Partially different? The importance of general equilibrium in health economic evaluations: An application to nocturia

Abstract Both the human capital approach and the friction cost approach are frequently used to quantify the productivity costs associated with illness, disability or death in health economic evaluations. In this paper we argue that these approaches have one major, but common shortcoming: they only capture partial equilibrium (PE) effects and therefore underestimate the true potential productivity costs associated with health conditions. They neglect the sizable, indirect, ripple effects in the economy captured by general equilibrium (GE) models. To demonstrate our point, we compare a traditional PE with a GE approach for the application to nocturia, a condition characterized by the need to frequently wake up at night to urinate. Nocturia is associated with substantial impairment of daytime functioning and work productivity. We employ large‐scale United Kingdom (UK) employer‐employee survey data to estimate the prevalence and productivity loss. These estimates are then used as shared inputs to drive both approaches. We find that the traditional PE approach underestimates the annual productivity cost of clinically relevant nocturia by around 16%. We propose a generalized GE/PE multiplier to approximate the GE effect for other health conditions. Our findings stress the importance of accounting for sizable GE effects when conducting health economic evaluations.

is endowed with labor and capital which it provides to firms in exchange for income. 32 1 www.gams.com 2 https://fred.stlouisfed.org/series/GDPDEF 3 https://data.oecd.org/conversion/purchasing-power-parities-ppp.htm#indicator-chart ington framework, which is commonly used in CGE modeling to allow for the cross-hauling of the same 34 goods (Armington, 1969). As Lanz and Rutherford (2016) explain in more detail, imports from different 35 countries include transportation services, which enter on a proportional basis, to reflect differences in 36 unit transportation margins across different goods and trading partners. The Armington composite 37 therefore involves trading both imported goods and associated transportation services.

38
Finally, as Figure 1 illustrates in red font, we capture the effect of nocturia as a change in the 39 effective labor-supply -discussed further in subsection 1.4.

41
Omitting country index r, Figure 2 provides more detail on the firms' production function in the product 42 market. The four main production sectors, discussed previously, are perfectly competitive economic 43 sectors that produce goods using a multi-level, differentiable, constant return to scale (CRS) production 44 function Y j = f (K j , N jq , L j ). Each sector demands the following inputs: capital K j , effective-labor L j , 45 and intermediate inputs N jq that are produced by sector q.

46
On the left-hand of Figure 2, firms demand input factors: skilled and unskilled labor and capital which they obtain from the factor market in Figure 1. At the lowest level, we aggregate skilled and 48 unskilled labor with an inelastic substitution elasticity σ = 0.5 that characterizes their differences in 49 skills. In the next level, aggregate-labor L j and capital K j are aggregated into a value added using a 50 Cobb-Douglas function, as usually applied in many macroeconomic models.

51
The right-hand side of Figure   In each country, the representative agent RA r is endowed with capital and labor, which they provide 60 to firms in exchange for income, and also collect (provide) taxes (subsidies) on domestic goods, and 61 tariffs on imports and export. With this income, they maximizes a multi-level CRS function, which   we compare the baseline with the counterfactual scenarios to obtain the cost of nocturia.

85
An increase in effective-labor supply is manifested through the removal of prolonged periods of 86 sickness or levels of presenteeism that reduce the effective-labor workforce. In our baseline scenario, 87 we normalize the productivity to the current nocturia prevalence levels by E = 1. In the counterfactual 88 nocturia threshold scenarios (v = 1+, 2+), we treat ("eliminate") nocturia for patients that have one or 89 more voids and two or more voids which raises the effective-labor productivity by E v = E + e v . Finally, 90 we compare the baseline with the counterfactual scenarios to obtain the productivity cost of nocturia.

91
The parameter e v is obtained by with prevalence rate θ v and work impairment α v , discussed and estimated econometrically in the 93 paper.

94
As there is uncertainty related to the parameter inputs, we further test our model assumptions by The table below provides the full regression result including covariates by nocturia definition.
100 Table 1: Associations between nocturnal voiding and work impairment due to absenteeism or presenteeism (% working time lost) (1)