An increasing amount of OECD countries are becoming reliant on immigration as a means of dealing with shortages of health care professionals. The affect of out-migration, especially in the continent of Africa where resources are extremely low, is becoming more concerning. This paper will examine the role of wages and per capita health expenditure in the migration decision and discuss the effects of changes in both variables on the density of health care professionals. We will also discuss the supply-side can do to limit the out-migration of physicians and nurses.
This paper will use wage and health expenditure differentials in the health care sector between source and destination countries (adjusted for PPP) to test the hypothesis that higher wages and health expenditure per capita lead to a larger density of health care professionals in a given country. Data is also presented on other important factors affecting migration. There exists a surprisingly lower correlation between wages and density than with per capita health expenditure and density.
This is the direct implication of the wage differentials between developing and OECD countries being so large that a small increase in wages in the native country are not likely to affect the decision of the health care worker. 1. 0 Introduction According to the WHO (2011) there are currently 60 million health workers worldwide, a density of 86 per 10,000 people. The WHO (2010) defines health workers as ones currently in employment, which excludes emigres who no longer practice their profession. Health workers are considered as individuals whose intentions are to enhance the health of another person or of a system.
They include professions such as doctors, nurses, pharmacists, and laboratory technicians. Support workers, volunteers and management roles such as hospital managers, financial officers, cooks, drivers and cleaners are also included. In 2000, according to a study by the United Nations, 175 million people (est) or 2. 9% of the global population were known to have been living outside their country of birth for longer than a year. These people are known as International Migrants. Of this number, 65 million are believed to be economically active1 (UN, 2001).
In absolute terms, the stock of migrant population2 has increased by 0. 3% from 2. 3% to 2. 9% during the years 1965 – 2000. When studying the effects of migration, it is important to consider both the positive and negative effects. Particularly for migrants who move from developing countries with the opportunity of securing better job, being offered higher wages abroad, remittances are an example of migration yielding positive externalities. Remittances provide income streams to many developing countries in need of foreign exchange.
In these developing countries, remittances often provide a larger fraction of foreign exchange than foreign investment, international trade and foreign aid. Turnell et al. discovered that in 2006, remittances made by an estimated 150 million migrants worldwide totalled around $300 billion (US) in the same year where the sum total of foreign direct investment, aid and international trade in developing countries was $270 billion (US). However, in cases where situations such as brain drain exist, migration also acts towards precluding economic and societal development. Stilwell et al. 2003) noted that one of the largest issues with increasing out-migration from developing countries is the loss of educated professionals to richer countries. His paper on developing ethical policies on the migration of health workers found that 65% of all economically active migrants who have migrated to developed countries are classed as “highly-skilled”. In order to be classified as a highly skilled professional, an individual is expected to have completed education up to a tertiary level. In the health sector workforce, this includes pharmacists, dentists, nurses and physicians.
This paper will be focusing on physicians and nurses. The increasing numbers of out-migration of health care professionals are having a significant impact on the welfare of health care systems worldwide, predominantly on lower income, resource-poor countries. The migration of health care workers is nothing new though; it has closely followed the trends in international migration since the boom of out-migration in the mid 1900’s (Stilwell, 2003). However, the concern that is unique to health sector migration is the dependency of developed countries on the recruitment of physicians and nurses from developing countries.
Many developing countries are now the main source of health care professionals for many developed countries. This pattern of migration has raised concerns, particularly in African countries where the health worker population as well as the health care system is being adversely affected (Vujicic, 2004). There are several different ways to calculate migration. Arah (2007) used the absolute number of migrants to migrate from a country to calculate the migration of health workers. This paper will be using a method used by Dacuycuy (2008), analyzing the increase/decrease in the density of health workers in a country.
In any given country, the infrastructure of the health care system is a crucial component of development. The migration of health professionals remains controversial because of the adverse affects it leaves on the health care systems in their native countries. One ongoing problem that exists however is very limited statistical evidence that can empirically validate this theory. One possibly point of inquiry is to identify the reasons behind the migration of health care personnel. What are the determinants of the migration of health professionals from developing countries?
What affect do foreign wages have on health care workers compared to domestic wages? Is there an empirical link between the density of health care workers and government spending? Figure 1 shows that as wages and per capita spending increases, so does the density of physicians. In the first fourteen developing countries, the wage rates fluctuate more than per capita spending, which suggests that individuals in source countries considering migrating abroad may only be looking to migrate to the developing countries we have at the end of the graph.
The wage differentials between developing and developed countries may be so such that wages offered in another developing country may not be enough to persuade a health worker to migrate. Figure 2 shows the same data as graph 1 but in the case of nurses. The main observation of this graph is that aside from Sierra Leone, the per capita expenditure and monthly wage seem to have a higher correlation than seen for physicians. Overall, with both graphs, we can see that there is a positive relationship between the monthly wage of physicians and nurses, per capita expenditure and the density of health care workers in the respective countries.
We will be studying the relationship between the three variables as well as the reasons why a higher positive correlation exists in some cases more than others. Though Fig 3 shows an obvious positive correlation that exists between physician wages and physician density, it also brings to like the size of the wage and density differentials of source and destination countries. Fig 1. Graph comparing physician wage and per capita expenditure to the density of physicians Fig 2. Graph comparing nurse wage and per capita expenditure on the density of nurses
Fig 3. A Scatter diagram of the wages of physicians, and physician density with a line of best fit 2. 0 Review of Current Literature The majority of the literature that I have reviewed has been sourced from the World Health Organization (WHO). The majority of papers published on the topic of health worker migration use the WHO as the most common primary source of statistics and data. The first point of interest is in regards to international migration of health workers and the resulting workforce shortages in developing countries.
The WHO reported in 2006 that African countries represent only 3% of the of health workers worldwide despite representing 11% of the global population. This is made even more concerning by the 24% of the world’s total disease count the continent harbors. This same study indicated that in order for a country to be able to provide an adequate healthcare system, the combined density of nurses and doctors should be at least 22. 8 per 10,000 (Verboom et al, 2006).
According to Levitt (2000), who analysed the impact on source countries as a result of health worker migration, immediate cessation of recruitment from countries with ailing health care systems should be considered. Pang et al. (2002) suggested a similar cause of action in migration reversal, citing that the brain drain effect on particular source countries are such that it weakens systems that are already short in funding, workforce and education. 2. 1 Migration Flow of Physicians In the USA, International Medical Graduates (IMGs) have become an extremely valuable part of the health care system.
The WHO (2006), through their yearly health statistics database, reported that from 1978 – 2004 the number of IMG physicians increased on average by 4,873 annually, totalling 215,576. This represents an increase in the proportion of IMGs from 22. 2% to 25. 6% compared to Local Medical Graduate (LMG) physicians. A report by Mullan (2005) who studied the metrics of the physician brain drain also shows comparatively similar ratios in the United Kingdom, Australia and Canada where 25%, 26. 5% and 23. 1% of physicians are IMGs. There are, however, larger differences in the percentage of IMGs that originate from developing countries.
According to the ILO (2008) 75% of the UK’s IMGs are from lower income countries, which is substantially larger than 60% in the USA, 43% in Canada and 40% in Australia. The composition of the IMG stock of physicians in developed countries such as the UK has changed over time according to Vujicic (2004). Due to shortages of physicians in many of destination countries, the main supply of migrant physicians are from developing countries. The OECD (2006) Health Project reported that in the UK for example, African IMGs make up 20% of the physician workforce, whereas 30% of the physicians in the USA are from Pakistan and India.
Also 10% of physicians in Canada graduated in South Africa. Despite the lack of statistics on health workers worldwide, we already get a picture of how dependent developing countries are on the supply of migrants from developing countries. One reason why out-migration of health workers is becoming a topic of great discussion and concern is because of the low density of health workers in developing countries. As Vujicic (2004) noted, the stocks of physicians, particularly in Africa, are so low that small absolute losses might still be extremely large in relation to a countries’ total physician workforce.
Research on emigration of health care workers in Africa by Clemens at al. (2006) found that Ghana and South Africa had lost 1,693 and 7,363 physicians respectively to countries that are considered destination countries, which include Canada, France, the UK and the USA. The concerning statistic however is that relative to their entire physician workforce, they lost 56% and 21% respectively. Arah (2007) studied the metrics behind such figures in Africa and said that this particular migration is “threatening to already weak systems”. 2. 2 Migration Flow of Nurses Lorenzo et al. 2007) studied nurse migration from the perspective of the source country. He highlighted that out of 193,223 registered nurses worldwide, 85% are working outside of their graduate country. He also pointed out that 30% of registered nurses in the USA were from the Philippines. This figure is the same as the aforementioned percentage of Indian and Pakistani physicians in America. Figures by Buchan (2003) who studied the trend of nurse international nurse migration over time also show that from 16,000 registered nurses in the United Kingdom, over 50% were IMGs compared to 25% in 1998.
He indicated that there were also major shortages in Canada and the USA. Four of the main source countries were India, Philippines, South Africa and Zimbabwe. Yamagata (2007) examined the securing of medical personnel in two source countries and two destination countries, and his data mainly focused on the deployment of nurses from the Philippines abroad. He concluded that while an upward trend in the density of migrant nurses abroad exists more in countries such as the UK, Ireland and Saudi Arabia, the slowdown in the US is predominantly down to policy changes and immigration initiatives.
Policy changes included budget cuts and a reduction in per capita spending in the health sector. While there may have been a slowdown, statistics from the US Department of Health Services show that while in the year 2000, the shortage of registered nurses in the US was estimated at 110,000 (6%), they expect the shortage to increase intensively to 29% in 2020. Aiken et al. (Vol. 23) predicted that the shortage of nurses in the UK would reach 53,000 by 2010. Statistics showing the number of nurses that leave developing countries are sparse but where they do exist, they still suggest an autonomous increase in migration flows.
For example, Buchan’s report showed that while 4,500 nurses emigrated abroad from the Philippines in 1996, 12,000 nurses emigrated in 2001, a third of which ended up in the United Kingdom. Lorenzo (2007) also reported that up to 30% of all the trained nurses in the USA were Filipino graduates. Matsuno (2008) showed the lengths at which migrants are willing to go to work abroad. Since 2000, 3,500 physicians have migrated from the Philippines only to work as nurses in the destination country. In 2005, 4,000 doctors completed nursing school abroad with another 6,000 currently enrolled. 2. 3 Labor Market
According to Clemens (2000) whose main focus was on the status of African health professionals abroad, high wage elasticities of labor supply aggravate critical situations in low income countries, most of which are in Africa. The consensual view of the impact of wages on migration is that higher wages offered in destination countries have a significant impact in the density of health care professionals in the source country. Vujicic et al. (2004) demonstrated that the differential in wages between source and destinations were so large that a small decrease would not affect migration flows.
Individual research by Stalker (2000) and Xaba (2001) also found that although wages were not the definitive factor, it is still a pivotal in the decision making process of whether to migrate. They both suggested that a more competitive salary in developing countries would motivate physicians to stay in the native country. To put it into context, the average monthly salary of a Filipino nurse in the US is $5,760 (US) compared to approximately $175 (US) in the Philippines (Matsuno, 2008). In both examples, the nurses are working in the public sector.
According to Matsuno who focused on nurse migration in Asia, wage differentials between a source and a destination country can be as large as 25 times. It is, however, important to consider that wages may not always be the main reason behind the migration of health workers. Matsuno (2008) reported that while the procedure of migrating to the UK and Ireland may take up to 4 to 6 months, it could take up to 2 years for a nurse to migrate to the US. Matsuno asserted that although salary prospects were often higher in the US compared to the UK, many prospective migrant nurses prefer to go to the UK. . 4 Health Expenditure Health expenditure of governments on health care is the direct result of policy implications. Dacuycuy (2008) noted that a critical variable not included by the ILO in analyzing migration of health care workers is health expenditure. If health expenditure turns out to correlate with migration figures, omitting it would cause significant bias. He asserts that one major implication on developing countries is that they lack the resources that would allow them to translate production possibilities to tax revenues that would finance public health care.
Or (2000) emphasized that it is of economic interest to measure the impact of health expenditures. He developed a model using physician density as a variable to explain changes in mortality rates. Dacuycuy notes that the including per capita health expenditure as a variable may be effective if there is a large portion of physicians in the public sector health workforce. He also mentioned that including the variable may be useful if reductions in per capita expenditure in the source country motivated physicians to migrate abroad. . 5 My Own Contribution Studies by Vujicic et al. (2004), Clark et al (2002) and Arah (2007) have all contributed to the formulation of my empirical analysis. Particularly in the case of Clark, he developed a formulae which determined the probability of individual i migrating from country y to country x. According to Clark, an individual will migrate if; Wf -Wd-C>Z Where Wf – Wd is the wage received in a foreign country minus the domestic wage, respectively. C is the direct financial cost of migrating abroad.
An individual will migrate abroad is the money left is larger than Z, which is the “compensating differential” which is all the non-wage factors in favour of staying in the source country. Clark noted that on average, Z would be positive so all else being equal, individuals prefer to remain in their country of origin. I have expanded on Clark’s original model and the main objectives of this paper are not only to analyse directly the likelihood of health worker migration from source to destination countries, but also the likelihood of migration occurring between source and destination countries themselves.
For example, in my initial specification, I expect to find that the higher the wage offered in the destination countries, the higher the density. By removing the destination countries from the sample, I hope to find out whether migrants are still as likely to migrate to a higher income developing country. The importance of studying the role of both wages and per capita spending is not only to investigate the possible reasons behind out-migration from developing countries, but also the reasons why health workers who are already abroad may/may not be willing to return. 3. Data and Limitations
Table 1. Variables Variables| Symbol (s)| Definition| Occupation| p, n| p: subscript denoting Physician; n: subscript denoting Nurse| Country| i| i: subscript denoting the country of origin from which an individual subsides| Source Country| s| s: subscript denoting figure representative of all source countries| Destination Country| d| d: subscript denoting figure representative of all destination countries| Health Worker Population| aDenip aDenin aDensp aDensn| Number of physicians/nurses in country i, with occupation p or n during year t given per 10,000| Wage | aWageip aWagein aWagesp Wagesn| Wage of physicians/nurses in country i, s or d during year t in nominal U. S Dollars (PPP)| Per Capita Expenditure | aPcsi aPcss aPcsd| Government per capita spending on health care given in nominal U. S Dollars (PPP) in all countries| One reason that very little data exists on the migration of health workers is because of the complex nature of migration itself. There are many reasons behind the decision to migrate and even so, they vary between individuals, cultures and countries. However, we took this as an opportunity to contribute new methodology to the study of health worker migration.
In order to quantify why health workers migrate from, we must analyse the supply and demand mechanism between the source and destination countries. In our model, the supply of health care workers is defined by the measure of doctors and nurses who wish to migrate abroad. The demand side of the mechanism is defined by the amount of medical IMGs that a destination country is willing to receive or accept. Each year, the WHO produce an updates version of their World Health Statistics publication. This paper extracted data using heir statistics (2010) on the density of physicians and nurses as well as the government health expenditure. Cross-referencing this data with a separate WHO (2008) database of health worker wages provided me with the metrics I needed to test my hypothesis. Given the lack of data and research available on health worker migration, I chose to study the effect of wages and per capita health sector expenditure on quantifying the correlates of health worker density from twelve developing source countries to five developed destination countries.
My initial hypothesis was that higher wages and per capita expenditure, the higher the expected density of a given country. For the initial wage data that I collected, I used the method proposed by Vujicic et al. (2004) to study the role of wages in the migration of health care workers from developing countries. In his conclusion, he suggested that if wages were to stay constant, improving working conditions in the developing countries are one possible way of limiting the supply of health workers to higher income countries.
When deciding which countries to include in my model, I reviewed a journal by Dumont (2005) on the mobility of immigrants between OECD countries. The findings indicated that Australia, Canada, France, UK and USA host more IMGs than any other OECD countries. As a result, my full dataset consists of 12 developing and the 5 OECD nations mentioned. The use of density in this paper is a new contribution to health worker migration analysis. In previous papers such as Arah (2007), it was used as an independent variable to explain the absolute number of emigrants on the African continent.
The reason for using the density as the dependent variable is that due to population differentials, it offers the most accurate representation of the amount of health care available in any given country. As density of a global population of health workers is my variable of interest, it is important to get a clear picture of the impact of wages and per capita spending through a direct comparison of density in source countries without the influence of destination countries.
What this will be able to show is the effect my independent variables have on the density of health workers if there was no destination alternative. For example, a comparison of my initial regression with the alternative regression might show that a nurse may be three times more likely to leave the Philippines when wages from destination countries are offered as supposed to when they are not. Borjas (1994) studied different factors that affect the decision to migrate in his publication, Economics of Immigration, in which he concluded that wages are one of the most pivotal factors in choosing to migrate.
Vujicic (2004), through his own research on the impact wages on migration noted that, “the supply of migrants is further complicated by the fact that migration is often a family decision”. In this paper, however, we choose to treat migration as an individual decision in order to simplify the findings. An interesting argument put forward by Matsuno is that the migration of physicians is not directly comparable to that of nurses. His asserted that as interest in the nursing profession in the Philippines grows due to the specific demand from abroad for Filipino nurses, interest in becoming a physician has fallen rapidly.
He reported that 39 medical schools in the Philippines have had to shut down due to a shortage in demand from prospective students. This paper will study the case of nurses and physicians separately to get a more detailed picture of migration. 4. Methodology 4. 1 Graphical Analysis This paper not only aims to present two possible determinants of health worker migration but also the reasons behind considering each variable. The supply curve (Clark et al. , 2002) depicted in Fig 4 shows the relationship between the supply of migrants and wages.
Wf/Wd is the price ratio of foreign to domestic wages and is known as the “wage premium”. The wage premium measures the intention of the individual to migrate. M is the totally amount of health workers that actively pursue migration. According to Clark, there exists diminishing marginal utility for every increasing unit of income. What this graph shows is that beyond a premium threshold, the supply curve becomes almost completely vertical. The empirical implications are that in this range of wage premium, any alterations of the wage premium on a health worker’s intention to migrate abroad.
If the case exists that the cost of moving and the compensating differential are both positive, then in order for the migrant supply to equal zero, domestic wages do not have to match foreign wages. Illustrated on Fig 4, the supply curve intercepting the y-axis above one is the direct result of a positive compensating differential; the preference to remain in the domestic country. Clark is assuming that all else equal, migrants will prefer to stay in their native country.
The purpose of Clark’s basic model is to illustrate that in source countries, there are a variety of policy options available to limit the out-migration of native health care professionals. One option is to increase domestic wage, which can be seen in the model as a movement along the supply curve. However, the effectiveness of the increase in domestic wage depends on the elasticity of supply. How an increase in wages affects the numbers of migrants in source countries will be addressed through data analysis in this paper. Another option according to Clark is to improve working a in the source country.
This would shift the supply curve to the left. Though Clark does not offer a method of measuring this variable, we measure it based on per capita spending on health expenditure. He notes, “Working conditions and wages in the health care sector are the most feasible policy levers”. Fig 4. Supply for migrants Source: Vujicic (2004) 4. 2 Testing the metrics behind physician density in a global population The first dependent variable to be investigated is the density of physicians. There will be three different equations to test for the endogeneity of the dependent variable.
It is important to note than an increase in density of physicians in country x offsets a decrease in density in country y. 4. 2. 1 Denip = B1 + B2aWagei The first linear equation with one regressor will be solely to identify the extent to which changes in the physician wage rate affects physician density; what effect does a 10% increase in the wage offered have on domestic physician density? Wages are measured in $(US) and density (per 10,000). Our hypothesis is that Wageip will be significant at 1% and have an R2 close to 1. 4. 2. 2 Denip = B1 + B2aPcsi
The second equation is concerned with the effect that per capita spending has on density; what affect does a 10% increase in government spending per capita have on the density in 2006? Per capita spending is calculated using an implicit foreign exchange rate so is given in $(US). Our hypothesis for this regression is that the R2 figure will not be as high as the first regression but still enough to explain the large majority of the physician density population. 4. 2. 3Denip = B1 + B2aWageip + B3aPcsi Here, we include both per capita spending and wages into the equation.
The purpose of this equation is to determine how much of physician density can be explained through a combination of wages offered and per capita spending. For example, an increase in wage of 10% may not affect physician density with per capita spending also in the equation. The significance of this regression is that, although the previous two regressions showed the two independent variables to be significant in the density of physicians in a country, they remain unrealistic because they are being considered independent of their effect on one another.
A survey held by the WHO (2003) discovered that according to domestic health workers currently in Uganda, 84% attributed salary and 54% work benefits as the joint two most important factors that would motivate them to stay in Uganda. 4. 3 Testing the metrics behind nursing density The second dependent variable to be investigated is the density of nurses. I will be using three formulae similar to the ones used to investigate physician density. 4. 3. 1Denin = B1 + B2aWagein The equation is investigating the expected density of nurses in a country given the wages offered. The first linear equation with one regressor will e solely to identify the extent to which changes in the physician wage rate affects physician density; what effect does a 10% increase in the wage offered have on domestic nurse density? Wages are measured in $(US) and density (per 10,000). 4. 3. 2Denin = B1 + B2aPcsi The second equation is concerned with the effect that per capita spending has on density; what affect does a 10% increase in government spending per capita have on the density of nurses? Per capita spending is calculated using an implicit foreign exchange rate so is given in $(US) 4. 3. 3Denin = B1 + B2aWagein + B3aPcsi
This equation includes both per capita spending and wages into the equation. The purpose of this equation is to determine how much of nurse density can be explained through a combination of wages offered and per capita spending. For example, an increase in wage of 10% may not affect nurse density with per capita spending also in the equation. The significance of this regression is that, although the previous two regressions showed the two independent variables to be significant in the density of physicians in a country, they remain unrealistic because they are being considered independent of their effect on one another. 4. Testing the metrics behind physician and nurse density in source countries What we are investigating in this part of the methodology is how likely physicians and nurses are to migrate between source countries if there was no allure of the destination countries. If there is limited correlation, what that will show is that either the higher wages offered by the higher income source countries is not high enough for physicians to migrate abroad. 4. 4. 1Densp = B1 + B2Wagesp + B3Pcsis This formula will measure the correlation between wages and per capita health expenditure with the density of the source countries included in my dataset.
The importance of this is to understand whether physicians are willing to move abroad to other source countries or whether an OECD country is the migration destination of choice. 4. 4. 2Densn = B1 + B2Wagesn + B3Pcsis This formula will measure the correlation between wages and per capita health expenditure with the density of the source countries included in my dataset. The importance of this is to understand whether nurses are willing to move abroad to other source countries or whether an OECD country is the migration destination of choice. 5. Results and Empirical Analysis 5. 0. 1 Monthly Wages and Per Capita Spending as a Function of Physician Density Table 2. Results from log Density (Denip) Variables | Denip (1)| Denip (2)| Denip (3)| Wageip| 0. 031***| -| -0. 001| Pcsip| -| 0. 008***| 0. 001***| | -| -| -| Intercept| -0. 152| 2. 62| 3. 02| R2| 0. 6492| 0. 8909| 0. 9329| N| 18| 19| 18| Our first regression involved using the variables Denip and Wageip. Denip is the measure of the density of physicians in country i in 2006 while Wageit are the wages offered to physicians in country i in 2006.
We expect that the higher the wages offered, the higher the density in the given country. The Wageit variable is significant; p-value of 0. 00, F-test at the 5% and 1% intervals. The t-test for Wageit means that the variable is statistically significant from 0. The coefficient is positive at 0. 003 which means that a $1000 per month increase in the wages offered would increase the density of physicians by 3 per 10,000. The R2 figure means that 65% of physician density can be explained through monthly physician wages. Though this figure is relatively high, I had expected higher.
My second regression involved using the variables Denip and Pcsip. Pcsip represents the per capita spend on healthcare in country i during 2006. We expect that the higher spend per capita, the higher the density of physicians. This reason behind this is because healthcare expenditure is directly related to the working conditions in a given country, which is one of the main factors behind the migration decision-making process. The Pcsit variable is proven to be statistically significant; p-value of 0. 00, F-test significant at 5% and 1% intervals. The coefficient is positive at 0. 78 which indicates that the higher the spending per capita, the higher the density in country i. This means that a $1000 per month increase in spending per capita would increase the density of physicians by 78. The R2 figure means that 89% of physician density can be explained through per capita spending, which is higher than monthly wages in the proceeding section. Though I had expected for Pcsip to be statistically significant, I had not expected it to be a larger impact on Denip than Wageip. My third regression involved using all three variables, Denip, Wageip and Pcsip.
We hypothesised that the R2 figure would increase but the coefficients of both the independent variables would decrease as a result of the multiple regression. The p-values suggest that the model is statistically significant. The R2 figure means that 93% of the density of physicians can now be explained through wages and per capita spending, which is what we predicted. However, the coefficient of wage is now -0. 0009 which was unexpected, which means that a $1000 increase in wages would decrease the density of physicians by 0. 9 per 10,000.
Though this was unexpected, this suggests that without an increase in per capita spending, physicians would not want to migrate abroad. These figures back up Dacuycuy’s claim that omitting the per capita spend variable would lead to a large omitted bias. 5. 0. 2 Monthly Wages and Per Capita Spending as a Function of Nurse Density Table 3. Results from log Density (Denin) Variables| Denin (1)| Denin (2)| Denin (3)| Wagein| 0. 035***| -| 0. 020***| Pcsin| -| 0. 025***| 0. 011***| | | -| -| Intercept| 1. 85| 15. 37| 6. 61| R2| 0. 8633| 0. 8473| 0. 8855| N| 17| 17| 17|
In our investigation into nurse density, the first regression involved using the variables Denin and Wagein. Denin is the measure of the density of nurses in country i during 2006 while Wageit are the wages offered to nurses in country i at the same time. We expect that the higher the wages offered, the higher the density in the given country. The Wageit variable is significant at the 5% and 1% intervals. The coefficient is positive at 0. 035 which means that a $1000 per month increase in the wages offered would increase the density of physicians by 35 per 10,000.
This is larger than the coefficient of physicians, but it should be expected because nurses receive a lower income, so a smaller wage differential compared to physicians may not be representative of the percentage increase in wages. The R2 figure means that 86% of physician density can be explained through monthly physician wages. This figure is what we had expected and shows that higher income for nurses does coincide with a higher density of nurses, in country i. My second regression involved using the variables Denin and Pcsin.
Pcsin represents the per capita spend on healthcare in country i during 2006. We expect that the higher spend per capita, the higher the density of nurses in said country. This reason for this is as a result of health expenditure being directly related to the working conditions in a given country, which is one of the main factors behind the migration decision-making process. The Pcsin variable is statistically significant at both 5% and 1% intervals. The coefficient is positive at 0. 025, which means an increase of $1000 per month in spending per capita would increase the density of physicians by 25.
The R2 figure means that 85% of the density of nurses can be explained through per capita spending. Interestingly, this is very similar to the R2 figure of Wagein. This means that a nurse migration figures may be just as influenced by wages as by per capita health expenditure. We can now predict that the R2 value of the following regression will be a similar figure to the first two. My third regression involved using all three variables, Denin, Wagein and Pcsin. The p-values suggest that the model is statistically significant. The R2 figure means that 89% of the density of nurses can now be explained through Wagein and Pcsin.
The value of R2 itself is similar to that of the first two regressions, which is what we had hypothesised. The distinct nature of this regression compared to the same regression with physicians is that both independent variables here are statistically significant. This evidence suggests that nurse density is more predictable from the countries that I have chosen to investigate than physician density. The possible reason behind is possibly that nurses are more likely to migrate between source countries because the wage allure of developing countries are not as large for nurses as for physicians. . 0. 3 Density of Physicians in Source Countries without the Allure of Destination Countries Table 4. Results from log Density (Densp) Variables | Densp| Wagesp| 0. 00132| Pcssp| -0. 00695| | -| Intercept| 1. 269099| R2| 0. 0991| N| 12| Throughout our investigation, we have tried to find possibilities as to why physicians choose to migrate abroad. This regression may not be considered as significant as the first set of regressions on physician density. However, the importance of these figures are underlined by the differences in motivations that source and destination countries offer.
In Vujicic’s (2004) paper on the role of wages in health worker migration, he concluded that wage differentials between source and destinations are so large that a small increase in wages offered would not motivate a health worker to migrate abroad. Therefore, we predicted that the R2 would be extremely low and the coefficient insignificant. Both Wagesp and Pcssp were insignificant as predicted. R2 was also significantly closer to 0 than in the initial regressions. 5. 0. 4 Density of Nurses in Source Countries without the Allure of Destination Countries The results of this regression, shown in table 4, were extremely surprising.
We had predicted that the coefficients would be insignificant and the R2 to be similar to that of the preceding regression. However, Wagesn was statistically significant at the 1% interval and 69% of the density of nurses in source countries could be explained through the two independent variables. This justifies our initial decision to regress physicians and nurses separately instead of under the “health care professional” bracket. Table 4. Results from log Density (Densn) Variables | Densn| Wagesn| 0. 0291***| Pcssn| -0. 0346| | -| Intercept| -1. 149| R2| 0. 6985| N| 12| 5. 0. Variation in Wages Across Source Countries Figure 4 and Figure 5 display the variation in wages across all the source countries in my dataset. The domestic nurse wage offered in Australia as almost 14 times the amount of the wages in Ghana, twice the amount offered in South Africa and 25 times that of Zambia. As for Physicians, wages offered in the USA are 4 times the amount offered in South Africa, 20 times the amount in Ghana and 25 times that of Ghana. As expected, the lowest physician density among these 4 countries is in Zambia with 0. 6 compared to 0. 9 in Ghana, 7. 7 in South Africa and 26. in the USA. Figure 4. Ratio of nurse wages (PPP$US), destination country to source country Source: Vujicic Figure 5. Ratio of physician wages (PPP$US), destination country to source country Source: Vujicic 5. 4. 1 A Case Study of South Africa and Ghana Source: WHO Fig 6. Shows that there is very little difference in the intent to migrate among health care professionals in S. Africa and Ghana. If we assumed that living, working and migration costs were similar in South Africa and Ghana, this would imply that the wage elasticity of their supply of migrants would be close to zero (Vujicic, 2004).
This means that if the supply of migrants in these two countries is relatively inelastic to changes in wages, then even large increase in wages will not affect the amount of health workers wanting to migrate abroad. Vujicic (2004) does note that South Africa is a “holding ground” for health care workers throughout Africa for those wishing to migrate to the UK, Canada and the USA. This means that the data on the density of physicians and nurses in South Africa may be inflated as they intend to migrate again to an OECD country.
One of the reasons for this is because of visa agreements that South Africa have in place with the UK which grants a migrant nurse full registration without having to pass a probationary period or extra training. The result of this is that, as wages and health expenditure do not act as a motivation for African migrants to move to South Africa, even if Ghana offered a higher wage premium, better working and living conditions, South Africa would still hold more health care professionals. This has been illustrated in Fig 7. Fig 7. Supply of Migrants in Ghana and South Africa Source: Vujicic 6. 0 Conclusion
This paper focuses on two of the determinants associated with the migration of health professionals; wages and health expenditure (per capita). It is important to highlight the limitations of the data and the analysis. Through our regression analysis, it became clear that the willingness to migrate, from developing countries to source countries, of health care workers was not as responsive to wage changes as we had hypothesised. It is important to note that our empirical analysis is, at best, only applicable to other source countries that share the same characteristics as those in our dataset e. . density of physicians ranging from 0. 4 to 11. 8 per 10,000. Due to the lack of data available, we cannot assume that other source countries share the same magnitude of difference between foreign and domestic wages. The lack of available data and statistics on health care wages and government expenditure in source countries limited the strength of the conclusions obtained on the affect of wages and per capita spending on the density of health care professionals. This is a concern that is also shared by Arah (2007) and Vujicic (2004).
We believe that in order to better the quality of the investigation into health worker migration, it is vital to go above and beyond the statistical and empirical data, to examine the reasons and experiences of source countries that have been able to implement retention reforms. Though our empirical analysis suggested that there was a high correlation between our variables, alternative ways of decreasing the supply of migrant health care workers must be suggested. Alternative variables could have included the impact of remittances, brain drain and living conditions in developing countries.
However, as is the case with wages and per capita expenditure, there is little data to test for the responsiveness of migrants with respect to these prospective variables. The WHO each year is increasing their scope of data and if this research were to be extended over the course of the next few years, it may be easier to identify the best practices. This paper has especially focused on out-migration from developing countries and the implications on the source country but one of the biggest imitations of the empirical side of the analysis in this paper is that the labour market for health care professionals is unlikely to ever be in equilibrium. This could suggest that out-migration from developing countries could increasingly demand-driven as supposed to supply side policies such as increasing domestic wages and domestic health care expenditure per capita. If this were to be the case, then there could be little that source countries can do for the time being to limit the flow of out-migration from developing countries.
Policy-makers may have to focus exceedingly on demand-side policies such as increased training capacity in developed countries, create a compensatory package to source countries to cover the cost of training. One major limitation of this paper, as well as many other publications on health worker migration is that the sample of destination countries often focuses solely on OECD (developing) countries, which may not correctly represent the data. As we found in our empirical analysis unexpectedly, the migration of nurses between source countries is quite common.
Countries such as the Philippines are regular suppliers of migrant nurses to other developing countries. This suggests that there may have been omitted bias by was of the sample of countries we chose to analyse. However, this is down to lack of data as supposed to lack of research when writing this paper. If we were to extend this research, new data would have to be made available. While I cannot conclude a definitive result, the results I have found have shown that wage differentials between source and destination countries are so large that it is damaging the health care systems of less well developed countries.
Bibliography 1. Arah O, (2007): The metrics and correlates of physician migration from Africa 2. Borjas G. (1994): The economics of immigration. Journal of Economic Literature, 32(4):1667-1717. 3. Buchan J, Parkin T, Sochalski J: WHOLIS database. International nurse mobility: trends and policy implications. Geneva, World Health Organ- ization 2003 [http://whqlibdoc. who. int/hq/2003/ WHO_EIP_OSD_2003. 3. pdf]. 4. Buchan, J. (2006): Filipino nurses in the UK: a case study in active international recruitment. Harvard Health Policy Review 7. 5. Clark X, Hatton T, Williamson J: Where do U. S. mmigrants come from and why? National Bureau of Economic Research Working Paper 8998 2002 [http://post. economics. harvard. edu/faculty/jwilliam/papers/ w8998. pdf]. 6. Clemens, M (2007): Do visas kill? Health effects of African health professional emigration. Center for Global Development Working Paper No. 114. Center for Global Development. 7. Clemens, M. and Pettersson, G. (2007) New data on African health professionals abroad. Working Paper Number 95, Center for Global Development. 8. Dumont, J. (2007) International migration of health professionals: new evidence and recent trends.
Powerpoint presentation at the Workshop on human resources for health and migration: Mobility, training and the global supply of health workers, Sussex. 9. Lorenzo, F, Galvez-Tan, J, Icamina, K and Javier, L. (2007): Nurse migration from a source country perspective: Philippine country case study. Health Research and Educational Trust. 10. Matsuno A. (2008): Nurse Migration: The Asian Perspective. ILO/EU Asian Programme on the Governance of Labour Migration Technical Note 11. Mullan F (2003): The contribution of international medical graduates to the physician workforce in the United States.
Presentation to the Working Group on Human Resources for Health. Geneva: World Health Organization 2003 12. Mullan, F. (2005): The metrics of the physician brain drain. The New England Journal of Medicine. 13. OECD (2007): The medical brain drain: myths and realities. International Migration Outlook:SOPEMI 2007. OECD. 14. OECD (2007): Immigrant health workers in OECD countries in the broader context of highly skilled migration. International Migration Outlook: SOPEMI 2007. OECD. 15. Or, Z. (2000): Exploring the effects of health care on mortality across OECD Countries.
Labour Market and Social Policy—Occasional Papers No. 46. OECD. 16. Stilwell, B, Diallo, K, Zurn, P, Vujicic, M, Adams, O and Dal Poz, M: Migration of health care workers from developing countries: strategic approaches to its management. 17. Vujicic M (2003): Recent trends in the nursing labour market in Can- ada. PhD dissertation. The University of British Columbia, Department of Economics 2003. 18. Vujicic M (2004): The role of wages in the migration of health care professionals from developing countries. [http://www. ncbi. nlm. nih. gov/pmc/articles/PMC419378/pdf/1478-4491-2-3. pdf] 19.
World Health Organization (2003): Migration of health professionals in six countries: a synthesis report. Brazzaville: World Health Organ- ization Regional Office for Africa 2003. 20. World Health Organization (2010): World Health Statistics 2010. WHO Library Cataloguing-in-Publication Data 21. United Nations Development Programme (2003): Human Development Indi- cators [http://www. undp. org/hdr2003/indicator/index. html]. accessed 12 September 2003 22. Yamagata, T. (2007) Securing medical personnel: case studies of two source countries and two destination countries. Institute of Developing Economies, Discussion Paper No. 105.