The distributional incidence of the Chilean water subsidy scheme is revisited by analyzing its evolution from 1998 to 2015. This is one of the only means-tested water subsidies in a developing country and is frequently used as an example in policy discussions and recommendations. Many changes have been introduced in the program since its inception and at least three different targeting instruments have been used to identify needy households in the last 20 years. We find that the incidence of the subsidy is progressive but moderate, with a Gini coefficient of close to 0.3. It has also remained stable between 1998 and 2015. The errors of inclusion and exclusion have also remained stubbornly high. These incidence results are surprising given the efforts made in the Chilean welfare system to target social benefits. Possible explanations for these results are given and compared to other developing country experiences.

Water affordability for the poor is a perennial subject in developing countries. Despite many advances during the last two decades, there are still many challenges to make piped water coverage universal and affordable among poorer households in these countries. Water subsidy policy is an essential element in this context. Countries or cities have approached the issue of water subsidy from very different perspectives. However, rational policymaking calls for these subsidies to be well targeted to benefit those most in need. Therefore, the distributional impact of water subsidies is of prime importance.

Gómez-Lobo & Contreras (2003) analyzed the distributional impacts of water subsidies in Chile and Colombia. Both countries have designed instruments to attempt to better target subsidies to poor households. In Colombia, a dwelling geographic classification system is used to distinguish the socioeconomic condition of households. In Chile, a means-tested welfare system using household interviews and other sources of information is used to classify households. The Chilean welfare classification system and its evolution is described further below. By means-tested, we refer to a subsidy scheme where beneficiaries are chosen individually according to their socioeconomic condition measured using administrative data or other instruments; means-tested schemes contrast with universal subsidies that benefit all users of a service. Gómez-Lobo & Contreras (2003) found that the Chilean water system is better targeted to the poor, having lower errors of inclusion. However, the Colombian system is more generous and has less errors of exclusion.

In this paper, the distributional properties of the Chilean water subsidy scheme are revisited and the targeting properties of this subsidy from 1998 to the 2015 are analyzed. We believe this is important for several reasons.

First, the Chilean water subsidy scheme is one of only few examples in the developing world of a formal means-tested subsidy in this utility industry. It is based on an explicit legal and regulatory framework, has national coverage, is funded from general government funds and benefits customers from many different regional water supply companies. As such, this scheme offers an interesting contrast to the informal and universal subsidies more commonly encountered in many utility industries. Gómez-Lobo & Contreras (2003) found using 1998 data that this subsidy had relatively good targeting properties. However, it is of interest to examine whether this is still the case after almost two decades have passed.

Second, and related to the above point, several changes have been introduced to the subsidy scheme since 1998. The scheme was expanded to increase coverage and a 100% subsidy was introduced for beneficiaries of an all-encompassing welfare program (‘Chile Solidario’) designed for very poor households. Indeed, poor households get their water for free.

Third, since 1998 Chile has experienced strong economic growth. According to World Bank figures, Chile's income per capita increased from 5,480 USD in 1998 to 13,793 USD in 2016. Consistently, poverty has markedly decreased over time. While in 1998 the official poverty rate was 22%, in 2015 this figure was close to 7%. Economic growth and an expansion in public policies are the main drivers to support this significant poverty reduction. Other demographic characteristics have also show significant change over that period. Household size decreased from 4.69 in 1998 to 4.02 in 2015. The significant reduction in the number of children during the period along with the increase in childcare services may help to explain the increase in female labour participation from 40% to 47% during the same period. This positive trend in terms of socioeconomic conditions is also observed in other Latin American countries. Finally, according to official figures, inequality has also shown an important decrease. In 1998, The Gini coefficient of per capita income was 0.534, while in 2015 this number was 0.486 (the Gini coefficient is the most common measure of inequality and fluctuates between perfect equality (Gini = 0) and perfect inequality (Gini = 1); see Deaton (1997) for further details).

This may have several implications for the water subsidy scheme. On the one hand, it may make socioeconomic means testing challenging as households become richer and harder to distinguish by dwelling characteristics or ownership of durable goods, something that will be discussed further below. On the other hand, increasing incomes may reduce the need for water subsidies. This paper will shed light on these issues for the case of Chile.

From a wider perspective, the targeting properties of water subsidies are widely discussed in literature. Fuente et al. (2016) present a summary of 21 studies in a broad set of countries, with different allocation schemes, and different data sets and methodologies. According to this evidence, in most of the countries the targeting properties of subsidies were classified as poor (the allocation of public resources was worse than if subsidies were equally or randomly distributed). In only a few cases, the allocation was identified as moderate, which means that targeting is marginally better than if subsidies were randomly distributed. In the conclusions to this paper, we will return to the international evidence to compare our results relative to the world experience.

The next section provides a description of the Chilean subsidy scheme and its evolution over time. This is followed by a description of the means-testing targeting instrument used in the Chilean welfare system for all social benefits, including the water subsidy and how it has evolved over time. The following section presents the distributional analysis of the water subsidy scheme using relative and absolute concentration curves of the total monetary transfers and the total number of beneficiaries. As will become clear, from a methodological point of view, we follow Gómez-Lobo & Contreras (2003). The paper concludes with some policy implications and discussion.

The Chilean water subsidy scheme was introduced in 1990 to counter the adverse social impacts of rising water charges to cost recovery levels.

The Ministry of Social Development (MDS) – created in October 2011 and which replaced the Ministry of Planning and Cooperation (MIDEPLAN), as it was known before that date – administers the subsidy program along with municipal governments. The ministry determines the number of subsidies that will be offered to each region the following fiscal year and the value of each regional subsidy. Municipalities are responsible for the management of the scheme.

The subsidy program gives eligible households the right to consume water, sewerage and wastewater treatment services at a lower price (expressed as a percentage of the full tariff) up to a certain monthly limit; in this paper, water tariff or bill refers to the charges for all three services (piped water, sewerage and wastewater treatment). Beyond this limit, additional water consumption is charged at the full tariff. Thus, the subsidy operates like a rising block tariff structure that subsidizes the first block of consumption. The MDS determines the consumption ceiling that the subsidy will cover (which has been fixed at 15 m3 per month per household in all regions for all the years analyzed in this study) and the percentage discount for this first block of consumption, which legally can vary between 25% and 85% of the full tariff. To put this into perspective, 15 m3 of water amounts to about 125 litre/person/day for a four-member household. If a household consumes less than the 15 m3 limit, it is billed for the actual consumption and the percentage discount of the subsidy applies to that bill. The number and intensity (percentage discount) of the subsidy is differentiated geographically (generally by municipalities) and among income groups, as will be explained further below.

There are several beneficiary groups for the subsidy. In its inception, the target population were households in the first quintile of the income distribution in each region. The program was subsequently expanded with the aim of reaching not only poor and indigent families but also other socioeconomic groups that had difficulties paying their water bill. Currently, the beneficiaries of the subsidy are households that live in permanent dwellings, that have a piped water connection, that are up to date in the payment of the service but would allocate 3% or more of their average monthly income to pay for a standard water bill (15 m3 of water including water, sewage and sewage treatment charges). Until 2001 the cutoff level was 5%; in 2002, this was reduced to 3% increasing the required number and intensity of subsidies (MDS, 2018). The Pan American Health Organization (OPS) recommends that no more than 3–5% of the monthly average income of a family group be destined to the payment of drinking water and sewerage. Subsidies also benefit households connected to rural water cooperative systems.

In 2004, an all-encompassing welfare program called ‘Chile Solidario’ was introduced (later to be replaced by ‘Seguridades y Oportunidades’). This program was designed for very poor households, and beneficiaries receive all State subsidies (housing, water, family allowance, amongst others) plus the assistance of social workers and psychologists, to break from the poverty trap. This group receives a 100% subsidy for the first 15 m3 of water consumption (Article 8, Law No. 19,949).

Finally, in 2014 the program was also expanded to include elderly households in the first two quintiles of income distribution (40% of poorest households) irrespective of whether the standard water bill is above or below 3% of monthly income.

Therefore, there are currently three beneficiary groups for the subsidy. By far the largest group is the first and most of the ensuing discussion refers to the determination of the number and intensity of the subsidy for this group (less than 10% of the number of subsidies are for households in ‘Chile Solidario’ or ‘Seguridades y Oportunidades’).

The number of subsidies and the intensity of each subsidy (percentage discount on the water bill for the first 15 m3 of consumption) is differentiated along two dimensions: first, on a geographical basis since household incomes and water tariffs differ across the country; second, subsidies are also different according to the socioeconomic segment the household belongs to. To explain this in more detail, the number of subsidies is determined considering the average income in each municipality obtained from the Socioeconomic Characterization Survey (CASEN) and information obtained from the private and public pension fund systems (AFP and IPS). This information is compared with the water bill for 15 m3 in each municipality to determine the number of households where water expenditure accounts for 3% or more of household income; thus, for example, a family with a maximum monthly income of Chilean Pesos (CLP$) 800,000 is potentially a beneficiary household in the municipality of Algarrobo in the Valparaíso Region, where water tariffs are high but, by contrast, the same family is not an eligible household in the municipality of Valparaíso in the same region, where the cutoff income level to be eligible for a subsidy reaches CLP$540,000 due to lower water tariffs (MDS, 2012). A special procedure is used to determine the number of subsidies in those municipalities that have more than one tariff group). This process determines the number of subsidies allocated to each municipality which is then aggregated at the regional level (there were 15 administrative regions in Chile until late 2017 when one region was divided in two).

The intensity of the subsidy within each locality is determined by the gap that occurs between 3% of average household income and the water bill. The difference is expressed as a percentage of the water bill for 15 m3. Since 2002, and due to recommendations from the budget management office, the MDS has differentiated the intensity of the subsidy by two broad socioeconomic segments. The first segment are those households with a monthly income lower than two basic food baskets – which is the definition of the poverty line in Chile – and the second are relatively richer households but those whose monthly water bill is still above 3% of income (the basic food basket represents a set of various foods, expressed in sufficient quantities to meet the calorific needs of an average household; in October 2000, it had a value of CLP$19.103 in urban areas and CLP$14.720 in rural areas. If a family has a per capita income lower than the value of a basic basket, the family is considered indigent; if the per capita income is between the value of one and two baskets, the family is considered poor; if the per capita income is greater than the value of two basic food baskets, the family is considered not poor). Table 1 presents an example taken from the municipality of Antofagasta.

Table 1.

Intensity of water subsidies in Antofagasta.

Socioeconomic segmentAverage household income (USD per month)3% of average household income (USD per month)Water bill for 15 m3 (USD per month)Difference of water bill and 3% of income (USD per month)Subsidy (USD per month)Intensity of the subsidy (%)
First segment (poor households) 334 10.5 46.3 35.8 35.8 78%
Second segment 770 23.1 46.3 23.2 23.2 50%
Socioeconomic segmentAverage household income (USD per month)3% of average household income (USD per month)Water bill for 15 m3 (USD per month)Difference of water bill and 3% of income (USD per month)Subsidy (USD per month)Intensity of the subsidy (%)
First segment (poor households) 334 10.5 46.3 35.8 35.8 78%
Second segment 770 23.1 46.3 23.2 23.2 50%

Source:MDS (2012). The first socioeconomic segment are households with monthly income lower than two basic food baskets and the second, relatively richer households but whose monthly water bill is still above 3% of income. The MDS differentiates the intensity of the subsidy according to these broad groups.

The number of subsidies and the value of each subsidy by region are determined annually, and the aggregate projected expenditure on the program is included in the national budget each fiscal year. The subsidy is funded entirely from general tax revenues and the water regulator (responsible for setting tariffs) is not involved in determining subsidy levels or in the operational aspects of the scheme. Thus, there is complete separation between the welfare policies applied in the water sector and the economic regulation of the industry. Subsidies are awarded to a household for a three-year period (although benefits accrue monthly), after which the household must re-apply. Therefore, the number of ongoing subsidies that were distributed in previous years determines a large fraction of the yearly budget for the subsidy program. Once a household is awarded a water subsidy, the service provider is notified and the client's subsequent monthly water bills will be net of the subsidy. Once a month, water companies bill municipalities in their service area to recover the subsidies given to beneficiary households in the last billing period.

Table 2 shows the number of subsidies awarded each year, the aggregate budget for the subsidy, and the average national value of the subsidy per household per month; we show the information for the same years as the CASEN surveys available, since we use the survey data to analyze the distributive properties of the subsidy. Between 1998 and 2015, there was a real increase of 138% in total expenditure, which is explained by a 54% increase in the number of subsidies awarded, particularly starting in 2002 due to the changes made in the subsidy scheme described above; in addition, there was a 55% increase in the average real monthly value of each subsidy.

Table 2.

Effectively expended annual budgets.

YearNumber of subsidies awardedBudget (USD millions)Average subsidy per household per month (USD per month)
1998 527,943 43.6 6.9
2000 536,320 51.9 8.1
2003 686,055 60.4 7.3
2006 748,424 75.8 8.4
2009 789,015 92.5 9.8
2011 820,187 94.2 9.6
2013 878,434 96.6 9.2
2015 811,233 103.8 10.7
% change 2015/1998 54% 138% 55%
YearNumber of subsidies awardedBudget (USD millions)Average subsidy per household per month (USD per month)
1998 527,943 43.6 6.9
2000 536,320 51.9 8.1
2003 686,055 60.4 7.3
2006 748,424 75.8 8.4
2009 789,015 92.5 9.8
2011 820,187 94.2 9.6
2013 878,434 96.6 9.2
2015 811,233 103.8 10.7
% change 2015/1998 54% 138% 55%

Source: MDS (this information was obtained directly from MDS and does not coincide with Table 4 of MDS (2018). Conversations with professionals at MDS revealed that the true figures are those presented above). The nominal budget values were first expressed in real terms (CLP of December 2016) and then converted to USD using the exchange rate for that month (667.17 CLP/1 USD). The values for 1998 do not coincide with those reported by Gómez-Lobo & Contreras (2003) because of differences in the exchange rate used.

It might seem odd that a country whose income per capita more than doubled between 1998 and 2015 would see an increase in the number and value of subsidies for water services. In part, this is due to the changes that have been introduced in the scheme during the last two decades. For example, the expansion of the target population to include beneficiaries of other social programs, such as ‘Chile Solidario’ (with a 100% subsidy for the first 15 m3 of consumption) and the reduction from 5% to 3% of income as the threshold for affordable water expenditure have both increased the number and value of subsidies.

The demographic changes that have occurred during the period may also explain part of the increase, at least in the number of subsidies. Decreasing average household size and population growth increased the number of households by 40.1% between 1998 and 2015. However, while this may explain the growth in the number of subsidies, it might be expected that the amount of each subsidy required would decrease as household size decreases, since average water consumption should be lower for a smaller household size.

The main explanation for the expansion of the budget for the program has been the continuing increase in real water tariffs across the country, due in large part to the need to fund the investments and operational cost of sewage treatment facilities that came into operation during the last decade (Chile now has 100% sewage water treatment after the reforms of the late 1990s; these included the privatization, or concessions to the private sector, of most water companies, with an explicit commitment to invest in sewage treatment plants).

Table 3 presents the real average tariff increase for the main water companies between 1998 and 2015. Average tariffs were measured as the total revenue of the company from water and sewerage operations (financial revenues and other revenues are excluded) divided by the cubic meters of water billed.

Table 3.

Real average increase in operational revenue per cubic meter billed per company, 1998–2015.

OperatorRegionPercentage increase in real operational revenue per m3 billed between 1998 and 2015 (%)
Aguas Andinas S.A. XIII (Metropolitan region) 137%
Essbio S.A. VIII and VI 114%
Esval S.A. 65%
Nuevo Sur S.A. VII 139%
Aguas Araucanía S.A. IX 91%
Essal S.A. X and XIV 142%
Smapa XIII (Metropolitan region) 72%
Aguas del Valle S.A. IV 57%
Aguas de Antofagasta S.A. II 69%
Aguas del Altiplano S.A. XV and I 37%
Aguas Cordillera S.A. XIII (Metropolitan region) 134%
Aguas Chañar S.A. III 131%
Aguas Magallanes S.A. XII 60%
Aguas Décima S.A. 89%
Aguas Patagonia de Aysén S.A. XI 76%
Semcorp Aguas Chacabuco S.A. XIII (Metropolitan region) 92%
Aguas Manquehue S.A. XIII (Metropolitan region) 34%
OperatorRegionPercentage increase in real operational revenue per m3 billed between 1998 and 2015 (%)
Aguas Andinas S.A. XIII (Metropolitan region) 137%
Essbio S.A. VIII and VI 114%
Esval S.A. 65%
Nuevo Sur S.A. VII 139%
Aguas Araucanía S.A. IX 91%
Essal S.A. X and XIV 142%
Smapa XIII (Metropolitan region) 72%
Aguas del Valle S.A. IV 57%
Aguas de Antofagasta S.A. II 69%
Aguas del Altiplano S.A. XV and I 37%
Aguas Cordillera S.A. XIII (Metropolitan region) 134%
Aguas Chañar S.A. III 131%
Aguas Magallanes S.A. XII 60%
Aguas Décima S.A. 89%
Aguas Patagonia de Aysén S.A. XI 76%
Semcorp Aguas Chacabuco S.A. XIII (Metropolitan region) 92%
Aguas Manquehue S.A. XIII (Metropolitan region) 34%

Source: Authors' calculations based on information from www.siss.cl.

As can be seen, real tariffs have increased significantly, in some cases more than doubling during the period. In comparison, the change in the average monthly income of households which received a water subsidy increased by only 38.4% in real terms between 1998 and 2013. Therefore, for most poor families, water tariffs increased faster than income during this period, thus justifying the increase in the number and value of water subsidies; we compared the income increase between 1998 and 2013 because the income data for the 2015 CASEN are not comparable to previous years. The 2015 survey was the first where households' income data were not adjusted to coincide with national accounts information. This does not affect the relative comparisons undertaken below but the absolute income figures are not comparable. In fact, the change between 1998 and 2015 drops to 23.4%. The change in real income does not consider the change in household composition between 1998 and 2013/2015. The average number of household members decreased during the period and, therefore, the increase in real individual incomes was higher than reported here.

To be eligible for a subsidy, a household must apply for the benefit in its municipality. Eligibility is then determined by socioeconomic need, based on the current instrument used to gauge the socioeconomic condition of households. These instruments have evolved over time and are used to target all subsidies in the Chilean welfare system, including the water subsidy.

The targeting of social benefits was introduced in Chile at the end of the 1970s when the selection of beneficiaries for cash transfers and other social benefits began to be made based on a means test-targeting instrument called the Communal Social Assistance Committees form (CAS form), whose first version qualified households on five levels according to information on housing, schooling and occupation of its members. However, it was a precarious instrument in terms of consistency and logistics and was easy to manipulate by users when interviewed by a municipal social worker. In 1987, a second version of the form was introduced, called CAS-2, the design of which was overseen by a group of social and statistical experts. It consisted of a set of 50 variables grouped into five dimensions, based on indicators of socioeconomic need and the application of principal component techniques and discriminating factors. This new form generated a continuous score (which varied approximately between 350 and 750 points) that was used to determine the priority of access to social benefits; a lower score was associated with higher poverty. A household's score was valid for two years before the household had to be reassessed. Information was collected during an interview conducted by municipal social workers at the dwelling of the potential beneficiary and there were well-defined procedures for the collection and processing of information.

As the country developed, dwellings improved and ownership of durable goods (refrigerators, televisions, etc.) soared amongst all income groups, and it became increasingly difficult for the CAS instrument to accurately discriminate wealth among households. In addition, this instrument was subject to manipulation by potential beneficiaries (for example, by taking all durable goods to a neighbour's house prior to an interview; see Herrera et al., 2010).

The socioeconomic changes experienced by the country in the 1990s and the need to include a new target population as part of the Social Protection System led to the introduction in 2007 of a new targeting instrument called the Social Protection Form (FPS). The FPS introduced a socioeconomic vulnerability approach to characterize families. Between 2007 and 2016, this was the instrument used to identify, characterize and stratify the national population to target social benefits given by the different government agencies. The FPS score was calculated based on the income generation capacity of household members, adjusted for the level of economic need. The income generating capacity measured the labour competencies of working-age household members, excluding students, mothers of minor children, disabled people and other groups that have objective difficulties in participating in the labour market. The income generating ability of each person was calculated based on characteristics such as years of schooling, work experience, type of work affiliation and other related variables. Contextual factors that affect the capacity to generate income were also included, such as the level of unemployment and other characteristics of the municipality or region of residence.

The FPS score did not include the possession of physical or financial assets, even when they are informative of a household's permanent income or wealth. This was a decision taken by the MDS since durable goods ownership could still be manipulated – as it was under the CAS system – and since poor households do not generally have substantial financial assets. On the other hand, household needs were quantified based on an index that considered the number of household members, adjusted for demographic characteristics, such as gender and age, and corrected by equivalent scales (being a demographic measure of household members that accounts for economies of scale, since the level of consumption expenditure increases less than proportionally to the number of people in the household). The needs associated with any household member having a physical or mental disability were also considered.

Just as the CAS score was open to manipulation, so was the FPS score. Both instruments were based on information self-reported by households and, as time passed and experience was gained, these households learned how to manipulate the system.

Although not strictly relevant for the time-period analyzed in this paper, it is interesting to note that the Household Social Registry (HSR) replaced the FPS in 2016. The HSR is an information system, managed by the MDS, that records, stores and processes information on the social, civil and socioeconomic attributes of households potentially eligible to be a beneficiary of the Chilean welfare system. To be more precise, the HSR includes information provided by the household (self-reported), which is obtained at the time of completing the entry form to the HSR, and information from administrative records held by the different State institutions, including: Internal Revenue Service (SII), Civil Registry, Unemployment Insurance Administrator (AFC), private Pension Funds (AFP), Social Security Institute (IPS), Superintendence of Health, Ministry of Education, National Health Fund (FONASA), among others. This information is used to construct a Socioeconomic Rating (CSE) for each household, which is made available to different public institutions to determine the eligibility for different social programs. Since this registry contains administrative data from State institutions, it is less susceptible to manipulation compared to purely self-reported information. Thus, we would expect the targeting properties of the water subsidy scheme to improve in the future as the HSR is used to allocate these benefits. An additional benefit is that the HSR significantly reduces the cost of the mean-tested targeting instrument insofar as household interviews are no longer required.

Although the targeting scheme used for the water subsidy seems complex and expensive, it must be borne in mind that it is used to distribute all state transfers, including educational, health and pension benefits amounting to several thousand million dollars each year. Thus, the institutional costs of the system are spread out among many social programs and the incremental costs on the system generated by the relatively modest water subsidy scheme is negligible.

This section evaluates the targeting properties of the subsidy program by analyzing the distribution of the monetary benefits among different households. The larger the proportion of benefits that accrue to lower income households, the better the targeting properties of the scheme. Besides the distributional impact, other complementary issues should be considered when making an overall evaluation of a subsidy scheme. The administrative costs of a program, the relative efficiency of the funding mechanism used, and the possible distortions generated in other markets are also relevant. In this paper, however, the focus is exclusively on the distributional incidence of these subsidies, that is, on how the monetary benefits of the scheme are distributed among different households in the income distribution.

There are two difficulties in estimating the distributional impacts of the Chilean subsidy. First, determining the target population is difficult since the explicit aim of the program is to benefit poor households that would spend more than 3% of their income on water services in absence of the subsidy. Therefore, the target population will be a function of water tariffs as well as household income and will differ across the country. In 2015, in some regions over 35% of households are eligible for the subsidy, whereas in other regions the share is much lower. In 2015, the Metropolitan Region of Santiago had the lowest share of households covered by the subsidy scheme (see Table 4). The Metropolitan Region (RM) accounts for over 40% of the population and is probably the largest tax base in the country. Therefore, since the subsidy is funded from general taxation, there would be an additional progressive impact from the funding side of the scheme1.

Table 4.

Subsidies as a percentage of regional number of households in 2015.

RegionNumber of assigned subsidiesTotal number of householdsShare of regional households covered (%)
I. Tarapacá 26,296 93,573 28.1
II. Antofagasta 39,334 165,373 23.8
III. Atacama 28,267 78,024 36.2
IV. Coquimbo 53,039 214,325 24.7
V. Valparaiso 108,520 565,537 19.2
VI. O'Higgins 44,389 280,440 15.8
VII. Maule 77,013 305,226 25.2
VIII. Bio-Bio 119,395 598,851 19.9
IX. Araucanía 68,115 244,472 27.9
X. Los Lagos 45,590 230,105 19.8
XI. Aysen 11,480 31,017 37.0
XII. Magallanes 13,333 51,574 25.9
XIII. Metropolitan 134,518 2,147,854 6.3
XIV. Los Ríos 24,761 97,333 25.4
XV. Arica y Parinacota 17,183 47,946 35.8
RegionNumber of assigned subsidiesTotal number of householdsShare of regional households covered (%)
I. Tarapacá 26,296 93,573 28.1
II. Antofagasta 39,334 165,373 23.8
III. Atacama 28,267 78,024 36.2
IV. Coquimbo 53,039 214,325 24.7
V. Valparaiso 108,520 565,537 19.2
VI. O'Higgins 44,389 280,440 15.8
VII. Maule 77,013 305,226 25.2
VIII. Bio-Bio 119,395 598,851 19.9
IX. Araucanía 68,115 244,472 27.9
X. Los Lagos 45,590 230,105 19.8
XI. Aysen 11,480 31,017 37.0
XII. Magallanes 13,333 51,574 25.9
XIII. Metropolitan 134,518 2,147,854 6.3
XIV. Los Ríos 24,761 97,333 25.4
XV. Arica y Parinacota 17,183 47,946 35.8

Source: MDS, CASEN 2015.

Because the total number of subsidies for each region is further differentiated among municipalities, a poor household in a municipality with low tariff rates might not receive a subsidy, and a relatively richer household in another municipality with higher water charges could receive a subsidy. This should not constitute a targeting error considering the explicit objectives of the subsidy.

The second difficulty relates to the fact that the survey data used, the Chilean National CASEN for years 1998, 2000, 2003, 2006, 2009, 2011, 2013 and 2015, can identify only a fraction of the total households that receive the subsidy. The differences between the number of households reported by CASEN and from administrative records from the Social Development Ministry are shown in Table 5. If the raw survey data are used, there may be important biases in the estimated distributional impacts. To control for this, the missing subsidies were distributed across households according to the same income distribution of beneficiary households per region as recorded in the survey for each year. To do this, we compared the number of subsidies by region in CASEN with the administrative regional data of subsidies and we amplified the households that reported to receive a subsidy by a regional constant to obtain the administrative number of subsidies. This was the only reasonable option because it assumes that the under-reporting of water subsidies was random. Conversations with professionals linked to the design and application of CASEN uncovered no reason to believe that under-reporting was biased among income groups. However, there was a notable increase in the number of subsidies reported in CASEN for the 2011, 2013 and 2015 waves of the survey, increasing the reliability of the distributional impact calculations for these three years (CASEN 2015 is the only year that does not include an adjustment of incomes according to National Account information. This does not affect our calculations because we present here relative curves for each year).

Table 5.

Number of subsidies reported by CASEN and by administrative data.

1998 527,943 222,851
2000 536,320 216,879
2003 686,055 350,154
2006 748,424 287,674
2009 789,015 269,189
2011 820,187 702,655
2013 878,434 649,851
2015 811,233 711,982
1998 527,943 222,851
2000 536,320 216,879
2003 686,055 350,154
2006 748,424 287,674
2009 789,015 269,189
2011 820,187 702,655
2013 878,434 649,851
2015 811,233 711,982

Source: Social Development Ministry and CASEN, for each year.

According to the raw survey data for 1998 CASEN, only 14.4% of households in the lowest decile of income distribution declared that they received a water subsidy (see Table 6). In the second lowest income decile, only 12.8% received the subsidy. In the 2015 CASEN survey, the raw data improved substantially: in this year, 26.3% of households in the lowest decile said that they received the subsidy compared to 23.0% of households in the second decile. That most households in the first quintile in both survey years do not receive the subsidy implies a very high error of exclusion – households that in principle should be eligible for the subsidy but do not receive it. The results improve somewhat once the data are corrected for under-reporting. All further results are based on the corrected data.

Table 6.

Percentage share of households receiving the water subsidy by deciles of per capita household income in 1998 and 2015.

Decile1998 (%)
2015 (%)
14.4 27.9 26.3 29.3
12.8 25.2 23.0 25.9
10.0 19.5 20.2 22.7
9.7 19.5 19.2 21.9
7.6 16.2 16.7 19.1
7.0 13.8 13.2 15.3
4.0 8.4 10.5 12.3
2.8 5.8 7.0 8.1
1.3 2.2 3.8 4.6
10 0.2 0.4 0.9 1.1
Total 6.6 13.2 13.8 15.7
Decile1998 (%)
2015 (%)
14.4 27.9 26.3 29.3
12.8 25.2 23.0 25.9
10.0 19.5 20.2 22.7
9.7 19.5 19.2 21.9
7.6 16.2 16.7 19.1
7.0 13.8 13.2 15.3
4.0 8.4 10.5 12.3
2.8 5.8 7.0 8.1
1.3 2.2 3.8 4.6
10 0.2 0.4 0.9 1.1
Total 6.6 13.2 13.8 15.7

Source: CASEN for each year and MDS data.

Table 6 also shows that there is a monotonically decreasing share of households that receive the subsidy as the income decile group increases.

Figure 1 presents relative monetary transfer distribution curves showing the percentage of total monetary transfers that accrue to households that are at or below a certain range of income distribution. The horizontal axis measures centiles of per capita income, from poorest to richest, and the vertical axis measures the accumulated percentage of total transfers. The higher and more concave the curve, the better the targeting property of the subsidy.

Fig. 1.

Cumulative monetary transfer curve per centile of per capita income. Source: Authors' computation based on CASEN for each year.

Fig. 1.

Cumulative monetary transfer curve per centile of per capita income. Source: Authors' computation based on CASEN for each year.

Close modal

Figure 1 shows that the relative monetary transfer curves are similar in all years. To examine their differences more closely and to see whether the differences are statistically significant, Table 7 presents the Gini coefficients for each year together with bootstrapped confidence intervals. The Gini coefficient is defined here as the area under the curve and above the 45° line divided by the whole area above the 45° line; if the curve is below the 45° line, then the area is measured as the negative value of the area below the 45° line and the curve. The Gini coefficient will be close to 1 if most monetary benefits accrue to the poorest households and close to −1 if most monetary benefits accrue to the richest households.

Table 7.

Gini coefficient and bootstrapped confidence intervals for cumulative money transfers.

YearGini index95% confidence intervals
1998 0.281 0.278 0.283
2000 0.295 0.293 0.298
2003 0.308 0.305 0.310
2006 0.288 0.286 0.290
2009 0.295 0.293 0.297
2011 0.275 0.273 0.277
2013 0.283 0.282 0.285
2015 0.284 0.283 0.286
YearGini index95% confidence intervals
1998 0.281 0.278 0.283
2000 0.295 0.293 0.298
2003 0.308 0.305 0.310
2006 0.288 0.286 0.290
2009 0.295 0.293 0.297
2011 0.275 0.273 0.277
2013 0.283 0.282 0.285
2015 0.284 0.283 0.286

The Gini coefficient for the relative monetary transfers first increased and then decreased over time. However, not all the differences between years are statistically significant according to the bootstrapped confidence intervals; for example, the difference between the Gini coefficients for 2015 and 1998 is not statistically significant.

The relative concentration curve with respect to the number of beneficiaries is another source of information on the targeting performance of a subsidy scheme (Figure 2). Again, the horizontal axis measures the centiles of the income distribution and the vertical axis is the accumulated percentage of the total number of beneficiaries. For a given point in the income distribution, the graph shows the percentage of beneficiaries that have incomes at or below that point; the higher and more concave the curve, the better the targeting property of the subsidy.

Fig. 2.

Cumulative number of beneficiaries curve per centile of per capita income. Source: Authors' computation based on CASEN for each year.

Fig. 2.

Cumulative number of beneficiaries curve per centile of per capita income. Source: Authors' computation based on CASEN for each year.

Close modal

Once again, Figure 2 shows that the curves are quite close together for the different years. Table 8 presents the Gini coefficient for each year and the bootstrapped confidence intervals. The Gini coefficient has increased over time and then decreased and the confidence intervals imply that differences between first and last years are statistically significant.

Table 8.

Gini coefficient and bootstrapped confidence intervals for cumulative number of beneficiaries.

YearGini index95% confidence intervals
1998 0.271 0.268 0.273
2000 0.294 0.292 0.297
2003 0.302 0.300 0.305
2006 0.287 0.285 0.289
2009 0.293 0.291 0.295
2011 0.275 0.273 0.277
2013 0.284 0.282 0.286
2015 0.285 0.283 0.287
YearGini index95% confidence intervals
1998 0.271 0.268 0.273
2000 0.294 0.292 0.297
2003 0.302 0.300 0.305
2006 0.287 0.285 0.289
2009 0.293 0.291 0.295
2011 0.275 0.273 0.277
2013 0.284 0.282 0.286
2015 0.285 0.283 0.287

What these relative concentration figures do not capture is the extent of any errors of exclusion. The relative concentration curves (Figures 1 and 2) show the relative targeting properties of both schemes, whereas errors of exclusion depend on the scale of each program.

This difference can be seen by analyzing an absolute concentration curve of beneficiaries – the analogue of Shorrocks (1983) absolute Lorenz curves – which is constructed by multiplying the relative concentration curves by the percentage of the population that receives a subsidy (Figure 3). The dotted line in Figure 3 represents perfect targeting. Since the explicit objective of the subsidy is to provide relief for households that spend more than 3% of income on water bills and that this depends both on the level of tariffs in each area as well as household income, there is no target population defined solely in terms of income groups. However, since in 2015 the number of subsidies was equal to 15.7% of households, we assume arbitrarily that this is the perfect targeting benchmark. Therefore, the perfect targeting curve in Figure 3 is a 45° line up to the 15.7th percentile of the income distribution and then flat afterwards. It reflects the fact that perfect targeting in this case implies that only 15.7% of the population are beneficiaries (as reflected in the vertical axis) and that all beneficiaries are concentrated in the 15.7% poorest households. Households above this limit are assumed not to be intended beneficiaries and so the curve is flat afterward. We could have used the number of subsidies as a percentage of households for any year, but 2015 seemed like a good year since the percentage of households receiving a subsidy is higher than in 1998. The percentage chosen is arbitrary and is not entirely correct since it does not consider that some higher income households may receive a subsidy if they reside in a high tariff area, and this should not constitute a targeting error.

Fig. 3.

Absolute concentration curve of number of beneficiaries (1998–2015).

Fig. 3.

Absolute concentration curve of number of beneficiaries (1998–2015).

Close modal

The distance between the perfect targeting curve and the empirical curve at the kink point reflects the errors of exclusion. For example, in Chile the cumulative absolute beneficiary curve (which should include 15.7% of households at the perfect targeting level) included only 5% in 1998, implying that about 10% of households that in principal should receive the subsidy did not. This is even worse in 2015, as less than 5%, rather than the 15.7% poorest households, received the subsidy. This implies that two thirds of poorer households ((15.7–5)/15.7) did not benefit from the subsidy in 1998 and even more in 2015.

Figure 4 presents the absolute concentration curve but considers the years 1998 and 2013. Since more subsidies were given in 2013 than in 2015, this may somewhat improve the errors of exclusion. In fact, it can be seen from Figure 4 that the error of exclusion was lower in 2013 than in 2015. However, this error was still high in 2013 and similar to the level in 1998.

Fig. 4.

Absolute concentration curve of number of beneficiaries (1998–2013).

Fig. 4.

Absolute concentration curve of number of beneficiaries (1998–2013).

Close modal

In summary, we find that the targeting properties of the Chilean subsidy scheme are moderate, with a positive but low Gini coefficient and high errors of exclusion. These errors of exclusion may be worrisome from a policy perspective because they may be indicating that many deserving poor households may not be receiving the subsidy. The flip-coin of this result is that many higher income households receive the subsidy when perhaps they are not poor enough to deserve it.

There are two major questions that come to mind regarding the results of this section. First, why are the distributive properties of the subsidy so low when so much effort is made in the Chilean welfare system to target subsidies to the most needy? Second, why have the targeting properties of the subsidy not improved over time despite the changes to the targeting instrument used to focus social benefits?

To answer the first question, it must be borne in mind that the Chilean subsidy scheme's stated aim is to keep water bills below 3% of household expenditure. Since tariffs differ across the country, some relatively well-off households may receive the subsidy in some localities where water tariffs are high while poorer households in low tariff areas may not. Therefore, measuring the incidence of the subsidy at the national level may not be consistent with the way the subsidy is designed. To explore this issue in more depth, Table 9 presents the Gini coefficient (of the water subsidy) by region for 2015, both for the relative expenditure on the subsidy as well as the number of beneficiaries. The table shows that for some regions the Gini coefficient improves when calculated using regional data rather than national data. This is the case for regions IV, VII, VIII, IX, X and XIV. However, for other regions, the results are worse than at the national level. Thus, this does not seem to be an explanation for the targeting results. Disaggregating the analysis further by municipality will probably not improve the coefficients much since in many regions there are only one or two different tariff areas.

Table 9.

Gini coefficients by region for the relative expenditure and number of beneficiaries of the water subsidy in 2015.

RegionExpenditureBeneficiaries
I. Tarapacá 0.267 0.267
II. Antofagasta 0.222 0.222
III. Atacama 0.245 0.245
IV. Coquimbo 0.306 0.307
V. Valparaiso 0.270 0.271
VI. O'Higgins 0.284 0.285
VII. Maule 0.309 0.310
VIII. Bio-Bio 0.333 0.334
IX. Araucanía 0.350 0.351
X. Los Lagos 0.313 0.314
XI. Aysen 0.241 0.241
XII. Magallanes 0.217 0.217
XIII. Metropolitan region 0.226 0.226
XIV. Los Ríos 0.332 0.333
XV. Arica y Parinacota 0.285 0.286
National average 0.284 0.285
RegionExpenditureBeneficiaries
I. Tarapacá 0.267 0.267
II. Antofagasta 0.222 0.222
III. Atacama 0.245 0.245
IV. Coquimbo 0.306 0.307
V. Valparaiso 0.270 0.271
VI. O'Higgins 0.284 0.285
VII. Maule 0.309 0.310
VIII. Bio-Bio 0.333 0.334
IX. Araucanía 0.350 0.351
X. Los Lagos 0.313 0.314
XI. Aysen 0.241 0.241
XII. Magallanes 0.217 0.217
XIII. Metropolitan region 0.226 0.226
XIV. Los Ríos 0.332 0.333
XV. Arica y Parinacota 0.285 0.286
National average 0.284 0.285

Source: CASEN 2015.

Other possible explanations include the widening of the target population to include elderly households irrespective of whether their water bill is above or below 3% of expenditure and the fact that households that have payment arrears in their water bills are not eligible for the subsidy. However, in the first case, the inclusion of elderly households in the target population was only introduced in 2014 and therefore does not explain the low targeting properties for the other years (many elderly live in larger households but are not the head of the household. Therefore, they may receive the benefit indirectly if the household receives a water subsidy although it would not count as a subsidy for the elderly. Therefore, the number of subsidies for the elderly is a small share of the total number of subsidies). For the second explanation, non-payment of water bills is quite low across the country (less than 1% on average) and therefore can explain at most a marginal impact on the incidence of the subsidy.

More important are probably two other factors. The first is that households need to apply for the subsidy at their municipality. It is unclear whether all deserving households do so. If a significant number of needy households do not apply for the benefit, then the subsidies will go to other less needy households. Operators are legally allowed to help households apply for the subsidy. However, it could be that companies target non-payers to reduce their commercial risks rather than the most-needy households. Second, the fact that the subsidy is given for a three-year period may also affect the targeting properties of the program. The socioeconomic condition of households can change in the interim period in such a way as to de-focalize the benefit. For example, the income of a household may improve in the ensuing years after receiving the subsidy or an undeserving household may become poorer and deserve the subsidy but there are no additional subsidies available in their municipality.

Of course, measurement error in the CASEN survey cannot be ruled out. Some household heads may wrongly state that they receive the subsidy when this is not the case, or vice versa.

Finally, another explanation may be that the instruments used to target subsidies in Chile during the 1998 to 2015 period were imperfect and unable to discriminate very well between deserving and undeserving households.

Clearly, more research is needed to explain the incidence results found in this paper. It will also be interesting to see whether the introduction of the HSR in 2016 has improved the targeting properties of the water subsidy.

This paper has examined the distributional impacts of water subsidy policy in Chile between 1998 and 2015. The Chilean targeting strategy for welfare programs is based on a means-tested welfare scheme. We have examined the targeting properties in a period of significant change in the way poorer households were identified generally and when significant changes were made to the subsidy scheme itself.

Distributive incidence is not the only important aspect of a subsidy program. For example, policy evaluation of the Chilean water subsidy scheme – using a propensity score matching estimator – shows that the subsidy has served to lower the incidence of non-payment, to lower the length of payment delays when non-payment occurs, and to lower the number and duration of service cuts due to non-payment (MDS, 2018). However, it is undeniable that distributive incidence is also an important issue.

Our results indicate that the Chilean water subsidy program is moderately progressive with a Gini coefficient that fluctuated between 0.27 and 0.30 between 1998 and 2015. This is somewhat disappointing since the Chilean welfare system explicitly aims to be well targeted. Another important finding of this paper is that targeting of subsidies has not improved over time. The concentration curves from 20 years ago look much like those in 2015. We discussed possible explanations for these moderate results in the last section. Future research should attempt to shed more light on the possible explanations for our results. Among them, that the instruments used to determine eligibility for subsidies in Chile – mainly self-reported information from households – were flawed. The recent introduction of the HSR, which combines administrative data from all public institutions and is less prone to manipulation, may improve the targeting properties of the Chilean welfare system in the future.

Another result of this paper is that, during a period of strong economic growth and rising real incomes, the water subsidy program did not go away or diminish in importance. Instead, the number of subsidies, the total fiscal expenditure and the average value of the subsidy per household increased. This is mainly explained by the increase in real average water tariffs across the country related to the investments in sewerage treatment plants. Rising tariffs outpaced real income growth, explaining the need to expand the program.

Finally, it is also relevant to note that, in practice, no subsidy will be perfectly targeted. The Chilean scheme and results need to be evaluated relative to the distributional incidence of similar subsidies in other countries. Although somewhat disappointing, the incidence of the Chilean water subsidy is probably better than many water subsidies. Fuente et al. (2016) summarize 21 water subsidy incidence studies across the world. Although the methodologies used in these studies differ across countries and are therefore not directly comparable to our results, 11 of the 14 cases with information were found to be poorly targeted, that is, the distribution of subsidies was worse than if benefits were randomly distributed. This implies regressive impacts (negative Gini coefficient) for these cases.

Banerjee & Morella (2011) and Banerjee et al. (2010) examined targeting properties in African countries. Using survey data, they evaluated the share of the subsidies received by the poor divided by the proportion of the population in poverty. For example, if the poor account for 30% of the population, then a neutral targeting mechanism would allocate 30% of a subsidy to the poor. A value greater (lower) than 1 implies that the subsidy distribution is progressive (regressive), since the share of benefits allocated to the poor is higher (lower) than their share in the total population (for instance, suppose that 30% of the population is poor and obtains 60% of the subsidy benefits; this implies that the poor receive twice as much subsidy as the population on average). Following this strategy, the evidence suggests that utility subsidies tend to be very poorly targeted. Although comparability issues are found among countries, on average the poor benefit only between one-fourth to one-third of what a household randomly selected in the population would receive. The authors concluded that most water subsidy mechanisms are poorly targeted, essentially because many poor households cannot even afford a connection to the piped water network, which can be a significant barrier to expansion for utilities. One policy recommendation therefore, is to subsidize household connections and then move to tariff design. (According to Banerjee & Morella (2011) and Banerjee et al. (2010), Africa remains a predominantly rural continent with a population of approximately 400 million people excluded from any form of safe water supply.)

Similarly, Burger & Jensen (2014) explored the efficacy of an increasing block tariff (IBT) subsidy in South Africa. They showed that even though cross-subsidization is possible with the IBT structure, if water access for poor households is low, a subsidy mechanism through the tariff system does not reach the intended beneficiaries.

Groom et al. (2008) presented evidence for China (Beijing), estimating, in a context of sporadic shortages of water, the welfare effects of alternative water pricing schedules. By using data taken from the Chinese Urban Household Income and Expenditure Survey (HIES), the authors estimated demand for water across different income groups and then performed policy simulations measuring welfare impacts. According to the evidence, the block tariff scheme in place subsidizes all water consumers, high users and low users, rich and poor, for the initial lifeline units of water consumption.

Foster & Araujo (2004) examined the effects of a major program of infrastructure reform in Guatemala. Among other things, the reform massively increased water connections. Thus, households traditionally excluded from basic services (poor, rural and indigenous populations) were twice as likely to be the beneficiaries of the new infrastructure than they had been previously. However, according to the authors, the substantial improvement achieved was not enough to offset their traditional disadvantages. As a result, these groups remained the least likely to receive services. The evidence also showed that the quality of the service was poor, mainly explained by low tariffs; consumers had little confidence in water quality, with three quarters of them either buying bottled water or undertaking some form of self-treatment (boiling drinking water).

Only in the case of Argentina (Foster, 2004) were impacts found to be moderately targeted. Walker et al. (2000) found that in some cities in El Salvador the impact of water tariff structures benefitted the poor but only because tariffs included an overcharge that non-connected poor households did not pay. In the case of Lima, Peru, a rising block subsidy scheme implied low levels of exclusion but very high levels of inclusion (Barde & Lehmann, 2014). Moving towards a means-tested subsidy would reduce the errors of inclusion but would increase the errors of exclusion, something we also found in the case of Chile.

We can conclude from all this that correctly targeting water subsidies are difficult to achieve even in cases where this is an explicit objective of the policy or, as in the case of Chile, of the general welfare system. In this light, our results indicate that the Chilean water subsidy, although moderately targeted in an absolute sense, is probably well targeted when compared to other schemes in the developing world.

This paper was motivated by a proposal from Dale Whittington. We are very grateful for his encouragement and very useful comments and suggestions. We are also thankful to Professor Joe Cook for many insightful comments, to Pieter van der Zaag, Associate Editor, and two anonymous reviewers for helpful comments and suggestions. Any errors and omissions in the paper are the authors' exclusive responsibility. Dante Contreras acknowledges the financial support provided by the Centre for Social Conflict and Cohesion Studies (CONICYT/FONDAP/15130009).

1

We thank Joe Cook for this insight.

Banerjee
S.
&
Morella
E.
, (
2011
).
Africa's Water and Sanitation Infrastructure Access, Affordability, and Alternatives
.
The World Bank
,
Washington, DC
.
Banerjee
S.
,
Foster
V.
,
Ying
Y.
,
Skilling
H.
&
Wodon
Q.
, (
2010
).
Cost Recovery, Equity, and Efficiency in Water Tariffs Evidence From African Utilities
.
Policy Research Working Paper 5384
.
Barde
J. A.
&
Lehmann
P.
, (
2014
).
Distributional effects of water tariff reforms – an empirical study for Lima, Peru
.
Water Resource Economics
6
,
30
57
.
Burger
C.
&
Jansen
A.
, (
2014
).
Increasing block tariff structures as a water subsidy mechanism in South Africa: an exploratory analysis
.
Development Southern Africa
31
(
4
),
553
562
.
Deaton
A.
, (
1997
).
The Analysis of Household Surveys: A Microeconometric Approach to Development Policy
.
The World Bank
,
Washington, DC
.
Foster
V
, . (
2004
).
Toward A Social Policy for Argentina's Infrastructure Sectors: Evaluating the Past and Exploring the Future
.
World Bank Policy Research Working Paper 3422
,
World Bank
,
Washington, DC
.
Foster
V.
&
Araujo
M.
, (
2004
).
Does Infrastructure Reform Work for the Poor? A Case Study From Guatemala
.
World Bank Policy Research Working Paper 3185
.
Fuente
D.
,
Gakii Gatua
J.
,
Ikiara
M.
,
Kabubo-Mariara
J.
,
Mwaura
M.
&
Whittington
D.
, (
2016
).
Water and Sanitation service delivery, pricing, and the poor: an empirical estimate of subsidy incidence in Nairobi, Kenya
.
Water Resources Research
52
,
doi:10.1002/2015WRO18375
.
Gómez-Lobo & Contreras
(
2003
).
Water subsidy policies: a comparison of the Chilean and Colombian schemes
.
World Bank Economic Review
17
(
3
),
391
407
.
Groom
B.
,
Liu
X.
,
Swanson
T.
&
Zhang
S.
, (
2008
).
Resource pricing and poverty alleviation: the case of block tariffs for water in Beijing
. In:
Coping with Water Deficiency
.
Springer
,
Dordrecht
, pp.
213
237
.
Herrera
R.
,
Larrañaga
O.
&
Telias
A.
, (
2010
).
La ficha de protección social, Documento de Trabajo, 2010- 3, Programa de las Naciones Unidas para el Desarrollo (PNUD) (The Social Protection Record, Working Document, 2010- 3, United Nations Development Program), Santiago, Chile
.
Ministry of Social Development (MDS)
(
2012
).
Manual Operativo y Estudio de Focalización del Subsidio al Pago del Consumo de Agua Potable y Servicio de Alcantarillado de Aguas Servidas. Secretaría Ejecutiva de Protección Social, Ministerio de Desarrollo Social (Operational Manual and Targeting Study of Subsidy for the Payment of Drinking Water Consumption and Sewage Service of Sewage. Executive Secretariat of Social Protection, Ministry of Social Development), Santiago, Chile
.
Ministry of Social Development (MDS)
(
2018
).
El Subsidio al Pago del Consumo de Aguas Potable y Servicio de Alcantarillado de Aguas Servidas en el contexto de los objetivos de desarrollo sostenible, Subsecretaría de Servicios Sociales, Departamento de Gestión de Proyectos, Informáticos y Procesos, Ministerio de Desarrollo Social (The Subsidy for the Payment of Drinking Water and Wastewater Sewerage Service in the context of the objectives of sustainable development, Undersecretary of Social Services, Department of Project Management, Information Technology and Processes, Ministry of Social Development), Santiago, Chile.
Shorrocks
A. F.
, (
1983
).
Ranking Income Distributions. Economica, New Series, Vol. 50, No. 197, February, pp. 3-17.
Walker
I.
,
Ordoñez
P.
,
Serrano
P.
&
Halpern
J.
, (
2000
).
Pricing Subsidies and the Poor: Demand for Improved Water Services in Central America
.
World Bank Policy Research Paper 2468
,
World Bank
,
Washington, DC
.