Improving Access to Food Stamps -- 9/1/00

Improving Access to Food Stamps:
New Reporting Options Can Reduce
Administrative Burdens and Error Rates
by Dorothy Rosenbaum

States, USDA, and non-profit organizations alike are concerned that eligible low-income working families are not participating in the food stamp program. Like health insurance, the EITC and child care, food stamps is an important benefit that can support work and help bolster workers' low wages. One of the primary reasons given for why many working poor families do not participate in the food stamp program is the time and burden associated with both obtaining and retaining food stamps.

On July 14, 1999, as part of an initiative to increase access to the food stamp program for working poor families, President Clinton announced several new options states can adopt to improve food stamp administrative procedures. These options address how households report changes in their income and circumstances during a certification period. The new options provide important opportunities for states to reduce the paperwork and other administrative burdens on both working families and state agencies. The options also can help states reduce their food stamp quality control (QC) error rates.

The food stamp program has traditionally required households receiving food stamps to report nearly every change in their income. For households with a stable income source, such as cash assistance or Social Security, this requirement may not be a significant burden because the income does not change often. For working families, on the other hand, it can take considerable time to contact the food stamp office to report fluctuating income and provide required pay stubs to verify income.

The example of a family of three with two school-aged children and a mother who works in a clerical job at minimum wage helps illustrate the importance of improved income reporting procedures. The number of hours per week that the employer offers the mother ranges from 25 to 35, depending on the office workload and the availability of other staff. As a result of her fluctuating hours, the mother's income varies from about $130 a week (or $560 a month) to $180 a week (or $780 a month). At any income level within this range the family would likely be eligible for at least $190 a month of food stamps.

Under traditional food stamp reporting rules regarding earnings, the state would have two choices: to require the household to report its income and other circumstances each month on a monthly report form, or to require the household to contact the food stamp office by mail, telephone, or in person, each time its monthly income changed by more than $25. Either option results in a significant burden on the household to make the report and on the state agency to process the information. If the household does not report the change, or if the state does not act correctly to adjust benefits to account for the change, a QC error can result.

USDA's new reporting options can significantly lower reporting requirements for households with fluctuating income:

USDA also announced that states may require families to come to the food stamp office only once every 12 months, even if the state requires the household to reapply for food stamps more often. If states choose to take advantage of these options to require fewer office visits and fewer reports of income and other circumstances, the administrative burdens of the food stamp program on households and state agencies can be significantly reduced.

In addition, less frequent reports of income should help states reduce their QC payment errors. According to USDA, about half of all QC error cases arise because of an incorrect determination of the household's income. If a household does not report a change that it is required to report, or if the state does not act correctly on reported information, a QC error will likely result. If a household was not required to report a particular change and does not choose to report it, however, that change cannot contribute to an error. In the above example, if the state uses quarterly or status reporting it will reduce the number of reports required of the household, which will both reduce the likelihood of errors from fluctuating income and give the state staff more time to focus on other factors that might cause errors and to carry out their other duties.

There likely are also QC advantages for states to have longer certification periods in conjunction with the new reporting requirements. At initial certification and recertification, states must collect information on all elements of eligibility and must verify many elements. Any information that the household fails to disclose, or that the eligibility worker fails to consider correctly, would contribute to an error. During the certification period, however, the state may limit the items of eligibility on which the household must report changes and provide verification. A state that uses three-month certification periods could reduce burdens on households and local agencies, and likely reduce its error rate, by replacing quarterly recertifications with quarterly reporting and longer certification periods.

Many states have taken advantage of these new changes, but about a dozen have maintained their old rules. In addition, some of the states that have adopted changes in this area may not be aware of the full range of options now available or the impact such options could have on participation among working poor families in their state and on QC error rates. This paper first discusses the food stamp QC system and why QC error rates and related fiscal sanctions influence how states structure their food stamp policies and procedures. It then describes current food stamp reporting procedures, including the new options, and concludes with some analysis of the advantages of the various options now available to states.

The number of food stamp participants fell dramatically — by almost 40 percent nationally — between 1994 and 1999. A large share of the decline occurred among low-income families with children. Some decline in food stamp participation was expected because of eligibility changes in the 1996 welfare law and the continued strength of the economy, but a consensus has developed that these two factors do not explain the bulk of the decline. An analysis by the U.S. Department of Agriculture (USDA) found that only about a quarter of the decline between 1994 and 1998 can be attributed to lower participation among unemployed childless adults and immigrants, the two groups whose eligibility was restricted in 1996.⁽¹⁾ Participation of elderly and disabled citizens was largely unchanged. The bulk of the decline occurred among families with children.

Partly because of the strong economy and partly because of states' cash assistance policies, a growing share of families that are eligible for food stamps have earnings, while a shrinking share receive public assistance grants. Studies by USDA have shown that historically, only about half of low-income working families that are eligible for food stamps and do not receive cash assistance actually obtain food stamps. Food stamps are, nonetheless, an important support for low-income working families. A family of three with a parent who works 30 hours a week at the minimum wage (with $250 in rent and utility expenses each month) is eligible for almost $230 a month in food stamps in 2000. Receipt of food stamps would increase the family's pre-tax monthly income by 34 percent. The expansions of the Earned Income Tax Credit (EITC) in the early 1990s were designed so that full-time year-round minimum wage earnings plus the EITC and food stamps would bring a family of four to the poverty line. Without food stamps the family's income is only 77 percent of the poverty line, and its typical monthly income is even lower because the EITC is usually paid as a lump sum once a year.

While there are many reasons that working families may not participate in food stamps, a contributing factor is that families with earnings often have circumstances that make complying with the program's procedural requirements more difficult. Work schedules may make it difficult for working people to appear at certification interviews, which are usually scheduled during daytime working hours. Some states require families with earnings to ask their employers to fill out forms that verify the households' earnings, a practice that some low-income workers may find stigmatizing.

In addition, because households with earnings tend to have fluctuating incomes and less stable circumstances in general, many states fear that serving working families could increase their error rates. By law, USDA must operate a QC system to measure errors in the issuance of food stamps. States with error rates above the national average face financial penalties. Because food stamp benefits are sensitive to minor changes in monthly income or expenses, households with earnings tend to be more error prone. In fiscal year 1998, every state had a higher error rate for households with earnings than for households without earnings. Many states seeking to reduce their error rates have targeted the working poor for increasingly burdensome procedures that cause many to drop off of, or not to apply for, food stamps. For example, states increasingly require working poor households to reapply for food stamps at welfare offices every three months, often forcing them to miss time from their jobs.

This makes last summer's USDA guidance, which offers states reporting options that can reduce state workloads and lessen burdens on working poor households while also helping states achieve or maintain low QC error rates, quite timely.

Unlike the other major state-administered assistance programs for low-income families, food stamp benefits are entirely federally-financed. To ensure that states spend federal funds properly, the Food Stamp Act understandably requires the most rigorous payment accuracy measurement system of the major benefit programs. The Act requires USDA to calculate penalties for states that perform below the national average and to pay extra administrative funding to states with very low error rates.

It is hard to overstate, however, the degree of emphasis that USDA places on payment accuracy. Too often a state's QC error rate is the only measure of performance in administering the program that receives any attention. USDA uses the error rates to determine fiscal sanctions and enhanced funding for each state annually. The states, in turn, often measure the performance of local food stamp offices, and even their individual eligibility workers and supervisors, with these same indicators. As a result of this emphasis, states place great weight on the potential impact of food stamp policy and administrative decisions on state error rates. Because of the enormous attention paid to QC error rates, it is critical to understand how the QC system works and influences state policy.⁽²⁾

To ensure that federal funds are spent appropriately, the Food Stamp Act authorizes a quality control system that measures the accuracy of the states' eligibility decisions and benefit amounts. Each month, after the state issues food stamps for the month, it randomly selects a specified number of cases from two "sample frames." The first, and the one that gets the bulk of the attention, is a sample of all cases that were issued benefits in a given month (known as active cases). The second is a sample of cases that were denied benefits, or whose benefits were terminated, in a given month (known as negative cases). Nationwide, states review about 50,000 active cases and about 30,000 negative cases each year.

For the active cases, the state's quality control reviewers conduct both detailed examinations of the case file and in-depth field reviews. The field reviews include an interview with an adult member of the household and contact, usually by telephone or mail, with other persons, such as landlords or employers, who have knowledge of the household's circumstances. The reviewer seeks the best available verification of the household's actual circumstances in the month under review. The purpose of the review is to determine, within the program's rules, whether the household was eligible and received the correct amount of food stamp benefits in that month. If, based on the results of the review, the QC reviewer determines that the household was ineligible or received an incorrect allotment, the case is cited with an "error" in the dollar amount of the incorrect issuance.⁽³⁾ The state's error rate for active cases is the estimated "error amount," in dollars, as a percentage of all benefits issued in the state.

In about one-third of the cases reviewed by the states, USDA's regional office conducts a second review to verify that the state's QC review was conducted with appropriate rigor. These federal reviews typically involve only an investigation of the eligibility worker's case record and the state QC reviewer's case record.

Each year, based on the reviews of participating cases, USDA issues a combined payment error rate for each state. The combined payment error rate has two components: the overpayment error rate and the underpayment error rate. The overpayment error rate counts benefits issued to ineligible households as well as benefits issued to eligible households in excess of what the program's rules allow. The underpayment error rate measures errors in which eligible, participating households received fewer benefits than they were eligible to receive. The underpayment error rate does not include the value of benefits that should have been provided to eligible households that were improperly denied or terminated. The combined payment error rate is the result of summing (rather than netting) the overpayment and underpayment error rates. For example, a state with an eight percent overpayment error rate and a three percent underpayment error rate would be reported as having a combined error rate of 11 percent. The net loss to the federal government, however, from the errors in that state's program (the benefits lost through overpayments minus those saved by underpayments) would be five percent.

States conduct a separate, much less rigorous review of "negative" cases — those denied or terminated from food stamps. These reviews usually consist only of a desk review of the eligibility worker's case record. If the reviewer is unable to verify that the state agency made the correct decision to deny or terminate a household's participation through the information in the case file or, in some circumstances, through telephone calls seeking further information, the case is considered invalid. The negative case error rate represents the proportion of all denials or terminations that the QC reviewer found invalid. For example, if 1,000 negative cases were reviewed and 70 denials and terminations were found to be invalid, the state would have a seven percent negative case error rate. Unlike the combined payment error rate, in the past federal officials have rarely reviewed the state's findings on negative cases except where, as discussed below, a state may qualify for enhanced funding. In July 2000, however, USDA announced that, beginning with the fiscal year 2000 negative case error rates, the USDA regional offices would conduct a separate review to validate state QC reviewers' findings on negative cases.

States are subject to financial sanctions if their combined error rate is above the national average. Conversely, states may receive enhanced administrative funding if their combined error rate is below six percent and they do not have a high negative case error rate. The Food Stamp Act sets the penalties on a sliding scale, with the more severe penalties applying to states that exceed the national average by the largest margin. For fiscal year 1999, in the absence of the adjustments discussed below, 24 states would have been subject to a total of $75 million in liabilities based on their payment error rates. (See tables 1 and 2 at the end of the report for 1998 and 1999 error rates and liability and enhanced funding amounts.)

Most states with high or rising error rates feel strong pressure to improve their performance. Because QC sanctions are triggered whenever a state's error rate exceeds the national average, by definition about half of the states will be subject to sanction each year. Furthermore, because the sanctions are based on a sample of cases, each state's error rate is subject to statistical fluctuation. The margin of error around most states' error rate is between one and two percentage points.⁽⁴⁾ In other words, a state with an error rate just below the national average faces a non-trivial chance that, in the following year, its error rate will rise above the national average simply because of an unlucky sample. In practice, this means that over 30 states each year are either subject to or at risk of QC sanctions. In fact, over the years 1995 through 1999, 38 of the 53 jurisdictions that participate in the Program had at least one year where their measured error rate exceeded the national average.

Moreover, a state cannot know until the end of the year how low its error rate needs to be to avoid sanction. If a few large states lower their error rates substantially and thereby reduce the national average, or if a state's sample randomly contains more cases with an error, a state that has consistently performed reasonably well may unexpectedly face a sanction.

Also, because the QC system offers enhanced funding for states with error rates below six percent, a number of states not at immediate risk of sanctions nonetheless are pressing to lower their error rates further to qualify for these payments. For 1999, six states are slated to receive $39 million in enhanced funding. Six other states had combined payment error rates between six and eight percent, giving them realistic prospects of achieving enhanced funding in the near future. Between the states potentially subject to liabilities and those aiming for enhanced funding, the overwhelming majority of states have strong incentives to seek policies that will lower their error rates.

The Food Stamp Act grants USDA broad flexibility on how much of the sanctions it will actually collect. Although automatic fiscal penalties for states with high QC payment error rates began in 1982, Congress waived virtually all sanctions accrued through fiscal year 1985. In 1993 USDA negotiated a large settlement agreement with the states that stipulated that USDA would waive 85 percent of the sanctions accrued over 1986 through 1991 and states would reinvest the remaining 15 percent in management improvement activities designed to reduce errors in the future. Since the early 1990s, USDA has offered states the alternative between paying the entire sanction amount and entering into agreements under which some of their potential liabilities are waived and some must be reinvested.

Although sanction amounts that are reinvested are not paid directly to the federal government, from the state's point of view reinvestment represents an appropriation of state funds it likely would not have made otherwise. States must expend the reinvested funds on error reduction activities USDA approves, such as additional administrative staff or staff training. Unlike most other administrative activities, where states receive a 50 percent match from the federal government, states do not receive any federal match for reinvestment activities. According to USDA, states have reinvested about $140 million since 1986.

Another strategy that USDA has used as a feature of liability settlement agreements with states is to enter into "at risk" agreements. Under an "at risk" agreement, a portion of a state's liability from one year is held "at risk" and is subject to collection or reinvestment in the future if the state does not meet a stipulated error rate target. Beginning in the mid-1990s, USDA began to cut back on the amounts of penalties it would waive outright, making states "reinvest" a larger portion of their potential liabilities and holding a share "at risk." For example, a state that was subject to a $500,000 penalty might reinvest $200,000 in the first year and have $100,000 potentially subject to collection or reinvestment in each of the next three years, depending on whether it met a designated error rate target. In settling the 1992 through 1997 liability amounts, USDA's placed $145 million "at-risk" in future years. For 1998 and 1999 liability amounts, however, USDA entered into only one new "at risk" agreement each year.

State administrators must justify to their governors and legislatures their requests for funds that need to be reinvested or paid to the federal government. Many find the task difficult, as it is often perceived as a negative reflection on their performance. In addition, because of the existing "at risk" agreements, some states that do not reach the ambitious error rate targets that they negotiate with USDA's regional offices may face large accumulated penalties.

USDA, states, and non-profit groups have recently begun to raise concerns that state efforts to reduce error rates rapidly can have an adverse effect on food stamp participation. Between 1994 and 1999, eight states reduced their error rates by four or more percentage points, which represented very large reductions, considering that the national average error rate is around ten percent. Food stamp participation in these states — Arizona, Florida, Indiana, Mississippi, Ohio, Texas, West Virginia and Wyoming — declined by 44 percent over those years, compared to a 30 percent participation decline in the remaining states. Other states achieving more modest declines in food stamp error rates did not have unusually sharp declines in participation.

One of the policies that a number of states have implemented to reduce error rates, often at USDA's recommendation, is to require "error prone" households, particularly households with earnings, to come to food stamp offices every three months to reapply. The states and USDA reason that the more often families with fluctuating circumstances have to fill out a new form and submit to a certification interview, the more likely that the QC reviewer will find that the eligibility and benefit determination for the family was based on accurate information if the case is drawn in a QC sample. Frequent visits to the welfare office can, however, be very inconvenient for households with earnings because many states assign interview times without regard to the household's work schedule, and most local welfare offices are not open in the evenings or on weekends. A USDA survey of food stamp recipients in 1996 found that the average applicant spent five hours applying for food stamps initially, and two to three hours applying for recertification.⁽⁵⁾

Between 1994 and 1998, several states sharply increased the proportion of working families with children required to reapply every three months. Nationally, data gathered through the food stamp QC system show that the proportion of working families with children required to come into food stamp offices at intervals of three months or less more than tripled between 1994 and 1998, rising from nine to 31 percent. Nine of the 11 states that increased the proportion of working families required to apply every three months by at least 50 percentage points experienced declines in participation for working families that exceeded the national average for this period. In addition, the five states with the largest declines in food stamp participation among working families between 1994 and 1998 all dramatically expanded their use of three-month certification periods during those years. (See tables 3 and 4.)

For many years staff in the USDA national and regional offices encouraged states to adopt three-month certification periods as an error-reduction tool. Some evidence that the agency may be backing off this policy can be found in the policies in the July 1999 guidance and in an Access Guide USDA issued last summer on serving working families.⁽⁶⁾

Barriers to participation are not limited to repeated, often lengthy, office visits that are scheduled during working hours. For example, some states call a worker's employer every three months to verify the employee's earnings. Low-wage workers who are concerned this may alienate their employers and endanger their jobs may conclude they do not wish to continue receiving food stamps. In addition, many states have phone systems that make it difficult for households to reach their eligibility worker, or otherwise have unreliable procedures for households to mail in or drop off verification or information about a change in their circumstances. Lack of adequate mechanisms in these areas can necessitate more trips to the welfare office by families. Also, some states have implemented overly-aggressive anti-fraud policies. Examples include fingerprinting all applicants and conducting unannounced home visits for a significant share of applicant households. Data are not available to measure the effects of these error-reduction policies and other obstacles in discouraging food stamp participation among low-income working families.

Table 3:
States that Increased Use of Three-Month Certification Periods for Food Stamps by at Least 50 Percentage Points
State	Proportion of Working Families with Children Required to Reapply Every Three Months		Change in Participation Among Working Families 1994 to 1998
State	1994	1998	Change in Participation Among Working Families 1994 to 1998
Arizona	1%	64%	-28%
Georgia	5%	62%	-1%
Indiana	1%	54%	-43%
Maine	24%	86%	-22%
Missouri	2%	92%	-22%
North Carolina	1%	76%	2%
Ohio	1%	78%	-30%
Oklahoma	11%	84%	-21%
Texas	3%	62%	-34%
Virginia	3%	71%	-13%
Wisconsin	1%	65%	-6%
U.S.	9%	31%	-5%

Table 4:
States with the Five Largest Declines in Food Stamp Participation Among Working Families, 1994-1998
State	Proportion of Working Families with Children Required to Reapply Every Three Months		Change in Participation Among Working Families 1994 to 1998
State	1994	1998	Change in Participation Among Working Families 1994 to 1998
Indiana	1%	54%	-43%
Texas	3%	62%	-34%
Ohio	1%	78%	-30%
Arizona	1%	64%	-28%
Utah	15%	50%	-26%
U.S.	9%	31%	-5%

When it determines the amount of liabilities it will assess, USDA may adjust the liability amounts downward to account for mitigating circumstances that may have caused an increase in a state's error rate. Last fall, in recognition that the emphasis on QC may conflict with the goal of serving working families effectively, USDA used this discretion to adjust the sanction amounts downward for those states serving relatively high or increasing shares of working households and households with immigrants.⁽⁷⁾ See Box 1 for a discussion of why households with earnings or immigrant members are more error prone and for detail on how USDA adjusted the liability amounts.

In 1998, every state had a higher error rate for households with earnings than for households without earnings, and almost every state with a significant immigrant population had a higher error rate for households with immigrant members. Because sanctions are determined based on a state's error rate compared to the national average, states that serve relatively more households with earnings or immigrants will tend to have higher error rates. This means that, in the absence of adjustments, states with strong welfare-to-work programs, states with state-funded food stamp replacement programs for immigrants,⁽⁸⁾ and states that maintain effective access for the working poor and immigrant families, will be at greater risk of fiscal sanction. To counteract this inequity, USDA's adjustments lower liability amounts to hold states harmless for the extent to which they have a growing share of households with earners or immigrants, or a share of such households that is above the national average.

Box 1
Detail on USDA's Liability Adjustments for Households with Earnings or Immigrants

Households with earnings. States have historically had higher error rates for working households because of the difficulty tracking earnings and correctly adjusting benefits when earnings fluctuate month to month. Nationally, in fiscal year 1998, the reported error rate was about six percentage points higher for working households than for those with no earned income (14.93 percent compared to 8.71 percent).⁽⁹⁾

Households with immigrants. For legal immigrants, by contrast, the higher tendency for errors appears to be a result of the increased complexity of eligibility determinations and benefit calculations introduced by changes in the 1996 welfare law. Prior to the 1996 legislation, most low-income legal immigrants were eligible for food stamps, and 64 percent of food stamp households with immigrant members also contained at least one U.S. citizen member. The welfare law eliminated food stamp eligibility for many legal immigrants. The new rules for determining whether a legal immigrant is eligible for food stamps are exceptionally complex. As a result of the change, many more households now contain both eligible and ineligible members. States must distinguish between the two types of members and handle the incomes and expenses of these households accordingly. States with food stamp replacement programs for legal immigrants face additional complexity in applying separate federal and state eligibility rules. Not surprisingly, states' reported error rates for households with at least one immigrant member climbed from 10.2 percent in 1996 to 16.6 percent in 1998, an increase of 63 percent. Nationally, states had a 16.6 percent reported error rate for households with immigrants, compared to 9.7 percent for households without immigrants.

USDA's liability adjustments. USDA's liability adjustments protect a state if its share of working poor or immigrant food stamp households exceeds the national average or if the proportion of such households receiving food stamps in the state (as a share of the state's total caseload) increased since fiscal year 1996, the year before the welfare law took effect. To calculate an adjusted error rate for 1998 for a given state, USDA used the state's 1998 error rates for working poor and immigrant families, but assumed that the shares of the caseload made up by working poor families and immigrants, respectively, were either the same as the national average or the same as in 1996, if such an assumption would lower the state's error rate. The state was subject to sanction if its error rate as adjusted in this fashion exceeded the (unadjusted) national average.

In addition, USDA removed from the liability calculations instances where the QC reviewer found the benefit to be off by $25 or less. This change recognizes that a state that serves an increased number of working families is likely to make more small errors as a result of fluctuating circumstances, and that these small errors do not threaten the integrity of the program. As a result of this adjustment, the average state's error rate was lowered by about two-thirds of a percentage point.⁽¹⁰⁾ USDA made the same set of adjustments for the working poor, immigrants, and small errors when it released fiscal year 1999 error rate and liability figures in April 2000.

As a result of the adjustments, of the 22 states originally subject to sanction in 1998, six states had their liabilities removed entirely, and the others' liabilities were reduced significantly. The total amount of sanctions across states with error rates above the national average declined from $78 million to $27 million. Similarly, in 1999, USDA's adjustments completely eliminated sanctions in eight of the 24 states subject to sanction, and the total sanction amount shrank from $75 million to $31 million.

USDA's liability adjustments were well-received by states. They were taken as a sign that the Administration is serious about increasing access to the food stamp program and about rewarding (or at least not penalizing) states that make efforts to serve the working poor. States should now be considerably less hesitant to take actions that would improve access to the working poor and immigrants because of the protections these adjustments provide.

The food stamp program's eligibility and benefit rules are set by statute or regulation and are largely uniform nationally. The states have considerable latitude, however, in how they set up their certification procedures. In determining the accuracy of the food stamp benefit, the QC reviewer takes into account both the eligibility and benefit rules and the state's certification policies.

To receive food stamps, a household must submit an application to a local food stamp office, be interviewed about its income, resources, and other circumstances, and provide verification (typically documentation) of many of these factors. If the state determines that the household is eligible, it issues benefits and assigns the household a fixed "certification period." The certification period is the length of time the household may receive food stamps before having to reapply (assuming it remains eligible over the period). States may set certification periods as short as one month under some circumstances, or, for most working families, as long as a year.⁽¹¹⁾ At the end of the certification period, to continue its food stamp benefits, the household must reapply, ordinarily must submit to a new interview, and must again provide extensive verification of its circumstances.

Over the months within the certification period, the household is responsible for making reports about its income and other circumstances. The state agency must adjust monthly benefit amounts in response to these reports or terminate food stamp benefits if the household becomes ineligible. As discussed in detail below, federal rules offer states several options for the reporting system they use. States may require different types of households to report under different systems. The food stamp office is responsible for informing a household about whatever reporting requirements it has decided to impose.

In conjunction with these certification and reporting systems, states have two options for how they determine eligibility and the level of food stamps for the household each month. They may use retrospective budgeting — where eligibility and the level of food stamps in a given month is based on the actual income and other circumstances reported for a previous month. Or states may use prospective budgeting — where eligibility and the level of food stamps in a given month is based on that month's anticipated income and other circumstances.⁽¹²⁾ Under prospective budgeting, income is taken into account only if "the household and state agency are reasonably certain [it] will be received."⁽¹³⁾

The choices a state makes about what certification, reporting, and budgeting procedures to use have several important ramifications. First, requiring frequent reapplication, interviews, or frequent reports of circumstances can be burdensome on both households and the state. A household that is required to appear at the food stamp office or submit verification frequently may decide that the costs of participation outweigh the benefits. Between filing the application, appearing for the interview, and dropping off needed verification, a household typically makes two to three visits to the welfare office to complete the certification or recertification process. Frequent paperwork requirements also increase the likelihood that eligible households will unnecessarily lose benefits because they fail to comply fully with cumbersome requirements. From the states' perspective, requiring households to maintain frequent contact, either through recertification or frequent reports, substantially increases the workload for eligibility workers.

Second, the recertification and reporting rules that a state adopts, along with the related budgeting rules, influence how well-aligned the food stamp benefit for a given month is with the household's actual circumstances in that month, and therefore how well the amount of assistance meets the household's need. In instances where the household's earnings or expenses fluctuate modestly from one month to another, it may not be of great concern if its food stamp benefits for a month do not correspond perfectly to its income in that month. If, however, a member of the household becomes unemployed, starts a new job, or experiences another major change in circumstances, it may be more crucial to adjust the food stamp benefit quickly to correspond with the household's new situation.

Third, the food stamp QC system considers as errors only discrepancies resulting from the eligibility worker mishandling known information or from a household's failure to comply with requirements to disclose information or report circumstances. If a household does not report a change in its circumstances, an error can result only if the state's rules required the change to be reported. At initial certification and recertification the household must report on all elements of eligibility. During a certification period, however, states may limit the information that households must report. A state that requires fewer items to be reported lowers its exposure to errors resulting from both a household's failure to report and the agency's failure to act correctly on a report. Thus, as discussed in detail below, by adopting procedures that require less frequent actions on the part of households (and that are easy for households to understand and comply with), a state can lower its QC error rate without resorting to measures that also tend to limit access to the program for needy households.

States currently have a variety of options regarding how households must report changes in their circumstances during a certification period. The two basic methods, which are described in the food stamp regulations, are monthly reporting and change reporting. The variations, which USDA has approved under waivers, are: quarterly reporting, which is a variation on monthly reporting; and three variations on change reporting — status reporting, the $80 (now $100) reporting option, and the 5-hour reporting option. See table 5 for information on what options states have chosen thus far.

Under a monthly reporting system, a household must submit a report on its circumstances and provide required verification each month even if nothing has changed. A subsequent month's eligibility and benefit level is usually based on the information for the earlier month contained in the monthly report. (In other words, the state uses retrospective budgeting.) For example, a household's benefits for May will typically be based on its income and other circumstances in March, as reflected on a monthly report form due with corresponding pay stubs in early April.

There are certain households that states may not require to report monthly. These households include migrant or seasonal farmworker households, households in which all members are homeless, and households without earnings in which all members are elderly or disabled.⁽¹⁴⁾ If a state chooses to require other households to report monthly, it cannot assign certification periods shorter than six months to such households without applying to USDA for a waiver of the rule.⁽¹⁵⁾ A household required to report monthly cannot be subjected to any other reporting requirements.⁽¹⁶⁾

Table 5:
State Reporting Waivers as of July 2000
Quarterly Reporting	Status Reporting	5-hour Reporting	$80 or $100 Earned Income Reporting	No Reporting Waivers
Alabama	Delaware	Maryland*	Arizona	Alaska
Arkansas	Florida	Michigan	Dist. of Columbia	Colorado
California¹	Georgia	Oregon*	Illinois*	Connecticut
Illinois*	Idaho	Vermont	Indiana	Guam
Louisiana*	Indiana		Iowa	Hawaii
Massachusetts	Missouri		Kansas	Maine
New York*	Nebraska		Kentucky	Minnesota
South Carolina*	Nevada		Louisiana*	Mississippi
Tennessee*	New Hampshire		Maryland*	Montana
Washington*	New Jersey		New York*	Pennsylvania
Wisconsin*	New Mexico		North Carolina	South Dakota
Wyoming*†	Ohio		North Dakota
	South Carolina*		Ohio
	Tennessee*		Oklahoma
	Texas		Oregon*
	Utah		Rhode Island
	Virginia		Virgin Islands
	Vermont		West Virginia*
	Washington*
	West Virginia*
	Wisconsin*
	Wyoming*
Notes: * State has more than one waiver, either because it has different policies for different types of households or because an earlier waiver (that is still in effect) has been superceded. † USDA reports that a waiver request is pending. ¹ California has an unusual waiver for a limited number of counties that, among other things, combines quarterly reporting and status reporting. Source: U.S. Department of Agriculture

Under the food stamp program's regulations, states may decide what information is included on the monthly report form and what verification must accompany it.⁽¹⁷⁾ When the state agency receives the monthly report, it determines if all necessary information has been provided, adjusts the household's future benefits to account for the income and circumstances reflected on the monthly report, and mails the household a new report form for the following month. If a household does not submit its monthly report or provides incomplete information, the state sends a notice informing it again that it must report on its circumstances by a date late in the month.⁽¹⁸⁾ If the household fails to comply by this extended deadline, its food stamps are suspended or terminated.⁽¹⁹⁾

For most of the 1980s, monthly reporting was mandatory for households with earnings or a recent work history, and states commonly required it of other households as well. In 1988, Congress repealed the mandate, and over 30 states have since dropped it. A handful of states, including California, Pennsylvania, and Mississippi, continue to require a large portion of their food stamp households to submit monthly reports. Studies of monthly reporting in the Aid to Families with Dependent Children (AFDC) program and food stamp program in the 1980s found that, contrary to expectations, monthly reporting did not lower the program's benefit costs or error rates and, in fact, increased administrative costs.⁽²⁰⁾ A major factor was the increased workload demands on welfare office staff associated with processing monthly reports. For many years, USDA has encouraged states to drop monthly reporting. In a recent waiver approval for parts of California, USDA noted:

We are very pleased to see that California is exploring options to its [monthly reporting/retrospective budgeting] system. We have long supported prospective budgeting as a more friendly budgeting system for working families and a simpler and less error prone system administratively for states.⁽²¹⁾

Under a quarterly reporting waiver, states follow the monthly reporting model but require households to file the report and related verification every three months instead of every month.⁽²²⁾

The state uses the information in the quarterly report to determine the household's eligibility and benefits over the subsequent three-month period. Households are not required to report any changes in their income or other circumstances between quarterly reports, although if they do report a change, the state agency must act on it.

States may choose to determine a household's food stamp benefits using either retrospective or prospective budgeting. In other words, benefits for the upcoming quarter can be based either on actual circumstances in the prior quarter or on anticipated circumstances for the upcoming months (as projected from the information in the quarterly report). To date, all states that have implemented quarterly reporting under a waiver have opted to use prospective budgeting. A state that uses prospective budgeting would require the household to report (on the quarterly report form) on any changes in its income and circumstances that are anticipated to occur over the succeeding three months. For example, if the household knew a family member was moving out to go to college, or one of the earners' jobs was scheduled to be suspended temporarily because of school summer vacation, this information would need to be reported. Changes that cannot reasonably be anticipated, such as reduced work hours due to illness or increased hours due to an unanticipated labor shortage, are not reported and would not affect benefits unless the household chose to report the change during the quarter.

States also have broad flexibility as to whether they require households to report on one or more months from the previous quarter on the quarterly report form. A state that requires a household to report on all three months of the quarter will have more complete information on which to base the succeeding quarter's benefits, but places a substantial burden on households to remember each month's circumstances and retain and submit the required pay stubs and other verification. If the household forgets or is confused about pertinent facts, an error may result. Such an approach also places burdens on eligibility workers to process substantial paperwork and verification.

States may instead require households to report on only one month's income and circumstances. For example, under Illinois' quarterly reporting waiver, if a household were participating on a January-to-March cycle, the state would send the household a new quarterly report form late in February and ask it to provide information and verification for February's income and circumstances, as well as any anticipated changes for March to June. The report would be due on March 7, and the state would use it to provide benefits for April through June. This would be the household's only reporting requirement until the state sent it another report form in late May.

In the early 1990s, USDA approved quarterly reporting waivers for five states: Alabama, Arkansas, New York, Oregon, and West Virginia. (West Virginia dropped its waiver relatively early. Oregon's limited quarterly reporting waiver expired in 1999.) In the late 1990s, when food stamp participation nationally was falling rapidly, with reporting and recertification requirements some states had imposed to address QC error rate concerns apparently playing a part, the four states that retained their quarterly reporting waivers experienced relatively modest participation declines. This suggests that quarterly reporting may be less burdensome for working households than some of the alternatives states have used in recent years to respond to QC error rate pressures. Nonetheless, in 1995 the Office of Management and Budget prohibited USDA from approving additional quarterly reporting waivers because of cost concerns.

As part of his July 14, 1999, food stamp initiative to address barriers for working families, President Clinton announced that USDA would again approve state requests to implement quarterly reporting of earnings for households with earned income. In later guidance provided to states in October 1999, USDA broadened the option so that in implementing a quarterly reporting system for earners, states could include items other than earnings on the quarterly report, thus rendering the quarterly report the only reporting requirement for such households. In addition, USDA has allowed states to consider as earners anyone with earnings in the past three months and does not require states to remove households from quarterly reporting during a certification period if the household loses its earnings.

USDA has not specified exactly what items of eligibility must be reported on the quarterly report, but staff have indicated that the six items that must be covered in change reporting systems by federal regulation⁽²³⁾ (discussed below) are both necessary and sufficient. USDA also has required that under quarterly reporting, states must provide the same warning notices and termination notices that are required under monthly reporting. The July 1999 USDA guidance contains a list of questions that states need to answer in submitting their waiver requests for quarterly reporting. This list indicates that USDA is willing to give states considerable flexibility in designing quarterly reporting systems that best meet their needs and concerns. As of July 2000, about a dozen states have adopted quarterly reporting (see table 5).

If a state does not require a household to submit monthly or quarterly reports, it must have a change reporting system in place.⁽²⁴⁾ Under change reporting, a household's food stamp benefits for a given month are usually based on the household's anticipated circumstances over that month. The eligibility worker uses the information obtained at the time of the household's most recent certification to anticipate the household's future income and circumstances. To adjust the food stamp benefit for changes in the household's circumstances, the state instructs the household that it must report certain changes within 10 days of the change. Food stamp regulations require households to report the following information during a certification period unless the state has a waiver modifying these rules:⁽²⁵⁾

Households may report other changes, and state agencies must adjust benefits accordingly. States must provide forms on which households can submit changes in writing, but they also must act on changes that are reported over the telephone or in person.⁽²⁷⁾ States must, within 10 days of receiving the household's report, determine if the change affects the household's eligibility or benefit amount. Changes that would lower benefits, such as an increase in income, need not be verified, but the household must provide proof of changes that would increase benefits, such as the loss of a job. If the state reduces or terminates the household's food stamp benefits, it must send a notice alerting the household of the action.⁽²⁸⁾ The household has 10 days to request a fair hearing appealing the change before the change goes into effect.⁽²⁹⁾

Under waivers, many states have adopted one of three variations on change reporting to limit the frequency of required reports for households with fluctuating wages and to simplify reporting requirements so they are easier for households to understand and less error-prone for states. As discussed above, under normal change reporting rules, a household must report any change in monthly income greater than $25. A person working at the minimum wage would have to work only five extra hours a month to trigger the obligation to report a change in income. With the change reporting variations, all of the reporting requirements are the same as under change reporting, except that a household is not required to report modest variations in earnings. Instead, it need only report to the state agency if it experiences a significant change in its employment status.

For several years, USDA has relied upon procedures set out in its December 17, 1996, proposed rule on change reporting in acting upon requests for such waivers.⁽³⁰⁾ That rule would require a change reporting household to report changes in the source of earnings (i.e., starting, stopping, or changing jobs) and one of the following, as selected by the state:

USDA's July 1999 guidance made the options more attractive for both households and states by:

According to USDA, as of the summer of 2000, about two-thirds of the states have one of the three new variations on change reporting for at least part of their caseload.

Recently, USDA began approving waivers for states to interview households once every twelve months, even if they are required to reapply more often.⁽³²⁾ States that choose to have short certification periods for working families because of their fluctuating circumstances may now choose to conduct the recertification through the mail or by telephone, rather than requiring a household member to come into a food stamp office for an interview each time the household is recertified. For example, a state could use three-month certification periods but require the household to come in for the interview every other, or every third or fourth, certification. Such a state could conduct the recertification interview by telephone or could eliminate the interview altogether (except for once a year) and instead collect all necessary information by mail.

The change can substantially lower the cost of food stamp participation on working households because they have to take time away from work to be interviewed much less often. It also can lessen the burden on state eligibility workers, as scheduling and conducting certification interviews is one of their most time-intensive duties. USDA reports that 15 states have received approval to waive some face-to-face interviews for certain categories of households.⁽³³⁾

The combination of stepped-up pressure from USDA on QC error rates and the increased earnings among food stamp-eligible households as a result of the strong economy and the 1996 welfare changes has caused many states to seek methods to lower their exposure to QC errors. The common strategy for reducing errors for many years, and one that USDA advocated, has been to require households to submit recertification forms and appear at the office for interviews more often. The new reporting options offer an alternative. Under these options, states are not held liable for small changes in earnings. This should allow states to return to or maintain longer certification periods without exposing themselves to QC errors.

The goal of the QC review is to determine whether the state issued the correct food stamp benefit in a given month, given the program rules and the various options the state has adopted. The QC reviewer conducts a thorough investigation of the household's actual income and circumstances during the review month and compares the benefit calculated based on that information to the benefit the household actually received. The procedures governing the QC reviews require, however, that some differences that a QC reviewer uncovers in the course of the review not be counted in the error rate calculation.

Perhaps most importantly, the state is not liable for errors if the household experienced a change in its circumstances that it did not report if the state's policies did not require the household to report the change. QC reviewers determine whether the eligibility worker used the correct budgeting system and reporting requirements for a sampled household, given the state's rules. If the eligibility worker used the correct procedures then, according to the QC handbook, "any change in circumstances that is not required to be reported cannot [be] included...in the QC error determination process if it was not reported."⁽³⁴⁾

For example, under a quarterly reporting system, if a household is required to report only once every three months, errors cannot result from unreported changes in the household's circumstances that occur between quarterly reports. Similarly, under status reporting, where the state limits the reporting requirement to changes in the source of income, wage rate, or full-time part-time status, any unreported changes arising only from a change in the number of hours worked would not be counted as an error. Thus, by reducing the frequency of reports, the new reporting options can significantly reduce states' exposure to QC errors for households with variable income for two reasons. First, because fewer changes are required to be reported, there are fewer opportunities for unreported changes to result in errors. Second, if welfare offices receive fewer reports, eligibility workers ought to have more time to process accurately the reports they do receive and to attend to their other duties.

Furthermore, any changes in elements of eligibility that are not included on the report form cannot contribute to errors because, again, the household was not required to report these changes. For example, consider a household that qualifies for and receives the child care deduction at the beginning of its certification period, but some months later no longer qualifies for the deduction because of a change in child care arrangements. Assuming the state does not require that changes in child care status be reported, it would be appropriate for the household to continue receiving the child care deduction until the end of the current certification period if it did not voluntarily report the change. The QC reviewer would not consider the use of the child care deduction in calculating the household's benefits to be an error.

For this reason, there may now be QC advantages for states to have longer certification periods. At initial certification and recertification, states must collect information on all elements of eligibility and must verify many elements. This can occur on the application, during the interview, or through subsequent communications from the household. Any information that the household fails to disclose, or that the eligibility worker fails to consider correctly, would contribute to an error. During the certification period, however, the state may limit the items of eligibility on which the household must report changes and provide verification. As discussed earlier, for change reporting and related variations, the food stamp regulations are very clear about the six elements of eligibility that states must include on their report form. Many of the states using quarterly reporting have developed forms that include a similarly abbreviated list of eligibility items. States are liable for errors that result from changes that occur during the certification period under two circumstances: if the household fails to report something that it was required to report, or if the eligibility worker fails to determine benefits correctly based on the information the household provided. Longer certification periods and fewer items on the report form can reduce the likelihood of both types of QC errors.

Consider a state that requires working households to reapply every three months. The household must submit a new application each time. Any items of eligibility that are not disclosed, or that the eligibility worker mishandles, can result in an error. By contrast, if a state has quarterly reporting and includes on the report form only the six items that are required by regulation to be included on a change report form, the state is not liable for any changes that have occurred since initial certification in other items of eligibility.

According to data from the QC system for 1998, in 67 percent of the cases with errors, the differences that led to the errors were classified as "agency" errors as opposed to "client errors." In other words, two-thirds of the differences were caused by the agency mishandling known information rather than the QC reviewer discovering information that the household had not disclosed or reported.⁽³⁵⁾ Moreover, over half of the differences in error cases were discovered by the QC reviewer from information that was in the eligibility worker's case file, as opposed to information that the QC reviewer unearthed in a field investigation.⁽³⁶⁾ Thus, it appears that some states could dramatically reduce their errors by increasing the amount of effort that eligibility workers spend on processing the information they obtain at initial certification or recertification and then limiting the items that must be reported over a six- or twelve-month certification period.

In addition, the QC review process requires that the time frames necessary for administrative processing be taken into account when QC reviewers determine whether a difference should be treated as an error. In general, under change reporting, households have 10 days to report the change, the state has 10 days to make a decision and inform the household of the decision, and the household has 10 days to appeal the decision.⁽³⁷⁾ To accommodate these timeframes, states are not held liable for unreported changes that the household becomes aware of within 30 days of the QC review date. The QC review date is usually the first day of the month for which benefits are being reviewed.

Assume, for instance, that May is the sample month for QC and, using status reporting, the state determines May's benefits prospectively based on the information it has about a household's circumstances through the end of April. If a household member was hired in a new job on April 10, the QC reviewer would not consider the case to be in error for May if the household did not report the change. A new job is a change the household is required to report under status reporting rules, but because it occurred within 30 days before the QC review date the case would not be considered in error.

Similar error exclusions apply if the household does report a change that it became aware of shortly before the review date. In the case of changes that would lower benefits, the QC reviewer would not count an error if the household report occurred within 20 days prior to the review date. In the case of changes that would raise benefits, that window is only 10 days because, in these cases, there is no need for the household to appeal the determination. In the above example, if the household reported its new job on April 15 but the state failed to lower May's benefits, the QC reviewer would exclude the difference because it fell within the 20-day period (10 days for the state to act and 10 days for the household to appeal the benefit reduction).⁽³⁸⁾

As a result of this QC rule, states are not subject to QC errors from changes in a household's circumstances that occur shortly before the state issues benefits.

In addition to the QC advantages outlined above, the new reporting options that USDA now allows can improve households' access to the food stamp program and ease state agencies' workload pressures. As discussed above, the new options allow states to:

States have broad flexibility even within some of the new reporting options. Below are two model approaches to certification and reporting procedures for households with earned income. They represent the least burdensome combinations of all the options available under current USDA policy.

While these policies represent the least burdensome options, not all states will want to adopt all of these policies in their entirety immediately. States may want to make changes incrementally because altering certification, budgeting, and reporting systems carries its own inherent costs for both households and states. Almost every state could take advantage of the options to make improvements by granting longer certification periods, requiring fewer office visits, and minimizing reporting requirements.

For example, states that currently have monthly reporting with 12-month certification periods could move to quarterly reporting, keeping their 12-month certification periods, as Illinois has done. States with change reporting or one of its variations and three-month certification periods could waive interviews at all but one recertification per year and could adopt status reporting with six-month certification periods. Both changes would substantially lower the amount and frequency of contact between households and the state agency. Alternatively, such states could move to quarterly reporting with 12-month certification periods and eliminate any reporting requirements between quarterly reports. Under such a policy households would have to come in to the food stamp office only once a year and submit abbreviated reports and verification only four times a year.

Several states, including Massachusetts, South Carolina, Tennessee, and Wisconsin, have recently moved from status reporting with three-month certification periods to quarterly reporting without switching to longer certification periods. Many of these states already waive the face-to-face interview at some recertifications. They use the recertification application as their quarterly report form. These states would benefit from using longer certification periods with quarterly reporting because they would not have to include all of the elements of eligibility on the quarterly report form. As discussed above, at recertification, states need to collect information on all items of eligibility and ask households to provide verification for many elements. On a quarterly report, these states could ask about fewer items of eligibility. This would reduce the burden on households and eligibility workers while reducing the potential for errors.

The discussion below analyzes some basic trade-offs among the various options available to states from the state perspective and then from the household perspective. Because state circumstances and current policies vary widely, no one policy is likely to work for every state. Furthermore, the states' specific implementation practices will significantly influence the effectiveness of their policies from both the state's and the household's perspective.

The local welfare office is responsible for all the application and certification aspects of food stamp administration. Staff in welfare offices must perform a multitude of tasks, including: taking applications, screening for expedited benefits, scheduling and conducting certification interviews, collecting and analyzing verification, determining eligibility and benefit levels, informing households of their reporting requirements, receiving change reports (or periodic reports such as monthly or quarterly reports), acting on the reported information, and informing households of their recertification responsibilities. Some of these activities can be automated or otherwise aided by technology, but many of them require a substantial commitment of staff resources. The various policies states adopt determine how eligibility workers structure their time. In general, the new reporting options and the opportunity to waive the face-to-face interview should ease workload burdens for states.

Under a monthly reporting system, most states use 12-month certification periods, and thereby keep to a minimum the amount of time eligibility workers spend on recertification interviews and processing. Considerable administrative costs accrue under monthly reporting, however, as a result of the time eligibility workers must spend processing the monthly reports. Every household the state requires to report monthly must file the report even if there is no change in its income or other circumstances. Depending on the degree of automation, states can also incur administrative costs from issuing reminder notices and notices of adverse action when the monthly report is not filed on time and from tracking the household's response to know whether or not to terminate benefits. States have indicated that monthly reporting often increases workloads because states frequently must take new applications (and conduct certification interviews) when a needy household's benefits are terminated for failure to file the report but the household subsequently reapplies for food stamps.

Compared to monthly reporting, quarterly reporting dramatically eases the eligibility worker's workload because the office receives, on average, one-third as many reports each month. The eligibility worker's workload can be further limited if the state adopts a quarterly reporting policy that requires the household to report on only one month of circumstances on the quarterly report, rather than two or three months.

Moreover, under quarterly reporting there are no other reporting requirements between quarterly reports. In addition to the QC advantages of limiting the number of required reports discussed above, under quarterly reporting each case demands very little attention from the eligibility worker in months other than the month in which the report is due. There will, however, be instances in which the eligibility worker must process a form for a household that experienced no change. And, as with monthly reporting, terminations of benefits for failure to provide the quarterly report may lead eligible households to lose benefits and subsequently reapply.

Like quarterly reporting, the new variations on change reporting limit the number of reports the state receives from working families. Instead of having to adjust benefits for a change in earnings as low as $25 a month, the state can inform households that they only need to report significant changes in their employment or earnings. Under status reporting, no reports would be required for changes in earnings that arise from fluctuations in the hours worked (rather than from a change in the job or in the wage rate). On the one hand, this should substantially reduce the number of reports the state receives, easing burdens on both households and food stamp offices. On the other hand, the timing of change reports is typically less predictable than the timing of quarterly reports, and the household still has to report changes in employment and wage rates, household composition, and a few other factors of eligibility other than earnings. In addition, under change reporting, eligibility workers will likely receive more oral reports of changes over the telephone or in person. To ease the workload on eligibility workers, some states have dedicated separate telephone lines and staff to receive change reports and forward the information to the appropriate eligibility worker.

In lieu of time spent processing periodic reports, states that use change reporting often spend more time on tasks related to recertification. Until its July 1999 guidance, USDA required states to use three-month certification periods for households covered by status reporting waivers. USDA now allows six-month certification periods for these households. Even where USDA does not require short certification periods, many states use them in conjunction with change reporting out of concern that households will not report changes and that the state will therefore be liable for QC errors. By limiting the number of required reports, however, the new change reporting alternatives should ease this concern and allow states to move to longer certification periods. In addition, USDA's new policy to allow waiver of the face-to-face interview at recertification except for once every 12 months allows states to cut back on the amount of time eligibility workers spend scheduling and conducting in-person certification interviews.

All of the policies that lower burdens on state welfare offices also will help to reduce the direct and indirect costs to working households of participating in the program. In particular, policies that require less frequent reports and fewer visits to the welfare office will benefit working families, whose schedules may make it difficult for them to comply with procedural requirements. States can further increase access for working families by designing short, clear applications, report forms, and notices, educating households about their reporting requirements, and allowing sufficient time for households to complete the forms and gather the required verification.

Quarterly reporting has the advantage for households that the requirements on when and what to report to the food stamp office are laid out clearly in the report form mailed to the household each quarter. In addition, the quarterly report form is the only reporting requirement. (Households may, however report changes in their circumstances if they choose; this opportunity protects them from losing out on food stamp benefits if they suddenly lose their jobs.) In some state-administered health insurance programs, the state agency provides the household with a pre-printed form at redetermination that contains the most recent information the state has about the household's relevant income and other circumstances. States could take this approach in food stamps for recertification applications and for quarterly reports.

Under change reporting, by contrast, the household must remember what changes trigger the requirement to report. For example, if a state uses change reporting with the $100 threshold, the household needs to know the level of income the eligibility worker anticipated for a given month so it can judge whether the income it actually received is different by more than $100. In addition, households in some states may have difficulty reaching their eligibility worker by telephone to report changes.

The change reporting options do, however, have the advantage that households do not have to file reports during periods when they have not experienced a change. Also, under quarterly reporting, when a family fails to file a complete report with all required verification (or, in some cases, if the state fails to process the report in a timely manner), the family's benefits can be terminated even though its income and other circumstances still make it eligible. Some households whose benefits are terminated may not understand that they remain eligible and consequently may not reapply. Even a family that does reapply will lose benefits for one or more months and faces the additional costs of a new certification interview and collecting a new set of verification documents. These costs include lost time from work, transportation costs, and child care costs, among others.

Finally, to the extent that any of the new options reduce errors, both states and households will benefit from the corresponding reduction in overpayments and underpayments. When the state discovers that it has improperly issued (or failed to issue) benefits, it must recoup the amounts (or restore them) in later months' allotments. This can be confusing and burdensome for state agencies and households alike.

In many respects, the details about how states actually implement their certification procedures will have more of an impact on QC error rates, agency performance, and household access than any specific policy option the state selects. The wide variation among the states in terms of error rates, administrative costs, and participation rates among eligible food stamp households confirms the importance of state implementation. Important implementation issues include:

Finally, states may have different reporting and verification requirements for other programs, such as TANF, Medicaid, and child care subsidies, that are often administered with food stamps. States should conform the reporting requirements among the programs where possible but should not allow the stringency of food stamp reporting requirements to affect receipt of other important benefits. For example, if a state adopts quarterly reporting for food stamps, it should not adopt a policy (either deliberately, or inadvertently in the way it programs its computers) that would result in the loss of Medicaid coverage for a household that fails to provide the food stamp quarterly report. The food stamp quarterly report should not become a condition of eligibility for other programs. On the other hand, there may be instances where the state can use information provided on the food stamp quarterly report to update eligibility in another program, such as Medicaid. Using the food stamp information could allow the state to require fewer contacts with the household for maintaining Medicaid eligibility.

The pressure that states believe the food stamp QC system puts on them to impose rigorous reporting requirements, to demand exhaustive verification, and require households to reapply frequently for benefits is at odds with the goals of reducing the burdens imposed on eligible low-income households seeking to participate. The conflict appears to be most serious with regard to the working poor, whose incomes often fluctuate from month to month and who consequently tend to be targeted in states' "error-prone profiles" and subject to greater reporting, recertification, and verification burdens.

The reporting options that USDA announced last summer offer states a unique opportunity to restructure their systems in ways that can benefit both states and low-income families. States should think creatively about ways to implement these reporting options that will reduce paperwork requirements on both households and states and also will limit states' exposure to payment errors.

**Table 1: FY 1998**
FOOD STAMP PROGRAM ERROR RATES, POTENTIAL & ADJUSTED LIABILITIES & ENHANCED FUNDING
	FY 1998 OVER-PAYMENTS (%)	FY 1998 UNDER-PAYMENTS (%)	FY 1998 COMBINED PAYMENT ERROR RATE (%)	POTENTIAL LIABILITIES	ADJUSTED LIABILITIES	ENHANCED FUNDING
STATE
CONNECTICUT	10.34	2.79	13.13	$897,739	$185,787
MAINE	7.43	2.72	10.15
MASSACHUSETTS	4.96	2.51	7.46
NEW HAMPSHIRE	5.74	4.46	10.19
NEW YORK	8.61	4.33	12.93	$6,980,940	$643,333
RHODE ISLAND	4.66	2.37	7.03
VERMONT	10.56	2.69	13.25	$206,958	$61,895

DELAWARE	9.71	2.74	12.45	$97,387	$18,640
DIST. OF COL.	7.41	3.25	10.66
MARYLAND	11.56	3.84	15.40	$5,846,661	$2,483,950
NEW JERSEY	8.70	3.21	11.91	$534,114	$66,352
PENNSYLVANIA	7.42	2.43	9.85
VIRGINIA	6.83	4.30	11.13	$55,574	NONE
VIRGIN ISLANDS	4.41	2.15	6.56
WEST VIRGINIA	8.51	2.88	11.39	$102,793	NONE

ALABAMA	6.55	1.12	7.67
FLORIDA	8.47	4.47	12.94	$4,003,351	$444,817
GEORGIA	9.90	3.75	13.65	$4,411,896	$2,241,851
KENTUCKY	4.53	2.87	7.40
MISSISSIPPI	3.70	2.31	6.01
NORTH CAROLINA	7.92	2.86	10.78	$3,191	NONE
SOUTH CAROLINA	6.60	1.46	8.07
TENNESSEE	6.58	2.16	8.74

ILLINOIS	11.04	3.00	14.04	$8,861,495	$837,706
INDIANA	4.98	1.81	6.79
MICHIGAN	13.13	4.55	17.67	$26,820,117	$15,756,414
MINNESOTA	3.35	1.83	5.18			$4,494,636
OHIO	6.19	3.10	9.29
WISCONSIN	9.28	5.30	14.58	$1,841,666	$689,391

ARKANSAS	4.96	1.01	5.96
LOUISIANA	5.52	2.16	7.67
NEW MEXICO	7.80	2.85	10.64
OKLAHOMA	7.65	3.22	10.87	$6,998	NONE
TEXAS	3.82	1.45	5.27			$19,742,234

COLORADO	7.67	3.02	10.69
IOWA	10.02	3.35	13.37	$735,441	$10,485
KANSAS	8.03	3.08	11.10	$13,018	NONE
MISSOURI	6.73	1.57	8.31
MONTANA	5.29	2.04	7.33
NEBRASKA	12.51	4.18	16.69	$2,281,180	$1,065,184
NORTH DAKOTA	6.32	3.03	9.36
SOUTH DAKOTA	1.59	0.52	2.11			$652,879
UTAH	7.69	2.01	9.70
WYOMING	3.48	1.33	4.81			$557,462

ALASKA	11.82	2.37	14.19	$570,064	$402,237
ARIZONA	4.32	1.58	5.90			$298,096
CALIFORNIA	8.17	4.35	12.52	$6,381,443	NONE
GUAM	8.15	2.17	10.32
HAWAII	3.23	1.58	4.82			$1,700,458
IDAHO	6.12	4.33	10.45
NEVADA	6.25	2.62	8.88
OREGON	11.47	1.98	13.45	$1,408,931	$232,011
WASHINGTON	12.04	3.16	15.21	$6,103,713	$2,050,783

TOTAL	7.63	3.07	10.69	$78,164,669	$27,190,838	$27,445,765

1/ Amount of liabilities adjusted to take into account the effect of high and increased proportions of earners and immigrants in States' caseloads and by removing errors under $25.00 2/ As announced on 6/30/99. Changes made after that date, including a change in the national average to 10.70, are not reflected. Source: USDA Food and Nutrition Service.

Table 2:
FY 1999 FOOD STAMP PROGRAM ERROR RATES
POTENTIAL & ADJUSTED LIABILITIES & ENHANCED FUNDING
	FY 1999 OVER-PAYMENTS (%)	FY 1999 UNDER-PAYMENTS (%)	FY 1999 COMBINED PAYMENT ERROR RATE (%)	POTENTIAL LIABILITIES	ADJUSTED LIABILITIES	ENHANCED FUNDING
STATE
CONNECTICUT	10.90	3.00	13.90	$2,446,892	$1,145,059
MAINE	6.77	2.02	8.79
MASSACHUSETTS	6.96	2.38	9.34
NEW HAMPSHIRE	9.96	2.90	12.86	$276,354	$96,396
NEW YORK	6.41	4.05	10.47	$515,975	NONE
RHODE ISLAND	4.43	2.61	7.05
VERMONT	9.82	2.27	12.09	$169,525	$69,006

DELAWARE	11.12	5.79	16.92	$1,620,967	$653,497
DIST. OF COL.	9.51	2.61	12.12	$407,204	$61,426
MARYLAND	10.07	3.54	13.62	$3,359,723	$1,019,283
NEW JERSEY	9.72	3.21	12.93	$3,255,003	$1,303,366
PENNSYLVANIA	7.75	3.04	10.79	$590,210	NONE
VIRGINIA	7.95	3.90	11.85	$1,109,062	$50,411
VIRGIN ISLANDS	4.17	1.69	5.85
WEST VIRGINIA	7.04	1.84	8.88

ALABAMA	9.59	1.70	11.29	$697,144	$218,846
FLORIDA	5.79	3.64	9.43
GEORGIA	7.84	3.03	10.86	$499,289	NONE
KENTUCKY	5.57	2.15	7.72
MISSISSIPPI	3.35	1.56	4.91			$5,161,155
NORTH CAROLINA	6.35	2.90	9.25
SOUTH CAROLINA	4.32	1.46	5.79			$757,997
TENNESSEE	6.86	1.79	8.64

ILLINOIS	11.00	3.79	14.79	$18,717,453	$3,860,944
INDIANA	5.80	2.32	8.11
MICHIGAN	12.38	5.21	17.59	$30,975,382	$19,772,853
MINNESOTA	4.56	2.12	6.68
OHIO	6.34	2.09	8.44
WISCONSIN	9.59	3.83	13.42	$1,570,194	$606,446

ARKANSAS	3.69	0.85	4.54			$4,099,364
LOUISIANA	5.14	2.20	7.35
NEW MEXICO	7.99	2.41	10.39	$37,959	NONE
OKLAHOMA	9.05	2.83	11.88	$896,549	$119,443
TEXAS	3.24	1.31	4.56			$27,941,372

COLORADO	6.29	2.73	9.02
IOWA	6.30	2.97	9.27
KANSAS	6.62	2.36	8.98
MISSOURI	6.38	2.19	8.58
MONTANA	5.67	2.43	8.10
NEBRASKA	10.93	3.29	14.22	$1,261,115	$528,674
NORTH DAKOTA	6.18	1.84	8.03
SOUTH DAKOTA	1.94	0.25	2.19			$714,305
UTAH	8.05	4.50	12.55	$528,193	$141,100
WYOMING	1.76	1.15	2.91			$548,871

ALASKA	11.82	4.13	15.94	$1,817,221	$979,902
ARIZONA	4.88	2.05	6.93
CALIFORNIA	7.92	3.43	11.34	$3,896,171	NONE
GUAM	7.97	2.17	10.14	$2,136	NONE
HAWAII	5.37	1.45	6.82
IDAHO	6.76	4.18	10.94	$51,526	NONE
NEVADA	5.59	2.55	8.14
OREGON	8.14	2.36	10.50	$74,099	NONE
WASHINGTON	6.05	2.50	8.55

TOTAL	7.03	2.85	9.88	$74,775,344	$30,626,652	$39,223,064

1/ Amount of liabilities adjusted to take into account the effect of high and increased proportions of earners and immigrants in States' caseloads and by removing errors under $25.00 2/ 8.11 was Indiana's assigned error rate as of 4/27/00. This rate was changed to 6.79 per Secretary Glickman's June 20, 2000 decision letter Source: USDA Food and Nutrition Service, as of July 31, 2000.

1. U.S. Department of Agriculture, Food and Nutrition Service, Office of Analysis, Nutrition, and Evaluation, Who is Leaving the Food Stamp Program? An Analysis of Caseload Changes from 1994 to 1998, August 1999.

2. This paper presents a very broad overview of food stamp QC. Section 16(c) of the Food Stamp Act (7 U.S.C § 2025(c)) provides the statutory basis for the QC system. The food stamp regulations at 7 C.F.R. §§ 273.2(d)(2), 275.1 - 275.4, 275.10 - 275.14, and 275.21 and the Food Stamp Program Quality Control Review Handbook (FNS Handbook 310) provide much more detailed information on QC procedures. For a discussion of the strengths and weaknesses of the food stamp QC system and related penalties and incentives see Dennis P. Affholter and Fredrica D. Kramer, eds., Rethinking Quality Control: A New System for the Food Stamp Program, National Academy Press, Washington, DC, 1987.

3. The error amount is excluded if the QC reviewer determines that the household was eligible but the benefit was off by a small amount ($5 or less prior to fiscal year 2000, $25 or less beginning in fiscal year 2000).

4. For example, in 1998, Pennsylvania's combined error rate was 9.85 percent. According to USDA, the 95 percent confidence interval would be 8.32 percent to 11.38 percent. The national average combined payment error rate in 1998 was 10.70 percent. Thus, while the state's measured error rate fell below the national average, and the state was therefore not subject to sanction, the statistical uncertainty around the state's measured error rate is wide enough that the state cannot be assured that sampling variation alone will not cause its error rate to rise above the national average in the future even if its performance does not change. (See Food Stamp Quality Control Annual Report, Fiscal Year 1998, USDA, Food and Nutrition Service, Program Accountability Division, Quality Control Branch, February 2000, p. 67.)

5. U.S. Department of Agriculture, Food and Nutrition Service, Customer Service in the Food Stamp Program, July 1999.

6. U.S. Department of Agriculture, Food and Nutrition Service, The Nutrition Safety Net: At Work for Families, A Primer for Enhancing the Nutrition Safety Net for Workers and their Children, 1999.

7. USDA has not yet announced whether it will continue these specific adjustments for fiscal years after 1999, though they have indicated they intend to adjust liabilities in the future in ways that help ensure access. Also, the adjustments apply only to liability amounts. USDA does not apply the adjustments in determining whether a state qualifies for enhanced funding.

8. The federal QC system does not apply to benefits provided under state-funded programs for federally-ineligible immigrants. However, many immigrant households contain both federally-eligible immigrants or citizens and ineligible immigrants. The QC system applies to the benefits provided to the federally-eligible members of such a "mixed" household. A state that has a state-funded program for legal immigrants will likely have greater participation among "mixed" households than a state that does not have such a program because the benefit the household could receive is larger and the eligibility rules (from the household's perspective) are less complicated.

9. State reported error rates reflect the error determination of the state QC reviewer. USDA uses the federal re-review results to produce the official payment error rates. USDA does not publish the official error rate for subgroups, such as households with earnings or immigrants, so the state reported error rates are used for subgroup analysis.

10. As part of the July 1999 guidance, USDA announced that beginning in fiscal year 2000 it would raise the threshold for small errors from $5 to $25 as a permanent feature of the QC system.

11. The 1996 welfare law amended the Food Stamp Act (7 U.S.C. § 2012(c)) to allow 24-month certification periods for households in which all adults are elderly or disabled. For all other households, the certification period may not exceed 12 months. The food stamp regulations at 7 C.F.R. § 273.10(f) have not yet been revised to reflect this statutory change.

12. 7 C.F.R. §§ 273.21 and 273.10; 7 U.S.C. § 2014(f). States may choose to determine eligibility prospectively but determine the level of benefits retrospectively.

13. 7 C.F.R. § 273.10(c)(1). When the amount of income that will be received, or whether it will be received in a particular month, is not certain, the portion of income that is uncertain is not counted. For example, if a household has a child support order in place, but the noncustodial parent does not pay regularly, the child support income cannot reasonably be anticipated and therefore should not be counted. In general, states use the amount of income received during the past 30 days as an indicator of the anticipated income, as long as no changes have occurred or are expected to occur.

20. Robert Greenstein and Marion E. Nichols, Monthly Reporting: A Review of the Research Findings, Center on Budget and Policy Priorities, 1989.

21. Waiver response from USDA to California for waivers numbered 2000044 and 2000048-2000050.

22. See letters from Susan Carr Gossman and Arthur Foley of USDA's Food and Nutrition Service to FNS Regional Administrators, July 1999 and December 1999 (www.fns.usda.gov/fsp/Clintoninitiative/

31. Under the proposed rule, whether an employee is full-time or part-time is to be based on the employer's definition of full-time and part-time or on a definition used in the state's public assistance program. It is common for employers to consider a worker to be full-time when the employer provides health insurance and other benefits to the employee. Thus, an employee who works on an hourly basis and does not receive benefits would likely be considered part-time and would not be required to report any changes in monthly wages that result from increased or decreased hours of work.

32. Under current regulation (7 C.F.R. § 273.2(e)(2)), states are required to waive the in-person interview on a case-by-case basis in certain instances where appearing at the office would cause a hardship. In these cases the eligibility worker still conducts a telephone or at-home interview with the applicant. USDA's new approach does not change the requirements for interviews at initial application, but allows states to implement a broad policy to waive recertification interviews for large categories of households and rely instead on a mail-in recertification form. The new approach can be found in proposed regulation in the February 29, 2000 Federal Register, Vol. 65, No. 40, p. 10865 (preamble) and p. 10897 (proposed 7. C.F.R. § 273.2(e)(1). Also see, Food and Nutrition Service, U.S. Department of Agriculture, The Nutrition Safety Net: At Work for Families. p. 13.

33. The states with waivers are: Delaware, Louisiana, Massachusetts, Maryland, Maine, Michigan, Missouri, New Jersey, South Carolina, Tennessee, Texas, Utah, Virginia, Wisconsin, and Wyoming. Not all of these states have implemented the waiver statewide or for all types of households that are assigned short certification periods.

34. U.S. Department of Agriculture, The Food Stamp Program Quality Control Review Handbook, FNS Handbook 310, September 1997. pp. 2-4.

35. USDA similarly reports that 56 percent of dollar error amounts are agency errors, as opposed to household errors. (In cases where there is more than one variance, the total error amount for each case is assigned to the first variance coded.)

36. See Food Stamp Program Quality Control Annual Report, Fiscal Year 1998, U.S. Department of Agriculture, Food and Nutrition Service, Program Accountability Division, Quality Control Branch, February 2000, p. 28, 31, and 35.

37. Some states allow slightly longer for households to appeal the reduction in benefits.

38. By contrast, if the household was applying for recertification in April for May benefits, and either the household failed to disclose the new job or the state agency failed to address the change correctly, these exclusions would not apply and the state would be liable for a QC error for May.

39. Models of the $100-per month and five-hour change reporting options are not included because they are inferior to status reporting in terms of both household burden and state error rate concerns. In particular, under both options, for a household to report accurately, its members must be aware of how much income or how many hours of work the eligibility worker is anticipating the household will have over the certification period. They must remember this figure and must remember to report a change if the actual income or number of hours worked in a given month is different by more than $100 or 5 hours. This can be confusing for households and, as a result, is likely to be error prone.