1. INTRODUCTION

One of the Millennium Development Goals (MDGs) was to increase global drinking water coverage from 77% in 1990 to 88.5% in 2015. The state needed to play an active role for this ambition to be realized. On occasion, extensive state intervention was required. The state has honoured its commitment to this by implementing various policies. As pointed out by ^{Gupta et al. (2002}), a country´s ability to achieve the MDGs is determined by its public expenditure policies. In order to expand the provision of water and sanitation in rural areas (^{Watkins, 2006}) and increase public investment, it was necessary to expand the fiscal space for investment in the MDGs (^{Vos et al., 2010}). In the case of Latin America, the question has arisen of to what extent the increase in water and sanitation coverage in recent decades is due to state intervention.

Theoretical formulations, such as those made by ^{Mehta et al. (2005}) and ^{Ginneken et al. (2011}), find that public financing has played a crucial role in paying for coverage increases in developing countries. For Mehta et al. (2005), funds from public financing can be used to develop or rehabilitate infrastructure and finance its operation and maintenance. It can also fund activities such as community capacity building, policy formulation, or supervision of the sector. Ginnekan et al. (2011) highlight the long life of infrastructure, and the need for sufficient maintenance funds to be available. On the other hand, redistribution and market failures require investment and necessitate public intervention.

Understood in this way, it can be seen that Latin American public sectors played a central role in increasing access to water and sanitation, especially in rural populations. It was always unlikely that an increase in coverage would originate from the private sector. Water and sanitation projects require a high initial investment, of both time and money, and offer low rates of return. Therefore, it was always unlikely that the private sector would cover the funding gaps in Latin America. In any case, any private sector participation would always need to be complemented by a strong public sector presence, in order to guarantee its efficiency (^{Zhang et al., 2005}, ^{Estache et al., 2005}, ^{Thoenen, 2007}). Water and sanitation are sectors with high lost costs and imperfect information, characteristics which tend to be unattractive to investors and can result in an undesirable consumption level (^{Ginneken et al., 2011}). Although there are other sources of financing, such as charging fees to users and private sector or community participation, the need for these sectors to be given high priority in public budgets is clear.

Undoubtedly, increased and more efficient public financing is crucial to cover the costs of higher water and sanitation coverage rates in most developing countries (^{Mehta et al., 2005}). The claim that the private sector has participated in the process of extending coverage for rural populations has little basis in either the theoretical literature or the historical record (^{Castro, 2007}). In this vein, ^{Lockwood (2002}) has highlighted the lack of rural water and sanitation provision projects in Latin America and pointed out that costs were generally covered exclusively by users. He has also advocated the need to find approaches that guarantee sustainable public financing.

Authors such as ^{Lockwood (2002}), ^{Schouten and Moriarty (2003}), or more recently ^{Hutchings et al. (2015}), stress that rural water projects, despite being well planned and managed by the communities themselves, require continuous external support, such as financing and technical and management advice, in order to to ensure the quality and success of coverage projects in the long term. In general, public bodies have had to provide this support.

The development and maintenance of public infrastructure is crucial (^{Sorenson et al., 2011}). It is argued that, due to the existence of important economies of scale, average production costs need to be substantially reduced. On the other hand, the high fixed costs and investment requirements must be covered by a natural monopoly. This is because the physical water supply and sanitation network can not be duplicated. If this natural monopoly were privately owned and unregulated, it would charge exorbitant prices and prioritize profit maximization over equity and efficiency. On the contrary, public sector provision would assume its main objective to be universal access and, ultimately, social welfare maximization (^{Swaroop, 1994}, ^{Ginneken et al., 2011}).

However, in most developing countries a large amount of the resources allocated to water supply and sanitation comes from users. This is a heavy burden for rural populations. High connection costs and the fact that physical access to the network is difficult have inhibited the expansion of running water and sanitation into peripheral communities. Under such circumstances, an increase in the percentage of the population with access to water and sanitation is only possible with increased public sector intervention (^{Wolf, 2009}).

Following this line of thought, Ginnekan et al. (2011) reiterate that no country has managed to increase access to drinking water or basic sanitation without a substantial increase in public spending. They provide examples such as the cases of Senegal, Burkina Faso, Niger, Benin and Mali, where large public investment programs were the basis for significant improvements in water supply levels. They also forewarn that, although the positive impact of public spending can be accepted as fact, strong institutions are necessary to ensure efficient use of resources and, therefore, certain precautions must be taken. Factors such as governance or low corruption levels are significant determinants of greater public efficiency in water and sanitation provision (^{Estache and Kouassi, 2002}, ^{Hunter et al., 2009}).

Ginnekan et al. (2011) find no relationship between expenditure levels and access to water or sanitation. However, they attribute this finding to the low quality of the data and to insufficient time allocated to the revision period. Undoubtedly, the usual methodological problems of investigating the social impact of public expenditure, in both small or heterogeneous samples, present themselves. There is also evidence which contradicts the idea that a country’s income level and the composition of public expenditure effect the influence public expenditure has (^{Bose et al., 2007}).

In an attempt to overcome these methodological challenges, several empirical studies have identified variables that explain the differences in countries. The ^{World Bank (1992)} and ^{Shafik (1994}) establish a positive and significant relationship with GDP per capita, While ^{Narayan (1995}) and ^{Prokopy (2005}) conclude that participation of rural beneficiaries is a positive indicator. Years later, ^{Hutchings et al. (2015}) would modify this conclusion by specifying that this participation must occur in communities which posses a clear collective initiative, strong leadership, and a transparent institutional framework.

^{Anand (2006}) finds GDP per capita, growth rate, and public social spending to be determinants. He highlights the legacy variable, which is when countries with low coverage rates appear to be anchored to them. ^{Estache et al. (2006}) find positive evidence regarding GDP per capita, privatization, low corruption and the size of the agricultural sector. Using Swaziland as a case study, ^{Mwendera (2006}), highlights the importance of long-term government financing for bringing existing funding levels up to the required level.

^{Wolf (2009}) finds that there are important differences between rural and urban populations. For example, population density and decentralization have a more decisive positive effect on the rural populations than on the urban populations. In fact, the identification of profound differences between urban and rural provision is not new and in a certain sense appears even obvious. ^{Donaldson (1972}) has already highlighted how rural water programs present a completely different set of challenges from those carried out in urban areas. It is impossible to transfer the techniques used in urban provision to small, low-density populations where financial resources are limited. Community participation, financing, technical assistance, and project scale are just some of the dimensions which differentiate the process of water and sanitation provision for rural populations from urban populations. Therefore, the determinants of their coverage levels also appear to be different.

^{Anillo et al. (2014)} highlight the role of governance, especially for rural areas. These variables are even more important than the quantity of water resources or public infrastructure. Along the same lines, ^{Luh and Bartram (2016)} speculate that the progress that has been made in rural coverage rates is related to changes in government policies and institutions’ capacity to continue to implement them efficiently and sustainably.

In regards to Latin America and using Bolivia as an example, ^{Sara et al. (1996)} stress the need for a coherent set of rules with appropriate incentives in order to ensure that beneficiaries’ targets are met. When comparing groups with similar incomes, ^{Soares et al. (2002)} find that rural populations have far more limited access to drinking water than urban population. ^{Pacheco (2013)} and ^{LaFleur (2014)} identify per capita consumption, the presence of public infrastructures, a female head of household, and a female head of household with a university degree as positive determinants of coverage. Living either in inadequate housing or in rural areas are identified as negative factors. ^{Gómez (2016)} points out that high-income urban populations are more likely to have access to water. This is especially true for populations located in the center of the country, as opposed to rural or peripheral areas. Events such as floods or landslides, or belonging to a specific race or ethnic group also reduce the likelihood of having access to drinking water.

It is in this framework that efforts are made to overcome, as far as possible, the limitations discussed and to empirically prove the effective positive relationship between increases in public social expenditure and a greater percentage of the rural population with access to drinking water and basic sanitation in Latin America. This study covers the period 1994-2014, as the existence of certain circumstances during this period make it a useful period to analyze when trying to validate the hypothesis.

Between 1990 and 2015, Latin America had to reduce the number of people without access to drinking water or sanitation by over 50% (^{Inter-American Development Bank, 2003}). This objective was met and even surpassed, thanks to the advances that were made in rural populations (^{WHO-UNICEF, 2015}). As Figure 1 shows, there was a clear increase in the percentage of the population with access to drinking water during the study period. Rural populations in particular benefited, with their coverage increasing from 62% to 83%, an increase of 21%. If we take into account that the combined rural populations in all the countries studied add up to nearly 113 million people for 2014, then during the study period 24 million people obtained access to water. Paraguay and Chile stand out in the country analysis, showing increases of 58% and 37%, respectively.

Note: SD-Standard Deviation. Source: Prepared by the authors

Figure 1 Population with Access to an Improved Source of Drinking Water (%) of Urban and Rural Population. Simple Mean and Standard Deviation (Latin America-15 Countries), 1994-2014. 

This increase stands in stark contrast to the slight advances in urban populations, where increase averaged just 4% over the study period, with some countries’ coverage rates even decreasing. It is also true that urban areas already had significant coverage rates, 93% on average, when the study period began. An inverse relationship was identified between drinking water coverage and its standard deviation. This trend does not apply to rural populations during the last years of the study period: instead of seeing convergence between their respective water levels, differences between them became wider.

As can be seen in Figure 2, lower basic sanitation coverage rates are lower when compared with drinking water coverage rates. Nevertheless, similar progress was made during the study period. Again, the progress in the rural sector is the most remarkable, increasing 25% from 43% to 68%. This means that 28 million people, all inhabitants of rural areas, gained access to sanitation during the study period. Paraguay, with an increase of 18%, as well as Nicaragua and Honduras, with increases of almost 14% in coverage for their rural populations during the study period, stand out among the countries who made the most progress.

Source: Prepared by the authors

Figure 2 Population with Access to an Improved Source of Drinking Water (% of Urban and Rural Population) Simple Mean and Standard Deviation (Latin America-15 Countries), 1994-2014. 

In terms of sanitation, countries are more heterogeneous and differences in terms of sanitation coverage for rural populations do not decrease, despite the progress in average coverage. This could be evidence of a certain endowment effect, which is when countries seem to be anchored to low rates of sanitation coverage. As ^{Pearce-Oroz (2011)} points out, despite the effort to elevate the profile of sanitation in government policies, the rural sanitation subsector is still one of the most unequal areas of Latin America in terms of enhanced solutions.

Authors such as ^{Vos et al. (2010}) suggest that although these advances were possible for most countries in the region, they would not have happened without the increase in public social spending that took place (^{Lora and Chaparro, 2007}). During the 1990s, the revival of social spending was a deliberate and shared policy designed with a view to fulfill social rights and reduce poverty (^{CEPAL-UNICEF, 2002}).

As can be seen in Figure 3, the "Latin American State" is generally far larger than it was at the beginning of 1994. The usual indicators for the size of the public sector seem to refute the belief that the rollback of the State was a fundamental part of the neoliberal program. While the remodeled political systems had tendencies towards reduction of state intervention, at the same time they also consecrated a wide range of social rights which necessitated an increase in current expenditure, especially social expenditure. Seen in this way, it is evident that general governments have increased current expenditure by an average of 7%. This increase, mainly in social spending, was made in an effort to fulfill neglected social needs and newly recognized social rights (^{Wiesner, 2002}; ^{Moncayo, 2006}; ^{Clements et al., 2007}).

Source: Prepared by the authors

Figure 3 Social Public Expenditure as a Percentage of GDP (Simple Average and Standard Deviation). Latin America (15-Countries), 1994-2014. 

As a whole, the 15-country region gave increased priority to public social expenditure, which increased from 7.93% of GDP for 1994, to 13.81% of GDP for 2014. Venezuela, El Salvador and Honduras stand out, with spending increasing by over 9% of GBP over the study period. It is hoped that the increased social participation of the public sector can be demonstrated as being positively related to a greater percentage of the rural population with access to water and sanitation.

The approach that was implemented established the extent of public participation that was a result of public social expenditure. This expanded any approach that limited public expenditure purely to housing or to the drinking water sector and urban or rural sanitation. For ^{Wolf (2009}), ^{Ginneken et al. (2011}), as well as ^{Manghee & Van den Berg (2012}), analyzing public expenditure destined exclusively for these sectors faces serious limitations. Definition, classification, and coverage may vary from country to country and make international comparison impossible. Additionally, the responsibilities of the sector are very divided, and their actors perform overlapping roles. The data is also usually incomplete or contradictory. This is aggravated by the fact that water and sanitation is not an autonomous public finance sector, something that would make it easier to draw international comparisons.

Generally, neither rural water nor rural sanitation services have specific policies, and are instead grouped together as “rural problems” (^{Pearce-Oroz, 2011}). On the other hand, there are various expenditure items which are not strictly housing, water, or sanitation, yet contribute either directly or indirectly to improved coverage rates. For example, public expenditure aimed at environmental protection contributes to improved water and sanitation coverage rates in water yet, strictly speaking, is not a housing expenditure (^{UDAPE, 2010}). Even international public management can play a decisive role (^{Barnaby, 2009}).

Therefore, the objective of the current paper is to determine the nature of the relation between public social expenditure and drinking water and sanitation coverage rates for a 20-year period (1994-2014), in the 15 countries of Latin America (Argentina, Bolivia, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru and Venezuela). A distinction will be made between rural and urban populations and a regressive analysis will be conducted using both fixed and random general effects models. It is important to rule out the possibility of this relationship being a spurious relationship. It is also important to confirm the presence of unitary roots in the series as well as their cointegration, that is, a long-term relationship between them.

2. METHOD

3. RESULTS AND DISCUSSION

4. CONCLUSION

Although there are several determinants, the importance of public expenditure for achieving coverage objectives in this sector cannot be denied. The existence of a definitive causal relationship can not be taken for granted. Strong institutions which ensure the efficient use of resources are needed.

Within this framework, the relationship between the volume of public social expenditure and the percentage of the population with access to drinking water and sanitation for 15 Latin American countries was investigated empirically over a 20-year period (1994-2014). In order to achieve this, a regressive analysis was conducted on four models which link the percentage of the population with access to improved water and sanitation sources with social public expenditure, as well as other control variables. This analysis was discriminant between urban and rural populations. The presence of unit roots in the series was also confirmed, as was their cointegration, that is, the existence of a highly-consistent long-term relationship.

At first glance, the results yielded highlight the differences between the four models. The contrasts between water and sanitation, as well as the location of the population (urban or rural), are important. This also supports the literature which calls for discriminant analysis. The processes of water and sanitation provision in rural and urban areas are clearly very dissimilar when the two are compared.

Despite the differences between populations, the public social expenditure variable appears to be positively related, significant and highly-consistent in all models. There is important evidence which indicates that the State’s social size is a factor that explains the differences and advances in coverage. The increase in public social expenditure has made it possible for countries in Latin American to reach their coverage goals, mainly in rural populations. Without access to supply or sanitation infrastructures, rural populations often use labor-intensive mechanisms which incur high costs. Therefore, obtaining water and sanitation involves a greater expenditure of money, time, and energy, especially for those without access to a nearby water source and who have not developed cooperation mechanisms for resource management. With income levels constant and tariffs in place designed to guarantee service continuity and efficient use of resources, increases in coverage were only possible with increased social intervention from the public sector.

The fact that rural populations’ water coverage and sanitation increased significantly during the study period is clear. To a large extent, this increase accounts for the overall increase in coverage for the total population. Coverage rates are highest in countries that have strong agricultural sectors. A larger agricultural sector seems to impact positively on resource management and the expansion of water infrastructure. Additionally, this may also explain higher income in rural sectors, a greater concern for the protection and efficient management of water sources, and the construction of better-organized rural sectors with stronger institutions.