:l l ~ l Strathmore UNIVERSITY Measuring the Impact of Universal Health Care on Longevity Risk (A case for Rwanda) MBAU LORNA, 100662 Submitted in partial fulfilment of the requirements for the Degree of Bachelor of Business Science in Actuarial Science at Strathmore University Strathmore Institute of Mathematical Sciences Strathmore University Nairobi, Kenya February 2021 This Research Project is available for Library use on the understanding that it is copyright material and that no quotation from the Research Project may be published without proper acknowledgement. l l .. l ' l J DECLARATION I declare that this work has not been previously submitted and approved for the award of a degree by this or any other University. To the best of my knowledge and belief, the Research Project contains no material previously published or written by another person except where due reference is made in the Research Project itself. © No patt of this Research Project may be reproduced without the permission of the author and Strathmore University LORt'-IA IV\.BAU . . . . . . . . . . . . . . . . . . . .. .. .... ..... .. ... .......... .. .. . ... ..... .. [Name of Canchdate] ... ~ . . ..... ... ......... ..... ........... .... ...... ...... [Signature] .. . \.Q. {f:~~ J -?.!?.~ ! ......... .. ....... .............. .... .... [Date] This Research Project has been submitted for examination with my approval as the Supervisor. · · · · · · · ;P. .\: \d~~t .. ;p. ~ ... ...... ........ .... .. [Name of Supervisor] . . . . .... .. ... . ~~ ....... ........ .... .. .... ..... [S1gnature] ... .• .. . ... ......•..... ... \ .t?. f~.t?, J.'q.._c:.3 . I. ............ [Date] Strathmore Institute of Mathematical Sciences Strathmore University 1 .J ) - I J ABSTRACT Longevity remains to be a complex risk that is ever evolving and requ1res great understanding. Improved healthcare through the roll out of Universal Health Care is also seen to enhance this risk through improved mortality rates . This leaves individuals, governments and benefit providers at risk because they can not appropriately make financial plans leading to major losses or inadequate funds and reserves. This study expounds on longevity risk and its impact on the interested pmties . 2 ~ J - [ .J J _j Table of Contents ABSTRACT List of Tables and Figures List of Abbreviations Chapter 1: Introduction 1.1 Background information 1.1.1 Definition and brief elaboration of key concepts 1.1.2 Identification of main developments in the study area and any gaps 1.1.3 Brief conceptualization of the study (context, trends) 1.2 Problem Statement 1.3 Research Objectives 1.4 Significance of the research 1.4.1 Identification of beneficiaries 1.4.2 How will they benefit? Chapter 2: Literature Review 2.1 Theor·etical and Empirical Framework 2.1.1 Measuring Universal Health Care 2.1.2 Modelling Mortality and Life Expectancy 2.1.2.1 Lee Cmter Model 2.1.2.2 Poisson Log-Bilinear Regression Model 2.2 Resear·ch Gap Chapter 3: Research Methodology 3.1 Description ofVariables 3.2 Research Design 3.3 J>opulation & Sampling 3.4 Data Collection 2 5 6 7 7 7 8 8 9 10 10 10 10 12 12 12 13 14 14 15 16 16 16 17 17 3 Chapter 4: Analysis, Results and Discussions 4.1 Sources ofData 4.2 Description of software used 4.3 Assumptions 4.4 Data Analysis 4.5 Fitting the model 4.6 Actuarial pmjections Chapter 5: Conclusions & Recommendations 5.1 Conclusion 5.2 Limitations 5.3 Recommendations List of References ; 19 19 19 19 20 21 23 25 25 25 25 27 4 ... ( J List of Tables and Figures Figure I: Life expectancy at birth in Rwanda Figure 2: Log death rates in Rwanda from 1960 to 2018 Figure 3: Percentage uptake ofCBHI in Rwanda Figure 4: Lee-Carter estimation of parameter a X Figure 5: SVD estimation of b and k X t Figure 6: k projections t 5 l .. l .J _j List of Abbreviations The following are the abbreviations and their respective definitions as used in this study: UI-IC- Universal health coverage WHO- World Health Organisation CBHI - Community Based Health Insurance 6 .l ! . ! Chapter 1: Intr·oduction 1.1 Background information 1.1.1 Definition and brief elaboration of key concepts According to the World Health Organization, Universal Health Coverage (UI-IC) means that "All people and communities can use the promotive, preventive, curative, rehabilitative and palliative health services they need, of sufficient quality to be effective, while also ensuring that the use of these services does not expose the user to financial hardship" (WHO, 2020). Over the past several years, we have observed an increase in life expectancy globally and Rwanda is no exception. Life expectancy is the average period that an individual may be expected to live. In 2019, the life expectancy at bi1th was 68.75 years compared to 52.33 years in 2004 when UHC was first rolled out (World Population Prospects [WPP], 20 19). This may be partly attributed to improved healthcare and better living standards. Increased life expectancy is a positive ,result but in turn, it enhances longevity risk to a great extent. However, other factors affect life expectancies such as lifestyle choices, genetics, nutrition, income, geographical location, diseases, terminal illnesses, and pandemics such as the one currently being experienced, COVID-19 and in turn, higher mo1tality directly reduces the life expectancy. Longevity risk may be defined as the risk that the actual survival rates and life expectancy exceeds the expected rates or any assumptions made earlier, resulting in greater cash flow needs. Longevity risk is not diversifiable so it might strain the resources available to individuals, pension schemes, life insurers and governments because they cannot adequately plan their future financial liabilities due to the unce1tainties involved . Increased life expectancy requires that governments, life insurers and defined benefits schemes keep paying out benefits for longer periods and individuals may outlive their income or savings. 7 l l J _j 1.1.2 Identification of main developments in the study area and any gaps Achieving Universal Health Care is a global concern as stated by the World Health Organisation (WHO) which is also in line with the Sustainable Development Goals rolled out in 20 15 . So, all WHO member states, Rwanda included, committed to achieving UHC by 2030 . Rwanda rolled out a community-b~sed health insurance scheme that enables them to subsidize costs for their citizens . Over 90% of the citizens are part of this scheme. There has been a lot of sensitisation to the citizens on disease prevention as well as a good breakdown of health care services to the grass-root leveL These steps have improved UHC in Rwanda and it is deemed to be do.ing well as a developing country. However, slightly more than half of the world 's population still does not have access to basic healthcare (McNeill & Jacobs, 20 19). The major challenge in making Universal Health Care a reality is lack of adequate funding . This exposes the citizens to some financial risk and hence we are not certain whether this program will be sustainable in the long term . This sector als.o lacks proper key progress indicators locally and internationally. 1.1.3 Brief conceptualization of the study (context, trends) Life expectancy in Rwanda is seen to be on the rise every other year, after the brutal genocide in 1994, which exposes them to increased longevity risks . This risk is driven by various factors that should be well understood for proper risk mitigation. Health interventions and social development are among the major factors affecting life expectancy (Chan & Kamala, 20 15). However, there is a lot of unce1tainty surrounding mortality projections (Brouhns, Denuit & Vermunt, 2002) as well as measures of progress in terms ofUHC. 8 l l -- ~ - ( -I J J Figure 1: Life expectancy at birth in Rwanda life expectancy at birth in Rwanda (years) 70 65 50 55 45 ·-···········-··-.. ···-···········-··············-................. ,,,_.,,.,, .... ........................ ,_, ···········-·····-·-··········---·-·······- 2.004 2.005 20;()5 2007 200S 2009 2010 2011 2012 20 13 2.014 2015 2015 2017 2018 Year Source: The World Bank 1.2 Problem Statement There has been an increased emphasis on healthcare that has led to the roll-out of UHC guidelines that assures all people of access to good quality, basic health care services without causing financial strain . UHC has therefore been embraced by all WHO member states which include developing countries such as Rwanda. Rwanda seeks to target the poorest members of the community by improving the quality and accessibility to healthcare services As we all know, with every action there is an implication and it has been a positive one in this case . The population is seen to be experiencing an overall increase in life expectancy due to an increase in the Healthy Life Expectancy as well as the Life Expectancy at Birth (Ranabhat, Atkinson, Park, Kim, & Jakovljevic, 20 18). 9 l I _) .J Increased life expectancy comes at a cost that is the increase in longevity risks. This affects individuals, the government, life insurers and defined benefits pensions as they might have to pay out benefits for longer periods . The financial strain of healthcare might have been taken away from the citizens however, this leaves the benefit-providers more exposed to financial risk. 1.3 Research Objectives This paper seeks to determine the impact of Universal Health Care on longevity risks by forecasting the Universal health care index assuming the current trends will continue. Thereafter, we will be able to see the financial impact on individuals, the government, life insurers and the defined benefit pension schemes. This study seeks to answer these questions: • What is the current trend in life expectancy? • What will the life expectancy trend be like as the universal health care changes? • What is the financial impact of the changes in UHC? 1.4 Significance of the research 1.4.1 Identification of beneficiaries Failure to acknowledge the impact of longevity risks could expose benefit providers to financial risks due to benefits being paid out for longer periods than had been projected . This study highlights the impact of UHC on life expectancy, which seeks to benefit individuals, the government, life insurers and defined benefits pensions. 1.4.2 How will they benefit? This study will enable the benefit-providers to see the financial implications brought about by longevity risks. Based on the highlighted issues, this will allow for better planning and allocation of financial resources while ensuring solvency and liquidity throughout while still having some emergency cash at hand. The benefit providers 10 . l .J should always be on the lookout for major factors affecting longevity risks such as UHC and therefore adequately plan for them . 11 l ~, I l J J I _j Chapter· 2: Literature Review This chapter reviews relevant literature and theories available on mortality risks, longevity risks and Universal Health Care. 2.1 Theoretical and Empirical Framework 2.1.1 Measuring Universal Health Care Achieving Universal Health Care was adopted as one ofthe Sustainable Development Goals however, we cannot measure its long term effect adequately (Carrin, James & Evans, 2005). Based on the WHO fact sheet on UHC, it is appropriate for the UHC index to be based on the degree of coverage. This is in terms of the financial protection being provided by this scheme, the proportion of the population that has been covered and the services being offered. However, most measures ofUHC focus on the costs paid out of pocket as a measure of coverage which has its shmtcomings (Adam, Daniel, Patrick & Leander, SPRING 2016). Based on the simple model proposed by Adam et al., 2016, we might measure UHC using an index that is the geometric mean of the two: UHC = SC 0 " 5 *FP0 " 5 if we weigh service coverage and financial protection equally. where SC is a service coverage index and FP is a financial protection index. Service Coverage is based on treatment and prevention. However, this is based on the need covered by the services, which is quite subjective considering the needs vary from one person to another. This makes this measure complex. WHO uses four categories to measure coverage in various countries : • Reproductive, maternal , newborn and child health 12 I I I I .l J . J J _j • Infectious diseases • Non-communicable diseases • Service capacity and access. Financial Protection considers whether the healthcare payments made may exceed the family's budget and whether the payments may push the family into financial distress (Wagstaff & van Doorslaer, 2003 ). Healthcare reforms aim to promote preventive healthcare and provide more prima1y healthcare services with the goal to improve the healthcare of the population (Jakovljevic, Arsenijevic, Pavlova, Verhaeghe, Laaser & Groot, 2017). Based on this progress we require appropriate demographic indicators, mortality indicators ~nd health-capacity indicators. Some of the indicators of an effective health system include antenatal care coverage, percentage of bi1ths attended by skilled bi1th attendants, child immunization rates, the prop01tion of children brought to a health-care facility for specific treatments, and the rate at which family planning needs are being met (WHO, 2014). 2.1.2 Modelling Mortality and Life Expectancy There is a lot of uncertainty surrounding m01tality projections (Brouhns, Denuit & Vermunt, 2002) as well as measures of progress in terms of UHC. However, we require appropriate projections of m01tality for the proper estimation of future costs (Brouhns et al., 2002). Mortality rates are not only a function of age. Other factors need to be considered and they are included as parameters when modelling mortality rates (Booth & Tickle, 1970). Due to this, the one-factor models may not be appropriate for this study. There are various models to project mo1tality such as Lee - Ca1ter and the Poisson log-bilinear regression model, which is an enhancement of the Lee-Ca1ter model. Proper m01tality projections reduce the longevity risks that arise . 13 l ~ I .I . I J J 2.1.2.1 Lee Car·ter Model This is a simple model that projects mortality ns a function of a single time index using least squares. It is also a better alternative to direct time-series forecasts as it is seen to have narrower confidence intervals (Lee & Cmter, 1992). The model uses a few parameters compared to the other models therefore, it is easier to use . Lee Ca1ter is currently the most popular method of moJtality forecasting. Lee & Ca1ter (1992) proposed: lnm = a + b k + E x;t x x t x,t ~herem is the central mortality rate at age x in year t; x;t a is average (over time) log-mortality at age x . So, exp a represents the X X general shape of the age-specific mortality profile; b measures the response at age x to change in the overall level of mo1tality over X time; k represents the underlying time trend in year t; t £ is the residual i.e. factors not captured by the model. They are liD with a x,t normal distribution. This model, however, assumes that the errors are homoscedastic which may not be accurate due to the great variance in mo11ality at older ages. Also having b as a constant X is quite debatable because the sensitivity is expected to change over time but despite the limitations, Tuljapuktar & Boe ( 1998), recommended using the Lee-Carter model to model mOitality rates. . J l ' ] ~ l -l . 1 J _ j 2.1.2.2 Poisson Log-Bilinear Regression Model This model is based on the number of deaths and the force of mot1ality. The number of deaths takes up a Poisson distribution because it is a counting random variable while the force of mot1ality takes up the log-bilinear form. D -Poisson (E ~~ ) ~t ~t ~ t where D is the number of deaths in year t of lives aged x x;t E represents the exposure to the risk of death x;t This tackles the problem of homoscedasticity in the Lee-Ca11er model. The assumption of homoscedastic errors is quite unrealistic. Instead, we make use of confidence intervals which helps take into account the various sources of variability (Brouhns et al. , 2002). 2.2 Research Gap Life expectancy is a reflection of the mortality rates and the healthcare system of a pm1icular region (Jakovljevic et al, 20 17). These models used to project future mortality rates fail to account for unexpected shocks and developments such as Universal Health Care that may impact mortality rates. This exposes individuals and benefit providers to greater risks which may not be adequately covered by the income, savings or reserves. This study aims to incorporate such developments for better mortality projections in order to protect individuals and benefit providers. The impact of UHC on mot1ality rates has not been tackled explicitly and we cannot precisely evaluate the relationship between these two factors (Ranabhat et al. , 20 18). This study aims to seek an appropriate relationship between these two factors . 15 l l J J Chapter 3: Research Methodology This chapter explains the various stages followed in the completion of the study. This is in terms of the collection, measurement and analysis of data. We can identify the various techniques used in the collection, processing and analysis of data in order to assess the impact of Universal 1-lealthcare and medical improvements on life expectancy as well as the financial implications on the benefit-providers. 3.1 Description ofVariables This study claims that improvement in UHC leads to an increase in life expectancy at birth. In turn, this exposes the benefit-providers to increase longevity risks. This affects their finances as they may end up paying out benefits for longer periods than expected. Improvements In UHC Increased Life Expectancy Increased Longevity Risk The independent variable is the improvement in mo11ality and longevity in relation to life expectancy. The dependent variable is the life expectancy of the general population . The intervening variable is the universal healthcare trends and its impact on life expectancy. 3.2 Research Design This research uses a quantitative approach to measure the progress in UHC as well as project the expected life expectancy. However, a qualitative approach will be used to show the impact of the development in UHC. 15 . ! _] _] Assumptions The data used is correct and bias free. The changes in mortality follow a constant pattern. Limitations The predicted future trends may be different from the actual future trends. Ce11ain factors may change and there may be unexpected events in future. 3.3 Population & Sampling The population under consideration is the Rwandan population. There is information readily available in the world bank database and the UNICEF global database. The sample size is also based on the available adult mo11ality data. This would be appropriate due to the law of large numbers so it would reduce any major deviations. 3.4 Data Collection The study uses secondary data on adult mm1ality 111 Rwanda and data available m similar studies. Measures of the UHC index are based on the information provided by the World Health Organisation. We aim to use time series analysis to study the trends in life expectancy since the health reforms stmied in Rwanda. Also in order to quantify the impact of Universal Health Care, we need to keenly study the reforms that have taken place in the health sector. We shall consider the features of the UHC index. The methodological approach that will be used to project mmiality is the Lee-Ca1ier model. 17 l ~ l Life Expectancy at Birth This is a measure of the average number of years one is expected to live from bi1th considering the current conditions remain (Zhao, Liang, Wenke Zhao & 1-lou, 2013) T The formula is:----(- "' where T is the total person-years lived the lives being considered X l is the number of lives aged x being considered X Lee-Carter Model (Lee & Carter, 1992) lnm x;t a +bk +£ X X t X,t where: m is the central m01tality rate at age x in year t; x;t a is average (over time) log-mortality at age x; X b measures the response at age x to change in the overall level of mortality over X time; k t represents the changes in mmtality in year t; £ is the residual. These are variations that are not captured by the model x,t The central rate of mortality of age x in year tis defined as; m =D E x;t x;t x;t where D is the number of deaths in year t of lives aged x x;t E represents the exposure to the risk of death x;t Vector a is the average of log rates of mmtality over time. 18 Vector b and k are based on the SVD of the residuals . These vectors will be able to reflect the changes that have arisen as a result of UHC. - ( , I I 19 1 .J J Chapter 4: Analysis, Results and Discussions 4.1 Sources ofData My analysis is based on Rwanda 's mortality data available from the World Bank Database and child mortality data available on the UNICEF database . The database provides information on several countries for different time periods in terms of infant m01tality, adult m01tality, mortality for every sex, crude death rates as well as other economic indicators. I chose Rwanda as opposed to Kenya because it is a third world country in Africa however, they rolled out UHC earlier on compared to Kenya therefore, more data is available for the study. 4.2 Description of software used Accurate mortality projections are required to make appropriate financial decisions. For data analysis, I made use of the RStudio sofuvare and Microsoft Excel. 4.3 Assumptions The following assumptions were made: • Life expectancy was used to measure health because it is inversely correlated with overall death (Wen, Tsai & Chung, 2008) • The age structure of the population is based on the percentages of the 2014 national census . • • Crude death rates represent m x;t The change in m01tality patterns is constant 20 . ! I 1 r • I J 4.4 Data Analysis The data available from the World Bank was imported through the Import Dataset code. The adult crude death rates (for eve1y 1000 people) between 1960 and 2018 were then plotted and the results are as follows : Figure 2: Log death rates in Rwanda from 1960 to 2018 Male death rates Female death rates Total rates ~ <>: : .,.; \ "' 1960 1980 2000 2020 1960 1980 2000 2020 1960 1980 2000 2020 Years Years Years ·' The results confirm a decline in death rates over the years. This is also seen after UHC was introduced in Rwanda in 2004. UHC in Rwanda is measured based on the uptake of CBHI and the financial protection offered by the insurance. The table below shows the percentage uptake of CBHI. 21 J Figure 3: Percentage uptake ofCBHI in Rwanda %of population with a health insurrance Visits/ person/ year 100% ····················-·····"""········----- ·"······· 90% 80% 70% 60% 50% 40% 30% 20% 10% 1.2 0.8 0.6 0.4 0.2 0% -------------~---------···-···-··-·- ....... ....... ... ............ -----------------------~--·-··-------····-···-- ----·------- ---·-- --------- 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 _.,_Enrolment in health insurrance (left axis) 4.5 Fitting the model The model used is the Lee-Carter model: lnm x;t a +bk +£ X X t ~t ~Health facility utilisation(right axis) Source: Science Direct a represents the general m01tality for each age x. It is calculated as the average ofthe X logarithm of the death rates over several years. The vanous values for different age groups are shown in the graph below. 22 l l 1 Figure 4: Lee-Carter estimation of par·ameter a X Measuring ax 0 0 0 •:0 ~ 0 0 U> ~ Q) ~ ~ .r_ 0 -ro <7 "' "0 "' "' .£ <7 0 1' 0 under5 age5to9 age10fo14 age15to19 age20to24 adults Age brackets b k is estimated using the singular value decomposition. We make use of the first right X t and the first left sin 0ou)ar vectors of the matrix (ln m - a )(Haberman & Renshaw, x;t x 2008). k reflects the effect of the calendar year while b reflects the impact as of a t X specific age group. 23 l J Figure 5: SVD estimation of b nnd k X t k! ~ ~---------------------------------------. '\(~' \/l 'I N 9 1!l60 1964 1968 1972 1976 1 ~80 1934 1988 1992 1996 2000 2004 2008 2012 2015 Years bx ~ 0 "'? 0 "' ~ "'l "'? 0 0 0 0 under5 age5to9 age10to14 age15to19 age20to24 adu~s Age groups The results reflect that those under 5 years of age and adults on average experience greater mo11ality rates . This may be attributed to their exposure to many health conditions and various diseases as they are more vulnerable. Also, the sudden hump in the 1990s may be attributed to the 1994 genocide which lead to increased death rates. 4.6 Actuarial projections a and b are assumed to be constant so, the forecast is based on projecting k . The X X t changes in k also reflect the expected improvements in UHC. The projections of k t t make use of the random walk with drift model. k =k +8+E t t-l t 24 l . i J k -k where 8= _r - 1 T-1 k is the last calculated k T t k is the first calculated k 1 t T - lis the time difference 8 is a positive value, 0 .0026, so kt is reflecting an improvement in mo1tality just as the levels ofUHC coverage are expected to be increasing with time. r I ::: I ::: I 0 .05 I o.o4 I 0.03 I 0.02 0.01 i 0 ; .................. .......... ...... .... . Figure 6: k projections t Kt 1 2o1s 2o2o 2021 202! 2025 202S 2030 2032 1034 The longevity risk facing the country is clearly highlighted and the life offices and the government should accurately account for this when making financial plans. Otherwise, this could lead to greater benefit payments. The reserves and premiums collected may not be adequate to cover and major losses may result as well. Longevity risk is seen to be developing slowly and it takes time before it can be realised . 25 l .. 1 - [ I .J Chapter 5: Conclusions & Recommendations 5.1 Conclusion This research study focused on Rwanda which is a developing country that had rolled out UHC in the early 2000s so it was a good representation of other third world countries embracing UHC. The study sought to determine the existence of longevity risk and the Lee-Carter model was used to project the mortality parameters. The results reflect a decrease in mortality and an increase in the UHC coverage as the years go by. So it is possible that the citizens may live for longer periods than expected. Failure to consider the longevity risks may lead to major losses due to inadequate premiums and reserves. 5.2 Limitations However, there are several limitations such as: • M01tality data is not readily available for all age groups so the study made use of the available grouped data. • The m01tality experience is assumed to be constant because the Lee-Catter model makes use of constant parameters and this may not be realistic. 5.3 Recommendations Longevity risk has been proven to exist so risk mitigation processes should be put iu place to counter its effects . Longevity risk is also accompanied by other risks such as interest rate risk, inflation risk, credit risk etc. Therefore, this makes longevity risk very complex. Therefore, parties involved should also be informed on the risk assessment, measurement and mitigation techniques. 26 . ! I _j Derivatives may be used by benefit providers to shield their funds from this risk . Insurance is also another viable option so the benefits provider passes on the risk to the insurer. It is essential that the m01tality tables are updated following these developments so as to prevent under the provision for future benefits . 27 ~l I I . J I I ._j List of References Adam Wagstaff, Daniel Cotlear, Patrick Hoang-Vu Eozenou, Leander R. Buisman (SPRING 20 16). Measuring progress towards universal health coverage: with an application to 24 developing countries, Oxford Review of Economic Policy, Volume 32, Issue I Booth, H., & Tickle, L. ( 1970). Mo1iality Modelling and Forecasting: a Review of Methods. Bojuan B. Zhao, Xiangliang Liang, Wenke Zhao, Delong Hou . (2013) Modelling of group-specific mortality in China using a modified Lee-Carter model. Scandinavian Actuarial Journal2013:5, pages 383-402. Brouhns, N ., Denuit, M., & Vermunt, J. K. (2002). Measuring the longevity risk in mortality projections. Bulletin of the Swiss Association of Actuaries (2) Carrin, G ., James, C., and Evans, D. (2005) . Achieving universal health coverage: developing.the health financing system. Geneva: World Health Organization Chan, M. F., and Kamala Devi, M. (20 15). Factors affecting life expectancy: evidence from 1980-2009 data in Singapore, Malaysia, and Thailand. Asia Pacific J. Public Health 27, 136-146. Haberman, S. , & Renshaw, A. (2008). New developments of the Lee-Carter model for mortality dynamics. Jakovljevic, M . M., Arsenijevic, J., Pavlova, M., Verhaeghe, N ., Laaser, U., & Groot, W. (20 17). Within the triangle of healthcare legacies: comparing the performance of South-Eastern European health systems. Journal of medical economics, 20(5), 483-492. https://doi.org/1 0.1080/13696998 .2016.1277228 Lee, R.D., & Cmter, L. (1992). Modelling and forecasting the time series of US mortality. Journal of the American Statistical Association 28 .J J McNeill, K., & Jacobs, C. (2019). Halfofthe world's population lack access to essential health services- are we doing enough? https://www.weforum.onz./agencla/20 19/09/half-of-the-worlcl-s-population-lack-access-to -essential-health-services-are-\ve-cloing-enough/ Ranabhat, C. L., Atkinson, J., Park, M. B., Kim, C. B., & Jakovljevic, M. (20 18). The Influence of Universal Health Coverage on Life Expectancy at Birth (LEAB) and Healthy Life Expectancy (HALE): A Multi-Country Cross-Sectional Study. Frontiers in pharmacology, 9, 960. WHO. (20 14 ). The health of the people: what works: the African regional health rep01t 2014. Rwanda Ministry ofFinance and Economic Planning (Revised 2012). Vision 2020. United Nations, Department of Economic and Social Affairs, Population Division (20 19). World Population Prospects 2019, Volume 11: Demographic Profiles Wagstaff A. & van D6orslaer E. (2003). Catastrophe and Impoverishment in Paying for Health Care: With Applications to Vietnam 1993-1998. Health Economics. Wen, C. P., Tsai, S. P., & Chung, W. S. (2008). A 1 0-year experience with universal health insurance in Taiwan: measuring changes in health and health disparity. Annals of internal medicine. WHO (2020). What is UHC? http://www. who . int/features/ga/un iversal health coverage/en/ World Health Organisation. Universal Health Coverage factsheet (20 19). https://www.who.int/news-room/fact-sheets/detai1/universal-health-coverage-(uhc) 29