Tackling the Stiglitz Report: Measuring Social Progress and Economic Performance under Various Climate Policy Scenarios

This paper attempts to connect the measurement of social progress from the Stiglitz report and climate change mitigation (CCM) by the Intergovernmental Panel on Climate Change (IPCC) assessment reports. Each report has been addressed insufficiently on the issue, although both reports have common in-terests in development patterns and pathways for the economy, humanity, and society. This study used our original integrated assessment model and applied for measuring various indicators for sustainable development, such as genuine savings (known as GS) and human appropriation of net photosynthetic primary production (HANPP). We expanded an analysis of sustainable development indicators of quality of life (QoL) and of the human development index (HDI) and introduced a modified quality of life indicator. These indicators expand on the “classical” GDP loss, which has been well analyzed in the majority of CCM literature. Our model’s main framework is based on the Regional Integrated model of Climate and the Economy (RICE) extended from Ramsey-Cass-Koopmans with a simplified climate model and added three original resource balance models with environmental consequences with a life cycle impact assessment (LCIA) model. We prepared various climate policy scenarios ranging from business as usual to economically efficient, CO 2 double stabilization, and targeting two degrees Celsius (DC). We believe this work has three contributions. First, in contrast with the World model by the Limits to Growth, our model has an economic foundation where genuine savings is introduced. Second, while the Stiglitz report only extrapolates the current troduces indicators of sustainable development in assessing climate policies.


Introduction
The Stiglitz report [1] rests on three main pillars for reforming global monetary systems after the financial crisis of 2007-2008-classical gross domestic product (GDP), quality of life (QoL), and sustainable development and environment (SD & E). GDP is the most used measure of economic activity globally. Although it primarily measures market production of economic activity in value-added terms, it is commonly used as a measure of economic well-being. Material standards of living, however, are more closely associated with measures of real income and consumption rather than production. Because of the various limitations (fully described in the report), alternative ways to measure social progress other than economic performance are presented in the report.
The measurement of QoL is categorized into three approaches: subjective well-being (SWB), capabilities, and welfare economics and fair allocations. SWB includes three separate aspects: life satisfaction, the presence (and absence) of positive feelings, and the presence (and absence) of negative feelings. Two representative measurement approaches are the World Value Survey (a qualitative survey) and the Gallup World Poll (a quantitative survey with a 0 to 10 quality-of-life scale). Quality of life is comprised of material living standards (e.g., income), health, education, personal activities (e.g., paid and unpaid work, commuting, leisure time, and housing), political voice and governance, social connection, environmental condition, and insecurity. These conditions can be functioning (people's doings, e.g., working and commuting, and people's being, e.g., healthy and educated) or freedom (e.g., political voice and participation). The human development index (HDI) has a capability approach in terms of a person's freedom to choose among the various combinations of functioning factors. The HDI is the representative indicator aggregating across those various domains, specifically health, education, and living standards. None of the qualitative in-depth investigations by IAMs, however, have been addressed in the report. Contrary to this, climate policy assessment has long been analyzed only by "classical" loss of GDP or the social cost of carbon (SCC) (calculated as the present value of discounted economic welfare loss by a marginal GHG emissions increase) [2]. These factors are categorized as "economic performance" in the report. This article tackles the projection issue under climate policy scenarios using the measuring indicators of QoL and SD & E in our IAM. The authors have long been studying the latter issue (i.e., measuring future projections of indicators in SD & E), especially on ANS (or GS) and HANPP 1 [4] [5] as a proxy of "carrying capacity" instead of EF 2 . Recently, Pezzey and Burke [10] addressed a "precautionary" approach by modifying the "highly aggregated" damage function in the original DICE model and deriving a recently well-defined SCC to amend ANS in the present day with and without the climate policy.
Compared with those results, the projection of the former (i.e., QoL indicators) has not been addressed to the authors' knowledge, a topic this article addresses intensively. We examine future projections of ANS and HANPP for the SD & E using well-defined HDI, modified indicators for QoL, and GDP for "classical" economic performance.
The organization of this article is as follows. Section 2 gives short reviews on the theoretical background and measurement issues for the economic and environmental indicators. Section 3 describes our IAM and calculation of the indicators. This is followed by results, discussion, and conclusions in Sections 4, 5, and 6, respectively.  [12], Stiglitz [13], and Hartwick [14] to more recent contributions by Weitzman [15], Pezzey [16], and Dasgupta [17]. Empirically, Pearce 1 Net photosynthetic production means the accumulation of biomass in plants [3]. Photosynthetic production is the production of organic compounds from atmospheric or aquatic carbon dioxide. It may occur through the process of photosynthesis, using light as a source of energy, or through chemosynthesis, using the oxidation or reduction of chemical compounds as a source of energy. 2 EF was excluded for three reasons. One was that we could not find an appropriate methodology to obtain the yield factor and the equivalence factor required to calculate EF using our model. On the contrary, the potential NPP to use to calculate HANPP is scientifically evident, as it is determined by temperature and precipitation. Another reason not to use EF was that EF is most appropriate for high-resolution analysis, such as those conducted at the country or mesh level [6] [7]. In our model with ten large regions could not calculate and present EF for such a high spatial resolution. The third reason is that previous studies have already presented the future trends of EF [8] [9]. and Atkinson (1993) [18] coined the term GS, succeeded by a series of contributions from Hamilton [19], Hamilton and Clemens [20], the World Bank [21] [22] [23], and Hanley et al. [24]. A detailed presentation of both theory and practical use of GS can be found in Hanley et al. [24].

Literature
GS is effectively the rate of change of the total wealth available in an economy.
This total wealth is understood as comprehensive wealth, that is, the economic value of all the capital stocks with and without market value in a given economy.
GS therefore integrates all of the changes (impacts) that alter the ability of an item to yield its services.
GS is grounded in the standard utility theory; as a result, it is easily incorporated into the DICE model [25]. Intergenerational well-being, V t , is expressed in Equation (2.1-1) by using instantaneous utility, U t , and the utility discount rate ρ: "Sustainable development" as presented in the WCED report [26] can be in- Two methods produce empirical estimates of the theoretical notions of "Genuine Savings" and "Comprehensive Wealth." The first method, Genuine Savings, which is outcome based, is used by the World Bank, while the second method, Comprehensive Wealth, which is capability based, has been used by the UN in a series of reports on inclusive wealth based on Arrow et al. [27]. See Hanley et al. [24] for more details. [10] offered a contribution that is quite close to the aim of this paper; they used the World Bank method but amended GS for the physical constraints associated with global warming and offered a more realistic estimation of the costs associated with uncontrolled climate change.

Carrying Capacity (HANPP)
Attention was first paid to HANPP in the 1970s to raise the concern over excessive human economic activities (Whittaker and Likens, 1973) [4]; that work triggered many quantitative analyses for HANPP. However, the results differed widely because of the application of different mathematical definitions of HANPP.
In the Special Issues of Ecological Economics in 2009, which included the paper by Erb et al. [5], HANPP was defined as the difference between the potential NPP and the actual level of NPP in land use, the same definition that Vitousek offered in 1986 [4].
NPP provides ecosystem services through agriculture and forestry, some of which can replace fossil fuel products (e.g., biofuels). Moreover, it also has a capacity to absorb exhaust emissions. NPP also serves as a buffer for waste products [5]. Since NPP is affected by changes in land use that have critical impacts on biodiversity, HANPP may be considered as a strong indicator of sustainability; it can express natural capital that is hard to substitute with physical capital.
HANPP also expresses resilience because it is related to the cycle of water, carbon, and nitrogen as well as the deposition of organisms in the soil that increases land productivity.

Quality of
where the EYS index (expected years of schooling) = ( ) ( ) Income index ln GNI cap ln 100 ln 75000 ln 100 This mathematical expression has well-known limitations. One limitation is the weighting among the three dimensions in applying the geometric mean; notably, that weighting the importance of the three dimensions implies is an ar-  [38]. The world economy is divided into n regions, and each region j is composed of identical individuals, who maximize utility through the consumption of a composite good: is the per capita utility of consumption in region j at time t. The parameter η is the elasticity of the marginal utility of consumption 3 . The total regional utility, ( ) , j t u c , is obtained by multiplying individual utility by , j t P 4 , the exogenously given population number for region j in time t. We then sum the regional total utility for all future time periods s over the time horizon T to obtain intertemporal well-being, , , where ρ is the pure rate of time preference, reflecting how future generations' well-being is taken into account. Each region is assumed to produce a single commodity, which can be used for either "generalized" consumption or investment as economic variables. The generalized consumption includes not only traditional market purchases of goods and services but also nonmarket consumption, such as enjoyment of the environment. Finally, regional-level intertemporal well-being, t V is maximized at the aggregate level via the function where t W 5 is the objective function weighted sum of social welfare, , j t V , for region j, t by Negishi weight, j Neg 6 . The use of Negishi weights means that the distribution of well-being is kept constant over time, preventing convergence in consumption levels.

Production
Gross output is determined by a nested production function, with capital, labor, and natural resources as inputs: where , j t A is the exogenously given total factor productivity term, where δ is the annual rate of capital depreciation. In line with our representa- 3 It also represents the curvature of the utility function, or the rate of inequity aversion, measuring the extent to which a region is willing to reduce the welfare of high consumption generation and to improve that of low consumption generation. 4 The given population number, , j t P , is taken from the SSP-2 scenario in order to coherently analyze climate change mitigation. The number is the largest among the five SSP scenario families [39], somewhat higher than the UN's [40] mid projection but still close to the central level compared with high and low projections.
tive agent assumption, we take population growth and technological change to be exogenous. Technological change in the model is divided in two parts: the exogenously given TFP and the evolution of the mix of inputs used in the production process.
( ) where , j t A , the TFP, is determined every period based on the exogenous TFP growth rate, , j t τ . Capital accumulation and natural resource inputs are then determined by maximizing the discounted utility flow over time constrained by the technology mix (the production function). Net output is then given by: The net output Equation (3.1-7) ties together the three components in Figure 1: The macroeconomic model in the red box determines gross output, based on the cost of resource acquisition, EXT are provided below. There is interregional trade of the final good, and trade is not balanced. Thus, the accumulated trade surplus/deficit of each region is not necessarily zero in any period, including the final period. The budget constraint for the representative agent in each region is therefore: With the imports,   Y that is selected is associated with a cost and a level of environmental impact in the blue and yellow parts of the model.

Costs of Production
The total cost of production, t TC comes from the three models of resource balance in the blue box ( Figure 1). These models tie together supply and demand to generate inventories. These inventories are then used as factor endowments in the production function ( The inventories generated by the resource balance models form the basis of the impact assessed here. The external cost is computed as: The external cost is best understood as a stock/impact/value relationship.   Three examples are provided as illustration. The impacts of global warming on human health are expressed by the relative risk increase due to a rise in the global mean temperature (T(t)) [48]. The economic impacts of land loss by sea level rise (SLR) (SLR(t)) are similar to those assessed by Fankhauser [53] and Tol [54]. Land use and land-use change are caused by biomass and food production, which can be expressed generally as LU(t). T(t) and SLR(t) are obtained from the same formulations in RICE 2010: total carbon emissions from fossil fuel combustion in energy systems [43], carbon released from land-use change [44] and exogenously given non-CO 2 GHGs.  [37], Tokimatsu et al. [55], Murakami et al. [56] and Kolstad et al. [57] for more details.

Data and Calibration
The discrete time step of the model is 10 years, and 10 regions are included: North The setting of the initial K value was obtained from the RICE 2010 model [42]. We used a nested CES production function inspired by Berndt and Wood [59] and Manne and Richels [60]  TFP was calibrated from data sources to fit the scenarios (level of production).
The form of function φ is increasing but diminishing in rate (

Scenarios
The model as presented so far is the baseline scenario. In this setting, all externalities are internalized, and all of the parameters are set at their base level. This is called the economically efficient scenario (Eeff). The incentives to reduce CO 2 emissions are based on their direct and indirect cost through

GS
We then computed GS ex post following the method used by the World Bank [22]. The method focuses on the definition of comprehensive investment as: In line with the theoretical definition in Equation ( Im was estimated using a power function defined using the World Development Indicators (WDI). GDP values were then entered into that function to estimate the value of investment in medical expenses per capita. This per capita value is then multiplied by the population size in time t to obtain the total investment value; • , j t Ie was estimated using a linear function defined using WDI. The , j t Y values were then entered into that function to estimate the value of investment in education per capita in year t. This per capita value was then multiplied by the population size in time t to obtain the total investment value; • , j t Inv was the natural resource stocks and inventories obtained from the LIME3 and RICE components of our model.
The shadow prices associated with produced, human, health, and natural capital were the optimal prices obtained from our model. The  1-11).
We have two sets of shadow prices, It should also be noted that due to the structure of the model, only We can now define Genuine Savings as the rate of change in total wealth, by 8 This stock and the respective investment are added in our computations to the World Bank methodology based on the suggestion of Arrow et al. [27]. first computing total wealth using the World Bank [22] method: where ρ is equal to 1.5 and C is defined as sustainable consumption, that is C minus , j t Im and , j t Ie . Gross genuine savings is therefore: This shows how wealth has increased between t and t + 1. We then adjusted this rate of change for population growth and technological progress (both exogenous in our model): With , 1 j t p + , the population growth rate 9 , and , 1 j t τ + , the technological progress growth rate, , 1 j t Wnt ± ∆ is the notation for the final fully adjusted rate of change in wealth, or the "GS rate."

HANPP
The denominator of HANPP is the potential NPP, determined by temperature and precipitation [75]. The numerator of HANPP is NPP consumption by human activities. NPP consumption includes the direct and indirect consumption of foods, trees, and their residues, as well as the potential loss of NPP due to land use and land-use change (LU and LUC), caused by resource production activities. The three levels of NPP in the numerator in HANPP were provided in the papers by Erb [45] and Vitousek [4] as follows: • Low estimate: Direct consumption (demanded quantity) of agricultural products (i.e., rice, wheat, corn), wood (i.e., logs, wood pulp, timber/boards, paper), and seafood eaten by humans and livestock; • Middle estimate: The harvested amount from agricultural land, grassland, forests, etc., that produces the direct consumption (i.e., low estimate). This level is calculated as the sum of the direct consumption and conversion loss (unused residuals); • High estimate: This is the sum of the middle estimate and the potential loss of NPP due to LU and LUC.
Direct consumption for the low estimate and the conversion loss in the middle estimate were calculated using our simplified land-use model. The potential NPP by 2100 that is needed to calculate the high estimate was obtained from the 9 As population enters both the maximum and the production function in the macroeconomic model, GS should be adjusted [17]. The literature offers two methods for this. First, Ferreira et al. (2008) [79] consider future population growth as a form of capital loss, as future total wealth should be divided among a larger number of individuals. GS are amended two ways to make up for the capital loss: a reduction of the discount rate and a wealth-based subtraction to the gross GS rate. Second, Arrow et al. [27] consider exogenous population growth as one of the external dynamics of the economy, just like technological change. The GS amendment is then to subtract the population growth rate in t. As our production-based computation of GS savings is the replica of the Arrow et al. [27] method, we use this adjustment. Note that our population growth rate is not constant between 10 years' time steps, but it is constant within those 10 years' intervals. Chikugo model [75], and it requires temperature and precipitation obtained from MAGICC/SCENGEN [76]. The changes in potential NPP due to LUC were calculated by multiplying damage factors for LUC with area change by LUC endogenously obtained from our land-use sub-model. This potential NPP is used as the denominator for the three estimates.
It should be noted also that our model does not have as high a resolution as the country or grid level using GIS, as seen in papers from the Special Issue of Ecological Economics in 2009 [5]. Instead, our model depicts the HANPP trend in line with the optimal economic development path.

Discussion
The model in this study is original; it diverges from similar studies and highlights the significance of applying multiple indicator dimensions, namely, "classical" GDP loss, SD & E in both WS (i.e., GS) and SS (e.g., HANPP), and QoL (e.g., HDI). Comparable studies explore future paths over the century of HDI by the World 3 model and the GS estimate by Pezzey and Burke [10].
Pezzey and Burke [10] modified the damage function in the DICE model to  This shortcoming, however, has been overcome by our modeling approach. The simulation technique (i.e., systems dynamics) employed by the World model "forecasted" the trajectories of "the economy", "HDI", "pollution", and "population". The nature of that technique nature led to "overshooting", leading to dismal outcomes. Compared to this, our model and GS are grounded by economic theory in a normative way (i.e., maximization of discounted utility flow).
Second, while the Stiglitz report simply extrapolates the current GS trend, we calculate the future trajectories of SD indicators based on a sophisticated IAM, as described above. Third, while the RICE model seeks the optimal climate policy in the sense of cost-benefit analysis, our model introduces SD indicators to assess climate policies. Nordhaus, the developer of the series of DICE/RICE models, had developed models for energy technologies [77] and copper [78]. Our IAM also includes state-of-the art models for energy, minerals, biomass and foods, in addition to an impact assessment model (i.e., LIME3) with dose-response (or cause-effect) functions in a bottom-up manner, contradicting the top-down aggregated damage functions.

Concluding Remarks
Our modeling exercise shows that we were not fully successful in operationalizing the QoL indicators for climate policy assessment because the trajectory of HDI is well synchronized to GDP level, and our original indicator QoL fa shows unexplainable behavior. The behaviors seen in SD & E indicators (GS, HANPP) are easier to interpret than QoL indicators, while GDP is the easiest to interpret.