brush pagan, Goldfield Guard:
and White test showed a nonsignificant result, proving to us that the problem, of varying
i variance have been removed/remedied and that may lead to correctness in the data and
accuracy in budget splitting policy. Normality testing were also made using Shapiro walk test
which show that only the transformed data proved to be normally distributed With a p-value
: = 0.0612. It can therefore be said that the more government is up to qpcurate data in
| allocating their budget, the higher the government is spending, the higher the level of
economic growth and the lower the government spending, the lower the level of economic
! growth in Nigeria, therefore, it recommends that government should pus more effort in
• having accurate data in terms of budget allocations.
Description
Keywords
Citation
Garba, Y.Y. (2023) CORRECTION OF HETEROSCADASTICITY USING BOX-COX TRANSFORMATION OF VARIABLE IN THE STUDY OF GOVERNMENT EXPENDITURE AND ECONOMIC DEVELOPMENT IN NIGERIA.
