Search Type Title Author/Investigator Search Advanced Search

Return to Previous page | Table of Contents | Previous | Next

Results

Elmira Results

Average per family net costs to government for the nurse-visited and non-visited groups are summarized in Table 2. Over the first 15 years of the study child’s life, the nurse-visited group used $56,000 (2001 dollars) less per family in government services than did the non-visited group. The visited group also paid $8,300 more per family in taxes than did the non-visited group. This resulted in a 393 percent recovery of the amount invested in the nurse-visitation program by the study child’s 15 year.

During the study child’s first 15 years, use of AFDC, Medicaid, and food stamps was substantially lower in visited families than in comparison group families and accounted for the greatest government cost savings (Table 3). Use of child protective services and Supplemental Security Income (SSI) were also low for visited families. Interestingly, education costs, both special education and repeated grades, were somewhat higher per family in the nurse-visited group than in the comparison group, although the difference was not statistically significant. Given that low-income families, and especially those headed by single mothers are at greater risk for use of government services, it is useful to determine whether the cost savings in the Elmira study were greater for low-income families. We therefore examined government costs for the treatment and comparison groups according to socio-economic status (SES) at randomization as expressed by Hollingshead classes I, II and III versus IV and V (Hollingshead, 1976). Hollingshead classes IV and V include semi-skilled and unskilled laborers. Classes I, II, and III include all other classes, including skilled craftsmen, clerical and sales workers, minor and technical professionals, and major business professionals. Table 4 shows per capita expenditures for the two Hollingshead groupings within each treatment group. It is apparent that most of the recovery of visitation program costs in Elmira occurred in the low SES group, with a five-fold return. The higher SES families recovered 150% of the visitation program’s cost. This finding coincides with that of the earlier Elmira cost study which found program savings at the study child’s fourth year to be concentrated in low-income families (Olds, Henderson, Phelps, et al., 1993). And it provides continuing justification for focusing this nurse home visitation program on low-income families.

Table 2: Average per Family Net Cost to Government, Period from Study Child's Birth to 15 Year, Nurse-visited and Comparison Group Families in Elmira, New York, 2001 Dollars

Type of Cost/Revenue	Comparison Group (N=134)	Nurse-Visited Group (N=87)	Difference	P-Value
Government Programs1	$148,998	$101,190	$47,808	p = .049
Tax Revenues2	45,157	53,494	8,337	p = .238
3. Net Government Cost (line 1 - line 2)	$103,841	$ 47,696	$56,145	p = .028
4. Home visitation program cost	---	$ 14,287	$14,287
5. Percent recovery	---	---	393%

1Government programs include Child Protective Services, AFDC, SSI, Medicaid, Food Stamps, Foster Care, Education (repeated grades), Special Education and New York State crime data. These figures include net government costs (government program expenses minus tax revenues) for the first 15 years of the study child’s life.[back] 2These figures include tax revenues for the period between the subject child’s birth and 15 year.[back]

Table 3: Government Program-Specific Costs Per Family for the Period from Study Child's Birth through 15 Year, Nurse-visited and Comparison Group Families in Elmira, New York, 2001 Dollars

Government Program	Comparison Group (N=134)	Nurse-visited Group (N=87)	Difference	P-Value
1. Aid to Families with Dependent Children (AFDC)	$65,354	$73,754	$ 8,400	p = .051
2. Child Protective Services	3,032	1,772	1,260	p = .143
3. Food Stamps	31,779	22,143	9,636	p = .027
4. Foster Care	2,052	599	1,453	p = .701
5. Medicaid	28,155	20,088	8,067	p = .066
5. Repeated Grades	3,775	4,352	(577)	p = .629
6. Special Education	2,525	3,436	(911)	p = .474
7. Supplemental Security Income (SSI)	8,953	3,543	5,410	p = .100
8. Costs of crime	3,373	2,274	1,099	p = .980
Total	$148,998	$101,190	$47,808	p = .049

Comparison of Current Economic Analysis with RAND Analysis

Prior to the current analysis, a study of the Elmira home visitation program was undertaken by the RAND Corporation (Karoly et al., 1998). There are several differences in the approaches taken by RAND and by the present study. A description of the most important differences is provided in the following paragraphs.

Both studies are net cost analyses or partial cost-benefit analyses, in that they monetize the benefits to government of having to provide less support to nurse-visited families. They fall short of full cost-benefit analyses because numerous benefits to society and to individuals, such as the reductions in child abuse and neglect and crime and improvements in children’s behavior cannot be monetized easily. Both studies adjusted for inflation in order to express values in real dollars. The differences between the two analyses lie in the scope and number of government programs analyzed and in some of the assumptions underlying the cost estimates. In terms of the number of programs analyzed, the RAND study included welfare programs, specifically AFDC, Medicaid, and food stamps, emergency room visits by the study child, taxes paid, and crime. The estimates of crime costs were especially detailed and inclusive.

Table 4: Comparison of Per Family Government Expenditures and Taxes Paid between Study Child's Birth and 15 Year According to Hollingshead Socio-economic Status Categories,*Elmira, New York, 2001 Dollars

Program	Comparison Group (n=82)	Hollingshead Categories IV and V			Comparison Group (n=52)	Hollingshead Categories I-III
		Nurse-visited Group (n=48)	Difference			Nurse-visited Group (n=39)	Difference
AFDC	$85,259	$55,536	$29,723	p = .038	$33,966	$27,533	$6,433	p = .833
CPS	3,845	1,648	2,197	p = .077	1,752	1,924	(172)	p = .967
Food Stamps	40,649	25,395	15,254	p = .021	17,792	18,141	(349)	p = .447
Medicaid	35,959	24,507	11,412	p = .053	15,848	14,650	1,198	p = .818
Education (repeated grades)	4,264	4,930	(666)	p = .694	3,004	3,640	(636)	p = .659
Special Ed	2,610	4,995	(2,385)	p = .188	2,391	1,517	(874)	p = .595
SSI	9,972	4,555	5,417	p = .369	7,347	2,298	5,049	p = .138
Crime	4,227	1,405	2,822	p = .745	2,027	3,343	(1,316)	p = .701
Foster Care	1,975	828	1,147	p = .364	2,174	317	1,857	p = .428
Total	$188,759	$123,800	$ 64,959	p = .033	$86,299	$73,363	12,936	p = .800
Taxes Paid	$40,216	$46,877	6,661	p = .261	$52,949	$61,637	8,688	p = .662
Total (Govt prog savings + taxes paid)			$71,620	P = .040			$21,624	p = .384
Per capita program cost			14,287				14,287
Percent recovery			501%				151%

 

With the exception of emergency room visits, the current study includes all of the categories explored by RAND; in addition it monetizes child protective services, foster care, Supplemental Security Income, repeated grades and special education. Our estimation of crime costs is much more limited than that in the RAND study. The source of all government program utilization for the RAND study came from the 15-year follow-up survey of program participants; the current study used data from that survey as well as administrative data for child protective services, foster care, and arrests.

Except for the costs of crime, the RAND study used more general assumptions than did the present analysis. For instance, RAND obtained monthly amounts for AFDC, Medicaid, and food stamps from Census Bureau documents. The report did not specify whether these were Chemung County, New York state, or national estimates nor whether they were averages. Also, the RAND study included welfare administration, thereby adopting an average cost analysisapproach, rather than a marginal cost approach. For the present study, we assumed that each family received the maximum AFDC and food stamps grant amounts. We obtained Chemung county AFDC grant amounts from New York State, and calculated each year’s amounts based on changes in the AFDC benefit. Over the period covered by the 15-year interview, items like home energy allowance, shelter allowance and supplemental home energy allowance were added to the AFDC benefit. We did not include welfare administrative costs because we conducted a marginal cost analysis, rather than an average cost analysis. We assumed that program impact on government programs, such as Medicaid or TANF, would be relatively small and thus would not affect the administrative costs for implementing those programs.

Another substantial difference in approach was in estimating the costs of crime. The RAND study provided a much more comprehensive analysis of the effects of crime reduction, by including not only the costs of arrests, but also cost of adjudication, jail operating costs, prison costs, tangible victim costs, and costs of being a criminal career for both the mother and study child. All arrest and conviction data came from the 15-year survey. All unit costs were based on published data. The current study used arrest data provided by New York State and estimated only arrest and jail costs for the study mother, because juvenile arrest cost data were not available from the state. Number of nights spent in jail were obtained from the 15-year survey.

Estimation of taxes paid by study families differed greatly between the two studies. Both studies obtained months employed from the 15-year follow-up survey (Olds et al., 1997). The RAND study used hourly income from published data on mean wages for women who had been on AFDC. The RAND authors assumed all taxes amounted to 25% of salary; they further assumed taxes in the amount of 10% of salary were paid by employers (for Social Security and Medicare). Also included were fringe benefits of 10%. The current study based income on wages reported in the 15-year survey. We assumed no benefits for this group, based on published reports about the likelihood of receiving benefits for women who had moved from welfare to work (Heymann, 1999). We also did not assume employer-paid taxes, such as the employer’s share of Social Security and Medicare taxes. We reasoned that the job filled by the respondent would have been filled by someone else, if it weren’t filled by the respondent, and that the employer would be paying those taxes anyway. These taxes would not be affected by the home visitation program. This is consistent with our marginal analysis approach.

We used income tax rates and credits in effect for each year during which the respondent reported income. We calculated income taxes using rates for household heads for each year. Most respondents were eligible for the earned income tax credit, which we also calculated.

Social security tax estimates were based on respondents’ reported income; we applied the fixed social security tax rate to estimate these taxes. Medicare taxes were estimated using the approach used for Social Security taxes.

Our approach resulted in much lower taxes paid by the study families than did the RAND study. RAND’s assumption that total taxes would be 35% of income seems on its face to be too high for these typically low-income families. The total tax rate in the United States, including not only the taxes included here, but also excise taxes, property taxes and sales taxes, amounts to less than 30% of Gross Domestic Product. These families, whose incomes were relatively low during the study period, are not likely to have paid 35% of their income in taxes, even considering the general regressivity of the U.S. tax system.

Denver Results

Table 5 summarizes the net costs of the home visitation program in Denver as of the study child’s 4 year of age. While the nurse-visited group accounted for over $1600 per family less in government support than did the comparison group, the paraprofessional-visited group required $618 more per family than did the comparison group. The nurse-visited group paid more taxes than did any other group, $874 more than the comparison group and about $500 more than the paraprofessional-visited group. As a result of these two factors, the nurse visitation program in Denver recouped 29 percent of its cost by the study child’s fourth year by virtue of lower net government expenditures and greater taxes paid. In contrast, the paraprofessional program, because the visited families used more government services than the comparison group, showed a negative recovery rate.

Table 5: Average per Family Net Cost to Government, Pregnancy to Study Child's 4 Year, Comparison Group, Nurse-visited, and Paraprofessional-visited Families in Denver, 2001 Dollars

Type of Cost/Revenue	Control (N=224)	Nurse (N=206)	Para- professional (N=211)	Difference (Control – Nurse)	P-Value	Difference (Control - Para- professional	P-Value
1. Government Programs	$14,964	$13,361	$15,582	$1,603	p = .378	(618)	p = .651
2. Tax Revenue	5,575	6,449	5,946	874	p = .957	371	p = .864
3. Net Government Costs (line 1-line 2)	$ 9,389	$6,912	$ 9,636	$2,477	p = .761	(247)	p = .396
4. Home Visitation Program Cost		$8,661	$5,838	$8,661		$5,838
5. Percent recovery				28.6%		-0.4%

 

The higher taxes paid by the nurse-visited group suggests that these mothers were farther along in the process of attaining economic independence than was the comparison group. Evidence of this is also provided by this group’s lower use of welfare programs, especially Medicaid and food stamps (Table 6). But the nurse-visited group used less subsidized child care than did the other two groups, a surprising finding, since child care is more often used by working mothers than non-working mothers. At this point in the Denver trial, none of the differences between the groups in individual government program costs is significant.

A continuation of these differences for the nurse-visited group could result in recovery of program costs over the next several years, although at a slower rate than occurred in Elmira for low-income women.

The performance of the paraprofessional-visited group suggests that such a program may not be a wise expenditure of limited tax revenues. While it is less expensive to deliver than is the nurse-visitation program, it does not show promise for future recovery of program costs, as does the nurse-visitation program.

Table 6: Detailed Government Program Costs per Family for Nurse-visited, Paraprofessional-visited, and Comparison Groups at Age 4, Denver, Colorado, 2001 Dollars

Government Program	Comparison (N=224)	Nurse (N=206)	Para- professional (N=211)	Difference (Comparison - Nurse)	P-Value	Difference (Comp. - Paraprof.)	P-Value
Survey Data:
Colorado Child	$12	$49	$55	($37)	p = .220	($43)	p = .070
Health Plan
Early Head Start	1, 210	1,158	610	52	p = .593	600	p = .953
Head Start	644	482	701	162	p = .314	(57)	p = .767
Energy Assistance
	57	44	52	13	p = .549	5	p = .968
SSI	652	492	216	160	p = .662	436	p = .233
Administrative Data:
AFDC/TANF	3,304	3,204	4,183	100	p = .638	(879)	p = .061
Subsidized Child Care	1,184	685	1,043	499	p = .842	141	p = .217
Family Preservation*	22	33	43	(11)	p = .782	(21)	p = .665
Out-of-home Placement*	69	75	261	(6)	p = .589	(192)	p = .602
Food Stamps	2,674	2,383	2,987	291	p = .903	(313)	p = .225
Job Training	113	77	140	36	p = .480	(27)	p = .084
Medicaid	5,023	4,679	5,291	344	p = .334	(268)	p = .745
Total	$14,964	$13,361	$15,582	$1,603	p = .378	(618)	p = .651

*Family preservation and out-of-home placement are programs within the Colorado foster care system and correspond to the Foster Care program described for Memphis and Elmira.

Memphis Results

Nurse-visited families in Memphis required $2,285 less in government expenditures than did the comparison group from the study child’s birth until age 4 ½, a significant difference (Table 7). They also paid $218 more in taxes than did the comparison group, resulting in a net government cost difference of $2,503. This represents a 25.7% recovery of program costs, somewhat lower than the recovery rate for nurse-visited families in Denver.

Table 7: Average per Family Net Cost to Government, Birth of Study Child to Age 4 ½ , Comparison Group and Nurse-visited Families in Memphis, 2001 Dollars

Type of Cost/Revenue	Comparison Group (n=456)	Nurse-visited (n=204)	Difference
1. Government Programs	$27,865	$25,580	$2,285	p = .041
2. Tax Revenue	1,872	2,090	218	p = .976
3. Net Government Costs	25,993	23,490	2,503	p = .053
4. Home visitation program cost		9,755	9,755
5. Percent Recovery			25.7%

 

Most (63.5%) of the difference in government program costs between the nurse-visited group and the comparison group resulted from significantly lower use of food stamps by the nurse-visited group (Table 8). The average cost of several other government services was also lower for nurse-visited families: AFDC, Foster Care and Medicaid. The nurse-visited families, however, accounted for significantly greater subsidized child care expenses. The higher taxes paid by nurse-visited families are consistent with greater use of subsidized child care, since their higher taxes may have been the result of holding more or better jobs than were held by comparison group families. The Child Protective Services and Foster Care costs are interesting. Child Protective Services agencies spent more on nurse-visited families than on comparison families, but nurse-visited families required much less in Foster Care expenditures than did other families. Earlier work in Elmira found that child maltreatment case-finding for nurse-visited families was greater than for comparison families (Olds et al., 1995), but that case severity was lower in nurse-visited families. This could be the explanation for the seemingly paradoxical results we found for Child Protective Services and Foster Care costs for the two groups of families in Memphis.

Table 8: Detailed Government Program Costs per Family for Comparison and Nurse-visited Groups at Age 4 ½, Memphis, 2001 Dollars

Government Program	Comparison Group (n=456)	Nurse-visited (n=204)	Difference
AFDC	$7,914	$ 7,234	$680	p = .053
Child Protective Services	643	879	-236	p = .958
Subsidized Child Care	1,026	1,612	-586	p = .011
Foster Care	728	39	689	p = .592
Food Stamps	14,060	12,608	1,452	p = .013
Medicaid	3,494	3,208	286	p = .097
Total	$27,865	$25,580	$2,285	p = .041

 

Recovery of the cost to deliver the home visitation program has proceeded somewhat more slowly in Memphis than in the other sites. This may be explained in part by differences in the populations among the three sites. As Table 1 shows, the Memphis participants were more likely to be unmarried, more likely to be under 19, and more likely to be a member of a minority group than were participants in either of the other sites. They were more likely to have low incomes than were the Elmira participants. This relatively high level of risk for negative outcomes could have placed more obstacles in the path toward labor force participation and self-sufficiency for the Memphis group than for the other two sites. Nevertheless, the nurse-visited group in Memphis earned more income, paid more taxes and used government programs less than did the comparison group. The nurse-visited group may yet recover visitation program costs, albeit at a slower rate than in Elmira or Denver. An analysis of the longer term effects of the Memphis nurse home visitation program should reveal whether this has happened.

 

 

Return to Previous page | Table of Contents | Previous | Next