Computable General Equilibrium Modeling and Its Application

Abstract

This chapter details Step 4 of the research framework: CGE modeling. CGE model simulations are run for each of the multiple random draws for each different hazard scenario. Relevant Direct Impact values are input into the USCGE model of the US economy, which captures the combined and interactive effects of these impacts through price changes and substitution effects across multiple economic institutions – 58 sectors, 9 household groups, government institutions, and international traders. GDP and employment impacts are generated for each of these multiple scenarios, and where relevant the Economic Structure of the impacted region is also factored in by scaling the national average results across three different example regional economy structures to render four times the number of original unique GDP and employment combination results.

Appendix 4A: Calculation of Input Data for Mild and Severe Influenza Outbreaks

We collected epidemiological data on pandemic influenza from various studies in terms of total population infected, the number of people that seek medical attention (either outpatient medical treatment or hospitalization), and the number of people that die from the influenza. Table 4.12 presents the data we obtained from the literature.

Table 4.12 Population infection, medical attention, hospitalization, and death data gathered from various studies

To determine the parameters for the two influenza outbreak scenarios, we first convert the data in Table 4.12 into infection rates, outpatient medical treatment rates, hospitalization rates, and death rates. In Table 4.13, we compute the rates of outpatient medical treatment, hospitalization, and death with respect to the total number of people infected. However, we find that different studies use different definitions in terms of the population being “infected” or “attacked”. For example, in Congressional Budget Office (CBO) (2006) and Verikios et al. (2010a, b) studies, “Attack Rate” is equivalent to “Infection Rate”. However, in Center for Disease Control (CDC) studies (2006, 2010), “Attack Rate” relates to the number of clinically ill cases, which are defined as “cases in persons with illness sufficient to cause an economic impact (e.g., half-day off work)”. According to the CDC definition, “infected persons who continued to work were not considered to have a clinical case of influenza.” The difference in the definitions regarding population infected (or attacked) makes comparison difficult across the studies. Therefore, in Table 4.14, we computed the rates on a consistent basis with respect to the total population.

Table 4.13 Population infection rate, and medical-treatment/infection, hospitalization/infection, death/infection ratios

Table 4.14 Population infection rate, and medical-treatment/population, hospitalization/population, and death/population ratios

Based on the range of ratios presented in Table 4.14, we determine the parameters to be used for the two outbreak scenarios in our analysis in Table 4.15. For the Attack Rate, based on the data presented in Table 4.14, we calculate the average of the lower-end percentages and the average of the higher-end percentages, and use them as the rates for our Mild Scenario and Severe Scenario, respectively. For Outpatient Medical Treatment Rate, we first eliminate the seemingly low estimate from Verikios et al. (2010a, b) as an outlier, and use the next lowest rate and the highest rate in Table 4.14 as the rates to be used for our Mild and Severe Scenarios, respectively. For the Hospitalization Rate and Death Rate, we first eliminate the outlier estimates (again from Scenario 1 of Verikios et al. 2010a, b) in Table 4.14, and then use the average of the lower-end rates and the average of the higher-end rates for our Mild and Severe Scenarios, respectively. For the Hospitalization Rate and Death Rate of the Mild Outbreak Scenario, we first eliminate the outlier estimates (again from Scenario 1 of Verikios et al 2010a, b) in Table 4.14, and then use the average of the lower-end rates. For the Hospitalization Rate and Death Rate of the Severe Outbreak Scenario, we used the rates of Class 5 pandemic influenza from Reed et al. (2013) and Meltzer et al. (2015). Since not all the studies based on which we collected data provide separate health outcome estimates by age group, in Table 4.15, we computed the rates for the entire population.

Table 4.15 Assumptions on epidemic parameters (all percentages are calculated with respect to U.S. total population)

4.1.1 4.A.1 Without Vaccination

By applying the epidemic parameters presented in Table 4.15 to the current U.S. population, we obtain the total number of people infected with symptoms, seeking outpatient medical treatment, hospitalized, and died, respectively. We then further break down these estimates among three age groups (0–17, 18–64, and 65+) based on data from CDC (2006, 2010) and Molinari et al. (2007) with respect to the mix of people from each age group in different health outcome categories. Table 4.16 presents the health outcome results for our two scenarios.

Table 4.16 Health estimates (number of people)

Table 4.17 presents the per-person lost productivity in days calculated based on the data from Molinari et al. (2007). Note that the analysis in Molinari et al. (2007) was primarily focused on seasonal influenza. We also compare the assumptions of per-person productivity loss in Molinari et al. (2007) with those in Meltzer et al. (1999), which focused on more severe outbreaks (e.g., influenza pandemic). The only notable difference is in the assumption of lost productivity for self-cured people between the two studies. Meltzer et al. (1999) assumed that the number of working-days lost for self-cured people is the same as working-days lost for people that received medical treatment (but no hospital stay). In this study, we decided to adopt the assumptions in Molinari et al. (2007), i.e., lost productivity for self-cured people is lower than people receiving medical treatment in both influenza outbreak scenarios. However, we note that as a sensitivity test, if we increase the productivity loss for the self-cured people by 100% for the Severe Outbreak scenario, the reduction in labor workforce participation will increase by only about 15% for the Severe Outbreak scenario.

Table 4.17 Per person lost productivity (days)

The workday losses due to own illness for the 18–64 age group for the two pandemic influenza scenarios are presented in Table 4.18. For each health outcome category, we multiply the number of patients in Table 4.16 by the corresponding number of lost working days per person, and adjust for the labor force participation rate of 62.8 % (BLS, 2014).

Table 4.18 Workday losses due to own illness (days)

Table 4.19 presents the workday losses due to caring of sick family members. These include caring for sick children in the 0–17 age group, sick spouse in the 18–64 age group, and sick elderly family members in the 65+ age group. When we calculate the workday losses due to the care of sick children, we adopt similar assumptions used in Dixon et al. (2010): (1) for any day that the children are sick at home, one full workday is lost for the caring parent; and (2) for any day that the children are hospitalized, half workday of the caring parent is lost. We also adjust the workday losses down according to the percentage of families with children that have no non-working parent. Based on the U.S. Census Bureau data, this percentage is 63 % in 2013 (U.S. Census Bureau 2014a). When we calculate the workday losses due to the care of sick spouses, we first apply the percentage of total population married with spouse present. This percentage is 50.7 % in 2013 (U.S. Census Bureau, 2014b). We next assume that 50 % of working people with spouses will decide to take sick days to care for their sick spouses during the outbreaks of the influenza. We further assume that: (1) if people in the 18–64 age group only experience flu symptoms, but do not seek medical treatment, no caring from their spouses is needed; (2) for any sick day due to outpatient medical treatment, half workday of the caring spouse is lost; (3) for any day that is lost because of hospitalization, half workday of the caring spouse is lost. When we calculate the workday losses due to the care of sick elderly family members, we first assume that 41.7 % of the sick old people will receive cares from their family members. This is based on the data presented in the National Alliance for Caregiving (2009) report, which indicated that about 43.5 million adult Americans provide unpaid family cares for someone 50+ years of age. This represents 41.7 % of the 50+ population. We also use the labor force participation rate to get the percentage of unpaid family caregivers that are in the labor forth. We then further assume that: (1) if people in the 65+ age group only experience flu symptoms, but do not seek medical treatment, no caring from their family member is needed; (2) for any sick day due to outpatient medical treatment, half workday of the caring family member is lost; (3) for any day that is lost because of hospitalization, half workday of the caring family member is lost.

Table 4.19 Workday losses due to caring of sick family members (days)

Table 4.20 presents the per-person medical expenditure for the three age groups calculated based on the data presented in Molinari et al. (2007). Because of lack of better data, we assume the per-person medical expenditures are same for the two scenarios. We believe the difference in the severity of the two influenza outbreak scenarios are mostly captured by the difference in the number of people received outpatient medical treatment, hospitalized, and died.⁴

Table 4.20 Per person medical expenditure (in 2003$)

In Table 4.21, we compute the total medical expenditures for the two pandemic influenza scenarios by multiplying the per person medical costs by the number of people seeking medical attention presented in Table 4.16. Note that in the mild scenario, the largest expenditures are for those who were hospitalized, while, in the severe scenario, those who were hospitalized and died incur the largest expenditures.

Table 4.21 Medical expenditures (in billion 2003$)

4.1.2 4.A.2 With Vaccination

In the following analysis, we incorporate into consideration the effects of one mitigation strategy, vaccination, on the economic impacts of the two major influenza outbreak scenarios. We make the assumptions on the vaccination coverage based on the vaccination coverage rate in the 2012–2013 flu season. We assume that compared with a regular flu season, the vaccination coverage rate for each age group will be 5 % higher in the mild influenza outbreak scenario and 10 % higher in the severe influenza outbreak scenario. The vaccination coverage assumptions are presented in Table 4.22.

Table 4.22 Vaccination coverage

The effectiveness of vaccination, which is defined as the percentage reduction in the number of illness cases in each age group, is calculated as the average of the influenza vaccination effectiveness in the U.S. from 2005 to 2013 (see Table 4.23), which include both non-pandemic and pandemic (H1N1) flu seasons. Since the age group designation in Table 4.20 is slightly different from the one we used in our analysis, we computed the population weighted average vaccination effectiveness rate for age groups 0–4 and 5–19 in Table 4.23 to be used for the age group 0–17 in our scenarios. The effectiveness rate for the age groups 20–64 in Table 4.23 is used for the age group 18–64 in our scenarios. The final vaccination effectiveness rates we used in our analysis are: 0–17 (52 %), 18–64 (50.6 %), and 65+ (34.9 %). In addition, we assume the same vaccination effectiveness rate across different categories of health outcome (i.e., outpatient visits, hospitalization, and death) within each age group.

Table 4.23 Vaccination effectiveness

The costs associated with vaccination are calculated mainly based on the data from Prosser et al. (2011). Table 4.24 presents the vaccination-associated costs by cost category and by age group. In Prosser et al. (2011), the authors assumed that two doses of vaccination for children aged 6 months to 10 years and one dose for individuals in other age groups are required for a full vaccination. The authors also made detailed assumptions regarding the variation in costs associated with the vaccination settings (mass vaccination setting, such as schools or physician office setting, vs. physician office setting) for different age groups. For the physician office setting, the authors also made the distinction between vaccination at an existing visit and vaccination-initiated visit. The cost associated with vaccination administration and the amount of time people spend on travel, waiting, and vaccination vary by vaccination setting and whether or not the physician office visit is vaccination-specific.

Table 4.24 Vaccination cost data

In Table 4.25, we computed the vaccination costs for the three age groups used in our analysis based on the data presented in Table 4.24. We use population in each sub age group in Table 4.24 as weights to calculate the costs for the broader age groups. The high-risk ratios for age groups 18–49 and 50–64 from Molinari et al. (2007) are used to aggregate the costs of high-risk and non-high risk sub groups within the same age group of these two age groups. The costs associated with any potential vaccination side effects are also computed based on the data from Prosser et al. (2011). In the last column of Table 4.25, we translated the costs associated with travel, waiting and vaccination time into the number of lost work hours using the hourly wage rate, $20.62, used in Prosser et al. (2011).

Table 4.25 Per person vaccination costs and associated lost work hours

Table 4.26 presents the health outcome results with vaccination. The workday losses associated with own illness for the 18–64 age group, caring for sick family members, and receiving vaccination are presented in Tables 4.27, 4.28, and 4.29, respectively. The incremental medical expenditures due to the treatments of illness and vaccination (including costs of vaccine dose, administration, and side effects) are presented in Tables 4.30 and 4.31, respectively.

Table 4.26 Health outcome estimates with vaccination (number of people)

Table 4.27 Workday losses (days) for own illness (with vaccination)

Table 4.28 Workday losses (days) for caring sick family members (with vaccination)

Table 4.29 Workday losses (days) due to vaccination

Table 4.30 Medical expenditures (with vaccination) (in billion 2003$)

Table 4.31 Medical expenditures associated with vaccination (in billion 2009$)