Two weeks ago, I blogged about some productive impacts of cash transfer programs. For these effects, as well as the myriad other blog posts and papers on this topic out there, a key point is that the benefits of these transfers extend well beyond the actual individual recipient of the transfer.
I came across a new working paper written by researchers at Google and Microsoft with the title “on the near impossibility of measuring the returns to advertising”. They begin by noting the astounding statistic that annual US advertising revenue is $173 billion, or about $500 per American per year. That’s right, more than the GDP per capita of countries like Burundi, Madagascar and Eritrea is spent just on advertising!
This week we're introducing our new series that we decided to call 'Ask Guido.' Guido Imbens has kindly agreed to answer technical questions every so often and we are thrilled. For this first installment, Guido starts by answering a question about standard errors and the appropriate level of clustering in matching.
One question that often comes up in empirical work concerns the appropriate way to calculate standard errors, and in particular the correct level of clustering. Here is a specific version of the question that someone posed, slightly paraphrased:
We are delighted that Dave Evans has agreed to be a semi-regular contributor to this blog, agreeing to post about once a month. David is a Senior Economist in the Chief Economist's Office for the Africa region of the World Bank, and coordinates impact evaluation work across sectors in the Africa region.
- For Valentine’s Day: of course there is a literature on this – from PhD comics comes abstracts of real papers such as “Me Do Wu My Val: The Creation of Valentine’s Day in Accra, Ghana”; and “Influence of Valentine’s Day and Halloween on Birth Timing”.
- Cyrus Samii on too much concern about “representativeness” and “generalizability”.
This week I finally got around to learning how to make a graph which displays the means of different treatment groups for a range of outcomes, along with standard error bars to show whether there is a significant difference between groups. Here is an example:
Imagine you are a local policy maker who just read about a new effective social program in another city and you want to determine whether this program would work for your area.