To pay or not to Pay? : a breakdown of how immigrant settling patterns affect their health-care usage : evidence from Dominican Republic

(1)

To Pay or Not to Pay?

A Breakdown of How Immigrant

Settling Patterns Affect their Health-Care Usage

Evidence from Dominican Republic

Master’s Thesis

Marisol Guzmán Benavides

Student Number: 11085975

Development Economics Track

Supervised by Prof. Menno Pradhan, PhD Academic Year 2015/2016

Abstract

We examine how the “pressure”-percentage of immigrants over total population- affects the type of health-care facility they go to when in need, as well as the probability of obtaining free health care. In order to study this, we apply probabilistic probit and multinomial logit regression analyses using data from the National Survey of Immigration (ENI 2012) from the Dominican Republic. We observed that pressure positively affects the probability of going to a public health-care facility but not of obtaining free health care, while individual characteristics, such as insurance tenure, Spanish level and employment status, make it more likely to attend a private facility and to pay for health-care services. We concluded that immigrants that settle in high-pressure areas are more likely go to public facilities, and this areas are associated with limited offer and lack of insurance. Key words: immigration, health care, utilization, OOP, public, immigrant pressure

(2)

Statement of originality

This document is written by Student Marisol Guzmán Benavides who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

What does the percentage of immigrants over a total population have to do with their health care utilization? The percentage of immigrants over total population or “pressure” could either positively or negatively affect their access to certain health-care facilities and payment options, depending also on social networks, supply responses from the healthcare system and solidarity from the receiving country’s government. There are several conditions in human development that can affect a person’s possibilities in life, and being an immigrant, for example, can determine an individual’s usage of health care, which can differ from that of local users because of legal rights or different health-care seek/usage behavior.

Now, how do governments respond to pressure? We have to begin by understanding where immigrants settle and what this means in terms of health-care demand and supply. Evidence from the United States suggests that immigrants concentrate (geographically) more than native populations do, controlling for age and ethnicity, and also that they usually settle in cities with large ethnic populations (Bartel, 1989). In other words, immigrants are likely to concentrate in specific regions of the country, like main cities or border areas. In largely populated cities, immigrants are less likely to generate pressure, because their weight in terms of population is not as significant. In smaller or less populated cities and border areas, immigrants are more likely to create a higher pressure because they make up for a bigger piece of the pie. This can be translated into higher immigrant weight over the public and private service health-care demand.

The way a country responds to increases in immigration depends on their own laws, social security systems and immigrant policies. A debate arises in this particular situation. On the one hand, governments can follow a restrictive approach, meaning that they will not take responsibility for the immigrants’ needs because it would imply helping a population that, in their opinion, does not contribute to funding the system. On the other hand, other countries follow a more cooperative and solidary approach, acknowledging that they do contribute to the country’s economy and provide some level of health care to this unprotected population (Dwyer, 2004).

The response to pressure does not only depend on the receiving country’s view on immigrants’ rights, but also on the type of system they have and, in general, the capacity of the country to finance the healthcare system, even if it has a solidary preference. In many developing

(5)

countries, the governments already struggle to finance health care for their local population, and universal health care is very difficult to achieve. In order to analyze how pressure affects immigrant’s health-care usage in a developing country, we studied the case of immigrants in the Dominican Republic, using a cross-section of the National Immigration Survey (ENI) of 2012.

The most relevant literature for this study helps identify which variables could affect the relationship between pressure and health-care usage by immigrants, as well as which methods are most useful to perform this estimation. For migration variables, Cerruti &Gaudio (2010) identify important differences in gender patters and Findley (1988) in age, which imply that health varies significantly after migrating depending on gender and age. These studies indicate that women are more likely to require healthcare, and more elderly people experience higher health declines after migration than youngsters do. Furthermore, the years of establishment have an important impact on health care. Evidence from the United Kingdom suggests that longer-established migrant populations are likely to have more children than groups that arrived more recently, who are more likely to be young and unmarried. This would generate an increase in the demand for maternal care in the receiving country (George et al, 2011).

Analyzing the supply side of health care, the geographical distance to the health-care facility is an important factor of usage decisions (Liu & Baker, 2006). This means that it is necessary to incorporate a distance variable that indicates either the distance of the person to the facility or the time required to get to the facility. Also, in this analysis we assume that health care is necessary (curative) and not preventive, following Pal’s (2012) differentiation, given that it is not possible to identify the type of health care they receive.

Our main findings indicate that pressure has a positive effect on the probability of an immigrant to seek health care at a public facility over a private one, but has no impact on receiving health care for free. Nonetheless, provinces with high pressure are likely to have positive supply of public free health-care responses from the government, but not from the private sector. This can suggest either a solidarity approach from the government or public health responsibility in specific or problematic areas. Individual characteristics, such as insurance tenure, better Spanish level, being employed, non-Haitian and older, make it more likely for an immigrant to attend a private facility and to pay for healthcare services or use an insurance. In other words, immigrants that have

(6)

managed to establish formally in economic terms are likely to migrate from public free health care to private payed services.

There are no other studies about impact of immigrant pressure in health-care usage to our knowledge, but, for our current research, we will also consider Arendt (2012) and Galárraga (2010), who examine immigrant health care usage in Denmark and Spain, focusing on differences between locals and foreigners, as references. This study will contribute to immigrant health-care usage literature in a developing receiving country and will help analyze the impact of weight of the population. Moreover, it will add new knowledge to the discussion about pressure impacts for government policies of service supply not only regarding health, but also in areas such as education, housing and security in the Dominican Republic.

The results can be extended to other developing countries that are big recipients of migration. It is also very relevant in low- and middle-income countries how the expenditure in health care affects the person, given that in most of these countries it disproportionately affects the poorest and the minorities (Galárraga et al., 2010).

2. Health Care System and Immigration in the Dominican Republic

Immigration in the Dominican Republic has increased at a high pace since the last decade (see Table 1). This has generated a lot of concerns in the government about the role that immigrants play in the country’s economy and society. They are located mostly in high concentration cities but, there are also clusters of immigrants in border areas and agricultural cities (see Figure 1 from Appendix).

Table 1: Immigrants and National Population in the Dominican Republic (2002, 2010, 2012)

2002 2010 2012

Status Frequency % Frequency % Frequency %

Dominican 8,466,308 98.8% 9,049,490 96.0% 9,292,789 92.0%

Immigrants _96,233 _1.2% 395,791 4.0% 803,268 8.0%

(7)

Most immigrants are from Haitian origin (73%)1 and many of their descendants are becoming part of the “ghost population”. This phenomenon of “ghosts” or “second class citizens” refers to people who are born in Dominican Republic, but whose parents are not legal residents. Therefore, they are not considered Dominicans by law and, consequently, are not granted birth certificates, so they cannot have their parent’s nationality either. This leaves the person living in the country they were born, but not as citizens and without the rights of being Dominican.

According to the National Constitution, Section I, Article 18, people born in the Dominican Republic territory who are sons or daughters of diplomats, consulates, foreign people “in transit” or that reside illegally in the Dominican territory are not considered Dominicans. This reform seems to be an answer to the Dominican Republic’s increased tendency for immigration, which goes hand in hand with the massive deportations in the last few years. Unfortunately for immigrants, the government’s rule for access to healthcare is that it will be universally provided to all legal residents of the country, which excludes a big percentage of the immigrant population.

Following the previous legal situation, it would be expected that a lot of immigrants are not being granted health care access (public and free), even though they are likely to be part of the poorest populations in the country (Arendt J. N., 2012). As Jiménez (2015) mentions in an Amnesty International report: “the multi-colored card [ID] may be small, but it makes the

difference between poverty and marginalization and a secure job, access to medical facilities, a school place and a chance in life.” (Jiménez, 2005, retrieved from: https://www.amnesty.org/en/latest/news/2015/11/a-legion-of-ghosts-of-haitian-descent-in-the-dominican-republic/)

The health-care system is currently a mixed one. With a reform in 2001, it is moving towards making the access to health care universal, protecting all the Dominicans and residents of the country, with no discrimination of health, sex, or social, political or economic condition, although in fact it has proven to have very poor coverage (Rathe, 2010). The system consists of public services, financed by taxing and insurance, and a big private sector that works through reimbursements or pre-paid platforms.

(8)

Currently, at the public sector, the system is divided into two levels: primary and specialized. For the primary level, the government has set to increase the health care coverage, implementing a model of primary health care units (UNAPs), which are small public health care points where they give basic services and promote prevention of diseases. For the second level, there are specialized units with more technology and human resources to deal with inpatients and required surgeries and are usually located in more densely populated areas. In level-divided systems, usually the person starts by the first level, unless it is an emergency. Once evaluated at the first level, if the doctors consider it requires a more specialized service, the person goes into a bigger facility/second level.

For areas far away from important demographic centers, the only service available is from the UNAPs which, as mentioned before, provide first level services only, such as treatment for infections and viral cases treatable with antibiotics or basic trauma issues. This is important as well, since access to basic health care is not the only relevant issue. On the contrary, access to

quality health care is an issue governments should also worry about. Although, it is clear that for

developing countries it is important to start by creating at least basic health care for everyone in the country.

The purpose of this study is to evaluate how immigrant pressure affects immigrants’ health-care usage and to calculate their probabilities of using the different systems depending on various factors. Given that there could be other mechanisms at work behind this, it is also relevant to examine what could increase or decrease their chances or receiving health care in spite of their immigrant status. By identifying this mechanisms, policy makers can make better decisions about budget distribution and better allocation of supply of free health care.

3. Data

The data used for this study is a cross-section of the 2012 “Encuesta Nacional de Immigración” - ENI (National Survey of Immigration) of the Dominican Republic, which is part of the country plan from the United Nations Population Fund (2012-2016) and is funded by the European Union. This survey data is representative of the immigrant population in Dominican Republic and analyzes the well-being of this population, asking about income, education, health, origin, and years since migration, as well as other more socioeconomic characteristics.

(9)

The survey is done with probabilistic sampling, covering 68,000 households, with 13,449 immigrants and 6,997 descendant from immigrants effectively interviewed, for a total of 20,446 people of foreign origin. The survey has national scope with five domains of statistical inference: provinces with high population concentration, frontier and neighbor provinces, provinces with sugar cane fields, provinces with rice or banana plantations and provinces with lower concentration of immigrants. Besides the ENI survey, we used the National Census of 2010 to identify the size of total population in each region in order to measure the pressure of usage. The Dominican Republic has 32 provinces, which are divided into 155 municipalities.

4. Empirical Strategy

Our research question is: how does immigrant pressure affect immigrant health-care usage? In order to answer it, we separately analyze the probability of using a public health-care facility and the probability of obtaining the service for free, of paying out-of-pocket (OOP), or through the insurance. We are interested in identifying how having a higher percentage of immigrants can affect both aspects and, by separating the analyses, we will also acquire insight into which aspects or characteristics matter for these decisions. We are testing two specific hypotheses:

Hypothesis 1: The higher the pressure, the more likely immigrants are to seek health care in a public health-care facility.

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = 𝑓𝑓(𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝_𝑗𝑗, 𝑊𝑊_𝑗𝑗, 𝑋𝑋_𝑖𝑖) (1)

In equation (1) we estimate the probability of the immigrant to go to a public facility versus to a private facility applying a probit model. We include our interest variable at municipality level (pressurej) with supply variables (Wj) also at municipality level and individual level controls (Xi).

Immigrant pressure points are likely to be in border areas (see Figure 2 in Appendix), as well as cultivation areas, which are not likely to have many private options. Instead, they usually only have public facilities and in a very limited number (see Figure 3 in Appendix), so it could imply that pressure positively affects the public facility usage. On the other hand, higher pressure of a population that is less likely to have legal status, ID card and other papers on rule could generate the opposite effect on public facilities, which could begin rationing the supply to this population, if they consider it necessary, in order to provide more services to the local population.

(10)

Another issue that could be relevant is the long-lasting discrimination against the Haitian population2. Given that the majority are Haitian descendants, there could also be discrimination in health-care services towards this population, which could imply that there’s a negative impact of pressure over public health-care centers. Basically, the impact could go both ways, and there are several underneath mechanisms working under the real impact, so in this case, we try to identify which force is stronger, either the positive or the negative mechanisms.

Hypothesis 2: The higher the pressure, the more likely immigrants are to obtain free health care.

𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = 𝑓𝑓(𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝_𝑗𝑗, 𝑊𝑊_𝑗𝑗, 𝑋𝑋_𝑖𝑖) (2)

We use equation (2) to test how a higher percentage of immigrants affects the probability of paying OOP or using insurance compared to obtaining health care for free. For this estimation, we use a multinomial logit, following a similar analysis to Fan et al (2011), applied to payment possibilities. Again, we include the pressure and supply variables at municipality level with the individual level controls. Following a similar logic as before, pressure could have a negative effect because of the crowding out effect, meaning that there could be a reduction or rationing of free health care to immigrants when they generate pressure. A possible positive mechanism of pressure over free health care is public health security. It is a matter of national interest to have healthy populations and this is a positive force for providing free health care. In addition, immigrants in Dominican Republic are likely to be part of the poorest segments of the population. Free health care is mostly given to vulnerable and poor populations, so it is likely that pressure points have high probability of receiving this service for free.

Furthermore, having connections and social networks (see Table 2) of immigrants could help to be more easily introduced into the system, making it easier for immigrants to cope with the bureaucratic processes to obtain free health care.

2_{Haiti and Dominican Republic share the Hispaniola Island in the Caribbean. Since the 1920’s, mostly male Haitian}

migrant workers have crossed the border to Dominican Republic as seasonal workers in the sugar cane industry. During 1952 and 1986, there were bilateral agreements between both governments to hire Haitian workers as sugar cutters. After some years of troubled sugar cane international market and fall of prices, anti-migrationdiscourse became predominant in the political field between both countries, limiting the number of Haitian migrants and restricting their descendant’s access to Dominican nationality (Amnesty International, 2015).

(11)

Table 2: Did you have a network or connection living in Dominican Republic before moving? Network or Connections High Concentration Border Sugar Cane Rice or Banana Low Concentration Total No 80.9% 72.1% 65.7% 80.9% 74.9% 73.9% Yes 19.1% 27.9% 34.3% 19.1% 25.1% 26.1%

Source: Data from ENI (2012)

5. Definition of Variables

Before we define the variables, it is important to mention that, even though in some cases self-report of health condition is used as a good approximation, it has several drawbacks. The need for health care can be difficult to measure since it depends on social norms and constructions of illness and perceptions of health (Harris, et al., 2011). In order to avoid bias3 of self-reporting illness, which is the question asked in the survey, we focus on immigrants who actually sought health care.

Health-Care Usage

There are the two dependent variables in our estimations. First, we have one dummy variable “public” for equation (1), which takes the value of one in case the immigrant attends a public facility and zero in case of attending a private facility. In this case, we focused on the relationship “private” versus “public,” given that it is not possible to identify the degree of emergency of the user or the type of services that they received. Second, payment of healthcare is indicated in a categorical variable for equation (2) that resumes the three options of health-care payment: out-of-pocket (OOP), through insurance, or free of charge.

Health-Care Supply

We created a variable of distance to hospital or clinic by estimating the weighted average distance of the people living in each municipality using the National Census of 2010, given that is average distance for the person that lives in each municipality, we assume there is no big difference

3

(12)

from 2010 to 2012. We also included a variable of number of public hospitals and clinics per municipality (this includes the UNAPs mentioned in section 2), which is a proxy for supply of health-care facilities from the government. These are the current hospitals (as of July 2016 according to the Health Ministry), but we assume that in four years (from 2012 to 2016) there has not been a significant change in the number of hospitals, given that these are not likely to be quickly built, changed or removed. Therefore, this is a good approximation of the amount of hospitals that were available in 2012.

Immigration Variables

The pressure variable is our main interest variable and, in order to create it, we divided the amount of immigrant population by the total population per municipality to obtain the weight of immigrants over the total inhabitants of each municipality. The denominator is from 2010 because the Census is the only available data source that divides population by municipality, while the ENI 2012 does it but only for immigrant population and both are nationally representative.

In addition, we included social networks, which indicate if the person had any connection or relatives in the country before moving. Furthermore, the Spanish level is also included, given that most immigrants are from Haiti, where Haitian Creole and French are the official languages, and this could represent a barrier to access health care. Spanish is also proxy for education given that the education level doesn’t include non-educated people (only as missing values), so half of the sample would be lost if we include education.

Income Measures

We included gross per capita income given that it is not possible to calculate if the person is paying taxes or what the final disposable income is. In order to create the income variable, we followed Bigsten, Kebede & Shimeles (2005) conversion method with averages to obtain a yearly income. For this research, we preferred yearly terms to avoid marked seasonality effect from poor rural immigrants working in agriculture4 (Ravallion, 2016). This variable includes earnings from their job or business in case they have one, alternative sources of income, such as side jobs, sales, etc., and remittances. Finally, we estimated the income per capita dividing total income of the household by the number of members of it.

(13)

Control Variables

We included a categorical control variable for age and dummies for gender, employment status and insurance tenure. Besides the control variables, dummies for the type of province (high concentration, border, sugar cane, banana or rice, low concentration) were included to control for heterogeneity.

6. Descriptive Statistics

Tables 3 and 4 present the percentages of some of the main variables for this study. The population analyzed in this research is only immigrant, and a profile of self-report and seek of health care is presented in Table 3. One of the most striking differences is the gender composition of the total immigrant population, with almost 61% of male members. This makes the whole immigrant population less likely to require health care because, according to previous studies, women are more likely to require it (mainly because of maternity). Also, almost 70% reside in urban areas and only 15% of the total immigrant population is insured.

Regarding self-reporting of illness, and keeping in mind this variable’s issues mentioned in section 5, women report ill or injured more often than men do. When analyzed by age groups, it is possible to see that young kids between 0-9 and older population of 65 or more appear to be ill or injured more often, which also responds to the general theory about health behavior5_.

The lowest income decile presents higher percentages of illness or injure self-report, which could be due to the lack of access to preventive care, and also because they are more likely to do hard labor than higher-income immigrants. As mentioned before, public health care in Dominican Republic is more curative than preventive, which means that if the person does not have the means to access preventive health care (which is more likely to be available in the private sector), then emergency health care will be used as a substitute. There are no significant differences between illness reporting by area (rural vs urban), insurance tenure or being Haitian or not.

(14)

In the case of seeking medical services, the gender and age variables were relevant. Being a man between the ages of 15 and 25 implies that you seek less health care. However, there are no significant differences for the other categories in the percentage of people who seek health care.

From Table 4 we can see that for the type of facility there are no differences between male and female in any kind of facility, but just as older people are more likely to use insurance or pay for health care, they are also more likely to attend to a private facility than any other group of ages. In general, for younger than 65 years old, immigrants attend more to public facilites. Also, almost 79% of the Haitians in need of health care use public facilities, while only 51% non-Haitians do. Public health-care facilities are more commonly used by the low-income deciles with almost 80% of the lowest decile using public facilities, while only 57% of the highest-decile immigrants use them. From this, it is possible to state that once the income allows it, immigrants prefer to attend a private facility. This could be due to the quality or level of the service, meaning that higher income allows immigrants to obtain better or higher-level services, and possibly even more specialized, in private facilities than in public ones.

Furthermore, we observe that there are no significant differences in type of payment by gender, but there are by origin, where 53% of Haitians obtain free health care, while only 38% non-Haitians do. Although is it clear that the poorest immigrants in Dominican Republic are more likely to be Haitian because of their own home-country situation. Insurance is not very likely to be used in immigrants under 65 years old, and instead OOP expenditures or free health care is more likely to be the paying situation. In terms of income distribution, it is also evident that the lowest deciles are the ones that obtain a higher percentage of free health care: the first three deciles with 62, 64 and 63%, while in the top three deciles, 48, 44 and 38% obtain free health care. This means that free health care is in general allocated to the people who need it the most.

Being from a rural area makes it more likely to obtain free health care, but it is also more likely to find only public facilities in rural areas. Then, for the non-insured immigrants about half of them end up paying and the other obtains it for free, while 60% of the insured ones used it as payment method.

(15)

Table 3: Immigrant Profile in Dominican Republic (2012)a)

Frequency* Percentages

Injured or Ill Seek Health Care

Yes No Total Yes No Total

Gender Male 466,593.52 60.69% 27.9% 72.1% 100% 89.61% 10.39% 100% Female 302,189.16 39.31% 36.87% 63.13% 100% 94.05% 5.95% 100% 768,782.68 100% Age Groups 0-9 161,152.91 20.99% 39.59% 60.41% 100% 95.70% 4.30% 100% 10-14 43,935.92 5.72% 25.84% 74.16% 100% 91.02% 8.98% 100% 15-19 66,752.69 8.70% 23.26% 76.74% 100% 86.20% 13.80% 100% 20-24 115,214.26 15.01% 27.84% 72.16% 100% 86.22% 13.78% 100% 25-29 122,891.87 16.01% 27.26% 72.74% 100% 90.59% 9.41% 100% 30-64 233,145.92 30.37% 31.29% 68.71% 100% 91.82% 8.18% 100% 65+ 24,551.44 3.20% 47.43% 52.57% 100% 94.10% 5.90% 100% 767,645.01 100% Origin Haitian 595,856.59 73.00% 31.61% 68.39% 100% 90.78% 9.22% 100% Non-Haitian 220,433.13 27.00% 33.10% 66.90% 100% 94.51% 5.49% 100% 816,289.72 100% Income Distribution Lowest decile 10% 34.25% 65.75% 100% 87.69% 12.31% 100% 2nd decile 10% 27.08% 72.92% 100% 93.58% 6.42% 100% 3rd_decile _10% _{24.11% 75.89% 100% 87.36% 12.64% 100%} 4th decile 10% 25.84% 74.16% 100% 90.79% 9.21% 100% 5th decile 10% 25.74% 74.26% 100% 85.17% 14.83% 100% 6th_decile _10% _{26.75% 73.25% 100% 90.94%} _9.06% _100% 7th decile 10% 26.67% 73.33% 100% 81.76% 18.24% 100% 8th decile 10% 26.58% 73.42% 100% 86.85% 13.15% 100% 9th_decile _10% _{25.33% 74.67% 100% 95.42%} _4.58% _100% Highest decile 10% 26.63% 73.37% 100% 90.52% 9.48% 100% Area Urban 568,549.74 69.65% 32.50% 67.50% 100% 92.07% 7.93% 100% Rural 247,739.98 30.35% 30.74% 69.26% 100% 90.68% 9.32% 100% 816,289.72 100% Insurance Yes 112,286.22 14.77% 33.52% 66.48% 100% 94.04% 5.96% 100% No 647,846.74 85.23% 31.77% 68.23% 100% 91.20% 8.80% 100% 760,132.95 100%

Source: Calculated with data from ENI (2012).

a)_{Includes all immigrants living in Dominican Republic in 2012, not only the ones who needed or sought health care.}

(16)

Table 4: Immigrant's Usage and Payment Conditional on Seeking Health Care

Payment Type of Facility

Insurance OOP Free Total Public Private

IDSS* or Low Level** Total Gender Male 9.78% 42.4% 47.8% 100% 72.96% 23.59% 3.44% 100% Female 11.62% 40.97% 47.42% 100% 72.20% 23.10% 4.70% 100% Age Groups 0-9 6.26% 40.25% 53.49% 100% 78.38% 18.38% 3.25% 100% 10-14 7.36% 33.55% 59.09% 100% 75.87% 20.38% 3.75% 100% 15-19 6.01% 37.76% 56.23% 100% 81.38% 16.68% 1.95% 100% 20-24 7.89% 48.20% 43.91% 100% 76.27% 22.57% 1.16% 100% 25-29 6.43% 45.78% 47.79% 100% 75.70% 21.24% 3.06% 100% 30-64 13.29% 42.78% 43.92% 100% 68.37% 27.00% 4.63% 100% 65+ 41.27% 29.66% 29.06% 100% 37.27% 45.09% 17.63% 100% Origin Haitian 5.48% 42.62% 51.90% 100% 79.09% 16.80% 4.11% 100% Non-Haitian 26.97% 38.96% 34.07% 100% 51.70% 44.19% 4.11% 100% Income Distribution Lowest decile 11.15% 26.48% 62.37% 100% 79.73% 15.81% 4.47% 100% 2nd decile 8.44% 26.58% 64.98% 100% 78.90% 13.92% 7.17% 100% 3rd decile 12.25% 24.51% 63.24% 100% 78.64% 13.11% 8.25% 100% 4th_decile _9.17% _25.69% _65.14% _100% _81.82% _11.82% _6.36% _100% 5th decile 10.78% 27.45% 61.76% 100% 78.30% 16.04% 5.66% 100% 6th decile 8.51% 38.72% 52.77% 100% 79.49% 14.96% 5.56% 100% 7th_decile _9.00% _36.97% _54.03% _100% _78.24% _18.06% _3.70% _100% 8th decile 15.42% 35.68% 48.90% 100% 76.39% 19.31% 4.29% 100% 9th decile 9.61% 45.41% 44.98% 100% 67.39% 30.00% 2.61% 100% Highest decile 19.37% 42.34% 38.29% 100% 57.52% 37.61% 4.87% 100% Area Urban 12.61% 46.38% 41.01% 100% 66.92% 28.62% 4.46% 100% Rural 5.98% 30.75% 63.26% 100% 85.74% 10.98% 3.29% 100% Insurance Yes 60.55% 22.10% 17.35% 100% 37.65% 51.81% 10.54% 100% No 1.24% 45.49% 53.27% 100% 79.21% 17.89% 2.90% 100%

Source: Calculated with data from ENI (2012).

*IDSS refers to entities from the Dominican Institute of Social Security which require insurance to be used. **Low level refers to military or church dispensary, where the service given is very basic.

(17)

7. Results

We divided our results by type of facility and by payment choices as analysed in Section 4 and include a goodness of fit analysis of the models at the end of each sub-section. For all models we included clusters at municipality level to allow for correlation within municipalities.

7.1 Type of Facility

In Table 5, we can see five different estimations of the probit model for utilization of public health-care facility conditional on seeking health health-care. We estimated different models for this variables in order to identify how different characteristics affect the decision. First, in Equation (1) we included only supply variables that result in not significance of the variable pressure and negative effect of income. For the Model (2) we included some individual characteristics such as gender and age, and on Model (3) we added socio-economic characteristics such as employment status, insurance tenure and also being Haitian. Models (4) and (5) include years since migration, having social networks and Spanish level, but the last one controls for type of province as well. We consider Model (5) as our complete and preferred model using the BIC criteria (we will address this at the end of this subsection).

Some variables are consistently significant even when adding controls (comparing the first and last models). From Model (5), pressure has a strong and significant effect over the probability of immigrants to attend a public health-care facility. This is consistent with our first hypothesis and it means that there is a positive mechanism behind the pressure, which increases the possibility of using public facilities.

Income is significant, but has the opposite sign from what would be expected, but the interaction term indicates that higher-income immigrants in high-pressure areas are less likely to seek public facilities. So, richer immigrants in pressured areas prefer to go to private facilities than to compete with the rest of the immigrants at public facilities. From Model (2) and (3) it is possible to see that adding the socioeconomic characteristics affects the sign of income. We have to keep in mind that going to a public facility does not mean obtaining health care for free, so for areas where there’s only public facilities, higher income also affects positively to the probability of going to a public facility in terms of opportunity cost (richer can afford not working one day).

(18)

Table 5: Estimation Results for Probability of Going to a Public Facility

(1) (2) (3) (4) (5)a)

Public Public Public Public Public

Income -0.0665 -0.0554 -0.000915 0.0814 0.0905* (0.042) (0.041) (0.046) (0.053) (0.050) Pressure 3.730 3.484 3.783 7.675** _7.493** (2.711) (2.777) (2.862) (3.206) (3.274) Inc*Press -0.259 -0.232 -0.285 -0.641** -0.662** (0.245) (0.249) (0.258) (0.284) (0.286) # Hospitals -0.00960*** -0.00968*** -0.00894*** -0.00687*** -0.00368* (0.002) (0.002) (0.002) (0.002) (0.002) Distance -0.00778 -0.00933 -0.000282 -0.000850 -0.00490 (0.012) (0.012) (0.013) (0.014) (0.012) Urban -0.309*** -0.319*** -0.254*** -0.251** -0.191* (0.090) (0.093) (0.093) (0.105) (0.108) Gender -0.0203 -0.0599 -0.0915 -0.0908 (0.038) (0.051) (0.062) (0.064) Age -0.00800*** -0.00522*** -0.00507 -0.00434 (0.002) (0.002) (0.003) (0.003) Haitian 0.505*** 1.058*** 1.118*** (0.095) (0.173) (0.176) Employed 0.0490 0.110 0.0875 (0.064) (0.091) (0.091) Insured -0.630*** _-0.717*** _-0.693*** (0.071) (0.094) (0.097) Social Net. 0.142 0.130 (0.096) (0.098) Spanish -0.0974** _-0.0913** (0.041) (0.040) Years Mig. 0.00734 0.00774* (0.005) (0.004) Constant 1.926*** _2.028*** _0.964* _-0.446 _-0.651 (0.459) (0.445) (0.510) (0.620) (0.592) N 4450 4450 3179 2453 2453 R2_p 0.0735 0.0832 0.123 0.135 0.142 AIC 3666.9 3632.8 2649.0 1980.2 1974.0 BIC 3711.7 3690.4 2721.8 2067.3 2084.3

a)_{Equation (5) includes type of province dummies, but none coefficient is significant.}

Standard errors in parentheses, all equations are calculated with clusters at municipality level.

(19)

On the other hand, being insured, having better Spanish level, non-Haitian and living in an urban area make it less likely for the immigrant to go to a public facility. Insurance tenure is associated with a more stable job, hence more stable income, which would also indicate the importance of the income effect, not only in levels, but also in terms of stability. It seems that insured immigrants are mostly used for private facilities, and if the person is not insured and had to pay for health care, it would be more affordable to pay public health care than private.

Worse income possibilities make immigrants seek public facilities over private, and less economic possibilities are also associated with areas far from the important cities, where the only options are to work either in sugar cane, banana or rice plantations. The same holds true for border areas, where people are mainly dedicated to informal commerce activities, although controlling for these kind of province is not significant.

On the supply side, the number of hospitals is significant at 10% as well as the urban variable. Urban areas are more likely to have a higher amount of supply, more types of services, both public and private. Having more available makes immigrants base their decisions in what they actually need, and if they can afford it. The fact that there are more choices, allows them to go to private facilities in case they are insured or can pay for it. Distance is not significant in any model although it is closely related with availability of facilities.

Years since migration becomes significant in our preferred model and it indicates that immigrants that have stablished for more time in Dominican Republic are more likely to go to a public facility. This could be explained because of two immigrant settling patterns. The first one is that immigrants that stay longer periods are mainly economic migrants, looking for better possibilities in terms of labour and settle in areas such as Border cities or agricultural cities, where there is a very limited supply of private health care facilities. Contrary to this, recently arrived immigrants might prefer to settle in cities where there are more job options and economic possibilities, allowing them access to more types of health care facilities, such as private clinics.

The second pattern is also described by George et al. (2011), where longer stablished immigrants tend to have more children than newly stablished. This can imply large costs in private facilities, which would induce them to seek health care in the public sector.

(20)

So, settlement of immigrants that causes high pressure seems to be associated with rural areas that have very limited health care supply, and if there is any supply is more likely to be public and very basic level, limiting their choices, regardless their income. When adding more choices and possibilities of health care, in big cities, or urban areas, immigrants might me prompted to look for specialized health care. As mentioned before, Haitian immigrants are likely to be part of the poorest population of the immigrants and the country in general, which reflects in the fact that they are more likely to attend to public facilities.

Insurance could be a main driver of the effect of pressure in public facility probability and this could bias our results, by making the effect of pressure stronger than it actually is. In Table 6 we make a comparison between immigrants that have insurance and those who don’t, using our preferred model from Table 5, equation (5), subject to insurance tenure. Results seem to be robust for immigrants that are not insured, this means that pressure has an effect for non-insured immigrants but not on insured ones in terms of probability of going to a public facility versus a private facility. Nonetheless, an important difference is present in supply of health care facilities. For non-insured immigrants, the more health care facilities available, the less likely to attend to a public facility. This is contrary to what we found in Table (5) and has opposite sign to what we would expect.

It is important to take this results very carefully since we are comparing two samples very different in size. The insured population is only 232 with higher standard errors, while the non-insured is 2221. The non-insured equation may be over fitted and causing biased estimations.

Goodness of fit for type of facility models

Comparing all five models from Table 5, the last two perform better in terms of BIC criteria, since they have a closer number to zero, and there is a difference of 17 between Model (4) and (5), so we consider it significant since it is bigger than 10, which makes the fifth more complete, and to have better fit for our estimations. Then, using the AIC criteria equation (5) also gives us the lowest estimate, which confirms our preferred model as the better fit.

(21)

Table 6: Preferred Model of Probability of Going to a Public Facility by Insurance Tenure Insured Public Not Insured Public Income 0.169 0.0873* (0.143) (0.052) Pressure 12.22 7.057** (11.188) (3.487) Inc*Press -1.146 -0.615** (1.017) (0.309) # Hospitals 0.00104 -0.00483** (0.004) (0.002) Distance -0.0764 0.000442 (0.054) (0.013) Urban -0.847*** -0.104 (0.181) (0.125) Gender -0.0957 -0.107 (0.226) (0.071) Age -0.00128 -0.00527 (0.008) (0.003) Haitian 0.872** 1.228*** (0.374) (0.218) Employed -0.333 0.131 (0.215) (0.103) Social Net. 0.445** 0.0753 (0.217) (0.097) Spanish -0.352*** -0.0584 (0.102) (0.042) Years Mig. 0.00817 0.00594 (0.008) (0.005) Constant -0.907 -0.746 (1.73) (0.604) N 232 2221 R2_p 0.237 0.096

Standard errors in parentheses, cluster at municipality level and both equations control for type of province.

*

(22)

7.2 Payment

In Table 7, we can see the results from the multinomial logit model for the payment possibilities with and without including interaction term between income and pressure. We compared the case of paying through insurance and OOP with obtaining health care for free.

Equations (1) and (3) are the same except for the interaction term and the same for equations (2) and (4) that also control from type of province. We included this four set of equations because income in this case plays a very important role so we want to see the differences of including and not including the interaction term. Also, as seen in the previous case, controlling for type of province makes a big difference. Panel A has the results of Insurance versus free health care, and Panel B has OOP versus free health care.

Starting with Panel A, pressure is not significant in any case, possibly because insurance usage responds more to individual characteristics than to the settlement arrangement or pressure that the immigrants could be causing to the system. The result of using insurance over obtaining free health care is also associated with very specific aspects such as supply and availability of mainly private clinics, which are the ones that generally require either OOP expenditures or insurance tenure.

Supply of health care facilities (# hospitals) is very significant in equations (1) and (3), but this disappears when controlling for type of province, although any type of province is significant either. This implies that the geographic location and population density are more relevant in terms of health care availability and defines better if a person has insurance, hence is more relevant in this probability. When type of province controls are applied, average distance to hospitals becomes significant. As Liu & Baker (2006) noted, distance to a hospital plays an important role in health care usage, in this case it has a positive impact in using insurance for health care services instead of obtaining health care for free.

Being Haitian is very significant in every case and it implies that if an immigrant is from Haiti, the relative log odds of using insurance to pay versus obtaining the service for free decrease between 1.4 and 1.5. As mentioned before, Haitians are likely to be the poorest immigrants and less likely to be insured or have a stable job and it reflects on more likeliness to obtain free health care than other nationalities.

(23)

Having an insurance is obviously very significant and the main driver in making this option more likely than getting free health care. Spanish level is also significant, the higher the Spanish level of the immigrant, the more likely to use insurance over obtaining free health care. This is also linked though higher education and hence better economic opportunities/stability.

Other socioeconomic and individual characteristics such as income, being from urban city, gender, age, possessing a Dominican ID, being employed, having social networks and years of migration are not significant. From this we can infer that being insured, non-Haitian, longer distance to hospitals (contrary to what expected) and a better level of Spanish increases the log odds of using insurance versus free health care.

Results of OOP payments compared to free are available in Panel B. Our main variable pressure has a negative and significant effect in equations (1) and (3), but the effect disappears when controlling for type of city. Given that both equations (2) and (4) are more complete and better fit, we chose them as our final models. For the first and the third, the coefficient is negative which would imply that the higher the pressure, the less likely the immigrant is to pay for health care, which can be also related to the previous analysis, where the higher the pressure, the more likely people are to seek a public health-care facility. Nonetheless, given that the effect disappears when controlling for type of city, we reject our second hypothesis of higher pressure increasing the probability of obtaining free health care over OOP or insurance.

Income is never a significant coefficient, but for equations (3) and (4) we observe expected sign. Also, the more available facilities there are, the more likely immigrants are to pay for health care than getting the services for free, although again the significance disappears when adding type of province controls. This could also be explained by the fact that, in general, there is more (national) demand in high concentration cities, which makes immigrants compete with other low-income local populations. For this reason, they might have to look outside the public area or, if they do go to the public facilities, they might still have to pay for the health care provided. This means that there could be some rationing of free health care for immigrants from the public sector.

Being from a border and rice/banana plantation province make it less likely to pay for health care, which also goes in tune with the previous reasoning. It is not possible to observe the effect of pressure directly with the pressure variable, but it could manifests through the type of

(24)

Table 7: Estimation Results for Payment Possibilities

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

Pay Pay Pay Pay

A. Insurance Income -0.198 -0.235 -0.0451 -0.0621 (0.237) (0.235) (0.111) (0.116) Pressure -10.19 -10.38 -0.218 0.835 (10.908) (10.728) (1.151) (1.348) Inc*Press 0.937 1.060 (0.978) (0.946) # Hospitals 0.0199*** _0.0105 _0.0196*** _0.0104 (0.005) (0.007) (0.005) (0.007) Distance 0.0313 0.0375** _0.0289 _0.0342* (0.022) (0.018) (0.022) (0.019) Urban -0.0404 -0.206 -0.0354 -0.197 (0.385) (0.390) (0.384) (0.387) Gender 0.340* 0.328 0.329 0.318 (0.203) (0.205) (0.202) (0.205) Age 0.00540 0.00409 0.00538 0.00412 (0.011) (0.011) (0.011) (0.011) ID -0.415 -0.393 -0.410 -0.385 (0.315) (0.283) (0.314) (0.282) Haitian -1.462*** -1.554*** -1.396*** -1.481*** (0.451) (0.473) (0.452) (0.477) Employed 0.180 0.234 0.215 0.270 (0.278) (0.275) (0.271) (0.269) Insured 4.852*** _4.827*** _4.834*** _4.808*** (0.382) (0.402) (0.373) (0.395) Social net. -0.478 -0.420 -0.476 -0.419 (0.293) (0.301) (0.296) (0.304) Spanish 0.273** 0.278** 0.274** 0.279** (0.109) (0.112) (0.110) (0.112) Years Mig. -0.00781 -0.00902 -0.00743 -0.00850 (0.012) (0.012) (0.012) (0.012) High Density 0.183 0.199 (0.554) (0.565) Border -0.885 -0.822 (0.568) (0.597) Sugar Cane -0.454 -0.418 (0.510) (0.525) Rice/Banana -0.560 -0.492 (0.631) (0.649) Constant -1.483 -0.634 -3.201** -2.601* (2.865) (2.747) (1.558) (1.526)

(25)

Table 7 (Continue): Estimation Results for Payment Possibilities

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

Pay Pay Pay Pay

B. OOP Income -0.0113 -0.0339 0.0443 0.0222 (0.070) (0.067) (0.047) (0.049) Pressure -6.922* -5.048 -3.201*** -1.277 (3.961) (4.036) (0.861) (1.024) Inc*Press 0.351 0.356 (0.368) (0.379) # Hospitals 0.0165*** _0.00249 _0.0164*** _0.00245 (0.005) (0.005) (0.005) (0.005) Distance -0.0514 -0.0367 -0.0522 -0.0379 (0.032) (0.023) (0.032) (0.023) Urban 0.205 -0.00587 0.203 -0.00727 (0.201) (0.206) (0.201) (0.206) Gender -0.0117 -0.0137 -0.0130 -0.0151 (0.123) (0.129) (0.122) (0.128) Age 0.00949 0.00529 0.00936 0.00510 (0.006) (0.006) (0.006) (0.006) ID 0.0374 0.0137 0.0359 0.0122 (0.087) (0.083) (0.087) (0.084) Haitian -0.376 -0.589 -0.359 -0.571 (0.410) (0.443) (0.409) (0.438) Employed -0.381*** -0.307*** -0.376*** -0.300*** (0.101) (0.108) (0.098) (0.104) Insured 0.347* _0.178 _0.341* _0.171 (0.209) (0.218) (0.207) (0.217) Social net. -0.120 -0.0610 -0.119 -0.0589 (0.129) (0.147) (0.129) (0.146) Spanish 0.278*** 0.275*** 0.278*** 0.275*** (0.077) (0.077) (0.077) (0.077) Years Mig. -0.0207** _-0.0224*** _-0.0205** _-0.0222*** (0.008) (0.008) (0.009) (0.008) High Density 0.468 0.461 (0.373) (0.373) Border -1.360*** _-1.365*** (0.368) (0.367) Sugar Cane -0.211 -0.206 (0.300) (0.302) Rice/Banana -0.979** -0.970** (0.408) (0.405) Constant -0.0947 0.884 -0.695 0.280 (1.065) (1.055) (0.824) (0.864)

(26)

Table 7 (Continue): Estimation Results for Payment Possibilities

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

Pay Pay Pay Pay

N 2514 2514 2514 2514

R2_p 0.231 0.263 0.23 0.262

AIC 3430.2 3305.9 3428.1 3304

BIC 3616.7 3539.1 3603 3525.6

Standard errors in parentheses, cluster at municipality level

*_{p < .1,}**_{p < .05,}***_{p < .01}

province, that is, the two previously mentioned provinces are really high-pressure areas on average.

Spanish level has a positive and significant effect on the odds of paying for health care over free. This is also positively associated with education level, which indicates higher chances of a more stable job. Also very relevant as in the previous cases, having migrated for longer time allows the immigrant to settle and learn how to use the system, but also through the two patterns mentioned in the case of insurance come to action and affect negatively the probability of paying for health care over free services.

Goodness of fit for type of payment

We analysed goodness of fit for the multinomial logit model by applying the BIC criteria as in the type of facility. Using the difference >10 units rule as well, models (2) and (4) are better than (1) and (3) respectably. Between model (2) and (4) we prefer the last one as well. The same conclusion is reached using the AIC criteria, where model (4) predicts better.

8. Sensibility analysis

Income provides an opposite expected sign in every relevant estimation, so we included another income variable, total income of the immigrant, which allows for zero income in case of unemployed and eliminates people under 14 (official working age). In Table 8 we can see the results of our main relevant variables from our estimations. Panel A repeats the same results

(27)

obtained in Tables 5 and 7 with per capita income, while Panel B includes the total income of the immigrant mentioned before.

Table 8: Sensibility Analysis of Different Types of Income in the Probability of Public Facilities

and Type of Payment

Public Payment

Insurance OOP

A. With income per

capita Income 0.0905* -0.235 -0.0621 -0.0339 0.0222 (0.050) (0.235) (0.116) (0.067) (0.049) Pressure 7.493** _-10.38 _0.835 _-5.048 _-1.277 (3.274) (10.728) (1.348) (4.036) (1.024) Inc*Press -0.662** 1.06 0.356 (0.286) (0.946) (0.379)

B. With total individual

income Income 0.0896* _-0.139 _-0.0119 _0.0166 _0.0999* (0.052) (0.216) (0.145) (0.086) (0.056) Pressure 9.164** _-7.312 _1.508 _-6.656 _-0.942 (3.589) (15.114) -1.324 (5.372) (1.101) Inc*Press -0.793*** 0.786 0.506 (0.307) (1.295) (0.44)

All models control for type of province and have cluster at municipality level. Standard errors in parentheses.

*_{p < .1,}**_{p < .05,}***_{p < .01}

For the type of facility estimation, results seem to be robust no matter the type of income included, and standard errors do not differ much either, so results are not sensible to the type of income and estimation maintains the same sign. For the payment estimations, results differ in significance and sign depending on the type of income used. This suggests that there could be an error in the specification of the model for equation (2).

9. Conclusion

We presented probabilistic analysis of impact of immigrant population participation in municipality level total population –pressure- over type of facility and payment for health care services. We concluded that the higher the pressure the immigrants make, the more likely it is for them to go to the public sector, but not necessarily higher odds of obtaining free health care. Characteristics, such as being insured, having better Spanish level, non-Haitian and living in an urban area, make it more likely to use private facilities and pay for the services. In sugar cane

(28)

provinces, companies insure workers, which makes it more likely for them to use the insurance to pay. The type of facilities immigrants choose also depends on the supplies available. Having social networks or Dominican ID do not affect health-care decisions. The type of province influences whether the person pays or not, but it does not seem to determine the type of facility the person is likely to use.

It is important to go back to the differentiation made in section 2 of this document about the level of health care that immigrants have access to. Most areas where immigrants generate higher pressure are rural areas, which are most likely to have public first-level facilities that deliver only primary health care. This is less expensive for a government than more specialized services; therefore, for many immigrants, this is the only solution available and they are mainly coping with their health issues through basic curative services like antibiotics and basic medicines. Also, it is more cost effective for a government to treat any contagious-infectious6 disease for free than having to deal with spreading diseases in their population.

In highly concentrated cities such as Santo Domingo or Santiago, they would be competing with more local population, so even if they are part of the poorest population and don’t have access to health care in any private facility, they are also competing with a lot of local population in the same situation, which decreases their chances of obtaining free health care, since there is a preference for the local and legal population.

Contrary to what would be expected given the legal situation of most immigrants in Dominican Republic, for most of the ones who are in need of health care, some degree of service is usually provided and a big percentage receives the service for free, even though they are very likely to be undocumented. This means that even though there is a policy that limits the access of health care for this population, they are still being taken care to a certain degree. We have to be very careful when making assumptions of the type of service they receive. Even though they may be granted access to a clinic, it doesn’t mean that they receive the level of service they actually require, especially given that in high pressure areas, very basic or UNAPs facilities are the only choice. Nonetheless, we can conclude that there are signs of an effort of universality of the access to health care in the Dominican Republic.

(29)

Furthermore, we find no evidence of discrimination in the reception of free health care. On the contrary, Haitian origin immigrants seem to be the most beneficiated from the public health care system, and they are also likely to be the poorest, which would indicate that the service is going to the people who need it the most.

In terms of public policy, there seems to be some rationing of free health care for immigrants in general in areas where they have lower pressure, like highly concentrated cities. Rationing public free health care could be a response from the government to crowding out of locals because of immigrant use. Also, for the public health care system is important to know what this means in terms of cost and their participation in the funding of the system. Moreover, it is important to identify if the behavior of immigrants is similar with the national population, then we could confirm that there is no discrimination based on nationality, but it would require to be complemented with an analysis of comparison national vs immigrant.

10. Limitations

In this study, there is no comparison with local use like in the case of Galarraga et al (2011). In order to make a more integral analysis, it would be important to compare the immigrant usage with Dominican’s behaviors, so as to identify any differences and gaps. Also, we can’t imply anything about real cost of usage by immigrants since there is no question about how much the actual cost is or what the immigrant payed for the services, in case they did pay. These data would give a more comprehensive picture of what immigrant’s health care represents to the public health-care system.

There were a few issues of estimation in the income variable, where several assumptions were required, which may affect the real income effect in the payment estimations. Besides this, we rely on self-reports of illness and self-report of attending a hospital or clinic for health care. Nevertheless, given that these are not institutional data, we can only assume that all health care was equally needed and we cannot identify the type of service.

(30)

Bibliography

Anderson, B., & Blinder, S. (2015, August 25). Who Counts as a Migrant? Definitions and their

Consequences. Retrieved from The Migration Obsevatory:

http://www.migrationobservatory.ox.ac.uk.

Arendt, J. N. (2012). The Demand for Health Care by the Poor under Universal Health Care Coverage. Journal of Human Capital, 6(4), 316-335.

Bartel, A. P. (1989). Where Do the New U.S. Immigrants Live? Journal of Labor Economics, 7(4), 371-391.

Bigsten, A., Kebede, B., & Shimeles, A. (2005). Poverty, Income Distribution and Labour Markets in Ethiopia. Sweden: Nordiska Afrika Institutet.

Cabieses, B., Tunstall, H., Pickett, K., & Gideon , J. (2013). Changing patterns of migration in Latin America: How can research develop intelligence for public health? Revista Panamericana de Salud Pública, 34(1), 68–74.

CEPAL. (2014). Tendencias y patrones de la migración latinoamericana y caribeña hacia 2010 y desafíos para una agenda regional (Patterns and tendencies of Latin American and Caribbean migration towards 2010 and challenges for a regional agenda) . CEPAL. Santiago: United Nations.

Cerrutti, M., & Gaudio, M. (2010). Gender Differences between Mexican Migration to the United States and Paraguayan Migration to Argentina. The Annals of the American Academy of Political and Social Science, Vol. 630, 93-113.

Cobo, S., Giorguli, S., & Alba, F. (2010). Occupational Mobility among Returned Migrants in Latin America: A Comparative Analysis. The Annals of the American Academy of Political and Social Science, Vol. 630, 245-268.

Dwyer, J. (2004). Illegal Immigrants, Health Care, and Social Responsibility. The Hastings Center Report, 31, pp. 34-41.

Elgar, F., & Stewart, J. (2008). Validity of Self-report Screening for Overweight and Obesity: Evidence from the Canadian Community Health Survey. Canadian Journal of Public Health / Revue Canadienne de Santé Publique, pp. 423-427.

Fan, J., Anderson, N., Foley, M., Rauser, E., & Silverstein, B. (2011). The persistent gap in health-care coverage between low-and high-income workers in Washington State: BRFSS, 2003-2007. Public Health Reports, 690-699.

Findley, S. (1988). The Directionality and Age Selectivity of the Health-Migration Relation: Evidence from Sequences of Disability and Mobility in the United States. The International Migration Review, Vol 22(3), 4-29.

Fussell, E. (2010). The Cumulative Causation of International Migration in Latin America. The Annals of the American Academy of Political and Social Science, Vol. 630, 162-177.

(31)

Galarraga, O., Sosa-Rubí , S., Salinas-Rodríguez, A., & Sesma- Vázquez, S. (2010). Health insurance for the poor: impact on catastrophic and out-of-pocket health expenditures in Mexico. The European Journal of Health Economics, 11(5), 437-447.

George, A., Meadows , P., Metcalf, H., & Rolfe, H. (2011). Impact of migration on the consumption of education and children's services and the consumptions of health services, social care and social services. London: National Instittute of Economic and Social Research.

Greene, W. H. (2012). Econometric Analysis. New York: Pearson.

Hadley, J., & Hargraves, J. (2003). The Contribution of Insurance Coverage and Community Resources to Reducing Racial/Ethnic Disparities in Access to Care. Health Services Research , 809-829. Hajizadeh, M., & Nghiem, H. (2011). Out of pocket expnditures for hospital care in Iran: Who is at risk of

incurring catastrophic payments? International Journal of Health Care Finance and Economics, 11(4), 267-285.

Harris, B., Goudge, J., Ataguba, J., McIntyre, D., Nxumalo, N., Jikwana, S., & Chersich, M. (2011). Inequities in access to health care in South Africa. Journal of Public Health Policy, 32, S102-S123. Jimenez, C. (2015, November 23). Amnesty International. Retrieved from A legion of "ghosts" of Haitian descent in the Dominican Republic: https://www.amnesty.org/en/latest/news/2015/11/a-legion-of-ghosts-of-haitian-descent-in-the-dominican-republic/.

Jimenez-Rubio, D., & Hernández-Quevedo, C. (2011). Inequalities in the use of health services between immigrants and the native population in Spain: What is dricing the differences? European Journal of Health Economics, Vol. 12(2), 17-28.

Lewis, M. (1993). User Fees in Public Hospitals: Comparison of Three Country Case Studies. World Bank. Liu, L., & Baker, J. (2006). The determinants of primary health care utilization: a comparison of three rural

clinics in Southern Honduras. Geojuornal.

Long, J. S. (1997). Regression Models for Categorical and Limited Dependent Variables. London: SAGE Publications.

Matibag, E., & Downing, T. (2011). Sovereignty and Social Justice: The "Haitian Problem" in the Dominican Republic. Caribbean Quarterly, 57, 92-117.

Migration Policy Institute. (2016, May 19). MPI. Retrieved from http://www.migrationpolicy.org/.

Olsen, J., Bloch, D., & Bloch, G. (2007). Controlling for Occasion-Specific Effects When Assessing the Test-Retest Reliability of Self-Report Health Questionnaires. Quality of Life Research, pp. 1399-1405.

Pal, R. (2012). Measuring incidence of catastrophic out-of-pocket health expenditure: with application to India. International Journal of Health Care Finance and Economics, 12(1), 63-85.

Rathe, M. (2010). Dominican Republic: Can universal coverage be achieved? World Health Organization. Ravallion, M. (2016). The Economics of Poverty. New York : Oxford University Press.

Secretaría de Estado de Trabajo. (2001, May 9). Ley No. 87-01: Ley de Seguridad Social. (C. N. Social, Ed.) Santo Domingo, Dominican Republic.

(32)

Wallace, S., Mendez-Luck, C., & Castañeda, X. (2009). Heading South: Why Mexican Immigrants in California Seek Health Services in Mexico. Medical Care, Vol 47(6), 662-669.

Appendix

(33)

Figure 2: Average Pressure by Type of Province in the Dominican Republic (2012)

Figure 3: Average Number of Public Hospitals and Clinics by Type of Province in the Dominican Republic (2012)

To pay or not to Pay? : a breakdown of how immigrant settling patterns affect their health-care usage : evidence from Dominican Republic