Version classiqueVersion mobile
OpenEdition Books

Mutations démographiques et sociales du Viêt Nam contemporain

 | 
Maria E. Cosio Zavala
, 
Myriam de Loenzien
, 
Bich-Ngoc Luu

Are non-farm jobs better for rural workers? A Panel Data Analysis of Earnings Gaps in Vietnam

Nguyen Huu Chi

Texte intégral

1In developing countries, agricultural employment accounts for a huge component of rural workforce. Literature have shown that the low agricultural productivity and great dependence of farm production on uncertain natural conditions are driving forces to low labour income in the agricultural sector which in turn results in high poverty incidence in rural area. Thus, a special attention is paid to the identification of rural pathways out of poverty. As summarized by McCulloch et coll. (2007), the main pathway out of poverty would be associated with increases in the productivity of rural poor, whether these increases are realized in farm job, in rural non-farm activity or by rural-urban migration. Traditionally, agricultural growth has been the main ingredient in rural development strategies in many countries. However, it is shown that the reliance on pro-poor agricultural growth as the main path out of poverty today is facing challenges due to a combination of factors leading to increase of risk, uncertainty and raising costs and/or lower return to agricultural investment (Dorward et coll., 2004). Scholars have for long emphasized the need for diversified approaches to fighting rural poverty in order to take the heterogeneity of the rural population into account (Jonasson, 2009). It is widely considered that rural non-farm employment (RNFE) is a potential path out of poverty as it can provide a source of income for rural workers, particularly those who are facing difficulties of securing income due to underemployment, insufficient of cultivable land. Additionally, non-farm employment appears to bring about better earnings than do agriculture jobs. Although this feature of non-farm employment has long been recognised as being among the most robust findings in comparative international labour economics (Hertz et coll., 2009), not much empirical evidence has been offered on it. This lack of evidence has sometimes been pointed out in the literature. For instance, Lanjouw (2008) argued that even though average earnings in the rural non-farm sector are higher than in agriculture, it is unclear whether income prospects are systematically better in non-farm activities than in agriculture. Winters et coll. (2008) also indicated that while agricultural jobs generally do pay less, there is still a considerable overlap between the farm and non-farm wage distributions in most countries.

2To measure farm - non-farm earnings gaps, some studies have been conducted based on both cross-sectional and panel data. For instance, basing on the method of propensity score matching, Dabalen et coll. (2004) estimate return to participation in the rural non-farm sector compared with that in the agricultural sector in Rwanda. They test whether people with similar attributes, but in different sectors, earn different incomes, and find that the self-employed in the non-farm sector earn significantly more than farm workers do. McCulloch et coll. (2007), in a study of pathways out of rural poverty in Indonesia, use panel data to trace the income changes of people who switch from agriculture to non-farm activities. They find that increased engagement of rural farmers in non-farm businesses has been the most promising way to improve incomes for workers and their household, thus constitute a path out of rural poverty. Jonasson (2009) provides further evidence on this issue in testing for existence of earnings gaps between rural non-farm and farm employment while controlling for skills of workers. His empirical findings indicate that there would be no premium for an unskilled worker who switches from agriculture to rural non-farm employment. By contrast, skilled workers are likely to do better in non-farm than in agricultural jobs since they have higher return to education. However, it is arguable that earnings gaps between non-farm and farm jobs depend not only on individual characteristics but also on the characteristics of the non-farm jobs in which workers engage in. So far, there has not been clear evidence on the variation of non-farm – farm earnings gaps by sector as well as status of non-farm employment.

3This paper aims at assessing earnings differentials between farm and non-farm jobs with a further attention paid to accounting for heterogeneity among non-farm employment in both formal and informal sector. This analysis is an extension of a work by Nguyen, Nordman and al. (2013) which measures earnings gaps within non-farm sector jobs in Vietnam. We distinguish different types of rural non-farm workers in order to allow earnings gaps to be measured specifically for each type of non-farm jobs compared to agricultural jobs. The main questions to be investigated are: is there a systematic earnings gap between rural non-farm and agricultural jobs when controlling for factors that potentially determine earnings of individuals, given the heterogeneity among rural non-farm activities as well as employment? Do all non-farm jobs provide pecuniary premiums over agricultural jobs? Do possible gaps vary remarkably along the earnings distributions?

  • 1 The definition of this so-called informal/formal employment divide is explained below (see 1.2).

4Taking advantage of the rich Vietnam Household Living Standards Survey (VHLSS) dataset in Vietnam (a Living Standard Measurement Study type household survey), in particular its three-wave panel data (2002-2004-2006), we assess the RNFE–agricultural earnings gaps in this country. Our empirical analysis consists of assessing the magnitude of different types of earnings gaps between non-farm and agricultural employment using Ordinary least squares (OLS) and quantile regressions. We use a worker level definition of informality, the so-called informal/formal employment divide1 (Hussmanns 2004). Standard earnings equations are estimated at the mean and at various conditional quantiles of the earnings distribution. Furthermore, we estimate fixed effects quantile regressions to control for unobserved individual characteristics, focusing particularly on heterogeneity within both the formal and informal non-farm employment categories. Our purpose is to address the important issue of heterogeneity at two levels: the worker level, taking into account individual unobserved characteristics; the job level, differentiating wage and self-employed workers. Estimations of earnings gaps from panel data would help preventing the risk of spurious results since we can take into account unobserved heterogeneity as well as, to some extent, the potential endogeneity concerning the selectivity process of job allocation across sectors.

5The remainder of this chapter is organized as follows. Section 1 presents the context, the data and some descriptive elements of income dynamics in the recent period, while Section 2 focuses on the econometric approach to assess rural farm - non-farm earnings gaps. Empirical results are presented in section 3 and discussed in Section 4.

Context, Labour Market Dynamics in Vietnam and Data

Context

6The growth model embraced by Vietnam during the last two decades, in an urbanization context, has prompted deep social economic transformation. The private sector has been thriving with the transition of a centrally planned economy towards a “socialist-oriented market economy” since the Doi Moi (Renovation) launched in 1986. Economic growth has helped reduce poverty considerably, but in the meantime, sparked increasing social inequality. The gap between regions, areas and social classes has widened (VASS 2010; Cling et coll. 2012). Market freedom, meanwhile, paved the way for the development of an informal economy.

7In the labour market, two main striking features are at stake: the rising rate of wage and non-farm employment; a sharp increase in real wages and labour incomes in recent years (Cling et coll. 2012). Vietnam’s impressive economic growth over the last decade has triggered a sharp increase in the rate of wage employment, which is one of the striking facts of the labour market developments in recent years: the rate rose from 19 percent in 1998 to 33 percent in 2006. Wage employment grew particularly sharply in the industrial sector (including construction) during the last ten years. This spread of wage employment has affected all population categories (urban/rural, male/female, skilled and unskilled), but substantial differences in level subsist. Wage employment is obviously more developed among the most skilled manpower (86 percent among the highly skilled as opposed to barely one-quarter among the unskilled), and it is also more prevalent among urban dwellers and among men (35 percent compared to 25 percent for women).

8The spreading of wage employment on the Vietnamese labour market has been accompanied by a steep decline in agricultural employment. From 1998 to 2006, the share of agricultural jobs has been reduced by 18 percentage points, from 67 to 49 percent (Table 1). This trend is due to a vibrant urbanization process (according to the latest population census conducted in 2009, the population has been growing by 3.4 percent annually in urban areas over the last decade, compared to 0.4 percent per year in rural areas; General Statistics Office and UNFPA 2009). But, at the same time, in all kinds of geographic areas, the proportion of out farm jobs has been on the rise, a shift particularly important in suburban areas (Cling et coll., 2010b). For instance, in the rural surroundings of the two biggest cities (Hanoi and Ho Chi Minh), agricultural employment has fallen down from 58 percent to 22 percent during the period. Despite an important rate of underemployment, wages gradually rose from 1998 to 2006. Sharp economic growth prompted a 56 percent increase in wage earners’ average annual remuneration over the period observed, which works out at an average annual growth rate of 5.7 percent.

Table 1- Changes in labour structure and earnings in Vietnam 1998-2006

Jobs (percent)

Real income (100 = 1998; wage only)

Sector

1998

2002

2004

2006

1998

2002

2004

2006

Agriculture

67.1

56.5

52.0

49.2

100

96.2

107.4

128.3

Secondary sector

13.9

19.7

21.7

23.0

100

109.4

119.6

134.3

Services

19.0

23.8

26.3

27.8

100

146.1

158.3

177.7

Total

100

100

100

100

100

121.2

137.1

155.7

Wage workers

17.5

28.6

31.0

33.1

-

-

-

-

Note: Secondary sector includes fishery, mining, manufacture and construction.

Source: General Statistics Office (GSO) Viet Nam Living Standards Survey (VLSS) 1998, Viet Nam Households Living Standards Survey (VHLSS) 2002, 2004, 2006; Cling et coll. 2012.

9Real wages grew at a slower pace in agriculture than in other sectors (28 percent vs. 34 percent and 78 percent for secondary sector and services) over the period. Wage dynamics was higher for the semi-skilled and high skilled workers than for unskilled workers (67 percent, 62 percent and 36 percent respectively). At the same time, the increase was lower for men than for women (respectively +51 percent and +60 percent from 1998 to 2006), mainly given the changes of the labour market structure (more in favour of female workers). This leads to a reduction in gender inequalities to some extent (Cling et coll., 2012).

10This context has brought a very optimistic view on the dynamism of the economy and of the labour market in Vietnam. Still, this analysis focuses only on the global trends and fails to take into consideration the informal economy. The on-going restructuring on the labour market clearly benefited the non-farm private sector: the formal sector (both domestic and foreign enterprises) but also the household businesses, of which the informal sector is the main part. The share of large enterprises in total labour force doubled, from a very low 4 percent in 1998 to 8 percent in 2006. At the same time, non-farm household businesses employment increased from 20 percent to 35 percent during the same period. In parallel, the informal jobs (not covered by social insurance scheme), still represent a massive component of the economy. The Labour Force Survey (LFS) conducted in 2007 gives a precise picture of these two dimensions of the informal economy (Cling et coll., 2010a). Informal sector jobs represent 23 percent of total jobs and nearly a half of non-farm jobs; informal jobs account for 82 percent of total jobs and two-thirds of non-farm jobs.

11Whatever the growth hypotheses in the years to come, the employment in the informal sector and its share in the total employment will increase even without the economic downturn of 2008-2009.This phenomenon is due to the limited capacity of the private formal sector (even if it continues to grow with the same frantic rhythm as prior to the crisis) to absorb the new entrants in the labour market and the workers who move from agricultural to non-farm activities. Therefore, understanding better the informal employment dynamics is a key challenge to design policies aiming at protecting its workers, improving the labour conditions and increasing productivity, keeping with it intrinsic flexibility.

Data

12The data used in this paper are drawn from three successive rounds of the Vietnam Household Living Standards Surveys (VHLSS 2002, 2004 and 2006). These surveys are Living Standard Measurement Study (LSMS) type surveys, probably one of the most popular household surveys in developing countries. Initially designed by the World Bank to measure and monitor poverty and inequality, LSMS became multi-purpose studies, covering almost all aspects of the economic and domestic activities of households.

  • 2 The primary sample units are the communes/wards, the secondary sample units are the census enumerat (...)

13In terms of sample design, the VHLSS are a classical three-stage stratified random survey, covering the ordinary households at the national level.2 The sample size is quite large even if it has been progressively reduced, from 75,000 in 2002 to 45,000 in 2004 and 2006 (Table 2). A detailed questionnaire (including expenditures and other subject specific modules) has been applied to a random sub-sample of 30,000 and around 9,000 households respectively. To track individual changes over time, a panel component has been implemented, selected among the three sub-samples. As in other studies, individuals have been matched between the three surveys using the common individual identifier across years, cross-checked with gender, age and other individual information. After undertaking thorough data cleaning including checking consistency of time-invariant variables between the three survey rounds, we have been able to recover a substantial number of new individuals, and to correct misclassified ones. In the end, our balanced panel includes 7,408 individuals matched between all the three rounds of VHLSS (Table 2); 10,891 individuals observed only in 2002 and 2004; and 9,529 individuals observed only in 2004 and 2006.

  • 3 We should be cautious to consider that all the jobs held by rural workers are rural employment sinc (...)

14As the major objective of our study is to investigate the question of earnings of rural workers, we retained only those individuals who are occupied workers aged 15 or more, residing in rural area.3 Finally, our empirical analysis is based on a panel of rural workers including 2,718 individuals observed in all three years (balanced part). In the unbalanced parts, there remain 3,820 individuals observed in both 2002 and 2004 but not in 2006, and 3,928 individuals who are observed as non-farm workers in both 2004 and 2006, but were not surveyed in 2002.

Table 2- Building the panel of individuals with VHLSS 2002, 2004 and 2006

2002

2004

2006

Full sample (household)

75,000

45,000

45,000

Detailed sample (household)

30,000

9,000

9,000

All individuals

Unbalanced Panel

18,299

27,828

16,937

Balanced panel

7,408

7,408

7,408

Rural workers aged 15 years or over

Unbalanced Panel

6,538

10466

6,646

Balanced Panel

2,718

2,718

2,718

Observed in 2002 and 2004

3,820

3,820

-

Observed in 2004 and 2006

-

3,928

3,928

Source: GSO VHLSS 2002, 2004, 2006; author’s calculations.

15In term of content, from an informal sector and informal employment perspective, the VHLSS does not allow to capture precisely these two concepts according to the international definitions (International Labour Organization, 2003; European Commission, et coll. 2008), as the survey has not been designed for such a purpose. In Vietnam, the informal sector is defined as all private unincorporated enterprises that produce at least some of their goods and services for sale or barter, are not registered (no business licence) and are engaged in non-farm activities. The informal employment corresponds to employment with no social security insurance. On the job side in the VHLSS, the formal/informal divide can only be computed for the wage workers. On the firm side, household businesses can be split between registered and not registered ones, but no information is available on the jobs generated by these businesses. Therefore, we created an informality proxy combining both job and firm approaches. Among non-farm workers, four main groups are distinguished, taking into account their job status. Among wage workers, informal ones are those who do not benefit from social security insurance. Among employers and self-employed, informal workers are those whose business is not registered. This classification provides the best available measures of informality in Vietnam, prior to the LFS 2007 (which unfortunately does not provide any panel component; Cling et coll., 2010a). Finally, our typology in four groups of non-farm workers is perfectly consistent with the International Labour Organization (ILO) definition of informal employment, but not with the informal sector one, as we are not able to distinguish among informal wage workers those who are working in the informal sector from those who are informally employed in the formal sector.

16Information on informality can be tracked in the questionnaire by the “Employment” and “Non-Farm Household Business” (NFHB) modules. Apart from our formal/informal variable, we compute the labour income associated with each remunerated job. For wage workers, earnings are obtained by summing the direct wage with all the supplementary benefits perceived in cash or in kind and converted into pecuniary equivalent (public holidays, bonuses, social allowance, etc.). For the self-employed, we compute their annual net income by subtracting all the expenses engaged (intermediary consumption, labour costs, taxes, etc.) to the production generated by the household business. Hourly earnings used in the estimations are deduced using the total number of hours worked during the year. Additionally, all the classical individual and household based socio-demographic variables are appended to our database.

17Finally, regional and time deflators have been elaborated to compute real earnings. As the regional deflators (16 locations, i.e. 8 regions in two areas, urban and rural) included in the VHLSS databases have been criticized for not being consistent over time (Mc Caig et coll., 2009), we combined the VHLSS 2006 regional deflators (supposed to be the most reliable one) with the provincial Consumer price indexes (CPIs) (63 provinces) provided by the General Statistics Office aggregated at the regional level. This adjustment is quite substantial, given the high differences in price levels and inflation: a difference of more than 77 percent in prices is observed between the lowest price level (rural North-East region, 2002) and highest one (urban South-East region, 2006), showing that markets are far from being fully integrated in Vietnam.

Econometric Approach to Measuring Earnings Gaps between different groups of workers

18The empirical analysis consists of assessing the magnitude of different types of farm - non-farm earnings gaps using OLS and quantile regressions with log hourly earnings from principal employment of occupied workers as dependent variable. Standard earnings equations are thus estimated at the mean and at various conditional quantiles of the earnings distribution. The models are regressed on a pooled sample of workers over years employed in the non-farm and agricultural sectors. The different covariates introduced into the regressions are the completed years of education, the years of potential experience (with quadratic profiles for these two regressors), a dummy for being married, a dummy for being a woman, seven regional dummies (in models estimated for the whole Vietnam), and two time dummies to control for macroeconomic trend effects on earnings.

19A number of studies based on data on African manufacturing firms have shown that wages are positively correlated to firm size, conditional on standard human capital variables (Strobl and Thornton 2002, Magda 2008, Söderbom, Teal and Wambugu 2005). In this paper, due to lack of information on the demand side characteristics, we cannot control for the size of the wage workers’ firms. The literature discusses numerous reasons why wages are positively correlated with firm size. One of the frequently made arguments is that firm size is correlated with omitted worker quality because large firms usually attract more productive workers. In this paper, we control for both observed human capital and time-invariant unobserved characteristics, thus mitigating the drawback of not accounting for firm size in the regressions.

20To account for non-farm vs. farm differences in earnings at the mean earnings level, we turn to Pooled OLS regressions across years and Fixed Effects OLS regressions (FEOLS), the latter accounting for time-invariant unobserved heterogeneity. The FE model can be written as

21yit = x'itβ + γIit+ τFit + αi + εit (1)

22Where xit denotes the vector of characteristics of individual i observed at time t (which includes a constant term), Iit and Fit represent dummies taking value one if person i observed at time t is a non-farm informal and formal worker respectively. αi is the time-invariant individual heterogeneity (or the individual fixed effect) and εit is an i.i.d. normally distributed stochastic term absorbing measurement error.

The estimated coefficient Image 100000000000001500000027889EBE12.png and Image 10000000000000140000001EC52667C5.png are interpreted as measures of the conditional earnings premium/penalty experienced by non-farm workers (informal or formal) compared to agricultural workers. However, as mentioned previously, the informal sector is extremely heterogeneous and a finer job divide should be considered. We then define four categories of rural non-farm workers split by job status (wage workers vs. self-employed workers) and institutional sector (formal vs. informal) and create four dummies taking value one if the individual i at time t is a non-farm informal wage worker (IWit), a non-farm formal wage worker (FWit), a non-farm informal self-employed worker (ISit) and a non-farm formal self-employed worker (FSit).

23Taking the agricultural workers (AWit) as the reference category, the model we estimate can be written as:

24yit = x'itβ + δIWit+θISit + μFW + λFSit + αi + εit (2)

The estimated coefficients Image 10000000000000B000000025AC091D82.png are interpreted, respectively, as the IW – AW, IS – AW, FW - AW and FS – AW conditional earnings gaps. Identification of these conditional earnings gaps relies on the presence in the sample of movers between employment states over time. Those movers can be compared to the stayers in terms of earnings.

25As an illustration, we consider a simple two-period example and ten cases of transitions out of the various possibilities of professional trajectories (which are 25 in a two-period example):

262 cases of stayers:

27E[yi2yi1|AWi1 = 1, AWi2 = 1] = Δ (3)

28E[yi2yi1|IWi1 = 1, IWi2 = 1] = Δ (4)

29With Δ = (x'i2 - x'i1)β

306 cases of movers:

31E[yi2yi1| AWi1 = 1, IWi2 = 1] = Δ + δ (5)

32E[yi2yi1| AWi1 = 1, FWi2 = 1] = Δ + μ (6)

33E[yi2yi1| AWi1 = 1, ISi2 = 1] = Δ + θ (7)

34E[yi2yi1| AWi1 = 1, FSi2 = 1] = Δ + λ (8)

35E[yi2yi1| ISi1 = 1, FSi2 = 1] = Δ + λ + θ (9)

36E[yi2yi1| IWi1 = 1, FWi2 = 1] = Δ + λ + δ (10)

37With Δ = (x'i2 - x'i1)β

38Equations (3) and (4) give examples of the changes in earnings for stayers, i.e. for workers that do not change their employment state between the two periods. Equations (5), (6), (7), and (8) illustrate the changes in earnings for those workers coming from an agricultural job and moving, respectively, into an informal wage job, a formal wage job, an informal self-employed job, and a formal self-employed job; equations (9) and (10) represent these earnings differentials for those coming from an informal self-employed/wage job and moving, respectively, into a formal self-employed/wage job.

39The identification strategy of FE on movers is quite standard but, in practice, one should verify that the number of moves across sectors is sufficient for a valid use of this estimator; which is the case, as shown in Table 4 in the next section.

40Finally, to allow the earnings gaps between job statuses to differ along the earnings distribution, we rely on Quantile Regressions (QR). Quantile earnings regressions consider specific parts of the conditional distribution of the hourly earnings and indicate the influence of the different explanatory variables on conditional earnings respectively at the bottom, at the median and at the top of the distribution.

41Using our previous notation, the model that we seek to estimate is:

42qϱ(yit) = x'itβ(ϱ) + δ(ϱ)IWit + μ(ϱ)FWit + θ(ϱ)ISit + λ(ϱ)FSit + α2 ,

43quel que soit ϱ appurtenant à l’intervalle [0,1] (11)

44where qϱ(yit) is the ϱth conditional quantile of the log hourly earnings. The set of coefficients β(ϱ) provide the estimated rates of return to the different covariates at the ϱth quantile of the log earnings distribution and the coefficients δ(ϱ), μ(ϱ), θ(ϱ) and λ(ϱ) measure the parts of the earnings differentials that are due to non-farm-farming job differences at the various quantiles. In a quantile regression, the distribution of the error term is left unspecified. The quantile regression method provides robust estimates, particularly for misspecification errors related to non-normality and heteroskedasticity.

We then turn to Fixed Effects Quantile Regressions (FEQR). The extension of the standard QR model to longitudinal data has been originally developed by Koenker (2004). More recently, Canay (2011) proposed an alternative and simpler approach which assumes that the unobserved heterogeneity terms have a pure location shift effect on the conditional quantiles of the dependent variable. In other words, they are assumed to affect all quantiles in the same way. It follows that these unobserved terms can be estimated in a first step by traditional mean estimations (for instance by FE OLS). Then, the predicted Image 100000000000002100000025F1AFB471.jpg are used to correct earnings, such as Image 10000000000000A8000000279A3B262F.png which are regressed on the other regressors by traditional QR.

45When running the regressions (2) and (11), we always provide robust standard errors using bootstrap replications. To reduce a possible bias due to measurement and reporting errors in the earnings and independent variables, we trim the data and drop influential outliers and observations with high leverage points from our sample that we identify by the DFFITS-statistic. As suggested by Belsley, Kuh and Welsch (1980), we use a cutoff-value

Image 10000000000001170000002DECC8DFF7.png with k, the degrees of freedom (plus 1) and N the number of observations. This procedure removes 658 observations from our initial unbalanced panel sample.

Descriptive statistics and validity checks for earnings gaps measurement

46Table 3 presents some basic summary statistics of the main characteristics of the panel data used in our analysis. These descriptive statistics are reported for sub-samples of rural workers in the whole country. The results obtained for average earnings as well as individual characteristics are in line with common findings in the literature.

Table 3- Summary Statistics (pooled waves 2002-2004-2006)

Non-farm formal

Non-farm Informal

Agricultural workers

Self-

Employed

Wage workers

Self-employed

Wage workers

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

Mean

Std. Dev.

Hourly earnings

2.013

0.76

1.711

0.62

1.607

0.73

1.391

0.50

1.221

0.82

Potential experience

25.87

10.69

19.54

10.94

28.11

13.85

19.84

11.80

27.89

13.25

Age

39.52

9.88

36.22

10.59

40.25

12.62

32.48

11.26

38.86

12.75

Female

0.537

0.50

0.405

0.49

0.598

0.49

0.258

0.44

0.466

0.50

Married

0.849

0.36

0.746

0.44

0.805

0.40

0.620

0.49

0.794

0.40

Position in the family

Head of household

0.404

0.49

0.392

0.49

0.383

0.49

0.417

0.49

0.425

0.49

Spouse

0.385

0.49

0.227

0.42

0.400

0.49

0.117

0.32

0.332

0.47

Children

0.183

0.39

0.352

0.48

0.198

0.40

0.445

0.50

0.223

0.42

Others

0.027

0.16

0.030

0.17

0.018

0.13

0.021

0.14

0.020

0.14

Education

No degree

0.037

0.19

0.025

0.16

0.137

0.34

0.100

0.30

0.292

0.45

Primary

0.277

0.45

0.101

0.30

0.326

0.47

0.322

0.47

0.317

0.47

Secondary

0.587

0.49

0.526

0.50

0.412

0.49

0.474

0.50

0.304

0.46

University & others

0.098

0.30

0.348

0.48

0.125

0.33

0.104

0.30

0.087

0.28

Industry

Food and beverage

0.076

0.27

0.040

0.20

0.111

0.31

0.058

0.23

-

-

Textile, leather, wood, handicraft

0.071

0.26

0.089

0.28

0.164

0.37

0.180

0.38

-

-

Construction

0.076

0.27

0.119

0.32

0.036

0.19

0.138

0.34

-

-

Whole sale

0.007

0.08

0.034

0.18

0.010

0.10

0.410

0.49

-

-

Retail sale

0.034

0.18

0.007

0.09

0.036

0.19

0.009

0.09

-

-

Hotel and restaurant

0.508

0.50

0.011

0.10

0.426

0.49

0.049

0.22

-

-

Transport. & warehouse

0.112

0.32

0.001

0.02

0.103

0.30

0.019

0.14

-

-

Other manufacture

0.080

0.27

0.038

0.19

0.077

0.27

0.066

0.25

-

-

Other services

0.076

0.27

0.040

0.20

0.111

0.31

0.058

0.23

-

-

Observations

589

1,606

2,420

2,345

4,538

Source: GSO VHLSS 2002, 2004, 2006; author’s calculations.

47Firstly, compared to non-farm workers, those rural workers holding farm jobs earn on average significantly less. Considering, for instance, the relative earning difference between farm and non-farm informal jobs, one can realize that on average farm workers earn 12 percent less than those informally employed in rural non-farm sector. Farm workers appear to have longer potential experience (which is calculated as age minus years of reported schooling minus five) than their counterparts working in the non-farm sector. Women have higher propensity to engage in farm and non-farm informal self-employed jobs. Workers having agricultural job tend to have also a lower level of education compared to those participating in non-farm job. For instance, 29 percent among those workers engaging in farm jobs has no degree at all, whereas, on the other extreme, 35 percent of non-farm formal wage workers are those who obtained undergraduate education.

48Secondly, regarding non-farm employment, the results show that workers holding formal jobs earn more on average than those engaged in informal jobs. Among each group of formal and informal workers, self-employed workers are those with higher earnings in comparison with wage earners. Informal workers tend to be younger than their formal worker counterparts, especially for wage workers. Self-employed workers exhibit on average longer potential experience than other non-farm workers on the labour market. As expected, workers having higher level of education are less likely to be engaged in informal employment and vice versa.

49At the aggregate level, the gender ratio does not vary between formal and informal jobs. However, when considering wage jobs, female workers represent more in formal jobs than informal ones or, in other words, male workers engage more in informal wage jobs. Finally, formal and informal workers are differently allocated across branches of activity. Specifically, informal employment is found more in trade, restaurants and transportation, while formal jobs are more concentrated in services. Interestingly, the share of manufacture is much higher for informal jobs than for formal ones (31 percent vs. 18 percent). Within institutional sectors, the distribution is also fairly unbalanced: formal wage workers are stubbornly engaged in services (60 percent), whereas formal self-employed workers hold transportation and hotel & restaurant jobs (12 percent and 52 percent respectively). Informal wage workers engaged prominently in construction (13 percent) and trade (35 percent) while informal self-employed job’s structure looks like the formal self-employed one. These significant differences in the distribution of job structure underline the importance of controlling for sectors of activity in our earnings gaps estimations.

  • 4 It is important to note that unemployment rate is stubbornly low in Vietnam’s rural area and this p (...)

50As presented in the previous section, the presence of movement between employment states overtime is important for the identification of conditional earnings gaps. Therefore, it is useful to have a first look at descriptive figures showing the movement on the rural labour market between the years studied. Table 4 reports the transition matrices of employment status between 2002-2004, 2004-2006 and 2002-2006 obtained from our unbalanced panel dataset. In order to provide a more general picture of the dynamics of switching between employment statuses, we present the results obtained from the panel of all individuals aged 15 or more. The categories shown in the matrices include then not only the four non-farm employment statuses and agricultural category, but also “not-working” (the latter category includes, to simplify the notation, those who are inactive or unemployed).4 This presentation allows identification of both transition flows between farm and non-farm jobs, as well as those within the non-farm sector employment. The figures in the second row and column of each matrix reveal that the former are not negligible. It is notable that among all transition flows, the most important ones are those between informal non-farm and farm jobs. These patterns of mobility would partly reflect the low entry barriers to both sectors as well as the fact that the majority of the workforce in Vietnam is still predominantly employed in agriculture. Another striking evidence on the flows of transition from non-farm employment is the rather high probability of becoming inactive (or unemployed) for those who were previously self-employed.

51Apart from revealing the movement between farm and non-farm jobs, the transition matrices show also that, on average, movement flows between non-farm job’s categories are not negligible. For the two-time periods, around one quarter of workers changed position from one of our four job’s status to another. Around 20 percent of the total sample moved from non-farm informal to formal jobs and the rates of formal-informal transition are about 40 percent. However, the flows are balanced in absolute terms. The fluidity between wage and non-wage jobs is smaller, but is far from negligible (from 13 percent to 15 percent of the total sample, depending on the years). Here again, the movements to and from non-farm wage jobs are relatively symmetrical. At a more disaggregated level, job mobility is at its highest for formal self-employed workers, where less than two thirds keep the same status in our different panels. Formal wage workers are the most stable (82 percent to 74 percent of stayers), while informal workers (wage and non-wage) are in between with a proportion of stayers ranging from 55 percent to 62 percent. Formal wage workers mainly move to informal wage jobs. When moving, non-farm informal wage workers tend to privilege self-employed positions, and secondarily formal wage jobs. Formal self-employed movers mainly get their business informalized (probably due to adverse conditions). A lower share of informal self-employed workers makes the inverse move, by formalizing their business. However, a substantial proportion also closes their business to become informal wage workers.

52All in all, the high consistency between our transitions matrices over different periods is a sound indicator of data quality. We would claim that the observed changes reflect real phenomena and do not mainly capture measurement errors. Furthermore, on the methodological side, the substantial numbers of movers is essential for our estimation strategy.

  • 5 We observe a slight difference between these results and those presented in Table 3 concerning the (...)

53To end this section on descriptive analysis, let us have look at the income dynamics by employment status of rural workers on the period 2004-2006. The first panel of Table 5 shows the level of real earning in 2006 by transition status, agricultural stayers being our basis. Consistently with Table 4, formal self-employed workers get the highest pay, while informal wage workers in non-farm sector together with those having farm jobs are at the lowest end of the earnings ladder.5

Table 4- Transition matrix of employment status between 2002, 2004 and 2006

2004

2002

Not-working

Agri-cultural emp.

Formal Wage

Informal

Wage

Formal

Self-employed

Informal

Self-employed

Total

Not-working

58.8

28.4

1.3

5.0

1.3

5.3

100 (16.7)

Agricultural emp.

6.5

81.7

1.6

4.6

0.8

4.8

100 (62.5)

Formal Wage worker

0.6

11.3

72.3

13.2

0.6

1.9

100 (4.4)

Informal Wage worker

1.4

20.6

3.8

66.0

3.4

4.8

100 (5.7)

Formal Self-empl. work.

4.1

12.2

4.1

2.7

51.4

25.7

100 (2.1)

Informal Self-empl. w.

4.6

21.6

0.3

6.9

8.9

57.7

100 (8.4)

Total

14.5

59.6

4.8

8.8

2.8

9.6

100 (100)

2006

2004

Not-wor-king

Agri-cultural emp.

Formal Wage

Informal

Wage

Formal

Self-employed

Informal

Self-employed

Total

Not-working

63.3

23.3

3.1

5.5

1.0

3.8

100 (14.5)

Agricultural emp.

7.1

82.0

1.7

4.4

0.9

3.9

100 (59.6)

Formal Wage worker

0.0

10.5

79.1

8.1

0.6

1.7

100 (4.8)

Informal Wage worker

1.9

23.3

7.3

61.2

1.9

4.4

100 (8.8)

Formal Self-empl. work.

11.0

14.0

2.0

1.0

46.0

26.0

100 (2.8)

Informal Self-empl. w.

6.3

21.3

1.7

3.5

5.8

61.5

100 (9.6)

Total

14.49

57.26

6.08

9.54

2.68

9.95

100 (100)

2006

2006

Not-working

Agri-cultural emp.

Formal Wage

Informal

Wage

Formal

Self-employed

Informal

Self-employed

Total

Not-working

53.5

24.4

4.8

9.4

1.2

6.8

100 (16.7)

Agricultural emp.

7.0

77.7

2.7

5.8

1.0

5.8

100 (62.5)

Formal Wage worker

1.3

10.1

73.0

12.6

0.6

2.5

100 (4.4)

Informal Wage worker

2.4

24.4

2.9

56.5

2.9

11.0

100 (5.7)

Formal Self-empl. work.

10.8

13.5

2.7

1.4

44.6

27.0

100 (2.1)

Informal Self-empl. w.

8.9

28.2

2.0

5.6

9.2

46.2

100 (8.4)

Total

14.49

57.26

6.08

9.54

2.68

9.95

100 (100)

Source: GSO VHLSS 2002, 2006, GSO; author’s calculations.

54Compared to the pooled sample, in 2006, informal non-farm self-employed workers reversed their position with formal non-farm wage workers, meaning that the earnings hierarchy between these two categories of workers is not fixed, but may vary over time. Furthermore, income levels are highly dependent on transitions. For instance, and as expected, whatever their job status in 2004, those who moved to informal wage jobs earn the less. Conversely, the workers who got the opportunity to open a formal business earn the most. The results are quite similar in terms of earning growth (second panel of Table 5). Systematically, moving to informal wage jobs is associated with the lowest increase in earnings over the period, whereas being able to change to a formal self-employed job is associated with the highest earnings growth. Of course, these unconditional averages should be controlled for observed and unobserved characteristics, which is the purpose of the following sections.

Table 5- Income dynamics by employment status between 2004 and 2006

Real income levels in 2006

2004\2006

Agri-cultural emp.

Formal Wage

Informal

Wage

Formal

Self-employed

Informal

Self-employed

Total

Not-working

100

82.7

69.8

143.4

96.5

97.0

Agricultural emp.

92.3

131.8

95.0

207.4

96.8

126.9

Formal Wage worker

79.4

74.8

84.1

213.3

117.0

87.0

Informal Wage worker

123.6

187.7

120.5

191.9

137.4

160.9

Formal Self-empl. work.

104.9

84.8

83.9

152.7

114.2

113.5

Total

99.4

118.4

80.4

143.4

96.5

103.7

Real hourly income growth 2004-2006

2004\2006

Agri-cultural emp.

Formal Wage

Informal

Wage

Formal

Self-employed

Informal

Self-employed

Total

Not-working

100

115.4

85.2

399.8

102.0

99.8

Agricultural emp.

87.1

116.1

126.8

240.1

211.4

117.9

Formal Wage worker

130.4

102.6

112.3

235.8

180.2

120.7

Informal Wage worker

156.5

439.4

57.0

111.5

129.8

123.2

Formal Self-empl. work.

157.4

109.7

90.5

235.8

120.0

125.9

Total

116.8

115.6

108.8

143.0

126.5

118.9

Note: base 100= Income level and income growth compared to agricultural workers’ stayers between 2004 and 2006.

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations.

Earnings gaps analysis

55In this section we discuss the earning gaps between rural non-farm jobs (formal and informal) and farm jobs at the aggregate level, estimated using the four estimations procedures presented in Section 3. As discussed earlier, the non-farm sector, and more broadly non-farm employment, is immensely heterogeneous. The theoretical literature suggests that a key divide should be considered between the informal and formal jobs, between wage workers and self-employed. If the point is now well established in the literature, formal job heterogeneity is rarely acknowledged. Therefore, to go beyond previous studies on this issue, we distinguish all rural non-farm workers between four groups, split by job status (wage workers vs. self-employed) and institutional sector (formal vs. informal). In the following discussion, we compare the four non-farm work status with farm workers, as our benchmark. We also investigate the gender issue.

Earnings gaps at the aggregate level

Non-farm Formal vs. Agricultural Workers

56At the aggregate level, the OLS estimate of the earning gap between formal non-farm and agricultural workers is near 50 percent (Figure 1, and Table 6). Taking into account the (time invariant) unobservable individual characteristics (UICs) through fixed effect OLS estimation (FEOLS) reduces the earnings penalty significantly, down to 31 percent. Thus, a remarkable proportion of the gap can be explained by unobservable characteristics, the most productive workers privileging the formal non-farm employment. As always, this standard feature does not tell us much about what specific factors are really at play. On the one hand, the innate ability or the “talent parabola” is commonly stressed in the literature. On the other hand, many other explanations can be put forward. For instance, UICs may have to do with more efficient social networks which help getting a formal non-farm job.

57However, the remaining 31 percent gap, once we control for UICs, highlights that formal non-farm jobs are able to bring about higher earnings per se. Here again, this result can be due to various factors which end up, at the firm level, in a higher productivity or market power, and/or, at the worker level, in a stronger bargaining power of formal workers to negotiate higher earnings.

58To go beyond average, we ran quantile regressions (QR). Results obtained show that agricultural workers are suffering earnings penalties at all levels of the conditional distribution with a larger gap at the bottom part. Concretely, from around 60 percent for the first two quartiles of income, the gap reduces to 38 percent at the upper-tier of the distribution (quantile q.90). The Fixed Effects Quantile Regression (FEQR) gap confirms the key role of UICs in reducing the “true” gap.The FEQR gaps are decreasing continuously along the earnings distribution for rural workers, from 47 percent for the bottom quantile to around only16 percent for the upper one. This would imply no much premium of moving out of farming employment for those having already high agricultural productivity.

Figure 1- Earnings gaps between Non-farm Formal and Agricultural Workers

Image 100000000000021A0000018A8EF42CD9.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations.

Non-farm Informal vs. Agricultural Workers

59As expected at aggregate level, those employed informally in rural non-farm sector are on average better-off than those engaging in farm jobs, but the gap is considerably lower than that observed between non-farm formal and agricultural workers (Figure 2 and Table 6). The OLS gap is significantly reduced to 17 percent when individual fixed effects are introduced. These results suggest that informal non-farm workers may have a disadvantage in terms of their unobserved productive attributes.

60Considering results obtained from quantile regressions, distributional pattern can be clearly identified no matter whether the UICs are controlled for or not (Figures 2, Tables 7, 8). The rural non-farm informal – agricultural earnings gap is decreasing continuously and significantly from 29.5 percent (quantile q.10) to 6.8 percent (quantile q.90), implying that agricultural workers conserve an earnings disadvantage in comparison with non-farm informal workers at any point in the pay ladder.

Figure 2- Earnings gaps between Non-farm Informal and Agricultural Workers

Image 100000000000021A0000018AAB3AACB9.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: GSO VHLSS 2002, 2004, 2006, GSO author’s calculations.

Heterogeneity among rural non-farm employment and the variation of farm - non-farm earnings gap regarding informality.

61Given the great heterogeneity among rural non-farm jobs as well as activities stressed in the literature (Foster and Rosenzweig 2004, Lanjouw and Murgai 2009), our analysis in this sub-section relies on results obtained from models in which non-farm workers are disaggregated by employment status (hired and self-employed) and informality

Non-farm Wage Workers (formal or informal) vs. Farm Workers

62Engaging in wage employment in rural non-farm sector appears to be the main trade-off against the traditional way of making a living in agriculture for rural workers. Whether rural non-farm wage employment can be distinguishable from agricultural jobs in terms of remuneration remains a question drawing considerable attention in the literature on rural economy in developing countries. Based on comparative analysis of wage distribution between rural agricultural and non-farm workers in various countries of different continents, Winters et coll. (2008) have pointed to a lower position of agricultural earnings distribution compared to that of rural non-farm wage workers in all of the African and Latin American countries. They have also mentioned the special case of Vietnam where it is harder to observe clear differences between the two earnings distributions. In our analysis, a disaggregation of rural non-farm wage workers into two groups (formal and informal) allows a finer reassessment of the earnings differentials between non-farm wage workers and agricultural workers.

Non-farm informal wage vs. farm earnings gaps

63As can be seen from Figures 3 and Tables A1, the conditional OLS gap is positive, with a significant premium of 12 percent for the non-farm informal wage workers. Nevertheless, the FEOLS models reduce the premium further to 8 percent for informal wage workers. Non-farm informal wage workers do have general disadvantage in terms of their unobserved productive attributes, which produce an overestimation of the premium associated with being a wage worker in rural non-farm sector compared to engaging as farm worker if this individual heterogeneity is not accounted for. When turning to quantile regressions (Tables 7, 8), the distributional profile of the gap presents a clear pattern, showing both positive and negative earnings gaps. The gap steeply decreases with income level, and is less and less in favour of the informal wage workers. In absolute terms, agricultural labourers suffer a penalty at almost but not entire points of the distribution. Positive earnings gap between informal wage workers and agricultural workers is observed of important magnitude at the lower quantiles (up to quantile q.80 in the pooled QR model, and quantile q.70 in the FEQR model). Afterwards, the gap is reversed into a significant premium, growing continuously up to around 39 percent for the richest decile. All in all, and given the size of the premium, we can conclude that rural informal wage employment may be less lucrative that agricultural alternatives, especially for the richest ones who have relatively high farming productivity. As a matter of consequence, we have good presumptions to assert that, in rural Vietnam, engaging in rural non-farm employment is not always a better way for workers to improve earnings.

Figure 3- Earnings gaps between Non-farm Informal Waged and Farm Workers

Image 100000000000021A0000018ABEEE5147.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations.

Non-farm formal wage vs. farm earnings gaps

64The results shown in Figure 4 and Tables A1, confirm what has been found for the raw earnings differentials reported in Table 4.

65Concretely, the conditional OLS earnings gap is positive, with a premium of more than twice that of the premium informal wage workers can have (28.5 percent vs. 12 percent), over farm jobs. When the individual unobserved heterogeneity is controlled for, the earnings premium to formal wage workers reduces to 20 percent. When turning to quantile earnings gaps, the results reveal some interesting points to be discussed. The FEQR gaps are decreasing steeply along the pay scale and turning to negative value from around the upper quantile (q.90),indicating that the farm workers do not appear to earn less than non-farm formal wage workers at these upper positions of the earning distribution.

Non-farm Self-employed vs. Farm Workers

66The earnings comparison of non-farm informal self-employed workers and farm workers is clearly in favour of the former, whatever the model chosen (Figure 5, and Tables 6, columns (3) and (5), Tables 7, 8).

Figure 4- Earnings gaps between Non-farm Formal Wage and Farm Workers

Image 100000000000021A0000018A3798A49D.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations.

67Compared to agricultural workers, earnings of informal self-employed workers are considerably higher (OLS gap of 42 percent, reduced slightly to 31 percent in FE model at the national level). The reduction of the earnings gaps between non-farm informal self-employed and agricultural jobs, when unobserved individual characteristics are controlled for, may reflect, to some extent, the disadvantages related to entrepreneurial skills. This indicates generally higher return to self-employment, even for those operating informally, in rural non-farm sector.

  • 6 The definitive assessment is even more complex as measurement errors in incomes are usually conside (...)

68The estimate of earnings gaps between non-farm formal self-employed and farm workers presents a 92 percent premium for former, reduced with fixed effects to 50 percent (Figure 6 and Tables 6, columns (3) and (5), and Tables 7, 8). Nevertheless, we should be cautious when interpreting these results, since the self-employed income may be overestimated for at least two reasons. First, the measure of income we computed remunerates both labour and capital factors, the latter being far from negligible in the informal sector (Cling et coll., 2010a). Second, the self-employed income includes the share that should be attributed to the productive contribution of unpaid family workers. As we do not have any order of magnitude of these two phenomena, it is difficult to exclude the possibility that such high premium we obtain may not be substantially reduced, once these two factors are taken into account.6

Figure 5- Earnings gaps (Non-farm Informal Self-employed vs. Farm Workers)

Image 100000000000021A0000018A85ED8343.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations.

69Turning to quantile regression, the results show that both types of self-employed workers are systematically in a better position than agricultural workers, all along the pay scale. Pooled quantile earnings gap is always positive in favour of formal self-employed workers (at least 88 percent). No distributional effects can be seen in the variation of earnings gap between non-farm formal self-employed and agricultural workers along the pay scale. The FEQR confirm this trend, the only difference is that the range of variation of the gap along the distribution is attenuated. Compared to the earnings gaps between formal wage and agricultural workers, the premium associated with working as self-employment (even for those operating informally) versus agricultural workers appear to be much higher. This implies that their unobserved productive attributes are better than those of the formal wage workers. As for the case of informal self-employed workers, their premium over agricultural workers is always higher than that of the formal wage workers no matter whether unobserved individual characteristics are controlled for or not. Overall, it seems that the Vietnamese labour market functions under a regime of wage repression. Whatever the reasons - macro pressures of international integration or deliberate policies to control inflation, or weak bargaining power of the wage workers, it seems globally preferable to work as an independent (even in the informal sector) than as a wage worker (at least in non-farm activities) when engaging in rural non-farm sector.

Figure 6- Earnings gaps (Non-farm Formal Self-employed vs. Farm Workers)

Image 100000000000021A0000018A407775AF.jpg

Note: Fixed Effects (FE) OLS are denoted by FEOLS and Fixed Effects Quantile Regressions (QR) by FEQR. Bootstrapped 95 percent confidence intervals are represented by the grey surface for QR and by dashed lines for the OLS.

Source: VHLSS 2002, 2004, 2006, GSO author’s calculations.

70Lastly, it is important to note the difference in the magnitude of earnings premium between two types of self-employment. Non-farm formal self-employed workers have systematically higher earnings premium than their informal counterparts when leaving agricultural employment, all along the pay scale (Tables 7, 8). This advantage of formal household businesses may be due to higher initial level of capital or more productive combination of factors (our models do not provide elements on this point), but it is compatible with the potential intrinsic benefits of getting formal (access to credit and markets) as found by Rand and Torm (2012) in the case of Vietnam.

A gender perspective

  • 7 For Africa, see Nordman, Robilliard and Roubaud (2010) for estimates of the gender earnings gap in (...)

71Taking into account the gender dimension in studying rural non-farm employment and informality is important for various reasons. First, labour market literature has generally shownstrong imbalances in the job structure regarding this dimension. Rural females are less likely to engage in non-farm employment, but when doing so, they are more prone to hold informal jobs than their male counterparts. Second, the raw gender earnings gap is in general significantly higher in the non-farm informal sector.7 Finally, and more importantly, the motivation to hold non-farm informal jobs is highly dependent on gender. Women may have a welfare function which is less dependent on income incentives, as they take more care of extra professional activities (as family life, children care, social relations, etc.), where informal jobs could be a more satisfying option. Without going into details, we highlight here the main findings displayed in Figures 7.

72Firstly, whatever the models’ specifications and the category of workers considered, rural females are always benefiting less premium when participating in non-farm jobs. For instance, at the aggregate non-farm informal/farm level, the estimated FEOLS earnings gap is 10 percent for females whereas the corresponding figure obtained is 22 percent for males. Such a feature is compatible with the idea mentioned above, that women may accept lower wages in the informal sector because it provides other non-pecuniary advantages, relatively more valuable to them. However, it can also reveal barriers or labour market segmentation, which would be more pronounced for women competing for non-farm salaried jobs.

73Secondly, in spite of differences in absolute levels, the distributional profile of the earnings gaps is quite similar across gender: no noticeable effect for non-farm formal self-employed workers, a decreasing premium for all types of non-farm workers, both formal and informal.

Figure 7- Estimated Earnings Gaps for Women and Men Separately by Fixed Effects OLS and QR (with reference to agricultural workers)

74Non-farm Formal – Agricultural Earnings gap

Image 10000000000001CA00000101013DF326.jpg

75Non-farm Informal – Agricultural Earnings gap

Image 10000000000001ED00000101A251B2CB.jpg

76Non-farm Informal Wage – Agricultural Earnings gap

Image 10000000000001BE000000FC1E4725E0.jpg

77Non-farm Formal Wage – Agricultural Earnings gap

Image 10000000000001ED000000FC572091E2.jpg

78Non-farm Informal Self-employed – Agricultural Earnings gap

Image 10000000000001D00000010CE1263635.jpg

79Non-farm Formal Self-employed – Agricultural Earnings gap

Image 10000000000001F90000010C1C71040E.jpg

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations

80Lastly, the sorting process in the allocation of men and women across sectors (which is partly revealed by the effect of controlling for unobserved individual characteristics) does not differ substantially across gender: informal wage workers have negative unobserved individual characteristics (in order to get a better income) vis-à-vis agricultural workers, while the unobserved skills are higher for non-farm self-employed workers (whether formal or informal). The only exception is for male wage workers, who have comparable unobserved individual characteristics along the formal/informal divide.

Conclusion

81This paper focuses on measuring and analyzing earnings gap between non-farm and farm workers in rural area of Vietnam in order to assess if the participation in non-farm employment is really able to improve earnings for rural workers, thus a path out of poverty for rural population. Taking advantage of the rich VHLSS datasets, the three rounds of panel surveys (2002, 2004 and 2006) give the unique opportunity to control for time invariant unobserved individual characteristics. Using both standard and fixed effects earnings equations estimated at the mean and at various conditional quintiles of the earnings distribution, we address the key issue of heterogeneity, at three different levels: at the worker level, taking into account individual unobserved characteristics; at the job level, comparing different types of non-farm workers with agricultural workers; at the distributional level. Gender issues are also examined.

82Our results suggest that the non-farm – agricultural earnings gap highly depends on the non-farm workers’ job status (wage employment vs. self-employment) and on their relative position in the earnings distribution. Rural workers can earn more when engaging in non-farm activities instead of keeping agricultural jobs, but this is not always the case for all types of non-farm jobs. In many cases, non-farm jobs (such as informal wage employment and even formal wage jobs at upper quintiles) are not as rewarding as agricultural jobs. In other words, engaging in non-farm jobs is generally more rewarding than agricultural jobs at lower quintiles. This feature is due to the relatively low wages of both formal and informal wage jobs in rural non-farm sector. However, the reason for such results should be investigated further by undertaking some case studies for refining non-farm job earnings by industry or in specific geographical areas. Second, females are always financially benefiting less when they are employed in non-farm employment, particularly when engaging in informal jobs. This feature opens the space for specific policies to align women’s functioning of rural labour market with men’s one (reduction in the barriers of entry to formal jobs, etc.).

83Our analysis raises further promising prospects, and could be extended in various directions. A first extension would be to better control for individual unobserved characteristics, by purging our earning estimations of differences in the amount of physical capital (for self-employed workers) and social networks. A firm based panel approach may be an interesting alternative entry in this respect. The purpose would be to focus on the adverse (or positive) effect of family and kinship ties and identifies the key determinants of potential redistributive pressure exerted by the kinship ties, extended family, and network capital on the performance of non-farm household businesses (NFHBs). A first draft of a paper on this issue has been undertaken (Nguyen and al. 2011). Preliminary results show that, when controlling for time-invariant unobserved heterogeneity of the NFHBs, being an informal NFHB corresponds to a penalty in annual value added of about 30 percent. Also for this type of household firms, we find evidence of a productivity differential between family and hired labour, with a gap of roughly 15 percentage points in favour of the latter. Using information on the entrepreneurs’ social capital, we obtain some results confirming the importance of unlocking financial constraints (on physical capital) and improving access to professional support for successful household entrepreneurship. First, we always find a positive effect on the informal business’ technical efficiency of benefiting from a loan from the kin. Second, for these informal entrepreneurs, being a member of a business association or having a friend producing the same product is beneficial in terms of efficiency, perhaps thanks to knowledge spill-over and/or shared clienteles. Professional network capital thus appears to be one important ingredient of informal business performance. Another potential extension would be to exploit further the nature of our data (three point panel) by estimating dynamic earnings equations. Lastly, our work could be usefully complemented by investigating the determinants of job’s satisfaction, to enlarge the perspective which relies exclusively on the earning outputs and to check for the robustness of our conclusions.

Bibliographie

Bargain Olivier, Kwenda Prudence, Is Informality Bad? Evidence from Brazil, Mexico and South Africa, IZA Discussion Paper, n° 4711, January 2010, p. 30.

Belsley David, Kuh Edwin, Roy Welsch, Regression Diagnostics: Identifying Influential Data and Sources of Collinearity, New York, John Wiley, 1980, 292 p.

Canay, Ivan A., “A simple approach to Quantile Regression for Panel Data” in The Econometrics Journal, 14, 2011, p. 368-386.

Cling Jean-Pierre, Nguyen Huu Chi, Razafindrakoto Mireille, Roubaud François, “Urbanisation et Insertion sur le Marché du Travail au Vietnam : Poids et Caractéristiques du Secteur Informel”, communication at the Regional Conference Trends in urbanization and peri-urbanization in South-East Asia, CEFURDS/IRD, Ho Chi Minh City, December, 9-11, 2012, p. 205-226.

Cling, Jean-Pierre, Nguyen Thi Thu Huyen, Nguyen Huu Chi, Phan Thi Ngọc Tram, Razafindrakoto Mireille, Roubaud François, The Informal Sector in Vietnam: A focus on Hanoi and Ho Chi Minh City, Hanoi, The Gioi Edition, 2010a, 248 p.

Cling, Jean-Pierre, Razafindrakoto Mireille, Roubaud François, « Assessing the Potential Impact of the Global Crisis on the Labour Market and the Informal Sector in Vietnam », in Journal of Economics & Development, n° 38, June 2010b, p. 16-25.

Dickens William et Lang Kevin, « A Test of Dual Labour Market Theory », in The American Economic Review, n° 4/75, p. 792-805.

dorward Andrew, Kydd Jonathan, Morrison Jamie, Urey Ian, « A Policy Agenda for Pro-Poor Agricultural Growth », in World Development, n° 1/32, 2004, p. 73-89

Dabalen Andrew, Paternostro Stefano et Gaëlle Pierre, The Returns to Participation in the Non-farm Sector in Rural Rwanda, World Bank Policy Research Working Paper WPS3462, 2004, 33 p.

European Commission, International Monetary Fund, Organization for Economic Co-operation and Development, United Nations, World Bank, The System of National Accounts 2008 (2008 SNA), New York, A Handbook, 2009, p. 722.

Falco Paolo, Kerr Andrew, Rankin Neil, Sandefur Justin, and Teal Francis, « The Returns to Formality and Informality in Urban Africa, CSAE WPS.2010-03, University of Oxford, 2010, 27 p. + appendices.

Fernández Rosa, Nordman Christophe, « Are There Pecuniary Compensations for Working Conditions? », in Labour Economics, n° 2/16, april 2009, p. 194-207.

Foster, Andew D.; Rosenzweig, Mark R.; 2004. Agricultural development, industrialization and rural inequality, 37 p. + appendices (dact.) accessed at http://www.ccpr.ucla.edu/ events/ccpr-seminars-previous-years/Sem04F%20Rosenzweig%20Agricultural%20Deve lopment%20Industrialization%20and%20Rural%20Inequality.pdf on 11/12/2014 15:33.

General Statistics Office, UNFPA, The 2009 Vietnam Population and Housing Census. Implementation and Preliminary Results, Hanoi, August 2009, 48 p.

Hertz Tome, Winters Paul, Quiñones Esteban, Azzari Carlo, Davis Benjamin, Wage Inequality in International Perspective: Effects of Location, Sector, and Gender, Paper presented at the FAO-IFAD-ILO Workshop on Gaps, trends and current research in gender dimensions of agricultural and rural employment: differentiated pathways out of poverty, Rome, 31 March – 2 April 2009.

Hussmanns Ralf, Measuring the informal economy: from employment in the informal sector to informal employment, Working Paper, n° 53, December 2004, p. 42.

International Labour Organization, « Guidelines Concerning a Statistical Definition of Informal Employment », in Seventeenth International Conference of Labour Statisticians, Geneva, 2003, 3 p. + appendices.

Jonasson, Erik, Earnings Differentials in the Rural Labour Market: Does Non -Agricultural Employment Pay Better? Working Papers, n° 7, 2008, 21 p.

Koenker Roger, « Quantile Regression for Longitudinal Data », in Journal of Multivariate Analysis, n° 91, 2004, p. 74-89.

Lanjouw Peter, « Does the Rural Non-farm Economy Contribute to Poverty Reduction? », in Transforming the Rural Non-farm Economy: Opportunities and Threats in the Developing World, Haggblade Steven, Hazell Peter et Reardon Thomas (Éds), Baltimore, John Hopkins University Press, 2008, p. 55-80

Lanjouw, Peter, Murgai, Rinku, « http://ideas.repec.org/p/wbk/wbrwps/4858.html », Agricultural Economics, 40, 2009, p. 243-263.

Lewis Artur, « Economic Development with Unlimited Supplies of Labour », in Manchester School, n° 2/28, Mai 1954, p. 139-191.

Magda, Iga, Wage Mobility in Times of Higher Earnings Disparities: It is Easier to Climb the Ladder?, ISER Working Paper Series, Institute for Social and Economic Research, 2008, 25 p.

Maloney William, « Does Informality Imply Segmentation in Urban Labor Markets? Evidence from Sectoral Transitions in Mexico », in World Bank Economic Review, n° 2/13, 1999, p. 275-302.

Maloney William, « Informality Revisited », in World Development, n° 7/32, 2004, p. 1159-1178.

Mccaig Brian, Benjamin Dwayne, Brandt Loren, The Evolution of Income Inequality in Vietnam, 1993-2006, unpublished manuscript, Australian National University and University of Toronto, 2009, 67 p.

McCulloch Neil, Weisbrod Julian, Timmer Petter, Pathways out of Poverty During an Economic Crisis: An Empirical Assessment of Rural Indonesia, Washington, World Bank, 2007, p. 50.

Nguyen, Huu Chi, Nordman, Christophe J., Household Entrepreneurship and Social Networks in Vietnam: Evidence from Panel Data, DIAL Research Paper DT/2014/22, 2014.

Nguyen Huu Chi, Nordman Christophe, Roubaud François, Who Suffers the Penalty? A Panel Data Analysis of Earnings Gaps in Vietnam, IZA Discussion Paper Series, n° 7149, Janvier 2013, p. 47.

Nordman Christophe, Robilliard Anne Sophie et Roubaud François, «Decomposing Gender and Ethnic Earnings Gaps in Seven cities in West Africa », in Urban Labour Markets in Sub-Saharan Africa, Philippe De Vreyer et Roubaud François (éd.), Africa Development Forum, 2013, p. 271-295.

Nordman Christophe, Wolff François-Charles, « Gender Differences in Pay in African Manufacturing Firms », in Gender Disparities in Africa's Labor Market, François-Charles Wolff, Christophe Jalil NOrdman ed., Washington D.C., World Bank, 2010, pp. 155-192.

Phung Du et Nguyen Phong, Vietnam Household Living Standards. Survey (VHLSS) 2002 and 2004 Basic Information, Vietnam General Statistics Office, unpublished manuscript, Hanoi, 2006, 47 p.

Rand John, Torm Nina, « The Benefits of Formalization: Evidence from Vietnamese SMEs», in World Development, vol. 40, n°5, 2012, pp. 983-998.

Roubaud François, L'économie informelle au Mexique : de la sphère domestique à la dynamique macro-économique, Paris, Karthala/Orstom, 1994, 453 p.

Söderbom Mans, Teal Francis, Wambugu Anthony, « Unobserved Heterogeneity and the Relation between Earnings and Firm Size: Evidence from Two Developing Countries », in Economics Letters, n° 2/87, 2005, p. 153-159.

Strobl Eric, Thornton Robert, Do Large Employers Pay More in Developing Countries? The Case of Five African Countries, IZA Discussion Paper, n° 660, 2002, 35 p. + appendices.

Vietnam Academy of Social Sciences, “Rapid Assessment of the Social Impacts of the Global Economic Crisis in Vietnam: Summary of First Round Research”, Oxfam Discussion Paper, July, 2010, 12 p.

Winters Paul, De la O Ana Paula, Quiñones Esteban, Hertz Thomas, Davis Benjamin, Zezza, Alberto, Covarrubias Katia, K Carletto, Stamoulis Kosta, Rural Wage Employment and Household Livelihood Strategies: A Multi-Country Analysis, ESA Working paper, unpublished paper, 2008, 36 p. + appendices.

Annexes

Table 6- Mean Earnings Regressions for Rural Workers in Vietnam

Dependent Variable: Log Hourly Real Earnings - VHLSS 2002-2004-2006

(1)

(2)

(3)

(4)

(5)

VARIABLES

Pooled OLS

Pooled OLS

Pooled OLS

Fixed Effects

Fixed Effects

Formal

0.397***

0.270***

(0.021)

(0.043)

Informal

0.244***

0.160***

(0.015)

(0.034)

Wage_inf

0.114***

0.079**

(0.016)

(0.036)

Wage_for

0.252***

0.189***

(0.022)

(0.050)

Self_inf

0.351***

0.271***

(0.019)

(0.045)

Self_for

0.653***

0.403***

(0.032)

(0.060)

Yearsch

0.021***

0.019***

0.006

0.027

0.027

(0.006)

(0.006)

(0.006)

(0.017)

(0.017)

Yearsch2

0.002***

0.001***

0.002***

-0.001*

-0.001

(0.000)

(0.000)

(0.000)

(0.001)

(0.001)

Expe

0.014***

0.015***

0.013***

0.001

0.001

(0.002)

(0.002)

(0.002)

(0.012)

(0.012)

Expe2

-0.000***

-0.000***

-0.000***

0.000

0.000

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

Female

-0.104***

-0.109***

-0.149***

(0.013)

(0.013)

(0.013)

Married

0.128***

0.133***

0.117***

-0.024

-0.029

(0.018)

(0.018)

(0.018)

(0.049)

(0.049)

Year 2004

0.328***

0.241***

0.241***

0.262***

0.265***

(0.016)

(0.015)

(0.015)

(0.020)

(0.021)

Year 2006

0.421***

0.407***

0.411***

0.430***

0.434***

(0.016)

(0.016)

(0.016)

(0.034)

(0.034)

Constant

0.497***

0.399***

0.514***

0.968***

0.977***

(0.037)

(0.038)

(0.038)

(0.266)

(0.266)

Observations

11498

11498

11498

11498

11498

R-squared

0.181

0.211

0.229

0.151

0.155

Number of id

6425

6425

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Source: GSO VHLSS 2002, 2004, 2006, author’s calculations

Table 7- Pooled Quantile Earnings Regressions for Rural Workers in Vietnam. Dependent Variable: Log Hourly Real Earnings - Vietnam VHLSS 2002-2004-2006. Source: GSO VHLSS 2002, 2004, 2006, author’s calculations

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

VARIABLES

Pooled .10

Pooled .25

Pooled .50

Pooled .75

Pooled .90

Pooled .10

Pooled .25

Pooled .50

Pooled .75

Pooled .90

Formal

0.472***

0.433***

0.402***

0.332***

0.322***

(0.041)

(0.027)

(0.025)

(0.031)

(0.042)

Informal

0.459***

0.335***

0.242***

0.115***

0.068**

(0.032)

(0.020)

(0.016)

(0.021)

(0.030)

Wage_inf

0.505***

0.316***

0.150***

-0.076***

-0.315***

(0.041)

(0.022)

(0.020)

(0.023)

(0.031)

Wage_for

0.446***

0.348***

0.280***

0.139***

-0.016

(0.040)

(0.029)

(0.025)

(0.029)

(0.043)

Self_inf

0.412***

0.357***

0.370***

0.356***

0.314***

(0.038)

(0.028)

(0.024)

(0.031)

(0.032)

Self_for

0.630***

0.687***

0.686***

0.658***

0.630***

(0.092)

(0.049)

(0.032)

(0.057)

(0.045)

Yearsch

-0.015

-0.012

0.005

0.045***

0.078***

-0.018*

-0.020**

-0.007

0.022***

0.033***

(0.011)

(0.008)

(0.006)

(0.008)

(0.012)

(0.010)

(0.008)

(0.005)

(0.008)

(0.010)

Yearsch2

0.003***

0.003***

0.002***

-0.000

-0.002***

0.003***

0.003***

0.003***

0.002***

0.001*

(0.001)

(0.000)

(0.000)

(0.000)

(0.001)

(0.001)

(0.001)

(0.000)

(0.000)

(0.001)

pe

0.014***

0.017***

0.018***

0.018***

0.017***

0.013***

0.016***

0.016***

0.016***

0.015***

(0.004)

(0.003)

(0.002)

(0.002)

(0.003)

(0.003)

(0.003)

(0.002)

(0.003)

(0.004)

Expe2

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

-0.000***

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

Female

-0.207***

-0.168***

-0.121***

-0.058***

-0.035

-0.203***

-0.184***

-0.166***

-0.148***

-0.148***

(0.024)

(0.015)

(0.013)

(0.019)

(0.026)

(0.026)

(0.019)

(0.015)

(0.018)

(0.024)

Married

0.141***

0.097***

0.085***

0.136***

0.195***

0.158***

0.100***

0.083***

0.093***

0.118***

(0.034)

(0.029)

(0.019)

(0.024)

(0.033)

(0.036)

(0.028)

(0.021)

(0.025)

(0.038)

Year 2004

0.367***

0.303***

0.202***

0.142***

0.146***

0.357***

0.296***

0.201***

0.137***

0.150***

(0.031)

(0.022)

(0.017)

(0.021)

(0.031)

(0.032)

(0.021)

(0.017)

(0.018)

(0.025)

Year 2006

0.468***

0.423***

0.336***

0.347***

0.381***

0.475***

0.427***

0.342***

0.321***

0.358***

(0.030)

(0.022)

(0.018)

(0.021)

(0.030)

(0.028)

(0.024)

(0.020)

(0.021)

(0.027)

Constant

-0.352***

0.082

0.450***

0.715***

1.075***

-0.357***

0.135**

0.568***

0.911***

1.346***

(0.072)

(0.050)

(0.037)

(0.048)

(0.066)

(0.075)

(0.055)

(0.042)

(0.049)

(0.066)

Observations

11498

11498

11498

11498

11498

11498

11498

11498

11498

11498

Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1

Table 8- Fixed Effects Quantile Earnings Regressions for Rural Workers in Vietnam Dependent Variable: Log Hourly Real Earnings - Vietnam VHLSS 2002-2004-2006. Source: GSO VHLSS 2002, 2004, 2006, author’s calculations

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

VARIABLES

FE .10

FE .25

FE .50

FE .75

FE .90

FE .10

FE .25

FE .50

FE .75

FE .90

Formal

0.386***

0.314***

0.270***

0.229***

0.154***

(0.023)

(0.016)

(0.000)

(0.014)

(0.028)

Informal

0.259***

0.201***

0.160***

0.123***

0.066***

(0.019)

(0.013)

(0.000)

(0.013)

(0.020)

Wage_inf

0.244***

0.153***

0.079***

0.019

-0.103***

(0.021)

(0.014)

(0.000)

(0.015)

(0.021)

Wage_for

0.355***

0.254***

0.188***

0.129***

-0.004

(0.026)

(0.020)

(0.000)

(0.017)

(0.026)

Self_inf

0.309***

0.282***

0.271***

0.263***

0.232***

(0.023)

(0.018)

(0.000)

(0.017)

(0.020)

Self_for

0.450***

0.431***

0.403***

0.396***

0.354***

(0.025)

(0.026)

(0.000)

(0.026)

(0.032)

Yearsch

0.001

0.020***

0.027***

0.037***

0.051***

0.007

0.021***

0.027***

0.035***

0.043***

(0.007)

(0.004)

(0.000)

(0.004)

(0.008)

(0.008)

(0.004)

(0.000)

(0.005)

(0.007)

Yearsch2

0.000

-0.001***

-0.001***

-0.002***

-0.003***

-0.000

-0.001***

-0.001***

-0.002***

-0.002***

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

Expe

-0.001

-0.000

0.001***

0.002

0.004*

0.000

0.000

0.001***

0.002

0.003

(0.002)

(0.002)

(0.000)

(0.002)

(0.002)

(0.002)

(0.002)

(0.000)

(0.001)

(0.002)

Expe2

0.000

0.000

0.000***

0.000

0.000

0.000

0.000

0.000***

0.000

0.000

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

(0.000)

Married

0.005

-0.027*

-0.024***

-0.020

-0.029

0.006

-0.027*

-0.029***

-0.033**

-0.041**

(0.022)

(0.016)

(0.000)

(0.013)

(0.021)

(0.022)

(0.015)

(0.000)

(0.015)

(0.020)

Year 2004

0.327***

0.261***

0.262***

0.214***

0.214***

0.314***

0.260***

0.264***

0.223***

0.226***

(0.021)

(0.012)

(0.000)

(0.012)

(0.018)

(0.019)

(0.013)

(0.000)

(0.012)

(0.016)

Year 2006

0.465***

0.420***

0.430***

0.381***

0.401***

0.455***

0.419***

0.433***

0.382***

0.410***

(0.019)

(0.013)

(0.000)

(0.013)

(0.022)

(0.020)

(0.014)

(0.000)

(0.013)

(0.019)

Constant

0.552***

0.843***

0.962***

1.113***

1.314***

0.519***

0.842***

0.986***

1.152***

1.418***

(0.044)

(0.031)

(0.000)

(0.028)

(0.048)

(0.046)

(0.031)

(0.000)

(0.032)

(0.052)

Observations

11498

11498

11498

11498

11498

11498

11498

11498

11498

11498

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Notes

1 The definition of this so-called informal/formal employment divide is explained below (see 1.2).

2 The primary sample units are the communes/wards, the secondary sample units are the census enumeration areas or villages and the tertiary sample units correspond to households. For more details, see Phung and Nguyen (2006).

3 We should be cautious to consider that all the jobs held by rural workers are rural employment since there should be a considerable number of rural workers undertaking jobs based on temporary and circular movement to cities. As a consequence, we consider in the analyzed sample the rural workers rather than rural employment.

4 It is important to note that unemployment rate is stubbornly low in Vietnam’s rural area and this phenomenon is less prevalent in rural areas than in urban areas.

5 We observe a slight difference between these results and those presented in Table 3 concerning the earning hierarchy of different groups. This is due to the different samples used: estimated average earnings in this table concern earnings in 2006, whereas the figures in Table 3 come from pooled sample of all the three waves.

6 The definitive assessment is even more complex as measurement errors in incomes are usually considered as more important for self-employed than for wage workers, the former not knowing their precise level of income (especially informal workers who do not have book accounts), and the richest ones tending to understate their level of activity.

7 For Africa, see Nordman, Robilliard and Roubaud (2010) for estimates of the gender earnings gap in the formal and informal sectors of different West African capital cities using household surveys, and Nordman and Wolff (2010) for formal sector gender earnings gaps using matched worker-firm datasets for seven African countries.

Auteur

UNE, CEPN Université Paris Nord

© Presses universitaires de Paris Nanterre, 2016

Conditions d’utilisation : http://www.openedition.org/6540

Acheter