As we approach Christmas, it is high season for charities who knock on doors and ambush pedestrians to ask for a donation. And while charitable giving is laudable (I do it as well, but not spontaneously, so stop sending me letters) I wonder how effective it is?
In particular, when it comes to common social problems like alleviating poverty, fighting hunger or providing shelter to homeless people, etc. there are fundamentally two ways to go about this. The first one is to redistribute income through taxes. Increase taxes to pay for social safety nets. The second way to do it is to leave taxes low and encourage people to donate some of their income to beneficial causes of their choice.
Especially in the Anglo-Saxon world where charitable giving has always been more prominent than in continental Europe, the answer of most people to that question is clear: Keep my taxes low and I will give my money to causes that I support. Enter Johanna Mollerstrom from the hotbed of libertarianism that is George Mason University and her colleagues from the University of Chicago and UC Berkeley. They did a little experiment to test how much money is given to charity in different settings. They grouped participants either into small groups of 4 people or into large groups of 200 people. The trick is that in a small group of 4, you tend to know each group member individually, while in a group of 200, individual group members tend to be more anonymous.
Once put into a small or large group, participants were asked to share a part of their wealth with other group members. In the first setup, called ‘giving to many’, participants were told that their donation would be split evenly amongst the other group members. In the second setup, called ‘giving to one of many’, participants were told their donation would be given to one person in the group chosen at random who deserves is particularly deserving. In the third setup, called ‘giving to one partner’, participants were told their donation would go to one specific person in the group that was known to the donor in advance.
The chart below shows the share of people who decided to donate a fixed amount of money. If a person was asked to donate to the entire group, the share of donations was smaller than if a person was asked to donate to a random individual or a pre-identified individual. Also, in large groups, the share of donations was always smaller than in smaller groups. We humans simply are reluctant to give money to strangers or pour our money into a black hole. Much better to help someone we know or where we know it is helping a specific individual. And in larger groups that are naturally more anonymous, this effect is more pronounced because we have a harder time identifying with individual group members in these larger groups.
But note also that there is a fourth setup. This one, called ‘vote on redistribution’, asked all participants in a group to vote in favour or against a redistribution scheme. If more than 50% of all group members voted in favour of redistribution a fixed amount of wealth would be taken from the members of the group and given in equal measure to other poorer people. Voting in favour of this proposal would mean that not only you but all other people in the group would be worse off in equal measure. And lo and behold, independent of group size about 50% of all participants were in favour of such redistribution schemes. Redistribution of income is generally more acceptable to people and raises more money to distribute to poorer people than relying on individual charity.
Obviously, the experiment did not account for government slack and inefficiency and one frequent argument against government redistribution is that the government wastes a lot of the money it collects through taxes while giving to charities is more efficient. That is likely to be the case, but note that the difference between the redistribution setup and the giving to many setup is large. So the government has to be very wasteful before this gap closes.
Donations in different setups and group sizes
Source: Mollerstrom et al. (2021).
Preface to my comment:
I am not against taxation as a means of income redistribution in principal. My comment below is about how to balance large-scale central intervention vs more localized efforts.
Comment:
Isn't another interpretation of the result that "Giving to One Partner" means some elements of charity should be organized by local communities for causes they regard as worthy. The "Giving to One Partner" result is very similar to the "Vote Redistribution" result and would suggest that the former is as effective a method as the latter and critically would eliminate wasteful governmental costs (but would clearly incur duplication of costs across many small communities). However, it has the added benefit of people seeing the results of their charity at a local level and maybe, over time, would outperform Vote Redistribution - or perhaps not.....
Too often I think governments try to solve micro problems with a macro solution - but that's just a feeling - it's not data.
Is there any data to support the notion that local community initiatives are as efficient or more efficient (cost, result, and satisfaction levels) than large government at solving some local problems, possibly because people can see the effects of their charity?
Ultimately we need both governments and communities, but what is the optimal balance between the two, and what are the implications for (a) devolution and (b) globalization vs localization?
Is it your view that giving money to the poor alleviates poverty so that the true measure of success is how much money gets redistributed? I don’t think the facts are on your side except in the very short term.