Instrumental Variables

[darina]

Hi, there!

So we are trying to establish whether (and what) is the link between disaster relief and development. When carrying out economics experiments, we usually resolve such problems by introducing Instrumental Variable regressions (IV).

What are your thought on using an indicator of sanitation (such as total % of population using improved sanitation facilities) as an instrument for disaster relief? Any ideas for a different variable which can link sanitation, disaster relief and development?

Thanks a lot!

[jerome]

Hey there,

Since sanitation and health are closely correlated, maybe a health indicator would be more accurate. For example, infant mortality rate and % of diarrhoeal diseases.

School attendance is also influenced by sanitation as students from areas without sanitation system is more likely to get ill and unable to attend schools so maybe literacy rate?

The problem with these indicators however is that there would be other confounding factors that needed to be filtered out.

J

[hui]

How about time? How long does it take for sanitation improvements to begin? How soon does it take for economic activity to take place? What type of activity is this? How long is it sustained for?

[yin]

Could we look at indicators of development that are affected by sanitation but in the mid-term?

For example, looking at indicators of improvement in health within (lets say...) 2 years after a disaster. Then could we not gauge how well the sanitation developed.

i.e. a sanitation programme that is good at "bridging the gap" would show signs of development through sanitation (indicated by health indexes) within a shorter amount of time than one that is bad at it (and hinders development).