Chapter 1

Estimating Growth of The Program Utilization Rates

Chapter 1

MEDICAL LEAVE INSURANCE (FAMLI) PROGRAM- PHASE II

In the second stage of Phase II, we are tasked to conduct an analysis of expected program claims and administration experience by studying and providing projections on:

• Expected volume of claims made on the state trust fund by employees employed in the state with predictions on the first year by month or quarter (2026), projections on at least an annual basis for later years (e.g., 2027 – 2030), and projections with break-downs of projected claims by leave type (medical, family care, birth of a child, military exigency)
• Expected cost and duration of claims with details by claim type (medical, family care, birth of a child, military exigency)
• Expected employer opt-out behavior, based on available information and data
• Expected self-employed individual opt-in behavior based on publicly available information from states that have mandated FAMLI benefits, based on available information and data.

1.1 Estimating the Volume of Claims in the First Year (2026)

To predict the volume of claims on the state trust fund by employees employed in the state by month or quarter in the first year (2026), we consider two potential trends, the seasonal trend (e.g., more claims in some months than others) and the trend in the first year after the payment starts (e.g. higher or lower claims since the program starts). We use data in other states to estimate the influence of these two trends. Table 1.1. shows the number of quarters that we used for our estimates in this chapter.

Table 1.1. Number of Quarterly Observations from Other States.

Note: For California, the dataset pertaining to bonding and family care spans from Quarter 3 in 2004 to Quarter 2 in 2022.  Meanwhile, the dataset concerning medical claims encompasses the period from Quarter 1 in 1974 to Quarter 2 in 2022.  Our seasonality analysis incorporates all the data available within these specified time frames. For the first-year trend analysis, we focus on the data from the inaugural year of the family care program, 2004, when payments for family care and bonding claims were initiated.

---

 

In our analysis, we first examine the seasonal effect. In doing so, we conducted two types of analysis to examine the seasonal effect. First, utilizing the data from four states (California, Rhode Island, and Washington), we examined whether there exists a seasonal effect for the three types of leaves for which data is available, including medical, birth of child/bonding and family care. Seasonal data for different types of leaves may vary across states. Quarterly data from DC is only available for submitted claims and therefore not included in the analysis.

We present the regression results examining the seasonal effects in Table 1.2. The results show that the dummy variables representing seasonality are not statistically significant. This suggests that the number of approved claims does not exhibit seasonality. To further validate this finding, we conducted additional analysis using data from California, where the dataset encompasses the largest number of seasons and takes advantage of the dataset’s longitudinal nature. A seasonality test was employed to assess whether seasonality significantly affects the number of approved claims each season. The test results confirm that seasonality does not exert a statistically significant impact on the number of approved claims. We further examined the seasonality on the monthly basis. Appendix – Figures 1.1-1.3 show that the number of approved claims does not show monthly seasonality either.

 

Table 1.2. Estimation of Claim Seasonality

Note: (1) Estimations are based on data on the number of approved claims from three states (CA, RI, and WA). The reference group is Season 4. (2) *p<0.10, **p<0.05, ***p<0.01.

---

Following this, our attention turns towards analyzing the number of approved claims after the initiation of program payments in other states where claim data is available. This analysis is conducted separately for medical, bonding, and family care claims for the first 12 months.

 

For medical claims, our analysis was conducted using data from California, Washington, and Rhode Island. As depicted in Figure 1, in Washington (WA) the number of medical claims, normalized based on the first 12-month average, increased significantly over the course of four months, followed by some fluctuations in months 4-7. Then it stabilized in the subsequent months. In contrast, data from California (CA) and Rhode Island (RI) exhibited a relatively flat monthly pattern.

This divergence in trends can be attributed to the fact that the paid medical leave program has been in existence for more than one decades in CA and RI, different from that in Washington: the claim data available for analysis starts from July 2004 for California and January 2014 for Rhode Island. With years’ experience in managing and finetuning medical leaves, as well as increased awareness of the availability of the paid medical leave program, the monthly variations reached stability overtime.

Figure 1.1 Trend in Approved Medical Claims in the First year (normalized based on the 12-month average)

Note: The index is calculated based on the average number of approved claims in the first 12 months, which is obtained by dividing the number of approved claims each month by the average in the first 12 months.

 

For bonding leaves and family leaves, we have data available for trend analysis again from CA and RI. It’s important to highlight that in the case of RI, the number of claims for the first four months was documented as a sum of the four months. Consequently, our RI data for Month 1 – 4 is represented as ¼ of the total number each month, i.e., evenly distributed, as we do not have a better way to assume differently. For Month 5-12, the data was recorded on a monthly basis. The summary of the data is presented in Table 1.3.

 

STOPPED FOR DEMO PURPOSES