I guess that it could be reverse causality (countries with a bad performance receive more aid). Anyway, check the literature in order to see how it addresses this potential endogeneity problem.

Furthermore, William Easterly has written on this topic extensively. See, e.g., Article Can Foreign aid buy Growth?

Thanks for the help José-Ignacio Antón. It makes perfect sense.

Apparently it is an endogeneity problem between the explanatory and dependentente variables. You will need to estimate with instrumental variables.

In a "perfect" (neoclassical) world one would expect the marginal investment to have a zero return, and when a country has a local savings deficit foreign savings can supplement local savings for productive investment. In the world in which we live, not a perfect world, foreign investments may compete with local investments for the projects that yield the best returns. The real world is messy, and microeconomics screw up the macro-economic perfect world scenarios.

As per my understanding, when you combining two variables, it will be correlated to included variables, i.e., multicollinearity, due to which the sign of the estimated variable becomes negative. Furthermore, signs could be negative or positive, based on the data which you are dealing. For example, as you know that countries are different (heterogeneity), which implicitly implies that, the determinants varies across countries.

Thank you for your clarification, Mahmood.

"Dead Aid" by Moya describes much of the Aid donated through large multi-country agencies through national governments in Sub Sahara Africa as not just a waste of resources but having a debilitating impact on the behaviour of government decision making. Poorer decisions result, people in charge are left to benefit themselves or their patrons that lead to projects which are wasteful at best and distorting and undermining of local initiative and skills development. She argues that withdrawing this aid will lead to greater accountability and in the long run facilitate higher levels of growth and development than without the outside funding. Those of us with a Keynesian or Harrod-Domar perspective see external funding as a necessary element in providing more capital to capital poor countries. But recognising the risk and planning for clearer transparency and monitoring the impact of aid is a necessary component in the process of making aid more effective.

Carolina Corrêa what do you think as a tool lag independent regressor values in turn for each specification or Vathanak Chet would have to introduce a new variable as IV?

Alexander Vilgelm, new instrumental variables bring important information, but it is not always easy to find good instruments.