Page 8 - Azerbaijan State University of Economics

THE JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.77, # 1, 2020, pp. 4-18

Variables are, respectively, the number of students enrolled in secondary education
(SEC), net (US dollar) government expenditure per student (SFIN), and real GDP
per capita (PGDP). Primary data on variables were obtained from the World
Bank[https://data.worldbank.org] and UNESCO [https://data.uis.unesco.org]. The
secondary data are the statistical institutions of the countries and the ministries of
education. Based on the defined variables, the models of the research are expressed
as follows using the method:

= + + + (1. Model) (1)

2

1

= + + + (2. Model) (2)

1
2

= + + + (3. Model) (3)

2

1

The variation of time data variables used in econometric studies as a unit may result
in nonlinearity. One of the variables used in this research is quantity, and two of
them are monetary expression. In order to ensure linearity, the data in models 1, 2,
and 3 were transformed as natural logarithms (Ln). Information about the log-based
data set is given in Table 2.

Table 2: Descriptive Statistics for the DataSet
lnSEC lnSFIN lnPGDP
Mean 14.23 8.48 10.01
Median 14.02 8.91 10.45
Maximum 17.11 10.19 11.42
Minimum 10.38 5.23 6.85
Std. Dev. 1.51 1.14 0.97
Skewness -0.21 -0.72 -0.95
Kurtosis 2.81 2.50 3.39
Jarqua-Bera 4.62 53.08 84.05
Probability 0.01 0.00 0.00

Due to the time data of the panel model, there may be regression errors in the findings.
Stationarity of data is provided to eliminate these regression errors. Unit root tests are
applied to ensure stationarity. All PP, ADF, LLC and IPS unit root tests [For detailed
information on unit root tests, see: Baltagi, B. H. (2013). Econometric analysis of
panel data. Chichester: John Wiley and Sons] were used to increase the validity of the
research. Stationarity in research models is expressed by the symbol "Δ".

When stationarity is applied to variables, their relationships are affected. In other
words, the relationship that really exists can disappear with stasis. Cointegration is
carried out to determine the existing relationship after stationary. Although
cointegrated variables have an unstable relationship in the short term, it can be stated
that they have real relations in the long term (Güvenek and Alptekin, 2010: 180). In
8