July 15, 2003

NEW ANALYSIS SHOWS "OVERCERTIFICATION" FOR FREE OR REDUCED-PRICE SCHOOL MEALS HAS BEEN OVERSTATED
By Zoë Neuberger and Robert Greenstein

Summary

PDF of this report

View Related Analyses

If you cannot access the files through the links, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader.

As the reauthorization of the child nutrition programs approaches, considerable attention has been paid to a USDA comparison that appears to show that substantial numbers of ineligible children are receiving free or reduced-price school meals.  USDA compared the actual number of children certified to receive free or reduced-price meals to an estimate of the number of eligible children, based on census data.  The number of children certified for free meals was 29 percent greater than the number who appeared to be eligible, indicating extensive “overcertification” in the school meals programs.

The USDA comparison had some methodological weaknesses, however, and a new analysis has now been conducted using a different census data source that allows for a more precise estimate of the number of children who were eligible for school meals at the time they were certified for the meals.  This new analysis shows that the number of children certified to receive free meals exceeded the number of eligible children by a much smaller amount — by two percent.  Moreover, the number of children certified to receive either free or reduced-price meals was 15 percent lower than the number of eligible children.

 

USDA’s Comparison

The USDA comparison indicated that for the 1998-1999 school year, the number of children approved to receive free lunches exceeded the estimated number of eligible children by 29 percent.[i]  The number of children approved to receive either free or reduced-price lunches exceeded the estimated number of children eligible for free or reduced-price meals by three percent.

New research shows that the number of children certified for free school meals exceeded the number of children eligible for these meals by two percent and that other figures sometimes cited as showing a larger discrepancy come from a comparison that is analytically flawed.

USDA’s estimate of the number of children eligible for free or reduced-price meals was based on annual household income as reflected in the Census Bureau’s Current Population Survey (CPS).  School meal applications, however, ask families to provide monthly rather than annual income.  In addition, in the USDA analysis, annual income for calendar year 1999 was compared to school meal certifications made in the summer or fall of the preceding year, 1998.

For the period that the USDA comparison covers, both the use of annual rather than monthly income and the timing lag (i.e., the comparison of income data for one year to meal certifications made in the summer or fall of the preceding year) tend to understate the number of eligible children.  Understating the number of eligible children results, in turn, in an overstatement of the degree to which the number of children receiving free or reduced-price meals exceeded the number of children who were eligible.

 

New Findings

A new analysis conducted by Dr. John Karl Scholz, Professor of Economics and Director of the Institute for Research on Poverty at the University of Wisconsin, uses monthly income data as collected in the Census Bureau’s Survey of Income and Program Participation (SIPP) to develop estimates of the number of children eligible for free or reduced-price meals.  Since applications for free or reduced-price meals distributed at the start of the school year ask families to provide their prior month’s income, families generally provide their income for August or September.  As explained below, the method employed in the USDA comparison of using annual Census income data understates the number of eligible children.  Dr. Scholz’ analysis shows that when household income for August or September of 1998 is used, 2.7 million more children are found to have been eligible for free meals than when estimates based on income for the following calendar year are used, as was the case in the USDA comparison.[ii]  This more accurate estimate of eligible children, based on Census data on monthly income, closely matches the actual number of children that were certified to receive free meals that fall.

These findings confirm that using annual income from calendar year 1999 to estimate the number of children eligible for free or reduced-price meals in the fall of 1998, as the USDA comparison did, substantially underestimates the number of children that were eligible at the time of certification and, as a result, overstates the extent to which ineligible children were certified.  Based on the more accurate estimates of the number of children eligible for meal subsidies in the fall of 1998 that were derived from the Census Bureau’s monthly income data, the number of children certified to receive free meals exceeded the number of eligible children by only a small amount.

 

Research Context

Dr. Scholz’ new findings are consistent with prior research.  Census data have consistently shown that many households are poor for a portion of the year but have income for the year as a whole that exceeds the poverty line.  The Census Bureau has found that 22 percent more people are poor in an average month than are poor based on annual income.[iv]  Thus, in any given month, there are some households with incomes below 130 percent or 185 percent of the poverty line — the eligibility limits for free and reduced-price meals, respectively — whose annual income will be above these levels.

When analyzing eligibility for the WIC program, the National Research Council (which is connected to the National Academy of Sciences) found that 21 percent more children under age five are in households with income below 185 percent of the poverty line in at least one month of the year than are in households with annual income below 185 percent of the poverty line.[v]  The Council recommended use of monthly Census data from the SIPP, rather than annual Census data from the CPS, in generating estimates of the number of people eligible for WIC.  (USDA had been using the annual income data to estimate the number of people eligible for WIC.)  In light of this prior research, it is not surprising that more children are eligible for free or reduced-price meals when August and September income is considered than when income over a full calendar year is used, and that use of annual rather than monthly Census data results in a sizeable underestimate of the number of children eligible for these meals.

Furthermore, during a period when unemployment is falling, wages for low-wage jobs are rising, and poverty is declining — as was the case in the late-1990s, the period the USDA comparison covers — incomes during the following calendar year will be higher, on average, than incomes in the fall of the previous year.  USDA’s comparison of an estimate of the number of children below the school meal income limits in calendar year 1999 to the number of school meal certifications made in late summer or early fall 1998 consequently resulted in a further understatement of the number of eligible children, and thus in a further overstatement of the degree of overcertification.

The National Research Council found similar problems with past analyses that used Census data on annual incomes to estimate the number of people eligible for WIC; the Council recommended using monthly Census data instead for this purpose.

It is important to bear in mind, however, that even though the number of children certified to receive either free or reduced-price meals did not exceed the number of eligible children, some ineligible children almost certainly were certified.  If the number of children certified exactly matched the number of eligible children but some eligible children were not certified, then the slots not filled by eligible children would have to have been filled by ineligible children.  We know that not every eligible child is certified to receive free or reduced-price meals although there are no data that allow for an accurate estimate of the extent to which eligible children are not certified.  Thus, even though the number of children certified to receive free or reduced-price meals does not exceed the number of eligible children, some ineligible children are being certified.  Similarly, even though the number of children certified to receive free meals exceeded the number of children eligible for free meals by only two percent, it is a near certainty that more than two of every 102 certified children were ineligible.

 

Conclusion

While existing data indicate that some ineligible children receive free or reduced-price meals, Dr. Scholz’ analysis of SIPP data indicates that the extent of erroneous certification is more limited than USDA’s initial analysis of CPS data suggested.  Past studies have shown that income verification requirements, which are designed to detect ineligible participants and to discourage ineligible families from applying for free or reduced-price meals, also result in fewer eligible children being certified for free or reduced-price meals.  If a household does not respond to a request for income verification — which is currently the case for one of every three households subject to income verification requirements — the children lose these meals even if their income is below the eligibility limit.  Past research found that the large majority of non-responding households were, in fact, eligible for free or reduced-price meals.

In the absence of better information about the extent of overcertification and how to prevent it without deterring participation by eligible children, policymakers need to approach this area with caution.

In the absence of better information about the extent to which ineligible children are being certified and how to prevent erroneous certifications without deterring participation by eligible children, policymakers need to approach this area with caution.  Policy proposals intended to reduce the certification of ineligible children carry a risk of leading to substantial increases in administrative costs, limited effectiveness in reducing erroneous certifications and — most serious — deterrence of large numbers of needy, eligible children from receiving free or reduced-price school meals for which they qualify.  Policy changes designed to reduce the certification of ineligible children need to be conducted, at least initially, on a limited basis and carefully evaluated to ensure that these efforts advance, rather than impede, achievement of the basic purposes of the school meals programs.


End Notes:

[i] USDA has said that the number of children approved to receive free lunches exceeded the estimated number of eligible children by 27 percent.  The USDA estimate included the territories.  When USDA’s data are adjusted to reflect only the 50 states and the District of Columbia, the number of children approved to receive free lunches exceeded the estimated number of eligible children by 29 percent.  This paper uses the 29 percent figure because the data from the Census Bureau’s Survey of Income and Program Participation and the new estimate of the number of eligible children that is derived from it reflect the 50 states and the District of Columbia but not the territories.

[ii] SIPP data from December 1, 1998 through November 30, 1999 were used as a proxy for calendar year 1999 data because data for December 1999 were not available.

[iii] The fact that the number of certified children is lower than the number of children estimated to be eligible when the reduced-price meal category is added to the analysis reflects two factors.  First, some children who are certified for free meals are not eligible for them but rather are eligible for reduced-price meals.  Second, it is likely that a lower portion of eligible children are certified for reduced-price meals than for free meals.

[iv] The Census Bureau computes the “average monthly poverty rate” by first comparing family income in each month to the poverty threshold to develop an estimate of the number of poor families in each month of a year.  Then the total number of poor families in each month of a year is added together and divided by 12 to develop an average monthly poverty rate.  Census Bureau, Trap Door? Revolving Door?  Or Both?,  P70-63, July 1998.

[v] National Research Council, Estimating Eligibility and Participation in the WIC Program, 2001, Table 4-3.