Property rights and water access: Evidence from

land titling in rural Peru ∗

Robyn Meeks†

December 2015

Abstract

Insecure land tenure and property rights are an impediment to the construc-

tion of water infrastructure in many developing countries. This paper explores

whether alleviating this impediment through a land titling program in rural Peru

is associated with improvements in water access. The economics literature on the

links between property rights and investment decisions has amassed yet, due to

the unique characteristics of water, it is not obvious how water service provision

would respond to improvements in land tenure. Utilizing the phased-in program

timing in a quasi-experimental design, we exploit the differences in implementa-

tion timing across space in conjunction with the differences in program impact

across households that held property titles prior to the project and those that

did not. Results from this modified difference-in-differences method indicate that

land titling is associated with small in magnitude, but statistically significant in-

creases in water access. As a robustness check, we also perform the analysis using

propensity score matching. Results from propensity score matching methods are

consistent with the findings from the modified difference-in-differences method.

We investigate the channels through which this improved access occurs and find

evidence suggesting it is due to government or water utility investments in infras-

tructure, rather than individual household investment. This is consistent with the

hypothesis that land tenure provides households with increased bargaining power.

∗I am grateful to Erica Field for sharing the PETT survey data and many helpful discussions. Ialso thank Rohini Pande and Bill Clark for tremendous support and guidance. All errors are my own.†School of Natural Resources and the Environment, University of Michigan. Email:

meeks@umich.edu.

1

1 Introduction

Insufficient access to safe drinking water is a problem that continues to plague many of

the world’s poorest populations, with 748 million people lacking access to safe drinking

water and, of those with water access, 1.2 billion are served by contaminated sources

(WWAP, 2015). The development literature pays much attention to the relationship

between poverty and water access and, indeed, this link is a tremendous concern.1

Much less attention is paid to the relationship between water and land rights, yet this

is a problem. The 2015 United Nations World Water Development Report highlights

this relationship, stating, “Access to water is also about access to land.” Indeed, in

many lower and middle income countries, populations living in informal settlements

and slums or on land without formal property rights are generally more likely to have

inadequate access to safe water (WWAP, 2015).

With 889 million people estimated to be living in slums and informal settlements within

the next few years (UN-Habitat 2010), understanding the relationship between land and

water access is crucial. The United Nations identified improvements in property rights

and land tenure security as a mechanism by which to increase water access. This could

happen through several paths. Land tenure improvements could increase the power of

the poor to negotiate improved water provision by the formal sector or it could increase

the poor’s own investments in building water infrastructure (WWAP, 2006).

The role of the state in setting up and protecting property rights is considered to be a

precondition for economic growth (Besley, 1995). Property rights and tenure security

improvements can come in various forms, such as land titling programs (Field and Field,

2006), land reform legislation (Besley and Burgess, 2000), and limited and full transfer

of property rights (Banerjee et al., 2002). Studies have shown that improved property

rights can have numerous economic benefits for households, such as increases in access

to credit (Field and Torero, 2004), household investments (Field, 2005), agricultural

investment (Goldstein and Udry, 2008), access to paid work (Field, 2007), and external

finance (Johnson et al., 2002. The question remains: what is the impact of property

rights on water access?

1For example, an analysis of household survey data from fifteen developing countries, showed thataccess to water services is positively and significantly correlated with households income (Komives etal., 2003).

1

In this paper, I use a quasi-experimental method to provide evidence on the follow-

ing questions: is improved land tenure security associated with improvements in water

access? If improved land tenure leads to differential access to water, what are the chan-

nels through which it occurs? Is the improved water access the result of investment by

individual households or government and/or water utilities?

Several theories have been proposed for the mechanisms through which land titles might

affect investments in agricultural land or housing infrastructure.2 First, land title pro-

vision reduces the risk of expropriation and therefore encourages longer-term immobile

investments. Second, it increases a household’s collateral and therefore their access to

credit. Third, it enables the renting and selling of land and thus increases their gains

from trade. Different impediments to investment occur through each of these paths of

investment in water infrastructure, but many are related to poor tenure security.

To answer these questions, I use a quasi-experiment in rural Peru, through which land

titles were provided to households previously lacking such land security. Historically,

impediments to gaining land tenure included expensive fees, long registration periods,

and other bureaucratic hurdles. Between 1994 and 2004, a nationwide titling program

provided households in rural Peru with improved property rights, through the Special

Rural Cadastre and Land Titling Project (PETT). PETT removed historical impedi-

ments to land tenure, making land titles available for the rural population. Due to the

administration needs of the project, PETT implementation was phased in over time.

The phased-in implementation of land title registration via the PETT program provides

a natural experiment to test the relationship between property rights and water access.

Prior to PETT, only the wealthiest households held title to their land. As a result,

a simple cross-sectional analysis comparing those households with and without titles

would be biased by unobservable factors correlated with both status of households’

property rights and water access. To overcome this identification challenge and investi-

gate the relationship between land tenure and water access, I exploit the differences in

timing of PETT program implementation across space. Specically, I use the variation

between villages in which PETT was (“program” villages) and was not implemented

2Besley (1995) discusses these three such theories in great detail.

2

(“non-program villages”) at the time of our survey, in conjunction with the variation

between households with (“non-squatters”) and without (“squatters”) titles prior to the

PETT project. I refer to this method as a modified difference-in-differences approach.

The intuition underlying this identification approach is that squatters in program areas

should be those most likely to benefit from the PETT program. In contrast, those

already holding formal property titles should see relatively little, if any, impact of the

project being implemented in their region. As a robustness check, I perform a simi-

lar analysis using propensity score matching, which provides the additional benefit of

avoiding the assumption that program effects are linear.

From these analyses, I find that the increases in land title provision are associated

with significant, albeit small in magnitude, improvements in household water access.

Title provision is associated with between five and a twelve percent increase in access

to improved water sources. Tests as to whether these improvements occur as a result

of increased investments in water infrastructure by individual households or through

changes in government or utility provision of water sources, indicate that the latter is

more likely the source of improved access. Specifically, I find increases in monthly water

payments but not in household infrastructure investments. This indicates that at least

some of this improved water access is due to investments either by the government or

utility. These results support the hypothesis that land titling provides poor households

with increased power to negotiate improved water provision, potentially due to the

the government’s or service provider’s greater willingness to serve areas where they are

more likely to recover the marginal cost of delivery.3

Due to the unique characteristics of water infrastructure and, thus the particular im-

pediments to water provision in informal settlements, ex ante it was not obvious that

investments in water infrastructure would perform similarly to other forms of invest-

ments. As such, this study contributes to the literature on property rights and invest-

ments in property by testing whether removing historic barriers to acquiring land titles

leads to improvements in water access.

The remainder of the paper proceeds as follows: Section 2 provides an overview on

3Using data on sources of household funding for water infrastructure investments, I test for evidenceof reduced risk of expropriation and increased access to credit; however, there is an insufficient numberof observations to provide clear support for either theory.

3

the theoretical relationship between property rights and property investments, a back-

ground on the general relationship between property rights and water access, and a

discussion of the unique characteristics of water that set it apart from typical house-

hold investments. Section 3 addresses related legal changes in Peru, including the PETT

program and changes related to water provision around this time. Section 4 describes

the PETT survey, the main dataset, and some initial tests. Section 5 describes the two

main empirical approaches, Section 6 covers results and Section 7 concludes.

2 Property rights and investments

2.1 Potential links: property rights and water infrastructure

In developing countries, informal settlements (clusters of households without land title)

typically face challenges in attaining basic services, including water provision. If the

impediments to government, utility, or household investment in water infrastructure are

indeed the cause of poor provision, then programs providing improved property rights,

land titles, and formal mapping and registration of properties could be associated with

an increase in formal water service provision.4 There are three main routes through

which improved access to water infrastructure in such informal settlements occurs, in-

cluding investments by: (1) the government; (2) utilities or private water companies;

and (3) households investment, through either individual investments or the collective

action of many households. Different impediments to investment occur through each of

these routes, many of which are related to poor tenure security.

Investments by government

In many countries, governments do not legally require the provision of water services

to informal settlements. Often there is concern that delivering basic services, such as

water and sanitation, could provide inhabitants with a legal claim to the land on which

they reside. As such, there can be government resistance to service provision in infor-

mal settlements (UNDESA, 2006). Another barrier to government investment may be

the wak bargaining power held by the inhabitants of the informal settlements. If these

populations lack land title, then they may alos likely lack voting rights, making the

4Such increases in services could impact households through several paths, such as lower volumetricrates for water, shorter distances to communal water standpipes, and less time spent collecting water.

4

extension of infrastructure to informal areas a low priority for politicians.

Investments by utilities

case study of water vendors in Nigeria, showed that households relying on such informal

water provision paid higher volumetric rates for their water than the populations served

by municipal water supply systems. The authors find that the willingness to pay of

such households is sufficient to cover the costs of extending municipal water supply into

these settlements (Whittington et al., 1991). What then is preventing utilities from

expanding? Many people believe the remaining barrier is legal; for utilities to ensure

returns on their investment of expanding the water supply system, they will limit them-

selves to supplying water to households with formal property titles (UNDP, 2006).

The large fixed costs of water infrastructure make utilities hesitant to expand their

services to areas with customers perceived as riskier and less reliable, particularly in-

formal settlements. Such households are considered less apt to pay their water bills or

more likely to relocate without notice. As a result, utilities must develop methods of

coping with non-payment, such as disconnecting non-payers from water supply systems

(UNDESA, 2006). In informal settlements, it can be uncertain who will be responsible

for water payments (in cases where several families or an extended family live in close

proximity), lack of official property boundaries or maps demarcating plots can make

it difficult to lay pipelines, and a lack of an official address system or record of house

locations can make payment collection difficult.

Investments by households

Poor households often face constraints in connecting to that system. To connect to the

water infrastructure typically requires households to make upfront fixed costs. In many

developing countries, water utilities may require an initial payment for the tap, a water

meter, and the pipeline extension from the main system to the individual household.

Households lacking property rights and with weak land tenure might not invest in water

infrastructure for two main reasons. Such costs may be difficult to cover for households

without access to credit. In addition, a risk of expropriation may also deter households

from investing in the fixed cost of infrastructure. Each of these could play a role in

improving access to water infrastructure.

Property rights insecurity (and the risk of expropriation) can act as a random tax on

5

a population (Besley, 1995). The presence of strong property rights, and land rights

in particular, provide individuals with the incentive to use their land efficiently and to

invest in land improvements (Feder et al., 1988). By decreasing the risk of expropria-

tion, titled households may have a greater willingness to invest in water infrastructure,

whether it be through a connection to communal water infrastructure or individual

household water storage or rainwater harvesting tanks.

If the individual’s land is easier to collateralize, that will cause the bank to charge a

lower interest rate and, as a result, individuals will invest more in their land (Besley,

1995). Land titles provide lenders with security to lend to individuals with whom they

are less knowledgeable, thereby increasing the circle of people to which they will lend

and possibly the number of people to whom they will lend. Feder formalizes these argu-

ments in his model, by assuming that credit available to a farmer is “positively related

to the value of their landholdings (as it would typically be when land is a collateral

for loans) and negatively related to their probability of land loss” (Feder et al., 1988).

In many developing countries, residents must purchase the tap, meter, and pipes to

connect to the municipal water supply systems. Households may find the upfront fixed

cost of connecting to be prohibitive. If newly titled households can access credit, they

might use it to connect to the water supply system.

Better property rights may be modeled as decreasing the cost of the exchange when

land is bought or rented. Trading costs are based on the owner’s ability to transfer the

property rights; strong property rights translate into lower costs of trade, whereas a lack

of property transfer rights is similar to infinite costs of trade. A marginal improvement

in the strength of property rights (and the ability to transfer those rights) leads to a

decrease in the cost of property transfer, thereby increasing the potential for the land

to be bought or rented (Besley, 19995). Adding a water connection or building water

infrastructure at a home would arguably increase the value of the household, thereby

increasing the potential gains from trade.

2.2 Unique aspects of water infrastructure

Existing research has empirically tested the links between property rights and invest-

ment in land and housing in numerous contexts, including the Brazilian frontier (Alston

et al., 1996), farms in Thailand (Feder et al., 1988), regions in Ghana (Besley, 1995),

6

and urban slums of Peru (Field, 2005; Field and Torero, 2004). The results of the em-

pirical studies in this literature have been mixed.5 A few features of water provision and

the associated infrastructure differentiate it from these other property improvements

previously researched, possibly complicating this relationship.

First, water can be considered to be a pure public good, a private good, or even a

common good. In places where water is only available through purchase from profit-

seekers, such as water vendors or water companies, then water is a private good and

households unable to pay will be excluded from access. When water remains in lakes

and rivers for recreational enjoyment and consumption by all who choose, then it is a

public good. When access to water is rivalrous but non-excludable, then it is considered

to be a common good. The public goods aspect of water clearly distinguishes it from

other commodities (Hanemann, 2005).

Second, humans need water to survive. This has led to much debate on whether basic

access to water should be considered a fundamental human right, which would require

governments or other parties to provide a minimum amount of the resource to people

when they are lacking access (Gleick, 2000). This issue ties into disagreements as to

how applicable economic concepts are to water. According to the fourth principle of

the Dublin Statement, created in 1992 at the United Nations Conference on Environ-

ment and Development (UNCED), “Water has an economic value in all its competing

uses and should be recognized as an economic good.” However, others in opposition

perceive water to be beyond economic analysis, as it is required by all species and is,

thus, sacred (Barlow and Clarke, 2002; Shiva, 2002).

When water provision is a service of the municipalities, views of water as a fundamental

human right make excluding non-payers from access politically infeasible. Hanemann

(2005) notes that the “special significance” of water exists in both developing and de-

veloped countries, making it politically challenging for publicly owned water utilities

to raise rates even the slightest bit. Certain situations, particularly those in which

exclusion is infeasible, can lead to a free-rider problem. Free riding, which occurs when

some of the users do not pay although they use the water provision services, tends to

5Although the articles mentioned here find some effect of improved land tenure on investment inland and/or property, there are also numerous publications presenting evidence of titles having noeffect on investments. Field (2005) provides a list of such research.

7

cause a commodity to be under-supplied (Hanemann, 2005).

The large fixed costs associated with water supply provision also make it unique from

other commodities. Water is expensive to transport, requiring extensive networks of

pipes and other infrastructure to be constructed prior to water provision. In addi-

tion, there are economies of scale for water provision. The large fixed costs associated

with water infrastructure lend themselves to a natural monopoly and thus a sole water

provider for a given location (Galiani, Gertler, and Schargrodsky, 2005). Or, as noted

by Hanemann (2005), these aspects of water supply lend themselves to public provision

(by a collective or by a monopoly seller) rather than individual, self-provision. House-

hold self-provision, such as is provided through wells, rainwater harvesting, and other

small-scale methods, is often prohibited by high costs (Galiani, Gertler, and Schargrod-

sky, 2005).

Finally, there are positive externalities associated with the provision of safe drinking

water and treatment, as it decreases the spread of contagious water-related diseases and

the spread of water pollution.6 Such positive externalities increase the incentives for

municipalities to ensure safe drinking water provision and invest in water infrastruc-

ture. For all of these reasons water infrastructure investments might not best be made

by individuals, but instead by municipalities or utilities. As a result, we may see no

significant household water infrastructure investments associated with property titles.

Other reasons also exist as to why we might not see evidence of individual household

investments in water infrastructure following titling. First, people might expect the

government or utility to provide such services and that acquiring title makes them

more entitled to formal water provision. Thus, titled households might be less willing

to invest their own time and resources into infrastructure construction. Second, certain

beliefs about water (including water as a human right and religious concepts of water)

might make people less willing to pay for water themselves. Finally, water infrastruc-

ture takes time to construct and thus the time period for which our data covers might

not provide enough time after titling to see investments to occur.

There are reasons to believe that titling might lead to improvements in water access

6Evidence on this is provided in Galiani, Gertler, and Schargrodsky (2005) and Watson (2006).

8

even if households are not investing themselves. The utility may be more willing to

serve households that are now perceived as less risky, as they are less likely to abruptly

move and are easier to bill with legal titles to their land. It is also possible that the

titled households better able to negotiate for coverage by the utility, as they are more

likely to cover the utility’s marginal cost of serving them.

3 Legal changes in Peru

3.1 The PETT Program

In the second half of the twentieth century, land ownership laws in Peru underwent a

series of changes that contributed to the scarcity of property titles in the early 1990s.

During the 1960s, a redistribution of land to peasants in the form of agricultural co-

operatives occurred. This land, received through the nationwide Agrarian Reform, was

prohibited from sale due to legislation changed in the 1970s. This was altered through

Legislative Decree 02, which allowed for the dissolution of cooperatives and the division

of land amongst cooperative members, however, this typically occurred without the re-

ceipt of property titles. In the early 1990s legislation was passed to legalize the sale of

land, but property titles were still relatively rare (Torero and Field, 2005).

In 1991, the Peruvian government passed Legislative Decree 25902, which led to the

onset of the Special Rural Cadastre and Land Titling Project (PETT). Initially this

project was designed to assist a subset of the country’s population;7 however, the titling

process was extended to include all rural estates in 1996 (Torero and Field, 2005). The

goal of PETT was to reduce the cost and duration of registration procedures required

to acquire land title and increase the proportion of titled rural estates. For this to

occur, PETT focused on surveying rural parcels of land and registering new property

titles (Field and Field, 2006).

The project was phased in through the following process. Beginning in 1993, PETT

initially focused on the country’s coastal region, before extending into the highlands

7This includes groups such as the recipients of the Agrarian Reform, owners of uncultivated land,and campesino and native communities.

9

and finally into the jungle in 2000 (Field and Field, 2006). By 2000, 1.9 million parcels

of rural land had been surveyed and 900,000 new property titles had been registered

(Torero and Field, 2005). Since receipt of land title prior to PETT required time-

consuming and costly registration, holding title prior to PETT is highly endogenous

to other household characteristics that also might determine access to water sources.

The aim of PETT was to increase property titles for all rural estates, therefore we

have no reason to believe that acquisition of title via PETT is endogenous to specific

household characteristics correlated with water access. However, the timing of PETT’s

introduction at a village-level might be endogenous. The concerns posed by this source

of endogeneity are discussed in greater detail below.

3.2 Water infrastructure in Peru

Over the past few decades, Peruvian water legislation underwent numerous alterations

and attempted changes, including a transfer of water administration and distribution to

user groups in the 1980s. Between 1991 and 1996, several attempts at privatizing water

services were made (Zegarra, 2004); however, numerous groups mobilized in opposition

to privatization (Isarra and Donner, 2004) and ultimately these privatization attempts

were defeated (Zegarra, 2004). As of the end of our study period in 2004, many of

the country’s water supply systems remained publicly managed.8 As a result, we need

not be concerned of any water policy changes that might simultaneously impact water

infrastructure and affect our estimations.

4 Data

4.1 PETT Survey

The data used for this study were collected through the 2004 Land Titling Special

Project Survey, a rural household survey implemented throughout Peru in the spring

and summer of 2004. The coastal, highlands, and jungle regions are all represented in

the survey sample, as the sample was stratified by the three regions (Field and Field,

8In 2004, the Inter American Development Bank (IADB) arranged a $60 million multiyear loan forthe Peruvian Government to reform priority regions of the country, primarily via privatization reforms(Isarra and Donner, 2004). This change, however, is outside our study period.

10

2006). At the time of the 2004 survey, PETT had been implemented in all three regions

of the country. The survey was developed specifically to be representative of areas in

which PETT was implemented and those in which it was not, in order to use the non-

program areas as a control group.

The survey sample includes 3,204 rural households. Data include household demo-

graphics, basic house characteristics, such as type of house and access to electrical,

water, and sanitation services, household assets and investments, participation in so-

cial programs, family health, household expenses and income, household titles, and

estate security. Data on water access and costs in the survey include: (1) the method

by which water is supplied to the home; (2) the cost of a household’s monthly payment

for water consumption, for the particular month; and (3) whether or not the household

purchased a connection to the rural water supply system, the cost of such a connection,

and whether the purchase occurred before or after the PETT titles was acquired.

4.2 Summary statistics

We calculate summary statistics for all households in out sample.9 According to the

survey data, approximately twenty percent of households in the sample have a PETT

title, whereas thirty-four percent of households in the sample have some sort of title

other than the PETT title. In total, nearly half of the surveyed population has some

form of title to their land. There is substantial heterogeneity in access to basic utility

services: forty-eight percent of the sampled households have access to tap water (ei-

ther indoor or outdoor tap) and an even smaller proportion (forty percent) of sampled

households have access to proper sanitation.

The average household size is 4.5 people with approximately one member of the house-

hold working. Households are highly literate, with eighty-seven percent of household

heads able to read and write. The average age of the household head is fifty-two and

approximately ten percent of household heads are female.

Table 1 shows the differences between PETT-titled households and those with either no

title or those with other forms of property titles. For variables such as amount received

9These are not baseline summary statistics in that they are not calculated using data prior to thePETT program onset.

11

from social programs, number of household members, household savings, and house-

hold assets, we cannot reject the null hypothesis that there is no difference between

households with no title and those with PETT titles. Similarly, I cannot reject the null

hypothesis that there is no difference between the households with a PETT title and

those with a non-PETT title.

Evidence indicates that the groups of households are statistically significantly different

from each other on many characteristics. Households with titles prior to the PETT

program implementation are expected to differ substantially from those without, as the

previous process to obtain title was costly. PETT, however, sought to make land titles

available to all rural households, and therefore it is less clear as to how households

with PETT titles should differ from households with no title (no PETT or other title).

Regarding differences in well-being, the results are mixed. Differences in the size of the

household’s land, the number of workers per household, whether the household head is

female, and the household head’s monthly income are all statistically significant.

4.3 Tenure security

To understand the incentives to participate in the PETT titling program, I analyze

data regarding household members’ perceptions about land tenure. As shown in the

Table 2, there is a statistically significant difference between household members’ per-

ceptions of tenure security depending on the property title types. Households with a

PETT title are more certain than households with no title that their house will not

be expropriated by the government or by another person and that their children will

inherit the house. Households with a non-PETT title feel even more confident than

PETT households that their house will not be expropriated by the government or by

another person. This supports our argument that households with property titles prior

to PETT had no incentive to join. It is not obvious that one type of PETT title is

stronger than another. However, the direction of the statistically significant differences

between PETT households and those without titles supports our assertion that non-

titled households had an incentive to participate in the program.

Of those households that have some property title, either PETT or other, fifty-eight

percent believe that having the title has increased the security level of their house

and seventy-four percent of respondents believe the title increases the probability of

12

protection in the court system. Tenure security is not perceived to impact neighborhood

safety; only six percent of respondents believe that acquiring title has reduced the

level of robberies. Most of these households also do not believe that acquiring title

has increased their access to public services or police protection. Of households with

property titles, one in five believe that acquiring title increased the taxes that they pay.

4.4 Differential access to water sources

Average access to the different water sources for PETT titled, non-PETT titled, and

non-titled households is depicted in the Table 3. Using the responses to the survey

question regarding the household’s water source, I create binary indicators for different

levels of water access. First, I create binary variables for each of the main types of water

sources: (1) indoor tap; (2) outdoor tap; (3) sink, (4) well; (5) river; (6) vendor; (7)

other. In addition, I created a combined indicator variable for all tapped water sources,

both indoor and outdoor, in addition to a variable to indicate access to an “improved

source” as defined by the development community’s standard for water projects.10

There is a statistically significant difference in access to all types of water, except for

outdoor taps and other sources, between (1) households with PETT titles versus house-

holds with no title, and (2) households with PETT titles versus households with other

types of titles. These results suggest that there is a relationship between land titles

and access to water sources, thus motivating this research as to whether acquiring land

title leads to improved household water access. Also, the lack of difference in access to

outdoor taps between the titled and non-titled households indicates that the presence of

communal (outdoor) taps may be better indicated by community characteristics rather

than household characteristics. In other words, it may be community action, or action

on the part of the municipality, that determines the presence of outdoor taps.

These results are supplemented with probit regressions and marginal effects regressions,

which test the impact of the PETT title on access to all types of water sources listed

above. The results from these regressions are shown in Table 4, panels A and B respec-

10According the HDR 2006, “improved” access is definition based on three dimensions: water quality,proximity, and quantity. The threshold for acceptable water source proximity is less than one kilometerand that for acceptable quantity is at least 20 liters per day. The definition of “improved” includesin-house connections, shared standpipes, pumps, and protected springs and wells; water provided byvendors, tanker trucks, streams, or unprotected wells and spring is not (HDR 2006).

13

tively.

The preferred specification (i.e. controlling for the abovementioned household char-

acteristics and the existence of an alternative type of title) provides evidence of a

significant relationship between acquisition of a PETT title and improvements in water

access. A household with a PETT title is five percent more likely to have access to an

indoor tap (statistically significant at the 95% level), four percent more likely to have

access to an outdoor tap, nine percent more likely to have access to any type of tap

water, and eleven percent more likely to have access to an improved water source (all

of which are statistically significant at the 99% level) than if they did not have such a

title. These results are preliminary evidence of a relationship between PETT titles and

water access.

5 Empirical Approach

I treat the PETT program as a quasi-experimental program, motivated by the fact that

PETT aimed to survey and register land for all rural landholders, regardless of wealth,

education, and other such household characteristics. As such, the program reduced the

endogeneity of land titles that had existed previously. The two overarching research

questions are explained below.

Is land titling via PETT associated with differences in access to water?

To answer this question, I investigate a subset of questions including: Is there a differ-

ence in overall access to water due to titling differences? Are there changes in the types

of water provision?

Through which pathways might differences in water access occur?

It is possible that provision of land titles could lead to greater investment in water

infrastructure, by the government, water utilities, or individual households. Changes in

certain types of water infrastructure, such as communal outside taps, might signify in-

vestments by the municipality and/or utility, whereas other investments, such as indoor

taps, would most likely indicate a private household investment. I look for evidence

as to whether infrastructure improvements occurred via individual households or mu-

nicipality/utility investments. I test whether titling is associated with an increase in

14

household investments in water infrastructure and/or larger monthly water bills. With-

out access to records of land sales transactions or property value estimates, I cannot

test for evidence of the gains-from-trade theory in this context.

I investigate the following questions: Is there a difference as to whether households

invest in a water connection? Is there a difference in the amount that households in-

vest in the water infrastructure? Results could indicate greater individual household

investments, which would be in line with the theory that titling leads to investments in

property. It would, however, identify whether investment occurred due to reduced risk

of expropriation or increased access to credit.

Finally, I investigate the possibility of water infrastructure investments by municipali-

ties or utilities. If a utility invested in new water pipelines or wells, one might expect

that to be reflected in the households’ monthly cost of water. We use responses to

the following questions as evidence as to whether utilities and/or municipalities make

investments. Is there a difference in whether or not a household makes monthly water

payments that is associated with title provision? Is there a difference in how much a

household spends in monthly water payments that is associated with title provision?

5.1 Difference-in-differences estimates

To determine if there is an effect of the PETT titling program on improved water

access, I use a modified difference-in-differences (DD) approach. A modified DD, com-

paring differential impacts across regions in which a policy is implemented with regions

in which the policy has not yet reached is similar to the methods used in Morduch

(1998) and Field (2005).11 I exploit the differences in the variation between villages

that PETT had reached (“program villages”) and those it had not yet reached (“non-

program villages”), in addition to the variation between the households that had no

titles prior to PETT (“no prior title”) and those that did (“prior title”). As the project

was implemented earlier within the coastal region than the other two regions, some

11Morduch (1998) assesses the impact of micro-credit provided by the Grameen Bank. To investigatethe effect of land titles on investment in urban slums in Peru, Field (2005) used a similar modified DDapproach in which a control group comprised of future program beneficiaries residing in neighborhoodsnot yet reached by the program.

15

program impacts may differ based on location. To control for this, I include regions

fixed effects. In addition, to account for the differential impacts of PETT over time, I

control for PETT title implementation year.

To measure the impacts of the PETT titling program on water access using this modified

difference-in-differences approach, I follow estimate the following equation:

qiv = τProgramv ∗NoPrioriv + βProgramv + δNoPrioriv + αXiv + γr + εiv (1)

where qiv is the water source of household i in village v, Program is an indicator equal

to 1 if the PETT program had reached village v at the time of the survey, NoPrioriv is

an indicator of whether household i in village v had title prior to the onset of the PETT

program, Xiv is a vector of household-level control variables, and γr are region fixed

effects. The coefficient on our interaction term Program ∗NoPrioriv is our estimated

treatment effect.

A standard difference-in-differences approach would require that the trends in water ac-

cess amongst “program” households are parallel to the trends in water access amongst

the “non-program” group prior over time. Our modified difference-in-differences method

requires a slightly altered parallel trends assumption: the households titled prior to

PETT and those not titled prior to PETT need not be the same; they just must not

systematically differ across early and late program villages. With only cross-sectional

data, I am unable to test the parallel trends assumption using pre-intervention peri-

ods. If the relationship between the titling program and water access is confounded by

wealth variables, the difference-in-differences estimate would be biased.12 Given these

potential biases of our modified difference-in-differences method, we test these same

questions using propensity score matching instead as well.

12For example, if richer villages, which would be more likely to have better water access regardless oftitles, were more likely to be program villages, then the difference-in-differences estimate would likelyoverstate the impact of the titling program on water access and the bias would be positive. Conversely,if poorer villages, which would be likely to have worse access to water services, were more likely to beprogram villages, then the difference-in-differences estimate would be downward biased and would belikely to understate the effect of the program.

16

5.2 Propensity score matching estimates

Propensity score matching methods have been used in several studies investigating the

impact of water access on child health (Jalan and Ravallion, 2003; Galdo and Briceno,

2005), an area of research in which randomizing access to treatment is often difficult due

to cost constraints, political feasibility, and implementation issues. Similarly, random-

izing access to the PETT titles would have been politically challenging to implement.

The propensity score is the probability of being assigned to the PETT program, condi-

tional on a vector of observed variables. Using propensity score matching to estimate

the causal effects of the PETT program on water access requires a two-step procedure.

First, we estimate the propensity score, which is the probability of participating in the

PETT program, based on the selected covariates. Second, we match households with

PETT titles to households without titles based on the scalar propensity score.

Using a propensity score to match, instead of matching on a vector of observables,

is particularly advantageous when there are many observables or when the indepen-

dent variables are continuous rather than binary (Angrist and Krueger, 1999). Also,

modifying for the propensity score balances the distribution of the observed covariates

between the treatment and control groups, based on their predicted probability of par-

ticipation, and removes bias due to the unobserved covariates (Rosenbaum and Rubin,

1983). Propensity score matching addresses one of the primary weaknesses of conven-

tional matching, the assumption that selection of the PETT and non-PETT households

does not occur on unobservables. Thus, instead of matching on all of the observable

characteristics, as in conventional matching, I match here on the scalar propensity score.

In our case, propensity score matching is preferred to conventional matching, for sev-

eral reasons.13 First, there are many observables on which to match, making PSM a

better fit. Second, there are possibly some unobservable factors that contributed to

the selection of the PETT and non-PETT households. I compare the PETT and non-

PETT households to estimate both the average treatment effect (ATE) and average

treatment on the treated (ATT), using nearest matching with replacement. I perform

these calculations for one, three, and six matches. PSM provides the additional benefit

of allowing the relationship between land titles and water access to be non-linear.14

13As a robustness check, we also performed conventional matching methods and found that resultswere not qualitatively different.

14Akin to the issues that Galiani, Gertler and Schargrodsky (2005) faced in their analysis of the

17

6 Results and Analysis

6.1 Land titling and differences in water access

The modified difference-in-differences results are shown in Table 5. The coefficient on

the interaction term for program and squatter gives us the effect of the PETT titling

program on access to water access. Results for two types of water sources are of partic-

ular interest: river water and improved water sources. When controlling for household

characteristics, whether the household had another title, and region, the results indi-

cate that access to titles leads to a four percent decrease in river water use (significant

at the 90% level) and six percent increase in improved source use (significant at the

95% level). However, when we include controls for year of titling, only improved water

sources remain significant, albeit the size of the effect is nearly the same.

Results from propensity score matching, which are shown in Table 6, serve as a check

for the difference-in-differences results. Matching results show a greater number of

significant results, all of which are with are in the expected direction. Overall, these

results provide similar estimates of the relationship between PETT titles and outdoor

tap water. The small increase associated with PETT could be a signal of communal

water supply systems and therefore a sign of community investment through either

collective action or, more likely, investment by the municipality or utility. An increase

in access to water through outdoor taps does not necessarily signify construction of an

altogether new water supply system, but could indicate the extension of an old system

to reach areas not previously connected.

Importantly, river water use shows a decline between nine and twelve percent and im-

proved water sources show an increase of between eight and twelve percent (all at the

impacts of privatization on child health, it is quite plausible that impact of titling are heterogeneousacross villages. For example, some villages may have stronger leadership or institutions and alreadyhave a high-level of water provision for its residents. In such a village, the PETT program may haveless of an impact on water access because the strong leadership has ensured that the people have ahigh level of water provision already. In contrast, a village that is without a strong organizer maybenefit more from the PETT program as it has much more to gain with respect to water provision.

18

99% significance level) due to titling. River water and improved water source use are

particularly important indicators of changes in water access. Increases in access to

improved water sources indicate steps towards achieving the water-related MDG. Ar-

guably the most interesting result in this section is for the binary variable indicating

use of rivers for water sources. This is perhaps the strongest indication that there are

improvements made in water access. Untreated river water tends to be the source of

last resort, when there is no access to formal water services in the area or if present

they are unaffordable to a particular household. The fact that people would be shifting

away from river water is a good indication that some other better source is replacing

river water for them, either due to improved reach of a formal water supply system or

lower relative prices for water provided by an existing one.

In addition, there is a slight decrease (albeit not always statistically significant) in ac-

cess to water via vendors. Water vending, as discussed previously, is typically more

expensive than access through municipal water systems. It would therefore be logical

that households would switch away from water vendors if presented with another al-

ternative; however, these results are not clear enough to make a claim in this regard.

Results for the variable encompassing other water sources are positive and statistically

significant, but it is difficult to make any greater assessments from that result, as “other

sources” can represent a wide variety of service provision, ranging from rainwater har-

vesting to use of polluted irrigation canals for drinking water.

Improved access to water sources was a designated outcome measures to meet goal

seven of the Millennium Development Goals (MDGs). It appears as though land titling

is indeed associated with the overall improvements in access to water that the MDGs

seek to cause. And, second, land titling could provide one possible method of working

towards improving water access. Thus, it appears as though land tenure insecurity is a

constraint to household access to improved water sources.

I test whether this difference in water access is likely due to the wealth effect of the

title or due to a substitution effect, by repeating the abovementioned regressions with

binary variables indicating ownership of certain household “luxury” goods, such as a

television and radio. The ownership of these items is significantly greater in PETT

households. Therefore we consider the changes in water access to be a wealth effect.

19

6.2 Pathways: land security influences water access

Given the evidence that improvements in water access occurred, both through an in-

crease in use of formal water provision via improved water sources and through the

decrease in unsafe water sources in the form of river water, I investigate whether in-

vestments in infrastructure might have been made by individual households or the

municipalities or utilities.

Evidence of improved water services

To investigate whether households differ with respect to investments in connections to

water services, I compare the dates on which the household received title with the year

in which the water supply connection was made. If the connections tend to occur after

or around the time that a household received title, then this reinforces the argument

that reducing the risk of expropriation will lead a household to increase its investment

in the land and household infrastructure.

I use difference-in-differences and matching estimates to see if the PETT title is associ-

ated with an increase in the probability of a household investing in water infrastructure

as well as the amount invested in water infrastructure, controlling for the region and

other household characteristics. In Table 7, difference-in-differences estimates show

that title provision is associated with a three percent increase in investments in wa-

ter infrastructure (statistically significant at the 10% level). However, when I include

dummies for year of titling this result is no longer statistically significant. Matching

results, as shown in Table 8, do not show much effect of titling on investments in water

infrastructure.

These results do not strongly support the development theory that land titles would

lead to increased investment in the household. However, given the unique aspects of

water this result is not surprising. These findings provide evidence that the PETT

title is leading to community improvements in water infrastructure, perhaps by the

municipality or utility, rather than through many individual investments in household

infrastructure. Together these results all support the hypothesis that although title

does lead to improved access to water sources, these improvements are not necessarily

occurring due to the private household investments.

20

Evidence of utility or government improvements to water services

Thus far, results indicate that titling is associated with improved water access, but

there is only weak evidence that households themselves increase their investments in

water infrastructure. What then is driving the improvements in water services?

Titling might enable investment through channels other than individual households,

such as utilities or governments. The data do not include such information, so I cannot

directly test for this. I can, however, test to see if there are differences in monthly water

bill payments, which might indicate improvements by a water utility or government.

The DD results in Table 9 show little effect of PETT titles on the probability that peo-

ple make monthly water payments. The propensity score matching results, shown in

Table 10, show impacts of a much greater scale than the DD results and are considered

an upper bound.

Although we have evidence that water access is improving and that it is occurring as

a result of municipality or utility investment, such improvements may translate into

a slightly greater proportion of households having to pay for their water and a larger

payment per month. The most likely explanation for this result is that whatever the

prior and current water sources, they probably both incur a fee and the newer one is

just slightly more expensive. This is the likely result if people are transferring away

from rivers, from which people tend to collect water for free, as one of their primary

water sources to a more formal water source, for which they have to pay for their water.

If this is the case, then it would be quite plausible that we would see an increase in

the proportion of households making monthly water payments associated with PETT

titles.

6.3 Credit-access or risk of expropriation hypotheses

Although we cannot test the gains-from-trade model here, we seek evidence of either

improved credit access or decreased risk of expropriation by investigating whether title

acquisition impacts the method of payment used (such as, formal or informal credit or

own household finances) for investments in water infrastructure. To test for any differ-

ences in access to credit, I look to survey responses regarding financing for household

connections to the public water supply. Respondents may reply to any of six types of

financing: (1) banking credit, (2) another type of formal credit, (3) informal credit, (4)

21

loan from family or friend, (5) own resources, (6) in exchange for a debt or service.

Unfortunately, the number of responses is too few for all categories of financing, except

for respondents replying “own resources.” Therefore, I am not able to test the access

to credit hypothesis with this data. However, I do test for impacts of PETT title on

whether a household uses their own finances to pay for connections to public water

supply systems. If I were to see a positive change in the use of own resources for water

infrastructure improvements associated with PETT titles, this would provide evidence

in support for the risk of expropriation hypothesis. It would signify that households

indeed had the ability to make the investments in water infrastructure but do not do

so until they have greater property security, as provided by the PETT title.

The difference-in-differences results, as shown in Table 11, do not show much of a

relationship between the titling program and using one’s own resources. These results

do not seem to provide evidence for or against either the credit access or security of

tenure models for investment.

7 Conclusions

The results presented within this paper support the hypothesis that land title provision

is associated with better water access, specifically increases in access to improved water

sources. One of the strongest indications of overall improvements in water access is the

significant decrease in the use of river water as a household source. The results appear

to indicate that at least some portion of this improved infrastructure is actually due

to public investments or investments by the utility in improving water infrastructure,

rather than investments by the individual households. We see significant positive effects

of titling on improved water sources, such as outdoor taps, which tend to be the result

of community or public financing. In comparison, we see no effect on the improved

sources that tend to be private and associated with one particular household, such as

indoor taps or wells.

The relationship found here between land title and water access could also be due to a

shift in the households’ attitudes towards access to services, particularly water. Newly

titled households might expect public investment in water infrastructure and formal

22

water service provision now that they have proper legal title to their land. This relates

to our discussions as to how investments in water infrastructure differ from other results

that find an effect of land titles on individual household investments in land and hous-

ing. Conversely, the positive relationship between PETT land titles and water access is

consistent with the hypothesis that newly-titled populations gain increased negotiating

powers, particularly with respect to demands for water services.

Although there are significant improvements in access to particular types of water

sources, the improvements measured by the modified difference-in-differences method

are small. We view these results as the conservative estimates, whereas the propensity

score matching results are perhaps an upper bound. It possible, however, that esti-

mates of the relationship between PETT land titles and water access are small due to

the amount of time between the PETT project implementation and the survey, which

is relatively short in comparison to the time it takes to plan and construct water in-

frastructure. Therefore, the magnitude of this relationship could potentially be larger

after more time has passed after the titling program.

References

[1] Alchian, A. and H. Demsetz. 1973. “The Property Right Paradigm.” The Journal

of Economic History, Vol. 33, No. 1.

[2] Alston, L., G.D. Libecap, and R. Schneider. 1996. “The Determinants and Im-

pact of Property Rights: Land Titles on the Brazilian Frontier.” Journal of Law,

Economics, and Organization, Vol. 12, No. 1.

[3] Barlow, M. and T. Clarke. 2002. Blue Gold: The fight to stop corporate theft of

the world’s water. The New Press.

[4] Banerjee, A., P. Gertler and M. Ghatak. 2002. “Empowerment and Efficiency:

Tenancy Reform in West Bengal,” Journal of Political Economy, 110(2), 239-280.

[5] Besley, T. 1995. “Property Rights and Investment Incentives: Theory and Evidence

from Ghana.” Journal of Political Economy, vol 103, no.5.

23

[6] Besley, T. & R. Burgess. 2000. “Land Reform, Poverty Reduction, And Growth:

Evidence From India,” Quarterly Journal of Economics, 115(2), 389-430.

[7] Field, A.J. and E. Field. 2006. “Globalization, Crop Choice and Property Rights

in Rural Peru.” WIDER Research Paper No. 2007/72.

[8] Demsetz, H. 1967. “Toward a Theory of Property Rights.” American Economic

Review, Vol. 57, No. 2.

[9] Feder, G., T. Onchan, Y. Chalamwong and C. Hongladaron. 1988. Policies and

Farm Productivity in Thailand. (Baltimore: Johns Hopkins University Press).

[10] Field, E. 2003. “Fertility Responses to Land Titling: The Roles of Ownership

Security and the Distribution of Household Assets.” Working Paper.

[11] Field, E. 2005. “Property Rights and Investment in Urban Slums.” Journal of the

European Economic Association Papers and Proceedings, April-May 2005, 3(2-3).

[12] Field, E. 2007. “Entitled to Work: Urban Property Rights and Labor Supply in

Peru.” Quarterly Journal of Economics, Vol. 122, No. 4.

[13] Field, E. and M. Torero. 2004. “Do Property Titles Increase Credit Access Among

the Urban Poor? Evidence from a Nationwide Titling Program.” Working paper.

[14] Galdo, V. and B. Briceno. 2005. “Evaluating the Impact on Child Mortality of a

Water Supply and Sewerage Expansion in Quito: Is Water Enough?” IADB Office

of Evaluation and Oversight Working Paper.

[15] Galiani, S., P. Gertler, and E. Schargrodsky. 2005. “Water for Life: The Impact of

Privatization of Water Services on Child Mortality,” Journal of Political Economy

113(1): 83-119.

[16] Gleick, P.H. 2000. The Human Right to Water. The World Water 2000-2001: The

Biennial Report on Freshwater Resources. (Washington, D.C.: Island Press).

[17] Goldstein, M. and C. Udry. 2008. “The Profits of Power: Land Rights and Agri-

cultural Investment in Ghana.” Journal of Political Economy, vol. 116, no. 6.

[18] Hanemannm, W.M. 2005. “The economic conception of water.” In Water Crisis:

Myth or Reality. Editors Peter P. Rogers & M. Ramon Llamas. A.A. Balkema

Publishers, The Netherlands. Pgs 61 - 91.

24

[19] Isarra, L. and L. Donner. 2004. “The Fight for Public Water in Peru.” Social

Policy. Winter 2004.

[20] Johnson, S., J. McMillan, and C. Woodruff. 2002. “Property Rights and Finance.”

American Economic Review, 92(5): 1335.

[21] Shiva, V. 2002. Water wars: privatization, pollution, and profit. (Cambridge, MA:

South End Press).

[22] Torero, M. and E. Field. 2005. Impact of Land Titles over Rural Households.

Working Paper OVE/WP-07/August 1, 2005. Office of Evaluation and Oversight

(OVE), Inter-American Development Bank, Washington, D.C.

[23] UN Millennium Project. 2005. “Health, Dignity, and Development: What Will it

Take?” Task Force on Water and Sanitation.

[24] WWAP (United Nations World Water Assessment Programme). 2006. The United

Nations World Water Development Report 2 (WWDR) 2006: Water: A Shared

Responsibility. Paris/New York, UNESCO/Berghahn Books.

[25] WWAP (United Nations World Water Assessment Programme). 2015. The United

Nations World Water Development Report (WWDR) 2015: Water for a Sustain-

able World. Paris, UNESCO.

[26] United Nations Development Programme. 2006. “Beyond Scarcity: Power, Poverty,

and the Global Water Crisis.” 2006 Human Development Report.

[27] Watson, T. 2006. ”Public Health Investment and the Infant Mortality Gap: Evi-

dence from Federal Sanitation Interventions on U.S. Indian Reservations.” Journal

of Public Economics, Vol. 90, Issues 8-9.

[28] Whittington, D., D.T. Lauria, and X Mu. 1991. “A Study of Water Vending and

Willingness to Pay for Water in Onitsha, Nigeria.” World Development, Vol. 19,

No. 2/3.

[29] Zegarra, E. 2004. “The Market and Water Management Reform in Peru.” CEPAL

Review, 83.

25

Tab

le1:

Mea

ns

ofE

xpla

nat

ory

Var

iable

s

Var

iabl

ePE

TT

No

Titl

eO

ther

TIT

LE

No

Titl

e O

ther

Titl

e A

mou

nt re

ceiv

ed fr

om so

cial

pro

gram

s 61

8.57

551

6.30

961

10.

825

0.98

9Si

ze o

f HH

land

(sq

met

ers)

1276

465

1804

3618

4717

0.01

70.

004

Wor

king

HH

mem

bers

0.98

10.

959

0.96

80.

003

0.01

HH

mem

bers

4.5

4.60

84.

537

0.27

30.

718

Chi

ldre

n ag

ed 5

-11

0.60

80.

918

0.76

20

0.00

1C

hild

ren

aged

12-

160.

577

0.62

30.

638

0.21

90.

125

Fem

ale

HH

hea

d 0.

168

0.11

80.

113

0.00

30.

001

HH

hea

d ag

e56

.148

.952

.91

0H

H h

ead

liter

ate

0.86

0.87

60.

871

0.31

80.

506

HH

hea

d ye

ars o

f edu

catio

n 5.

202

5.74

15.

627

0.00

40.

035

HH

hea

d in

com

e 46

8.9

372.

845

8.6

0.01

0.80

2H

H sa

ving

s 0.

341

0.32

20.

374

0.38

30.

173

Ow

n ho

use

0.63

0.21

20.

658

00.

233

Val

ue H

H a

sset

s67

50.1

9757

.572

30.5

10.

892

Not

es: M

eans

cal

cula

ted

usin

g da

ta fr

om th

e 20

04 L

and

Titli

ng S

peci

al P

roje

ct S

urve

y.

Mea

ns o

f HH

Cha

ract

eris

tics (

By

Titl

e T

ype)

T-T

EST

S w

ith B

ase

Cas

e of

"P

ET

T T

itle"

26

Tab

le2:

Per

cepti

ons

ofE

xpro

pri

atio

nR

isk

T-T

EST

: T

-TT

EST

:

Surv

ey Q

uest

ion

SQue

stio

n P

ET

T T

itle

Oth

er T

itle

No

Titl

ePE

TT

vs.

NO

TIT

LE

T

ITL

E v

s. PE

T

How

cer

tain

are

you

that

you

r hou

se w

ill

not b

e ex

prop

riate

d by

the

gove

rnm

ent?

85.0

494

.14

73.7

20.

000

0.00

0

How

sure

are

you

that

you

r hou

se w

ill

not b

e ex

prop

riate

d by

ano

ther

per

son?

88.5

796

.45

76.8

90.

000

0.00

0

How

sure

are

you

that

you

r hou

se w

ill

be in

herit

ed b

y yo

ur c

hild

ren?

96.5

988

.48

76.2

0.00

00.

000

Not

es: M

eans

cal

cula

ted

usin

g da

ta fr

om th

e 20

04 L

and

Titli

ng S

peci

al P

roje

ct S

urve

y.

Perc

ent h

ouse

hold

s rep

lies:

"cer

tain

" or

"ve

ry c

erta

in"

27

Tab

le3:

Pro

por

tion

ofH

ouse

hol

ds

wit

hA

cces

sto

Wat

erSou

rces

by

Diff

eren

tT

itle

s

Var

iab

leP

ET

TN

o T

itle

TIT

LE

No

Tit

le

Oth

er T

itle

tap

indo

or0.

210.

426

0.36

40.

000

0.00

0ta

p ou

tdoo

r0.

305

0.34

50.

351

0.51

20.

455

Fou

ntai

n0.

204

0.40

60.

389

0.00

00.

000

Wel

l0.

218

0.48

90.

293

0.00

00.

017

Riv

er0.

114

0.64

50.

241

0.00

00.

000

Ven

dor

0.07

70.

558

0.36

50.

000

0.01

2im

prov

ed s

ourc

e0.

217

0.42

90.

354

0.00

00.

000

tap

wat

er

0.22

10.

417

0.36

20.

000

0.00

0

Not

es: M

eans

cal

cula

ted

usin

g da

ta f

rom

the

2004

Lan

d T

itli

ng S

peci

al P

roje

ct S

urve

y.

Pro

por

tion

of

HH

s u

sin

g w

ater

sou

rce

T-T

ES

TS

wit

h B

ase

Cas

e

28

Tab

le4:

Pro

bit

and

Mar

ginal

Eff

ects

Res

ult

s:L

and

Tit

les

and

Acc

ess

toW

ater

Sou

rces

Indo

or

ta

pO

utdo

or

ta

pS

hare

d

ta

pIm

prov

ed

sour

ceS

ink

Wel

lV

endo

r R

iver

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Pan

el A

: P

robi

t Res

ult

s: T

itli

ng

and

Acc

ess

to W

ater

Sou

rces

PE

TT

-0.0

20.

210.

050.

330.

170.

080.

06-0

.47

(0.0

6)(0

.10)

**(0

.06)

(0.0

8)**

*(0

.07)

**(0

.07)

(0.2

3)(0

.09)

***

TIT

LE

0.11

0.11

0.14

0.35

0.22

-0.0

90.

19-0

.38

(0.0

5)**

(0.0

9)(0

.05)

***

(0.0

6)**

*(0

.06)

***

(0.0

6)(0

.14)

(0.0

7)**

*

Pan

el B

: M

argi

nal

Eff

ects

Res

ult

s: T

itli

ng

and

Acc

ess

to W

ater

Sou

rces

PE

TT

-0.0

10.

020.

020.

070.

050.

020.

00-0

.08

(0.0

2)(0

.01)

*(0

.03)

(0.0

2)**

*(0

.02)

**(0

.02)

(0.0

0)(0

.01)

***

0.04

0.01

0.06

0.08

0.06

-0.0

20.

00-0

.07

TIT

LE

(0.0

2)**

(0.0

1)(0

.02)

***

(0.0

1)**

*(0

.02)

***

(0.0

1)(0

.00)

(0.0

1)**

*

Obs

3182

3182

3182

3182

3182

3182

3182

3182

Not

es: S

tand

ard

erro

rs in

par

enth

eses

(*

sign

ific

ant a

t 10%

; **

sign

ific

ant a

t 5%

; ***

sig

nifi

cant

at 1

%).

All

regr

essi

ons

incl

ude

regi

on

fixe

d ef

fect

s an

d co

ntro

ls f

or h

ouse

hold

cha

ract

eris

tics,

incl

udin

g: y

ears

of

HH

hea

d’s

educ

atio

n, w

heth

er th

e H

H h

ead

is f

emal

e,

num

ber

of p

eopl

e li

ving

in th

e H

H, n

umbe

r of

wor

kers

livi

ng in

the

HH

, num

ber

of c

hild

ren

betw

een

5 an

d 11

yea

rs o

ld, w

heth

er th

e fa

mil

y ow

ns a

hom

e bu

sine

ss, w

heth

er th

e H

H r

ecei

ves

any

inco

me

from

ren

ts.

29

Table 5: Difference-in-differences: Water Sources

Improved source

Outdoor tap

Indoor tap

All taps River Sink Well

(1) (2) (3) (4) (5) (6) (7)

program*no prior 0.06 0.02 0.03 0.05 -0.04 0.03 -0.01(0.03)** (0.02) (0.04) (0.04) (0.03)* (0.03) (0.03)

program 0.02 0.01 -0.01 0.00 -0.04 0.02 0.01(0.02) (0.01) (0.02) (0.02) (0.02)** (0.02) (0.02)

no prior title -0.07 -0.02 -0.02 -0.04 0.07 -0.05 0.01(0.02)*** (0.01)* (0.02) (0.02)* (0.02)*** (0.02)** (0.02)

Obs 3182 3182 3182 3182 3182 3182 3182

Notes: Standard errors in parentheses: * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions confixed effects and controls for household characteristics, including: years of HH head’s education, whether the HH head is feof people living in the HH, number of workers living in the HH, number of children between 5 and 11 years old, whether thowns a home business, whether the HH receives any income from rents.

30

Tab

le6:

Pro

pen

sity

Sco

reM

atch

ing:

Wat

erSou

rces

1mat

ch3

mat

ches

6 m

atch

es

1mat

ch3

mat

ches

6 m

atch

es

tap

wat

er

0.02

9.0

86**

*

.100

**

.094

***

.095

***

.0

97**

*

tap

indo

or-0

.012

.050

**

.067

**

.0

56*

.

064*

*

.062

**

tap

outd

oor

.041

***

.0

36**

*

.034

**

.037

**

.031

**

.035

***

sink

.045

**

0.03

1.0

38*

.0

42*

.0

42*

.0

37*

wel

l 0.

004

-0.0

07-0

.018

-0.0

2-0

.005

-0.0

03

rive

r-.

087*

**

-.12

1***

-.

129*

**-.

118*

**

-.1

29**

* .

-.13

1***

vend

or-0

.005

-0.0

01-0

.003

-0.0

07-.

010*

*

-.00

9*

impr

oved

sou

rce

.078

***

.1

10**

*

.121

***

.1

15**

*

.132

***

.1

30**

*

AT

TA

TE

Not

es: V

aria

bles

use

d to

det

erm

ine

prop

ensi

ty s

core

s in

clud

e: y

ears

of

HH

hea

d’s

educ

atio

n, w

heth

er t

he H

H h

ead

is

fem

ale,

num

ber

of p

eopl

e li

ving

in th

e H

H, n

umbe

r of

wor

kers

livi

ng in

the

HH

, num

ber

of c

hild

ren

betw

een

5 an

d 11

ye

ars

old,

whe

ther

the

fam

ily

owns

a h

ome

busi

ness

, whe

ther

the

HH

rec

eive

s an

y in

com

e f

rom

ren

ts.

31

Table 7: Difference-in-differences: Titles and Water Infrastructure Investments

(1) (2) (3) (4)

program*no prior 0.04 0.03 5.57 5.34(0.02)** (0.02)* (7.77) (7.87)

program 0.01 0.01 -1.57 -2.17(0.01) (0.01) (5.25) (5.32)

no prior title -0.03 -0.03 -10.96 -10.57(0.01)*** (0.01)*** (4.80)** (4.91)**

Obs 3204 3182 3204 3182

Notes: Standard errors in parentheses: * significant at 10%; ** significant at 5%; *** significant at 1%. "Water connect" is a binary variable for whether or not the HH invested in water infrastructure. "Water invest" is the amount a HH invested in water infrastructure. All regressions include region fixed effects and controls for household characteristics, including: years of HH head’s education, whether the HH head is female, number of people living in the HH, number of workers living in the HH, number of children between 5 and 11 years old, whether the family owns a home business, whether the HH receives any income from rents. In April 2004, 1 USD = 3.5739 Peruvian Nuevo Sol; all amounts reported in Sol.

Water connect Water invest

32

Tab

le8:

Pro

pen

sity

Sco

reM

atch

ing:

Tit

les

and

Wat

erIn

fras

truct

ure

Inve

stm

ents

1mat

ch3

mat

ches

6 m

atch

es

1mat

ch3

mat

ches

6 m

atch

es

Indi

cato

r fo

r w

heth

er H

Hs

inve

sted

in w

ater

con

nect

ion

0.00

50.

010.

011

0.01

0.00

80.

007

Tot

al a

mt i

nves

ted

in w

ater

co

nnec

tion

2.34

92.

676

1.12

82.

175

2.07

1.15

6

Am

ount

inve

sted

in w

ater

co

nnec

tion

if in

vest

ed a

t all

72.7

23*

33

.916

31.3

845.

866

11.0

5220

.64

AT

TA

TE

Not

es: V

aria

bles

use

d to

det

erm

ine

prop

ensi

ty s

core

s in

clud

e: y

ears

of

HH

hea

d’s

educ

atio

n, w

heth

er t

he H

H h

ead

is f

emal

e,

num

ber

of p

eopl

e li

ving

in th

e H

H, n

umbe

r of

wor

kers

livi

ng in

the

HH

, num

ber

of c

hild

ren

betw

een

5 an

d 11

yea

rs o

ld, w

heth

er

the

fam

ily

owns

a h

ome

busi

ness

, whe

ther

the

HH

rec

eive

s an

y in

com

e f

rom

ren

ts. I

n A

pril

200

4, 1

US

D =

3.5

739

Per

uvia

n N

uevo

Sol

; all

am

ount

s re

port

ed in

Sol

.

33

Table 9: Difference-in-differences: Titles and Payment for Water Access

(1) (2) (3) (4)

program*noprior 0.05 0.06 -6.95 -6.39-0.04 -0.04 (4.14)* -4.17

program 0.03 0.03 6.07 6.41-0.02 -0.02 (2.79)** (2.82)**

nopriortitle -0.08 -0.07 1.47 0.61(0.02)*** (0.02)*** -2.55 -2.61

Observations 3204 3182 3204 3182

Has monthly water bill payments

Monthly payment amount

Notes: Standard errors in parentheses: * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include region fixed effects and household characteristics, including: years of HH head’s education, whether the HH head is female, number of people living in the HH, number of workers living in the HH, number of children between 5 and 11 years old, whether the family owns a home business, whether the HH receives any income from rents. In April 2004, 1 USD = 3.5739 Peruvian Nuevo Sol; all amounts reported in Sol.

34

Tab

le10

:P

rop

ensi

tySco

reM

atch

ing:

Tit

les

and

Pay

men

tfo

rW

ater

Acc

ess

1mat

ch3

mat

ches

6 m

atch

es

1mat

ch3

mat

ches

6 m

atch

es

Indi

cato

r fo

r w

heth

er H

H m

akes

m

onth

ly w

ater

pay

men

ts

.060

*

.084

***

.

106*

**

.091

***

.0

70**

*

.0

80**

*

Am

ount

of

mon

thly

pay

men

ts8.

931*

*

8.30

1**

8.14

8**

18

.030

**

10.8

37 1

0.15

3**

AT

TA

TE

Not

es: V

aria

bles

use

d to

det

erm

ine

prop

ensi

ty s

core

s in

clud

e: y

ears

of

HH

hea

d’s

educ

atio

n, w

heth

er t

he H

H h

ead

is f

emal

e, n

umbe

r of

pe

ople

livi

ng in

the

HH

, num

ber

of w

orke

rs li

ving

in th

e H

H, n

umbe

r of

chi

ldre

n be

twee

n 5

and

11 y

ears

old

, whe

ther

the

fam

ily

owns

a

hom

e bu

sine

ss, w

heth

er th

e H

H r

ecei

ves

any

inco

me

fro

m r

ents

. In

Apr

il 2

004,

1 U

SD

= 3

.573

9 P

eruv

ian

Nue

vo S

ol; a

ll a

mou

nts

repo

rted

in

Sol

.

35

Table 11: Difference-in-differences: Titles and Financing Household Investments

(1) (2)

program*no prior 0.03 0.02-0.02 -0.02

program 0.01 0.01-0.01 -0.01

no prior title -0.03 -0.03(0.01)*** (0.01)***

Observations 3204 3182

Whether household invested own money in infrastructure

Notes: Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include region fixed effects and controls for household characteristics, including: years of HH head’s education, whether the HH head is female, number of people living in the HH, number of workers living in the HH, number of children between 5 and 11 years old, whether the family owns a home business, whether the HH receives any income from rents.

36