Even at low prices (and in poor places) the demand curve slopes downward

My UM econ classmate Austin Davis sent me a link to this great piece by Michael Kremer and Rachel Glennerster in the Boston Review, which summarizes what we’ve learned from RCTs in developing countries and how the results generalize to human behavior around the world. The entire thing is worth reading – they capture, better than I probably could myself, the reasons why I’m working in development economics and the sorts of issues I study.

The best part, though, is this plot of the (inverse) demand curves for a bunch of different products across various settings:

Both uptake and utilization of a variety of goods declines monotonically with their price

I like it for two reasons. First, as the authors point out, it’s fairly common to argue that “people won’t use [whichever good] if it’s free”. All the evidence I’m aware of indicates the opposite: not only the rate at which people take goods, but also the extent to which they use them, is drastically higher if the price is zero.

Second, I’ve occasionally been asked (including once by my roommate who is a fellow economist) whether we “really know” the demand curve for anything. The answer is yes: field experiments that randomly vary prices can identify actual demand curves, eliminating the unavoidable uncertainty that comes from estimating them observationally. From there we can figure out the optimal price to set in terms of cost-effectiveness (since we can use user fees to fund the provision of goods to more people) and, if we’re willing to make some strong assumptions, do social welfare analysis. Beautiful.


About Jason Kerwin

This entry was posted in Uncategorized. Bookmark the permalink.

3 Responses to Even at low prices (and in poor places) the demand curve slopes downward

  1. Austin Davis says:

    First, let me give credit where credit is due; the article was actually sent to me by my roommate and fellow UM econ student Ed Fox. Maybe Ed will weigh in here and correct any misrepresentation of his views but he approaches RCT with a good deal of skepticism. He wonders whether the results from RCTs are broadly applicable and, Ed points out, details matter; the precise manner in which an experimental intervention is presented or implemented can have a large affect on its uptake / utilization. To take an example from the article, the authors mention that “[i]nforming families about the returns of education increased student enrolment [sic] in the Dominican Republic and Madagascar.” Who informed the families? How? What information was relayed? For these reasons, he questions whether it’s wise to base policy decisions on a small number of experiments.

    Let preface my own opinion by announcing my own ignorance; although I am very interested in development, I don’t have a great familiarity with the literature. Let me also say that the article is long and rich in ideas and my response will certainly be incomplete. I think there are limits to the generalizability of RCTs but that doesn’t mean the results of such studies apply only to the context in which they were conducted. By considering the context in which a certain intervention was attempted examining the mechanism that determined its success or failure, perhaps careful researchers can predict whether an experimental result has implications beyond the place and time of the experiment itself. And, of course, the same intervention can be tested in different contexts.

    Another idea that caught my attention: the article gave a number of examples in support of their point that “people are not good at making even small upfront investments in order to obtain a stead stream of modest future benefits.” Setting aside the debate about whether such behavior can be rationalized, a few poorly formed explanations came to my mind, including constraints, positive externalities, attitudes towards risk, lack of information, and imperfect foresight. The authors added something else:

    “Perhaps the tough decision, however, is not whether to take action, but when. If people place particular weight on costs and benefits today, investing tomorrow always looks preferable — then benefits will still be there in the long run but the costs will be delayed. The catch is that when tomorrow comes it is again tempting to delay, and yet people fail to anticipate this.”

    I found this theory compelling perhaps because I notice myself doing the very same thing. As the authors note, maybe all it takes is a small upfront incentive to encourage people to make the investment today instead of postponing until tomorrow.

    Finally, I responded much differently to those inverse demand curves than you did. While I appreciate the vindication of basic micro theory, I had some technical objections. The “de-worming in Kenya” and “soap in India” series were constructed with a grand total of two data points each. “Chlorine in Kenya” has three, “mosquito nets in clinics, Kenya” weighs in with four, and we get the highest resolution from “chlorine in Zambia,” which has five. It’s not clear how well the lines fit the data and the experiment that yielded the most information, “chlorine in Zambia,” shows two reversals, i.e. two cases where a higher price resulted in higher … higher what? I don’t know because there is no label on the y-axis. Come on.

    A few nits aside, great article. Though provoking. Much more to discuss.

  2. Jason Kerwin says:

    I neglected to post this earlier, and it came to mind again: many of your issues with the graph are well-taken, but it’s not quite right to say that the curves are based on a handful of datapoints. I’m not sure about the sources for all the data, but the bednet plot is the mean uptake plotted against the price – they had lots of villages at each price offered. I’m unsure how heterogeneous the prices were, I think the X axis really does have just a few observations. As I recall (it’s been a while) there’s a negative and significant relationship between price and uptake when you look at all the uptake data instead of just the conditional means for each X.

    It’s still possible to criticize this if you think there are weird little nonmonotonicities in uptake with respect to price. Ideally you’d vary the prices close to continuously. I know of some research that does that, usually by incorporating travel costs, but I’m unsure about the plot.

  3. Pingback: Heterophobia | MethodLogical

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s