Page 71 - Azerbaijan State University of Economics

Swaty Sharma, Munish Gupta: Does the Rise of Emerging Technologies Transform Digital
Entrepreneurial Activity? Evidence from OECD Nations

5.1 Model Analysis
The study uses two models to examine the impact of different variables:

Table 2: Slope Heterogeneity Test Results
Model Variables Stat adj Stat Sig

Model -1 ICT, Internet, GDP, EST-BUS, financial Risk, 3.001 5.009 0.000
TEN
Model -2 ICT, RR&D, R&D, EST-BUS, financial Risk, 2.998 4.001 0.000
TEN
Author’s Creation

Table-3: Estimated Intercepts of Key Variables at Level 0 and Level 1
Variables Intercept at level 0 Intercept at level 1
ICT -1.786044 -3.894527***
Internet -2.142222 -3.597857***
EST_bus -2.598764 -3.897242***
TEN -2.009009 -4.211489***
GDP -1.298765 -2.884287***
RR&D -1.276547 -2.898473***
Financial Risk -1.654 -3.789***
R&D -1.049555 -3.890233***
Author’s Creation

After the descriptive analysis, the research proceeds to test the slope heterogeneity in
the two models. The outcomes in Table 2 indicate considerable differences in the slope
models, and heterogeneity is present (p < 0.05). In particular, the statistics [Delta =
3.001, Deltaadj = 5.009] of Model 1 and [Delta = 2.998, Deltaadj = 4.001] of Model
2 using the test suggested by Pesaran & Yamagata (2008) indicate a statistically
significant variance among countries. Although no individual slope t-tests were
carried out, these global test statistics give very solid evidence that the association
between dependent and independent variables is not the same across OECD countries.
This confirms the ability to use fixed-effects modeling, and it also suggests including
heterogeneity in the econometric specification.

The research proceeded to check stationarity across observed variables during its
subsequent stage. variables. Stationarity means that statistical properties like the mean
and variance do not experience changes with time. The stationarity testing is a crucial
part of the panel data models since the non-stationary variables may induce spurious
regression. The estimated relationships can be statistically significant when it is not
really related in any meaningful sense when the data are non-stationary. When
variables are identified as non-stationary at the level, they are usually transformed
(most often by taking first differences or logarithms) to stabilise the variance and
produce valid results in the subsequent econometric estimation step (Gujarati &

71