The primary data used for analysis are Colorado-specific. We use data from other states for comparisons.

To study the effect of interest on payment behavior in Colorado, we use data from four primary sources.

Most of the collections data from other states were obtained from the Federal Office of Child Support Enforcement's Report to Congress. We use information from 1998 because that is when the survey of state interest usage was conducted.

Several different methodologies were used to answer the research questions. One of the more frequently used methodologies was to compare differences in collections between interest-assessing counties and non-interest assessing counties. A similar comparison is made between interest-assessing states and non-interest assessing states. In effect, these are natural experiments because roughly half of the Colorado counties assess interest and roughly half of the states assess interest. This methodology was used to gain insight into interest's impact on arrears payments (research question 2); interest's impact on payment of current support (research question 3); and interest's impact on different groups of obligors (research question 5). However, there are at least two limitations to this approach. First, whether and when a county assessed interest is subject to the interviewer's response, who may not recall this information precisely. Second, counties that assess interest do not assess it uniformly. Some counties assess interest on judgments only, others assess it in conjunction with an arrears calculation, and still others assess it differently. The comparison does not capture the impact of these variations between counties. In additional analysis, we attempted to capture these county variations by grouping counties with similar interest policies (e.g., counties that assess interest on judgments only in one group and counties that assess interest on judgments and non-judgments in another group). We did not detect statistical differences between these subgroups.

The second most commonly used method was to apply information from other states. This was used in part to answer how much interest would likely be collected (research question 2); the impact of interest on difference groups (research question 5); and the effect of notification on collections (research question 6). The limitation of this approach is that Colorado might not replicate another state's experience.

Two other methods were used. The first question, which addresses interest's impact on statewide accounts receivable, is answered by projecting accounts receivable using statistics from 1999 State Monitoring Reports and State performance standards. The projection is based on three major assumptions: (1) if the State begins charging interest statewide, the amount of arrears would be equivalent to that of January 1, 2000; (2) counties will achieve the State performance standards for percent of current obligation paid; and, 3) that interest arrears will be paid at the same rate as 1999 arrears. Generally, this assumes an optimistic scenario and is likely to underestimate interest's impact on accounts receivable.

Another method was used to analyze the impact of interest on negotiations. This consisted of compiling the information collected from the negotiation survey.

Exhibit 9 shows the number of counties that assessed interest in 1997, 1998, 1999 and January 2000. As of January 2000, 26 counties assess interest and 37 counties do not. As is evident in Exhibit 9, the number of interest-assessing counties has declined since 1997. In 1997, 32 counties assessed interest and 31 counties did not. Counties have stopped assessing interest due to a variety of issues. Some counties found it too burdensome to compute and did not feel they had the automation to support it. Other counties were concerned about equity issues. For example, one interviewee pondered whether it was fair to assess interest at case closure when the obligor had not been previously notified of the amount of interest that was accruing. Most recently, Garfield County suspended interest assessment due to a lawsuit pertaining to interest that is currently being appealed.

Although there are more counties that went from assessing interest to not assessing interest, there are a few counties that have just begun to assess interest in the last few years. One of these counties reported that their County Commission drove their recent policy change.

As discussed earlier, counties differ in when they assess interest and on what type of arrearages they assess interest (e.g., those reduced to judgments). Among the counties assessing interest, most of them only assess it when they are preparing a case for a specific enforcement action (e.g., preparing for a lump-sum negotiation). The only exceptions are two counties that assess interest monthly on judgments among paying cases.

As discussed in the last chapter, interest arrears account for about 7% of Minnesota's total IV-D arrears, and about 14-17% of Virginia's total IV-D arrears.^[9] It is not clear what causes the precise difference between Minnesota and Virginia. One explanation may be time. Virginia began assessing interest in July 1995. Although Minnesota began assessing interest in 1993, it was not routinely assessed until implementation of the statewide automated system in 1997.

Another difference between Minnesota and Virginia is their interest rate. The interest rate in Minnesota is based on the secondary market yield of U.S. treasury bills. It is currently about 7% per annum and assessed monthly. Virginia assesses a monthly simple interest rate of 9% per annum.

Exhibit 10 displays an estimate of how statewide assessment of interest could inflate Colorado arrears in its first year of implementation. The estimate assumes that:

We make no assumptions about whether interest affects the percent of current collections paid, we simply assume that the current collection goal, 52.5%, is met. (The impact of interest on current collections is addressed in the third research question.) If the goal for current collections is unmet, the projection underestimates interest's impact on arrears balances. Furthermore, if collections on interest is less than 6.41%, the projection also underestimates interest's impact on arrears balances. (As discussed in the next research question, 6.41% is considered the high-range estimate.)

We project interest accrual for two different scenarios: (1) interest accrues only on past-due current support after January 1, 2000; (2) interest accrues on all arrears. For comparison, we also project interest accruals using simple interest on all arrears.

As evident in Exhibit 10, if interest accrues only on past-due current support after January 1, 2000 by the end of the year, interest arrears would comprise about $1.1 million dollars and 0.09 percent of the total arrears. If interest accrues on all arrears, the year-end forecasted interest arrearages would be about $134 million if interest is compounded and $128 million if it is calculated using simple interest (about 10 percent of total arrears in either circumstance). These are one-year estimates. In subsequent years, these interest arrearages will likely multiply.

The State Auditors already consider Colorado arrears high relative to other states. Over $1 billion is owed in Colorado child support. The State Auditor's report shows that arrears per case average $4,400 in Colorado while the national average is $2,263.^[10] Statewide interest assessment will increase the gap between Colorado and the national average, particularly if interest is assessed on all arrears.

This is a two-fold question. It addresses what proportion of interest arrears is likely to be collected. It also addresses whether interest charges could change the amount of total arrears collected. As discussed in Chapter II, Virginia is the only state where we were able to obtain numbers on how much interest arrears was collected. In one year, Virginia collected $460,00 from about $77.5 million due in interest (0.6%). Virginia currently collects about 11% of its total arrears (i.e., interest and principal arrears).

Based on Virginia's experience and the experiences of Colorado counties assessing interest, which is shown later in Exhibit 11, we assume that the percent of interest arrears collected will not exceed the proportion of total arrears currently collected in Colorado. In other words, the proportion of interest balances collected will be less than the proportion of total arrears collected. In 1999, Colorado collected 6.4% of its child support arrears. Thus, we assume this is the maximum percentage of interest that Colorado can expect to collect. We also assume that the minimum percentage of interest Colorado can expect to collect is the same rate experienced by Virginia (0.6%).

We are unable to develop a point estimate between the range in estimated potential interest collections (0.6-6.4%). We had hoped to gain some additional insight into this issue by tracking the payment records of obligors who were assessed interest, but we were unable to identify a large enough sample of cases with interest assessment. As discussed earlier, we had randomly selected 400 arrears cases from counties assessing interest. Only 76 cases had interest charges posted on ACSES and most of these were closed due to locate, invalid case, disability or some other factor that suggested child support was not being collected on the case. Nonetheless, even if we had the data to derive a point estimate there would also be a margin of error (e.g., a point estimate +/- 5%).

To gain more insight into whether interest would affect collection of total arrearages, we compared Colorado counties that assess interest to those that do not. Exhibit 11 compares the average arrears per case, the proportion of cases with arrears and the percent of arrears paid between counties that assess and counties that do not assess interest. There are no statistical differences in the percent of arrears paid between counties that assess interest and those that do not. For example, as shown in Exhibit 11, from January through October 1999 counties assessing interest collected 7.0% of their arrears and counties not assessing interest collected 6.9% of their arrears. This suggests that interest has no effect on the percent of arrears paid (i.e., it does not increase or decrease it).

A similar comparison is made between states that assess interest and those that do not based on information provided in the Federal OCSE Reports to Congress. States that assess interest collect about 16% of the amount past due on average. In comparison, states that do not assess interest collect about 13% of the amount past due. The difference is not statistically significant.

Exhibit 11 also compares the average arrears per enforcement case. It shows that counties that assess interest, on average, have significantly higher arrears per case statistically than counties that do not assess interest. As evident in Exhibit 11, the difference in arrears is $1,087 to $1,386 per case depending on the time period examined. On the one hand, this is consistent with the theory that interest inflates arrears (i.e., counties that assess interest have higher total arrears than counties that do not). On the other hand, since the interest-assessing counties do not always assess interest, it may suggest that these counties have higher arrearages to begin with due to a more difficult caseload. There may be other reasons that also explain these differences.

The average percentage of enforcement cases with arrears is also compared between counties assessing interest and those that do not in Exhibit 11. This comparison shows no statistically significant differences.

One of the rationales for assessing interest on child support arrears is that it puts child support on a par with other debts the obligor may owe. Moreover, it is commonly believed that interest motivates obligors to stay current. If this is true, we would expect to see a higher proportion of the current support paid in counties and states that assess interest than in those that do not. Exhibit 12 shows the differences between Colorado counties that assess and do not assess interest on the percentage of current support paid. It should be noted that percent of current support paid is also a Federal performance indicator.^[11]

Exhibit 12 shows that counties that assess interest generally perform less well on the Federal performance indicator, percent of current obligation paid. The percent of current support paid between January and October 1999 is 49.8% in counties that assessed interest as of October 1998 and 54.0% in counties that did not assess interest as of October 1998. This constitutes a gap of 5.8 percentage points in current support collected, which is statistically significant, but in the opposite direction that the theory projects. The gap is somewhat narrower for the other time periods considered and in some instances it is statistically insignificant. Nonetheless, the evidence, that shows that current support paid is not more in interest-assessing counties than counties that do not assess interest, does not support the theory that interest increases payment of current support.

A similar comparison between states that assess interest to those that do not, shows that interest-assessing states collect 51% of their current support on average, whereas non-interest-assessing states collect 52% of their current support on average. The difference is not statistically significant. Again, the fact that evidence shows that current support paid is not more in interest-assessing states than states that do not assess interest, does not support the theory that interest increases payment of current support.

In addition, we caution using the comparisons in Exhibits 12 to suggest that interest decreases current support paid. These bivariate comparisons do not identify what is the cause and what is the effect. For example, if interest is the cause and decreases in the percentage of current support paid is the effect, than interest could be interpreted as decreasing payment of current support. On the other hand, counties with lower percentages of current support paid may assess interest as an attempt to increase overall collections. This issue of cause and effect is analogous to the issue of whether visitation increases child support payments or whether child support payments increase visitation. It is also as difficult to determine.

Several of the counties provided additional comments in the survey about the effectiveness of interest as a negotiation tool.

Interest is a good tool for negotiation; we don't want to lose that.

Let counties keep interest as a negotiation tool; interest is a good tool, it's fair.

The survey results indicate that interest is used as a negotiation tool infrequently, but when it is used it can net big results. Interest was used as a negotiation tool 26 times during October 1999, by the six counties that participated in the negotiation survey. In somewhat less than half of these incidences interest was used to negotiate lump-sum payments that averaged $6,297 and ranged from $784 to $16,000 per payment. One negotiation that was outside the study period timeframe resulted in a lump-sum payment of $40,000. Part or all of interest was waived in these negotiations.

Interest was also used to negotiate other things beside lump-sum settlements. Some of the negotiations involved determining the amount of retroactive support and payment of current support. One county uses interest in conjunction with drivers' license suspension to guide delinquent obligors into a workforce program. Unemployed and underemployed obligors subject to drivers' license suspension who participate in the county workforce program have their interest waived for six months if they continue to participate in the program.

The inability of some obligors to pay child support has become a national issue. Dr. Elaine Sorenson of the Urban Institute estimates that 16-32% of young noncustodial fathers not paying support are impoverished.^[12] In a report to the legislature exploring options for child support arrears forgiveness and pasthrough of payments to custodial parents, Minnesota found that 50% of its obligors with arrears have gross monthly incomes of $1,500 or less.^[13] Furthermore, 20% of the obligors with arrears have gross monthly incomes of $500 or less; their arrears averaged $3,055. This report also estimates that it would take an average of 8 years for an obligor to pay his or her arrears assuming he or she pays the maximum amount allowable. (Minnesota's income withholding statute allows an additional 20 percent of current support to be withheld to pay off arrears.)

This leads us to question why we are assessing interest when the obligor cannot even pay arrears. In fact, could the assessment of interest push some obligors away from paying because they never feel they can get caught up in their child support? We explore this empirically by comparing the percentage of arrears cases with a payment between counties assessing interest and those not assessing interest. Theoretically, interest charges would drive some obligors underground, so there would be a smaller proportion of paying cases in counties that do than in those that do not assess interest.

Exhibit 13 presents the results of this comparison. It shows that generally counties that do not assess interest have a higher percent of arrears cases with payment than those that assess interest. For example, during January-October 1999, 61.8% of the arrears cases in counties that assessed interest as of October 1999 had a payment. In comparison, 65.2% of the arrears cases in counties that did not assess interest as of October 1999 had a payment. The difference is statistically significant with a confidence of 95%. For other time periods, the difference is statistically significant with 90% confidence.

In part, the findings in Exhibit 13 provide some weak evidence to support the hypothesis that some obligors stop paying due to interest. The evidence is weak because, the gap is not always statistically significant; and, when it is statistically significant, it is usually significant at a 90% confidence level rather than a 95% confidence level.

In determining whether notification increases child support collections, we rely on the experiences of other states. Massachusetts reports that its annual notification, which was implemented for the first time in 1999, resulted in payments from 2,000 obligors averaging $400 each. Virginia's notification (1995) resulted in payment from 27,000 obligors who were not previously paying. After receipt of the notice, $12 million in arrears was collected from these obligors ($444 each on average).