The methodological issues of this study was discussed under the following sub-headings: Research design, nature and sources of data, and model specification.
3.1 Research Design
The purpose of a research design is to verify that the information gathered allows the researcher to effectively address the research problem as logically and unambiguously as possible. Acquiring information relevant to the research problem in social science research, typically includes determining the type of facts required to test a theory, to evaluate a programme, or accurately characterize and evaluate context related to an observable phenomenon. This study adopted the quantitative method and descriptive research design using already existing data to provide empirical answers to the research problems. Descriptive research designs help provide answers to the questions about who, what, when, where and how connected with a research problem. A descriptive research design cannot conclusively establish answers to the why problems associated with a research. It is used to generate information on the current state of the phenomenon and to explain what exists with respect to variables.
3.2. Nature and Sources of Data
The data used in this study were gathered from secondary sources. These data were time series data collected using the desk survey approach from Central Bank of Nigeria (CBN), the Debt Management Office (DMO), World Bank and IMF statistical database. The macroeconomic variables on which data were collected included the Real Gross Domestic Product (RGDP), External Debt Stock (EDS), Domestic Debt Stock (DDS), Debt Service Payments (DSP), Foreign Reserve Position (FRP) all in millions of United States Dollars, effective Interest Rate (INTR), Gross Fixed Capital Formation as a percentage of GDP (GFCF) and Foreign Direct Investment inflow as a percentage of GDP (FDI). All variables were taken on annual basis in nominal terms and in rates running from 1980-2018 making a total of 312 observations. Data on RGDP and INTR were sourced from the Central Bank of Nigeria, EDS, DSP FRP, GFCF and FDI were sourced from the World Development indicators while DDS was sourced from the Debt Management Office.
3.3: Econometric Specification.
To investigate the impact of government debt on economic growth in Nigeria, an open multivariate debt-growth model allowing for key control variables was specified following the lead of Gomez-Puig & Sosvilla-Rovero (2017) with slight modifications to suit the requirements of the current study. The model explored the linear relationship between economic growth represented by RGDP and disaggregated components of public debt indicators. This disaggregation was informed by the need to evaluate the individual effects of various indicators of public debt on the long and short-run economic growth of Nigeria. Such a rich environment can overcome variable omission bias, thus allowing for efficient estimates of the test statistics. The ARDL form of the regression equation estimated is specified in equation 1 as follows:
Where: RGDP = (Proxy for economic growth) Dependent variable.
EDS, DDS, DSP, FRP, INTR, GFCF and FDI = Independent variables of the model.
β0 = Constant. β1, β2, β3, β4, β5, β6 and β7 = Long-run coefficients to be estimated while 8 until 15 represent the short run coefficients of the respective variables in the model.
ECM = Error Correction Term which measures the speed of adjustment and t = time trend consisting of years from 1980 to 2018.
Note: Macroeconomic variables at levels tend to show geometric growth and required taking their logarithms to linearize their movement through time. The study therefore, transformed RGDP, EDS, DDS, DSP and FRP into their natural logarithm functional form specification to reflect the elasticity of the respective variables.
In accordance with economic theory, it is expected that β1, β2, β5 and β7 can either be positive or negative, that is > or < 0. β4 and β6 are expected to be positive, that is, > 0 and β3 negative, that is < 0 .
3.4: Data Estimation Technique
The study uses the Autoregressive Distributed Lag (ARDL) approach to co-integration proposed by Pesaran & Shin (1999) and Pesaran, Shin & Smith (2001) to empirically analyse the long and short run impact of public debt on economic growth in Nigeria. This method presents three significant advantages over the two alternatives commonly used in empirical literature: the single-equation procedure developed by Engle & Granger (1987) and the maximum likelihood method postulated by Johansen (1991, 1995) which is based on a system of equations. First, the ARDL bounds testing approach allows the analysis of long-term relationships between variables, regardless of whether they are stationary at levels, [I(0)], or first difference, [I(1)] or fractionally co-integrated. Second, the ARDL method allows for the simultaneous estimation of the short-run and long-run components, eliminating the problems associated with omitted variables and the presence of autocorrelation. Finally, the short and long-run parameters estimated using this approach are consistent in small samples such as the one used in this study.