A Framework of the Economic Model in HIA

A number of comprehensive reviews are available covering a range of studies on various aspects of HIA in non-health policy areas, including implementation issues (Danneberg et al. 2008, Mindell et al. 2008, Birley 2002). A recent review on HIA stated (page 299, Metcalfe and Higgins 2009):

“…By its very nature, HIA has been developed as a method of informing

healthy public policy and seeking to predict the health consequences of implementing different policy options. It is therefore a support tool for decision-makers to address both potential health impacts and health inequalities in identified policy, programmes or projects.”

However, there is little review-level evidence available to demonstrate if and how the HIA approach informs the decision making process and, in particular, if it improves health and reduces health inequalities (ACHEIA 2004, HDA 2002). The present decision making in regard to public investments in Australia does not quantify health impacts and costs systematically, mainly due to the lack of an appropriate approach and/or model. Therefore, this study proposed that an economic

model be incorporated into the HIA framework, to assist policy decision makers in bridging this knowledge gap.

Figure 4.2: The Proposed Economic Model in a HIA Framework

Here, HI = Health Inequality, Edu = Average education attainment of the population; Y = Average income level of the population; H0 = Current health status of the population; UnEmp = unemployment level of the population; AQ = Air Quality; M = use of health care services; X = household consumption excluding health care.

The economic model developed is incorporated in the HIA framework by establishing an association between identified economic and non-economic factors using the Health Production Function (HPF) concept in order to explain the linkage between policies and health of target population group. The proposed economic model is a part of a HIA framework, which is presented in Figure 4.2.

Health Inequality Mortality

Health Production Function: H = f (M, HI, AQ, X, Edu, Y, H0)

Household Choice For Better Health HI = F (Edu, Y, UnEmp, AQ,

Race, Location, Gender)

Morbidity Health Cases

Economic Value

Economic Cost of estimated Life Lost

Economic Cost of estimated Morbidity

Health Impacts: Total Cost

Epidemiological Evidence: RR Ratio of disease Health Exposure

In the above HIA framework, the economic model is shown by two equations - a Health Production Function (HPF) and a Health Inequality (HI) equation. The framework presented above starts with a damage function approach using dose- response relationships to estimate the health impacts (health cases – mortality and morbidity charts) of a policy initiative (e.g.: air pollution reduction). An economic model can be then used to determine the factors affecting the health cases and to estimate health inequalities. This economic model is finally used for placing monetary values on these health effects - by estimating a HPF, where health is determined by HI, important macroeconomic and demographic variables. A literature review has shown that usually HIA studies use the cost-of-illness (COI) or willingness to pay (WTP) approach to estimate the monetary values of reduced illness (or, morbidity) (Bellavance et al. 2007, BTRE 2005, Viscusi and Aldi 2003). In the case of mortality, estimates are generally based on meta-analysis of values of statistical life (VSL) to reduce premature mortality.

It is now widely recognised that a variety of factors can affect health and health inequalities outside of the formal health services and structure (Mahoney and Durham 2002, Milner and Marples 1997, Dahlgren 1995). It has also been suggested that HIA offers a practical way to consider health and inequalities within the decision making process at policy and other levels (Mahoney and Durham 2002, European Centre for Health Policy 1999).

Based on the evidence available, as of now, from the studies conducted in Australia and overseas, the study identified factors affecting public health with reference to specific policies (e.g., transport policy). The aim of the study, in this context, was to examine whether these identified variables could explain the public health status of the target population. For this purpose, an economic model was

developed incorporating the two-way relationship between public health status and policy decision (parameters) by specifying a Health Production Function (HPF).

In the present thesis, HPF is not intended to be an equity, social justice based concept. Rather, HPF is used as a concept that identifies the important health and non-health factors that are responsible for the current health status of an individual or a representative household in the population. In this study, the existing health inequality - that is, the disparities in the prevailing health conditions of the target population - is shown as one of the important factors contributing to population health, which can be reduced through a range of non-health policy interventions, such as, transport emission abatement policies.

Similarly, non-health policy of active transportation investments that increase land value as well as generate new jobs, also can lower existing health inequality indirectly through increased household access to health care services as new health care investments happen and household income rises. However, higher land value may, in effect, increase health inequality further by displacing lower income households from their land as a result of gentrification induced by transport investments (Lin 2002; Kahn 2007, Pollack et al. 2010, Dominie 2012, Zuk et al. 2015). Consequently, displaced low income households are likely to have limited access to both a healthy neighbourhood and labour market as they move to a cheaper area. In this context, the HPF can capture these impacts of active transportation investments by including appropriate indicators of income inequality, demand for health care services, employment level, house prices or land value.

The Health Production Function concept used in this economic model was grounded in Grossman’s HPF approach, developed in 1972, that showed a functional relationship between health status (output) and its determinants (inputs). In fact, Becker (1965) first introduced households as “producers of commodities” in his theory of the allocation of time. Accordingly, this study presents an economic model of health behaviour underpinned by a consumer choice approach, where households are assumed to produce health output (Grossman 1972) which provides:

 A health oriented choice model, where individuals are viewed as producers of health capital goods (H).

 Individuals maximise their utility (U) from consumption of H and X, other non-health consumption goods: U = U ( X, H )

Subject to the HPF: H = f (M, alpha, delta)

Where: M denotes medical care; alpha is an environmental indicator like air quality (AQ); and delta represents economic variables like education, income, unemployment and lifestyle factors.

However, equity in population health is a core policy outcome in the HIA framework and hence, existing population health inequalities (HI) need to be considered as an endogenous variable in the proposed HPF. In this context, one important criteria of the policy effectiveness is a reduction in the observed health inequality. Thus, the HPF can be expressed as H = f (M, HI, AQ, X, Edu, Y, H0)

Here HI denotes the estimated health inequality of the target population, which plays an important role in determining the population health (H) in a HIA framework. A review of empirical studies indicates a strong association between health inequality and socioeconomic indicators – such as employment status, income

and health (Lahelma et al. 2005, Stronks et al. 1997). There is strong evidence to support a link between less education and poorer health (Laplagne et al. 2007, Stanwick et al. 2006, Turrell et al. 2006) indicating higher education policy is likely to reduce health inequalities by improving population health. Studies have also found an association between the level of unemployment and a range of health concerns, including low self-rated health, cardiovascular disease, and drug and alcohol abuse (AIHW 2007, Benach and Muntaner 2007, Cummins et al. 2005, Morrell et al. 1998). Therefore, HI is expected to depend on the income, employment and education levels of the population, in addition to their current health status, demographic and environmental factors.

HI = F (Edu, Y, UnEmp, AQ, Race, Location, Gender)

Here, Edu = Average education attainment of the population; Y = Average income level of the population; H0 = Current health status of the population; and

UnEmp = unemployment level of the population.

In a health production function, the individual’s health choice is usually emphasised; whereas in a public policy context, policy impacts are supposed to cover a target group of people, defined by the policy coverage. This means that the macro- micro linkage needs to be identified while estimating the HPF for policy impacts analysis. This study proposed to address this by estimating an aggregate HPF that showed a long run relationship between population health and its identified determinants.

The choices of health production output variables are usually concentrated on life expectancy, mortality rate and related variables. The United Nations consider life expectancy, infant mortality rate and children’s mortality rate to be three major

indicators of the health achievements of a country or a region. In estimating the health system performance of its member countries in 1997, WHO took disability - adjusted life expectancy (DALE) and children’s mortality rate as indicators of the general health achievements (WHO 2000). Life expectancy, DALE and mortality rate are all health output variables widely applied in the Health Production Function studies (Zhang et al. 2006). In this study, life expectancy and mortality rates were adopted as the health production output variables.

This paper defined the health production input variables in the narrow sense as medical and health inputs and health spending being the main inputs in the health production system (Grossman 1972). As direct inputs into the health production system, doctors (with physicians and nurses) per capita, hospital beds per capita and medical equipment availability can be considered as the factors directly influencing health production output (Retzlaff-Roberts et al. 2004). As there may not be particular data on medical equipment, the number of health technicians per 1000 people and the per capita health spending (including government budget expenditure, public health spending and private health spending) can be used as input variables.

In the case of socio-economic factors, strong evidence exists showing a link between less education and poorer health (Laplagne et al. 2007, Stanwick et al. 2006, Turrell et al. 2006), as mentioned previously. Education can lead to better quality jobs, and this may be a protective factor against poor health. A study by Australian Institute of Health and Welfare (2007) concluded in support of a positive association between the level of unemployment and a range of health concerns, including low self-rated health, cardiovascular disease, and drug and alcohol abuse. A similar association was also reported by other studies (Benach and Muntaner 2007, Cummins et al. 2005, Morrell et al. 1998), which found unemployment to be

associated with low self-esteem and mental health problems. Therefore, education level was included as an explanatory variable in both HPF and HI equations in this study’s proposed model.

The model may be estimated using an econometric simultaneous equations system based on available socio-economic, health status and macroeconomic data. For example, a two stage ordinary least square (2SLS) model could be constructed to identify the factors that determine public health gains associated with a policy intervention. In evaluating the explanatory variables in this economic model, this study was interested in the presence and nature of statistical relationships and the statistical significance of the exhibited association among variables. This study was also interested in the practical importance of the statistical relationship, measured by the elasticity of the explanatory variables. In order to obtain elasticities directly, all explanatory variables could be transformed into their natural log forms. The coefficients obtained for each explanatory variable could be then read as the elasticity of the public health variable (H) with respect to the explanatory variable.

Estimated health production functions for the target population can then be directly used to calculate the economic costs and benefits charts of identified health impacts of the policy intervention in question. The next step would be to conduct a sensitivity analysis to test the robustness of the economic model developed.