UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Effects of rice prices on agricultural wage rates in Vietnam Nguyen, Diem Kieu 2006

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2006-0598.pdf [ 2.66MB ]
Metadata
JSON: 831-1.0092733.json
JSON-LD: 831-1.0092733-ld.json
RDF/XML (Pretty): 831-1.0092733-rdf.xml
RDF/JSON: 831-1.0092733-rdf.json
Turtle: 831-1.0092733-turtle.txt
N-Triples: 831-1.0092733-rdf-ntriples.txt
Original Record: 831-1.0092733-source.json
Full Text
831-1.0092733-fulltext.txt
Citation
831-1.0092733.ris

Full Text

EFFECTS OF RICE PRICES ON AGRICULTURAL WAGE RATES IN VIETNAM by DIEM KIEU NGUYEN B.Sc, National Economics University, Vietnam, 1998 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Agricultural Economics) THE UNIVERSITY OF BRITISH COLUMBIA August 2006 © Diem Kieu Nguyen, 2006 ABSTRACT The thesis aims to examine the relationship between agricultural wage rates and rice prices in Vietnam. This study is inspired in part by the fact that agricultural wage rates and rice prices are important factors affecting rural welfare; especially the rural poor who have little land holdings and rely on labor markets for the dominant share of their incomes. Rice farming does not only provide the staple food for over 90 percent o f Vietnamese households, but also absorbs the greatest portion of the rural labor force in Vietnam. It is also o f substantial interest how rural wage rates are actually determined, and to what extent rural versus urban factors are involved. In other words, to what extent are rural and urban labour markets integrated? We hypothesize that changes in rice prices w i l l have a significant impact on agricultural wage rates. Furthermore, many previous studies show that there has been in some periods a significant gap between the domestic and world prices o f rice due to the control over rice exports of the Vietnam government. These regulations have been relaxed since 2001, and therefore the domestic price o f rice would now be more affected by external fluctuations. To answer how rice price changes affect agricultural wage rates, we make use of data at the commune level in Vietnam over the period from 1993-2002. We found that the effects o f rice prices on real agricultural wage rates vary by year and from region to region. Generally, the regression results provide empirical evidence to support that agricultural factors have a sizeable effect on agricultural wage rates in Vietnam. But we found that non-agricultural factors also play an important role in increasing the wage rate for agricultural laborers. The non-agricultural wage rate, urban unemployment rate and the economy's real growth rate were found to have significant, albeit smaller, effects on the agricultural wage rate in Vietnam. II T A B L E O F C O N T E N T S Abstract >> Table of contents - i» List of tables v List of figures vi Abbreviations vii Acknowledgments viii 1. Introduction 1 1.1 Rationale 1 1.2 Problem statement and research questions 2 1.3 Thesis organization 3 2. Theoretical background 4 2.1 Existing theories of agricultural wage determination 4 2.2 Empirical studies on the relationship between agricultural prices and agricultural wage rates in other countries 7 2.3 Some related studies in Vietnam 10 3. Rice production and rice prices in Vietnam 11 3.1 Rice production in Vietnam 11 3.2 Rice prices in Vietnam • 14 4. Agricultural labor market in Vietnam 16 5. Factors affecting agricultural wage rates in Vietnam 18 6. Methodology, model specification, and data sources 21 6.1 Methodology 21 6.2 Model specification 23 6.3 Data description ; 26 7. Empirical results 27 7.1 Regression results for the pooled model 28 7.2 Regression results for each year 31 7.3 Regression results for each region 33 8. Conclusions 38 8.1 Summary and conclusions 38 8.2 Policy implication 40 8.3 Thesis limitations and recommendations for future study 41 Bibliography 43 Appendix 47 iv LIST OF TABLES Table 3.1: Share of paddy area 11 Table 3.2: Distribution of rural household by farm ize : 12 Table 4.1: Share of agricultural employment by region 17 Table 5.1: Independent variables and expected relationship with the agricultural wage rate 21 Table 6.1: Results of Box-Cox test... 24 Table 6.2: Descriptive statistics of variables 27 Table 6.3: Data description by year : 47 Table 6.4: Data description by region with real values 48 Table 7.1: Regression results for equation (2) : 29 Table 7.2: Regression results of equation (3) 49 Table 7.3: Results o f F-tests (1) 30 Table 7.4: Results of F-tests (2) 31 Table 7.5: Regression results for each year .- 32 Table 7.6: Regression results for each region 35 Table 7.7: Regression results by region for year 2002 37 Table 7.8: Regression results,by region with inclusion of interaction terms 50 LIST OF FIGURES . Figure 2.1: Labor market equilibrium under perfect competition 5 Figure 2.2: Labor market equilibrium under monopsony 6 Figure 3.1: Changes in paddy yields by region in Vietnam during 1990-2002 13 Figure 3.2: Real rice prices of paddy by region in Vietnam during 1992-2002 15 v i LIST OF ABBREVIATIONS 1. R R D : Red River Delta 2. N E : Northeast 3. N W : Northwest 4. N C S : North Central Coast 5. S C C : South Central Coast 6. C H : Central Highlands 7. M R D : Mekong River Delta 8. IFPRI: International Food Policy Research Institute 9. G S O : General Statistic Office o f Vietnam 10. O L S : Ordinary Least Squares 11. US$ : United States Dollar 12. V N D : Vietnamese Dong 13. V L S S : Vietnam L iv ing Standard Survey ACKNOWLEDGEMENTS I would like to send my special thanks to my supervisor, Dr. Richard Barichello, for his patient assistance from the very beginning of choosing the thesis topic to the completion of the thesis. I am very grateful to all the other members of my thesis and examining committees Dr. Craig Riddell , Dr. Timothy Beatty, Dr. Katherine Baylis, and Dr. Andrew Riseman for their comments and assistance. I am also thankful to Hanoi Agricultural University, the Ford Foundation and the Center for Educational Exchange with Vietnam, for having offered me an opportunity to pursue a Master's degree in Agricultural Economics at U B C and provided support for my stay in Canada. vin 1. Introduction 1.1 Rationale Vietnam is a poor agriculturally based economy with over 67 percent of the labor force working in the agricultural sector (World Bank, 2004), and rice playing the dominant staple crop in agricultural production. Rice farming not only provides the staple food for almost all Vietnamese households (over 90 percent), but also is the largest agricultural sector and absorbs the greatest portion of the rural labor force in Vietnam. About two thirds o f all Vietnamese households are involved in growing rice'. On average, rice production contributes 67 percent of household income (IFPRI 1996). For the poorest fifth o f the population, rice makes up around 35 percent o f total household spending ( C 1 E M , 2004). Therefore, agricultural wage rates and rice prices are important factors affecting rural welfare, especially for the rural poor who have little land holdings and rely on labor markets for the dominant share of their incomes. Higher rice prices almost certainly have a negative impact on non-farm wage earners. But the effect of higher rice prices on the agricultural workers is ambiguous because they are both producers and consumers of this staple food (Minot and Goleti , 2000). Moreover, the Vietnam government hesitates to completely liberalize rice trade to keep domestic rice prices low because they are concerned that increases in the domestic rice prices would have a negative impact on domestic rice consumers, particularly the poor. It is documented by several studies that the control over rice exports by the government has at times caused a significant gap between the domestic and world prices of rice. But those regulations w i l l be removed when Vietnam joins the W T O , and therefore the domestic price of rice wi l l be expected to be more directly affected by external fluctuations. Thus, it is important to investigate the effects of changes in rice prices on the consumers and producers of this staple food. However, there are broader questions to address in the interest of alleviating rural poverty. A variety of policy tools exist, such as raising agricultural prices by price controls, subsidy or providing border protection. Acknowledge of the determinants of rural wage rates is important in order to use the most appropriate policy tools to raise incomes of the poor. Consequently, the focus of this thesis is to examine how changes in rice prices affect hired agricultural labor through the impacts of rice prices on agricultural wage rates and in the context of a larger model of those wage rates. 1 Markets and Development Bulletin, Nov 2004, Central Institute for Economic Management (CIEM) 1 1.2 Problem statement and research questions How do agricultural wage rates respond to changes in rice prices in Vietnam? This question has remained unaddressed for rural areas of Vietnam, where a growing number of hired laborers primarily depend on labor markets for their livelihood and agricultural wages are an important source of their income. Moreover, there is considerable evidence to suggest that agricultural wages and agricultural prices, i.e prices of staple food grains, are often important determinants of rural poverty 2. Since a significant share of the income of poor hired agricultural laborers comes from selling their labor in the agriculture sector, this would strengthen the poverty-reducing effect of rice price policies i f agricultural wage rates were found to respond proportionally to rice prices. According to Bale and Phung (2001), about 55 percent of the hired laborers in agriculture in Vietnam still live below the World Bank's poverty line. Therefore, this question is an important empirical issue for not only the prospects of hired laborers from rural areas but also for agricultural policy in the context of poverty alleviation efforts in Vietnam. While the responsiveness o f agricultural wages to prices is an important determinant of how to improve the standard of living o f the poor as well as its implications for rural poverty reduction policies, there is relatively little related empirical work on this issue for Vietnam. Hence, the central concern of this thesis is to investigate the impacts of changes in rice prices on agricultural wage rates for hired laborers in Vietnam by estimating the elasticity of the agricultural wage rate with respect to the price of rice. Besides that, we are also interested in finding the answers to the following questions: (2) Does the price of rice have significant impacts on the agricultural wage rate in Vietnam or are other factors more important? (3) What role does the agricultural yield play in determining the agricultural wage rate at those study communes? (4) Is the non-farm wage rate for unskilled labor a significant determinant of agricultural wage rates in Vietnam? (5) How does the urban unemployment rate affect the agricultural wage rate? The answers to those above questions are very important, especially for hired agricultural laborers relying on their wages for income, and policy makers in the effort to reduce poverty in Vietnam. If agricultural wage rates are now dominated by non-farm factors, such as wage rates in other sectors rather than rice prices, then changes in agricultural policy, particularly rice price policy, w i l l not have such an important effect on the agricultural wage rate. Since most of the hired agricultural 2 See Datt and Ravallion (1998); Martin Ravallion (2000) and Felix Paukert (1988) 2 workers belong to the low-income group in Vietnam, it is particularly important for policy makers to know i f and how changes in rice price policies affect those people for the purpose of poverty reduction. Then the question in this context is, should government use agricultural policy, particularly rice price policies, as a poverty alleviation tool, or should it focus its efforts in other policy areas? Or, i f the government uses government policies to raise rice prices, who is likely to benefit? Furthermore, i f the non-agricultural wage rate for unskilled labor is a significant determinant of agricultural wage rates in Vietnam now, then it may be implied that the Vietnamese government can have an important influence on labor earnings for agricultural workers through improving the wage rates for labor in other sectors. In addition, the empirical analysis of this thesis w i l l attempt to test: (1) Whether the effects of rice prices on agricultural wage rates are the same over three study years 1993, 1998 and 2002? And (2) whether the effects o f rice prices on agricultural wage rates vary across regions? Given the regional cost of living, wage differentials across regions have a large impact on inter-regional labor mobility. Thus, i f the null hypothesis that the effects of rice prices on agricultural wage rates are the same across regions is rejected, then it w i l l provide empirical support to confirm the hypothesis put forward by Barichello (2004) that there is a lowidegree of labor market integration among sectors and regions in Vietnam, probably due to a low level of labor mobility. If we fail to reject this null hypothesis or the urban unemployment rate is.found to have a significant negative effect on the agricultural wage rate, it would be inferred that the degree o f labor mobility and urban-rural labor market integration in Vietnam might not be as low as previously suggested. 1.3 Thesis organization This thesis is organized as follows. Following the introduction is section two, which provides an overview of the theoretical issues and previous related studies. Section three and four briefly describe the situation of rice production and rice prices as well as agricultural labor markets and agricultural wages in Vietnam. Factors that may have effects on the agricultural wage rates are discussed in the following section. Data description, econometric model and empirical methodology are introduced in section six. Section seven discusses the empirical results from our estimates of the econometric model. Conclusions, policy implications and recommendations for future study are drawn in the final section. 3 2. Theoretical background 2.1 Existing theories of agricultural wage determination Agricultural wage determination models in the literature range from fixed wage models to others with complete wage flexibility, and the choice of an appropriate model remains debated especially for the labor markets of developing countries (Dart and Olmsted, 1998). There are several alternative theoretical explanations on the process of agricultural wage formation including theory of subsistence wage, theory of efficiency wages, and market theory of wage determination. The subsistence wage theory emphasizes the supply aspects and neglects the demand aspects of the labor market. The theory states that wages cluster around the subsistence level of workers, and changes in the labor supply are the fundamental force that drives real wages to the minimum required for subsistence. The subsistence wage theory is usually assumed to apply in the agricultural sector, but it is not adequate. There are many antecedent empirical studies of agricultural wage determination in developing countries like Bangladesh, Egypt, and India provide consistent evidence that labor demand and labor supply are factors that determined the agricultural wages. For example, several previous studies on agricultural labor markets support the^ view that the demand side does play an important role in agricultural wage formation, especially in the casual employment sector, such as Bardhan (1979); Bardhan and Rudra (1981); Ahmed (1981); Binswanger and Rosenzweig (1984). According to the competitive paradigm, employers wi l l pay hired workers a wage, which clears the market for labor. The efficiency wage hypothesis argues that wages, at least in some markets, are determined by more than simply supply and demand. The fundamental idea underlying the theory o f efficiency wage is that employers may gain some benefits from paying their workers a wage that is higher than necessary. Hence, efficiency wage is a wage, which minimizes labor cost per efficiency unit and which is higher than the wage that employers need to pay to attract a given quantity of labor. The efficiency wage model is based on the idea that labor productivity may depend directly on the wage paid by the employer. Thus a profit-maximizing employer that can choose both the wage and the level of employment wi l l hire all the labor he wants at a real wage satisfying the "Solow condition", which is the elasticity of effort with respect to the wage is unity. 4 Fidel Ezeala-Harrison (2004) stated that under casual employment arrangements the employer would not adopt an efficiency wage policy since there are no incentives to shirk. Therefore, as for casual hired agricultural workers, the appropriate wage and employment level should be based on the labor market forces. In addition, Ahmed (1981) and Hossain (1990) established that the efficiency wage hypotheses were not sufficient in explaining actual wage rates. They found that the agricultural wage rates are responsive to the demand and supply factors and also the prices for the case of casual employment sector. From the view of the market theory o f wages, agricultural wage rates are determined by the demand and supply factors. Different market structures wi l l have different degrees of market power on each side of the labor market. A competitive labor market is a market in which there are many buyers and sellers, so that no single buyer or seller can individually exert any influence on the wage rate. Under the conditions o f perfect competition labor market (each unit of labor is homogeneous, buyers and sellers have perfect knowledge of the market and are perfectly mobile), the wage rate and level of employment are determined by the intersection o f the market supply and demand curves. In particular, in competitive equilibrium the labor market clears. D E* Employment Figure 2.1: Labor market equilibrium under perfect competition 5 If the wage rate is above the equilibrium, the quantity of labor supplied exceeds the quantity demanded and a surplus occurs. In this case, the existence of unemployed workers w i l l be expected to result in downward pressure on the wage rate until an equilibrium is achieved. If the wage rate is below the equilibrium, a shortage of labor w i l l occur. Competition among employers for workers is expected to result in increases in the wage rate until an equilibrium is reached. Generally, in competitive labor markets, adjustment is assumed to be so rapid that the adjustment process itself can be ignored, and workers w i l l be hired at the point when the marginal revenue product is equal to the wage rate. A t equilibrium, where market demand and supply of labor intersect as illustrated in figure 2.1, the equilibrium wage is W * and the equilibrium level of employment is E * . Monopsony occurs in the labor market where there is a single buyer of labor and many sellers of labor. Under this type of labor market, a monopsonist faces the entire market labor supply curve and can offer any wage rate it chooses. The profit-maximizing monopsonist wi l l equate the marginal factor cost of labor ( M F C ) with its marginal revenue product ( M R P ) , which, for any given wage rate, is represented by the demand curve. O u_ B 00 i (U I W* VV0 E 0 E* Employment Figure 2.2: Labor market equilibrium under monopsony In figure 2.2, the optimal level of employment is equal to E 0 that occurs at the point where M R P = M F C . The equilibrium wage rate under this type of labor market, however, must equal the wage rate that is paid to hire E 0 workers. This wage rate is determined by the labor supply curve. Thus, the 6 M R P employer (the monopsonist) w i l l choose to employ at level of employment E 0 and pay a wage rate equal to W 0 . Therefore, the intersection between the M R P and M L C curves determines the level of employment while the labor supply curve determines the wage rate that must be paid in a monopsony labor market. If the M R P curve is the same as the demand curve for the perfectly competitive labor market, it is obvious that the outcome of the monopsony labor market is associated with a lower wage rate and a lower level of employment than under perfectly competitive labor market. That is, E 0 < E * and W 0 < W * as shown in figure 2.2. Besides perfectly competitive and monopsony labor market structures, there are other structures for labor markets such as monopoly and oligopolistic. However, in the real world, it is unlikely that most agricultural labor markets w i l l conform to the extreme types, but are usually characterized by intermediate degrees of market power (Bosworth et. al). Since different market structures wi l l result in a range of different outcomes, those characteristics of the agricultural labor markets in developing countries wi l l be important to modeling the wage rates and employment for agricultural labor. In fact, many previous studies argued that agricultural labor markets in the developing countries justify the competitive assumption, such as Rosenzweig (1980, 1984), Adubai and Delgado (1999). 2.2 Empirical studies on the relationship between agricultural prices and agricultural wage rates in other countries. Our understanding of the agricultural wage determination process in developing countries is less than satisfactory (Hossain, 1990). This results in conflicting theoretical findings on explanations of wage formation in rural agriculture. Thus, the subject of wage determination in rural agricultural labor markets in developing countries has continued to attract debate while remaining totally unresolved (Ezeala-Harrison 2004). While there is no antecedent study of agricultural wage formation in Vietnam, there are a considerable number of studies that analyzed various aspects of agricultural wage formation in Bangladesh, Egypt, Ghana, India, and Indonesia. For the case of Bangladesh alone, there have been at least a dozen studies on this matter. Many authors have argued that increases in the price o f agricultural product, in particular staple food-grain, could lead to higher real agricultural wages in the long run. For instance, Lipton (1984) stated about the wage-price relationship in India that " in the long run, high prices of agricultural product are in the interests of agricultural laborers and low prices against them". Khan (1984) concluded that improved terms of trade for agriculture result in higher 7 real wages. He also showed that real wages responded positively to higher agricultural productivity. However, some studies argue that in countries such as India, Bangladesh and Vietnam low staple food prices have an ambiguous effect upon poor landless laborers (World Bank, 1986; Minot and Goleti , 2000) because their main source of income comes from selling their labor to agricultural production, and rice consumption takes a major share of their expenditure. After having defined the limitations of the past studies on agricultural wage determination in Bangladesh and estimated a new dynamic model, Boyce and Ravallion (1991) pointed out that about 20 percent o f an increase in rice prices was passed on to the agricultural wage rates within the a short term and 50 percent in the long run. They also suggested that lower rice prices help more than lower wages hurt. Their conclusion was "an increase in the price of the staple foodgrain (relative to other goods) wi l l thus immediately reduce the rice purchasing power o f agricultural wages, and this wi l l persist in long-run equilibrium". They also denied the conclusion made by Khan (1984) about gains in agricultural productivity having a positive effect on real agricultural wage rates in Bangladesh. Palmer-Jones and Parikh (1998) re-examined the relationship between rice prices and wages of agricultural laborers in Bangladesh using the co-integration and error correction approach. They found that agricultural wages have strong positive long run relationship with rice prices, which is not much different from the estimates by Ravallion (1994), and by Boyce and Ravallion (1991), and the short run response of wages, estimated consistently with the long run relationship, to rice price is small. However, they did not agree with the argument of Boyce and Ravallion that it takes several years for agricultural wages to adjust fully to a rice price change. A recent study by Rash id (2002) pointed out that there are at least two reasons to be cautious about past studies on the responsiveness of wage rates for hired agricultural laborers to changes in rice prices in Bangladesh. That is, the problem of inconsistency of parameter estimates due to using O L S estimation with non-stationary explanatory variables, and the problem of endogeneity of right hand side variables. Rashid (2002) introduced a new dynamic model to re-investigate and analyze dynamics of agricultural wages and rice prices based on co-integration techniques using data from 1949 to 1999. Contrary to previous conclusions of alarming downward trends in agricultural wages to rice prices by Boyce and Ravallion (1991) and Ravallion (1994), the results suggested that agricultural wages was strongly co-integrated with rice prices until the late 70s, but is no longer a significant determinant of agricultural wage formation in Bangladesh for the period from 1977 to 8 1999. Rashid's study also suggested that lower rice do not lead to lower agricultural wages presumably because demand is substantial for labor outside of the rice sector. For the case of Egypt, a study o f agricultural wages and food prices in Egypt by Dart and Olmsted (1998) indicates that nominal wages adjust slowly, and there is a significant negative initial impact of rising food prices on real wages, though wages wi l l follow up in the long run. A study by Abdulai and Delgado (1999) investigates empirically the factors that influence real agricultural wage rates in Ghana by estimating a dynamic agricultural wage model. They found that a 1 percent change in the relative agricultural price leads to a 0.48 percent change in the real agricultural wage rate in the short run and a 0.83 percent change in the long run. There are also a number o f papers providing evidence to suggest that higher relative prices of staple food w i l l hurt the poor during the agricultural policy reforms in India. Such studies are Saith (1981), Desai (1985), and Sen (1996). The data of real wage rates and relative prices of staple food in India for the period 1958-1994 supports the claim that there is a negative correlation between prices staple food on farm output and real wage rates in India (Ravallion, 2000). However, Ravallion wrote: "the strong negative correlation between the real agricultural wage rate and the relative price of food appears to be spurious; it disappears as soon as one allows for dynamics of real wages and the strongly adverse effects of inflation. It is specious to conclude from these data that policy reforms that entail a sustained increase in the relative price of food are a permanent threat to India's poor." 3 In this section, we provided an overview of three main theories including subsistence theory, market theory and efficiency theory that can be used to explain the determination of agricultural wage. The range of different outcomes from different types of labor markets suggests that market structures wi l l be crucial to modeling agricultural wage rates and employment. However, many empirical studies on agricultural wage rates and food prices show that efficiency wage model is not suitable for the case of casual agricultural labor in developing countries where agricultural labor is abundant. Instead, they proposed to follow the market theory in modeling agricultural wage rates and agricultural prices. Although there are a large number of empirical studies on agricultural wage determination in many developing countries with differences in the literature about the choice of the appropriate model and estimation procedure, almost all of the empirical studies examine the relationship between agricultural wage rates and prices of agricultural product, i.e staple foodgrain, using the reduced 3 Martin Ravallion (2000), p. 363. 9 form of agricultural wage model. Many of these studies also allow for past wage rates and/or rice prices in the previous year to affect the current wage rates, such as Boyce and Ravallion (1991), Dart and Olmsted (1998), Abdulai and Delgado (1999), and Rashid (2002) etc. The empirical results of these studies also provide contradicting findings. While some studies found that higher prices o f staple foodgrain would hurt the poor by lowering agricultural wage rates, many others found a positive relationship between agricultural wage rates and prices o f staple foodgrain. 2.3 Some related studies in Vietnam Although there is some literature on agricultural labor market and agricultural wages for Vietnam, most of the literature is descriptive in nature. Moreover, there has surprisingly no previous work on the modeling of agricultural wage ratesin Vietnam. Brassard (2004) made use of the data from V L S S 1998 to demonstrate the differences of the average daily agricultural wages between different types of agricultural work for main staple food crop. The focus of her paper was placed on the roles of wage and labor regulation in poverty reduction in Vietnam. Gallup (2002) investigated another aspect of the labor market in Vietnam. He used the data from two surveys V L S S 1993 and V L S S 1998, and employed econometric approach to estimate a model of wage on education levels and work experience of wage labor markets in Vietnam. His study emphasizes on how wage employment contributes to the changes in inequality. Edmonds and Pavcnik (2004) studied the relationship between rice price increases and economic activities, such as wage work, work within the household and market work. The emphasis of their study was placed on whether rice price changes due to the liberalization of rice markets in Vietnam are associated with changes in participation in agricultural and non-agricultural activities. They also made use of the data from two surveys V L S S 1993 and 1998 to estimate the relationship between the participation in economic activities and rice prices using a linear regression framework. However, they did not consider how changes in rice prices affect the agricultural wage rates in Vietnam. Barichello (2004) argued that lower agricultural prices are likely to have a depressing effect on agricultural wage rates. He also emphasized that rice is an important crop in Vietnam, and changes in rice price can be expected to influence rural wage rates to a measurable extent due to a low degree of integration between industrial and agricultural labor markets as well as between rural and urban labor markets. Minot and Goleti (2000) suggested, in their theoretical study, that the use of hired 1 0 7 labor is less common in Vietnam than other developing countries, so the effect of rice prices on income via agricultural wage rates is likely to be weaker. Therefore, the main objective of this thesis is to examine the effects of changes in rice prices on agricultural wage rates in Vietnam. We wi l l employ an econometric approach to estimate the relationship between agricultural wage rates and rice prices for hired agricultural labor using a linear regression framework. The thesis wi l l make some contribution to improving the understanding of the adjustment o f the agricultural labor market for policy making particularly in the context o f poverty alleviation in Vietnam. In the next two sections, we wi l l provide a brief overview of rice production, rice prices and agricultural labor market in Vietnam. 3. Rice production and rice prices in Vietnam 3.1 Rice production in Vietnam Vietnam is divided into eight regions, and rice accounts for a majority of the farm area in every region except the Central Highlands. Rice production in Vietnam is characterized by multiple cropping, labor-intensive practices and small size farms (Minot and Goleti, 2000). O f the total cultivated land in Vietnam, the share of paddy area is over 60 percent in 2002. The Red River Delta and Mekong River Delta with barely 15 percent of the national territory account for more than two-thirds of Vietnamese rice production. The proportion o f land allocated to rice in 2002 was around 90 percent in the Mekong River Delta, and more than 74 percent in the Red River Delta. A s a result, only these deltas of eight regions in Vietnam produce rice surpluses. Table 3.1: Share of paddy area Unit: % Year 1995 2000 2002 Year 1995 2000 2002 Whole country 67.3 64.0 61.6 Whole country 67.3 64.0 61.6 Red River Delta 76.4 74.1 74.4 South Central Coast 65.4 61.5 59.4 Northeast 53.8 48.9 47.7 Central Highlands 28.4 16.3 15.8 Northwest 45.7 37.7 36.0 Southeast 34.9 34.9 31.2 North Central Coast 62.6 57.4 58.4 Mekong River Delta 89.0 90.6 89.5 (Source: GSO) 11 Table 3.1 shows that the share o f paddy areas has decreased in every region over the period 1990-2002. Slight reduction in the share o f paddy area occurred in the Red River Delta and Mekong River Delta. The highest decline in share o f paddy area is in the Central Highlands, about 12 percent less. The average labor-use in rice production in Vietnam ranges from 116 to 134 person-days per hectare per season, and is different across regions. The labor intensity o f rice cultivation in the Red River Delta is more than 200 person-days per hectare per season while in the Mekong River Delta is from 85 to 100 person-days per hectare per season. Because o f the large-scale rice production and the increasing problem of landlessness in the Mekong Delta, the use o f hired labor in rice production in this region is much higher. Accordingly, the cash cost o f hired labor for rice production in this region was 16 percent of total cash cost compared with only 5 percent in the Red River Delta (IFPRI, 1996). In general, the farm size in Vietnam is very small. Table 3.2 presents the distribution o f rural household by farm size. In 1995, the average farm size o f an agricultural household is only 0.49 hectares of agricultural land. The proportion o f rural households with no agricultural land is 1.2 percent. About 71 percent o f rural households have less than 0.5 hectares o f agricultural land. Only 11.7 percent o f Vietnam rural households have more than 1 hectare. In 2001, the number o f rural households with no agricultural land has increased, accounting for 4.2 percent o f the total Vietnamese rural households. Table 3.2: Distribution of rural household by farm size Farm size Unit 1995 2001 N o land % 1.2 4.2 Less than 0.2 hectare % 27 24.8 From 0.2 to 0.5 hectare % 44 39.6 From 0.5 to 1 hectare % 16.2 16.4 From 1 to 3 hectares % 10.5 13 • More than 3 hectares % 1.2 2 (Source: GSO) The proportion of rural households with little agricultural land holdings (less than 0.5 hectares) is 74.4 percent. And only 15 percent of rural households have more than 1 hectare of agricultural land in 2001. The average farm size is quite small in the Red River Delta, about 0.25 hectare, and intensely cultivated. The cropping intensity ratio for rice is 1.8, and rice is grown by 95 percent of 12 the rural households and accounts for 18 percent of national rice production. During the period 1990-2002, the increase in rice output of this region was mostly due to improvement in paddy yield (94 percent) and higher crop intensity. Compared with the Red River Delta, the Mekong River Delta has larger average farm size of 1.26 hectares. Rice is large-scale cultivated here, but less intensely. The Mekong Delta accounts for more than half of Vietnamese rice production. Unlike the Red River Delta, the increase in rice output of this region during the period 1990-2002 was because of increase in cultivated area (52 percent from crop intensity and 12.5 percent from paddy land). In other regions, a large majority of rural households grow rice, but the cropping intensities and paddy yields are lower than the two deltas. While cultivated area of paddy increased in most of the regions, the area of paddy in two regions Northwest and South Central Coast reduced during the period 1990-2002. Figure 3.1: Changes in paddy yields by region in Vietnam during 1990 - 2002 0.00 - • — 1. RRD T * ^ - 2 . N E — 3. NW -0— 4. N C C - • — 5 : S C C - A — 6. C H 7 . N E S - O — 8. M R D Over this period, the average annual growth rate of rice production in Vietnam was about 4.9 percent. This result was because of the increases in both paddy yield and total cultivated area of paddy. However, the improvements in the paddy yield as well as the growth rate of paddy output were different across the regions. The paddy yields were increased from 1.7 ton/ha in 1990 to 3.2 ton/ha in 13 2002 for the Northwest and from 2.4 ton/ha in 1990 to 4.5 ton/ha in 2002 for the North Central Coast. The total paddy area increased by 24 percent during the period 1990-2002, but this change was mostly because of cropping intensity since the total agricultural land for growing rice was almost unchanged, increasing only 1 percent. 3.2 Rice prices in Vietnam The effect of changes in agricultural policy on household behavior and welfare is usually through changes in agricultural prices. Given the importance of rice as a source of staple food consumption, farm households' income, and export earnings, rice price policies have direct and powerful impact on the living standards of ordinary Vietnamese people, especially the poor. However, rice policy continues to be the subject of intense debate in Vietnam. The government has to face a trade-off between increasing foreign exchange earnings from rice exports and maintaining low prices for domestic consumers (Minot and Goleti, 2000). Policy makers have given high priority to ensuring food security, and are concerned that an increase in the domestic price of rice would hurt the domestic rice consumers, particularly the poor (Ryan, 1999). Therefore, the Vietnamese government has kept a tight control on the volume of rice exports since it re-entered the international rice market as an exporter in 1989 (Nielson, 2003). Under this policy regime, the government restricted the movement of rice across provinces and controlled the volume of rice exports by setting an annual export quota based on estimates of domestic supply and needs. In the early reform period, the movement of rice from south to north was impeded4, and so there existed the time restrictions on movement of rice across provinces, which create additional costs and prevent spatial arbitrage from bringing the price differences down to the cost of transportation and marketing.. State-owned trading enterprises enjoyed a legal monopoly on rice exports under the national quota. Thus, rice prices in Vietnam were distorted by such marketing restrictions. These government interventions on rice prices may have created bias effects of rice prices on the agricultural wage rates 4 The procedure for a firm in the North to buy and transport rice from the South included: (1) apply for a license from the Ministry of Agriculture and Rural Development, (2) register with the Market Control Department, (3) buy rice at the place determined by VINAFOOD II, (4) register with local market control sections in the South, (5) register at the ports of arrival and departure, (6) register with local market control sections in the North. 14 during this period. However, the government has relaxed some important restrictions on rice exports, such as the removal of the rice export quota as o f M a y 1, 2001, allowing more state-owned and private firms to export rice. In spite of that, the government still maintains some control over rice exports by requiring rice exporters to have its official approval on the volume of rice exports annually. Under this more open rice trade policy, the price of rice in Vietnam is determined by market forces. Some have argued that there is a fiscal incentive to maintain the current system since the rice exporting state-owned enterprises contribute part of their profits to the government (Minot and Goleti , 2000). While others argued that rice price controls in Vietnam are not necessarily bad since, besides public finance considerations, it can have the effect of insulating domestic markets from external fluctuations and thus reducing transaction or adjustment costs (Ghosh and Whalley, 2002). Figure 3. 2: Real price of paddy by region in Vietnam during 1992-2002 1. R R D 2. N E 3. N W K — 4. N C C • — 5 . S C C i— 6. C H —— 7. N E S 8. M R D 0 -I r 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 With the relaxation of domestic and international restrictions on the trade of rice in the 1990s and concurrent with the trade liberalizations, the price of rice increased nearly 30 percent relatively to the consumer price index. However, the domestic real prices of paddy have tended to decrease, from V N D 1,600/kg during 1996-1999 to V N D 1,300/kg during 2000-2002. Decreases in the real price of paddy benefit the net rice consumers but cause a loss for net rice producers. Since the demand and supply of rice and the degree of integration into national and international markets differ across 15 communities, the changes in the rice price also vary regionally (Edmonds and Pavcnik, 2004). The fluctuations of the real paddy prices in eight regions of Vietnam are illustrated in figure 3.2. From this graph we see that the difference in the real prices of paddy between two main rice-producing . regions, Red River Delta and Mekong River Delta, has been getting larger over the period from 2000 to 2002. 4. Agricultural labor market in Vietnam Like many other developing countries, the agricultural labor market in Vietnam is typically characterized by the extremely abundance of agricultural labor and low wage rates. The household registration system has stifled the transfer of labor from agriculture to other sectors and created the distinctions between rural and urban. Under this control on residence permits, rural residents could not freely move to urban areas. Therefore, labor markets for agricultural workers in Vietnam are also characterized by the lack of alternative job opportunities and low labor mobility due to low degree of integration between industrial and agricultural labor markets as well as between rural and urban labor markets (Barichello, 2004). In 2003, approximately 76 percent of Vietnam's labor force are in rural labor, and as much as 90 percent of the poor work in the rural areas. Although there have been structural changes in the labor market under reforms, agriculture is still an economic sector that employs a vast proportion of Vietnam's labor force. The total employment in 2000 was 36.2 million, of which agricultural employment was 22.66 million^. Of eight regions, the mountainous areas in the North have the highest share of agricultural employment, e.g Northeast (81.4 percent) and Northwest (86.9 percent). In the two largest rice production deltas, agriculture sector provides over 60 percent of total employment. Since it includes HCM City, the biggest city of Vietnam, the Southeast region has the lowest share of agricultural employment of 31.4 percent in 2000. Declines in the employment share of agriculture for each region over period 1997-2000 are shown in table 4.1. ^ r - ^ ^ ^ ^ „ d reused b , U S D e p a n n - of Labor, Bureau of ,n,er„a,io„a, Labor Affairs and U.S Embassy in Hanoi, 2002. 16 Table 4.1: Share of agricultural employment6 by region Regions 1997 1998 1999 2000 Whole country 65.8 65.3 64. J 62.6 Red River Delta 68.3 65.3 64.3 63.2 Northeast 82.4 83.3 82.5 81.4 Northwest 89.6 89.5 87.9 86.9 North Central Coast 75.6 76.0 73.7 71.3 South Central Coast 62.9 64.6 62.6 60.7 Central Highlands 77.1 74.8 76.2 76.7 Southeast 34.3 33.9 32.6 31.4 Mekong River Delta 63.8 63.6 63.0 61.5 (Source: GSO) In the early 1990s, most of the Vietnamese farmers were self-employed, and only a small number of them worked for wages as hired agricultural laborers. However, the portion of hired agricultural laborers in Vietnam's rural labor markets has been increasing as the problem of landlessness is growing, especially in the Mekong River D e l t a 7 . The proportion of landless households in the Mekong Delta has increased from less 0.7 percent in 1994 to 5.7 percent in 1998 and even more serious recently (Minot and Goleti , 2000). In poorer regions, the proportion of landless peasants is as high as 35 percent. Most of these landless peasants and near landless (with little land holdings) work as tenant fanners or hired laborers for more prosperous landowners. In the 1996-farm survey, hired labor represented about 5 percent o f total labor use in the Red River Delta, but from 33 to 39 percent in the Mekong River Delta (IFPRI, 1996). From the 2001-rural household survey, the number o f 6 Agricultural employment share in this table includes employment in agriculture, forestry and fishery. However, the employment shares of forestry and fishery are very small. In 2002, the total labor in forestry is 64.5 thousand and in fishery is 1,346 thousand while in agriculture is 25,326 thousand. 7 According to the 1994 Agricultural Census, about 1.2 percent of the agricultural households had no land. But from the 2001 rural household survey, the proportion of landless households was 4.2 percent (419,704 households) and the proportion of households having less than 0.2 hectares of agricultural land was 24.8 percent (2,503,233 households). Also see http://www.hrw.org/reports/l997/vietnm/Vietn97d-01 .htm, and http://www.wsws.org/articles/1999/janl999/viet-i23.shtml. 17 hired workers in agriculture was about 2,059 thousand, accounting for 9.3 percent of total agricultural laborers in Vietnam. Hired agricultural laborers usually work in the crop fields doing such jobs as preparing land, planting crop, tending or harvesting crop. The majority of hired agricultural workers in Vietnam are employed on a casual or temporary basis. Unl ike permanent laborers, who are hired to work year-round, casual laborers are hired for a shorter time (usually during harvesting and planting seasons). In addition, unionization of the agricultural labor force and minimum wage legislation is almost completely not effective for hired agricultural workers, and thus the wage rates for this type of laborers in Vietnam are influenced greatly by market forces. Casual hired labor is predominately paid by daily rate at the end of each day worked (75 percent) or on a task basis (22 percent) with very low rates. Consequently, most of waged workers in agriculture live on pr below the poverty line. From the V L S S survey in 1993, the mean daily wage of agricultural laborers was V N D 9,270 (less than US$ 1/day) while the mean paddy price was almost V N D 1,092 per kilogram. In 1998, the average agricultural wage rate was doubled, V N D 18,660/day (less than US$ 2/day) when the average price of paddy was V N D 2,259 per kilogram, higher than the mean price of paddy in 2002. On average an agricultural worker was paid a nominal daily wage of V N D 19,970 (still less than US$ 2/day) in 2002 while the price for a kilogram of paddy was V N D 1,755. This information can be found from table 6.3 in the appendix. Table 6.4 provides deflated values of mean daily agricultural wage rates for each of the seven regions in Vietnam over three study years. On average, an agricultural worker in Red River Delta, North Central Coast and Central Highlands received a real daily wage of approximately V N D 13,000. However, the real price o f paddy in the Red River Delta was second highest, around V N D 1,459 per kilogram. O f seven regions, agricultural labor in Northern Uplands was paid the lowest with a mean daily wage of V N D 10,650 while the real price of rice in this region was the highest of almost V N D 1,478 per kilogram. In contrast, the average of real daily wage for agricultural labor in the Mekong Delta was the highest of,approximately V N D 17,000 when the real price of paddy was the lowest of V N D 1,163 per kilogram. 5. Factors affecting the wage rates of hired agr icul tural laborers in Vie tnam < Although non-farm wage earners almost certainly wi l l not benefit from higher rice prices, it is not clear how increases in rice prices affect hired agricultural laborers. This is because their main source 18 of income is from selling labor in the sector, but they also spend a large share of their income consuming this staple food. Since rice production in Vietnam is labor-intensive, higher rice prices would be expected to increase the demand for labor and hence causing the real wage rates to rise (positive effects). But for hired agricultural laborers, who are net rice consumers, higher rice prices may also have negative effects on them because an increase in rice prices wi l l result in a temporary decline in the food purchasing power of their wage rates, and thus reduces real agricultural wage rates. Thus, the effect of rice prices on the real agricultural wage rates ambiguous and whether the induced positive effects upon real agricultural wage rates could outweigh the negative effects wi l l be answered through our empirical investigation. Besides price of rice, there are other factors from both demand and supply sides of the labor market that have different effects on determining the agricultural wage rates of hired labor in Vietnam. If the demand effects are stronger, then it w i l l result in higher wage rates. On the other hand, stronger supply effects may depress the wage rates. In this section, we w i l l discuss some main factors other than rice prices that may affect the wage rates o f hired agricultural laborers in Vietnam. * A g r i c u l t u r a l yield Higher agricultural yield may be associated with increasing capital investment, better technology, or higher productivity of agricultural labor, etc. If improvements in agricultural yield are due to higher productivity of agricultural labor, that may help to increase the demand for labor and push the real wage rates up. Therefore, it is likely that a rise in the agricultural yield would be expected to have a positive effect on real wage rates. On the other hand, i f an increase in agricultural yields is mostly due to capital investments or technological changes that involve a shift from labor intensive to labor saving technologies, then it may lead to a decline in the wage rates for agricultural labor. In this case, a negative relationship between agricultural yield and agricultural wage rate is very likely. * Agr i cu l t u r a l labor force On the supply side, the size of agricultural labor force is an important factor that may force the agricultural wage rates down. In a country like Vietnam, where there is lack of alternative jobs for agricultural workers, also there are some restrictions on migration, the abundant supply of casual agricultural labor would remain in the agricultural sector. A n abundance of the agricultural labor 19 force may result in weakening their bargaining position, and therefore, that may lead to lower wage rates. We thus expect a negative relationship between wage and the size of agricultural labor force. * Non-agricultural wage rates for unskilled labor The availability of alternative jobs and wages paid to unskilled workers in other sector may also affect the agricultural wage rates of casual hired labor. If they are offered better pay in other sectors, agricultural hired workers w i l l want to move to non-agricultural sector to work. That wi l l reduce the supply of agricultural labor, and so that can make the agricultural employers be wil l ing to offer agricultural hired labor higher wage rates. If the wage rates o f unskilled labor in non-agricultural sectors are low, then the wage rates for hired agricultural labor may be also low. * Urban unemployment rates Abundant rural labor often goes to towns or cities to look for non-agricultural work. If the urban unemployment rates are high, then there is less employment opportunity for rural agricultural labor. This wi l l bring pressures on the labor market. Consequently, agricultural wage rates may be driven down since too many suppliers of labor while there is insufficient employment. On the other hand, i f the unemployment rates in urban are low, it w i l l be easy for rural agricultural labor to find non-agricultural jobs. That can reduce the supply of agricultural labor in the rural labor market. Therefore, lower urban unemployment rates may lead to higher agricultural wage rates. * GDP per capita G D P per capita may be also a factor that affects the agricultural wage rates. If G D P per capita increases, it means that the economy is growing well , and thus more jobs are created .in both agricultural and non-agricultural sectors: There wi l l be more demand for labor. These effects may help to raise wage rates. It can be argued that the inclusion of variable non-agricultural wage rates for unskilled labor would reflect the effects of changes in G D P per capita on the agricultural wage rates. However, that is only true i f non-agricultural wage rates were perfectly measured. So we included variable G D P per capita to help pick up what is left off that non-agricultural wage rates variable did not capture 20 It is plausible to argue that variable farm size should be added into the model because this factor may affect the demand for hired labor. But it is quite usual in Vietnam that small farm-sized owners also hire labor, especially during the peak season. Moreover, owning a large farm does not always mean that the owner would have to hire more labor since he/she could use the agricultural services provided by machinery. In addition, our analysis is at the commune level; therefore inclusion in the model of the variable, averaged communal farm size would not provide us useful information on the effects of this variable on the agricultural wage rates. Table 5.1: Independent variables and expected relationship with the agricultural wage rate Independent variables Description Expected sign P P R Price o f paddy +/_ N A W Non-farm wage rates of unskilled labor + Y L D Y i e l d of paddy +/_ G D P G D P per capita L + A L B Agricultural labor force U E R Urban unemployment rates -6. Methodology, model specification and data description 6.1 Methodology Generally, the agricultural wage rates are determined by the intersection between supply and demand for agricultural labor under the assumption of a competitive rural labor market. The agricultural labor demand and supply functions are of the form: D t = a | W t + Zb,mX,mt + U , t • S, = a 2 W t + Ib2nX2nt + U 2 t where D and S are labor demand and labor supply respectively; W is the wage rate; X ' s are exogenous variables capturing the impacts of supply and demand sides of the labor markets; U ' s are disturbances; and a's and b's are parameters; subscript (t) indicates t t h observation; n and m indicate the number of independent variables. 21 On the demand side for agricultural labor, daily agricultural wage rates, agricultural product price (rice price), agricultural yield, and G D P per capita are assumed to influence the amount o f labor that agricultural producers want to hire. The aggregate demand for agricultural labor may therefore be specified as: D, = a , . D W G , + b „ . P P R | , + b 1 2 * Y L D + b n . G D P + U„ On the supply side of agricultural labor, agricultural wage rates, wage rates for unskilled labor working in other sectors, urban unemployment rates, and agricultural labor force are assumed to affect the decision to participate in the agricultural labor market. The aggregate supply of agricultural labor may be specified as: S, = a 2 . D W G , + b 2 1 . N A W + b 2 2 . U E R + b 2 3 , A L B + U 2 t In the labor demand and supply equations, D W G is short for daily agricultural wage rates and replaced letter W in the general demand and supply functions. Actual data on agricultural employment are not available, so we can not estimate demand-supply relationships for the agricultural labor market. Based on the availability of data, and also the characteristics of the agricultural labor markets in Vietnam, we employ the market theory of labor demand and labor supply to explain the relationship between agricultural wage rates and rice prices, which is used by authors of many previous studies. In particular, we adopt a partial equilibrium approach and use comparative static analysis to examine the agricultural wage rates when we bring supply and demand functions together under the competitive market structure of the agricultural labor in Vietnam. The agricultural labor market is assumed to be competitive, and then the market-clearing agricultural wage rate can be determined in a reduced form as follows: D W G , = a + £ p k t X , + e, where X s are independent variables including the price of rice, agricultural yield, G D P per capita, wage rates for unskilled labor working in other sectors, urban unemployment rate, and agricultural labor force. In order to answer the research questions, we wi l l formulate a model of agricultural wage determination for hired laborers in Vietnam, and make use of the data from three surveys to empirically estimate the wage model. 22 Box-Cox regressions wi l l be used for testing the functional form of the model. This transformation covers a wide variety of relationships that can exist between the response and the regressor variables. The Box-Cox procedure wi l l estimate X and/or 0, which wi l l be used to transform either or both dependent variable and independent variables, from the data. Formally, we define the response to be W, where W is given by: f Y - 1 W = < L n Y ifX = 0 After the variables are transformed based on the Box-Cox transformation, we wi l l use O L S method to estimate the relationship between agricultural wage rates and rice prices in Vietnam. 6.2 M o d e l specification Given the previous discussion, to identify the relationship between rice prices and agricultural wages, our model of agricultural wage w i l l include variables capturing factors of both sides of the labor market, labor demand and labor supply. We wi l l estimate the determinants of agricultural wage rates with multiple linear regressions, where the dependent variable W is the daily wage rates of hired agricultural workers and X ' s are independent variables. Parameters P, a and A, are to be estimated. Subscript (i) indicates the i" 1 community, and subscript (k) is the k"' independent variable. DWGi = p 0 + S p k X k i + S i * B o x - C o x regressions for testing the functional form of the model The results of Box-Cox regressions show that a semi-log specification is rejected whereas transformation of variables on both sides would significantly add to the regression at the 0.01 level. The output of the likelihood-ratio tests on three standard functional form specifications also say that the linear and multiplicative inverse specifications are both strongly rejected. A natural log specification can not be rejected at the 0.01 significance level. 23 Table 6.1: Results of Box-Cox test Restricted LR statistic P-Value Test Ho log likelihood chi2 Prob > chi2 lambda = -1 -2500.019 200.480 0.000 lambda = 0 -2399.860 0.160 0.686 lambda = 1 -2479.196 158.840 0.000 Thus our agricultural wage model wi l l be of log-log functional form. In particular, our modified model with transformed variables wi l l have L n W , the natural log of daily wage rates of hired agricultural workers, to be the dependent variable and lnX ' s are the natural log o f independent variables. L n D W G , = Bo + Z p k l n X k i + u, (1) To reflect the fact that the population may have different distributions over time periods or across regions, we wi l l include dummy variables to control for the time and regional effects. L n D W G , = Bo + Z p k l n X k i + Z a M D M + IVJRJ + U I (2) where D m is year dummies with m = 1998 of 2002, and D | 9 9 8 = 1 for year 1998 and 0 otherwise; D2002 = 1 for year 2002 and 0 otherwise, and year 1993 is chosen as the base year; R, are the regional dummies and Rj = 1 i f region = j , and Rj = 0 i f region ^ j , for j = 2, 3,.. . , 7. The seven regions and their orders are as follows 8 : Red River Delta (region 1), Northern Uplands (region 2), North Central Coast (region 3), South Central Coast (region 4), Central Highlands (region 5), Southeast (region 6), and Mekong River Delta (region 7). After estimating equation (2) we wi l l want to test whether the time and regional factors have important effects on the agricultural wage rates in Vietnam. The null hypotheses of these tests are: 8 Based on the most recent classification, Vietnam is comprised of eight administrative regions, but we will follow the previous classification of seven regions for our analysis and regressions. This is because the sampling of VLSS 1993 and 1998 were based on this, and there may be insufficient observations in the divided regions for significant statistical results if the eight regions are utilized in our analysis. 24 (1) Ho (2) Ho (3) Ho Ctl998 — 0^2002 72 = Y3 = 74 = 7s = Y6 = Y? = 0. ai998 = CC2002 = Y2 = Y3 = 74 . = Ys = Y6 = 77 = 0. Running a single regression for pooled data means that we assume the parameters of some independent variables in the model to remain constant for all sub-samples of observations. To test whether this assumption is true for the case of how rice prices affect agricultural wage rates over time and across regions in Vietnam, we wi l l interact each year and region dummy with variable rice price. Therefore, our modified model wi l l be: L n D W G , = Po + SPk-i l n X k . u + I a m D m + EyjRj + SA,m ( D M . InPPRj) + (Rj. InPPR,) + u, (3) The null hypotheses to test are that all the estimated coefficients on the interaction terms between year dummies or/and region dummies with variable rice prices are simultaneously equal to zero. In particular, we wi l l test the following: (4) Ho : ^-1998 = ^-2002 = 0 (5) H 0 : 4>2 = <|>3 = <t>4 = <|>5 = d>6 = 4>7 = 0 (6) H 0 : A.,998 = A-2002 = fa = fa = 4>4 = <t>5 = <K> = <t>7 = 0 We also want to investigate the immediate and short-term effects of rice price policies on the real agricultural wage rates. To find out how agricultural wage rates respond to changes in rice prices during the same year, we wi l l estimate the elasticity of wage rates with respect to rice prices for each study year. Region dummies are added to the wage model of each year to take into account the regional effects. In equation (4), t indicates the year. LnDWGi(t ) = po + E p k lnX k i ( t) + E Y j Rj + u,(t) (4) With regard to regional differences, we also want to look at the relationship between rice prices and agricultural wage rates for each region in the short-run. In this case, year dummies w i l l be included in the regional models to control for the time effects. These models are as follows, where r indicates the r t h region. LnDWGj(r) = p 0 + Z p k lnX k i ( r ) + Z a m D m + e,(r) (5) 25 6.3 Data description To estimate the relationship between agricultural wage rates and paddy prices in Vietnam we make use of data from three main sources: VLSS-1993 , VLSS-1998 and VLSS-2002 carried out by the General Statistic Office (GSO). The first survey was conducted between September 1992 and October 1993. The second nationwide survey began in December 1997 and finished in November 1998. Both V L S S 1993 and 1998 have three different surveys including commune survey, price survey and household survey, but V L S S 2002 does not include the commune and price surveys. In addition, we also use the supplementary data from previous studies, the G S O ' s website and the M O L I S A ' s labor census. The data on agricultural wage rates as well as non-farm wage rates of unskilled labor are collected from the commune surveys. Agricultural wage rates are communal daily averages computed from daily wage rates paid for four types of work, such as preparing land, planting crop, tending crop, and harvesting crop. Non-farm wage rates for unskilled labor are obtained by taking the average of daily wage rates received by hired laborers for doing simple services such as letter delivery, mining, construction, and transportation (loading and unloading goods), etc. These variables are measured in thousand Vietnamese Dong. Real wage rates are obtained by using both spatial (regional) price indices and Consumer Price Indices (CPIs, with base year 1994). The deflation factors are the products of regional price indices and yearly CPIs. By doing this we can take into account the discrepancy in the costs of living among regions. The spatial price indices are obtained from the V L S S and the CPIs are collected from various volumes of Vietnam Statistics Yearbooks, published by G S O . The data on rice prices of 1993 and 1998 are collected from the price surveys at commune level. Rice prices are the averages of paddy prices from three observed times. Since there was no price survey in V L S S 2002, rice prices for this year are obtained by taking average of paddy prices received by all households that sold a portion of their output in each commune. Nominal rice prices are used in all regressions except for the regression in column 2 of table 7.1. When estimating the wage model using real rice prices, the price deflators are the products of regional price indices and yearly CPIs with base year 1994. Data on G D P per capita, agricultural labor force and urban unemployment rates are at provincial level since we do not have this type of data at commune level. We took the provincial data on these 26 variables and applied to each commune that belongs to the province. Communal averages of paddy yield are used to represent the agricultural productivity since rice production is a dominant crop and plays an important role in Vietnam's agriculture. The unit of this variable is the amount o f paddy per sao, which is commonly used in the country and equivalent to 360m 2 . Table 6.2: Descriptive statistics of variables9 Variable Unit of measurement Obs Mean Std. Dev. D W G '000 V N D 927 19.12 5.75 PPR V N D / k g 927 1745.67 315.30 N A W '000 V N D 927 24.79 17.78 Y L D Kg/sao (360 m 2) 927 164.72 39.71 G D P '000 V N D 927 5153.43 5101.20 A L B '000 person 927 521.47 256.45 U E R Percent 927 5.75 0.80 Data description of variables for the full sample is shown in table 6.2. The descriptive statistics o f variables used in the regression analysis for each year and for seven regions are presented in table 6.3 and 6.4 from the appendix. 7. Empirical results The data are analyzed using the ordinary least squares. A l l regression results in this thesis are estimated with Stata. White's correction of standard errors wi l l be applied for the possibility of heteroskedasticity. Since our data is collected from three years for seven regions in Vietnam, we wi l l first examine the relationship between agricultural wage rates and rice prices in Vietnam by running regressions for pooling cross sectional data. Then we also want to look at how rice prices affect agricultural wage rates in each year and each region by running separate regressions by year and by region. 9 In this table, the means of the first three variable's are nominal values. 27 7.1 Regression results of the pooled model In this section, we use the full sample consisting of 927 communities to estimate the short-run elasticity of real agricultural wage rates with respect to rice prices. The White's general test statistic for presence of heteroskedasticity when using real values is x2(94) = 172, hence the null hypothesis o f homoskedasticity can be rejected at the 0.01 significance level. For the regression with nominal prices of rice, the null hypothesis of constant variance is also rejected at the 0.01. Therefore, we w i l l estimate model (2) using White corrected standard errors. The regression results of equation (2) using real prices o f rice, which are obtained by using the products of the regional price indices and yearly CPIs as deflation factors, are presented in column (2) of table 7.1. The reported R 2 is approximately 38.1 percent. The signs o f all explanatory variables are consistent with our expectation. A l l estimated coefficients are statistically significant except for variable rice prices. Overall, the short-run elasticity of real agricultural wage rates with respect to rice prices is 0.16. This result implies that real agricultural wage rates increase by 0.16 percent i f rice prices rise by '1 percent. This finding also suggests that the positive effects of an increase in rice prices tend to outweigh the negative effects on the real agricultural wage rates during the data period 1993-2002. The effects o f non-farm wage rates are found to be significantly positive although not large. Real agricultural wage rates increase by 0.05 percent i f the wage rates for unskilled labor in other sector are raised by 1 percent. Furthermore, the impacts of the agricultural labor force variable and the urban unemployment rates are negative and statistically significant. A decline of 1 percent in the agricultural labor force or in the urban unemployment rates w i l l increase the agricultural wage rates by 0.072 percent or 0.125 percent respectively. These findings indicate that non-agricultural sources of labor demand and the integration of urban and rural labor markets in Vietnam are important factors to help boost the real agricultural wage rates. 28 Table 7.1: Regression results for equation (2) Ln(DWG) Regression with real rice prices Regression with nominal rice prices 1 0 Ln(PPR) 0.160 0.118 (0.1) (0.103) LnfNAW) 0.050*** 0.05*** (0.014) (0.014) Ln(YLD) 0.050* 0.052** (0.03) (0.029) Ln(GDP) 0.071*** 0.074*** (0.023) (0.023) Ln(ALB) -0.073*** -0.072*** (0.017) (0.017) Ln(UER) -0.121** -0.125** (0.056) (0.055) D1998 0.251*** 0.200*** (0.06) (0.097) D2002 0 ] 9 7 * * * 0.148*** (0.044) (0.062) Region 2 -0.234*** -0 217*** (0.033) (0.031) Region 3 0.023 0.037 (0.031) (0.032) Region 4 0.053* 0.047* (0.029) (0.029) Region 5 -0.018 -0.04 (0.039) (0.033) Region 6 0.103*** 0.093*** (0.036) (0.035) Region 7 0.230*** 0.216*** (0.034) (0.032) Intercept 0.898 1.175* (0.702) (0.716) R 2 0.381 0.379 Values in parenthesis are White's corrected standard errors. *** Significance at 1% level or better ** Significance at 5% level or better, * Significance at 10% level or better. The nominal rice prices here are simply raw prices with no deflation whatsoever. 29 Over the study period, agricultural yield is found to have a positive effect on the real agricultural wage rates. The short-run elasticity o f the agricultural wage rate with respect to this variable is 0.052 and significant at the 0.1 level. The positive coefficient of variable G D P per capita suggests an increase in the real agricultural wage rate by 0.074 percent as a result of the economic growth. Theses following tests are to verify the importance of time and regional factors when investigating the relationship between the real agricultural wage rates and rice prices in Vietnam during the period 1993-2002. The test results suggest that we are unable to reject the null hypotheses that regional and/or time effects do not influence the real agricultural wage rates at the 0.01 level. Table 7.3: Results of F-tests (1) Hypothesis F-statistic P-value (1) H 0 : 0C|998 = CC2002 = 0 F ( 2 , 912)= 11.35 Prob > F = 0.000 (2) H 0 : y 2 = 73 = 74 = Ys = Ye = Y? = 0 F ( 6 , 912) = 23.43 Prob > F = 0.000 (3) H 0 : a,998 = CX2002 = Y2 = 73 = 74 = Ys = Ye = Y7 = 0 F( 8, 912) = 24.09 Prob > F = 0.000 There are slight differences in estimated coefficients of equation (2) when using real prices of rice and nominal prices but signs of estimated coefficients are basically the same. The elasticity of agricultural wage rates in terms of rice prices become smaller in the regression with nominal prices. To examine the effects of the rice prices over three years and across seven regions, O L S is used to estimate equation (3). The regression results of this model are shown in tables 7.2. The F-statistic computed from the data for testing the null hypothesis that the effects of rice prices on agricultural wage rates are the same over three years is F(2, 904) = 2.95. The probability for the calculated F-statistic to exceed the critical value is 0.081. This implies that the null hypothesis can be rejected at the 0.1 significance level, but not the 0.05 level. Similarly, the test for the null hypothesis that the effects of rice prices on agricultural wage rates do not vary across seven regions in Vietnam yielded F(6, 904) = 2.33, which is above the critical value, suggesting a rejection of the null hypothesis at the 0.01 significance level. The hypothesis that the elasticity of agricultural wage rates with respect to rice prices does not change over time nor differ across regions also can be rejected at the 0.01 significance level. 30 Table 7.4: Results of F-tests (2) Hypothesis F-statistic P-value (4) Ho : A,i998 = X,2oo2 = 0 F ( 2 , 904) = 2.95 Prob> F = 0.081 (5) H 0 : <)>2 = <(>3 = <|>4 = <t>5 = <t>6 = <t>7 = 0 F( 6, 904) = 2.33 Prob > F = 0.002 (6) H 0 : A.,998 = A.2002 = <P2 = 4>3 = 4>4 = <|>5 = 4>6 = 4>7 = 0 F( 8, 904) = 2.14 P r o b > F = 0.001 7.2 Regression results for each year A s we mentioned before, there are differences in sub-sample sizes when estimating the relationship between agricultural wage rates and rice prices separately for each year. The number of observations in sub-samples 1993 (65 observations) and 1998 (68 observations) are very small compared with the size of sub-sample 2002 (794 observations). Moreover, with the restrictions on rice trade policies until 2001, that may have created some degree of endogenity in the rice prices and thus caused the biasedness in the estimated effects of rice prices on agricultural wage rates in 1993 and 1998. For those reasons, we wi l l therefore place our analysis focus on the estimated wage model of 2002. The regression results for each year are shown in table 7.5. The model of 1993 has the highest R 2 (52.4 percent), and R 2 of regression models for 1998 and 2002 are 45.1 percent and 30.9 percent respectively. The F test for the null hypothesis that all the regression coefficients are simultaneously equal to zero is rejected in all three models. The estimated elasticities of agricultural wage rates with respect to rice prices in 1993 and 1998 are -0.267 and -0.308 respectively, but statistically insignificant. These findings appear to suggest the depressing immediate effects of increases in rice prices outweigh the positive effects upon real agricultural wage rates in 1993 and 1998, but these results are, after all , insignificant. They may imply a slow response of agricultural wage rates to changes in rice prices that is typical of the wage adjustment process in most labor markets. Or it is likely that the estimated elasticities could have been biased because of the endogeneity in the rice prices due to the restrictions on rice trade set by the Vietnam government during this time. As mentioned before, several important regulations on rice exports were relaxed by 2001, including the removal of rice export quota system and allowing more firms to export rice. With these changes in the rice policy, rice prices have been determined more by market forces, and so we argue that the endogeneity of this variable would not be a problem for our regression results in 2002. Consequently, the estimated coefficient on the rice price variable is found 31 positive and significant in 2002. This suggests that a 1 percent increase in the rice price wi l l result in a 0.234 percent increase in the real agricultural wage rate for this year, a relatively large effect. Table 7.5: Regression results by year Ln(DWG) Year 1993 Year 1998 Year 2002 -0.267 -0.308 0.234*** Ln(PPR) (0.368) (0.334) (0.091) 0.283*** 0.011 0.042*** L n f N A W ) (0.128)' (0.109) (0.013) -0.129* -0.192 0.036 •Ln(YLD) (0.07) (0.24) (0.034) 0.155 0.091 0.053** Ln(GDP) (0.119) (0.129) (0.024) 0.027 0.083 -0.098*** L n ( A L B ) (0.088) (0.112) (0.017) -0.033 -0.3 -0.061 L n ( U E R ) (0.17) (0.337) (0.06) 1.722 5.291 0.771 Intercept , (2.502) (3.364) (0.723) Rz 0.524 0.451 0.308 Number of observations 65 68 794 Al l three regressions in this table include region dummies, but their coefficients are not reported here. Values in parenthesis are White's corrected standard errors". *** Significance at 1% level or better ** Significance at 5% level or better * Significance at 10% level or better Non-farm wage rates of unskilled labor are an influential factor affecting the real wage rates for hired agricultural labor in 1993 and 2002. The positive coefficients of this variable in these years are more economically and statistically significant than that of year 1998. A n increase in the wage rates " These will be applied to all tables of regression results hereafter. 32 for non-farm unskilled labor by 1 percent would help to raise agricultural wage rates by 0.283 percent in 1993 and 0.042 percent in 2002. The impact o f agricultural yields on agricultural wage rates was found to be positive in 2002 and negative in 1998, but insignificant for both years. Only in 1993 was an improvement in the agricultural yield found to have a significant contribution to labor demand in increasing the real wage rates for hired agricultural laborers. This may be explained by the land reforms occurred in Vietnam during this time. The land reform functioned as an incentive for increasing physical productivity of labor. In the later years, it is more likely that the agricultural yield is improved by more capital investment in fertilizers or chemicals. Contradicting our earlier expectation that agricultural labor force would have a negative effect on agricultural wage rates; this factor was found from the regression results to be positive but it was insignificant in 1993 and 1998. However, in 2002, a decline in the agricultural labor force or economic growth helped raise real agricultural wage rates significantly. The estimated elasticity of agricultural wage rates with respect to this variable in 2002 is -0.098 and significant at the 0.01 level. A s we expected, the urban unemployment rates have negative effects on the agricultural wage rates. However, the estimated coefficients on this variable are statistically insignificant in all three years. 7.3 Regression results for each region In this section first we wi l l look at the relationship between agricultural wage rates and rice prices in every region in Vietnam. Then we place our focus on the regression results by region for year 2002. Because the number of observations for the Central Highlands is too small for meaningful regression results, we wi l l combine this region with the Northern Uplands. These two regions are both remote mountainous areas with large ethnic minority populations and are pockets of poverty. We also combine two coastal regions into one. These combined regions are named the Uplands and Central Coast. When controlling for the time effects, the regression outcomes by region in table 7.6 show that the effects of rice prices on agricultural wage rates vary across regions. A n increase by 1 percent in rice prices wi l l lead to higher wage rates in the Central Coast and Southeast regions by 0.502 percent and 0.453 percent respectively. However, when using aggregated data of three years the empirical findings are quite surprising for the two main rice producing deltas of Vietnam. While rice prices have a modest positive impact on real agricultural wage rates in the Mekong River Delta, they have negative, but highly insignificant, effects for hired laborers in the Red River Delta. In the 33 Northern Uplands that has the largest proportion of poor households, and Central Highlands where the poverty is most severe, the estimated elasticity o f real agricultural wage rates with respect to rice prices is -0.317, but again insignificantly. The impacts of non-agricultural wage rates for unskilled labor are found to be positive and significant in all regions except for the Uplands. These results indicate that higher wage rates for unskilled labor working in the non-farm sectors wi l l attract laborers out of the agricultural sector, thus bidding up the wage rates for hired workers in this sector. Agricultural yields are found to have negative effects on agricultural wage rates in the Red River Delta and Central Coast, but the elasticities of agricultural wage rates with respect to agricultural yields in the Uplands and Mekong Delta are 0.11 and 0.103 respectively, and significant at the 0.1 level. The negative and significant coefficient of the agricultural labor force variable from the regression results indicates that a decrease in the agricultural labor force wi l l help increase the real agricultural wage rate in most regions of Vietnam. The effects of urban unemployment rates on agricultural wage rates are varying across regions. Increases in the urban unemployment rates wi l l depress the real wage rates of hired agricultural laborers in three regions having large cities, namely Red River Delta, Central Coast, and Southeast. These findings suggest that in those regions there is some connection between rural and urban labor markets. A low rate of the urban unemployment w i l l attract rural labor to move out of agricultural sector, and thus reduce the agricultural labor force, which can exert a depressing influence on agricultural wage rates. The coefficient of this variable is found to be positive but insignificant in regions that do not have large cities. When estimating the agricultural wage rate-rice price relationship for each region using only data in 2002, the regression results seem to reflect better our prior expectation. The regression results are shown in table 7.7. In the Red River Delta, non-agricultural wage rates are the most important factor positively affecting the wage rates of agricultural laborers. The estimated elasticity of agricultural wage rates with respect to rice prices is 0.316. Reducing the agricultural labor force and lowering the urban unemployment rates wi l l help to improve the wage rates for agricultural labor of this region in 2002. 34 Table 7.6: Regression results for each region Ln(DAW) Red River Delta Uplands Region Central Coast Region Southeast Region Mekong River Delta Ln(PPR) -0.107 -0.317. 0.502** ' 0.453*** 0.079 (0.264) (0.127) (0.167) (0.174) (0.137) Ln(NAW) 0.058*** -0.002 0.092*** 0.06** 0.033* (0.024) (0.054) (0.037) (0.042) - (0.019) Ln(YLD) -0.096 0.110* -0.037 0.021 0.103* (0.07) (0.058) (0.079) (0.078) (0.056) Ln(GDP) 0.106** 0.363*** -0.014 0.034 -0.072 (0.052) (0.087) (0.083) (0.028) (0.109) Ln(ALB) 0.001 -0.117*** -0 127*** -0.097** -0.069* (0.045). (0.043) (0.034) (0.05) (0.04) Ln(UER) -0.131 0.104 -0.24 -0.332 0.011 (0.089) (0.204) (0.232) (0.258) (0.112) D1998 0.414* 0.333 0.113 0.203 0.104 (0.257) (0.146) (0.19) (0.182) (0.13) D2002 0.284* 0.296 0.108 0.174* 0.138 (0.176) (0.19) (0.15) (0.093) (0.12) _I intercept 2.781 1.602 0.236 -0.249 2.473** (1.816) (1.725) (1.459) (1.233) (1.127) R 2 0.147 0.425 0.270 0.414 0.083 Number of observations 260 109 204 84 270 For the Uplands region, increases in rice prices and agricultural labor force w i l l significantly reduce the real agricultural wage rates in this year. The estimated coefficients on variables non-agricultural wage rates and urban unemployment rates are insignificant. This implies a weak link between rural and urban labor markets as well as a low integration between the agricultural labor market with labor markets in other sectors of this region. However, G D P per capita appears to have a significant positive impact on improving the wage rates for agricultural labor here in 2002. There is not much difference in the 2002-regression results except for non-agricultural factors become less influential in the Central Coast regions. Both non-agricultural and agricultural factors were found to have significant effects on 2002-agricultural wage rates in the Southeast. This can be explained by higher degree of labor market integration and labor mobility in this region, where Ho Chi M i n h City, the biggest economic center of Vietnam locates. In the Mekong River Delta, the effects of rice prices and non-agricultural wage rates on agricultural wage rates are positive and significant at the 0.1 level. However, urban unemployment rates did not have a substantial influence on the wage rates of agricultural labor in this region. This finding may reflect the fact that disadvantageous conditions of transportation, infrastructures, and education in this region are hindering the link between rural areas with towns or economic centers. Moreover, the regression results in table 7.8 do not provide evidence to support an increasing degree of labor mobility and labor market integration in Vietnam during this period. The estimated coefficients on the interaction terms between year dummies and variables rice prices, non-agricultural wage rates and urban unemployment rates are mostly insignificant. Only the estimated coefficients on the interaction term [year2002*Log(rice price)] in the Mekong River Delta and the interaction term [year2002*Log(urban unemployment rates)] are significant at the 0.01 level. 36 Table 7.7: Regression results by region for year 2002 Ln(DWG) Red River Delta Uplands Region Central Coast Southeast Region Mekong River Delta Ln(PPR) 0.316 -0.591*** 0.539** 0.523*** 0.244* (0.203) (0.174) (0.265) (0.198) (0.138) Ln(NAW) 0.059*** 0.03 0.05 0.075* 0.034* (0.024) (0.05) (0.033) (0.045) (0.019) Ln(YLD) -0.138 0.035 -0.043 0.043 0.09 (0.142) (0.096) (0.084) (0.089) (0.058) Ln(GDP) 0.057 Q 312*** -0.13 0.038 -0.197* (0.064) (0.103) (0.102) (0.027) (0.116) Ln(ALB) -0.037 -0.139*** -0.160*** -0.136*** -0.059 (0.041) (0.042) (0.037) (0.053) (0.043) Ln(UER) -0.036 0.056 -0.15 -0.716** 0.145 (0.096) (0.206) (0.321) (0.317) (0.123) Intercept 0.566 4.814*** 0.981 0.105 2.242* (1.792) (1.779) (1.934) (1.376) (1.279) R 2 0.077 0.263 0.167 0.218 0.067 Number of observations 227 82 175 69 241 8 Conclusions 8.1 Summary and conclusions i The main objective of this thesis is to investigate the relationship between agricultural wage rates and rice prices in Vietnam in the context of a broader model of agricultural wage rates. This study is inspired by the fact that agricultural wage rates and rice prices are important factors for the rural poor in Vietnam. Furthermore, the Vietnam government is concerned about the impact o f rice market liberalization on the poor because they claimed that increases in rice prices would have negative effects on the domestic rice consumers, particularly the poor. The relationship of agricultural wage rates and rice prices is analyzed in a reduced form, where the wage rate is determined by the supply and demand of labor in the agricultural sector under a competitive market structure. We examined this wage-price relationship in each o f the three study years, by each region, and pooled cross sections. Therefore, the answers to our research questions vary by year and from region to region. Overall, the regression results provide fairly strong empirical support for our general hypothesis that higher rice prices significantly increase real agricultural wage rates in Vietnam in the short term. The short-run estimated elasticity of real agricultural wage rates with respect to real rice prices is between 0.16 (not significant) and 0.24 (highly significant for year 2002) using data from all regions. Within several regions the elasticity becomes larger (e.g., 0.5) and more significant (but with one case being negative 0.6). This hired agricultural laborers who are net consumers of rice are not expected to benefit from an increase in rice prices because the budget share of income spent on rice by the poor in Vietnam is over 35 percent 1 2. Furthermore, it is also evident from the empirical results that when the abundance of agricultural labor rises or where there is high urban unemployment, downward pressure is exerted on agricultural wage rates in Vietnam within the period 1993-2002. 1 2 According to Ravallion (1990), the necessary and sufficient condition for net rice consumers and net suppliers of agricultural labor to benefit from a small increase in rice price is that the elasticity of agricultural wage rates in terms of rice prices should be greater than the budget share for rice. 38 Other agricultural variables such as the rice yield and the agricultural labour force were sometimes significant. The results were variably significant for the yield, although always of positive sign, but the size of the agricultural labour force always had a significantly negative effect on the agricultural wage rate, as expected. Similarly to the agricultural variables, the non-agricultural variables exerted a significant influence on agricultural wage rates. However, despite their tendency toward statistical significance, their influence, as measured by their elasticities, was small. The non-agricultural wage rate was a significant determinant of agricultural wage rates in all years and most regions, but its elasticity was usually about 0.05. The level of G D P was less consistent in its effect but when it was significant it had a similar elasticity as the non-agricultural wage rate. The urban unemployment rate performed mostly as expected (a negative sign) but it was more typically not significant. Depending on the regression results by year or by region the elasticity of agricultural wage rates with respect to rice prices varied. When studying how wage rates respond to changes in nominal rice prices in the same year, it is possible for the failure of real wage to catch up with changes in rice prices. That might be the reason for a decline, although insignificant, in real agricultural wage rates for 1993 and 1998. However, the regression results show that increases in rice prices raised real wage rates significantly for agricultural labor in 2002. And as mentioned earlier, the 2002 results are far more important than those for the two earlier years because that year's survey included more than ten times the number of observations. Non-farm wage rates for unskilled labor are found to be a significant factor affecting the real agricultural wage rates in both 1993 and 2002. Rice prices were found to have significant positive effects on agricultural wage rates of hired labor in the Central Coast and the Southeast. But higher rice prices have a significant negative impact on the wage rates for hired agricultural labor in the Northern Uplands and Central Highlands, the two poorest regions of Vietnam. Non-agricultural wage rates and urban unemployment rates did not appear to have any influence on the agricultural wage rates in those two poorest regions. This may be because of the lack of access to job market information and education, and limited infrastructure in the regions. The extreme abundance of agricultural labor, persistence of high underemployment rates together with weak physical and/or social links with major cities or economic centers wi l l make disadvantages in economic and employment for agricultural laborers in these poor regions. 39 The empirical findings on the relationship between agricultural wage rates and rice prices in the two main rice producing deltas were surprising. A 1 percent change in rice prices would increase the agricultural wage rate by 0.079 percent in the Mekong River Delta, but it had an insignificant effect on the agricultural wage rate in the Red River Delta. Non-farm wage rates for unskilled labor turn out to be a significant determinant of agricultural wage rates for hired labor in the Red River Delta, Central Coast, and Mekong River Delta. The size of the agricultural labor force has a significantly negative influence on the wage rates in all regions except for the Red River Delta. In three regions having large cities, namely Red River Delta (Hanoi), Central Coast (Da Nang), and Southeast (Ho Chi M i n h City), lower urban unemployment rate helped raise the real wage rates of hired agricultural laborers, but insignificantly. These findings suggest that in those regions there is some integration between rural and urban labor markets, but not yet at a high level. The mobility of labor in Vietnam, especially in Hanoi, H C M city and Da Nang, has been improved due to the relaxation of numerous regulations on migration that prevailed in the past, improvements in infrastructures and access to information about job opportunities. However, policy-induced barriers to geographic mobility of labor still exist (Nguyen Thang, 2002). This explains the substantial differences in the wage elasticity among regions given that the discrepancy in the regional cost of l iving was already taken into account and by the insignificant relationship between the urban unemployment rate with agricultural wage rates. 8.2 Policy implications Effects of rice prices on real agricultural wage rates vary in the course of time and across regions, therefore changes in the rice price policies should take into account this fact. In the short-run, higher rice prices would not increase the real wage rates for hired agricultural laborers in the Northern Uplands and Central Highlands, where most of the poor live. This suggests that, the negative effect of an increase in rice prices upon real agricultural wage rates outweighed the positive effect in these two poorest regions of Vietnam 1 3 . However, higher prices of rice would raise real agricultural wage rates for hired labor in the southern and central regions. Overall, agricultural yield does have a positive effect on real agricultural wage rates for hired laborers within the period 1993-2002. Higher agricultural yield can improve labor earnings for hired agricultural workers in the Uplands and 13 Thus, this result may suggest that in the short-run poor hired agricultural laborers in Vietnam who are net rice consumers will be better off with lower rice price policies. 40 Mekong River Delta, however it is not a significant determinant of agricultural wage rates in other regions. These empirical findings show that agricultural factors generally do not play the most important role in improving agricultural wage rates. Moreover, the factor with the most depressing effects on agricultural wage rates is the agricultural labor force. Therefore, agricultural policies such as rice price policies may not be the most important influences in improving Vietnam's agricultural wage rate. Instead, the empirical results indicate the relatively important roles of non-agricultural factors and policies in improving the wage rates of agricultural laborers. Non-agricultural employment provides many different sources of labor demand from other sectors of the economy. Without barriers to labor mobility and high integration level of labor markets between agricultural and other sectors, increases in non-farm wage rates of unskilled labor w i l l attract laborers out of agricultural sector, thus helping raise real agricultural wage rates substantially. Therefore, policies that prevent labor mobility or tend to suppress non-farm wage rates for unskilled labor wi l l depress the real wage rates for hired labor in agricultural sector. There are always more job opportunities for both skilled and unskilled labor in big cities, towns, and economic centers. Improving labor mobility as well as the integration level between rural and urban labor markets by creating better physical and social accesses to satellite cities, towns and economic centers wi l l therefore help to narrow the wage gap between agricultural and non-agricultural work as well as wage differences across regions in Vietnam. 8.3 Thesis limitations and recommendations for future study There are several limitations of this thesis as follows. (1) Since we use data on the average daily agricultural wage rates to estimate the model, we can not study the seasonality and gender effects on agricultural wage rates. (2) We only study daily wage rates of hired labor in agriculture because it is very difficult to calculate exactly how much a self-employed agricultural laborer earns for a working day. However, we recognize that the labour component of the income of a self-employed farmer is a. reflection of the market wage rate for hired labour. (3) The rice prices in the model are domestic prices that were received by the farmers for ordinary paddy 1 4. Besides ordinary rice, there are some other types of rice planted in Vietnam such as glutinous rice and specialty rice but the proportions of total output and grown area for these types of paddy are relatively small compared with those of the 14 paddy is unprocessed rice with husk 41 ordinary paddy. (4) It would be more accurate to study the response of agricultural wage rates to changes in rice prices in the long run and under a dynamic context, but we only can do our study for the short term and in a static model because of the limiting availability o f time-series agricultural wage data in Vietnam. (5) The full sample is obtained by combining data from three years of the Vietnam L iv ing Standard Surveys ( V L S S ) , but the number of observations obtained from the surveys in 1993 and 1998 are relatively small compared with the number of observations in 2002. (6) For some variables we have to use provincial level data instead of commune level because there are no such commune level data available. This is the case for G D P per capita, urban unemployment rates, and the provincial agricultural labor force. (7) L ike most of.the empirical studies on agricultural wage rate determination, the estimated empirical model wi l l be in a reduced form. Wage adjustment processes are typically sluggish; therefore there are large differences between the long-run and short-run wage responses. A l lowing for current and lagged effects o f rice prices and wages to affect current agricultural wage rates wi l l not only make our modeling o f the wage adjustment more realistic, but also enables us to disentangle the short-run effects from the long-run effects of changes in rice prices on agricultural wage rates. Since our model is in the reduced form, we can not tell much about the induced changes in labor demand and labor supply upon wages. Estimating the wage-rice price relation in a reduced form instead of estimating simultaneous equations may potentially encounter problems with endogeneity. Moreover, seasonal and gender effects are important in determining agricultural wage rates, but we did not study these factors in our model. Therefore, estimating a dynamic model that allows lagged and current effects of rice prices and wages on current agricultural wage rates using time series or household data wi l l better explain the relationship between rice prices and agricultural wage rates for hired labor in Vietnam than a cross-sectional wage model. These elements of modeling and measuring the determinants of agricultural wage rates are priorities for future work. 42 Bibliography 1. Abdulai , Awudu and Delgado, Christopher L . [1999] " A n empirical investigation o f short and long run agricultural wage formation in Ghana", M S S D discussion paper N o . 37, IFPRI. http://www. ifpri. org/divs/mtid/dp/papers/dp3 7.pdf 2. Ahmed, Iqbal [1981] "Wage determination in Bangladesh Agriculture", Oxford Economic Papers, V o l . 33, No.2, pp. 298-322. http://www.jstor.Org/view/00307653/di015349/01p02235/0 3. Akerlof, George A . and Yellen, Janet L . [1986] "Efficiency wage models of the Labor Market", Cambridge University Press. 4. Bales, S. and Phung D. Tung [2001] "Sectoral changes and poverty", in Haughton D. and Nguyen Phong (eds.), "L iv ing standards during an economic boom: The case o f Vietnam", Statistical Publish House, Hanoi, pp. 47-61. 5. Bardhan, P . K . [1979] "Wages and unemployment in a poor agrarian economy: A theoretical and empirical analysis", Journal o f Political Economy, V o l . 87, No.3.1, pp. 479-500. 6. Bardhan, P .K . and Rudra, A . [1981] "Terms and conditions of Labor contracts in Agriculture: Results of a survey in West Bengal 1979", Oxford Bulletin o f Economics and Satistics, V o l . 43, N o . l , pp. 89-111. 7. Barichello, Richard [2004] "Agricultural development and poverty reduction in East As ia : The impact of O E C D agricultural policies", in Kiichiro Fukasaku et. al (eds.), "Pol icy coherence towards East Asia : Development challenges for O E C D countries", 2005 O E C D Publishing, Chapter 5. http://www. oecd. org/dataoecd/2 7/61/3185 7434.pdf 8. Benjamin, D. and Brandt, L . [2002] "Agriculture and income distribution in rural Vietnam Under economic reforms: A tale of two regions", Working paper N o . 519, The Wil l iam Davidson Institute at the University of Michigan Business School. http://www.chass.utoronto.cci/ecipa/archive/UT-ECIPA-BENJAMIN-02-01.pdf 9. Binswanger, H.P and Rosenzweig, M . R (eds.) [1984] "Contractual arrangements, employment and wages in rural labor markets in As ia" , Yale University Press, pp. 1-40. 10. Bosworth, D., Dawkins, P., and Stromback, T. [1996] "The economics of the Labor Market", Addison Wesley Longman. 11. Boyce, Jame K . [1989] "Population growth and real wages of, agricultural laborers in , Bangladesh", Journal of Development Studies, V o l . 25, No.4, pp. 467-489. 43 12. Boyce, Jame K . and Ravallion, Martin [1991] " A dynamic econometric model of agricultural wage determination in Bangladesh", Oxford Bulletin of Economics and Statistics, V o l . 53, No.4, pp. 361-376. 13. Brassard, Caroline [2004], "Wage and labor regulation in Vietnam within the poverty reduction agenda", Working paper, National University of Singapore. http://www. spp. nus. edu. sg/docs/wp/wp58.pdf 14. Datt, Gaurav and Omlsted, Jennifer [1998] "Agricultural wages and food prices in Egypt: A governorate-level analysis for 1976-1993", F C N D discussion paper N o . 53, IFPRI. http://www. ifpri. org/divs/fcnd/dp/papers/dp5 3.pdf 15. Desai, G . M . [1985] "Trends in rural poverty in India: A n interpretation o f Dharm Narain, in Mellor , J. and Desai, G . M . (eds.), "Agricultural change and rural poverty", John Hopkins University Press. 16. Edmonds, Eric V . and Pavcnik, N . [2004] "Trade liberalization and the allocation of labor between households and markets in a poor country", Working paper, Dartmouth College. http://www. dartmouth. edu/% 7Eeedmonds/hhspec.pdf 17. Ezeala-Harrison, F. [2004] "Analysis of wage formation process in rural agriculture", The Journal of Development Areas, V o l . 38, N o . l , pp.79-92. http://inuse.jhu. edu/journals/journal_of_developing_areas/v038/38.1 ezeala-harrison.pdf 18. Ezeala-Harrison, F. [2005] "Two-tier wage systems in rural agriculture: Evidence from Indonesian micro data", Southwestern Economic Review, V o l . 32, N o . l , pp.1-12. http://www.ser.tcu.edu/2005/SER2005%20Ezeala-Harrison%201-12.pdf 19. Gallup, John L . [2002] "The wage labor market and inequality in Vietnam in the 1990s", W B policy research working paper N o . 2896. http://papers.ssrn.com/sol3/papers.cfm? abstract_id=636259 20. Gosh, Madanmohan and Whalley, John [2002] "Are price controls necessarily bad? The case of rice in Vietnam", Working paper, University of Western Ontario. http://www.ssc.nwo. ca/economics/econref/workingpapers/researchreports/wp2002/wp2002_l .pdf 21. Hossain, Akhtar [1990] "Real wage determination in Bangladesh agriculture: an econometric investigation", Applied Economics, V o l . 22, pp. 1549-1565. 22. Khan, A . Rahman [1984] "Real wages of agricultural workers in Bangladesh", in A . Rahman Khan and Eddy Lee (eds.), Poverty in Rural Asia , I .L.O Bangkok, Thailand. 23. Kompas, T. [2002] "Market reform, productivity and efficiency in Vietnamese rice production", Working paper, Australian National University. http://ideas.repec.Org/p/eab/tradew/587.html 44 24; Lipton, M [1984] "Urban bias revisited", Journal of Development Studies, V o l . 20, No.3 , pp.139-166. 25. McCarty, Adam [1999] "Vietnam's labor market in transition", Paper presented at Law and labor market regulation in As ia conference at University o f the Philippines, Nov 12, 1999. http://129.3.20.41/eps/lab/papers/0110/0110001.pdf 26. Minot , Nicolas and Goleti , Francesco [2000] "Rice market liberalization and poverty in Vietnam", IFPRI research report, Washington, D . C . http://www. ifpri. org/pubs/abstract/114/rrl 14.pdf 27. Nielsen, Chantal P. [2003] "Vietnam's rice policy: recent reforms and future opportunities", Asian Economic Journal, Vol .17, N o . 1, pp. 1-26. http://www.foi.kvl. dk/upload/foi/docs/publikationer/working%20papers/2002/8.pdf 28. Nguyen Thang [2002] " A note on goods and labor mobility in Vietnam", unpublished paper. http://www. crefa. ecn. ulaval. ca/develop/thang.pdf 29. Oxfarm Great Britain and Vietnam Farmer's Association [2005] "Landless and near-landless farmers in two provinces of Tra V inh and Dong Thap: Problems and Solutions", Project report. http://www. oxfamgb. org/eastasia/documents/viet_landless_report. doc 30. Palmer-Jones, Richard [1993] "Agricultural wages in Bangladesh: What the figures really show?", Journal of Development Studies, V o l . 29, No.2, pp. 277-300. 31. Palmer-Jones, Richard and Parkh, Ashok [1998] "The determination of agricultural wage rates in Bangladesh", Journal of Agricultural Economics, V o l . 49, N o . l , pp. 111-133. 32. Rashid, Shahidur [2002] "Dynamics of agricultural wage and rice price in Bangladesh: A re-examination", M S S D discussion paper N o . 44, IFPRI. http://www. ifpri. org/divs/mssd/dp/papers/mssdp44.pdf 33. Ravallion, Martin [1989] "Do price increases for staple foods help or hurt the rural poor?", W B working paper. 34. Ravallion, Martin [1990] "Rural welfare effects of food price changes under induced wage responses: Theory and evidence for Bangladesh", Oxford Economic Papers, V o l . 42, No.3 , pp. 574-585. http://www.jstor.Org/view/00307653/di015384/01p0332d/0 35. Ravallion, Martin [1994] "Agricultural wages in Bangladesh: What do the figures really show?", Journal of Development Studies, V o l . 31, No.2, pp. 334-345. 45 36. Rosenzweig, Mark R. [1980] "Neoclassical theory and the optimizing peasant: A n econometric analysis of market family labor supply in a developing country", Quarterly Journal of Economics, Vol .95 , pp.31-55. http://ideas. repec. org/a/tpr/qjecon/v94yl980ilp31-55. html 37. Ryan, G . James [1999] "Assessing the impact of rice policy changes in Viet Nam and the contribution of policy research", Impact Assessment Discussion Paper No.8. International Food Policy Research Institute (IFPRI), Washington, D C . http://www.ifpri.org/impact/iadp08.pdf 38. Saith, A . [1981] "Production, prices and poverty in rural India", Journal of Development Studies, Vol . 19, pp. 196-214. 39. Sen, A . [1996] "Economic reforms, employment and poverty: Trends and options", Economic and Political Weekly, V o l . 31. 40. Tran Dinh Thao and Nguyen Ngoc Que [2004] "Rice industry in Vietnam" Research report. 46 Appendix Table 6.3: Data description by year Variable 1993 1998 2002 Obs Mean Std.Dev. Obs Mean Std.Dev. Obs Mean Std.Dev. D W G 65 9.27 3.29 68 18.67 6.63 794 19.96 5.04 P P R 65 1091.54 153.35 68 2259.41 315.18 794 1755.22 219.09 N A W 65 9.77 4.35 68 17.68 6.07 794 26.63 18.39 Y L D 65 112.44 36.20 68 142.04 29.45 794 170.94 36.82 G D P 65 1985.72 1232.66 68 4172.50 4394.86 794 5496.75 5254.43 A L B 65 453.15 238.93 68 471.87 217.95 794 531.31 259.79 U E R 65 5.69 1.27 68 6.60 0.85 794 5.68 0.70 47 Table 6.4: Data description by region with deflated values Variable Region 1 Region 2 Region 3 Region 4 Region 5 Region 6 Region 7 Obs Mean Obs Mean Obs Mean Obs Mean Obs Mean Obs Mean Obs Mean DWG 260 13.25 94 10.87 99 13.41 105 14.51 15 13.54 84 16,86 270 17.26 PPR 260 1485.37 94 1511.17 99 1345.74 105 1342.58 15 1297.34 84 1309.05 270 1183.65 N A W 260 17.96 94 14.61 99 18.85 105 19.67 15 15.37 84 20.49 270 20.75 Y L D 260 185.00 94 145.26 99 168.02 105 159.18 15 155.80 84 144.21 270 159.77 GDP 260 4856.09 94 3116.35 99 3555.75 105 4290.88 15 3364.47 84 12066.98 . 270 5018.71 A L B 260 545.88 94 451.88 99 805.83 105 374.91 15 530.27 84 299.32 270 543.53 UER 260 5.90 94 5.91 99 5.95 105 5.71 15 5.30 84 5.76 270 5.51 48 Table 7.2: Regression results of equation (3) Ln(DWG) Coefficients and Robust Standard errors Ln(PPR) -0.333 (0.314) L n ( N A W ) 0.051*** (0.014) L n ( Y L D ) 0.045 (0.029) Ln(GDP) 0.063*** (0.024) L n ( A L B ) -0.081*** (0.017) L n ( U E R ) -0.096* (0.055) D1998 0.385 (2.828) D2002 -3.137* (1.812) R2 -1.797 (2.631) R3 -0.756 (2.425) R4 -3.998** (1.859) R5 -0.095 (1.429) R6 -2.902* 91.639) R7 1.005 (1.459) D1998*Ln(PPR) -0.007 (0.392) D2002*Ln(PPR) 0.473* (0.258) R2*Ln(PPR) 0.214 (0.360) R3*Ln(PPR) 0.107 (0.336) R4*Ln(PPR) 0.570** (0.260) R5*Ln(PPR) 0.010 (0.199) R6*Ln(PPR) 0.425* (0.230) R7*Ln(PPR) -0.113 (0.204) Intercept 4.483** (2.219) Values in parenthesis are White's corrected standard errors *** Significance at 1% level or better ** Significance at 5% level or better * Significance at 10% level or better 49 Table 7.8: Regression results by region with inclusion of interaction terms L n ( D W G ) Red River Delta Uplands Region Central Coast Region Southeast Region Mekong River Delta Ln(PPR) -0.475 -0.18 0.924** 0.21 -1.202*** (0.61) (0.74) (0.377) (0.251) (0.383) Ln(NAW) -0.148 -0.375 0.856*** 0.192 0.21 (0.172) (0.566) (0.083) (0.122) (0.132) Ln(YLD) -0.114* 0.086 -0.017 0.045 0.102* (0.061) (0.078) (0.081) (0.09) (0.055) Ln(GDP) 0.091* 0.367*** -0.079 0.034 -0.11 (0.054) (0.09) (0.09) (0.026) (0.106) L n ( A L B ) -0.003 -0.129*** -0.140*** -0.121** -0.074* (0.041) (0.043) (0.034) (0.051) (0.041) Ln(UER) -0.487*** 0.782** 0.163 0.36 -0.051 (0.168) (0.359) (0.25) (0.299) (0.162) D1998 4.253 0.972 9.509* 9.309 0.149 (8.855) (6.492) (5.128) (6.66) (3.071) D2002 -6.356 3.914 •5.986* 0.042 -9.555*** (4.512) (4.58) (3.253) (2.217) (2.654) D1998*Ln(PPR) -0.643 0.137 -0.927 -0.656 -0.037 (1.127) (0.91) (0.713) (0.681) (0.548) D1998*Ln(NAW) 0.246 0.279 -0.768*** -1.305*** 0.064 (0.233) (0.57) (0.224) (0.264) (0.29) D1998*Ln(UER) 0.483 -1.233 -0.36 -0.135 0.595 (0.513) (0.932) (0.56) (0.678) (0.79) D2002*Ln(PPR) 0.746 -0.414 -0.454 0.31 1.436*** (0.635) (0.753) (0.465) (0.303) (0.402) D2002*Ln(NAW) 0.209 0.409 -0.808*** -0.119 -0.178 (0.175) (0.572) (0.089) (0.128) (0.133) D2002*Ln(UER) 0.429** -0.752 -0.382 -1.037** 0.129 (0.184) (0.407) (0.394) (0.447) . (0.191) Intercept 6.656 0.189 -5.038* -0.037 11.281*** (4.182) (4.105) (2.656) (2.013) (2.535) R 2 0.191 0.479 0.387 0.484 0.127 50 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0092733/manifest

Comment

Related Items