Version classiqueVersion mobile
OpenEdition Books

Mutations démographiques et sociales du Viêt Nam contemporain

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

Exploring the heterogeneity of informal household businesses in Vietnam: from macro dynamics to micro characteristics and functioning

Le Thi Thuy Linh, Mireille Razafindrakoto et François Roubaud

Texte intégral

1Despite its economic weight, knowledge of the informal economy is extremely limited in Vietnam as it is in most developing countries. Researchers, whether Vietnamese or foreign, have paid little attention to the subject. This situation is due to a number of factors. First of all, the concept of what constitutes “informal” is vague with a multitude of definitions having been put forward by different authors. Second, measuring the informal economy is a difficult task since it operates on the fringes of the economy. Third, the informal economy is often neglected by the authorities as it does not pay (or pays little) taxes and is seen more as a nuisance (especially in the towns) and a mark of underdevelopment inevitably doomed to extinction by the country’s economic growth. These elements explain why there has been no real significant effort to date to improve knowledge in this area. Our work sets out to amend this situation by providing accurate statistical data and in-depth analyses on the informal sector and informal employment in Vietnam. It draws on the results of several statistical surveys conducted with support from the authors and largely refers to different studies recently published on this subject (Cling et coll., 2010a & 2012; Demenet et coll., 2010; Nguyen Huu Chi et coll., 2010; Razafindrakoto et coll., 2011; Razafindrakoto et coll., 2012; Nguyen Huu Chi et coll., 2013).

2Prior to 2007, statistical data on the informal economy (in terms of labor, income and production) in Vietnam was scarce. Acknowledging this, the General Statistics Office (GSO) launched a joint research project with the French Institute of Research for Development (IRD-DIAL) in 2006. The prime objective was to set up a statistical system that would measure Vietnam’s informal sector and informal employment in a comprehensive and sustainable way, and in-keeping with international recommendations.

3First, an operational definition of both the informal sector and informal employment has been adopted. 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 license) and are engaged in non-agricultural activities. Informal employment is defined as employment with no social security (especially social insurance). All employment in the informal sector is thus considered to be informal employment, as is part of the employment in the formal sector. In keeping with the ILO (2003), both the informal sector and informal employment are defined as belonging to the informal economy.

4Second, in line with these definitions, data collection and analysis providing sound statistical indicators of the informal economy have been conducted, following the recommended two-phase (or mixed household/enterprise) survey methodology. The Labor Force Survey (LFS) has been redesigned to capture accurately employment in the informal sector and informal employment. The LFS has been implemented nationwide in 2007, 2009 and 2010. Additionally, a specific Household Business & Informal Sector Survey (HB&IS) was grafted on to the LFS and carried out by interviewing household business heads identified by the LFS in 2007 and 2009 in Hanoi and Ho Chi Minh City. In 2009, open-ended questions were added to capture more precisely the advantages and disadvantages that the heads of household business (HB) put forward regarding their activity. Thus the survey has information from household businesses which are representative of the informal sector in the two cities.

5This chapter aims at investigating the characteristics of the production units which operate in the informal sector in Vietnam. One of the main questions is to what extent all informal Household businesses (IHB) constitute a homogeneous group. In the economic literature, there exist three dominant schools of thought which characterize the informal sector: the Dualist, the Structuralist and the Legalist schools. The “dualist” approach is an extension of the work by Lewis (1954) and Harris and Todaro (1970). It is based on a dual labor market model where the informal sector is considered to be a residual component of this market totally unrelated to the formal economy. It is a subsistence economy that only exists because the formal economy is incapable of providing enough jobs. Unlike the dualist school, the “structuralist” approach focuses on the interdependencies between the informal and formal sectors (Moser, 1978; Portes et coll., 1989). Under this neo-Marxist approach, the informal sector is part of, but subordinate to the capitalist system; by providing formal firms with cheap labor and products, the informal sector increases the economy’s flexibility and competitiveness. The “legalist” or “orthodox” approach considers that the informal sector is made up of micro-entrepreneurs who prefer to operate informally to evade the economic regulations (de Soto, 1989); this liberal school of thought is in sharp contrast to the other two in that the choice of informality is voluntary due to the exorbitant legalization costs associated with formal status and registration.

6In fact, the informal sector presents a “multi-segmentation” phenomenon, whereby a number of very different categories of IHBs co-exist. In earlier work (Cling et coll., 2010) based only on quantitative data, we have identified three specific IHB groups. The first is the “Survivors” (about 40% of the total) which are the most precarious and insecure, and most of the heads of these IHBs have ended up in this business because they could not find a job elsewhere. The second is the “Resourceful” (51% of the total), which are better off and most of the IHBs in this group were created for reason not related to labor market constraints. The third is the “Professionals” (10% of the total), which constitute the high-end group and for almost all of these IHBs, their creation stems from their head's desire to be their own boss.

7Thanks to the richness of the data collected in Hanoi and Ho Chi Minh City in 2009, the objective of this chapter is to combine qualitative and quantitative approaches, using in particular open-ended questions, in order to explore further the heterogeneous nature of the informal sector.

8The chapter is structured as follows. Section 1 gives an overview of the context, in particular the recent trends in the labour market in Vietnam and the weight of the informal sector. In order to explore the heterogeneous nature of the informal sector, Section 2 presents the methodology and some descriptive statistics derived from the open-ended questions. Section 3 discusses the results of the textual correspondence and cluster analysis.

The labor market in Vietnam: the context

9This section presents the general trends in the labour market (in normal time) and analyses the extent of the impact of the international crisis in 2008 and 2009. Similar to other Asian countries, Vietnam has been affected by the international crisis and economic growth has slowed down. During 2004-07, the average annual growth rate was 8.5%, which decreased to 6.5% in 2008 and 5.3% in 2009. However, the impact has been relatively moderate compared to other countries in the region. Along with China, Vietnam has been one of the few Asian countries to not go into recession in 2009. Empirical evidence from the LFS suggests that the Vietnamese labour market reacted remarkably well during the crisis.

The flexibility of the labor market in Vietnam

10There was no increase in global unemployment rates (with a structurally low level of 2.0% in 2007 and 1.7% in 2009, too small a change to be statistically significant). But while the global level was stable, a significant downward trend can be noticed in the youth unemployment rates from 2007 to 2010 (a decrease from 3.6% to 1.6% in urban areas for the age group 15-24 years).

11While the main structures of the labor market remained unaffected overall, the principal variable of adjustment during the slowdown has been the working hours and the multi-activity. The economic slowdown resulted in a drop in the average number of hours worked (from 43.9 to 42.6 hours per week between 2007 and 2009) and by a rise in part-time employment (less than 35 hours per week): 13.2% of workers were concerned by this in 2007 and 26.7% in 2009. Paradoxically, this average evolution also went hand in hand with a lengthening of working hours for the most vulnerable part of the population, resulting in another form of “invisible” under-employment: the percentage of the labor force working more than 60 hours per week rose from 5.6% to 9.3% in two years. Finally, a very high rise in the number of workers with more than one job was observed: having more than one job constituted a strategy to compensate for the reduction in working hours by seeking an alternative source of income. The percentage of workers with more than one job thus rose from 18.2 in 2007 to 25.4 in 2009 and 27.3 in 2010.

12Indeed, the flexibility of the labor market in Vietnam (both in the formal and informal sector) allows to mitigate the negative impact of the global crisis. The informal sector plays an important role particularly in absorbing the shocks at the macro level. However, at the individual level, the affected workers and households have fully endured the negative impact of the crisis (Cling et coll., 2010; Demenet et coll., 2010). More structurally, the predominance of informal sector jobs and informal employment, characterized by poor labor conditions, is an issue that needs to be addressed.

A resilient job’s distribution by institutional sector

13The distribution by institutional sectors shows that the informal sector represents 24% of the total employment in 2009 in Vietnam (Table 1). The number of jobs in this sector grew between 2007 and 2009 (500,000 new jobs, which corresponds to a 4.9% increase). But in general, the overall structure did not change significantly with the crisis. Agricultural jobs, which represented 48% of the total employment in 2009, continued their declining pattern (-2.3 percentage points compared to 2007), in line with previous trends. Public jobs also reduced their share (less than 10% in 2009), also consistent with the previous trends. The informal sector and, more surprisingly, the private formal sector benefited from the shrinking share of the two above mentioned institutional sectors. The foreign direct investment (FDI) sector experienced the highest job growth with an increase of 52% (mainly in rural areas), even if it still accounts for less than 3% of the labor force. Domestic enterprises follow, with one million additional jobs (40% and a 2 percentage point increase in the distribution), and finally the formal household business jobs, witnessing an increase of 100,000 new jobs in two years.

Table 1- Employment by institutional sector and area, 2007 & 2009 (%)

2007

2009

Urban

Rural

Total

Urban

Rural

Total

Public

23.8

6.1

10.5

20.2

5.7

9.7

Foreign Enterprise

3.4

1.5

2.0

3.8

2.5

2.9

Domestic Enterprise

11.6

3.8

5.7

14.5

5.1

7.7

Formal HB

16.9

4.7

7.7

15.1

5.0

7.8

Informal sector

31.5

20.8

23.4

31.6

20.7

23.7

Agriculture

11.1

63.0

50.4

14.7

60.9

48.1

Total

100.0

100.0

100.0

100

100.0

100.0

Sources: LFS, 2007 & 2009, GSO. Total: Occupied population; authors’ calculation.

The informal economy will still be predominant in the following years

14At the national level, informal employment represented around 80% of the total jobs in 2006-2010 (table 2). The rate of informal employment varies a great deal among sectors, obviously higher in the informal sector and agriculture. But formal sectors are not spared (9% of employment in the public sector and 38% in domestic enterprises, which are still relatively high even if a downward trend can be noticed).

Table 2- Informal employment in the main job by institutional sector, 2007 and 2010

Number

Informal employment by institutional sector (%)

of jobs

(1,000)

(%)

Public sector

Foreign enterprise

Domestic enterprise

Formal

HB

Informal sector

Agricul-ture

2007

37,705

81.9

12.3

17.2

52.9

48.0

100

99.0

2009

38,288

80.5

12.6

12.9

48.0

51.6

100

98.6

2010

39,539

79.1

9.2

11.4

38.0

52.5

100

98.5

Note: In 2010, 79.1% of the jobs (those declared as the main job by Vietnamese workers) are informal. When each institutional sector is considered, 11.4% of the jobs in foreign enterprise are informal. By definition, 100% of the jobs in the informal sector are informal.

Source: LFS 2007, 2009 & 2010, GSO; authors’ calculations.

15In the long term, it is expected that a country’s development is accompanied by a progressive reduction of the weight of its informal sector, in conformity with the observation of the marginal weight of this sector in developed countries (La Porta et coll. 2008). However, this mechanism only works in the long term, as stated by Bacchetta et coll. (2009). Given the rapid rates of growth of the Vietnamese economy since the 1980s and the launch of Doi Moi, one would thus think that the informal sector’s weight on the job market would have tended to diminish somewhat. But in spite of the expansion of the private formal sector, the informal sector will continue to grow in Vietnam, the consequence of rapid agrarian, urban and demographic transition. Forecasts for employment in 2015 (using pre-crisis trends) show that employment in the informal sector and its relative weight in total employment will continue to grow over the next few years (Cling et coll., 2010b). Regarding labor supply, firstly Vietnam is in a period of “demographic dividend” with a huge number of young people arriving on the labor market (more than one million per year) and the situation will last until 2015 (Table 3). At the same time, the growth of the formal private sector, though rapid, is not high enough to absorb all the new arrivals on the labor market (in particular given the progressive decline of agricultural employment, which represents nearly half of the total employment). Even if a return to strong growth scenario is envisaged for the Vietnamese economy, once the effects of the crisis are over, these forecasts suggest that the informal sector will continue to represent a considerable share of employment in the years to come.

Table 3- Projections for the evolution of employment from 2007 to 2015: number and share by institutional sector

2007

(LFS adjusted)

2015

(Projections)

Institutional sector

Number (1 000)

Structure (%)

Number (1 000)

Structure (%)

Public sector

4,921

10.8

4,810

9.1

Foreign Enterprise

902

2.0

2,522

4.8

Domestic Enterprise

2,628

5.7

5,883

11.1

Formal Household Business

3,560

7.8

3,801

7.2

Informal Household Business

10,794

23.6

14,444

27.2

Agriculture

22,957

50.0

2, 570

40.7

Total

45,762

100.0

53,031

100.0

Unemployment

1,043

2.2

1,209

2.2

Active population

46,805

100.0

54,240

100.0

Sources: LFS2007, GSO, RGPH1999-2009, GSO; Projection of population by age, United Nations, 2009. See Cling et coll., 2010b.

Global characteristics of the informal sector compared to other institutional sectors

16Given the weight of the informal sector in terms of employment in Vietnam, the objective is to further investigate its characteristics. The data from the LFS provide some stylized facts (Table 4):

  • The informal sector is characterized by low level of education and low incomes; precarious labor conditions; vulnerability of informal household businesses, which operate almost without capital and mostly without professional premises.

    • 1 See Cling et coll. (2010a) for a more detailed presentation of informal household businesses in Vie (...)

    The informal sector consists mainly of micro-businesses (self-employment) and is not strongly integrated into the rest of the economy. Purchases from and sales to the formal sector are marginal. The main supplier of the informal sector is the informal sector itself. Its main market is households and household businesses; sales to the formal sector and sub-contracting are marginal and IHBs mainly compete with each other.1

Table 4- Characteristics of the workforce and of employment by institutional sector in Vietnam, 2009

Institutional sector

Structure

(%)

Migrant

(%)

Salaried workers (%)

Average monthly income (1 000 VND)

Professional premises (%)

Public sector

9.7

10.4

99.7

1 964

96.4

Foreign enterprise

2.9

32.1

99.9

1 735

97.6

Domestic enterprise

7.7

16.0

93.6

2 093

86.4

Formal household business

7.8

8.4

36.4

1 805

33.8

Informal sector

23.8

5.6

26.7

1 273

7.8

Agriculture

48.0

2.4

9.6

703

1.1

Total

100

6.3

33.6

1 185

23.8

Source: LFS2009, GSO; authors’ calculations.

17Figure 1 represents job satisfaction levels expressed by workers. There is a very clear hierarchy depending on the institutional sector, which consists of three main categories. The public sector is at the top of the ladder: nearly three-quarters (72%) of employees in the public sector (civil servants or salaried workers in public or para-public enterprises) declare themselves to be satisfied or very satisfied with their job. Next are workers from the private formal sector, of which a little more than half (52%) report being at least satisfied, without any significant difference between those who work in foreign, domestic or individual enterprises? Finally, workers in the informal sector and in agriculture are the most critical; the proportion of those declaring being satisfied is around one third, with the informal sector doing relatively better (38% and 29% respectively).

Figure 1- Job satisfaction levels by institutional sector in Vietnam, 2009

Figure 1- Job satisfaction levels by institutional sector in Vietnam, 2009

Note: The balance of satisfaction is defined as the difference between the proportion of respondents having expressed a positive opinion (satisfied & very satisfied) and the proportion of respondents having expressed a negative opinion (unsatisfied & very unsatisfied).

Source: LFS2009, GSO; authors’ calculations.

A quali-quanti approach to explore the heterogeneity of the informal sector in Vietnam

18The following sections aim at furthering the analysis and identifying more precisely the characteristics of the different types of HBs in the informal sector. The main finding is that the informal sector can be considered as a continuum of HBs in the multi-dimensional space. It is far from homogeneous, with different groups of HBs encountering different constraints and advantages as well as having different characteristics. It also appears to be more complex than the dual segmentation view.

Methodology

19We employ a combination of quantitative and qualitative approaches (i.e., so-called quali-quanti approach). The quantitative approach is appreciated for its precision and ability to generalize, while being criticized for its restricted freedom and imposed a priori conceptions of researchers. Meanwhile, the qualitative approach often raises the problem of subjectivity and lack of generalizability. However, it gives the respondents the freedom to express their own experience, thus suggesting new insights (Hughes, 2012). When used together, these methods can be complementary. In this research, the interaction between these two approaches is reflected in both data collection and data analysis. First, a representative statistical survey with random sampling (i.e. a quantitative approach) permits to collect qualitative data through open-ended questions along with quantitative data. Second, qualitative data are quantified by a quantitative analytical tool (i.e., DTM-Vic software). Also, the correspondence and cluster analysis on texts is supplemented by categorical variables. Moreover, we use the findings from one type of study to check against the findings derived from the other type (i.e., “logic of triangulation” - Bryman, 2006).

20The analyses are performed mainly in DTM-Vic version 5.6, a program developed by L. Lebart, which is devoted to the exploratory analysis of multi-dimensional data. The basic tools that are employed in this chapter are correspondence and clustering analysis.

  • 2 More precisely, we use multiple correspondence analysis (MCA) which permits to consider qualitative (...)

21Correspondence analysis (CA)2 is a multivariate statistical technique that is conceptually similar to Principal Component Analysis (PCA). It involves a mathematical procedure that transforms a number of possibly correlated variables into a smaller number of uncorrelated variables called principal components. The components are created to account for maximal variation among the original variables, i.e., the first principal component accounts for as much of the variability in the original variables as possible, and each succeeding component accounts for as much of the remaining variability as possible (Lebart et coll., 2006).

  • 3 Agglomerative clustering algorithm starts with one-point (singleton) clusters and recursively merge (...)

22Clustering is a division of data into groups of similar objects. Clusters are formed such that HBs in the same cluster are as similar to each other as possible, and that HBs in different clusters are as distinct as they can be (Wanner, 2004). Clustering can be performed on the principal components of PCA “to denoise the data” (Husson et coll., 2010). In DTM-Vic version 5.6, clustering is based on a hybrid method using both bottom-up hierarchical (agglomerative) and k-means clustering (Lebart et coll., 2006).3

23According to Lebart and Mirkin (1993), CA and clustering are practically complements and it seems wiser to use both. They argue that CA could result in shrinkages and distortions due to both the projection onto the principal dimensions and the possible lack of robustness of the global fit (sensitivity to outliers). It is thus desirable to complement it with a classification performed in the whole space. Being derived in a much higher dimensional space allows them to provide the information that could have been obscured by the projection onto a low dimensional subspace. Moreover, it is much easier to describe a set of clusters than a continuous space.

24Interestingly, these techniques can be applied to textual analysis. This extension requires a transformation of texts into numerical data. This step cross-tabulates a lexical table with columns being distinct words used and rows being HBs (i.e., “contingency table”). The cell value is thus the frequency with which a certain word appears in a response (Lebart et coll., 1998; Yelland, 2010). Correspondence and clustering analyses are then implemented just as with categorical data. In other words, results of these data analyses are based only on word frequency, and other aspects of the corpus (context, discursive aspect) are not taken into account in CA. However, our interpretation of results is done based on the semantic aspect of words. Categorical data (from closed question in the survey) are also considered in order to characterize each cluster. A v-test is computed in order to assess whether the percentage of respondents in a class holding a modality is significantly different from that of the whole sample or not.

  • 4 See also Pham et coll. (2008) for an overview of the Vietnamese language.

25Economic research employing these text mining techniques seems to be rare. It is even non-existent for Vietnamese texts. Until now, statistical programs such as DTM-Vic have not been able to work properly with Vietnamese. This is due to specific features of this language as explained below:4

  • The Vietnamese alphabet contains special letters that are non-existent in the classical Latin alphabet such as ê, ă, â, ô, ơ, ư, đ.

  • Vietnamese has six tones: “level,” “hanging,” “sharp,” “asking,” “tumbling,” and “heavy.” They are expressed by diacritic marks. Words with similar letters may have totally different meanings while marked by different tone marks. For example, “tôi” (I, myself) changes its semantic content to be “bad” (tồi), “dark” (tối) or “crime” (tội). Therefore, ignoring those marks in the analyses is impossible.

  • Unlike English or French, where “word is a single distinct meaningful element of speech or writing, used with others (or sometimes alone) to form a sentence and typically shown with a space on either side when written or printed” as stated in the English Oxford dictionary, in Vietnamese and other Asian languages, whitespaces are not used to identify the word boundaries. Indeed, Vietnamese is monosyllabic in nature. Each “syllable” (i.e., “tiếng”) tends to have its own meaning and is written with a space before and after. However, "Vietnamese words" are often made of two or more syllables. Thus, morphosyllables in Vietnamese are considered to have the status of morphemes (Tran et coll., 2007). Words made from two or more morphemes can only be recognized by context. DTM-Vic may consider morphemes as words, thus misunderstanding their implications. For example, in the sentence: “Gia đình tôi gặp khó khăn về tài chính” (My family faces financial difficulties), “tài” could imply “talent,” “chính” separately means “main,” but a combination of “tài chính” turns out to be “finance.” In this context, it must be the whole “tài chính” as a single word to be meaningful.

26Cutting-edge research in linguistics and informatics has developed various methods to tackle the issue of word segmentation in Vietnamese, such as Dinh et coll. (2001), Nguyen, Cam Tu et coll. (2006) and Nguyen, Thanh et coll. (2006). Yet, none of them are perfect. Moreover, the first two problems remain in DTM-Vic. Taking into account the sample size and the response length; we decided to code the texts manually in a way that addresses the aforementioned problems:

  • The alphabet problem (i.e., “chữ cái”) was solved by replacing special letters by a pair of Latin letters that imply no other meaning in Vietnamese: ă = aw; â = aa; đ = dd; ê = ee; ô = oo; ơ = ow; ư = uw

  • The tone marks problem (“dấu”) was tackled by the same logic: placing a specific Latin letter, which never ends a word in Vietnamese, for each tone: ` (“hanging”) = f, ´ (“sharp”) = s, (“asking”) = r , ~ (“tumbling”) = x, . (“heavy”) = j. The idea to deal with the two first problems comes from the Vietnamese typing rule with the English keyboard of Vietkey and Unikey software programs.

  • Separate morphemes of a compound word are identified and linked with each other, making compound words continuous strings of pure Latin letters. E.g., “cạnh tranh” = canhjtranh (competition), “rảnh rỗi” = ranhrroix (free, idle)

Descriptive statistics

  • 5 The two open-ended questions were analyzed in parallel but separately to avoid misinterpretation (g (...)

27The textual data mobilized for the analysis are the precise answers provided by the head of HBs to two open-ended questions: on advantages and disadvantages they want to emphasize upon regarding their activity.5 The qualitative data reveal copious information that not only directly answers the questions asked, but also provides explanations, inferences and attitudes of the HBs. One HB, for example, puts forward difficulty due to price increase, after explaining its causes and mechanisms: “Gold price increases, gasoline price goes up day by day so goods prices follow the trend.” Another HB considers that their grocery store is impossible to sell because their main customers are manual workers, who are unemployed at the moment. Meanwhile, some HBs report very job-specific problems such as: “There was once a customer who ate our sweetened porridge and had a belly-ache, she laid the blame on us and scolded us loudly.” This kind of information is hardly available in a quantitative survey.

28Basic statistics resulting from numerical coding of texts are provided in Table 5. We report the number of responses (i.e. observations or head of HBs who have provided answers) to each question and count the total number of words and distinct words contained in the responses. We then remove all the words with frequency equal or less than two, as these words contribute little to the text corpus. Table 5 shows that the response rate of the “disadvantage” question is significantly higher than that of the “advantage” question (72.2% versus 61.9%). Non-responses appear to be random, as most of the key variables (i.e., HB size, value-added, educational attainment of HB heads) are very similar between the respondents and the non-respondents (Table 14 in appendix). Table 5 also indicates that the number of words and distinct words used by HBs to present their disadvantages is about 50% higher than for advantages. These figures, to some extent, reflect the low level of satisfaction of HBs in the informal sector. This result is consistent with the findings mentioned at the end of the first section.

Table 5- Numerical Coding of Texts: summary results

Disadvantages

Advantages

Total number of responses

505

433

Total number of words*

5,188

3,498

Number of distinct words

867

585

Frequency threshold chosen

2

2

Kept words

4,551

3,045

Distinct kept words

336

229

* The same word might be counted many times as it can be used by different respondents.

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

29A closer look at the responses gives us a list of the most frequent words used by HBs (Table 6). In terms of advantages, “home” matters the most. Two possible explanations can be put forward: either a HB owner has a premise at home and she saves rent, or she is happy working at home to take care of her family. The first explanation seems to be more plausible since “premise” and “location” also appear very frequently. “Stability” is also one of the main advantages revealed, which reflects the risk-aversion of these HBs.

30As far as disadvantages are concerned, the HB&IS survey already includes a close-ended question asking the main difficulties faced by the HBs (Table 7). A comparison of the results from these two types of questions could give an insight into the methodological problem regarding question choice (interesting examples are provided in Piau, 2004).

Table 6- Most Frequent Words Used by Informal HBs

Disadvantages

Frequency

Advantages

Frequency

Customers

102

Home

78

Prices

79

Customers

67

Capital

56

Stable

57

Location

35

Premise

54

Competition

32

Location

50

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

31As suggested by the close-ended question (Table 7), “competition” is the most predominant difficulty encountered by the HBs. “Lack of customers” and “access to loan” come in the second and third place. Reasonably, these three issues also appear in the top five produced by textual analysis. However, the order is different. “Customers”, instead of competition, becomes their primary concern. Interestingly, “prices” emerges as a major issue in the open-ended question, whereas it is non-existent in the close-ended question. This may be due to the fact that “prices” is not explicitly mentioned in the latter. It is possibly grafted into “inflation and exchange rate” category, which might appear as an abstract concept to informal HBs. Moreover, this category is positioned towards the end of the list, inducing the “primacy effect”, i.e., when the question is long, respondents tend to choose the first categories (Schuman & Presser, 1996).

Table 7- The Main Difficulties Faced by Informal HBs

The difficulty faced

Percentage

Competition

30.7

Lack of customers

14.7

Access to loan

14.2

Premise, space

12.7

Cash flow

6.5

Machine, equipment

5.4

Material supply

4.6

Crime, theft, disorder

3.4

Inflation, exchange rate

2.8

Transportation

2.3

Other

1.8

Access to land

0.8

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

The heterogeneous nature of the informal sector: Results of multivariate data analysis

Correspondence Analysis on two open-ended questions

Overview

  • 6 We performed also an analysis mixing “advantages” with “disadvantages” in a preliminary step but th (...)

32There are two distinguished groups of variables involved in this analysis. Active variables, which are in this case the two open-ended questions, will contribute to the performance of CA and explain the factors derived from the analysis. In these questions, HBs were asked about the advantages and disadvantages in the course of operating their business. Such questions reveal the satisfaction and well-being of HBs in a much less restricted manner than close-ended questions, thus the analysis based on these textual data may provide richer and interesting results. CA is performed separately for each question to avoid the confusion if we mix “advantages” with “disadvantages.”6 Besides, supplementary variables, which are in this case categorical variables constructed from the quantitative data, are added into the analysis to provide more informative results.

  • 7 The Kaiser-Gutman rule for CA indicates that we should keep the factorial axes whose eigenvalues gr (...)
  • 8 Actually, any axis contributing more than 0.4% (1/229) could be considered for the analysis.

33The explanatory power of the principal components is reflected through their corresponding eigenvalues as well as percentage in the total variance. However, the larger the dimensionality of the original data, the lower the eigenvalues and percentages (Table 8)7. Acknowledging the number of distinct words contained in “advantages” and “disadvantages,” which is equivalent to the number of textual dimensions (229 and 336, respectively), the figures below are relatively large. The first factor of “advantages,” for instance, accounts for nearly 2% of the total variance, whose explanatory power equals to that of four and a half variables in the original data set.8

Table 8- The First Five Eigenvalues of CA

Advantages

Disadvantages

Factor

Eigenvalue

Percentage

Eigenvalue

Percentage

1

0.5848

1.96

0.5114

1.39

2

0.5597

1.88

0.4925

1.34

3

0.5311

1.78

0.4878

1.32

4

0.5093

1.71

0.4578

1.24

5

0.4940

1.66

0.4537

1.23

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

  • 9 Although the other axes are also interpretable (i.e. meaningful), we only focus on the two most imp (...)
  • 10 We eliminate several words that can be considered as “outliers”, i.e., which do not belong to the s (...)

34Hereafter we consider only the two first factors.9 The interpretation of principal axes is based on the words that have extreme contributions to their creation (Tables 9, 10 and 11).10

Advantages

35The first factorial axis clearly differentiates “social” and “non-social” logics (Table 9 and Figure 2). On the negative side, informal HBs take advantage of their social networks such as local authority, friends and family. On the positive side, informal HBs find the advantages right from the nature of their business (i.e., “small” scale, no or little “tax” duty, “premise,” “initiative,” “time” and so on). The second factorial axis contrasts “maximizing” against “satisfying” logics (Table 9 and Figure 2). Some informal HBs, on the one hand, actively “make use of” as many opportunities as they can. This is demonstrated through words such as “try,” “initiative” and economic advantages that they mention like “tax,” “premise,” “production,” “expenses.” On the other hand, other informal HBs seem to lower their ambition to achieve job satisfaction. These people choose less demanding jobs which are “easy,” “light,” “simple” and maybe appropriate with their “health” and “age,” in order to earn a just “sufficient” and “acceptable” living.

Table 9- Selective Points with Extreme Contributions to the First Two Principal Axes CA on “Advantages” of Informal HBs

Axis 1

Axis 2

Words

Translation

Coordinate

Words

Translation

Coordinate

Giúp đỡ

help

-11.85

Giúp đỡ

Help

-15.66

Chính quyền

authority

-10.23

Chính quyền

Authority

-14.85

Địa phương

locality

-8.46

Thuế

Tax

-10.50

Bạn bè

friends

-6.51

Cố gắng

Try

-4.11

Gia đình

family

-5.72

Kinh doanh

Business

-2.34

Quen biết

accquainted

-5.26

Mặt bằng

Premise

-1.58

Cố gắng

Try

-4.47

Tận dụng

make use of

-1.51

Thu nhập

income

-4.27

Sản xuất

Production

-1.33

Quen

familar

-3.19

Kinh phí

fund, expense

-1.26

Nguồn

sources

-3.01

Chủ động

Initiative

-1.21

Thời gian

time

1.80

Tuổi

Age

1.33

Chi phí

costs

1.85

Tạm

Acceptable

1.49

Chủ động

initiative

2.62

Con

Children

1.67

Địa điểm

location

2.83

Ổn định

Stable

1.79

Nhà

home

3.08

Sức khỏe

Health

1.86

Kinh doanh

business

4.43

Sống

Live

1.91

Thuê

hire

4.57

Đủ

Sufficient

1.99

Mặt bằng

premise

5.15

Thời gian

Simple

2.04

Nhỏ

small

6.19

Nhẹ nhàng

Light

2.05

Thuế

Tax

12.82

 

Dễ

Easy

2.35

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

36It is interesting to see the link between HBs’ advantages with the motivation of heads of HBs to set up their business and to choose their business activity. Figure 2 suggests that HBs which were established as a voluntary choice (i.e., to get better income, to have higher profit than other activities) rather than as a constraint, either employment constraint (i.e., because the HB head did not find a job in another HB) or expertise constraint (i.e., because that is the profession that they know) are located towards the bottom right of the dimensional space formed by the two first principal axes. In other words, “voluntary HBs” are less dependent upon personal networks such as family, neighbors and friends in their business operation. They also take advantage of economic opportunities to develop their bus²iness.

Disadvantages

37The first factorial axis distinguishes “entrepreneurship” from “personal” perspectives (Table 10, Figures 3 and 4). Those who are located towards the negative side of the axis are more likely to raise their concerns in an entrepreneurship perspective. These informal HBs stress more economic problems such as “capital,” “finance,” “scale” and “prices of materials.” In contrast, those who are positioned towards the positive side of this axis tend to put forward their difficulties in a personal perspective. Their disadvantages, for instance, are the fact that they have to “get up” “early” in the “morning” to work, which is “strenuous” as far as they are concerned.

38The second factorial axis separates out “internal” and “external” disadvantages: (i) those that are directly linked to the informal HBs themselves and that they can actively improve (e.g., “strenuous,” “finance,” “family” and “production”); and (ii) those that are external and almost cannot be changed by HB’s own actions (e.g., “street,” “petrol,” “dust,” “rain” and so on).

39Given these interpretations, we could expect that those located in the bottom left of the plane view formed by the two first factorial axes (i.e., have negative values on both axes) would be the “high-end” informal HBs and vice versa. They could be more likely to operate on a larger scale, have a better economic performance and more stable working conditions. As a validation of this hypothesis, we project the significant quantitative modalities onto this plane (Figure 4). Impressively, we observe an evolution of informal HBs from the top right to the bottom left: scale is enlarged (from “size=1,” “size=2” to “size3-6”), business turnover increases (from the first quintile to the fifth quintile), business premise is stabilized (from unstable premise to premise at home), their attitude toward their business becomes more optimistic (from thinking that their HB has no future to the reverse), and so on.

Table 10- Selective Points with Extreme Contributions on the First Two Principal Axes CA on “Disadvantages” of Informal HBs

Axis 1

Axis 2

Word

Translation

Coordinate

Word

Translation

Coordinate

Vốn

capital

-7.32

Sớm

Early

-16.33

Tài chính

finance

-7.01

Dậy

get up

-15.89

Vay

borrow

-4.97

Thức

Awake

-13.14

Hỗ trợ

support

-4.88

Sang

Morning

-10.72

Tay nghề

workmanship

-4.48

Vất vả

strenuous

-6.23

Mở rộng

enlarge

-4.43

Vốn

Capital

-5.69

Diện tích

area, surface

-4.26

Tài chính

Finance

-3.76

ổn định

stable

-3.04

Vay

Borrow

-3.06

Quy mô

scale

-3.02

Gia đình

Family

-3.04

Máy móc

machines

-3.00

Sản xuất

production

-2.94

Kinh tế

economy

-2.85

Mở rộng

Enlarge

-2.76

Mặt bằng

premise

-2.84

Diện tích

area, surface

-2.53

Giải tỏa

land clearance

-2.66

Già

Old

-2.4

Vật giá

prices of materials

-2.65

Chồng

Husband

-2.04

Địa điểm

location

-2.52

Con

Children

-1.82

Mất

lose, lost

2.45

Thị trường

Market

2.07

Trung Quốc

China

2.51

Chất lượng

Quality

2.20

Bụi

dust, dusty

2.81

Bụi

dust, dusty

2.25

Xăng

petrol

3.38

Siêu thị

supermarket

2.28

Xe ba bánh

tricycles

3.41

Nắng

Sunny

2.57

Kẹt

(traffic) jams

3.74

Nước

Water

2.57

Chất lượng

Quality

3.85

Mưa

Rain

2.68

Cấm

Ban

4.36

Khách hàng

customers

2.79

Vất vả

Strenuous

4.37

Dọn dẹp

tidy up

2.83

Khách hàng

Customers

4.54

Buýt

Bus

2.92

Đường

road, street

4.94

Kẹt

(traffic) jam

3.53

Sang

Morning

8.31

Xăng

Petrol

4.09

Thức

Awake

9.40

Xe

Vehicle

4.65

Dậy

get up

10.87

Đường

street, road

5.36

Sớm

Early

11.46

Giá

Prices

6.64

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

40A broader view on both informal and formal production units further elucidates the heterogeneity of the urban HBs (Table 11, Figures 5 and 6). It should be noted that the majority of the analyzed sample of non-farm HBs is informal (68%) and that medium, large and formal enterprises are excluded from the HB&IS Survey (Cling et coll., 2010a). For this reason, these results obtained here are very relevant to the informal sector.

41The first factorial axis is mainly on disadvantages which affect personally each HB: It opposes those who face individual constraints (e.g., “get up” “early” in the “morning”) to those who are more integrated into the market economy and put forward business and economic constraints (on the supply side) in terms of “capital”, “premise,” “production,” “interest rate” and “machines”, for examples.

42The second factorial axis appears to be more on disadvantages faced globally by HBs: from economic and market-related difficulties (on the demand side) such as “prices,” “competition,” “market,” and “taxes” to non-economic, non-market factors (e.g., “weather,” “rain,” “season,” “flooded”).

Table 11- Selective Points with Extreme Contributions on the First Two Principal Axes CA on “Disadvantages” for all HBs (Informal and formal)

Axis 1

Axis 2

Word

Translation

Coordinate

Word

Translation

Coordinate

Sớm

early

-13.10

Giá

prices

-10.47

Dậy

get up

-12.57

Cạnh tranh

competition

-7.22

Thức

awake

-10.44

Gia công

processing

-6.5

Sang

morning

-9.74

Đối thủ

competitors

-4.06

Khách

customers

-5.99

Thị trường

markets

-3.31

Xe

vehicle

-5.69

Lợi nhuận

profit

-3.15

Đường

street, road

-4.97

Điện

electricity

-3.05

Vất vả

strenuous

-4.51

Chi phí

costs

-2.85

Cấm

forbid

-4.06

Tiêu thụ

consume

-2.69

Quản lý

manage

-3.67

Dân cư

inhabitants

-2.67

Bụi

dust

-3.42

Thuế

tax

-2.60

Xe ba bánh

tricycles

-3.23

Tay nghề

workmanship

-2.51

Xăng

petrol

-3.16

Sản phẩm

products

-2.45

Kẹt

(traffic) jam

-3.05

Doanh thu

revenue

-2.39

Lề

edge

-2.52

Chất lượng

quality

-2.29

Chật hẹp

narrow

3.14

Bụi

dust

2.96

Quy mô

scale

3.65

Xe ba bánh

tricycles

3.61

Máy móc

machines

4.09

Nước

water

4.02

Lãi suất

interest rate

4.10

Cấm

forbid

4.37

Địa điểm

location

4.32

Nắng

sunny

4.6

Sản xuất

production

4.44

Dọn dẹp

tidy up

5.26

Ngành nghề

industry

4.50

Vất vả

strenuous

5.38

Diện tích

surface area

4.7

Sáng

morning

5.67

Mặt bằng

premise

5.42

Ngập

flooded

6.42

Di dời

move

5.56

Đường

street, road

6.50

Vay

borrow

6.99

Mùa

season

7.37

Mở rộng

enlarge

7.31

Dậy

get up

8.98

Tài chính

finance

8.40

Sớm

Early

9.42

Kinh doanh

business

9.03

Trời

Weather

13.35

Vốn

capital

10.81

Mưa

Rain

15.21

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

43As a further attempt to visualize the heterogeneity of the HBs, we project the significant words and quantitative modalities onto the planes of the first two principal axes above. Figure 6 shows that from the top left of the figure, we have the “low-end,” i.e., the most precarious, the poorest, the least educated, the less integrated into the economy; and at the bottom-right of the figure we have the “high-end” businesses, i.e., the biggest, the best in terms of performance, the more educated, the more embedded in the economy and so on. This figure is the clearest illustration of our argument that the informal sector is a continuum, rather than a homogeneous sector, or a dual segmented phenomenon. Business size, for instance, gradually grows from “size=1” in the top left quarter to “size3-6” and then to “size>6” in the far bottom right of the plane. School attendance of HB head, in the same manner, escalates from primary to secondary school, and finally high school.

Figure 2 - The Space of Informal HB Characteristics (Projection of Illustrative Variables on the First Factorial Plane of CA on Advantages)

Figure 2 - The Space of Informal HB Characteristics (Projection of Illustrative Variables on the First Factorial Plane of CA on Advantages)

Note: The explicit meaning of the categorical variables is provided in the appendix

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

Figure 3 - The Space of Disadvantages Put Forward by Informal HBs (First Factorial Plane of CA)

Figure 3 - The Space of Disadvantages Put Forward by Informal HBs (First Factorial Plane of CA)

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

Figure 4 - Space of Informal HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages.

Figure 4 - Space of Informal HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages.

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

Figure 5 - The Space of Disadvantages Put Forward by HBs (First Factorial Plane of CA)

Figure 5 - The Space of Disadvantages Put Forward by HBs (First Factorial Plane of CA)

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

Figure 6 - The Space of HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages)

Figure 6 - The Space of HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages)

Note: The explicit meaning of the categorical variables is provided in the appendix.

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

44By viewing these characteristics of HBs in parallel with their disadvantages (Figure 5), insightful policy lessons can be drawn: Depending on the types of HBs, the policy to support them is very different. Moving downward from the top left to the bottom right, government first of all should adopt anti-poverty policy for the households, then for the HBs which are not the most precarious but with limited performance, policies should focus more on infrastructure (e.g., roadways) and finally policies for the “professional” ought to be exercised on business environment (i.e., macro-economic management such as taxes, monetary policies to confine inflation, and so on).

Cluster analysis

45Based on the dimensional space generated by CA, cluster analysis is then applied to obtain a typology of informal HBs. Lebart et coll. (2006) propose a test with the null hypothesis assuming that the average coordinate of a cluster on a factor is equal to zero, i.e., the profile of that cluster on this factor does not differ from the profile of the whole sample. If a cluster has a test absolute value greater than or equal to 1.65, the null hypothesis is rejected at 5% level of significance. The number of clusters can be controlled by cutting the dendrogram at a certain height. In this analysis, five distinct clusters of informal HBs are identified, based on their disadvantages.

46Table 12 shows that the profile of the clusters is significantly different from the global (i.e., their test absolute values are greater than 1.65), except the second cluster on the first factorial axis. The position of clusters in the factorial space in visualized in Figure 7. Each cluster is characterized by their representative responses and then depicted by significant quantitative categories (Table 13).

Table 12- Coordinates of Clusters in the Dimensional Space –Disadvantages of Informal HBs

Clusters

Coordinates

Test-values

Size

1

2

1

2

1

260

-0.39

-0.23

-12.6

-7.6

2

103

0.08

0.45

1.35

7.23

3

89

0.58

0.3

8.42

4.42

4

42

0.43

0.57

4.05

5.54

5

10

 

2.7

-3.63

 

12.05

-16.5

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

47The first cluster covers “the impervious HBs,” who are more likely to declare that “There is no difficulty” obstructing their business. More than half of the HBs in this cluster have their premise at home, which appears to be a major advantage as suggested by the previous section. This is the most populous among the five groups, with 260 HBs being classified in this cluster.

48The second cluster comprises “the optimistic HBs,” who are more inclined to put forward their disadvantages in terms of prices, costs and profit. A substantial share of HBs in this group (41%) thinks that there is a future for a business like theirs. HBs in this cluster are more prone to achieve very high economic performance (the fifth quintile of value added).

49The third cluster is made up of “the desolate HBs,” which includes the HBs who are more likely to face a lack of customers. They can be a tailor, who complains that people are increasingly buying cheap ready-made clothes in market-places; or a motorbike taxi driver, who has less and less customers because his vehicle is too old. Many of them have a turnover of the lowest quintile (38%), compared to the average (27%).

Table 13- Categorical Characteristics of the Clusters – Cluster Analysis on Disadvantages of Informal HBs

Characteristic categories

v.test

Percentages

Weight

Prob.

cla/mod

mod/cla

global

Cluster 1

51.49

260

Supplier_individual

2.59

83.33

5.77

3.56

18

0.0048

Premise_home

2.5

57.63

52.31

46.73

236

0.0062

Cluster 2

20.4

103

va_quin5

2.68

37.78

16.5

8.91

45

0.0037

Future_yes

2.64

28.19

40.78

29.5

149

0.0041

Cluster 3

17.62

89

turnov_quin1

2.5

25.19

38.2

26.73

135

0.0061

Cluster 4

8.32

42

Premise_unstable

4.98

16.33

76.19

38.81

196

0

Future_no_will_chang

2.45

15.31

35.71

19.41

98

0.0072

Cluster 5

1.98

10

Future_catb_4

2.87

50

20

0.79

4

0.0021

Note: Cla/mod indicates what share of all individuals with this category is found in this class. Mod/cla shows the percentage of all individuals in this cluster who have a certain modality.

See in the appendix for a detailed presentation of the categories

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

50The fourth cluster is characterized by “the precarious HBs,” who typically complain about slow consumption of their goods due to bad weather (i.e., rain) and urban police expelling street vendors. This is consistent with their characteristic modalities. A majority of HBs in this group (76%) have unstable premises. These have among the most precarious working conditions. Predictably, they are nearly twice likely than the average to conceive that there is no future for their business, and to think of changing activity.

  • 11 For this reason, we do not illustrate this cluster in the factorial plane.

51The fifth cluster covers “the strenuous HBs,” who generally complain that they have to get up very early in the morning to work. Interestingly, up to half of these HBs keep silent about their future prospect, i.e., neither upset nor optimistic. There are only ten HBs classified into this cluster11.

52Overall, the results obtained by the clustering analysis do provide further insights into the continuum of HBs suggested by CA. This continuum is not as “smooth” as a usual continuous variable, since there is no absolute order of different clusters from the low-end to the high-end HBs. We could not tell, for example, which one among cluster 3 and cluster 4 is of lower end than the other. In fact, a HB has a variety of characteristics, some of them may be considered as “high-end,” whereas others are “low-end.” Therefore, it is hard (if not to say impossible), and unnecessary to have an absolute ranking of HBs in the continuum. In practice, government policies should be designed to target different groups of HBs with their own characteristics, which encounter specific types of disadvantages as we can identify in the factorial space.

Figure 7 - Cluster View of Informal HBs on the First Factorial Plane of CA on Disadvantages

Figure 7 - Cluster View of Informal HBs on the First Factorial Plane of CA on Disadvantages

Note: The explicit meaning of the categorical variables is provided in the appendix

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.

Conclusion

53As the informal sector is here to stay, and since there is a strong connection between the informal sector and urban poverty, public policies cannot ignore this sector. Nonetheless, the State’s ambivalent and inconstant attitude to the informal sector constitutes a source of uncertainty that needs to be lifted if the productive effort of informal entrepreneurs is not to be constantly frustrated. In Vietnam, there are currently no policies targeting the informal sector.

54To improve that situation, Vietnamese government should first of all understand the sector’s features and its diversity. This is the aim of our analysis. Based on a unique quali-quanti analytical approach, we acutely point out the heterogeneity of this sector. Textual CA shows that HBs in the informal sector are located along a continuum generated in a multi-dimensional space of their disadvantages or advantages. HBs at the lower end of the continuum are generally poorer, smaller, more precarious and less integrated into the economy. Their owners are less educated and more pessimistic about their business future. These HBs often lower their ambition and choose less demanding work in order to achieve job satisfaction. Yet, low-end HBs still put forwards their constraints in a personal and passive perspective. Reversely, the higher-end HBs have better economic performance, bigger size, more stable working conditions and are more integrated in the economy. Their owners have higher educational attainment and are more optimistic about their business future. These HBs tend to stress their difficulties in an entrepreneurship perspective, for that they can grasp opportunities and actively improve their situation. Furthermore, we conduct a cluster analysis on disadvantages faced by HBs and obtain a typology with five categories of informal HBs, namely: “the impervious,” “the optimistic,” “the desolate,” “the precarious” and “the strenuous.”

55Our research stands out for a number of originalities. Firstly, the analyses are based on representative textual data, which measure the direct utility of HBs in an open and unrestricted manner. Secondly, we employ exploratory analytical methods (i.e, CA and clustering) to understand the underlying structure of these data, which help limit the pre-conceived ideas of researchers. Last but not least, this is the first economic paper that treats properly the linguistic problems associated with Vietnamese texts, for which factor analysis can be applied.

56Nonetheless, some limits still exist in our analysis. First, the codification process is performed manually, which would be impractical for larger data sets. This calls for the development of statistical packages that incorporate cutting-edge linguistics’ and informatics’ achievements in order to properly treat Vietnamese language in a systematic way. Second, textual analysis is sensitive to data cleaning, which necessarily depends on our subjective understanding of texts. CA, in particular, is sensitive to “outliers,” whose identity could be controversial. Future research could accordingly conduct a number of sensitivity tests by modifying the list of words to be deleted (i.e., “stop-words”), grouping different sets of synonyms as single items and removing some potential “outlier” HBs. In addition, we could take advantage of the panel dimension of the data to further explore the dynamics of the informal sector. In addition, given the meaningfulness of the principal axes generated by textual CA, it is possible to employ HBs’ coordinates on these factors as continuous variables in further studies.

57In conclusion, our analysis directly suggests that targeted policies should take into account the heterogeneity of the informal sector in Vietnam. A “one size fits all” scheme would not be appropriate as there is not one single reason for working in this sector, and different categories of informal HBs experience different kinds of difficulties as well as advantages. Instead, a policy package could include, for examples, anti-poverty policies for lower-end HBs and policies related to the business environment for professional and higher-end HBs; or premise stabilization for “precarious HBs,” vocational training for “desolate HBs” and inflation control for “optimistic HBs.”

Bibliographie

Bacchetta Marc, Ernst Ekkehard, Bustamante Juana, Globalization and informal jobs in developing countries, Geneva , OIT et OMC, 2009, 187p.

Bryman Alan, «Integrating Quantitative and Qualitative Research: How is it done? », in Qualitative Research, 6(1), 2006, p. 97-113.

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

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

Cling Jean-Pierre, Razafindrakoto Mireille, and Roubaud François, «To be or not to be registered? Explanatory factors behind formalizing non-farm household businesses in Vietnam », in Journal of the Asia Pacific Economy, n° 4/17, 2012, p. 632-652

De soto Hernando, The Other Path: The Invisible Revolution in the Third World, New York: Harper and Row, 1989, 271 p.

Demenet Axel, Nguyen Thi Thu Huyên, Razafindrakoto Mireille, Roubaud François, Dynamics of the informal sector in Hanoi and Ho Chi Minh City 2007-2009, GSO-IRD, UKaid, World Bank, December (available at http://www.worldbank.org/en/country/vietnam/research, reference DT N°6174 112/2010), 2010.

Dinh Dien, Hoang Kiem, Nguyen Van Toan, « Vietnamese Word Segmentation », in Proceedings of the Sixth Natural Language Processing Pacific Rim Symposium (NLPRS2001), Tokyo, 2001, p. 749-756.

Guttman Louis, « Some necessary conditions for common factor analysis », in Psychometrika, n° 19, 1954, p. 149–161.

Harris John, Todaro Michael, «Migration, unemployment, and development: A two-sector analysis », in American Economic Review, n° 1/60, 1970, p. 126-42.

Hughes Christina, Qualitative and Quantitative Approaches, URL: http://tinyurl.com/bmztxp8, 2012.

Husson François, Josse Julie, Pages Jérôme, «Principal Component Methods - Hierarchical Clustering -Partitional Clustering: Why Would We Need to Choose for Visualizing Data », in Technical Report-Agrocampus, Applied Mathematics Department, 2010, p. 1-17.

ILO, « Guidelines Concerning a Statistical Definition of Informal Employment », in Seventeenth International Conference of Labour Statisticians, ILO, Geneva, 2003.

Kaiser Henry, « The application of electronic computers to factor analysis », in Educational and Psychological Measurement, n° 20, 1960, p. 141–151.

Kaiser Henry, « A second generation Little Jiffy », in Psychometrika, n° 35, 1970, p. 401–417.

La porta Rafael, Shleifer Andrei, «The Unofficial Economy and Economic Development», Brookings Papers on Economic Activity, Economic Studies Program, The Brookings Institution, vol. 39(2 (Fall)), p. 275-363.

Lebart Ludovic, Morineau Alain, Piron Marie, Statistique Exploratoire Multidimensionnelle, Paris, Dunod, 2006, 480p.

Lebart Ludovic, Mirkin Boris, «Correspondence Analysis and Classification », in Multivariate Analysis: Future Directions, Cuadras Carlos María, Rao Calyampudi Radhakrishna (éd.), Amsterdam: North Holland, 1993, p. 341-357.

Lebart Ludovic, Salem André, Berry Lisette, Exploring Textual Data, Dordrecht, Kluwer Academic Publisher, 1998, 222p.

Lewis W. Arthur, « Economic development with unlimited supplies of labor », in Manchester School, n°2/28, 1954, p. 139-191.

Moser Caroline, « Informal sector or petty commodity production: dualism or independence in urban development », in World Development n° 6, 1978, p. 1041-1064.

Nguyen Huu Chi, Nguyen Thi Thu Huyen, Razafindrakoto Mireille, Roubaud François, Vietnam labour market and informal economy in a time of crisis and recovery 2007-2009 ; Main findings of the Labour Force Surveys (LFS), Hanoi: GSO/IRD, UKaid, World Bank, December (consult the site : http://www.worldbank.org/en/country/vietnam/research, reference DT N°6175 of 1/12/2010), 2010.

Nguyen Huu Chi, Nordman Christophe, Roubaud François, « Who suffers the penalty? A panel data analysis of earnings gaps in Vietnam », in Journal of Development Studies, n°12/49, 2013, p. 1694-1710.

Nguyen Cam Tu, Nguyen Trung Kien, Phan Xuan Hieu, Nguyen Le Minh, ha Quang Thuy, « Vietnamese Word Segmentation with CRFs and SVMs: An Investigation », in Proceedings of the 20th Pacific Asia Conference on Language, Information and Computation (PACLIC 2006), Paris: Organization for Economic Co-operation and Development, 2006.

Nguyen Thanh V., Tran Hoang K., Nguyen Thanh T. T., Nguyen Hung, « Word Segmentation for Vietnamese Text Categorization: An Online Corpus Approach », in the 4rd IEEE International Conference in Computer Science, 2006.

Pham Giang, Kohnert Kathryn, Carney Edward, «Corpora of Vietnamese Texts: Lexical effects of intended audience and publication place», Behavior research methods 40.1, 2008, p. 154-163.

Piau Claire, Quelques expériences sur la formulation des questions d’enquête. CREDOC, Département Conditions de vie et Aspirations des Français, Cahier de recherche n°206, 2004, 65 p.

Portes Alejandro, Castells Manuel, Benton Lauren (éd.) The Informal economy: Studies in advanced and less developed countries, Baltimore MD, The John Hopkins University Press, 1989.

Razafindrakoto Mireille, Roubaud François, Nguyen Huu Chi, « Vietnam Labor Market: An Informal Sector Perspective », in Vietnam Annual Economic Report 2011: The Economy at a Crossroad, Nguyên Duc Thành (éd.), Hanoi, Edition Tri Thuc, 2011, p. 223-258.

Razafindrakoto Mireille, Roubaud François, Wachsberger Jean-Michel, «Travailler dans le secteur informel : choix ou contrainte? Une analyse de la satisfaction dans l'emploi au Vietnam », in L'économie informelle dans les pays en développement, Cling Jean-Pierre, Lagree Stéphane, Razafindrakoto Mireille, Roubaud François (éd.), Paris, Edition AFD, «collection conférence et séminaire», 2012, p. 47-66.

Schuman Howard, Presser Stanley, Question and answers in attitude surveys: Experiments on question form, wording and context, New York, Academic Press, 1996, 392 p.

Tran Q. Tri, Pham T. X. Thao, Ngo Q. Hung, Dinh Dien, Collier Nigel, «Named entity recognition in Vietnamese documents », in Progress in Informatics, n° 4, 2007, p. 5-13.

Wanner Leo, «Introduction to Clustering Techniques», International Union of Local Authorities, 2004, 32p. (URL : http://www.iula.upf.edu/materials/ 040701wanner. pdf.)

Yelland Phillip, «An Introduction to Correspondence Analysis », in The Mathematica Journal, n° 12, 2010, p. 1-23.

Annexes

Table 14- Mean Tests for the Non-responses to Open-ended Questions

Disadvantages

Advantages

Non-respondents

Respondents

Difference

Non-respondents

Respondents

Difference

Total size

1.48

1.55

-0.07

1.38

1.63

-0.25**

Value-added

6507.52

5458.01

1049.509

5291.29

6030.65

-739.36

Educational attainment

7.79

7.84

-0.04

7.78

7.85

-0.08

**: significant at 5% level

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations

Table 15- Categorical Variables Used in Correspondence Analysis

Description

Modalities

Abbreviation

Characteristics of HB

 

1

Branch of activity

Manufacturing

manuf

Trade

trade

Services

service

2

Informality of HBs

Formal

formal

Informal

informal

3

Accounting method

Professional accounting

acc_no

Simplified accounting

acc_simp

Personal record/ notes

acc_pers

No accounts

acc_no

4

Type of premise

Unstable premise

premi_unstab

Premise at home

premi_home

Professional premise

premi_prof

5

Has wage earners or not

No

wearner_no

Yes

wearner_yes

6

Total size of HB

1 person

size=1

2 persons

size=2

3 to 6 persons

size3-6

More than 6 persons

size>6

7

Quintile of monthly turnover of HB

First quintile

turnov_q1

Second quintile

turnov_q2

Third quintile

turnov_q3

Fourth quintile

turnov_q4

Fifth quintile

turnov_q5

8

Quintile of monthly value-added of HB

First quintile

va_q1

Second quintile

va_q2

Third quintile

va_q3

Fourth quintile

va_q4

Fifth quintile

va_q5

9

Quintile of monthly equipment costs of HB

First quintile

equip_q1

Second quintile

equip_q2

Third quintile

equip_q3

Fourth quintile

equip_q4

Fifth quintile

equip_q5

10

Main customer of HB

Public or para-public sector

cust_pub

Enterprises

cust_enterp

Household businesses

cust_hb

Individuals

cust_ind

11

Main supplier of HB

Public or para-public sector

supp_pub

Enterprises

supp_enterp

Household businesses

supp_hb

Individuals

supp_ind

12

Age of HBs

Less than 3 years

Agehb<3

From 3 to 7 years

agehb3-7

From 8 to 15 years

agehb8-15

More than 15 years

agehb>15

13

Gender of HB heads

Male

male

Female

female

14

Age of HB

head

Less than 40 years old

Age<40

From 40 to 45 years old

age40-45

From 46 to 55 years old

age46-55

More than 55 years old

age>55

15

Educational level of HB head

Primary school

edu_prim

Secondary school

edu_sec

High school

edu_highsc

University or more

edu_univ

16

Apprenticeship of HB head

Technical school

tech_school

On-job training

on_job_train

Self-trained

self_train

Others

aprt_others

17

Immigration status of HB head

Has lived in this province/ city for at least 10 years

immig_no

Has migrated from another city to this province/ city for less than 10 years

immig_city

Has migrated from a countryside to this province/ city for less than 10 years

immig_rural

18

Residence status of HB heads

Registered in the same province (KT1-KT2)

KT1-2

Temporary registration for a period of 6 months and more (KT3)

KT3

Temporary registration for a period of less than 6 months (KT4)

KT4

No registration at the destination

res_no

19

Why you do not register your HB?

Not obligatory

ynotreg_incompul

Do not know if it has to be registered

ynotreg_dontknow

Other reasons (e.g., too complicated procedures, too expensive)

ynotreg_others

20

Reasons to set up or manage this HB

Did not find a salaried job in an enterprise

a4d_nojob_enterp

Did not find a salaried job in a HB

a4d_nojob_hb

To earn higher income

a4d_better_inc

To be independent

a4d_indep

To follow family tradition

a4d_fam_trad

Others

a4d_others

21

The main reason to choose this business activity

Family tradition

g1_fam_trad

The profession that you know

g1_prof_know

Have higher profit than other activities

g1_higher_profit

More stable returns than other products

g1_stabler_return

Others

 

22

Willingness to register your HB

Yes

wilg_reg_yes

No

wilg_reg_no

Do not know

wilg_reg_dontknow

Already registered

registered

23

Do you think there is a future for a HB like yours?

Yes

futur_yes

No, and think of changing activity

nofutur_change

No, but do not think of changing activity

nofutur_notchange

Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL

Notes

1 See Cling et coll. (2010a) for a more detailed presentation of informal household businesses in Vietnam.

2 More precisely, we use multiple correspondence analysis (MCA) which permits to consider qualitative as well as quantitative data in the same procedure.

3 Agglomerative clustering algorithm starts with one-point (singleton) clusters and recursively merges two most similar clusters. The similarity matrix is then updated to reflect the pairwise similarity between the new cluster and the remaining clusters. These steps are repeated until a single cluster remains. The output of this procedure is a hierarchical tree or “dendrogram.” In K-means clustering method, however, k data points are randomly chosen as the initial centroids. All points are then assigned to their closest centroids and centroid of each newly assembled cluster is recomputed. This process is repeated until the centroids do not change.

4 See also Pham et coll. (2008) for an overview of the Vietnamese language.

5 The two open-ended questions were analyzed in parallel but separately to avoid misinterpretation (given the fact that the analysis presented is a first exploratory phase), but they could be combined in further steps of the analysis.

6 We performed also an analysis mixing “advantages” with “disadvantages” in a preliminary step but the results were very difficult to interpret and not convincing since the technique used refers mainly to words and not to sentences and the same word can be mentioned to stress an advantage or a disadvantage.

7 The Kaiser-Gutman rule for CA indicates that we should keep the factorial axes whose eigenvalues greater than average eigenvalue. This is equivalent to eigenvalue-greater-than-one criterion for PCA (Guttman, 1954; Kaiser, 1960, 1970)

8 Actually, any axis contributing more than 0.4% (1/229) could be considered for the analysis.

9 Although the other axes are also interpretable (i.e. meaningful), we only focus on the two most important factorial axes, since the objective is to provide an illustration of .the heterogeneous nature of the informal sector and the multidimensionality is taken into account in the cluster analysis.

10 We eliminate several words that can be considered as “outliers”, i.e., which do not belong to the systematic meaning of the axes.

11 For this reason, we do not illustrate this cluster in the factorial plane.

Table des illustrations

Titre Figure 1- Job satisfaction levels by institutional sector in Vietnam, 2009
Légende Note: The balance of satisfaction is defined as the difference between the proportion of respondents having expressed a positive opinion (satisfied & very satisfied) and the proportion of respondents having expressed a negative opinion (unsatisfied & very unsatisfied).
Crédits Source: LFS2009, GSO; authors’ calculations.
URL http://books.openedition.org/pupo/docannexe/image/4401/img-1.png
Fichier image/png, 41k
Titre Figure 2 - The Space of Informal HB Characteristics (Projection of Illustrative Variables on the First Factorial Plane of CA on Advantages)
Légende Note: The explicit meaning of the categorical variables is provided in the appendix
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.
URL http://books.openedition.org/pupo/docannexe/image/4401/img-2.jpg
Fichier image/jpeg, 660k
Titre Figure 3 - The Space of Disadvantages Put Forward by Informal HBs (First Factorial Plane of CA)
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations
URL http://books.openedition.org/pupo/docannexe/image/4401/img-3.jpg
Fichier image/jpeg, 320k
Titre Figure 4 - Space of Informal HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages.
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.
URL http://books.openedition.org/pupo/docannexe/image/4401/img-4.jpg
Fichier image/jpeg, 652k
Titre Figure 5 - The Space of Disadvantages Put Forward by HBs (First Factorial Plane of CA)
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.
URL http://books.openedition.org/pupo/docannexe/image/4401/img-5.jpg
Fichier image/jpeg, 296k
Titre Figure 6 - The Space of HB Characteristics (Projection of Illustrative Variables in the First Factorial Plane of CA on Disadvantages)
Légende Note: The explicit meaning of the categorical variables is provided in the appendix.
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations
URL http://books.openedition.org/pupo/docannexe/image/4401/img-6.jpg
Fichier image/jpeg, 688k
Titre Figure 7 - Cluster View of Informal HBs on the First Factorial Plane of CA on Disadvantages
Légende Note: The explicit meaning of the categorical variables is provided in the appendix
Crédits Source: HB&IS HCMC survey 2009, GSO-ISS / IRD-DIAL; authors’ calculations.
URL http://books.openedition.org/pupo/docannexe/image/4401/img-7.png
Fichier image/png, 63k

© Presses universitaires de Paris Nanterre, 2016

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

Acheter