Published on Development Impact

Weekly links October 4, 2024: why optimal taxes may look different in developing countries, teaching IV, is everything development now, and more…

This page in:
Young boxers at the White Collars Boxing Match 2019, taken by Mariajose Silva Vargas

·       The latest VoxDevList is now out, on Taxation and Development, edited by Anders Jensen, Anne Brockmeyer and Lucie Gadenne. “In the domain of effective taxation and enforcement, there are three main lessons. The first is that the existence of third-party reporting improves enforcement by creating paper trails on the activities of firms and citizens….The second insight is that the lower third-party coverage in low- and middle-income countries can lead optimal tax policies to look very different than they do in high-income countries. For example, once we account for enforcement constraints, it may be desirable for optimal tax policy to implement policies that distort firms’ and households’ economic choices relatively more if they lead to less evasion and therefore greater revenue collection. The third insight is that tax authorities must use evidence to balance between statutory reforms (e.g. changing the tax rate) on the one hand, and direct investments in enforcement, on the other hand.” The review also discusses evidence on administrative reforms and the functioning of the tax authority, as well as issues of equity and distributional considerations.

·       On VoxEU (not VoxDev!), Ravi Kanbur has the provocatively titled piece “The end of development economics” where he argues that the improvement in incomes in developing countries and homogenization of economics methodology means the distinction between development economics and economics in general no longer is necessary. Part of his argument is that topics like market failures, the role of markets versus state, and the role of culture and social conventions are now seen as key topics in developed countries too. So whether the label is needed or not, he is still arguing that the questions and issues remain very important.

·       If you are looking for a new example to teach instrumental variables with, how about this paper by Frimmel, Halla, and Winter-Ebmer just published in the Journal of Public Economics, which looks at the impact of parental divorce on children’s long-term outcomes (in Austria). The instrument for divorce is “idiosyncratic variation in the extent of gender balance in fathers’ workplaces. Fathers who encounter more women in their relevant age–occupation–group at work are more likely to divorce. This result is conditional on the overall proportion of female employees in a firm and on detailed industry affiliation”. As evidenced by lots of discussion on social media, this paper seems great for discussions about assumptions and treatment heterogeneity:

o   Relevance and Mechanism: the authors find that going from 0 to 100% of the other workers in a fathers’ age-occupation group being women increases the chance of divorce by 2 percentage points (relative to an overall average of 13.5% of kids experiencing their parents divorce by age 18). The F-stat is around 16. So this instrument is correlated with higher divorce rates. Twitter/Bluesky delighted in thinking of this as a “men are dogs” instrument, and the authors see their identification as coming from “Gender-balanced workplaces reduce the cost of extramarital search and allow married individuals to meet alternative mates, which increases the likelihood of divorce. Thus, we aim to identify the causal effect of divorce for the child whose father left the family because he met a new partner at work.”. But they do not show this directly, and one could also think of other channels like discussions with co-workers influencing happiness with marriage, and divorce occurring without it being through leaving the spouse for a co-worker.

o   What is the treatment effect being identified/who are the compliers? Good for discussion about LATEs, and whether the effect of divorce on kids whose parents divorce because of a father’s exposure to more women at work is likely to differ from the effects of divorce arising from other reasons.

o   Could there be defiers? E.g. could there be some men who only get divorced when lots of their co-workers are men, and not when lots are women? One can think of possible channels such as peer effects of having more male co-workers who are either divorcing or who appear happier in their marriages than the person being considered influencing how happy a worker is in their own marriage. Then this could lead to discussions of what happens if there are some defiers, but more compliers than defiers, as discussed in this old post.

o   Discussing the exclusion restriction: the authors put in lots of controls and fixed effects, including industry fixed effects and the age-specific share of women in the firm – so they then need to assume that the share of women in an age-occupation category within a firm is exogenous. There could be discussion about whether this seems plausible. But then more discussion on whether we think this share should only affect child outcomes through divorce – what happens if there is infidelity without divorce, or if exposure to more other fathers versus mothers in your age-occupation provides information and norms that affect how non-divorced fathers interact with their children.

  • IV estimation: forgot to add this when I posted - the authors don't use 2SLS, but use a logistic to estimate the first stage, and then this non-linear control function approach - so there could be discussion on when/whether to prefer this over linear IV as well.

o   External validity – this is from kids in Austria whose parents divorced between 1976 and 2005 – so there could be discussion about whether students think the same mechanisms might apply in their country and in today’s world.

·       For those getting deluged with requests from high school students, here is a nice page by astrophysicist Katie Mack on students, school projects, and scientists – particularly when this is coming from students being asked to contact an expert as part of a class assignment.


David McKenzie

Lead Economist, Development Research Group, World Bank

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000