The Relationship Between Remittances and Health Care Provision in Mexico

Abstract

Objectives. We examined whether remittances sent from the United States to Mexico were used to access health care in Mexico.

Methods. Data were from a 2006 survey of 2 localities in the municipal city of Tepoztlán, Morelos, Mexico. We used logistic regression to determine whether household remittance expenditure on health care was associated with type of health insurance coverage.

Results. Individuals who lacked insurance coverage or who were covered by the Seguro Popular program were significantly more likely to reside in households that spend remittances on health care than were individuals covered by an employer-based insurance program.

Conclusions. Improving the coverage and quality of care within Mexico's health care system will help ensure that remittances serve as a complement, and not a substitute, to formal access to care.

Every year, Mexican migrants living in the United States send home billions of US dollars in the form of remittances, also known as “migradollars.” Since the 1980s, the flow of remittances has increased annually so that in 2006, Mexicans in the United States remitted 10 times the amount that was remitted in 1990, with an average annual growth rate of 15%.¹ According to the Central Bank of Mexico, remittances to Mexico during 2006 totaled $23.1 billion.² In 2007, the sum of remittances appears to have leveled off: a survey conducted for the Inter-American Development Bank estimated total remittances for 2007 to be $23.4 billion.³

Existing evidence suggests that remittances are used largely to purchase basic necessities.^4–7 In this context, US remittances provide migrants and their families in Mexico access to goods and services that they could not otherwise afford, including food, housing, and consumer goods. Another possibility that has received less attention in the literature is that remittances are used to access health care.

Mexico is marked by large inequities in health care access and consumption.⁸ The provision of health care is governed by a hybrid combination of employment-based formal insurance for salaried workers, publicly provided services for those excluded from formal insurance programs, and a much smaller percentage (2%) with private insurance coverage.⁹ According to the 2005 Mexican Conteo de Población y Vivienda (Count of Population and Housing), approximately 40% of the Mexican population had access to a formal insurance program.¹⁰

The rest of the population falls outside of the purview of formal social insurance programs and includes those who are self-employed, nonsalaried, unemployed, and informal workers; the majority of which are poor and are served by the Secretaría de Salud (SSA; Mexican Ministry of Health).⁹ The uninsured largely access health care through out-of-pocket payments paid either to public clinics or to a large, mostly unregulated private sector that provides fee-for-service care.¹¹ Out-of-pocket payments represent the highest proportion of health care expenditure in Mexico.¹² In an effort to reduce out-of-pocket spending and promote more equitable resource distribution, a new program called Seguro Popular (Popular Health Insurance Program) was introduced in 2004 and continues to expanded across the country.⁸ The program's aim is to increase financial protection of individuals outside the formal insurance programs by providing coverage for essential interventions (249 as of 2006) and selected catastrophic treatments.¹¹ The package of covered services will be expanded and updated annually. Currently, the program includes ambulatory care and hospitalization for basic specialties (e.g., internal medicine, general surgery, obstetrics and gynecology, pediatrics, geriatrics).⁸ As of September 2006, roughly 4 million families were enrolled in the program.¹¹

Migrants living in the United States (i.e., international migrants) and their families in Mexico may be more reliant on out-of-pocket spending on health care, which may be funded by remittances. Because they are less likely to have salaried jobs within the formal economy in Mexico, they are less likely to have access to an employer-based formal health insurance program in Mexico.¹³ There is some existing evidence that remittances are used to purchase health services in Mexico, although estimates vary widely. One study based on a community survey in Oaxaca estimates that as little as 1% of remittances are spent on health care.¹⁴ Estimates from a national-based survey of migrant households put the number closer to 50%.¹⁵ Such variability in estimates is likely caused by differences in sampling frames and survey methodology (e.g., whether respondents were allowed to list multiple uses for remittances or only 1). One of the few existing studies that looks explicitly at the relationship between remittances and household expenditure on health care used data from the Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH; National Household Income and Expenditure Survey).¹⁶ The study found that households receiving remittances spend more money on primary care expenses than do those that do not receive remittances.¹⁶ Another study using the ENIGH data also found a similar effect and reported that households receiving remittances exhibit a 44% rise in the share of the household budget spent on health care.¹⁷ Data from the Encuesta Nacional a Hogares Rurales de Mexico (ENHRUM; Mexico National Rural Household Survey) found that rural households with international migrants spend more income on health than do either internal migrant households or households with no migrants.¹⁸

The pattern between remittance receipt and health expenditure coupled with Mexico's large uninsured population (estimated at over 50%) raises the possibility that remittances may be used as a compensatory mechanism to address gaps in health care provision among Mexican migrant families.^8,19 If remittances serve as a method for redressing gaps in health care coverage, we would expect to find differences in how remittances are used by health insurance status. We examined whether remittances sent from Mexican immigrants in the United States to relatives in Mexico are used to pay for health care in Mexico. We tested this possibility by examining patterns of household remittance expenditure by individual health insurance status.

For all analyses we used data from a household census completed in the municipal city of Tepoztlán, Morelos, Mexico. The survey included an extensive module on US migration experience and remittance expenditure as well as health and health care, which allowed us to examine the associations between remittances, health insurance status, and health status at a level of detail that previously had not been possible.

METHODS

The Instituto Nacional de Salud Pública (INSP; National Institute of Public Health of Mexico) is responsible for carrying out most health and social program surveys in Mexico. Between October 2006 and December 2006, the INSP conducted a census in 2 communities, Santa Catarina and Tepoztlán, in the state of Morelos. The 2 communities were chosen because of their geographic proximity to the INSP and for their close approximation to the national average on several migration-related indices. Morelos represents the midpoint of Mexican states in terms of its level of remittance receipt (14th out of 32 states).¹ Both communities are close to the national average in the percentage of households that receive remittances and the percentage of households with emigrants in the United States.^10,20 Conducting the analysis in an average-level migration-sending region allowed us to more broadly generalize our findings than would have been possible in a high-level migration-sending region in which the patterns we observed would likely be more exaggerated.

The survey was unique in that it included a migration module in addition to a complete health module. Information was gathered on the US migration experience of each household member and on household receipt of remittances from the United States, including the specified uses of all remittances received.

The surveys were administered using computer-assisted personal interview survey technology. The complete migration and health modules were administered to a total of 2044 households. The nonresponse rate for the survey was 9.2% of households across the 2 communities. Each interview lasted approximately 60 minutes.

The models were estimated using multinomial logistic regression to predict the odds of residing in a household that does not receive remittances versus residing in a household that (1) spends remittances on health care or (2) receives remittances but does not spend them on health care.

The key explanatory variable in this set of models was the insurance status of the respondent. Insurance status was determined with the 4 major divisions discussed earlier that make up the Mexican health care system. Respondents were first categorized by whether they were enrolled in an employer-based formal insurance program extended to salaried workers and their families or if they reported receiving care from the SSA, which includes the new Seguro Popular. The employer-based category was further divided by whether coverage was provided for private-sector workers and their families by the Instituto Mexicano del Seguro Social (IMSS; Mexican Social Security Institute) or for government employees and workers in certain government industries and their families by the Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado (ISSSTE; Mexican Social Security and Services Institute for Government Workers) or companies such as Petróleos Mexicanos (PEMEX; the Mexican Oil Company). The small number of respondents who reported they were covered by private insurance (less than 1%) were placed in the same category as those insured by a government employee insurance program. The final insurance category encompassed all respondents that were not affiliated with any health insurance coverage.

Model 1 tested the relationship between household remittance receipt and individual insurance status. Model 2 examined this relationship after control for individual demographic variables, including gender, age, and health status, and for household-level variables, including whether the household head was male, had completed at least 9 years of schooling, was married, or was employed. We included a measure of whether the house was owned by the respondent and the quality of the household infrastructure. Houses with no electricity or with dirt floors were coded as having poor household infrastructure. A measure of whether any children resided in the household was included, as was a measure of per capita total household income, which was divided into quartiles, with the fourth quartile indicating the highest income level. A separate category was created for the nearly 20% of the sample that were missing on income information. All models controlled for clustering at the household level and were run in Stata version 9.2 (StataCorp, College Station, TX).

RESULTS

Table 1 presents the descriptive statistics for the sample, stratified by household remittance receipt and expenditure. The majority of individuals in the sample lived in households that reported never having received remittances (around 90%). Of those households that had received remittances, over twice as many had spent remittances on health than had not (7% versus 3%).

TABLE 1.

Sample Characteristics, by Household Remittance Receipt and Expenditure: Santa Catarina and Tepoztlán, Morelos, Mexico, October–December 2006

		Received Remittances			Significant Difference,a χ2 or F
	Total	Used for Health Care	Not Used for Health Care	Did Not Receive Remittances
Total, no.	7764	536	259	6969
Total, %	100.0	6.9	3.3	89.8
Insurance, %					23.25*
ISSSTE	10.0	6.0	10.0	10.3
IMSS	12.4	11.2	11.2	12.5
Seguro Popular	12.7	17.2	8.9	12.4
None	65.0	65.7	69.9	64.7
Age, y, mean	30.68	29.7	30.0	30.8	0.94
Men, %	48.5	43.7	44.4	49.1	7.65*
Head of household characteristics, %
≥ 9 y of education	57.8	64.7	58.3	57.3	11.33*
Married	86.3	84.0	81.1	86.6	9.12*
Employed	79.8	78.9	81.1	79.9	0.53
Income quartile, %
Missing	20.4	16.8	17.4	20.7	32.05*
1st (lowest)	19.9	18.5	14.3	20.3
2nd	19.9	18.3	19.3	20.1
3rd	20.0	19.6	26.3	19.8
4th (highest)	19.9	26.9	22.8	19.2
Children in household, %	73.1	74.6	71.0	73.0	1.21
Own house, %	71.9	80.0	66.0	71.4	22.68*
Poor house infrastructure,b %	6.1	3.5	3.5	6.3	9.98*
Self-reported health status, %
Very good/excellent	97.0	95.7	95.8	97.2	5.32
Good/fair/poor	3.0	4.3	4.3	2.8

Note. ISSSTE = Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado; IMSS = Instituto Mexicano del Seguro Social.

^aThis is the difference by household remittance expenditure. The χ² test was used for categorical variables, and the F test was used for continuous variables.

^bHouses with no electricity or with dirt floors were considered to have poor household infrastructure.

^*P < .05.

Sixty-five percent of the sample reported having no insurance coverage at all, a number that corresponds with estimates from the 2005 Mexican Count of Population and Housing for these localities.¹⁰ The percentage of the sample who reported being part of the Seguro Popular program is higher than that reported in the 2005 population count of Tepoztlán and Santa Catarina. Because the new program was initiated in 2004 and continues to expand across the country, the increased number of respondents who reported that they had received care under the Seguro Popular in the 2006 Tepoztlán survey likely reflects this ongoing expansion.

After we stratified insurance status by household remittance receipt, we found that individuals residing in a household that did not receive remittances or those residing in a household that did receive remittances but did not spend them on health care were more highly represented in the employer-based formal insurance program categories (i.e., ISSSTE or IMSS) compared with those who received remittances and spent them on health care. Compared with the other groups, individuals living in households that spent remittances on health care were more likely to be covered by Seguro Popular. Individuals living in households that received remittances but did not spend them on health care were more likely to be uninsured.

Remittance-receiving households appeared to be more advantaged than nonremittance receiving households in terms of per capita household income level, with individuals living in remittance-receiving households more highly represented in the top 2 income quartiles. Remittance-receiving households were also more likely to have better household infrastructure. Individuals in households that spend remittances on health care were more likely to own their own home, suggesting that some migrant families were able to translate remittances into an investment. Heads of these households were also more likely to have completed secondary school (defined as 9 or more years of schooling). There were no significant differences across household types regarding the presence of children, employment status of household head, health status, or mean age.

Table 2 presents the results from the set of models examining the relationship between household remittance receipt and insurance status. Remittance-receiving households were differentiated by whether remittances were spent on health care or not. The reference group for both categories was households that had not received remittances. Remittance expenditure on health care was significantly associated with insurance status. Individuals covered by the SSA (i.e., Seguro Popular) were over 2 times more likely to reside in households that spent remittances on health care than were individuals covered by insurance for government employees (e.g., ISSSTE; odds ratio [OR] = 2.39; 95% confidence interval [CI] = 1.17, 4.87).

TABLE 2.

Odd Ratios (ORs) For Using or Not Using Remittances for Health Care Among Households That Received Remittances: Santa Catarina and Tepoztlán, Morelos, Mexico, October–December 2006

	Model 1		Model 2
	Remittances Used for Health Care, OR (95% CI)	Remittances Not Used for Health Care, OR (95% CI)	Remittances Used for Health Care, OR (95% CI)	Remittances Not Used for Health Care, OR (95% CI)
Insurance
ISSSTE	1.00	1.00	1.00	1.00
IMSS	1.552 (0.771, 3.125)	0.923 (0.380, 2.245)	1.739 (0.848, 3.566)	0.918 (0.379, 2.225)
Seguro Popular	2.388* (1.171, 4.867)	0.735 (0.243, 2.223)	3.263** (1.538, 6.923)	0.801 (0.264, 2.431)
None	1.755 (0.995, 3.095)	1.111 (0.556, 2.221)	2.221* (1.196, 4.123)	1.166 (0.566, 2.404)
Age			0.999 (0.994, 1.004)	0.996 (0.990, 1.002)
Gender
Women (Ref)			1.00	1.00
Men			0.818** (0.718, 0.931)	0.838 (0.689, 1.020)
Head of household charactersticsa
Male			0.627 (0.337, 1.168)	0.849 (0.361, 1.995)
≥ 9 y of education			1.338 (0.863, 2.074)	0.935 (0.528, 1.655)
Married			1.061 (0.566, 1.987)	0.694 (0.322, 1.495)
Employed			0.416 (0.140, 1.236)	1.738 (0.524, 5.770)
Income quartile
Missing			0.541 (0.260, 1.126)	0.317 (0.090, 1.109)
1st (lowest)			0.291* (0.089, 0.959)	1.070 (0.336, 3.409)
2nd			0.745 (0.415, 1.337)	0.504 (0.211, 1.208)
3rd			0.739 (0.436, 1.250)	0.685 (0.330, 1.421)
4th (highest; Ref)			1.00	1.00
Poor household infrastructureb			0.521 (0.126, 2.153)	0.545 (0.140, 2.119)
Children in household			1.115 (0.709, 1.754)	0.972 (0.535, 1.766)
Own house
Yes			1.820* (1.134, 2.921)	0.786 (0.454, 1.361)
No (Ref)			1.00	1.00
Self-rated health
Poor/fair/good (Ref)			1.00	1.00
Very good/excellent			0.837 (0.623, 1.125)	0.894 (0.604, 1.323)

Note. ISSSTE = Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado; IMSS = Instituto Mexicano del Seguro Social; OR = odds ratio; CI = confidence interval. There were 7764 total observations.

^aThe reference groups for this category were female head of household, fewer than 9 years of education, not married, and not employed.

^bHouses with no electricity or with dirt floors were considered to have poor household infrastructure.

*P < .05; **P < .01.

Similarly, individuals who did not report any insurance coverage were 2 times more likely to live in households that spent remittances on health care than were individuals that were covered by government employee insurance. This effect became significant at the .05 level after sociodemographic characteristics were controlled in model 2 (OR = 2.22; 95% CI = 1.20, 4.12). Between individuals covered by different types of employer-based insurance (e.g., IMSS vs ISSSTE), there were no significant differences in the likelihood of spending remittances on health care.

The significant differences in health insurance status among individuals residing in households that spent remittances on health care were not replicated among individuals residing in households that received remittances but did not spend them on health care. In these cases, there were no significant differences in health insurance status between individuals residing in households receiving remittances but not spending them on health care and those not receiving remittances.

In model 2, we controlled for sociodemographic characteristics as well as a measure of self-rated health. The controls strengthened the relationship between health insurance status and remittance receipt among individuals living in households that spent remittances on health care. Once differences in sociodemographics and health status were accounted for, individuals that did not have health insurance or had Seguro Popular were even more likely to reside in households that spent remittances on health care than were individuals covered by an employer-based formal insurance program.

Men were less likely to live in a household that spent remittances on health care. Individuals living in households in the lowest income quartile were significantly less likely compared with those in the highest income quartile to spend remittances on health care. Home ownership increased the likelihood that remittances were spent on health care. Individual health status was not significantly associated with remittance receipt and expenditure on health care. There were no significant differences in insurance status or socioeconomic background between respondents residing in households that received remittances but did spend them on health care and those that did not receive remittances.

DISCUSSION

Our findings suggest that one of the uses of remittances sent from the United States is to gain access to health care for individuals who do not have formal access through an employment-based insurance program in Mexico. This relationship is salient for individuals who report coverage by Seguro Popular and for individuals who do not have health coverage. More frequent use of remittances on health care by individuals with these kinds of insurance supports the possibility that US remittances provide a way for some migrant families to access health care in Mexico.

The finding that uninsured individuals are significantly more likely to live in households that spend remittances on health care likely indicates the use of remittances for pay-for-fee services. The finding that individuals enrolled in the new Seguro Popular program are even more likely to live in a household that spends remittances on health care than are uninsured individuals raises several possibilities. The Seguro Popular program was designed to limit out-of-pocket payments on health and promote more efficient and equitable health care provision to uninsured Mexicans.^8,11 In addition to contributions from the federal and state governments, the program requires a progressive prepayment from beneficiaries that should not exceed 5% of disposable income. Individuals in the 2 lowest income deciles are not required to contribute in monetary terms. It is possible that international migrants and their families, who are not in the poorest income deciles, are subject to the enrollment fee and that remittances from abroad may be used for this purpose.

Alternatively, individuals covered by the Seguro Popular program may use remittances to access services not provided by the program. Without more information on how remittances used for health care are spent, it is not possible to determine from these data if remittances are used to pay the Seguro Popular enrollment fee or to pay for services not provided by the SSA. The latter possibility is concerning because reduction of out-of-pocket spending on health care services is one of the principal aims of the new program. More detailed information on health expenditure and remittance receipt would clarify how remittances are used to access health care. Such data would help clarify whether remittances are used as 1-time payments during health emergency situations, as payment for pay-for-fee services, or as a way to gain access to Seguro Popular.

Migrants and their families are unique in the sense that they are not poorest of the poor but they are uninsured at higher rates because they are excluded from employer-based insurance programs in Mexico. It will be important to evaluate the effects of the expanding Seguro Popular in the general population as well as among segments that may be unique, such as international migrants and their families.

The pattern we found is suggestive of a causal relationship whereby either no insurance coverage or Seguro Popular coverage leads members of migrant families to spend remittances on health care. Longitudinal data of national scope is required to verify the cross-sectional association. Tepoztlán, where the present survey was conducted, is an area with average migratory intensity.²⁰ How these patterns may differ in areas with different levels of migratory intensity is not yet known. Households in communities with high or very high migratory intensity may be even more reliant on remittances as a compensatory mechanism for limited access to health care. This possibility gains added urgency in light of recent evidence suggesting a decrease in remittance volume.^3,21 If the patterns presented here reflect truly causal relationships, then decreased remittance levels will cause immediate concern for the health of migrants and their families. Improving coverage and quality of care within the national health care system will help insure that remittances complement but do not serve as a substitute to formal access to care.¹¹