This discussion focuses on the Income Shares model originated by Dr Robert Williams in the 1980s, and the underlying national economic data sources that he uses to feed it. As of this writing at least 31 states use this model and the underlying economics. Many of the presumptions used in states using other models come philosophically from the same thinking advocated by him.
I want to thank Roger Gay and Greg Palumbo for their contributions and advice on the content of this paper.
When one fully grasps the economic methodology used in today's guideline development, there becomes a recognition that it is impossible to really know what guideline numbers are appropriate or what assumptions are used to determine state-specific child support obligations. Only through getting at just those things would states be consistent with the tone of the 1996 report from the Office of Child Support Enforcement (OCSE) where they said:
"Surprisingly, few States reviewed their core guideline model or methodology. Rather, guideline reviews focused on issues relating to income, adjustments to income, adjustments to the guideline amount, and deviations from the guideline amount."What is needed is to get outside the paradigm that existing models (using the same 'economic studies') are the only way of determining appropriate and just child support awards in each state.
The reality is that the base economic studies used in child support schedule development were not planned for nor conducted with child support considerations in mind. Rather, they were designed for significantly different purposes, never intended to be specifically applied to individual situations such as child support. Highlighting this fact is that none of the studies measure what federal law says we need to do in each state, and that is to fully understand the impact on both parents' ability to continue to provide for their children in two separate households, fully considering the involved second parent's expenses.
The United States Bureau of Labor Statistics which gathers the base expenditure data used in the 'Income-Shares' model espoused by Williams, actually cautions against the use of such generalized data to apply to any individual situation, exactly what is done in the vast majority of the states, including Kansas. Most states have also been reviewing the logic of their child support guidelines based entirely within the bounds of the guideline logic itself. Over time, it is likely in each state that one can't tell for certain what assumptions have truly been included in the model or not, each of which directly affects the child support schedule values. Often, the state review committee has simply gone back to see if the modified guidelines still conform to the original developer's personal preferences. In the 14 years since the federal government mandated development of state-wide child support guidelines, additional research has been conducted, including new scientific approaches that solve many of the problems both in the baseline data used in state child support models, as well as in the models themselves. The flaws inherent in the current child support estimate methodologies are being addressed in this research. These should be fully explored consistent with what the OCSE urges be done regarding methodology review.
Williams describes his model in the Health and Human Services publication, 'Child Support Guidelines: The Next Generation', published in April 1994:
"The Income Shares model is based on the concept that the child should receive the same proportion of parental income that he or she would have received if the parents lived together. A basic child support obligation is computed based on the combined income of the parents (REPLICATING TOTAL INCOME IN AN INTACT HOUSEHOLD) {emphasis added}. This basic obligation comes from a table which is derived from economic estimates of child-rearing expenditures, minus average amounts for health insurance, child care, and child's extraordinary medical expenses. The basic child support obligation is divided between the parents in proportion to their relative incomes. Pro-rated shares of child care and extraordinary medical expenses are added to each parent's basic obligation. IF ONE PARENT HAS CUSTODY, THE AMOUNT CALCULATED FOR THAT PARENT IS PRESUMED TO BE SPENT DIRECTLY ON THE CHILD. FOR THE NONCUSTODIAL PARENT, THE CALCULATED AMOUNT ESTABLISHES THE LEVEL OF CHILD SUPPORT {emphasis added}."He continues,
"The Income Shares model was developed by the staff of the Child Support Guidelines Project, which was funded by the U.S. Office of Child Support Enforcement and administered by the National Center for State Courts. It utilizes several concepts from the earlier Washington (State) Uniform Child Support Guidelines, but diverges in basing its numerical parameters explicitly on a different and more recent body of economic analysis."The reader is urged to keep in mind a few key points from Williams' description of his Income Shares model as they will be addressed in the balance of this paper.
"A written finding or specific finding on the record that the application of the guidelines would be unjust or inappropriate in a particular case as determined under criteria established by the State, shall be sufficient to rebut the presumption in that case."It further specifies that guidelines:
"shall be reviewed at least once every 4 years to ensure that their application results in the determination of appropriate child support award amounts."In other words, the table values established within the guidelines are 'presumed' to accurately reflect the situation of parents and their children at the various income levels. In theory at least, federal law enables parents the possibility of pointing out to the court why the guideline numbers should not apply in their particular case (rebutting the presumption). Practice and theory though, are very different.
Economic studies used in the Income Shares model are based on total family expenditures in intact families. These are estimates of spending that might occur if the parents were living together, sharing all the expenses of a single household. Spending on children in split households has a random relationship to the combined income of the parents. The income of both parents can be appropriately considered in the award decision only if that consideration is consistent with the fact that the parents do not live together and therefore do not use their income jointly. Joint income, and table values related to joint income, have no relationship at all to family economic circumstances in the context of a child support award decision. In a particular state, even assuming that the sample of data is appropriate (and that is dubious as I will show below), individual case circumstances (those which deviate from the circumstances presumed in developing the guideline, such as separate households and continued dual parent involvement), cannot be adequately considered unless the numeric table is categorically divided (food, clothing, shelter, transportation, entertainment, etc.). Without an explicit and clear conceptual basis for the award, a parent attempting to rebut the presumptive amount on the basis that it is unjust or inappropriate must do so without knowing what just and appropriate means.
The only way to properly apply mathematical decision models within the context of Constitutional justice is to fully disclose the nature of the mathematics, the underlying reasoning, and the assumptions in such a way as to make their review practical in comparison with the circumstances of each case. Currently state committees reviewing the models and the underlying data, the judges making awards using the resultant support schedules, and attorneys and parents living with the results of them, are not able to directly tie the schedule to specific cases. It literally is impossible! Federal law (and the Constitution) requires a just and appropriate award in each case, and the goal for states is to construct guidelines that are sufficient to do so in every circumstance to which they are applied. It is also required that judges can identify inappropriate and unjust results and that attorneys and parents can argue for deviation when a formula fails. My state of residence, Kansas, is a joint custody preference state by statute, which logically entails some degree of joint parental involvement. Even those states without such a preference generally apply some minimal level of parenting time (visitation). Separation/divorce inherently means separate households. Therefore, use of the existing guidelines based on intact family expenditures without inclusion of involved non-custodial parent expenditures on children, is inappropriate in all such cases. Guidelines have continued to fail to take the reality of parenting expenditures of an involved second parent into account.
Most state guidelines utilize expenditure data developed from the Bureau of Labor Statistics (BLS) annual Consumer Expenditure Survey (CES) in the development of the child support obligation schedules. It consists of 5,000 household surveys conducted each quarter, totalling 20,000 surveys/year. (The BLS said that the 5,000 surveyed is a staggered pool concept. The whole annual sample is the same 5,000 households for 3 quarters, and a new 5,000 for a 4th quarter.)
Recognizing that each state is to have guidelines appropriate to that specific state, I called the regional BLS office in Kansas City, as well as their main office in Washington, and asked how many of the sample actually came from Kansas. All they could tell me was that it was "somewhat less than 100 surveys" (with all but a few out of the Kansas City Metro area, and the remaining coming from Lawrence, KS. None came from any other cities in the state, including Wichita which has the largest population in Kansas). Therefore, this state's guidelines, to specifically apply to its child support cases, are based upon generalized data, virtually all of which comes from out of state, and again which are derived from intact family expenditures. With this small national sample size, this has to be the case in each state. The bigger problem however, is the sample data itself.
The BLS publishes a list of 'Frequently Asked Questions' regarding the CES. Number 15 specifically asks and answers:
"What are some of the Limitations of the Data?"Along these same lines, Kansas Guidelines review committee economist Dr. Walter Terrell admitted to me in a letter in April 1998 in response to a request for a detailed break out of expenditure areas at various income levels, that:"The Interview and Diary Surveys are sample surveys and are subject to two types of errors, non-sampling and sampling. Non-sampling errors can be attributed to many sources, such as differences in the interpretation of questions, inability or unwillingness of the respondent to provide correct information, mistakes in recording or coding the data obtained, and other errors of collection, response, processing, coverage, and estimation for missing data. THE FULL EXTENT OF NON-SAMPLING ERROR IS UNKNOWN. (All caps added for emphasis) Sampling errors occur because the survey data are collected from a sample and not from the entire population. Tables with coefficients of variation and other reliability statistics are available on request. However, because the statistics are shown at the detailed item level, the tables are extensive."
"CAUTION SHOULD BE USED IN INTERPRETING THE EXPENDITURE DATA, ESPECIALLY WHEN RELATING AVERAGES TO INDIVIDUAL CIRCUMSTANCES. (All caps added for emphasis) The data shown in the published tables are averages for demographic groups of consumer units. Expenditures by individual consumer units may differ from the average even if the characteristics of the group are similar to the individual consumer unit. Income, family size, age of family members, geographic location, and individual tastes and preferences all influence expenditures."
"Given the same total level of spending due to children, the component parts will vary from family to family. That is, families, say, with a focus on dental and health care might show above average child spending on these items, and below average spending on children's clothing. This applies to the USDA (United States Department of Agriculture) estimated components as well, i.e., no measures of variation are presented for the component parts."The application of this generalized data currently utilized in Kansas and most other states, both from the federal government agency gathering the baseline data, as well as a committee's economist, shows no direct relationship with specific circumstances around individual child support scenarios in the state. The BLS explicitly discourages such application of data potentially riddled with non-sampling errors, and an expert economist admits that such detail necessary in order to potentially rebut, is not discernible from the model. What is also being pointed out is that there are absolute limits to what can be derived from the CES. Most state guidelines currently in use, stake their entire logic on inferences from the CES. But the CES itself has no way of telling us what the right redistribution of income actually is. It is necessary to supplement the statistical work with what the OCSE report has pointed out is missing in state reviews, the fundamental logic of the guidelines must also be further developed."In short, if the total amount of child support that is supposed to be spent due to children is in fact spent for that purpose, then the component parts are irrelevant. Further, about 75 to 80 per cent of expenditures on children involve jointly consumed goods, e.g., home, auto, utilities, etc This further complicates the question of how much is spent (on average) for each spending class."
Continuing, the Income Shares model incorporates the CES data as repackaged by the United States Department of Agriculture (USDA) in their report, 'Expenditure of Children By Families' which is published each year. From the 5,000 household quarterly CES data, the USDA culls it down based on the following qualifiers:
(12,850 Husband and Wife households / 3,395 Single-parent households)
Only intact husband/wife households are utilized due to sample size limitations
There is a two child assumption per Husband-Wife household.
The country is then divided into regions; West/Northeast/South/Midwest, and a general US Rural category. (Kansas for instance, is part of their Midwest Region which also includes: Illinois, Indiana, Iowa, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin. According to the USDA, among the regions, the Midwest is the lowest for child-rearing expenses)
The Expenditure Categories are: Housing; Food; Transportation; Clothing; Healthcare; Child care and education; and Miscellaneous
The expenditures for Clothing, and Child care and education only apply to children and are divided equally between them, and exclude adult-related expenses. Food expenditures are determined from the USDA Food Plans to allocate among the various family members. Healthcare expenditures are derived from the National Medical Expenditure Survey, and allocated among the family members by age, etc.
Expenditures for Housing, Transportation, and Miscellaneous goods and services however, are allocated on a per capita basis (divided equally among the members). This has the effect of minimizing the costs to adult members, while raising the level of expenditures on children. According to the USDA, this is done as they say there is no research base for allocating these expenses, and they reject the marginal cost method as well for that reason. (The study itself however, addresses the marginal cost basis in some detail in the report appendix, referring to actual studies that show that use of the marginal cost basis can reduce Housing expenditures by 28-44%, and the Miscellaneous category by 28%.)
The per capita methodology employed for these categories also shows problems when reviewing what is specifically included in these expenditures. For instance, Miscellaneous specifically includes such things as manicures, make-up, hair styling, health club memberships, country club memberships, etc. Surely, many of the expensive costs associated with maintaining adults should not be equally distributed amongst all family members including children since they are not costs associated with raising children.
The Transportation cost share as determined by the USDA included vacation travel expenses as well as automobile transportation expenses that were calculated by subtracting the costs associated with travel to work. That is they subtracted the mileage associated with getting to work from the costs of the automobile, insurance, maintenance, etc., and then divided this amount by the number of members in a family. For instance, a car costing $12,000 the day before a baby is added to an expectant family, is allocated at $6,000 for each parent. The next day, with baby arrived, the cost of the car attributed to the baby suddenly on the scene is $4,000! Arguably, the mileage directly associated with transporting children would be more accurate than USDA estimates, which seem to be grossly exaggerated.
The same type of treatment occurs for Housing. Using the expectant family example, the day before the baby's arrival, the cost for housing is divided equally between the two adults. Upon the baby's arrival, the cost of housing is suddenly divided equally between the adults and the baby. The child's 'portion' is then summed and used directly in the calculations for state child support guidelines. Are these children supposed to be buying their own cars and living in their own apartments? Or are they living in a parent's residence and being transported sometimes, including family outings to places the parent would be going anyway? On the face of it, it is obvious that such allocations are questionable. These points have not been so obvious in the past because the estimates have not been separated to the point that anyone on any state review committee understood what the numbers in the guideline mean. With today's guidelines and their underlying data, how in the world can parents, attorneys, and judges begin to understand them within the context of due process?
Regionalizing data creates problems as well for the figures used to calculate specific state child support schedules. Tax rates differ in each state, along with differing costs of living. Although the survey says that the measure of expenditures for items is after tax (arguing that it is therefore then held constant across the country), the reality is that the level of income available based on after tax and cost of living differs across the country, let alone each region as income available to spend varies as a result. (This also highlights an additional area of concern regarding available income for expenditure on children in the non-custodial home, which as established earlier, remains unaccounted for and unmeasured.)
As stated previously, to get to the schedules that have been developed in states using the Income Shares model (and most others as well), the CES data has been utilized to feed the model. I have pointed out many of the problems inherent in that data being used to determine child support schedules. I have also addressed several flaws in the Income Shares model itself.
First off is that actual expenditures on children by families is not addressed. The methodology used to identify family expenditures do not actually track all costs per person through a marginal cost accounting basis, and thus do not reflect true costs. What is reported are total 'intact' family expenditures, which are then broadly allocated to children, and then which are entered into the state economic model based on the parameters established by the modeler (certainly not any specific individual child support case at hand).
Additionally, the model also fails to account for costs incurred while the non-custodial parent exercises his/her parenting time. Lastly, the model purports to accurately reflect what it costs to raise a child in a particular state (such as Kansas in my experience, based on a household expenditure survey sample size consisting of "somewhat less than 100" Kansas participants!). An argument may be made for instance, that since about 30% of Kansas cases involve interstate child support orders, regional and national data is fine. However, orders under the jurisdiction of a particular state should be based on what it costs to raise a child specifically in that state as the starting point.
Some believe that it might be too costly to conduct state-specific studies, to include data collection. Their inclination would be to continue on using the same flawed methodology and data, falling back on the false belief that what we have today is the best we can do. However, this cost is arguably quite negligible compared to the impact of our current approach. To simply continue doing the same old thing would be ignoring our responsibility to be thorough in guideline development and review. Opening our minds to alternative approaches of child support determinations could prove to be less costly than might be believed.
Lastly, I have not mentioned the impact that today's approach to child support guidelines and enforcement have on our country's children. Focusing solely on 'financial' child support while failing to emphasize 'emotional' child support is destroying our children's lives by depriving them of someone that they desperately need to have involved in their lives, their other parent. All credible research shows that for the vast majority of children, the best parent to children of separation and divorce is quite simply both parents.
Instead of seeking out ways of facilitating dual parent involvement, our current public policy has established economic and legal roadblocks, merely because it is easy to 'garnish' income. The end result in this crazy social experiment is increased juvenile suicide, teenage pregnancy, juvenile delinquency, and teenage drug abuse, among many other childhood pathologies, all sharing a common variable in most of these instances, an absent parent. Continued parental involvement of the second parent (the one currently not included in the studies of the true ongoing costs of parenting), does indeed cost money. Such involvement is not a free good. This country is beginning to awaken to the damaging effects of having frustrated dual parent involvement for so many years. Reforming the child support enforcement public policy from the ground up needs to occur. Only then will we begin to turn the corner in the direction of our country's most important assets, our children.
Received - 11 September 1998