BUILDING CAPACITY FOR AN EFFECTIVE SOCIAL WELFARE SYSTEM IN MONGOLIA

This working paper explores four themes in the design and delivery of social welfare in Mongolia: (i) the rationale for social welfare, (ii) whether welfare dependency is an issue, (iii) how the Government of Mongolia can strengthen the shock-responsiveness of its social protection system, and (iv) how a global innovation in social protection can help vulnerable Mongolians graduate from poverty and achieve economic inclusion. It synthesizes key findings from four policy briefs prepared at the request of Mongolian social protection policymakers with Asian Development Bank technical assistance (Building Capacity for an Effective Social Welfare System, 2019–2023). The first two themes include an overview of social welfare (or social assistance) for general readers and provide context and a framework for the Mongolian system. They also include evidence and examples of international experience to help respond to commonly held public perceptions. The remaining themes are examples of ideas that are being tested in-country, supported by global evidence, and well-aligned with the directions promoted by the Ministry of Labor and Social Protection.


I. INTRODUCTION
The Asian Development Bank (ADB) has been a leading partner in the development of Mongolia's social protection system for more than 20 years, helping to build the foundations of social assistance and social insurance. 1Since the mid-1990s, a combination of ADB loans, grants, and technical assistance (TA) projects has helped Mongolia mitigate the adverse impacts of economic crises; strengthen the sustainability of social expenditure through better targeting and consolidation of social transfers; and improve social protection strategies, programs, and services, including through capacity building.However, despite good progress, the reform agenda is still evolving and far from complete.A key lesson from ADB's Social Welfare Support Program is the importance of providing continuous support because reforming a sensitive area such as social protection requires long-term effort and commitment. 2 To respond to this lesson, from May 2019 to April 2023, ADB supported Mongolia's Ministry of Labor and Social Protection (MLSP) with a TA grant to help build capacity for a more effective social welfare system. 3One of the project's workstreams supported social welfare policy research, policy dialogue, and advocacy, aiming to be flexible and demand-driven, with research themes determined collaboratively with MLSP.In the past, evidence-based policy dialogue on existing and proposed social welfare programs was limited.A lack of clear monitoring and evaluation systems had resulted in insufficient assessment of programs and services, resulting in a limited evidence base on what works and whether there might be better, more efficient, or more effective alternatives.Consolidating fragmented social welfare programs and improving poverty targeting have both been discussed for a long time and some progress has been made, but not enough, in part because both social welfare and the reform process can be highly politicized.This makes evidence-based decision-making and advocacy around the purpose and effectiveness of particular programs-and the system as a whole-very important.
A major test of the project's demand-driven and flexible approach came with the outbreak of the coronavirus disease (COVID-19) pandemic in early 2020.Mongolia took early and decisive action as early as January 2020 to prevent the spread of COVID-19, and while the direct health impact was initially limited, economic data and several rapid assessments confirmed that the socioeconomic consequences of the virus containment efforts-including border closures, trade disruptions, and school and business closures-were substantial. 4The economy contracted by 7.3% in the first three quarters of 2020 compared to the same period in 2019.The resulting household-level shocks were severe and long-lasting, particularly on the poor and vulnerable people.As early as May 2020, (i) nearly three-quarters of all Mongolian households and 85% of poor households reported experiencing some sort of economic shock; (ii) nearly two-thirds (64%) of all households reported an increase in food prices; (iii) nearly three-quarters (73%) of self-employed workers experienced income loss; and (iv) 70% of farmers and herder households reported a decline in income compared to the previous year. 5Against this backdrop, the Government of Mongolia's shock-responsive social protection (SRSP) measures took center stage, and the project work plan was rapidly adjusted to respond to the government's request for a detailed assessment of the effectiveness of the measures.
The TA produced five policy briefs that examined social welfare challenges and themes and offered evidence-based policy recommendations to inform future directions. 6This working paper synthesizes four of those policy briefs and explores the following themes: (i) the rationale for social welfare in Mongolia and recommendations for the future; (ii) whether welfare dependency is an issue in Mongolia (it is not); (iii) how Mongolia can strengthen the shock-responsiveness of its social protection system; and (iv) how a global innovation in social protection can help vulnerable Mongolians graduate from poverty and achieve economic inclusion.
Each section presents key messages before delving into the discussion and more detailed recommendations.Sections 2 and 3 provide an overview of social welfare (social assistance) for general readers and provide context and a framework for the Mongolian system.They also present evidence and international experience to help respond to commonly held public perceptions, which in some cases, are clear misperceptions.Sections 4 and 5 focus on innovations in social welfare, providing examples of ideas supported by global evidence that are being tested in-country given their alignment with ideas that MLSP has promoted for some time.The themes of shock-responsiveness and graduation also link very well with what many in line ministries and local governments see as the role of government in helping households build and sustain basic capacities and resilience.The themes addressed in this working paper have a broader applicability beyond Mongolia, as more ADB developing member countries recognize the need to build and strengthen nascent social protection systems in anticipation of future shocks.

II. EXAMINING THE RATIONALE FOR SOCIAL WELFARE Key Messages
(i) Mongolia's legal framework defines social welfare as an instrument to reduce poverty and vulnerability and support people's livelihoods.(ii) The international literature emphasizes ethical, social, and economic justifications for investing in social welfare.The economic rationale includes (a) addressing market failures, (b) enabling people to make investments with higher returns, and (c) efficiently dealing with large-scale crises.(iii) Investment in social welfare in Mongolia should align with the country's key challenges, including climate change, demographic transition, and low skills and productivity as the key determinants of poverty.

A. Objectives and Functions of Social Welfare
Mongolia's 2012 Social Welfare Law defines social welfare as an instrument to protect poor and vulnerable people and ensure their basic needs are met (Article 3.1.1).The revised social welfare law-still under process of approval at the time of publication in mid-2023-expands on this definition to emphasize social protection as key to improving people's livelihoods.
Reasons for investing in social welfare.International policies and theories make similar assertions on the objectives of social welfare. 7These are influenced by ideologies and ethical principles, as well as theories on the causes and drivers of poverty, which then determine the role of the state and its people.
The three main reasons to invest in social welfare are (i) ethics and social justice, (ii) social cohesion, and (iii) economic efficiency.
(i) Ethics and social justice.This first broad category includes moral and ethical arguments.Many countries recognize the moral imperative to ensure that no one is in a state of great deprivation.The principle of social justice tries to ensure that everyone can succeed in life and that the state should compensate for certain disadvantages.This category also includes social protection with support across the life cycle as a fundamental human right.(ii) Social cohesion.Some countries make the case for social welfare explicitly to ensure social cohesion and stability.High inequality can be a source of conflict, and social welfare can mitigate this through income redistribution.Social welfare to protect people against risks is a way to ensure solidarity and create common ground.This category also includes the use of social welfare to support state-building and reforms.(iii) Economic efficiency.This justification holds that social welfare provides an efficient way to support economic growth and address market failures, and is based on the economic efficiency of social welfare interventions at the household, community, and national level.
Functions of social welfare.The functions of social welfare include protecting minimum income, protecting against risk, supporting improvement of people's circumstances, addressing disadvantage, reducing inequality, and increasing social inclusion.
Economic rationale for social welfare.The ethical and social reasons for welfare are important, but these tend to change slowly since they reflect a country's history and the evolution of moral principles.In contrast, the economic rationale and related empirical evidence can be used to underpin and help develop social welfare in response to macro trends and challenges, such as climate change, aging, migration, and urbanization.The economic justification for social welfare is based on (i) the presence of certain market failures, (ii) social welfare as an investment that has significant returns, and (iii) social welfare as an efficient way to deal with crises.
(i) Addressing market failures.Markets alone are unable to cover certain risks, which justifies social protection interventions to provide social insurance and intervene with social welfare for protection against catastrophic losses.In such situations, social welfare can be efficient because it prevents certain events from forcing people into less productive arrangements.At the household level, social welfare can address life-cycle risks (like sickness, unemployment, and disability) by supporting affected people to recover quickly and ensuring that there is a return to the labor market and productivity (e.g., through support for health and nutrition).(ii) Enabling investments and activities with higher returns.People are often trapped in poverty because they cannot adequately invest in human capital (education, health, and nutrition) or in productive assets and, instead, pursue activities that have low risk, but also low returns.Social welfare can support such households to move to different trajectories and thus overcome poverty traps, by enabling investment in human capital and productive assets and increasing productivity.In many countries, there is resistance to the idea of providing support to able-bodied people of working age, as this is thought to create work disincentives.However, one of the basic observations of poverty analysis, including in Mongolia, is that a very large share of poor people are working poor, and that poverty is explained by low productivity, lack of skills, and high demographic dependency ratios (the ratio of people in a household who are working to those who are not, such as children, older people, or people with disabilities).In these circumstances, social welfare can help people overcome such situations.(iii) Dealing with shocks and crises.The arguments for dealing with risk at the individual and household level also applies at the macroeconomic level.Supporting consumption in times of crisis enables countries to overcome recessions and recover more quickly.The shock caused by the COVID-19 pandemic from 2020 to 2022 was of an unprecedented scale and many countries decided to respond by increasing social welfare and especially cash transfers.While the full lessons from the social protection response to COVID-19 are still under study and will be for many years, countries with functioning and advanced social welfare systems were in a better position to respond to the crisis.During the Asian financial crisis of the late 1990s, many countries had poor social welfare systems, which caused a painful and prolonged recession.For example, it took Indonesia 8 years (1997-2005) to return to its precrisis level of real gross domestic product per capita.Both at local levels and in the overall economy, social welfare programs can support demand and have positive spillovers, which enhance economic growth.8 Poverty impact of social welfare.There is strong global evidence that social welfare reduces poverty in the short term and increases human development in both rural and urban areas, specifically through increasing school attendance.Less research, however, has been carried out on the longer-term impacts on sustainable graduation from poverty and improved job prospects. 9Evidence is strong on the productive impact of cash transfers, including that they enable more investment in agricultural inputs and tools, resulting in increased production and higher sales.Similarly, evidence suggests that cash transfers increase the ownership of livestock and diversification of off-farm activities. 10

III. IS WELFARE DEPENDENCY A PROBLEM IN MONGOLIA? Key Messages
(i) Welfare dependency implies unnecessary reliance on welfare benefits for a long time, linked to a suspicion that welfare benefits discourage work and create a culture of dependency.(ii) The welfare dependency debate has a long history in many western countries and is associated with "welfare-to-work" policies.However, the structure of benefits and the nature of poverty and economic conditions in those countries are very different from those in developing countries.(iii) Empirical studies in developing countries find no evidence that welfare benefits reduce work among beneficiaries.This is due to both lower benefit amounts and different targeting mechanisms.(iv) In Mongolia, the risk of welfare dependency is low.The Food Stamp Program is the main benefit targeted to poor people, and studies do not find evidence of welfare dependency.(v) The only risk of welfare dependency is in social welfare pensions for people with disabilities.One way to minimize this risk is to ensure that eligibility is not associated with not working, and the definition of disability is kept separate from the ability to work.

A. Defining Welfare Dependency
The concept of welfare dependency has two main facets: unnecessarily long-term reliance on welfare benefits, and a culture of reliance and exploitation of welfare benefits. 15The long-term reliance on welfare does not necessarily have a negative connotation, but some people receive assistance when they might not need it and could instead help themselves.Welfare is thought to discourage work and create a mentality or culture of dependency.The concept is also associated with the risk that welfare benefits can create poverty traps whereby people choose to not work or to work less so they can keep receiving benefits, and thus remain in low living standards.This behavior is explained by specific work disincentives that can be created by the design of social welfare benefits, such as automatic links between employment income and the benefit amount.This can create a mentality of dependence, whereby people do not work to maintain the security of social welfare benefits.
Welfare duration and dynamics.Longitudinal data allows the empirical study of welfare duration and dynamics.Studies in the United States have been particularly influential, followed by similar analysis in many European countries.In general, these empirical studies find a relatively short duration of social assistance, but indicate that a minority of recipients did receive welfare for a long period of time.However, it is difficult to link long duration to welfare dependency and relying on benefits even when people could be self-sufficient.This is difficult to prove because it is unclear to what extent the factors associated with longer receipt of welfare-such as low skills and qualifications, low experience, disability and ill health, or having many dependents-are fully responsible, or whether longer receipt of welfare is the result of people not seeking work.
Three behavioral models.Understanding the possible risk of welfare dependency is important to shape the way benefits are designed and the policies that should be put in place to complement welfare benefits.Understanding welfare dependency requires explaining people's behavior.Three main types of behavioral models explain welfare dependency: rational choice, expectancy, and cultural models. 16) Rational choice models.These are based on economic theory.Classical economic theory would predict that if a person is offered welfare income, this could affect their decision on whether to work, and how many hours they work.People's choices are affected by the extra income, and as a result, they reassess the optimal combination of work and leisure to maximize their utility.Importantly, this effect depends on the amount of the transfer, and on the level of earnings that a person receives or could receive with their qualifications, work experience, age, number of young children, care needs of other family members, disability and ill health, and so on.However, unlike in a simplified model, a person's actual decision to work will be determined not just by personal conditions, but also by those of the household, the flexibility of the labor market, and other constraints that people might face.
In rational choice models, welfare dependency can arise if work incentives are affected by inadequate benefit design.Within a certain benefit design, some people are expected to need welfare, and the same people are likely to remain in such need.Actual dependency would only arise if people remained in welfare because of wrong incentives, such as bad financial incentives.For example, benefit eligibility and amount can depend on the household's income, so that everyone is guaranteed a minimum income and the amount of the benefit is higher the lower the household income.In such cases, certain households with low earning potential and high needs would see no difference in their income in a situation where they work or not, because the difference would be provided by the benefit.These households face a 100% tax rate on income earned from work because the welfare benefits would be reduced to an amount equivalent to that earned, thus making no change to their overall income.
(ii) Expectancy models.These emphasize the importance of people's confidence and control over their lives.Failures can shape people's motivations and aspirations and result in a loss of confidence which, in turn, could encourage welfare dependency.Within such behavioral theories, receipt of welfare benefits is more likely in people with low qualifications and skills as predicted by rational choice models, but there will also be other factors that could affect expectations.In these models, the solution to welfare dependency is not so much about financial incentives, but more about interventions that could boost people's self-confidence and provide more information to people about opportunities.
(iii) Cultural models.These argue that people's behavior can be affected by the environment in which they live, and that living at the margins of society can result in a vicious cycle of stigmatization that reinforces exclusion and antisocial behavior.Within that environment there can be an acceptance that receiving welfare is a perfectly normal way to live.Cultural models describe situations in specific geographic areas with high levels of poverty and deprivation, where poverty is often associated with crime, and alcohol and drug use.According to cultural models, receipt of welfare would be geographically concentrated and very stable in deprived areas, but much more dynamic in other areas.Welfare dependency should be tackled by providing more economic opportunities, and by making the level of support conditional on certain obligations to push people into work.
These three different theories of welfare dependency can have divergent predictions of what we should expect in terms of welfare duration, movement in and out of welfare programs, and policies that could be implemented to influence people's behavior.

B. Evidence from Developing Countries
In many countries of the Organisation for Economic Co-operation and Development (OECD), welfareto-work policies have improved work incentives by changing the design of benefits and reducing the high marginal tax rates associated with guaranteed minimum income benefits.This has followed interventions primarily suggested by rational choice models, but expectancy models are also considered and used to shape some of the policy responses. 17It is important to understand that means-tested benefits in OECD countries are very different from the type of benefits predominant in developing countries.Poverty and labor market conditions are also significantly different.For Mongolia, therefore, it is more useful to focus on the literature from developing countries.Key differences between the OECD and developing country contexts include the following: (i) Transfer amount.The level of support provided by welfare benefits in developing countries is relatively low compared to means-tested benefits in OECD countries.This is not only in absolute terms, but also in relative terms (as a percentage of beneficiaries' overall income or consumption).An international review of the level of social welfare transfers in developing countries found their average value to be 10% of beneficiary income or consumption, reaching 19% if the focus is on the value of cash transfers received by the poorest quintile. 18In European countries, the average guaranteed minimum income is about 40% of the median national income, which means that the value of the benefit as a percentage of poorer beneficiaries' income is even higher. 19(ii) Targeting approach.While means testing is common in OECD countries, it is rare in developing countries and there is no automatic or perfect substitution between increased income from work and a reduced value of benefits.Instead, developing countries more commonly use indirect living standards measures (proxy means testing) or categorical, community, and geographic targeting.(iii) Frequency of reapplication.In OECD countries, there is often a frequent recertification of people's eligibility to receive support, based on the assumption that support should usually be temporary and linked to episodes of unemployment.In developing countries, recertification tends to be done less regularly, which entails a prolonged period of support.(v) Labor markets and poverty characteristics.The level of informality and the number of working poor in developing countries is substantially higher than in OECD countries.
Poverty is more widespread in developing countries than in OECD countries and it is not necessarily associated with unemployment, but rather with low-paid jobs and specific household characteristics, e.g., large households with many children or dependent members.Developing countries also face more problems of incomplete markets, lack of access to financial markets, and child labor.
In developing countries, there is no evidence to support a negative impact of social welfare on work countries finds no sign of welfare dependency. 20Another study reviewed data from randomized control trials in seven countries and found no evidence that government cash transfer programs discourage work. 21Instead, welfare transfers are more likely to have the positive effect of reducing child labor or encouraging people to invest in more productive economic activities.Many rigorous studies find no evidence of welfare dependency.

C. Social Welfare Benefits in Mongolia
Having defined welfare dependency and its determinants and reviewed the evidence from developing countries, we are better equipped to address the question of the relevance of social welfare dependency in Mongolia.However, this requires understanding the main characteristics of social welfare benefits in Mongolia, considering the benefit amount, and identifying the targeting or eligibility criteria.
Categorical benefits.The main social welfare benefits in Mongolia are categorical benefits, which are benefits provided to certain groups of people, but not in relation to their income or living standards.There are transfer programs with eligibility based on age (children and older people), disability, motherhood, single parents with many children, and so on.For example, the CMP is universal, as it is provided to all children under 18.Other benefits are given to pregnant and breastfeeding mothers, or mothers who have more than four children.Eligibility for these benefits is not constrained by the income of the person or of the household in which the person lives.The amount of such benefits supports household expenditure and needs, but it is far from being the main source of income. 22cial welfare pensions.Social welfare pensions are provided to older people who do not receive any other social insurance pension and who are above a certain age or have certain disabilities.In this case, the benefit is "pensions-tested" and involves a significant amount, but it is provided to people who are already of pension age.There is, therefore, no work disincentive.For social welfare pensions provided to people with disabilities who do not qualify for a social insurance pension, there is a risk of labelling such people as "unable to work" and thus making the receipt of support conditional on not working.The disability assessment determines the percentage of "lost ability to work," so that if a person is considered to have lost more than 70%, they cannot work, and receive instead the social welfare pension, while those labelled as having lost 50%-70% of their ability can theoretically combine work and receipt of social welfare pension.Nevertheless, associating disability with a loss of ability to work goes against the principles defined in the Mongolia's 2016 Law on Persons with Disabilities and could discourage incentives to work.food stamps, strict requirements to participate in employment services and accept job offers might compromise receipt of support (footnote 14).It found no evidence suggesting that people reduce work because they receive benefits.Further investigation of the risk of welfare dependency in relation to the FSP requires analysis of the factors associated with people's participation in the labor market.
In the 2018 Household Socio-Economic Survey (footnote 12), the discrepancy between recipients and nonrecipients of food stamps in the share of people employed is stark: men who receive the FSP have an employment rate 11.4 percentage points lower and women 25.9 percentage points lower than non-FSP recipients (see Figure 1).However, as documented in the ADB FSP assessment, the characteristics of FSP beneficiaries are very different: they have lower education, qualifications, and skills; a disproportionate number have disabilities; and many live in large households with many young children.For this reason, it is important to examine whether, after accounting for various observable characteristics, we still find that receiving food stamps is associated with lower employment rates.For men, the effect of the FSP is close to zero after controlling for age, household income, educational attainment, marital status, place of residence, having functional limitations or other household members with disabilities, number of children, and belonging to a household in the bottom part of the consumption distribution.For women, FSP still has an effect after controlling for observable characteristics, but it is greatly reduced: the employment rate is only 4 percentage points lower.As a comparison, having a serious functional limitation reduces the employment rate by 40 percentage points.The key takeaway from this is that receiving the FSP is not associated with reducing work.
No evidence of welfare dependency.Mongolia's welfare benefits are not likely to generate a problem of welfare dependency, both because of the eligibility rules and the relatively low benefit amounts.The only benefit that poses a significant risk is the social welfare pension for people with disabilities.This benefit could discourage work if the entitlement is linked to people's nonparticipation in the labor market, and if disability is defined as a lost ability to work.This is a key area for attention under the reform of disability assessments and the level of support that should be provided to people with disabilities who are willing and able to work.

IV. SHOCK-RESPONSIVE SOCIAL PROTECTION: PANDEMIC LESSONS AND POLICY IMPLICATIONS Key Messages
(i) Shock-responsive social protection complements disaster risk management by using existing social protection systems to respond to external shocks that affect large segments of the population.(ii) Social protection was prominent in Mongolia's response to COVID-19, helping to avoid an increase in poverty and providing the basis for a quicker recovery.(iii) As the country moves toward transition and recovery, policy measures should be adapted to support those sectors that remain badly affected by the crisis.Active labor market policies may support transition to alternative jobs.(iv) Beyond COVID-19, better systems are needed to respond to recurrent crises, such as dzuds (especially harsh winters) and macroeconomic shocks.(v) Better shock-responsiveness of social protection systems requires legislative and regulatory amendments, improved financing arrangements, and strengthened institutional capacity to quickly deploy emergency measures.

A. Defining Shock-Responsive Social Protection
SRSP is the use of social protection systems and programs explicitly to respond to, prepare for, or mitigate the impacts of covariate shocks.Covariate shocks are the large, external shocks that tend to affect entire communities. 23There is a fundamental difference between large, external shocks and those that tend to affect people at an individual or household level or are to some extent predictable because they are part of the life cycle, such as aging.Traditionally, social protection deals with individual and household-level shocks by addressing vulnerabilities across the life cycle.However, the increasing frequency of disasters has prompted the need to develop better ways to respond to sudden crises affecting whole communities, countries, and regions.
In many countries, responding to emergencies is usually considered part of disaster risk management (DRM) and in some contexts is supported by humanitarian aid.SRSP does not replace DRM and humanitarian aid, but is a way to complement it.Social protection can intervene in two main ways in relation to disasters: (i) building household resilience before a disaster, and/or (ii) supporting affected households and prevent the use of harmful coping strategies after a disaster. 24Following a disaster, SRSP can support the emergency and recovery phase through two means: (i) Vertical expansion.This involves providing additional support to existing beneficiaries of social protection programs for a limited duration, e.g., "top-ups" for cash transfer recipients.(ii) Horizontal expansion.This involves adding beneficiaries by using existing programs to reach more people for a limited duration (temporary expansion) or developing new programs using existing beneficiary identification and benefit delivery systems.
Even though the existing literature emphasizes the role of social welfare (or social assistance) within SRSP, we deliberately speak of social protection because measures to respond to shocks should involve all three components of social protection, i.e., social insurance, social assistance, and labor market interventions.Their use depends both on the country context and the nature of the crisis.For SRSP to be effective, countries need to develop several key elements well in advance of a shock: legislative and regulatory frameworks, financing, beneficiary identification, and benefit delivery mechanisms.

B. Lessons from Mongolia's Pandemic Response
The COVID-19 pandemic was an unprecedented global shock and social protection provided a fundamental instrument to reduce the negative economic effects of the crisis.Almost all governments adopted new social protection interventions in 2020 as a pandemic response.Social assistance measures-such as cash transfers, waiver of utility payments, and in-kind transfers-were the most popular, followed by social insurance measures, such as unemployment benefits and paid sick leave; waiving, reducing, or deferring social insurance contributions; and labor market policies like wage subsidies and adjustments to labor regulations.Most responses involved horizontal expansion (enrolment of more beneficiaries in existing programs) and creation of new programs. 25e Government of Mongolia delivered a significant social protection response.Social welfare measures included increasing CMP payments from Mongolian togrog (MNT) 20,000 (about $7) to MNT100,000 per child per month, starting in April 2020.Initially for 6 months, these top-ups were extended several times and continue at the time of writing in 2023.Other responses included increased social welfare pensions and FSP payments, as well as subsidies for cashmere farmers.The social insurance response included waivers of mandatory social insurance contributions.
Through the CMP top-ups, the government was able to quickly provide support to four out of five Mongolians (all people living in a household with at least one child).This was a justified intervention especially in the short term because of the widespread effect of the crisis and the ability to deliver payments quickly using the CMP's electronic direct deposits to bank accounts in the children's names.Mothers are the custodians of the children's accounts in about 86% of households, so the CMP top-ups had an important gender benefit as well.
In case of a shock, the rationale for a social protection intervention in addition to traditional disaster response is to prevent the adoption of harmful coping strategies that could put both households and the country in a much longer negative spell and increase the time for recovery.Therefore, in the context of a shock, a social protection intervention has both a moral and an efficiency justification.
Trajectories of recovery.The trajectory of recovery from a shock with and without social protection interventions is graphically simplified in Figure 2. If people lack the ability to cope with the crisis, fall into poverty, and are forced to adopt negative strategies, such as selling productive assets, this could have medium-to long-term consequences.Panel survey data from Mongolia shows that exposure to extreme winters persistently slowed children's growth among herding households, with negative effects still evident 3-4 years after the shock. 26In contrast, if consumption levels are supported through social protection, a much quicker recovery becomes possible.Shock-responsive social protection and poverty reduction.The economic effects of the pandemic (border closures, lockdowns, and other measures adopted to contain the spread of the coronavirus) would have increased the share of people below the poverty line from 28% to 37% in 2020, according to an ADB study (footnote 4).However, timely shock response through social protection prevented such an increase and likely even reduced the overall level of poverty.The same ADB analysis projected that the CMP alone reduced poverty to 25%, and all social protection measures together further reduced it to 18%.While these results are simulations that depend on several assumptions, they still show the likely impact of social protection in preventing poverty increase and the significant contribution to containing the crisis.The likely positive impact of social protection measures on poverty was confirmed by a study that used value-added tax data to indirectly assess poverty and inequality in 2020. 27 many respects, the global pandemic-induced recession was exceptional.The nature of the shock affected a very large share of the population, including people who were previously not poor.This means that interventions that reached a large percentage of the population were appropriate to reach the new poor. 28ration of support.A vertical expansion, such as the one provided through the CMP and that has lasted for 2 years, nevertheless raises several questions.Was the increase in the benefit amount clearly presented as temporary relief?How will people react when the amount is reduced?What is the strategy to adopt when moving from acute emergency to transition and recovery?As the pandemic situation improves, one path is for policy interventions to focus on the economic sectors that remain affected, such as the tourism and hospitality industries, and on developing active labor market policies to support employees to find alternative job opportunities where needed.SRSP policies need to adapt to the changing context.
Suitability of existing programs.Another important question is the suitability of using existing social protection programs to accurately target those affected by shocks.While increasing the benefit amount of a universal program such as CMP was the correct policy intervention for a short period and as a response to a pandemic, this is not necessarily always the best option.Climate-related shocks, such as dzuds could become more concentrated and clearly affect only certain population subgroups, such as herders.This calls for a wider assessment of suitability of various social protection programs for shock response, in parallel to DRM responses.Prior to the pandemic, a pilot project of the United Nations Children's Fund (UNICEF) in Zavkhan Aimag aimed at preventing negative effects due to dzuds by identifying soums (districts) most at risk, supporting the preparation of the response to a crisis, and then providing actual support in an expanded list of soums. 29The pilot focused on providing a CMP top-up to children under the age of 5 years.This geographically and categorically targeted pilot was assessed using postdistribution monitoring surveys after each payment, and while these do not offer a full evaluation, the effort represents a good example of the type of adjustments that might be required to improve the responsiveness of the system to a recurring shock in Mongolia.

C. Policy Implications
Recurrent shocks in Mongolia: is there a strategy?The country's recurrent shocks tend to be due to dzuds and volatility of global mineral commodity prices, which can induce significant macroeconomic shocks.Mongolia is prone to extremely harsh weather conditions that tend to drastically affect the livestock sector, which employs a large share of the population.Dzuds not only cause direct livestock losses, but also can determine sharp price movements.The frequency of extreme weather events is expected to increase in the future, and climate change predictions have negative implications for the livestock sector.
The National Emergency Management Agency in Mongolia plays an important role in DRM, but has a restricted remit: it works primarily in emergency response and focuses on direct losses, such as livestock loss, whereas the effects of dzuds are also indirect through inflationary pressures.In this context, social protection can play a significant complementary role beyond emergency response in supporting affected households during recovery.
Legislative gaps.For SRSP to be effective, timeliness is critical.This can be achieved if the system is well-prepared in advance of exposure to shocks.This means that legislation allowing shock response through social protection needs to be in place in addition to financing arrangements and systems for the accurate identification of potential beneficiaries and the delivery of benefits.Mongolia's current legislation does not allow for the flexible use of social protection instruments to respond to a crisisand indeed the response to the COVID-19 pandemic required the development and approval of emergency provisions. 30Especially for recurrent shocks, there is a strong rationale to allow the current social protection system to be deployed to assist people in need, and some simple provisions in the legislation could make this significantly easier.Amending the legislation could also create the basis for better financing in case of shocks, since at present only the National Emergency Management Agency and local governments can access reserve funds.The Integrated Household Database.Mongolia has the advantage of a good household registry system, the Integrated Household Database, which covers a large share of the population.This system is used to target poor households for the FSP, based on a proxy means test score, and household registration in the database is a prerequisite for receiving the CMP.It could also be the basis to provide information in an emergency.However, keeping the database updated is an ongoing challenge.It is also important to think in advance how this information could be used to operationalize such an emergency response program, including needs assessments, benefit amounts, payment systems, communication, and grievance procedures.As in the UNICEF pilot, there could be scope to consider the use of preventive payments before the shock hits in areas that are particularly at risk.
Toward a more effective shock-responsive social protection system.Mongolia's pandemic response has shown the importance of SRSP.Prompt interventions that provide adequate support and coverage can mitigate the effect of recessions and enable a quicker recovery.However, there are important lessons to consider in increasing the potential to use social protection to respond to recurrent shocks, such as dzuds.These are summarized in the table below, which shows the key building blocks of an effective SRSP system and the recommended actions to develop them.

Building Blocks of a Shock-Responsive Social Protection System and Recommended Actions in Mongolia
SRSP System Building Blocks Recommended Actions Legislative and regulatory frameworks Include in the social welfare law the possibility of using social protection as a tool to respond to shocks.

Financing arrangements a
Develop clear financing arrangements to support the deployment of emergency social protection responses and the use of reserve and contingency funds by the Ministry of Labor and Social Protection.
Explore financing agreements with international financial institutions that could be promptly activated in case of shocks, ensuring the clarification of conditions.

Systems to identify potential beneficiaries affected by shocks
Strengthen the Integrated Household Database as an ideal system to identify beneficiaries both in the case of a shock and potentially before a shock, for transfers that support household preparedness.
Further invest in dynamic needs assessment and in the interoperability of the database.
Increase the use of such data across programs both to improve its quality and to justify its cost.

Preparedness to deploy SRSP solutions
Determine the key design features of emergency programs before a shock: whom to cover, length of support, and determination of amount of support.
Implement clear communication campaigns.

Delivery of social protection emergency measures
Though electronic delivery of benefit payments is well developed in Mongolia, explore how to ensure money is readily available and not automatically locked in saving accounts, as well as the possibility of internet banking.

V. THE GRADUATION APPROACH: A PROMISING INNOVATION IN SOCIAL PROTECTION Key Messages
(i) The graduation approach delivers coordinated welfare, training, employment, and financial services to poor and vulnerable people with an emphasis on coaching, mentoring, and psychosocial support to build confidence.(ii) The government's National Program for Targeted Household Development and Protection was a type of graduation program that worked directly with households, but implementation was varied, given differences in staff capacity and a lack of dedicated program resources.(iii) Graduation programs implemented by nongovernment organizations usually perform well because of their resource-intensive nature, including dedicated staff and budgets.To be more successful, government-led programs require strong multisector coordination and a commitment to reducing capacity constraints among staff.(iv) Graduation programs should be integrated into a social protection system that offers "graduated" households the option of reapplying for social assistance benefits if they experience a shock or fall back into poverty.(v) Graduation programs should be designed to support access to formal social insurance.
Also known as productive economic inclusion, the graduation approach in social protection works with poor and vulnerable households, building on a foundation of social welfare benefits with holistic support to tackle the multiple facets of deprivation.The objective is to gradually move households out of poverty and on to a trajectory of diversified livelihoods, improved living standards, and better resilience to shocks.The approach starts from the premise that poverty has many dimensions, so no single intervention by itself (such as education, training, or cash support) will be able to eliminate poverty completely.Instead, a multidimensional approach is needed that starts with, but goes beyond consumption support, such as cash grants.The programs pioneered in Bangladesh in 2002 targeted "ultra-poor" households.Over time, more than 200 graduation programs and similar economic inclusion interventions have been adapted across 75 countries, reaching nearly 92 million people. 31Rigorous impact evaluation has shown both immediate benefits and sustainable results for some time after the program interventions ended. 32listic and sequenced support.Graduation programs offer poor participants a carefully sequenced set of solutions that are time-bound, usually for a period of 18-36 months (see Figure 3). 33These programs support household consumption in the short term while aiming to develop sustainable and resilient livelihoods in the long run.This is in contrast to other types of social assistance programs that provide cash or other benefits, but do not make assumptions on the trajectory out of poverty, and the duration of participation is not strictly limited.Without explicitly calling it graduation or productive economic inclusion, the NPTHDP essentially adopted a graduation-like approach, recognizing that poor households need multiple forms of support to transition into sustainable livelihoods.MLSP implemented the program through local social welfare offices, identifying 55,000 targeted households (13,500 per year) based on their living standards and their potential for improved livelihoods.Social workers used a "household development triangle" to determine households' baseline status considering (i) basic needs, such as housing, food, consumption, finance, and employment; (ii) capacities and opportunities of the household members; and (iii) morality and respect, or family values, social relations, and participation.Household members, in collaboration with social workers, were then expected to follow a tailored household development plan to work on improving their standard of living.This tailored plan aimed to highlight the process through which households would achieve improved livelihoods based on their individual situation, needs, resources and capabilities.Social workers were to assess household progress annually.
Challenges in past programs.An ADB-supported assessment examined some of NPTHDP's challenges to help inform future social protection programs aiming to graduate households out of poverty and into sustainable livelihoods: 36 (i) While program implementers and participants generally perceived the program to be relevant and saw the household development plan as a useful tool, it was rarely used in practice primarily due to insufficient training for social workers.(ii) Most participants were not aware of the assessment process using the household development triangle; a simplified process coupled with more training for social workers may have ensured greater awareness and uptake.(iii) Social workers connected households to existing services and training programs and recorded their attendance.A more effective measure of progress would have been a direct measure of outcomes (such as improved standard of living or poverty reduction) rather than using participation in two or three services as a proxy.(iv) Program implementation varied considerably across the country mainly due to differences in staff capacity, reported attitudes of beneficiaries, and complementary community initiatives.
A more consistent roll-out of a plan, training, and routine monitoring would ensure more standard implementation.(v) Supporting household development plans, linkages to services, and monitoring progress required social welfare staff to maintain a sustained high level of engagement and case management.However, key staff issues included high workload, staff turnover, limited motivation, and limited training.Reducing workloads and caseloads could have increased staff retention and time spent with households while boosting staff motivation and beneficiary outcomes.(vi) The lack of a dedicated budget associated with the NPTHDP made it very difficult to achieve the program's ambitious targets.While some households could be linked with existing services, others required tailored services.(vii) Motivated and dynamic households were more likely to take advantage of the services and training, while uptake was lower for less engaged households (those that likely needed it the most).Common lessons.The ADB assessment of the NPTHDP and other pilots that leverage existing government social assistance programs present some common lessons in five broad categories: (i) Resources and staffing.Graduation programs implemented by NGOs usually perform well because they are well resourced with dedicated field staff, well-designed training, and capital-intensive interventions that provide participants with productive asset transfers, such as livestock or equipment and supplies for small businesses.Scaling-up this concept to a government-run program requires complex multisector coordination.A reliance on social welfare staff may not be feasible, given their many competing responsibilities.Case management for a graduation program can place a significant demand on their time if not well integrated into their work plan and caseload.This points toward a need to think through the implications of scaling up complex programs while relying on government systems by putting in place the required resources and integration of services.Government purchasing or outsourcing of service delivery to NGOs is also a viable solution, as partly envisioned in the draft revisions to the law on social welfare.This is the model being tested under the ADB-MLSP Mongolia Graduation Pilot Program, 2021-2023, with impact evaluation results expected in 2024.(ii) Sustainability.The sustainability of graduation depends on the ability of governments to institutionalize such programs and adopt a long-term approach.This means systematically building capacity of the social welfare workforce and strengthening processes to identify households, then linking them with existing social services, such as counselling, access to finance, and health, education and training services.For many recipients of social assistance, existing social and employment services need to be tailored so that they are easier to access and better suited to specific needs, e.g., for people with disabilities.Finally, Mongolia's NPTHDP demonstrated that for such programs to be institutionalized, a significant financial commitment is necessary in the form of a dedicated budget at the central level, as well as funds for operational expenses, such as travel and communication costs for social welfare staff.(iii) Ensuring that social assistance has a "revolving door."It is important to be clear what graduation means, and how it should and should not be used: the graduation approach does not intend to exit all poor households from all social welfare programs.There will always be some households, especially those that are labor-incapacitated or labor-constrained, which may never be able to adopt sustainable livelihoods and achieve a higher standard of living.These could be households headed by older persons, or households with many children or other members with high care needs, who cannot participate in the labor market.The inherent characteristics of such households combined with the precarious nature of their typical livelihoods (e.g., raising livestock) means that they will require significant social assistance and continued social services for long periods of time.In addition to life-cycle risks, chronically poor households are more susceptible to falling back in poverty when exposed to shocks, such as disasters.Graduation programs should be integrated into a social protection system that provides a "revolving door" so that households that fall back into poverty are able to be reassessed and can access social assistance when they need it. 37he extensive reach of Mongolia's social workers coupled with frequent surveys and ondemand updates to the social registry (the Integrated Household Database) should make this easier to implement.(iv) Graduating into what?Another key consideration is what households are graduating into when they are assessed to have achieved the program's graduation criteria, which can vary significantly from program to program.Graduation programs can help encourage a sustained improvement in livelihoods by strengthening links across social assistance, labor market programs, and social insurance schemes.Many of the households receiving social assistance include working-age individuals engaged in the informal economy and, therefore, do not receive social insurance. 38Informal workers earn less than workers in formal sector jobs, in part due to lower skills.Women in Mongolia's informal economy are more often found in the most vulnerable work situations. 39Households that participate in graduation programs are expected to achieve a sustained improvement in standard of living typically as self-employed individuals.It is important to ensure that these livelihoods do not involve precarious work in the informal economy, which is highly correlated with the probability of falling back into poverty if faced with a shock, such as death, illness, or disaster.(v) Social insurance.Graduation programs should ideally be designed to support workers' transition into the formal economy, where they can contribute to social insurance and, in case of need, access social insurance benefits.This is particularly relevant for Mongolia given the relatively large size of the informal economy: in 2019, before the pandemic, informal employment was 41.5% of total employment including agriculture, and 22.8% excluding agriculture (footnote 10).Mongolia has made efforts to improve coverage of social insurance, including allowing self-employed workers, herders, and informal workers to participate in voluntary social insurance schemes.Graduation programs should be designed so that both participating and graduated households can be guided toward forms of voluntary social insurance.This should complement efforts to reform the contributory social insurance system to allow informal workers to participate despite irregular contributions.Finally, graduation programs can seek linkages with employment services-such as those for people with disabilities-or labor market schemes like public works programs which can act as safety nets for households who are exposed to shocks.
Moving forward with a new pilot.Learning from these and other lessons, a new government-led pilot supported by ADB is being implemented from 2021 to 2024 in Ulaanbaatar, targeting 1,520 working-age adults from poor households.Rigorous impact evaluation through a randomized controlled trial will assess the impact of layering graduation elements on top of social welfare benefits.The MLSP has engaged an NGO implementation partner, the Mongolian Red Cross Society, to work in the three districts of Bayanzürkh, Chingeltei, and Songino Khairkhan, selected for their relatively larger poor and vulnerable populations.Eligible households are those receiving both food stamps and child grants.The design has drawn on international and national lessons learned, including from the NPTHDP.The team includes 34 coaches assigned to work closely with households, and detailed market assessment by the Independent Research Institute of Mongolia has informed asset selection and developed tools for household profiling and enterprise matching.Impact evaluation results are expected in 2024.
As Mongolia finalizes revisions to its social welfare law and aims to build and consolidate capacity for a more effective social welfare system, it will be important to learn lessons from previous graduation experiences, both local and international.The rich experience from graduation and similar approaches in Mongolia provides clear lessons for better linking such programs to the wider social protection system to create sustained improvement in the lives of poor and vulnerable Mongolians.

Figure 1 :
Figure 1: Percentage of Employed among the Working-Age Population

25U.
Gentilini et al. 2020.Social Protection and Jobs Responses to COVID-19: A Real-Time Review of Country Measures.Washington, DC: World Bank.

Figure 2 :
Figure 2: Illustrative Trajectories of Recovery from a Shock with and without Social Protection 33

( i )
Social protection.Delivered through social welfare transfers (usually cash) to help families meet their immediate consumption needs.(ii) Coaching and mentorship.This is done by a trained facilitator or social worker who provides support tailored to each participant through regular home visits and aligned with their family development plan.(iii) Livelihoods promotion.Done through an asset transfer selected by the participant from a list of options, based on a detailed local market and value chain analysis, and with focused technical and business skills training to develop the capacity to turn the asset into a productive enterprise.(iv) Social empowerment.This helps participants grow their confidence, integrate more into their communities, and develop new life skills.(v) Financial inclusion.This involves financial literacy training, savings encouragement, and connection to formal banking institutions where needed.

Figure 3 :
Figure 3: The Graduation Approach

36T.
Bolormaa and B. Enkhtsetseg.2021.Assessment of the National Program to Support Targeted Household Development and Protection.Manila: ADB (TA 9893-MON).Unpublished.This is a working paper prepared by Oxford Policy Management for MLSP that included a survey of program staff and beneficiaries.
The poverty-targeted Food Stamp Program.The only benefit that is currently given in relation to a household's poverty condition is the Food Stamp Program (FSP).While it supports poor households to smooth consumption, the benefit amount is very small: in 2018 it made up only 6% of recipient households' total consumption.People can work and still receive support.The benefit amount is not automatically affected by the level of income of the household: either a household qualifies (scoring below the proxy means test cut-off) and receives the full FSP amount, or it does not qualify and receives nothing.The amount varies depending on the number of adults and children in the household.An ADB assessment of the FSP in 2022 examined new employment conditions initiated by MLSP and concluded that, while employment services are needed and utilized by the people who receive 20 S. Baird, D. McKenzie, and B. Özler.2018.The Effects of Cash Transfers on Adult Labor Market Outcomes.Policy Research Working Paper.No. 8404.Washington, DC: World Bank.21 A. Banerjee et al. 2017.Debunking the Stereotype of the Lazy Welfare Recipient: Evidence from Cash Transfer Programs.World Bank Research Observer.32 (2).pp.155-184.
The typical program components (with variations across different countries and contexts) include the following: 34 31 C. Andrews et al. 2021.The State of Economic Inclusion Report 2021: The Potential to Scale.Washington, DC: World Bank.Banerjee et al. 2015.A Multifaceted Program Causes Lasting Progress for the Very Poor: Evidence from Six Countries.Science.348 (6236). 32A.