Let’s assume that we want to estimate the effect of tech inn…
Let’s assume that we want to estimate the effect of tech innovation on individual wealth in a nation. We can measure individual wealth in a nation with GDP per capita. We can measure tech innovation with the number of patents or the number of science journal publications. In addition, we want to measure the Covid-19 pandemic effect on individual wealth as well. Therefore, we construct the dataset as follows: Patent: the number of patents in a country GDPcapita : GDP per capita = GDP/population GDPgrowth: GDP growth rate SCJournal: the number of scientific journal publications HighTechExport: the share of high-tech export in GDP Unemploy: Unemployment rate Covid: 1 data comes from 2020 (during the pandemic); 0 data comes from before the COVID-19 pandemic period. Then, we conducted correlation test among variables as follows: we want to put both the patent variable and SCJournal variable as independent variables into a linear regression. However, the high correlation between patent and SCJournal variables (i.e., 0.87) can cause _______________. In this case, we cannot trust the p-value for the coefficients of independent variables in the sample linear regression; therefore, we usually drop one of the correlated independent variables in the linear regression model.
Read Details