Supermarket Entry and the Survival of Small Stores

Abstract

We analyze the effect of supermarket entry on the exit of small stores in the food retailing sector in Montevideo between 1998 and 2007. We use detailed geographical information to identify the link between supermarket entry and the exit of nearby small stores. Entry of supermarkets using small- to medium-size formats creates a competitive threat for the existing small stores, decreasing their probability of survival. The result is robust to several model specifications and varying definitions of what constitutes a supermarket. The impact of supermarket entry is unequivocal for groceries, bakeries, fresh pasta shops, and butcher shops.

Appendices

Appendix 1: Data Sources

We collected seven different data bases. The first is from the Servicio de Regulación Alimentaria (SRA) of The Intendencia Municipal de Montevideo (IMM), the executive branch of the province’s government (home of the “Intendente”). SRA is the municipal authority that is in charge of keeping a record of all food businesses, and of controlling their sanitary condition. SRA establishes definitions; it sets standards for management and personnel, food operations, and equipment and facilities; and it handles inspections and suspensions in order to safeguard public health and provide consumers with food that is safe, unadulterated, and honestly presented. The database indicates, for each year between 1998 and 2007, which stores that deal with food were open. All data about “small businesses” comes from this database; the rest of the databases concern data and definitions about supermarkets.

The second database, is from the Instituto Nacional de Carne (INAC, the meat regulatory body). Every food business in Montevideo is required to obtain a SRA permit to start operations, with the exception of butchers who need permission from the INAC. The INAC issues permits that are valid for 2 years. The data base from INAC contains two types of data: one is the list of permits for butcheries within supermarkets, and another is the list for small butcher shops.

Our third data base is from the Dirección General de Comercio (DGC) from the Ministry of Economics and Finance. DGC provides a public list of supermarkets that are required to report prices of a consumer goods basket on a monthly basis. Each supermarket must submit its price levels to DGC if it has more than three cash registers or if it belongs to a chain of supermarkets with four or more business units.

The fourth and fifth databases are about supermarket size and were purchased from Ciudata and from IDRetail, which are two local consulting firms. They include information about size in square meters, and the number of cash registers for each supermarket. The data from IDRetail were collected in 2005, and the data from Ciudata were surveyed in 2008. We encountered some differences in the observations that were present in each database. In the case of large scale supermarkets, some differences were relatively large. In order to make a quality evaluation of each data source, we interviewed staff members from IDRetail and Ciudata. An important observation is that in the case of IDRetail, the information that is related to the large supermarket chains that operate in Montevideo was reported by Asociación de Supermercados del Uruguay (Uruguayan Supermarket Association; ASU), which is the trade association for the majority of the supermarket chains that operate in Uruguay (with the exception of Tienda Inglesa). In the case of large-chain supermarkets, we used information that was provided by ID Retail (i.e., originally from ASU). In the case of smaller supermarkets, we used information that was provided by Ciudata. We found that the Ciudata database contained information for a larger number of supermarkets than did the IDRetail database. Additionally, for small supermarkets, differences in area information were less significant than for large-scale chain supermarkets. Data about the number of cash registers are based on information that was obtained from Ciudata and Dirección General de Comercio.

Our sixth data base is the “Encuesta Continua de Hogares” (ECH, the national household survey). ECH is prepared by Instituto Nacional de Estadística (INE) and contains socioeconomic information at the individual and household level. From the ECH we collected data about average income, wealth and population in each CCZ in each year. Care should be taken when working with the income data that are present in the ECH. It is well recognized that, in general, people tend to under report their income levels. Additionally, the information on population that is contained in the ECH presents problems that are due to recent changes in the sampling procedures. In particular, the sample is not representative of the total population of Montevideo.

Finally, our seventh database is the construction permits from the IMM. Since any new construction or remodeling needs a permit from IMM, this database contains year of request, square meters of remodeling or construction, and year of authorization.

Appendix 2: Data Processing

The SRA provided five files in xls format: (i) Companies, list of active companies; (ii) Annulled, list of annulled companies, which are currently not active; (iii) Line of Business, list of the main line of business, defined on an ad hoc basis, (iv) Street codes, list of codes in order to locate stores; (v) Complexes, list of housing projects (that do not have street numbers, and where some stores are located). Along the analyzed period, there are 8,871 active companies, and annulled and inactive companies are 16,944 in the SRA database.

To begin, all establishments that operate in lines of business that do not compete with supermarket services were eliminated from the list of active and annulled companies. We were particularly interested in the following areas: groceries, bakeries, pasta shops, and kiosks. The data base shows for each store: (i) Main line of business (bakery, grocery store, fish market, etc); (ii) N$^{o}$ of procedure (the number that the SRC assigns to each request for a permit); (iii) Street code (each street has a code, and for each store, the street code is the code of the street on which it is located); (iv) Street number; (v) Complex address (if the store is located in a housing project, the address of the store is just the name of the project in which it is located); (vi) Centro Comunal Zonal (CCZ, this information was obtained from the database that was processed by Dubra and Ferrés (2006)); (vii) Opening dates (registration, authorization, and renewal dates); (viii) Closing dates (registration, authorization, renewal dates and annulment dates).

Next, we generated a dummy variable, where a 0 stands for a year in which the company was not active or closed, and 1 for a year in which it was active in the market. This was done following criteria about the date of registration, authorization, renewal, and annulment, as in Dubra and Ferrés (2006).

Next, all of the companies that were in the list of annulled companies that were closed and then reopened, were discarded from the list of annulled companies, modifying the corresponding dummy variable in the list of active companies. Last, all of the companies that were never active in the period considered were discarded. For completeness, some details about when a company was considered active follow.

From the list of active businesses, we deleted all of those that should have renewed their permits before December of 1997, but did not. We assumed that these stores were closed before the start of our study (379 were deleted in this stage). We then eliminated approximately 2.500 stores that had opened more than 5 years before January of 1998, and had no renewal dates (we assumed that they were not active at that date). Those stores that opened between 1993 and 2002, but had no date of renewal for their permits, were imputed a closing date of 5 years after their opening. Finally, we “merged” all stores that had opened and closed on the same location. For the purposes of this study, we do not care whether the ownership of a store changed during the period, only if it ceased operations.

To define each category of food shop we use the definitions from the SRA that are in Table 6.

Table 6 Definitions of retail food shop categories

We did an extensive review of the data. We checked the date of opening for all of the supermarkets in our database. We did this for the main chains (Disco, Devoto, Tienda Inglesa, TATA, Multiahorro), and some other individual supermarkets. We also checked web pages to determine whether these data were available. For those supermarkets for which we could not obtain information about the date of opening from these sources, we maintained the data from SRA.

Finally, we checked the consistency of the supermarket area data, as our two database had, in some cases, huge differences. Multiahorro gave us the exact area of each supermarket in its chain, and in the case of Macromercado we found the information in its institutional web page. We also used information that the supermarkets provided to the Supermarket Commission. Some supermarkets suggested that we use IdRetail data on store area because it was more reliable. In those cases for which we could not check the data, nor was IdRetail information available, we used the data purchased from Ciudata.