Greetings all,
I am writing a paper on the Dynamic Analysis of Profitability Determinants in Germany's for the Top 50 Banks(by total assets) for the period 2010-2022, with a SGMM Dynamic Panel Data Model Approach" taking into account time-invariant unobserved heterogeneity. I am aware the model should be T<<N, but thats how i should operate. My dependent variable is Return On Average Assets (roaa), and the possible explanatory variables are: cost to income (cti), equity/ total assets (eqas), loan loss reserves as deposit ratio (lodep), the ratio of loan loss reserves to gross loans (ll_gl), impaired loans/gross loans (implgl), total assets (ta), total equity (te), size= proxy for total assets, net interest margin (nimarg),Total Assets ratio (loanta) a measure of the proportion of a bank’s assets that are funded by loans, recurring earning power (repower), z score ( measure for bank’s financial soundness Z-score is measured as Z=(ROA+(EQ/TA) )/σROA), gdp growth, unemployment, inflation.
I could bring to the model more appropriate variables but for now i want to engage with SOME of these. The problem i have is i still find it very difficult to distinguish between the included and excluded, and the exogenous and the endogenous. The command I am trying to use is:
Xtabond2 y X's, gmm() iv() two robust (twostep)
Is there any way I could distinguish which variables to include to to the model, which to include to gmm and to th iv. I was considering the gdpgr and unemployment to be the exogenous, te the endogenous. But many problems appeared as soon as i was inserting the command to stata ( the coefficient of the lagged variable wasn't between 0.73 - 1.045 and the AR(1) test and the number of instruments.
Any help would be appreciated.
I am writing a paper on the Dynamic Analysis of Profitability Determinants in Germany's for the Top 50 Banks(by total assets) for the period 2010-2022, with a SGMM Dynamic Panel Data Model Approach" taking into account time-invariant unobserved heterogeneity. I am aware the model should be T<<N, but thats how i should operate. My dependent variable is Return On Average Assets (roaa), and the possible explanatory variables are: cost to income (cti), equity/ total assets (eqas), loan loss reserves as deposit ratio (lodep), the ratio of loan loss reserves to gross loans (ll_gl), impaired loans/gross loans (implgl), total assets (ta), total equity (te), size= proxy for total assets, net interest margin (nimarg),Total Assets ratio (loanta) a measure of the proportion of a bank’s assets that are funded by loans, recurring earning power (repower), z score ( measure for bank’s financial soundness Z-score is measured as Z=(ROA+(EQ/TA) )/σROA), gdp growth, unemployment, inflation.
I could bring to the model more appropriate variables but for now i want to engage with SOME of these. The problem i have is i still find it very difficult to distinguish between the included and excluded, and the exogenous and the endogenous. The command I am trying to use is:
Xtabond2 y X's, gmm() iv() two robust (twostep)
Is there any way I could distinguish which variables to include to to the model, which to include to gmm and to th iv. I was considering the gdpgr and unemployment to be the exogenous, te the endogenous. But many problems appeared as soon as i was inserting the command to stata ( the coefficient of the lagged variable wasn't between 0.73 - 1.045 and the AR(1) test and the number of instruments.
Any help would be appreciated.
