Abstract
This Selected Issues Paper (SIP) uses the dynamic macroeconomic Sustainable Development Goals (SDGs) financing framework to estimate the infrastructure financing gap in São Tomé and Príncipe (STP). Human capital and infrastructure investments are key strategic priorities of STP’s National Development Plans.2 We will: i) estimate STP’s human capital and infrastructure gap in 2030 based on the authorities’ current policies; ii) estimate how active policies—improved domestic revenue mobilization, enhanced spending efficiency, and efforts to attract private investment—could reduce this infrastructure gap; and iii) determine the residual financing to be sought from STP’s development partners in grant financing.
Human Capital and Infrastructure Financing GAP1
This Selected Issues Paper (SIP) uses the dynamic macroeconomic Sustainable Development Goals (SDGs) financing framework to estimate the infrastructure financing gap in São Tomé and Príncipe (STP). Human capital and infrastructure investments are key strategic priorities of STP’s National Development Plans.2 We will: i) estimate STP’s human capital and infrastructure gap in 2030 based on the authorities’ current policies; ii) estimate how active policies—improved domestic revenue mobilization, enhanced spending efficiency, and efforts to attract private investment—could reduce this infrastructure gap; and iii) determine the residual financing to be sought from STP’s development partners in grant financing.
A. Introduction
1. Promoting strong and inclusive growth in São Tomé and Príncipe (STP) would require narrowing the large human capital and infrastructure financing gap. While STP fares slightly better than the average sub-Saharan African (SSA) countries on some infrastructure and social indicators, ensuring universal access to basic services remains a challenge. For instance, close to 40 percent of the population do not have access to electricity and more than 30 percent do not have access to drinking water.
2. The COVID-19 pandemic risks delaying STP’s reaching the SDGs. The pandemic has severely impacted STP’s main economic activities (leisure and hospitality sectors), exacerbating preexisting inequalities and social vulnerabilities. To avoid lasting economic scarring from COVID-19, the authorities have formulated a two-year National Development Plan (NDP) focused on infrastructure and social spending. The cost of the NDP is estimated at US$84.3 million (15.8 percent of GDP), with an identified financing gap of US$57 million (10.7 percent of GDP). To bolster the credibility of the NDP framework, it needs to be rooted in a sustainable medium-term framework, backed by feasible revenue and spending measures.3
3. This SIP aims to inform the authorities policy considerations on meeting SDGs. The paper is organized as follows: section II presents stylized facts; section III presents STP’s human capital and infrastructure financing needs based on the authorities’ current policies; section IV estimates how active policies—improved domestic revenue mobilization, enhanced spending efficiency, and efforts to attract private investment—could reduce the financing gap, Section V presents simulation results to determine the residual financing to be sought from STP’s development partners in the form of grants, and Section VI discusses policy implications.
4. We use the dynamic macroeconomic SDG financing framework developed by Benedek et al. (2021) to estimate the infrastructure financing gap (see Annex 1). It is a long-term, macroeconomically consistent, dynamic framework in which output growth is driven by investment in physical and human capital. Like Benedek et al. (2021), we use this model to (i) assess the role of the public and private sectors to generate the funding to achieve the SDGs in five sectors: education and health (human capital sectors), and electricity, roads, and water and sanitation (infrastructure sectors); and (ii) assess various financing scenarios to close the SDGs financing gap. In this context, active policies are estimated as follows:
Improved domestic revenue mobilization (DRM). To assess the extent to which enhanced tax policy and revenue administration measures could increase tax revenues. We estimate STP’s tax effort (i.e., the level of tax revenue collected relative to the frontier) and derive the tax potential (i.e., how much more revenue can be collected at full potential), using Stochastic Frontier (SF) models (see Annex 2).
Improved public spending efficiency. We use SF models to estimate the efficiency of public spending in the sectors captured by the SDG financing framework. It estimates models for public spending on health, education, and investment.
Leveraging private sector participation. We assume that the government takes measures to improve the regulatory framework and business environment, leading to one percentage point increase in private sector financing of SDGs.
Mobilizing grant financing. With these active policies, STP would still need additional financial support from its international development partners to close the human capital and infrastructure financing gap.
B. Stylized Facts
5. São Tomé and Príncipe is a small, remote, island state economy, with a population of about 200,000, of which about a third lives in extreme poverty. Subsistence agriculture and fisheries are core sectors of the economy. The key exports are cocoa products and tourism services, and the country imports food, fuel, and other essentials. With weak tax revenues, grants remain an important source of financing for economic and social development. Obsolete fossil fuel-based electricity generation and loss-making state-owned energy enterprises (SOEs) hinder the efficient functioning of the economy and pose social challenges. 4 Strong capacity building efforts are critical to strengthen institutions and support reforms.
6. STP’s GDP growth rate has declined steadily during the last fifteen years (Figure 1). The growth performance reflects weaker structural and trade policies relative to peer countries. While access to electricity and the quality of trade and transport-related infrastructure in STP is above the SSA average (Figure 2), the quality of education provision is lower with a shortage of trained teachers. Both tax revenue and official development assistance (ODA) have declined steadily over the last decade (Text Figure 1) narrowing the fiscal space for social and development spending.
Structural Indicators
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Sources: São Tomé and Príncipe authorities’ data, World Economic Outlook and IMF staff estimatesStructural Indicators
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Sources: São Tomé and Príncipe authorities’ data, World Economic Outlook and IMF staff estimatesStructural Indicators
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Sources: São Tomé and Príncipe authorities’ data, World Economic Outlook and IMF staff estimates
Structural Indicators (concluded)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Source: IMF In vestment and Ca pita l Stock Data base, WB World Development Indicators, WB Logistics Performance Index.Structural Indicators (concluded)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Source: IMF In vestment and Ca pita l Stock Data base, WB World Development Indicators, WB Logistics Performance Index.Structural Indicators (concluded)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Source: IMF In vestment and Ca pita l Stock Data base, WB World Development Indicators, WB Logistics Performance Index.
Tax Revenue and ODA Inflows
(In percent of GDP)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Tax Revenue and ODA Inflows
(In percent of GDP)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
Tax Revenue and ODA Inflows
(In percent of GDP)
Citation: IMF Staff Country Reports 2022, 096; 10.5089/9798400206320.002.A001
C. Human Capital and Infrastructure Financing Needs
7. STP needs about 14 percent of GDP in additional annual investment to achieve the human capital and infrastructure SDGs by 2030. Considering the country’s investment spending efficiency, the initial capital stock, and the capital depreciation rate, to meet its SDGs by 2030, STP needs additional annual investment of 2.7 percent of GDP in health, 3.5 percent of GDP in education, 1.3 percent of GDP in electricity, 2.0 percent of GDP in water and sanitation, and 4.7 percent of GDP in infrastructure (e.g., roads).5
8. However, São Tomé and Príncipe’s tax revenue performance is not sufficient to meet large social and development needs. To this end, implementing the VAT in 2022 will help to generate domestic resources to support growth-enhancing social and infrastructure development programs. These efforts need to be complemented by other active policies, including improvements in public investment efficiency.
9. The authorities recognize the need to mobilize additional domestic revenues to finance STP’s social and development goals. Their strategy is guided by IMF Technical Assistance (TA) on tax policies and revenue administration with a focus on: (i) implementing VAT by end 2022 (preparations are at an advanced stage); (ii) improving tax compliance (boosting registration, intensifying the use of technology, and improving management and strategic planning); (iii) adopting modern compliance risk management practices, including audit programs that make use of information from third parties; and (iv) overhauling of the current performance monitoring framework, including key performance indicators and a rewards program.
D. Scenarios for Financing the Infrastructure Gap
Domestic Revenue Mobilization
10. STP’s estimated tax effort is more than 30 percent below the frontier. To assess the extent to which enhanced tax policy and administration could increase tax revenue, we estimate STP’s tax effort (i.e., the gap between the level of tax revenue collected and the frontier) and derive its tax potential (i.e., how much more can be collected if STP reaches its full potential), using Stochastic Frontier (SF) models (Text Table 1).
Sao Tome and Principe: Tax Effort and Tax Potential
Sao Tome and Principe: Tax Effort and Tax Potential
| Tax Effort Range (%) | Tax Potential Rang (% of GDP) | ||||
|---|---|---|---|---|---|
| Min | Max | Min | Max | ||
| Total (Model Aggregation) | 65% | 69% | 5.6 | 6.6 | |
| Goods and Services Tax Model | 34% | 38% | 3.5 | 4.1 | |
| Trade Tax Model | 77% | 81% | 1.4 | 1.7 | |
| Income Tax Model | 85% | 87% | 0.6 | 0.8 | |
Sao Tome and Principe: Tax Effort and Tax Potential
| Tax Effort Range (%) | Tax Potential Rang (% of GDP) | ||||
|---|---|---|---|---|---|
| Min | Max | Min | Max | ||
| Total (Model Aggregation) | 65% | 69% | 5.6 | 6.6 | |
| Goods and Services Tax Model | 34% | 38% | 3.5 | 4.1 | |
| Trade Tax Model | 77% | 81% | 1.4 | 1.7 | |
| Income Tax Model | 85% | 87% | 0.6 | 0.8 | |
11. The tax revenue-to-GDP ratio could increase by up to 6.6 percentage points to reach the frontier. With appropriate tax policy and administration measures, STP can increase its tax revenue-to -GDP ratio from the current level of 13 percent to 19 percent. In the SDG financing simulations, we assess the impact of increasing tax collection between 2023 and 2025 by the maximum tax potential (i.e., 2 percentage points per year). It is expected that the VAT would generate 4 percent of GDP, if implemented with a 15 percent rate and limited exemptions.
Public Spending Efficiency
12. Improving efficiency of public spending could help to narrow the infrastructure gap. Our analysis uses SF models to estimate the efficiency of public spending in the sectors captured by the SDG financing framework. We estimate three models for public spending on health, education, and investment. The estimated efficiency parameters are fed into the SDG financing model and simulations are made to assess the impact of increasing public spending efficiency on the SDG financing gap as well as on the timing of achievement of the SDGs.
13. STP has ample room to improve the efficiency of public spending, especially on investment projects (Text Table 2). STP appears to be close to the frontier when it comes to heath spending efficiency, but educational outcomes could be improved by about 4 percent with the same amount of public education spending, given the country’s initial conditions. Furthermore, STP could save about 14 percent of GDP on public investment spending.
Sao Tome and Principe: Public Spending Efficiency and Potential Savings
Sao Tome and Principe: Public Spending Efficiency and Potential Savings
| Efficiency of Public Spending (%) | Potential Savings (% of GDP) | |||
|---|---|---|---|---|
| Min | Max | Min | Max | |
| Investment | 86% | 92% | 14.0 | 14.0 |
| Education spending | 86% | 93% | 3.7 | 3.7 |
| Health spending | 97% | 97% | 2.7 | 2.7 |
Sao Tome and Principe: Public Spending Efficiency and Potential Savings
| Efficiency of Public Spending (%) | Potential Savings (% of GDP) | |||
|---|---|---|---|---|
| Min | Max | Min | Max | |
| Investment | 86% | 92% | 14.0 | 14.0 |
| Education spending | 86% | 93% | 3.7 | 3.7 |
| Health spending | 97% | 97% | 2.7 | 2.7 |
Private Sector Participation
14. Financing could also be leveraged from fostering private sector participation to close the infrastructure gap. This could be done through public private partnerships and/or by enhancing the clarity and transparency of the regulatory and legal frameworks to improve the business environment. To this end, governance would need to be strengthened further to remove bottlenecks that hinder the development of private sector activities. Against this background, we analyze a scenario which assumes that the government takes measures to improve the business environment, leading to an additional one percentage point increase in private sector financing of SDGs.
E . Simulation Results
15. Active policies would help narrow the human capital and infrastructure financing gap. In the baseline scenario, without active policies and additional ODA financing, STP would not meet the human capital and infrastructure SDGs before 2051 (Text Table 3). To achieve the MDGs by 2030, STP would need about 14 percent of GDP in additional investment per year:
On the one hand, taking measures that increase tax collection by 2 percentage points of GDP each year between 2023 and 2025 lowers the additional infrastructure investment need per year to 9.4 percent of GDP to achieve the MDGs by 2030.
On the other hand, boosting spending efficiency to the frontier for investment, education, and health, lowers the additional investment needs to 13.3 percent of GDP. This scenario leads to a significant improvement in the debt-to-GDP ratio in 2030 (by 3.8 percentage points).
Together, revenue and spending efficiency measures could cover more than 40 percent of the SDG financing needs.
Gradually increasing private sector investment by one percentage point by 2023 also helps reduce the human capital and infrastructure financing gap. Under this scenario, the annual investment needs to close the gap by 2030 is lowered by 0.8 percentage points of GDP.
Altogether, active policies could reduce the additional annual investment needs to 7.7 percent of GDP.
Sao Tome and Principe: Dynamic Infrastructure Financing Framework Scenarios (in percent of GDP)
Sao Tome and Principe: Dynamic Infrastructure Financing Framework Scenarios (in percent of GDP)
| Scenarios | Infrastructure SDG targets met | Additional grants per year | Residual infrastructure gap in 2030 | Additional investment need per year | Public debt in 2030 |
|---|---|---|---|---|---|
| Baseline settings | 2051 + | 0.0 | 62.7 | 0.0 | 56.8 |
| Baseline settings | 2030 | 0.0 | 0.0 | 14.0 | 53.2 |
| Domestic revenue mobilizat | 2030 | 0.0 | 0.0 | 9.4 | 53.7 |
| Increased spending efficienc | 2030 | 0.0 | 0.0 | 13.3 | 49.4 |
| Combined active fiscal polic | 2030 | 0.0 | 0.0 | 8.6 | 49.9 |
| Private sector participation | 2030 | 0.0 | 0.0 | 13.2 | 52.9 |
| Active policies without grant | 2030 | 0.0 | 0.0 | 7.7 | 49.8 |
| Active policies plus grants | 2030 | 7.7 | 0.0 | 0.0 | 49.8 |
Sao Tome and Principe: Dynamic Infrastructure Financing Framework Scenarios (in percent of GDP)
| Scenarios | Infrastructure SDG targets met | Additional grants per year | Residual infrastructure gap in 2030 | Additional investment need per year | Public debt in 2030 |
|---|---|---|---|---|---|
| Baseline settings | 2051 + | 0.0 | 62.7 | 0.0 | 56.8 |
| Baseline settings | 2030 | 0.0 | 0.0 | 14.0 | 53.2 |
| Domestic revenue mobilizat | 2030 | 0.0 | 0.0 | 9.4 | 53.7 |
| Increased spending efficienc | 2030 | 0.0 | 0.0 | 13.3 | 49.4 |
| Combined active fiscal polic | 2030 | 0.0 | 0.0 | 8.6 | 49.9 |
| Private sector participation | 2030 | 0.0 | 0.0 | 13.2 | 52.9 |
| Active policies without grant | 2030 | 0.0 | 0.0 | 7.7 | 49.8 |
| Active policies plus grants | 2030 | 7.7 | 0.0 | 0.0 | 49.8 |
16. Even with active policies, STP would need support from international development partners to close its human capital and infrastructure financing gap by 2030. Additional grants needed from development partners would amount to 7.7 percent of GDP per year if all active policies are implemented as described above. Without this support and if STP authorities commit to closing the SDG gaps by 2030, the alternative would be incurring public debt to invest in infrastructure. In the extreme scenario in which no policy action is taken, the government would need to increase debt by at least 14 percent of GDP per year between 2022 and 2030, which would lead to a significant increase in public debt, relative to the baseline scenario. However, this option is difficult to envisage given the fact that fiscal space is currently limited and public debt is in distress.
F. Policy Implications
17. STPs SDG financing needs are large. We estimate that, under current policies, the amount of investment spending needed to achieve the human capital and infrastructure SDG targets in 2030 amounts to about 63 percent of GDP. Considering the country’s investment spending efficiency, the initial capital stock, and the capital depreciation rate, this estimate translates into an average additional annual investment need of about 14 percent of GDP between 2021 and 2030.
18. STP needs strong policies to meet its SDGs Our estimates suggests that, out of the annual additional infrastructure financing need of 14 percent of GDP, enhancing domestic revenue mobilization could cover 4.6 percent of GDP, and strengthening spending efficiency and private sector involvement could each cover about 1 percent of GDP. Still, STP’s SDG needs (14 percent of GDP per year) would far exceed potential domestic public and private resources (6.3 percent of GDP per year), suggesting that development partners may have to contribute to financing the residual human capital and infrastructure investment gap (7.7 percent of GDP).
19. The authorities’ strong efforts could focus on a combination of domestic revenue mobilization, higher public spending efficiency, favorable environment for private sector financing, and collaboration with development partners:
Improve revenue mobilization. Implement the VAT in 2022 as scheduled, with a 15 percent rate and limited exemptions. This would broaden the tax base and achieve a tax-to-GDP ratio of 20 percent, in line with bes t-performing peer countries. Increasing the tax-to-GDP ratio by 6 percentage points in the medium term is an ambitious but achievable aspiration for STP. In general, an efficient VAT would also boost the buoyancy of other taxes.
Improve spending efficiency. Strengthening institutions governing project appraisal, selection and management would be important to enhance the effectiveness of spending and reduce resources required to achieve the SDGs.
Enable private investment. Strengthen institutional frameworks to enhance the clarity and transparency of the regulatory and legal frameworks. Boosting private finance to bridge the SDG financing gap would increase tax revenues, bring efficiency gains, and enhance risk sharing between the public and private sectors.
Seek an increase in ODA financing. Closing the residual human capital and infrastructure financing gaps would require an increase in ODA from the current level of about 5.5 percent to about 13 percent of GDP per year. This level of grant financing for projects is achievable, given the track record in the early 2010s.
Annex I. A Dynamic Macroeconomic Framework for SDG Financing
1. The macroeconomic framework developed by Benedek et al. (2021), to evaluate the financing strategies to achieve the SDGs, consists of a set of accounting and behavioral equations covering the real, fiscal and external sectors of the economy, with the overriding objective of ensuring macroeconomic consistency while maintaining flexibility and tractability. The framework focuses on the ability of public and private economic actors to mobilize funding to achieve the SDGs in five key areas, namely education, health, roads, electricity, water and sanitation. The framework ensures that economic growth is consistent with human and physical capital investment and follows demographic developments. The model is used to simulate the effect of policies over the 2020–50 horizon. The main features of the framework are described below (see Benedek et al. 2021 for a detailed description of the model):
2. On the real side, the model relies on the IMF’s Debt, Investment and Growth (DIG) model that addresses the public-investment-growth nexus and fiscal adjustments in low income and emerging economies. The production function is given by:
3. Where H is human capital, A is total factor productivity, KG,b and KG,nb are public bankable and public non-bankable (i.e., financed with private resources) capital stocks respectively. Kp is private capital stock. L is the labor force and
4. Investment (Ii,t) and depreciation (δi,k) determine the dynamics of capital stocks according to the following law of motion:
5. Where 0 < ε ≤ 1 is the efficiency with which investment spending is transformed into effective capital.
6. Similarly, schooling and improvements in health, represented by ξ > 0, and depreciation (δh) determine the dynamics of human capital according to:
7. Where ω ∈ (0,1) is the rate at which human capital increases with the previous period of schooling. Human capital generated through schooling and health accumulates according to this law of motion:
8. Where h is the annual nominal spending on health and education, which translates into new human capital according to an efficiency parameter e > 0, with elasticity ϕ > 0. n is the share of school-age population and γ > 0 is the elasticity of schooling to n.
9. The fiscal balance determines the amount of resources available for SDGs spending, according the following identity:
SDG resources = Revenue – NonSDG Expenditures – Net public lending
10. In the framework, Gaspar et al. (2019)’s quantification of SDG targets are used to derive the gap between the actual annual investment spending in infrastructure and the spending required to meet the SDG targets. Thus, the framework calculates the amount of additional financing (on top of resources in staffs baseline scenario) needed to reach the SDG goals within a given timeframe.
Annex II. Stochastic Frontier Models for Estimating Tax Potentials and Spending Efficiency
Methodology
1. A stochastic frontier analysis uses econometric models to link measures of input (or resources) with a measure of output, while controlling for determinants of the output variable other than the input variables, with the final goal to assess whether the inputs produce the highest level of output (maximum efficiency). The stochastic frontier model used in this paper (see Greene, W. H., 2008; Parmeter, C. F. and Kumbhakar, S. C.,2014 for more details) specifies a production technology, f(Xi, α) using inputs for country i,
Tax Potential Analysis
2. Following the literature in this line of research, economic and socio-political variables are used to explain the behavior of several tax revenue indicators (Total tax, Goods & Services, Income, and Trade) in several sub-group of countries to which STP belongs (Fragile states, Low-income countries, and Sub-Saharan Africa). The estimations use panel datasets with varying numbers of countries depending on the subgroup, over the period of 1996–2019. Specifically, the explanatory variables include the PPP-adjusted real GDP per capita, consumption, CPI inflation, financial deepening index, share of urban population, agricultural value-added, investment, and government effectiveness. Data was obtained from World Economic Outlook, World Development Indicators, International Financial Statistics, and World governance indicators.
3. Estimation results for the total tax revenue model are presented in Annex 2 Table 1 1 Model results are used to compute the tax effort (TEi), the tax frontier (TFi) and the tax potential (TPi) respectively as: TEi = εi,
Spending Efficiency Analysis
4. The efficiency of public spending is analyzed in three key sectors: education, health, and infrastructure (investment). For the education sector, the outcome indicator combines measures of out-of-school children, mean schooling years, and school enrollment and attainment. For the health sector, the outcome indicator combines measures of life expectancy, infant, child and maternal mortality, and treatment outcomes of tuberculosis, diphtheria and measles. Following IMF (2015), measures of public investment efficiency include measures of coverage (World Development indicators’ measures of access to electricity, water, telecommunications) and a measure of infrastructure quality (from the World Economic Forum). Explanatory variables for each model include public spending, private spending (where data is available), the level of development measured by real GDP per capita, as well as other relevant determinants of the outcome variable. Data are obtained from the World Economic Outlook, World Development Indicators, International Financial Statistics, World Health Organization, and Unicef.
5. Estimation results for the Public Investment Spending model are presented in Text Table 2.2 Model results are used to compute the efficiency of public spending.
References
Benedek, D. and E. Gemayel, A. Senhadji, and A. Tieman (2021), A Post-Pandemic Assessment of the Sustainable Development Goals. Staff Discussion Note 21/003. International Monetary Fund, Washington, D.C.
Gaspar V., D. Amaglobeli, M. Garcia-Escribano, D. Prady, and M. Soto (2019), Fiscal Policy and Development: Human, Social, and Physical Investment for the SDGs. Staff Discussion Note 19/03. International Monetary Fund, Washington, D.C.
Greene, W. H. (2008), The Econometric Approach to Efficiency Analysis. In Fried, H. O., Knox Lovell, C. A., and Schmidt, P., editors, The Measurement of Productive Efficiency. Oxford University Press, New York and Oxford
International Monetary Fund (2015), Making Public Investment More Efficient, (Washington). https://www.imf.org/external/np/pp/eng/2015/061115.pdf.
International Monetary Fund (October 2021), Fiscal Monitor, (Washington).
Parmeter, C. F. and S. C. Kumbhakar (2014), Efficiency Analysis: A Primer on Recent Advances, Foundations and Trends in Econometrics, 7(3–4), 191–385.
Prepared by Koffie Nassar. The author would like to thank Laurent Kemoe and David Baldini for their guidance and Chie Aoyagi and Victor Duarte Lledo for constructive comments.
Building infrastructure is a key strategic priority of STP’s National Development Plan (NDP) 2020-24. The COVID-19 pandemic has caused delays in implementing the NDP. As a result, an interim NDP has been prepared to address the COVID-19 pandemic during 2020-2022.
The IMF’s Fiscal Monitor (October 2021) demonstrates that medium-term framework credibility can lower financing costs for countries and increase fiscal space in the near term.
See Selected Issues Paper: Assessing Fiscal Risks and Implications for the Energy Sector.
These estimated costs are based on benchmarking of STP with its comparison group and an extrapolation from exercises conducted in other countries. Estimates are in percent of 2030 GDP.
Results from the three other models are available from the authors upon request.
Results from the Education and Health models are available from the authors upon request.
