Thoughts on improving governance in developing countries

The re­cent EA pod­cast on char­ter cities made me think about this ques­tion—these are just my very rudi­men­tary thoughts on the sub­ject.

The prob­lems that gov­ern­ments in de­vel­op­ing coun­tries face are quite fun­da­men­tal. I’ll use Tamil Nadu (a state in In­dia) as an ex­am­ple, as that’s where my par­ents are from. Cor­rup­tion is ram­pant at all lev­els of gov­ern­ment, even for village lead­ers elected demo­crat­i­cally. The gov­ern­ment of­ten ig­nores prob­lems un­til they be­come too dire, such as ne­glect­ing the man­age­ment of wa­ter­sheds un­til there was a drought.

I also don’t think this is an iso­lated thing. I’m not able to find quan­ti­ta­tive data on this sub­ject, but from what I’ve read, these gen­eral pat­terns ap­pear to be com­mon through­out de­vel­op­ing coun­tries. There are pock­ets of great gov­er­nance from some mo­ti­vated in­di­vi­d­u­als but that ap­pears to be in the minor­ity. This sum­mer, I’ll be do­ing COVID co­or­di­na­tion work with a coun­try in South Amer­ica and an­other in Africa, so I’ll get a more first-hand view.

There ap­pear to be three sources of bad gov­er­nance:

  • Apa­thy (i.e. not do­ing a job one knows one should do)

  • Mal­ice (i.e. steal­ing money, cor­rup­tion)

  • Ig­no­rance (i.e. not know­ing one should do the job)

Each of these sources would have differ­ent solu­tions.

Or one could look at it through a differ­ent lens:

  • A peo­ple prob­lem

    • Solu­tion: re­search into bet­ter re­cruit­ment strategies

  • A cul­ture/​in­cen­tives prob­lem (i.e. peo­ple get cor­rupted)

    • Solu­tion: mod­ify cul­ture/​in­cen­tives us­ing be­hav­ioral science

I’ve looked into some aca­demic liter­a­ture on this sub­ject, but I haven’t found pa­pers that try to iso­late the cause. There’s an EA prob­lem pro­file on im­prov­ing in­sti­tu­tional de­ci­sion-mak­ing, but it seems to fo­cus on pre­dic­tion cal­ibra­tion and ra­tio­nal­ity tech­niques. I’d ap­pre­ci­ate any thoughts or re­sources on the sub­ject.