The Dominican Republic’s Solidaridad conditional cash transfer program provides its monetary transfers to poor families in the form of a debit card that can only be used at a network of grocery stores affiliated with the program (it does this in part to ensure they spend the money on food). The typical monthly transfer is about $36, which is 17% of median monthly food expenditure. In order to make it attractive for stores in poor neighborhoods to be formal, get a card reader, and accept the program debit card, the authorities limited the number of stores that could offer this service depending on the number of beneficiaries in each district. This led to concerns that retailers were using their market power to keep prices higher than they could be.
A very nice working paper by Matias Busso and Sebastian Galiani asks what happens when you allow more competition? It is innovative in actually randomizing the amount of competition faced in different markets.
Key details
They work in 72 districts of the country, with an average of 630 consumers using the CCT debit card in each district and an average of 6 stores already accepting these cards in each district. The districts had other retail stores operating in them which weren’t in the program.
The treatment then consists of randomly allowing 0, 1, 2, or 3 more existing retailers to join the system. A total of 61 additional retailers joined, representing a 26% increase in the number of retailers offering the CCT service in treated markets.
Beneficiaries use the debit cards for only a very limited number of products, with 15 products - bread, rice, pasta, cooking oil, sugar, flour, powdered milk, onions, eggs, beans, cod, canned sardines, chicken, salami and chocolate – accounting for 85% of non-perishable food products and 60% of all food spent. The authors therefore focus on examining changes in prices and quality offered in these products. For those stores who do accept the debit cards, CCT transfers account for about half of total sales, so this is a significant source of revenue for the stores.
Results
This is a very well executed study that is the first I’ve seen to experimentally manipulate competition and measure its effects. My main thoughts concern the tricky problem of measurement, mostly with some ideas for what might be done in future such studies.
Measuring prices:
As most people who have shopped in developing countries have found out, often there is not a single price for goods offered by small stores. Indeed Busso and Galiani note that only 41% of retailers in their sample have prices posted, and only 44% say they never bargain with customers over prices. This makes it difficult to measure “the” price. The authors’ solution to this issue is to ask retailers for information on the price, variety and brand of the cheapest available option of a product. But even conditional on this, it is unclear how retailers that would charge different prices for the same variety and brand would respond – perhaps with the lowest price they charge to any customer. The authors then alternatively use a household survey of beneficiaries to measure total expenditure on a product and the number of units, and take the ratio as the price – but then this may be comparing different brands/varieties.
Consider two possibilities here:
The authors don’t observe what different types of clients pay for the same product at the same store, and can’t distinguish between these two stories. Either explicitly asking retailers about the prices charged to different clients, and/or surveying different types of clients would allow this to be measured better in future work – the general point is that measuring retail price data is hard.
Measurement period: the authors measure impacts 6 months after entry. It would be nice to have time series on how firms respond to see how quickly competition has effects, and whether the market has come to a new equilibrium by 6 months, or if it takes longer for firms to figure out their new optimum pricing strategies.
Firm effects: it would be interesting to measure what happens to the profits and sales of existing CCT network firms and of those of the new entrants to the network after this policy change. The authors look at the number of customers that retailers have on their best day, finding a negative but insignificant effect, and don’t have these other measures. It would also be interesting to know whether more competition leads to more efficiency in terms of sales moving from less productive to more productive firms.
This is an area where there is very little in the way of experimental evidence, so this paper is a great contribution to the literature. I look forward to seeing more work which can rigorously isolate the impacts of policies to influence competition.
A very nice working paper by Matias Busso and Sebastian Galiani asks what happens when you allow more competition? It is innovative in actually randomizing the amount of competition faced in different markets.
Key details
They work in 72 districts of the country, with an average of 630 consumers using the CCT debit card in each district and an average of 6 stores already accepting these cards in each district. The districts had other retail stores operating in them which weren’t in the program.
The treatment then consists of randomly allowing 0, 1, 2, or 3 more existing retailers to join the system. A total of 61 additional retailers joined, representing a 26% increase in the number of retailers offering the CCT service in treated markets.
Beneficiaries use the debit cards for only a very limited number of products, with 15 products - bread, rice, pasta, cooking oil, sugar, flour, powdered milk, onions, eggs, beans, cod, canned sardines, chicken, salami and chocolate – accounting for 85% of non-perishable food products and 60% of all food spent. The authors therefore focus on examining changes in prices and quality offered in these products. For those stores who do accept the debit cards, CCT transfers account for about half of total sales, so this is a significant source of revenue for the stores.
Results
- Having an extra grocery store in the district that is able to accept the debit cards causes a 5 percent decline in prices of these key products
- The authors examine several dimensions of quality including brand offered, store cleanliness, and customer satisfaction and find no significant change in the quality of products. Given the difficulty in defining quality, they do a nice job considering a range of measures here.
This is a very well executed study that is the first I’ve seen to experimentally manipulate competition and measure its effects. My main thoughts concern the tricky problem of measurement, mostly with some ideas for what might be done in future such studies.
Measuring prices:
As most people who have shopped in developing countries have found out, often there is not a single price for goods offered by small stores. Indeed Busso and Galiani note that only 41% of retailers in their sample have prices posted, and only 44% say they never bargain with customers over prices. This makes it difficult to measure “the” price. The authors’ solution to this issue is to ask retailers for information on the price, variety and brand of the cheapest available option of a product. But even conditional on this, it is unclear how retailers that would charge different prices for the same variety and brand would respond – perhaps with the lowest price they charge to any customer. The authors then alternatively use a household survey of beneficiaries to measure total expenditure on a product and the number of units, and take the ratio as the price – but then this may be comparing different brands/varieties.
Consider two possibilities here:
- Third-degree price discrimination: retailers charge different prices if you are using the debit card than if you are not (just like some small stores charge different prices for cash versus credit cards). Since customers not using the debit card have more stores available to them, they have more elastic demand, so the same stores charge higher prices to CCT customers than non-CCT customers.
- Monopolistic competition: here stores charge the same price to all customers, but distance among stores along with the existence of the program gives them some ability to set prices that differ across stores. Here the stores that are part of the CCT program will set higher prices, mostly serve CCT customers, and then stores that aren’t part of the program will have lower prices. Some non-CCT customers will still pay the higher price to shop at the CCT store due to it being closer/more convenient to them.
The authors don’t observe what different types of clients pay for the same product at the same store, and can’t distinguish between these two stories. Either explicitly asking retailers about the prices charged to different clients, and/or surveying different types of clients would allow this to be measured better in future work – the general point is that measuring retail price data is hard.
Measurement period: the authors measure impacts 6 months after entry. It would be nice to have time series on how firms respond to see how quickly competition has effects, and whether the market has come to a new equilibrium by 6 months, or if it takes longer for firms to figure out their new optimum pricing strategies.
Firm effects: it would be interesting to measure what happens to the profits and sales of existing CCT network firms and of those of the new entrants to the network after this policy change. The authors look at the number of customers that retailers have on their best day, finding a negative but insignificant effect, and don’t have these other measures. It would also be interesting to know whether more competition leads to more efficiency in terms of sales moving from less productive to more productive firms.
This is an area where there is very little in the way of experimental evidence, so this paper is a great contribution to the literature. I look forward to seeing more work which can rigorously isolate the impacts of policies to influence competition.
Join the Conversation