Microsoft Word FINALSCN.DOC


1 Food Stamp EBT Systems and Program- Eligible vs. Non-Eligible Food Items REPORT TO CONGRESS Contract No. 53-3198-6-029 August 1998 Prepared for Ken Offerman, Project Officer U.S. Dept. of Agriculture Food and Nutrition Service 3101 Park Center Drive Alexandria, VA 22302 Prepared by John Kirlin, Abt Associates Inc. William Adam, CCMI

2 The U.S. Department of Agriculture (USDA) prohib its discrimination in all its programs and activities on the basis of race, color, national origin, gender, religion, age, disability, political beliefs, sexual orientation, or marital or family status. (Not a ll prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communica tion of program information (Braille, large print, audiotape, etc.) should contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room 326-W, Whitten Building, 14th and Independence Ave nue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider and employer.

3 Table of Contents Summary ... ... 1 Introduction... ... 3 The EBT System Environment at the Point of Sale ... ... 4 POS Terminal Configurations ... ... 4 Identifying Eligible Food Items ... ... 5 Linking EBT Terminals to Retailers’ ECRs ... ... 7 Integrated Terminals and ECRs... ... 7 Stand-Alone POS Terminals ... ... 8 Estimated Costs... ... 11 Linking EBT and Scanner Systems in the WIC Program ... 17 Preventing the Purchase of Ineligible Items ... ... 18 Exhibits Exhibit 1: Stores with Scanning Capabilities, by Store Type ... 12 Exhibit 2: Estimated Costs for Stores with Scanning Systems ... 14 Exhibit 3: Estimated Costs for Stores without Scanning Systems ... 16 Exhibit 4: Total Estimated Costs, by Store Type... ... 16

4 Food Stamp EBT Systems and Program-Eligible vs. Non-Eligible Food Items Summary lleviate poverty-related hunger by increasing the food The Food Stamp Program (FSP) is designed to a purchasing power of low-income households. Program be nefits are therefore restricted to the purchase of As a result, the use of program benefits to buy eligible food items, as defined by program regulations. of benefits for cash (trafficking) represent major areas of concern non-eligible items and the exchange cidence of these problems, the U.S. Congress desires to within the program. In an effort to reduce the in know whether electronic benefits transfer (EBT) syst ems can be configured to differentiate between program-eligible and non-eligible items. The only technology currently able to differentiate between eligible and non- eligible items is food retailers’ optical scanning systems, which identify ite ms based on bar-coded information affixed to their packaging. An estimated 32 percent of all program -authorized retailers currently use scanning systems, and these stores—which are mostly supermarkets—red eem approximately 74 percent of all food stamp benefits. Because many scanning systems are programmed to identify FSP-eligible items, the accidental purchase of non-eligible items with FSP bene fits should be reduced in these stores. The mere presence of an optical scanning system, howev er, is not sufficient to prevent the purchase of non-eligible items with FSP benefits. For an EBT system to differentiate between eligible and non- eligible items, two factors must be present in program-authorized stores. First, the store must use a scanning system. Second, the store’s EBT terminals must be linked to the store’s electronic cash registers at the point of sale (POS) so that information on scanned and eligible items can be passed to the EBT system. Only then can EBT purchase amounts be limite d to the total dollar value of program-eligible items. Achieving this linkage is technically feasible in about 95 percent of all authorized retailers. Indeed, the percentage of FSP stores using newer, integrated POS required linkage already exists in a relatively small systems and in stores participating in an EBT demonstration for the Special Supplemental Food Program for Women, Infants and Children (WIC). Introduci ng the required hardware and software in all FSP- 1

5 authorized stores, however, will be quite expensive. The estimated initial cost is $4.60 billion, of which $3.30 billion is for the estimated 68 percent of program-authorized stores that currently do not scan. To maintain this functionality over time would require an additional expenditure of $752 million per year to equip new stores entering the FSP and to establish ser vice agreements for equipment in stores that currently do not use scanning equipment. These cost estimates are not indicative of the costs the WIC Program would face in implementing EBT, where costs would likely be lower due to a smaller retailer base, a larger percentage of WIC stores already scanning, and differences in technical requirements between the two programs. No assumptions are made regarding how the estimated cost of $4.60 billion might be shared between the public and private sectors. Although many of the e quipment upgrades envisioned would benefit retailers most smaller food retailers have been unwilling to as well as help achieve the desired EBT functionality, make the investment to date. Further, when c onsidering non-scanning stores, the hurdles to be overcome in introducing a scanning system within a store extend well beyond price. Implementing a scanning system requires much more than simply installing new hardware and software. The store must be committed to creating an accurate computerized price f ile, working with suppliers to keep the price file up-to-date, training staff on how to use scanners, a nd generally dealing with and working through the problems and hassles that often accompany the introduction of a new technology. Therefore, in order to achieve Congress’s desired intent across all program-authorized stores, a strong commitment would be ffort to encourage program-authorized stores to adopt needed in terms of both financial resources and an e newer technologies and equipment. Finally, investing in new hardware and software will not eliminate trafficking or the purchase of ineligible items with food stamp benefits. First, scanni ng is either not feasible or not practicable in about 5 percent of all program-authorized stores (mostly route venders, produce stands, and residential facilities), so no controls would be added there. Second, even where scanners are used, the new technological controls could be circumvented by reta ilers who wish to traffic or scan non-eligible food items as eligible in order to access food stamp benefits. 2

6 Introduction The Food Stamp Program (FSP) is currently transitioning to a new method for issuing and redeeming food stamp benefits. Paper food stamp coupons are bei ng replaced with electronic benefits transfer (EBT) systems. In an EBT system, food stamp recipi ents receive an EBT card that provides access to a computerized account in which program benefits are stored. When the card and its associated security code are used at a special EBT terminal at a stor e’s checkout lane, the dollar amount of the food stamp purchase is electronically subtracted from the recipient’s EBT account and credited to the store’s account. In March 1993, just over 1 percent of food stamp households nationwide recei ved benefits through EBT systems. By March 1998, 37 percent of all food st amp households were receiving benefits through EBT systems in 30 states. As part of the Personal R unity Reconciliation Act of esponsibility and Work Opport ssue and redeem food stamp benefits with EBT 1996 (PRWORA), Congress has mandated that all states i by October 1, 2002. One provision within PRWORA indicates Congress’ desire that EBT systems be able to differentiate between program-eligible and ineligible food items . The specific language requires that state EBT systems include: (ii) effective not later than 2 years after the da te of enactment of this clause, to the extent practicable, measures that permit a system to differentiate items of food that may be acquired with an allotment from items of f ood that may not be acquired with an allot- 1 ment; ible from ineligible food items. Neither coupons The FSP currently relies on cashiers to determine elig nor existing EBT systems indicate what is being purch ased and paid for with food stamp benefits. Thus, they cannot limit what is being paid for with program benefits. The only automated systems in place in the retail f ood industry that can identify the items purchased are optical scanner systems. Using special bar codes placed on food items, scanner systems can identify what is being brought to the checkout lane for purchase. Scanner systems, however, are not configured to place limits on the value of EBT transactions. Indeed, in most retail stores in EBT states and counties, there is no direct link between the EBT system and the retailer’s optical scanner system. This report explores the feasibility and potential cost of enabling EBT systems to differentiate between program-eligible and ineligible items. It considers th e cost of upgrading systems in stores that now have 1 Public Law 104-193, Section 825 (a)(2)(D)(ii). 3

7 scanners and the cost of installing new systems in stores without scanners. The report also examines the with the introduction of new technological controls. potential for the purchase of ineligible items even The EBT System Environment at the Point of Sale Before identifying ways in which a link between an optical scanning system and an EBT system might be achieved, it is necessary to understand how retailers’ point-of-sale (POS) systems at the checkout lane rminal configurations a nd how retailers currently currently operate. We therefore discuss below POS te identify food stamp-eligible food items. POS Terminal Configurations EBT transactions, together with credit card and debit card transactions, form part of a class of electronic financial transactions known as electronic funds transf ers, or EFT. Any retail food store wishing to accept a credit, debit, or EBT card for payment must have a terminal at the point of sale with EFT room” computer controlling multiple terminals) capabilities. The POS terminal (or a separate “back formats a message to the EFT processor seeking paym ent authorization for individual transactions initiated with the credit, debit, or EBT card. Excluding for the moment terminals that can handle EB T transactions, retailers ha ve two general types of POS terminals capable of handling EFT transactions. The most common is a “stand-alone” terminal, separate from the store’s cash register. To use a st and-alone terminal, the cashier needs to manually key- enter the purchase amount on the terminal’s keypad. Fa r less common are “integrated” POS terminals, in which all EFT transactions are processed directly by th e store’s electronic cash register (ECR). With an integrated terminal no manual entry of purchase amount is required; the total amount of the sale is automatically incorporated into the EFT transaction request. With the addition of EBT, three different terminal configurations are possible: · stand-alone terminals that process only EBT transactions; · stand-alone terminals that process credit and debit card transactions, as well as EBT trans- actions; and · integrated terminals, in which all electroni c funds transactions are processed by the store’s ECR. The configuration in place in any given store depends on several factors. FSP-authorized stores using 4

8 government-supplied equipment may participate in an EBT system without paying any program-related ts an EBT system, stores without electronic payment fees or charges. That is, when a state implemen terminals (i.e., those stores not accepting credit or debit cards) can elect to receive stand-alone, EBT-only terminals from the state free of charge. Unless the store’s food stamp redemptions exceed 15 percent of ill be equipped with free EBT terminals is limited to total food sales, however, the number of lanes that w 2 one free terminal for each $8,000 or $11,000 in monthly food stamp redemptions. Stores having no EFT capabilities can negotiate if they wish with the EBT vendor (or another vendor) for installation of stand-alone EFT terminals that can accept EBT, credit, and debit cards. Unlike the EBT- only stand-alone terminals, retail stores need to pay for fully-functional EFT terminals. They may, rough per-EBT-transaction charges negotiated with the however, be able to recover some or all costs th EBT vendor. Finally, EBT functionality can often be added to stores with existing credit and debit card capability, whether their existing POS terminals stand alone or ar e integrated with the lane’s ECR. Typically, the software controlling the debit and credit transactions must be modified so that EBT transactions are 3 routed to the state’s selected EBT vendor, either directly or through the store’s selected EFT processor. that terminal software is upgraded to handle EBT It is the store’s or processor’s responsibility to see transactions. Again, per-transaction charges may be a pplied to EBT transactions handled by the terminal. These charges would be paid by the EBT vendor. Identifying Eligible Food Items Before turning to the issue of how EBT systems might be able to differentiate between program-eligible and ineligible items, we discuss how scanning systems identify eligible items. It is unlawful to use FSP benefits to purchase anyt hing other than “eligible foods.” According to regulation, eligible foods include: · Any food or food product intended for human consumption except alcoholic beverages, tobacco, and hot foods and hot food products prepared for immediate consumption; and 4 Seeds and plants to grow foods for the personal consumption of eligible households. · 2 The monthly redemption figure for supermarkets is $11,000; for all other stores it is $8,000. 3 Some stores process their own EFT transactions, whereas others contract with a third party for this service. le foods” also includes meals prepared and delivered or served by various CFR Section 271.2 (Definitions). “Eligib 4 institutions, as specified in the regulation. 5

9 Food stamps cannot be used to buy: · Any non-food items, such as: – pet foods – soaps, paper products, and household supplies – grooming items and cosmetics · Alcoholic beverages and tobacco · Vitamins and medicines · Any food that will be eaten in the store · Hot foods that are ready to eat · Any food marketed to be heated in the store In many stores it is up to the cashier to enforce these limits on program-eligible food items. That is, if the customer says that food stamp benefits will be used to pay for all or a portion of the items brought to the register, the cashier is supposed to ensure that food stam p benefits are not used to pay for ineligible items. Although many cashiers are diligent about ensuring that food stamps are used only for eligible items, the possibility clearly exists that ineligible items will sometimes be purchased with program benefits, either intentionally or accidentally. This is the case rega rdless of whether the customer is using food stamp coupons or an EBT card. In more and more food stores, however, optical scanning devices are being used to scan bar codes on purchased items to ring up the item’s price. Bar c odes on most pre-packaged items (e.g., cereal, milk, canned fruits and vegetables, frozen dinners) adhere to a Universal Product Code (UPC) format that 5 Food retailers with scanning systems also enables the store’s POS system to uniquely identify the item. create their own unique bar codes for non-UPC items such as meat, cheese, or fruits and vegetables supplied by local independent supp liers. These latter bar codes are referred to as price look up (PLU) codes. PLU codes generally contain coded informa tion about the item’s category or department (e.g., meat) and its unit price. A store’s primary purpose for installing a scanning system is to improve productivity at the checkout ce file” that lists each UPC in its inventory and its counter. When the store maintains a computerized “pri associated price, the store’s POS system can automa tically ring up an item’s price when the item’s UPC 6 code is scanned. Scanning systems also provide a means of tracking inventory. More important for this 5 The UPC is typically a 12-digit number co mposed of a six-digit manufacturer’s c ode, a five-digit item number (selected by the manufacturer), and a single check digit. Some versions of the code remove embedded zeros; the remaining portion of the code follows an accepted pattern for indicating wh ere the zeros are located within the overall code. e item’s unit price, coded within the PLU, by its weight or Prices for items with PLU codes are computed by multiplying th 6 size, as determined at the checkout counter. 6

10 report, however, is the fact that most stores with scanning systems also have a means of tracking which items being purchased are food stamp-eligible. The sam e computer file that contains a price associated ain a food stamp eligibility “flag” associated with with each UPC code also can be programmed to cont ile, the flag is set whenever the item is eligible. each item. As the store builds and maintains its price f The same process occurs for items with PLU codes. Then, as items are scanned, the store’s POS system eligible items. After all items are rung up, the POS tracks both total price and the price of all food-stamp- system can display a food stamp subtotal if the customer plans to use food stamp benefits to pay for all or part of the purchase. This process reduces potential cash ier error or confusion as to whether an item is 7 FSP-eligible or not. Linking EBT Terminals to Retailers’ ECRs To enable EBT systems to distinguish between eligible and ineligible items, two conditions must be met. First, the store must have an optical scanning system in place, with a price file that flags food stamp- eligible items. Second, because the scanning system is linked to the store’s ECR, the EBT terminal must communicate with the ECR as well. Operationally, with such linkage the ECR would first determine the food stamp total based on the total price of all food stamp-eligible items scanned. Then, the ECR would generate the EBT transaction request without requiring manual entry of the food stamp total. Limiting the transaction request total to the summed value of FSP items would ensure that food stamp benefits 8 were not used to buy ineligible items. Integrated Terminals and ECRs The desire that food stamp benefits be limited to food stamp-eligible items is already being met in stores 9 In such systems, the POS terminal may be integrated with a store’s ECR with integrated POS terminals. through cabling, or the terminal and ECR may be built t ogether as a single unit. In either situation, the ECR’s EFT software module will normally be configured to prevent a food stamp EBT transaction amount from exceeding the subtotal of food stamp-e ligible items. For example, once the cashier presses 7 The POS software that identifies FSP-eligible items has been created independent of USDA involvement. Programming and updating the software to identify which items are eligible are the responsibility of the retailer and product suppliers. USDA has made no effort to monitor or verify the accuracy or tim eliness of this software to ensure that items are properly flagged. 8 This, of course, assumes no overt fraudulent activity, such as scanning eligible items that th e customer does not intend to purchase. The potential for fraud is exam ined further later in this report. 9 It should also be pointed out that, in the EBT demonstration in Wy oming, a small number of st ores have integrated POS terminals that limit the purchas e of food items in the Special Supplemental Food Program for Women, Infants and Children nd EBT are described later in am to link scanner systems a (WIC) to WIC-prescribed items. The efforts of the WIC Progr this report. 7

11 the “Food Stamp EBT” tender key on the ECR, the ECR will format a message to the EBT vendor requesting authorization for a food stamp transaction not to exceed the food stamp subtotal of scanned items. Stand-Alone POS Terminals stand-alone EBT terminals. The EBT terminal is Most retail food stores in EBT locations, however, have vice. Instead, the cashier key-enters the requested not linked to either the ECR or the optical scanner de EBT transaction amount into the EBT terminal. In st ores with scanners, the appropriate dollar value of food stamp-eligible items is likely to be subtotaled by the ECR. In other locations, the clerk must determine the appropriate subtotal. For current purposes, stand-alone POS terminals that can process commercial debit or credit transactions as well as EBT transactions are similar to EBT-only te rminals; the EBT purchase amount still needs to be 10 The following discussion therefore speaks of both types of manually key-entered into the terminal. terminals as simply “stand-alone” terminals. The manual key entry of the EBT payment amount on a stand-alone terminal clearly has the potential for key entry error, even in stores with scanner systems. Furthermore, with manual key entry, there is no automatic means to ensure that food stamp benefits are not being used to purchase ineligible items. To e owner with a stand-alone terminal would have achieve the functionality required by Congress, a stor three options: (1) Replace the stand-alone terminal (and the re tailer’s current ECR, most likely) with an integrated terminal/ECR with appropriate software; (2) Physically connect the existing terminal and ECR and make necessary software mod- ifications; or (3) Transfer EFT processing to the existing ECR by making necessary software modifications, and scrap the old stand-alone terminal. Of course, software modifications would most likely be done by hardware or software vendors, not by the retailer. The first option takes us to where this section star ted—an integrated terminal/ECR already programmed to meet Congress’ requirements. Integrated system s, however, are a very expensive solution for the tinction between stand-alone and integrated terminals. 10 This is the principal dis 8

12 11 individual retailer. As such, it seems a viable option only for retailers who are about to upgrade their POS systems anyway. The remainder of this section discusses the other two options noted: physically connecting the terminal and ECR, and transferri ng all EFT processing (including EBT) to the ECR. to an EBT payment terminal is to connect the two One way to transfer food stamp subtotal information devices physically, for example by connecting the EBT terminal to the ECR with an RJ45 or RS232 cable. The ECR would need to be programmed to pass the EBT subtotal amount and program type to the EBT terminal once the ECR’s “EBT Payment” key w as pressed. The EBT terminal would also require programming to accommodate the EBT tender total and program type information passed from the ECR. and program type would automatically be present With this reprogramming, the EBT subtotal amount within the EBT terminal when the clerk (or client ) pressed the EBT terminal’s “send” button, thereby initiating EBT processing of the purchase. The EBT terminal’s programming would also need to allow for an “override” function for food stamp or cash assistance purchases, allowing the clerk to manually key-enter a lower dollar amount for the EBT transaction, but not a higher amount. The manual override would be used when the client’s remaining EB T balance was less than the purchase amount (for cash assistance transactions) or food stamp subtotal (for food stamp transactions), or when the client decided for other reasons to use another payment form (e.g., cash) for part of the food stamp-eligible portion of the total purchase. To complete the tender process, the customer would enter his or her security code, and the transaction would be sent to the EBT processor for approval. Once the EBT transaction was authorized, the EBT terminal would pass the acknowledgment on to the E CR. Upon receipt of the EBT acknowledgment, the ECR would apply the EBT payment amount to its record of the transaction. There are several substantial hurdles to be faced in achieving the above physical connection and functionality. First, older ECRs do not have an ava ilable communications port for a connecting cable. In addition, software revisions to many of these olde r ECRs are no longer supported by the ECR vendor. In these situations, retailers would likely need to replace th eir ECRs with newer, integrated models to meet the goal of linkage. Even when the EBT terminal and ECR can be c onnected, both require reprogramming to achieve the desired linkage. Contacts with EBT vendors suggest th at EBT terminal software development would cost system is likely to cost at least $8,000 per lane. 11 As discussed later in the report, an integrated 9

13 at least $35,000 to $40,000. This estimate, however, does not include the cost of installing the updated software in the EBT terminals. This cost estimat e also assumes that the information being passed from the ECR would be in a standard format, regardless of ECR manufacturer or model. Without such a protocol, separate reprogramming would be needed for model of ECR to which EBT terminals are each models of ECRs for which separate programing connected. There are at least several dozen popular might be required. Finally, the ECR needs to be reprogrammed. The pr ogramming itself should not be too difficult. It is likely to be time-consuming and difficult, however , to get the numerous manufacturers and retailers involved to undertake the programming. Furthermor e, as noted above, the programming specifications should adhere to a standard interf ace protocol with the EBT terminal. The above discussion suggests that the physical connec tion approach may be quite difficult, achievable only on an interface-by-interface basis. Looking to the future, however, several independent software vendors offer payment system software that will allow th e integration of the terminal to the ECR. Even this software would need to be modified, however, to ensure that the EBT total did not exceed the food stamp subtotal for food stamp purchases and to allow manual override when the EBT transaction amount needed to be less than the food stamp subtotal. Du e to the expense involved, these software solutions tend to be targeted to multi-lane s upermarkets with newer POS systems. Another option for “linking” the terminal and ECR involves processing all EFT transactions directly from r EBT) terminal. At least two companies provide the ECR, eliminating the need for a separate POS (o sactions, directly routing debit and credit transactions software that allows the ECR to control all EFT tran not now have the capability to process EBT trans- for approval. Although these software products do esumably be added. This appro ach, however, would seriously narrow actions, this functionality could pr retailers’ choice of software vendors unless the other major vendors of ECR software decided to develop and offer similar packages. Unless this happened, the physical connection approach would probably gain wider acceptance among retailers, ECR manufacturers, and software vendors. Estimated Costs Approximately 183,300 retailers in the United States are authorized to participate in the FSP. Of these, about 173,700 represent retail food outlets where opti cal scanning systems could be used to identify 12 In January 1998, these stores redeemed $1.44 billion in food stamp program-eligible food items. 12 The remaining locations include about 5,500 route vendor s and produce stands—places where optical scanning systems has no affixed UPC codes. cannot be installed or where installation would make no sens e because the product being sold 10

14 benefits, or over 99 percent of that m onth’s total redemption of $1.45 billion. ates, having established in the prior section that it By October 1, 2002, when EBT is mandated for all st rminals, the next question is what costs would be should be technically feasible to link ECRs to EBT te incurred to achieve this linkage. This section assembl es available information to provide an estimate of the cost to upgrade EBT systems. The cost calculati ons are meant to be conservative, providing a lower- EBT terminals in nearly all program-authorized bound estimate of the cost to link ECRs, scanners, and stores. No attempt is made to discuss how these costs, which are substantial, might be allocated or shared between the public and private sectors. Exhibit 1 presents the distribution of the 173,700 program-authorized food stores in the United States by rcentage of stores equipped with optical scanning store type, together with two estimates of the pe systems. The first estimate is based on USDA’s Store Tracking and Redemption Subsystem (STARS) database. This estimate is a known lower-bound estimat e; retailers self-report their scanning status when d at re-authorization. Overall, the STARS database they are authorized, but this information is not update indicates that about 18 percent of program-authorized stores use scanners. The second estimate of stores with scanning systems is based on figures presented in industry trade publications and, where figures are not available, on the judgement of an independent contractor with 13 experience in retail scanning systems. Trade publications indicate that about 93 percent of chain-based supermarkets use scanning systems, compared to a bout 83 percent of independent supermarkets. The estimate of 88 percent in Exhibit 1 assumes that chai n-based and independent supermarkets are evenly distributed across program-authorized stores. These pub lications also indicate a substantial recent jump in the percentage of convenience stores with scanners —up to 33 percent. This estimate seems skewed by the small sample of convenience stores responding to an industry survey, however, so we have used a more conservative estimate of 25 percent. For the ot her store types, no evidence regarding the prevalence of scanning systems is currently available. The assumed values in the exhibit are meant to be con- servative while acknowledging that, over time, more stores are installing scanning systems. With these newer figures, the estimated proportion of all program-authorized stores with scanning systems is 32 percent. Another 3,900 locations are not retail food outlets; they incl ude places like drug and alcohol treatment centers, women’s shelters, group living homes, communal di ning facilities, and food wholesalers. a subcontractor to Abt Associates Inc. 13 Consumer Card Marketing, Inc. (CCMI) is for the study under which this report has been prepared. CCMI markets card-based “frequent s hopper” systems to stores with scanner systems. 11

15 Exhibit 1 Stores with Scanning Capabilities, by Store Type Percentage Store Type Percentage Percent with Percent with Number of Scanners of Total of Total Scanners Authorized Stores (lower- Redemp- (updated Stores a b tions bound) estimate) 79.2 78 18.0 Supermarket 31,295 88 0.1 Military commissary 88 244 75 0.1 25.7 10.9 5 44,691 Medium to small grocery 20 3.0 Convenience store 4 27.4 25 47,665 12.5 10 5 Grocery/Gas station 1.1 21,792 47 2.8 60 4,844 Other grocery combination 1.1 3 15,619 Specialty food 10 9.0 3.6 1.4 4 0.3 10 General store 2,488 0.5 861 Grocery/Restaurant 0.1 2 5 7 0.9 Health/Nutrition food 0.1 1,580 10 Other stores 5 0.4 10 1.5 2,621 100.0 18 32 173,700 100.0 ALL STORES a Based on STARS data. b Based on recent trade publications and judgement of independent contractor. Note: of 100.0 due to rounding. Percentages may not sum Stores with scanning systems can be grouped into thr ee categories, depending on what needs to be done in the store to achieve linkage between the POS/EBT te rminal and the ECR. Nothing needs to be done in stores whose ECRs and POS terminals are already integr ated. In stores with newer ECRs and stand-alone terminals, the terminals and ECRs need to be physi cally linked, and software needs to be upgraded. Finally, in stores with older ECRs, physical linkage often is impossible because the ECR has no available communications port for a cable to the EBT/POS terminal. Linkage in these stores is possible only by replacing the older systems with newer ones. In estimating the cost of achieving linkage, we have at is, we assume that 20 percent of stores with assumed a 20/40/40 distribution across these categories. Th scanning systems have integrated POS systems, 40 percent have terminals and ECRs that can be 14 physically linked, and 40 percent have systems requiring complete replacement. 14 No industry source tracks the distribution of scanning systems acco rding to these categories, eith er for all stores as a gro up or for specific store types. One EBT vendor stated that, in th e states it served, about half of the ECR systems were so old 12

16 Exhibit 2 shows the estimated cost of achieving linkage among stores with scanning systems. Total For the estimated 22,387 stores needing a physical estimated cost for these stores is $1.30 billion. connection between their ECRs and EBT/POS terminals, the total cost is $15.0 million, or an average of ng new equipment, however, the total cost is $1.29 $670 per store. For the estimated 22,387 stores requiri billion, or an average of $57,410 per store. These cost estimates are based on the following assumptions developed by the study’s subcontractor: · For stores requiring new equipment – For multi-lane stores, cost for a system controller averages $15,000 per store, including 15 regular POS software. – Cost for a basic integrated ECR/terminal, with regular POS software and an optical scanner, averages $8,000 per lane. – The cost to modify ECR software averages $40 per store. – The average cost to modify EBT terminal software is $25 per store. r day, with two to five days needed per – Cost for installation and training averages $800 pe store, depending on size. – All lanes are equipped, even if EBT terminals are not present in all lanes. · For stores needing an upgrade – The cost to modify ECR software averages $40 per store. – The average cost to modify EBT terminal software is $25 per store. – The cost to install software and physically link the ECR and terminal averages $200 per lane. – In stores with one or two lanes, all lanes are upgraded. In stores with three or more lanes, 16 one-half of all lanes, on average, are upgraded. that no communication ports existed. The assumed split of 20/40/ 40 is a bit more conservative in terms of estimated total costs required to upgrade or replace ECR syst ems in stores with scanning systems. 15 No controller is needed for single-lane stores. aded. Under federal EBT regulations, about 47 percent of all 16 The assumption is that all lanes with EBT terminals are upgr lanes are eligible for free, government-provided EBT terminals. The actual percent is higher because all lanes are equipped in some stores, whether by the EBT vendor or store owner. 13

17 Exhibit 2 Estimated Costs for Stores with Scanning Systems System Replacement Required Store Type Total Cost Upgrade Required (000s) Total Cost Average Total Cost Average Cost/Store Cost/Store (000s) (000s) $1,005,569 Supermarket $ 90,323 $10,577 $ 960 $ 994,992 9,554 1,213 9,449 104 109,923 Military commissary 75,443 334 20,767 Medium/Small grocery 74,247 1,196 Convenience store 102,465 100,749 21,137 1,716 360 315 21,607 Grocery/Gas station 18,835 362 19,150 61,338 72,109 71,303 Other grocery combination 807 694 Specialty food 9,401 15,047 192 307 9,593 22,102 37 2,236 2,200 General store 368 19,385 343 334 Grocery/Restaurant 340 6 Health/Nutrition food 22 1,415 353 1,393 22,031 38 Other stores 22,359 2,344 2,383 367 $ 670 ALL STORES $1,285,245 $ 57,410 $1,300,255 $15,010 Columns may not sum to reported totals due to rounding. Note: Note that the assumptions regarding lane coverage diffe r for the two groups of stores. In stores in which linkage can be achieved with a software upgrade, th e ECR software in each lane gets upgraded (at an average cost of $40 per store), but cabling is required only in those lanes where EBT terminals are present. Thus, costs are reduced to the extent that EB T lane coverage in a store is less than 100 percent. need to be upgraded even if all lanes are not In contrast, for stores needing upgraded ECRs, all lanes e would be operating with a mix of both new and old equipped with EBT terminals. Otherwise, the stor ECRs and software. A store could not operate effici ently with two different POS systems, so we must assume that all lanes are equipped with new ECRs. This accounts for the high cost of totally replacing the store’s POS and ECR terminals. Turning to stores without scanning systems, their equipment and software needs are similar to scanning stores with older POS systems. That is, they need a complete ECR and POS system at each lane. Thus, 14

18 the same assumptions are used in estimating equipment costs for these stores. These stores, however, also rs. In particular, a store-specific price file needs have to develop an infrastructure to support use of scanne puter file with one record for each type of item in the to be created and maintained. The price file is a com store’s inventory. Each record typically includes a short descriptor for the item, its bar code identifier (which is the item’s UPC or PLU code), a food stamp- eligibility flag, and the item’s price. Whenever the store wishes to change the price of an item or add a ne w item to its inventory, the price file needs to be updated. The cost of creating a price file will vary depending on the size of the store and how data are collected and added to the file. In any event, it is typically a labor-intensive process. The estimated cost here varies between $4,000 and $7,500 per store, based on information provided by the study’s subcontractor. Exhibit 3 presents the estimated costs for installing sca nner systems with the desired linkage in stores that currently do not use scanning systems. The average cost for a single store, $28,028, is much lower than for scanning stores needing equipment upgrades because stores without scanners tend to be much 17 smaller. Even with a lower per-store cost, however, th e total amount is $3.30 billion. With the number of program-authorized non-scanning stores estimated at nearly 118,000, any attempt to introduce improved technology at the checkout is going to be expe nsive. When the cost of $3.30 billion is added to nning stores needing upgrades or new equipment, the the estimated cost of $1.30 billion for already-sca total cost is $4.60 billion. Exhibit 4 shows how this total cost is distributed across the different store types within the FSP. Even though 88 percent of supe rmarkets already scan, the cost of upgrades at these stores and new equipment at the remaining supermarkets represents 30 percent of the total estimated cost for all stores. This expense ari ses because the average number of checkout lanes in supermarkets (nearly nine lanes per store, based on data in the STARS datab ase) is much higher than in other stores. Costs are high at small- to medium-grocery stores and conve nience stores because of the large number of these stores participating in the FSP. 17 Based on the STARS database and this re port’s assumptions regarding the percentage of stores with scanning equipment, non-scanning stores average 1.8 lanes per stor e. In contrast, stores with scanning sy stems average 5.2 lanes per store. This of store and the most likely to scan. commissaries are both the largest types arises because supermarkets and military 15

19 Exhibit 3 Estimated Costs for Stores without Scanning Systems New System Required Store Type Total Cost Average (000s) Cost/Store $ 367,367 $ 97,823 Supermarket 3,441 117,423 Military commissary 956,991 26,767 Medium/Small grocery 25,137 898,609 Convenience store 25,607 Grocery/Gas station 502,228 Other grocery combination 126,588 65,338 Specialty food 267,743 19,047 26,102 58,446 General store 23,385 19,124 Grocery/Restaurant 26,031 Health/Nutrition food 37,022 Other stores 26,359 62,184 ALL STORES $3,299,743 $ 28,028 Exhibit 4 Total Estimated Costs, by Store Type Store Type Percentage of Total Cost Total Cost (000s) $1,372,936 30 Supermarket Military commissary 12,995 <1 1,032,434 22 Medium/Small grocery Convenience store 1,001,074 22 Grocery/Gas station 11 521,378 Other grocery combination 4 198,697 277,335 6 Specialty food 60,682 1 General store 19,464 Grocery/Restaurant <1 Health/Nutrition food 38,437 1 Other stores 1 64,567 16 ALL STORES $4,599,999 100

20 The total cost estimate of $4.60 billion represents the cost to achieve a linkage between EBT terminals The cost estimate, which equals approximately and store ECRs in all currently authorized FSP stores. $26,000 per store, will change somewhat to the extent that the number of FSP-authorized stores changes would be expected to decrease as more stores introduce in the near future. Over time, estimated costs newer scanner and POS systems on their own. Any expect ed reduction in total costs is likely to be small, nner systems increases dramatically, especially in however, unless the percentage of stores adopting sca supermarkets, medium- to small-grocery stores, and convenience stores. any costs which would be required to maintain It is also important to consider the linkage between EBT and ECR systems in the future. In recent years, an average of about 21,000 new stores have been authorized annually. If one assumes that these new stores have the same distribution of store types and existing scanning equipment as current program-authorized stores, then the estimated cost of linking these 18 stores’ EBT terminals with their ECRs would be about $556 million per year. All stores, both new and old, would also need to establish service agreements to maintain their equipment in working order. 19 Providing service agreements for all stores would cost about $371 million per year. Of course, stores with existing scanning equipment and ECRs already pa y for their own service agreements, so the total cost for service agreements in stores that do not currently scan would be lower—about $196 million per year. In addition, in future years, the scanning equipment and ECRs in many stores may need to be is report, however, does not attempt to estimate updated as improved technology becomes available. Th the cost of future upgrades. Linking EBT and Scanner Systems in the WIC Program In recent years, the Special Supplemental Food Pr ogram for Women, Infants and Children (WIC) has begun to explore EBT as a means to automating its paper-based transaction process. WIC benefits, however, comprise a specific “prescription” of food items tailored to the nutritional needs of the participant. This makes WIC’s requirements much di fferent from those of the FSP. In particular, because the prescribed quantity and list of authorized foods va ries from one participant to another, an EBT system for the WIC Program must be able to recognize what is being purchased before authorizing payment to the retailer. That is, an automated screening process fo r eligible foods must be incorporated within the 18 The estimated cost is based on an average cost of $26,482 per store. 19 This cost estimate assumes that annual service agreement costs would average $1,000 per lane in small stores, $750 per lane in medium-sized stores, and $500 per lane in supermarkets and military commissaries. The costs are based on information supplied by a firm providi ng POS service agreements. 17

21 system. An EBT demonstration in Natrona Count y, Wyoming showed the technical feasibility of and other states are planning similar demonstration integrating scanner technology with an EBT system, projects. At this point it is too early to determine how the WIC Program’s efforts to integrate EBT systems and scanner systems might affect efforts to achieve a similar functionality in the FSP. The two programs are very different. For instance, whereas there are over 173,000 FSP-authorized food retailers, WIC has only rthermore, although detailed information is not about 45,000 retailers authorized nationwide. Fu available, a larger percentage of WIC versus FSP retailers are supermarkets where scanning systems are already likely to be present. The technical requirements for integration also diffe r between the two programs. The total number of WIC-eligible food items across the nation is estimat ed to be on the order of 10,000, whereas the number of FSP-eligible items is at least ten times greater. Furthermore, within any given state, the number of WIC-eligible items can be as few as 2,000. Thus, it is possible to develop an EBT system for the WIC Program without linking the EBT terminal to the retailer’s scanner system and associated host price file. Instead, a scanning device can be hooked directly to th e EBT terminal, and information about the limited a separate EBT file. This approach bypasses the set of WIC-authorized food items can be stored in 20 difficult task of integrating the EB T terminal with the retailer’s ECR. Due to these differences in e cost estimates presented in the previous section retailer populations and technical requirements, th should not be extended to the WIC Program. Preventing the Purchase of Ineligible Items At a total estimated initial cost of $4.60 billion, plus an additional cost of $752 million per year, about 95 percent of all FSP-authorized stores could be configur ed so that EBT benefits could be used to purchase only program-eligible food items. These stores redeem over 99 percent of all food stamp benefits. The final question addressed by this report is how effectiv e this investment might be in limiting the use of program benefits to the purchase of eligible items. Although we are unable to quantify this effectiveness, it must be acknowledged that the technological controls offered by scanning systems can be circumvented in several ways. The first circumvention arises simply because it is not feasible to link EBT terminals to scanning systems 20 Several retailers participating in the Wyoming WIC demonstra tion have had their ECRs integrated with the EBT terminal, interface for each different POS system in the demonstration but the EBT vendor had to custom design and build a separate area. 18

22 in about 5 percent of the 183,300 establishments authorized by the FSP. Two percent of these estab- 21 lishments are not retail outlets and therefore do not sell bar-coded items. There is nothing to scan in these establishments. The other 3 percent represen t route vendors and produce stands. These estab- lishments either sell large quantities of non-bar-coded items, or they operate in an environment where scanning would be very difficult to implement. A ll together these route vendors, produce stands, and non-retail establishments redeem about 0.8 percent of Although this percentage all food stamp benefits. is small, it still represents approximately $135 million in annual food stamp benefits that would not be subject to scanning controls. For the estimated 173,700 retail food outlets where scanners either are or could be installed and linked to technological controls would have at least two means EBT terminals, retailers who wish to circumvent the to do so. First, we note that the retailer’s price file is a critical element to ensuring that only eligible food ce file needs to be programmed to include a food items are purchased with program benefits. The pri stamp eligibility flag for each item in the retailer’s inventory of goods sold. If an ineligible item was flagged on the file as eligible, then the EBT system would have no way of limiting the purchase of that item with food stamp benefits. In some cases the fl ag for an ineligible item could be mistakenly programmed to indicate eligibility, thereby allowing the in advertent use of program benefits for ineligible items. In other cases, however, a retailer could fraudulently flag one or more ineligible items as eligible. could be prevented only by government monitoring of (The programming is straightforward.) Such abuse monitoring would require an entirely new set of the eligibility flags on retailers’ price files. This compliance procedures that would be both intrusiv e and expensive. For instance, it could entail unannounced, periodic government review of the food stamp eligibility flags on store price files. Even if monitoring procedures were implemented, trafficking in program benefits would not be eliminated. Retailers who wished to traffic could simply affix a series of UPC codes for eligible food items to a sheet of paper or cardboard kept near the EBT terminal and scanner. When a client wishing to traffic benefits came to the store, the retailer c ould scan any combination of the UPC codes until the desired total amount of the trafficked benefits was r eached. The retailer could then give the client an agreed-upon percentage of the total “sale” in cash a nd receive the full amount of the sale during the EBT system’s normal daily settlement. By using different combinations of valid UPC codes each time, the retailer could avoid having the tran sactions identified as suspect during any subsequent investigation or agency review of EBT transaction data. s, and other providers of prepared meals. shelters, rehabilitation centers, group living quarter 21 These establishments include 19

23 It is beyond the scope of this report to estimate how often either of the above means to circumvent scanning controls might be used. The important point to recognize, however, is that trafficking would scanners, as would the actual purchase of ineligible still be possible after EBT terminals were linked to items. 20

Related documents