Comments and Suggestions

Post Reply
Larue_Bruno
Posts: 8
Joined: Tue Jun 05, 2018 8:28 pm
Contact:

Comments and Suggestions

Post by Larue_Bruno » Thu May 30, 2019 10:17 pm

Comments on «Can Urbanization Improve Household Welfare? Evidence from Ethiopia » authored by Kibrom A. Abay, Tsega G. Mezgebo, Meron Endale and Helina Tilahun.

The main hypothesis is that greater luminosity at night indicates economic activity which benefit households living close by. Rural areas tend to be dark and the new light generally appears in urban areas. From this, it is inferred that urbanization is taking place. The analysis relies on daily information about luminosity and household surveys (from the Living Standard Measurement Study-Integrated Surveys on Agriculture, LSMS-ISA for Ethiopia) from which changes in consumption can be measured and labor outcomes observed. Urbanization in this paper is about growth of small rural towns. Nearby farmers can benefit from new employment opportunities. It is not about farmers leaving their farms to work several hundred kilometers away in a large city.

Household consumption (real per adult equivalent) is regressed on luminosity to infer a causal effect. Time-invariant household characteristics are accounted for by household fixed effects. Time-varying household characteristics such as household size are included as controls. Time fixed effects capture the effects of time-varying variables that do not vary across households like macroeconomic indicators. Enumeration area fixed effects control for all time invariant area-specific variables.

Some of the older papers cited argue that urbanization in Africa is accompanied by poverty and does not spur economic growth. The authors find that luminosity induces increases in consumption and improves labor outcomes. As such, the study reverses earlier results and support the evidence reported more recently. The paper also investigates the relationship between night light and welfare inequality.

Urbanization has been a most studied phenomenon for quite some time. The use of satellite data to study urbanization is recent and getting increasingly popular.

Main comments:
The usual interpretation of inequality is that it is bad. However, this needs not be the case. If all households are poor before there was light and that some households are “a bit better off” and other households “substantially better off” once exposed to light, then we have a Pareto improvement AND more inequality. Still, no household would want to go back to darkness in this example. The concern about the distributional effects of the treatment is about the number of losers/how many households experience drops in consumption, the magnitude of the losses and who loses (the poorest ones or the ones that are well off?). The descriptive statistics about consumption in 2011 and 2013 suggest that consumption is going down. If the poorest households exposed to more light suffered a larger drop in consumption than similarly poor households exposed to the same amount of light over time, then light has indeed bad distributional effects. Is this the case? Quantile regressions indicate weakly positive light effects across all quantiles. The poorest households benefit less than households with the highest consumption, though the difference is perhaps not statistically significant, and the households in middle quantiles are not statistically impacted. The coefficients for middle quantiles have large confidence intervals spanning positive and negative values. So let there be light!

The area covered by light is said to measure 10 km. Is the 10 km referring to the diameter or the radius of the area covered or is the area 10 km2? How big in squared km are the rural towns and how big are they in terms of population?

Since consumption is used as a welfare proxy, are goods produced by farm households factored in? There is no mention about this in the paper.

Most of the households surveyed live in rural areas and there does not seem to be evidence of urbanization given that the urban dummy has a mean of 0.12 for 2011 and 2013. Urbanisation has a different meaning from the one typically used in the literature which is about households abandoning their farm to move to a city. In this case, the benefits from urbanization are for small town residents and farmers living within an enumeration area spanning a small town.

You report up to 7 model specifications (tables 4 and 5). While this is useful to gauge the robustness of the light effect on welfare, at some point you should indicate which is your preferred specification and discuss the effect of other variables on welfare from this preferred specification.

Results in Tables 6 and 7 indicate that light increases labor supply and wages, which is intuitive. However, why is access to formal credit impacting negatively the labor supply and wages (and consumption in other tables)?

The coefficients for light about food prices in table 9 are small. They measure the %change in food prices from a one-unit change in light. It would be useful to use the average change in light in small towns to get a better grip on food inflation. The descriptive statistics about daily luminosity suggest that the change in luminosity in small towns between 2011 and 2013 would be a fraction of a unit. If I am right, increases in food prices would be very small and hence not worth worrying about.

Minor comments:

p. 6: The term EA has not been defined.

p.8: (nominal) consumption vs real expenditure. It would be best to use the term real from the start. You use it later on.

p.8: Consumption expenditure. It should be plural: consumption expenditures.

p.9: footnote 7 should be clearer about the specific census-based indicators that the authors are alluding to.

p.10: The main shortcomings from using satellite images should be listed. The references could then be used if someone needs more information.

In your regression tables, you have yes, no and – to inform the reader about the presence or absence of fixed effects. What is the difference between no and -?

See my copy of your paper for additional typos.
Attachments
PMMA 20210_K_A_Abay.pdf
(418.57 KiB) Downloaded 13 times

Post Reply

Return to “PMMA-20210 - Can Urbanization Improve Household Welfare and Provide Inclusive Opportunities? The Case of Urban Expansion in Ethiopia”

Who is online

Users browsing this forum: No registered users and 1 guest