The impact of minimum wages on employment of low-wage workers

The impact of minimum wageson employment of low-wageworkersEvidence from Vietnam1

Cuong Viet Nguyen**National Economics University, Hanoi, Vietnam: and Mekong Development Research Institute,

Hanoi, Vietnam. E-mail: [email protected]

Abstract

This study provides empirical evidence on the impact of a minimum wage increaseon employment of workers in the formal sector who have wages below the mini-mum level in Vietnam. Using the difference-in-differences with propensity scorematching and the Vietnam Household Living Standard Surveys of 2004 and 2006,the article finds that the minimum wage increase in 2005 reduced the proportion ofworkers having a formal sector job among low-wage workers. Most workers wholost formal sector jobs became self-employed.

JEL classifications: J31, J23, D31.Keywords: Minimum wage, employment, income, Vietnam, difference-in-differ-ences, propensity score matching.

Received: July 21, 2010; Acceptance: January 28, 20131 I would like to thank two anonymous referees for their very helpful comments and suggestions on thepaper. I would also like to thank John Gallup, Paulette Castel, and workshop participants at the Institute ofLabor Science and Social Affairs (ILSSA), Ministry of Labor, Invalids and Social Affairs of Vietnam (MOLISA)in April 2009, workshop participants at MOLISA in July 2009, participants in an IR Mini-Talk workshop inInternational Labour Organization (ILO) Office in Hanoi in November 2010, participants of the ILO‘Regula-tion on Decent Work’ workshop in Geneva, July 2011, for their useful comments and discussions on thispaper.

Economics of TransitionVolume 21(3) 2013, 583–615DOI: 10.1111/ecot.12022

� 2013 The AuthorEconomics of Transition � 2013 The European Bank for Reconstruction and Development.Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA

1. Introduction

A minimum wage is the lowest hourly, daily or monthly wage that a governmentrequires employers to pay to employees. The main objectives of the minimum wageare to increase the living standards of labourers, especially the poor and vulnerable,and to prevent exploitation of labourers. In addition, the minimum wage has otherpositive effects such as promotion of labourers’ work and productivity, reduction ofthe number of people covered in subsidy programmes, increasing consumption,aggregate demand and generation of multiplier effects (Dowrick and Quiggin, 2003;Freeman, 1994; Gunderson, 2005).

In addition to positive effects, the minimum wage can have unexpected effects.The main negative effect is to increase unemployment, especially among unskilledand low-wage workers. In the traditional economic theory, an increase in labourcost will reduce demand for labour. It means that the increased minimum wage nor-mally leads to unemployment (Hamermesh, 1986). Before the 1990s, there was con-sensus on the adverse impacts of the minimum wage on employment. Most of theempirical studies in the United States at the time showed that a 10 percent increasein the minimum wage resulted in a 1–3 percent reduction in employment (seeBrown, 1999; Brown et al., 1982; Card and Krueger, 1995).

However, in the ‘new minimum wage research’, there is no consensus on thedirection of the effect of the minimum wage on employment (Bazen, 2000; Lemos,2004; Neumark and Wascher, 2007). In economic theory, the minimum wage canhave negligible effects on employment if the minimum wage is close to the competi-tive wage, or employers have bargaining power so that the elasticity of labourdemand is less sensitive to labour cost (Dickens et al., 1995, 1999). In the monopsonymodel, the increased minimum wage can lead to an increase in employment.

Recent empirical findings on the impact of the minimum wage on employmentare not consistent. Negative effects of the minimum wage on employment, espe-cially for young workers, are found in studies on the United States and developedcountries (Abowd et al., 1999; Burkhauser et al., 2000; Campolieti et al., 2005; Man-gan and Johnston, 1999; Singell and Terborg, 2006; Neumark and Wascher, 1992,1994, 1995, 2000, 2002, 2003). In developing countries, the minimum wage is alsofound to have adverse effects on employment (Bell, 1997; Gindling and Terrell,2004; Harrison and Scorse, 2005; Rama, 2001). However, other studies found thatthe minimum wage had positive effects on employment (Card, 1992; Card andKrueger, 1994, 2000; Dickens et al., 1999; Katz and Krueger, 1992; Montenegro andPag�es, 2004; Rama, 2001; Reich and Hall, 2001; Turner and Demiralp, 2001).2

Vietnam is a developing country which has achieved high economic growth,with annual GDP growth rates of around 6 percent over the past 10 years. Povertyrates have declined remarkably from 58 to 16 percent between 1993 and 2006. To

2 For a detailed review of studies on the minimumwage and employment, see Neumark and Wascher (2007).

� 2013 The AuthorEconomics of Transition � 2013 The European Bank for Reconstruction and Development

584 Nguyen

improve the living standards of the low-wage workers, the government hasincreased the nominal minimum monthly wage from 120,000 to more than 730,000VND during the period 1994–2010.

There are continuing debates about positive and negative impacts of minimumwage increases in Vietnam. There are a large number of advocates of minimum wages,who argue that the minimum wage should be increased to compensate low-wageworkers for the loss in real wages caused by high inflation. Increased wages can leadto an increase in aggregate demand and economic growth, especially in the context ofeconomic slowdown (Dan Tri, 2009a; Duy-Tuan, 2009). On the other hand, there arecritics who argue that an increased minimum wage can result in high inflation (BaoMoi, 2009; Dan Tri, 2009b). An increased minimum wage can also increase productioncosts and unemployment, and add burdens to enterprises, especially when there is on-going economic stagnation (Thai-Uyen, 2009).

The above arguments against or in favour of minimum wage increases are oftenmade without empirical evidence on the impact evaluation of minimum wageincreases. The question of the impacts of minimum wage increases on employmentas well as income of workers in Vietnam remains so far unanswered.

The main objectives of this article are to present summary statistics of workersbelow the minimum wage and to examine the impact of the minimum wageincrease on employment of workers in the formal sector including state and privateenterprises and organizations. The method of impact measurement used in this arti-cle is difference-in-differences with propensity score matching, and the data arefrom Vietnam Household Living Standard Surveys (VHLSS) from 2004 and 2006.The article is expected to make an empirical contribution to the minimum wage lit-erature by providing evidence on the effect of the minimum wage increase in Viet-nam. There are only a few studies on relationships between minimum wages andemployment in Asian developing countries, and Vietnam is a developing and transi-tion country with economic conditions very similar to many countries in Asia.

The remainder of this article is organized as follows. Section 2 introduces thedata sources used in this study. Section 3 presents data on minimum wages in Viet-nam. Section 4 describes the methodology. Section 5 presents the impact estimatesof the minimum wage increase. Finally, section 6 presents the general conclusions.

2. Dataset

The study relies on data from two recent VHLSSs, which were conducted by theGeneral Statistical Office of Vietnam (GSO) with technical support from the WorldBank (WB) in the years 2004 and 2006. The 2004 and 2006 VHLSSs cover 9,188 and9,189 households, respectively. The samples are representative for the national, ruraland urban, and regional levels. The 2004 and 2006 VHLSSs have a panel of 4,216households, for which data are available for both years.


The Impact of MinimumWages On Employment 585

For both surveys, the time of data collection took place mainly in June and Sep-tember. Around 45 and 35 percent of sampled households were surveyed in Juneand September, respectively. The remaining 20 percent of the sample were surveyedin other months, mainly in July and October.

The surveys collected information by household and community level question-naires. Information on households includes basic demography, employment andlabour force participation, education, health, wage, income, expenditure, housing,fixed assets and durable goods, and participation of households in poverty allevia-tion programmes.

Expenditure and income per capita are collected using detailed questions.Regarding income, household income can come from any source. Income includesincome from agricultural and non-agricultural production, salary, wage, pensions,scholarship, income from loan interest and house rental, remittances and socialtransfers. Income from agricultural production comprises crop income, livestockincome, aquaculture income, and income from other agriculture-related activities.

3. Minimum wages and wage earners in Vietnam

3.1 Minimum wage adjustments

In Vietnam, there are only minimum monthly wages. There are neither minimumdaily nor minimum hourly wages. The government adjusts the minimum wageswhen there is price inflation and economic growth. According to the Labour Law ofVietnam, the minimum wage is adjusted as the prices of commodities and serviceschange. When the economy experiences high economic growth, the minimum wageis also increased to improve the living standard of workers. In addition, the pay-ment capacity of the state budget is considered, since workers’ wage in the state sec-tor is set based on the minimum wages.

There were nine adjustments of the minimum wage during the period 1993–2010. The timing and the minimum wages are presented in Figure 1. It shows thatthe nominal minimum wage increased by 508 percent from 120,000 to 730,000 VNDduring the period 1994–2009. However, the real minimum wage (in terms of theprice of 1999) increased by 120 percent.

It should be noted that the minimum wages presented in Figure 1 are appliedfor the governmental sector and the domestic sector. Minimum wages applied forthe foreign sector including foreign joint-venture enterprises, foreign-investedenterprises, international individuals, institutions and organizations are higher.

Since the available data are from VHLSSs 2004 and 2006, this article will examinethe impacts of the minimum wage increase from 290,000 to 350,000 VND in October2005 on the income and employment of workers in the domestic sectors includingpublic and private sectors. During the 2004–2006 period, there was no change in theminimum wages for the foreign and international sectors.


586 Nguyen

3.2 Minimum wages and workers’ wages

In Vietnam, people can be self-employed, or wage employed in formal or informalsectors. In this article, wage earners are divided into formal and informal employ-ment sectors. In VHLSSs, there is no information on whether a worker is enrolled inthe social insurance system. Thus, we define informal workers as those who workfor other households. These people often do not have labour contracts and are notenrolled in social insurance. Workers in state and private enterprises/organizationsoften have labour contracts and are enrolled in social insurance; they are defined asworkers in the formal sector.

Table 1 presents the distribution and monthly wage of workers aged from 15 to60 years by employment sector. The monthly wage is from the main employment ofworkers during the past 12 months. We do not consider wages from secondaryemployment, since most secondary employments are part time and informal. Thedistribution of workers by employment sector was almost unchanged during 2004–2006. In 2006, there were around 20 percent of people who were not involved in pro-ductive activities, such as students, retired or unemployed workers. Self-employedworkers accounted for 52.2 percent of the labour force. The percentage of peopleworking for other households was 12.7 percent of the labour force. The proportionof people working for state and private enterprises/organizations in 2006 was 8.4and 6.7 percent, respectively.

Regarding wages, the state sector had the highest average wages, followed bythe private formal sector. However, the private formal sector had a smaller fraction

Figure 1. Minimum monthly wage in Vietnam (Thousand VND)

Source: Government of Vietnam (2006, 2008, 2009).



Tab

le1.

Distribution

andnom

inal

mon

thly

wag

eof

peo

ple

invo

lved

inproductiveactivities

Peo

ple

from

15to

60ye

ars

old

2004

2006

Percent

Mon

thly

wag

e(‘00

0VND)

%withwag

elower

than

290thou

sand

VND

%withwag

elower

than

350,00

0VND

Percent

Mon

thly

wag

e(‘00

0VND)

%withwag

elower

than

290thou

sand

VND

%withwag

elower

than

350,00

0VND

Not

working

18.6

––

–20

.0–

––

Working

fortheir

hous

eholds(self-

employ

ed)

54.3

––

–52

.2–

––

Working

forothe

rho

useh

olds

(inform

alsector)

12.8

671.9

6.5

11.4

12.7

738.7

6.8

15.2

Working

forstate

sector

(formal

sector)

8.5

1,09

1.3

5.1

7.6

8.4

1,25

2.8

4.2

7.9

Working

for

privatesector

(formal

sector)

5.9

918.9

3.7

7.6

6.7

985.0

3.0

6.7

Total

100

856.8

5.5

9.4

100

953.9

5.1

10.9

Num

berof

observations

25,655

25,708

Source:E

stim

ationfrom

VHLSS

s2004

and2006.


588 Nguyen

of workers below the minimum wage. The percentage of workers below the mini-mum wage in the formal private sector was 3.7 and 3.0 percent in 2004 and 2006,respectively. Meanwhile, the corresponding numbers for the state sector are 5.1 and4.2 percent. Workers who were employed by private households had the lowestaverage wages. The percentage of workers below the minimum wage in this groupwas 6.5 in 2004 and 6.8 percent in 2006.

Table 1 also presents the percentage of workers below the minimum wage dur-ing the next period. In 2004, in the formal sector, the proportion of workers belowthe level to which the minimum wage rose in the next year was 7.6 percent. This isthe group likely to be most affected by the minimum increase. This article willmeasure impacts of the minimum wage increase on workers in the formal sector.For the informal sector, the corresponding percentage is 11.4. The article does notevaluate the impact on this group, since the informal sector rarely follows the mini-mum wage regulations.3

Table 2 examines how the monthly wage and employment of workers belowthe minimum wage of the next period changes over time. Among workersemployed in 2004, the proportion of people involved in productive activities in2006 was 95.5 percent. The corresponding fraction for workers who in 2004 hadmonthly wages below 350,000 VND in the informal and formal sectors was 90.6and 97 percent, respectively. It is not clear that workers below the minimum wagein the formal sector were unemployed after the minimum wage increase. However,workers with wages below 350,000 VND in 2004 tended to move out of the formalsector in 2006. Among workers in the formal sector in 2004, the fraction of thoseremaining in the formal sector in 2006 was 58.4 and 78.5 percent for workers withwages below and above the 350,000 VND wage level, respectively.

Table 2 shows that workers with low wages in 2004 constituted a large propor-tion of those workers having wages lower than the minimum wage in 2006. Amongthe workers who had their wages below 350,000 VND in 2004, the fraction of work-ers having wages in 2006 also below 350,000 VND was 15.3 and 27.3 percent for theinformal and formal sectors, respectively. However, workers with low wages in2004 experienced higher average growth rates of wages and consumption expendi-tures than workers with high wages over the period 2004–2006.

Table 3 examines changes in wages and expenditure of workers in differentemployment sectors during the period 2004–2006. The real monthly wage of work-ers in the formal sector in both 2004 and 2006 increased remarkably by around19.9 percent during 2004–2006. Workers who moved from the informal sector tothe formal one experienced an increase of only 4 percent in monthly wages. It ispossible that these workers accepted lower wages in the short term with an expec-tation of a long-term formal job. Workers who moved from the formal sector tothe informal one increased their wages by around 26.1 percent. Workers in the

3 Not only the minimum wage but also other regulations such as health insurance, social insurance andlabour contracts are not often implemented in the informal sector.



Tab

le2.

Employm

enta

ndmon

thly

wag

esof

workerswithwag

esab

ovean

dunder

350,000VND

Workersfrom

15to

60ye

arsold

Wag

esin

2004

(‘00

0VND)

No.

ofob

servations

Percent

%peo

ple

working

in20

06

%peo

ple

workingin

theform

alsector

in20

06

%workers

withwag

ebelow

350,00

0VND

in20

06

Mon

thly

wag

ein

2004

(‘00

0VND)

Mon

thly

wag

ein

2006

–in

2004

price

(‘00

0VND)

Working

for

inform

alsector

in20

04

Wag

e<35

016

96.4

90.6

10.7

15.3

261.9

546.6

Wag

e≥3

501,06

941

.395

.69.1

4.4

741.7

759.1

Working

for

form

alsector

in20

04

Wag

e<35

011

24.3

97.0

58.4

27.3

246.8

532.8

Wag

e≥3

501,20

648

.095

.978

.51.1

1,08

4.6

1,28

9.3

Total

2,55

610

095

.544

.74.0

885.5

1,02

9.0

Notes:W

orking

forothe

rho

useh

oldsin

2004

(inform

alsector).Mon

thly

wag

esarein

thou

sand

VND

atthe2004

price.

Source:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.


590 Nguyen

Tab

le3.

Mon

thly

wag

esan

dexpen

diture

ofpeo

ple

invo

lved

inproductiveactivities

intheform

alan

dinform

alsector

Workingpeo

ple

from

15to

60ye

ars

old

No.

ofob

servations

Percent

%workers

withwag

elower

than

350,00

0VND

in20

06

Mon

thly

wag

ein

2004

(‘00

0VND)

Mon

thly

wag

ein

2006

–in

2004

price

(‘00

0VND)

%increa

sein

mon

thly

wag

e

Working

ininform

alsector

inbo

th20

04an

d20

06

723

35.6

5.7

670.6

722.6

7.8

Working

ininform

alsector

in20

04bu

tform

alsector

in20

06

110

5.8

5.8

772.9

803.8

4.0

Working

inform

alsector

in20

04bu

tinform

alsector

in20

06

113

6.2

3.8

823.4

1,03

7.9

26.1

Working

inform

alsector

inbo

th20

04an

d20

06

1,01

552

.42.8

1,05

1.2

1,26

0.9

19.9

Total

1,96

110

04.0

885.5

1,02

9.0

16.2

Note:Mon

thly

wag

esarein

thou

sand

VND

andin

the20

04price.

Source:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.



formal sector in both 2004 and 2006 also experienced a growth of 7.8 percent inmonthly wages over the period 2004–2006.

The fact that there is a proportion of workers in the formal sectors having wagesbelow the minimum level suggests that several employers do not comply with mini-mum wage regulations. To examine the minimum wage binding, we graph the ker-nel density with different bandwidths of nominal monthly wages of workers from15 to 60 years old in different sectors for the year 2004 (Figure 2) and 2006 (Fig-ure 3). Kernel density graphs are widely used to examine the minimum wage bind-ing (see Cunningham, 2007; Dinardo et al., 1996; Heckman and Pag�es, 2003). Theminimum wage is more likely to be binding if there is a bulge right after the mini-mum wage. The Figures 1 and 2 do not show strong compliance with the minimumwages in either 2004 or 2006.

Figure 2. Kernel density of wages by sector in 2004 with different bandwidths

Source: Estimation from the 2004 VHLSS.


592 Nguyen

4. Methodology of impact evaluation

4.1 Parameter of interest

The main objective of impact evaluation of a programme is to assess the extent towhich the programme has changed the outcomes of subjects. In this study, we aimto measure the effect of minimum wages on employment of workers in the formalsector. Denote D as a binary variable indicating exposure of a person to a minimumwage increase, that is D equals 1 if she will be affected by the minimum wageincrease, and D equals 0 otherwise. In this article, D equals 1 for workers who hademployment in the formal sector and monthly wages below 350,000 VND in 2004.These people are expected to be affected by the increase of minimum wage from

0.0

005

.001

.001

5

0 1000 2000 3000 4000 5000x

kdensity Households kdensity Public_sectorkdensity Private_firms

0.0

005

.001

.001

5

0 1000 2000 3000 4000 5000x


0.0

005

.001

.001

5

0 1000 2000 3000 4000 5000

x


0.0

005

.001

.001

5

0 1000 2000 3000 4000 5000

x


Figure 3. Kernel density of wages by sector in 2006 with different bandwidths

Source: Estimation from the 2006 VHLSS.



290,000 to 350,000 VND in 2005. Further let Y denote the observed value of an out-come of interest. This variable can have two potential values depending on the valueof D, that is, Y = Y1 for D = 1, and Y = Y0 for D = 0.4 For consistency with the litera-ture relating to impact evaluation, in this article, workers who are affected by orexposed to the minimum wage increase are sometimes called participants or treated,and workers who are not affected by the minimum wage increase are sometimescalled non-participants or untreated. In addition, the minimum wage increase itselfis sometimes called the programme.

The most popular parameter of the programme impact is Average TreatmentEffect on the Treated (ATT) (Heckman et al., 1999), which is the expected impact ofthe programme on the actual participants5 :

ATT ¼ EðY1 � Y0jD ¼ 1Þ ¼ EðY1jD ¼ 1Þ � EðY0jD ¼ 1Þ: ð1Þ

This parameter can be varied across a vector of the observed variables X:

ATT Xð Þ ¼ EðDjX;D ¼ 1Þ ¼ EðY1jX;D ¼ 1Þ � EðY0jX;D ¼ 1Þ: ð2Þ

Estimation of ATT is not straightforward, since E(Y0|D = 1) is not observed andcannot be estimated directly. E(Y0|D = 1) is called counterfactual which indicatesthat it would have been the expected outcome if participants had not been affectedby the minimum wage.

4.2 Difference-in-differences with matching method

When panel data on the treatment and control groups before and after the treatmentare available, the difference-in-differences estimator can be used to estimate theimpact of the treatment programme. The difference-in-differences estimator can becombined with matching to control differences in observed variables between thetreatment and control groups. The basic idea of the matching method is to find acomparison group that has the same (or at least similar) distribution of the variablesX as that of the treatment group.6 Compared with difference-in-differences regres-sion, the main advantage of the matching method is that it does not require anassumption on the functional form of outcome. Thus, it can avoid bias caused bymisspecification of outcome functions.

4 Y can be a vector of outcomes, but for simplicity we consider a single outcome of interest.5 There are other parameters such as average treatment effect (ATE), local ATE, marginal treatment effect oreven effect of ‘non-treatment on non-treated’which measures what impact the programme would have on thenon-participants if they had participated in the programme, etc.6 There is a large literature on matching methods of impact evaluation, for example, Dehejia and Wahba(1998), Heckman et al. (1997), Rosenbaum and Rubin (1983), Rubin (1977), and Smith and Todd (2005).


594 Nguyen

To describe the method, let Y20040 denote the outcome in 2004, that is before the

2005 minimum wage increase. After the minimum wage increase, the potential out-comes in 2006 are denoted by Y2006

0 ;Y20061 corresponding to the states of no minimum

wage increase and minimum wage increase. The ATT(X) after the minimum wageincrease is defined as follows:

ATTðXÞ ¼ EðY20061 jX;D ¼ 1Þ � EðY2006

0 jX;D ¼ 1Þ: ð3Þ

The difference-in-differences with matching relies on an assumption that condi-tional on X, the difference in the expectation of outcomes between the participantsand non-participants is unchanged before and after the minimum wage increase,that is,

EðY20040 jX;D ¼ 1Þ � EðY2004

0 jX;D ¼ 0Þ ¼ EðY20060 jX;D ¼ 1Þ � EðY2006

0 jX;D ¼ 0Þ:ð4Þ

Under this assumption, the conditional parameter ATT(X) can be identified bythe matching method, since

ATTðXÞ ¼ EðY20061 jX;D ¼ 1Þ � EðY2006

0 jX;D ¼ 1Þ� EðY2006

0 jX;D ¼ 0Þ � EðY20040 jX;D ¼ 0Þ� �

þ EðY20060 jX;D ¼ 1Þ � EðY2004

0 jX;D ¼ 1Þ� �¼ EðY2006

1 jX;D ¼ 1Þ � EðY20060 jX;D ¼ 0Þ� �

:

� EðY20040 jX;D ¼ 1Þ � EðY2004

0 jX;D ¼ 0Þ� �ð5Þ

The unconditional parameter is also identified, since

ATT ¼ZXjD¼1

ATTðXÞdFðXjD ¼ 1Þ: ð6Þ

To estimate the programme impact, the non-participants are matched with theparticipants based on their variables X before and after the programme. Thematched non-participants will form a control group. To find a control group whohas similar variables X, there is a common support assumption:

0\PðD ¼ 1jXÞ\1; ð7Þ

which states that there are non-participants who have the X variables similar tothose of the participants.

A problem in the matching is how to match non-participants with participants.When there are several conditioning variables X, finding ‘close’ non-participants to



match with participants is not straightforward. Following Rosenbaum and Rubin(1983), a widely used method of finding the matched sample is propensity scorematching.7 Non-participants and participants are matched based on the closenessof their propensity score, which is equal to the probability of being assignedinto the programme. In this article, matching based on propensity score isemployed.

The matching estimator is based on Equation (5). It is equal to the difference-in-differences in outcomes between the treatment group and the control group beforeand after the minimum wage increase. Depending on the number of non-partici-pants matched with a participant, we can have different matching estimators. In thisarticle, we use five nearest neighbours and kernel matching to examine the sensitiv-ity of the impact estimates. In the five nearest neighbours matching, each participantis matched with five non-participants which have the propensity score closest to thepropensity score of the participant. Each matched non-participant receives equalweight in constructing the average outcome that serves as the counterfactual for theparticipant. The kernel matching matches a participant with one or several non-par-ticipants which have a distance from their propensity score to the participant’s pro-pensity score lying within a selected bandwidth. The formulas for the estimators arepresented in Appendix A. The standard errors are calculated using bootstraptechniques.8

5. Impact estimation

5.1 Construction of treatment and control groups

In this article, we estimate the impact of the minimum wage increase in 2005 onemployment of workers who worked in the formal sectors (state and private) andhad monthly wages below 350,000 VND in 2004.9 These workers are expected to be

7 Other matching methods can be subclassification (e.g. Cochran, 1968; Cochran and Chambers, 1965) and co-variate matching (Rubin, 1979, 1980).8 This bootstrap is implemented by repeatedly drawing samples from the original sample of the VHLSS paneldata. Since the VHLSSs sample selection follows stratified random cluster sampling, communes (i.e. primarysampling units) instead of households are bootstrapped in each stratum (Deaton, 1997). In other words, thebootstrap is made of communes (i.e. clusters) within strata. The number of replications is 500. We also tried tobootstrap households instead of communes, and the results of both are very similar. Abadie and Imbens(2006) show that bootstrapping can give invalid standard errors for the nearest neighbour matching estimator.However, there has not been evidence on the validity of standard errors for other matching estimators usingbootstrapping. Most empirical studies rely on the bootstrap method to estimate standard errors of matchingestimators.9 As mentioned in the previous section, we do not evaluate the impact on workers in the informal sector. Theminimum wage before 2006 (including the minimum wage increase in 2005) was applied for workers in theformal sector (Government of Vietnam, 2005, 2006). The minimum wage is now applied for all the workers,but the informal sector rarely follows the labour regulations (Ha, 2013; Pham, 2013).


596 Nguyen

exposed to the effect of the minimum wage increase. The treatment group does notinclude workers in the foreign sector, since there was no adjustment of minimumwage for this sector during the 2004–2006 period.

The control group includes workers who were in the formal sector and hadmonthly wages from 350,000 to 650,000 VND in 2004. The matching is performedbetween workers with monthly wages below and above 350,000 VND. This is simi-lar to the method based on discontinuity design (Hahn et al., 2001; Van derKlaauw, 2002). Ideally, we should have the control group and treatment group justaround 350,000 VND. However, there are few observations around that level inthe dataset, and we have to use all the observations below 350,000 VND as thetreatment group and all the observations from 350,000 to 50,000 VND as the con-trol group. The number of observations in the treatment and control groups is 112and 351, respectively.10

It should be noted that we assume that workers with monthly wages above350,000 VND in 2004 are not affected by the minimum wage. This assumption mightnot hold if there are numeraire effects of the minimum wage increase.11 In this case,wages and employment of workers above the minimum wage are also affected.There are no clean control groups, and the difference-in-differences estimators can-not be applied.

We do not include self-employed workers in any control group, since the defini-tion of employment can be different between self-employed workers and employedworkers and there are no data on wages for self-employed workers. All individualsin the treatment and control groups are from 15 to 60 years old in 2004. In this arti-cle, we do not estimate impacts separately for the state or private sectors, since thenumber in each sector is very small.

The 2004 data are regarded as the baseline data of the 2005 minimum wageincreases, while the 2006 data are regarded as the post-treatment data of the mini-mum wage increase. It should be noted that the minimum wage was also increasedin October 2006 (section 3.1). However, as mentioned in section 2, the 2006 VHLSSwas mainly conducted in June and September 2006. Thus, the 2006 VHLSS was notaffected by the minimum wage increase in October 2006.

It should be noted that the treatment group and the control group differ in their2004 wages. Thus, the treatment group and the control group can differ in bothobserved and unobserved characteristics. Matching can eliminate the difference inobserved characteristics, and difference-in-differences estimation can eliminate thedifference in time-invariant unobserved characteristics between the treatment and

10 We examined the sensitivity of the impact estimates to the definition of the control group by changing thewage level used define the control group from 650,000 VND to 600,000 and to 500,000 VND. The results arequite similar. We do not present results of impact estimations using these other treatment groups in this arti-cle. However, the results can be provided on request.11 Neumark et al. (1998) find numeraire effects of the minimum wage in the United States. Other studies suchas Arango and Pach�on (2004), Cunningham (2007), Fajnzylber (2002), Gindling and Terrell (2005), Maloneyand Nu~nez (2001), Neri et al. (2000) find numeraire effects of the minimum wage in developing countries.



control groups. The main assumption for the difference-in-differences with match-ing method is that there is no difference in the expectation of time-variant unob-served variables between the treatment and control groups.

The estimates will be biased if the above assumption does not hold. Althoughthe treatment and control groups both have low wages, the treatment group still haslower wages than the control group. It is possible they are different not only in time-invariants but also time-variant unobserved variables. Unfortunately, we are notable to predict the direction of the bias. For example, low-wage workers tend tohave lower non-cognitive skills and experiences than high-wage workers (Pierre,2012). If the gap in non-cognitive skills between the treatment and control groupstends to increase over time, then our difference-in-differences will overestimate theeffect of the minimum wage increase. Without the minimum wage increase, low-wage workers are still more likely to lose jobs than high-wage workers.

There can be unobserved effects from the labour demand side. For example,low-wage workers are more likely to be employed by small firms in Vietnam(Pierre, 2012). Large firms and small ones can have different growth and demandfor labour overtime. As a result, firm size can have an effect on employment ofworkers, thereby causing the estimate to be biased. However, it is expected thatthe treatment and control groups are not very different, since the wage gapbetween the treatment and control groups is not very large. Most importantly,unobserved variables such as non-cognitive skills are time-invariants during theshort time period 2004–2006.

To match the treatment and control groups, we predict the propensity score; thatis, the probability of having a monthly wage below 350,000 VND. Since the depen-dent variable is binary, a logit regression is used. The explanatory variables includeage, sex, married, ethnicity, education and occupation, households’ land, regionaland urban variables. These variables are expected to affect the wage and employ-ment of workers. To ensure the explanatory variables are exogenous to the mini-mum wage increase, they are all taken before the minimum wage increase, that is inthe 2004 VHLSS. The conditioning variables are presented in Table B1 in AppendixB. Table B2 presents the logit regressions. The large model uses all the availableexplanatory variables, while the small model keeps only variables which are statisti-cally significant at the 10 percent level (using stepwise regressions).12 Once the pro-pensity score is estimated, the matched control group is constructed by five nearestneighbours matching, kernel matching with bandwidths of 0.01 and 0.05.

5.2 Impact estimation

Table 4 presents the impact of the minimum wage increase in 2005 on employmentof the workers in the formal sector and with monthly wages below the 350,000 VNDin 2004 using three matching estimators including five nearest neighbours and

12 Both backward and forward stepwise regressions result in the same models.


598 Nguyen

kernel matching with a bandwidth of 0.01 and 0.05. The control group is matchedwith the treatment group based on the propensity score that is estimated from thesmall model. The estimates from the three matching estimators are very similar. Itshows that the impact of the minimum wage increase on overall employment is verysmall and not statistically significant. However, the minimum wage increase hasnegative and statistically significant effects on employment in the formal sector. Theminimum wage increase reduces the proportion of workers having a formal sectorjob from 70 percent to 58 percent.

Workers with low wages tend to lose jobs in the formal sector and move to theinformal sector, either self-employed or employed by other households. To examinewhether workers losing jobs in the formal sector can find a wage job in the informalsector or become self-employed, we estimate the effect of the minimum wageincrease on self-employment. In 2004, by definition of the treatment and controlgroup, the proportion of self-employment was equal to zero. In 2004, the proportionof self-employed in the treatment and control groups is around 27 and 15 percent,respectively. The minimum wage adjustment increases the possibility of being self-employed among low-wage workers in the formal sector by 12 percentage points.Thus, the majority of workers who lost their formal sector job became self-employed instead of getting a wage job in the informal sector.

For sensitivity analysis, we also match the control and the treatment groupsusing the propensity score that is estimated from the large model (Table B2 in theAppendix). The impact estimates are presented in Table C1 in the Appendix. Theseestimates are very similar to those presented in Table 4.

In addition to the control group with monthly wages from 350,000 to 650,000also try a control group that includes workers with monthly wages from 350,000 to550,000 VND. This control group has closer wages to the treatment group. This con-trol group has 209 workers. Table C2 in the Appendix presents the estimate of theminimum wage on employment using this control group definition. The propensityscore used for matching is estimated from the small model. The impact estimatesare very similar to those obtained from the control group with monthly wages from350,000 to 650,000 VND.

For additional robust analysis, we run parametric difference-in-differencesregressions (Table C3 in the Appendix). Similarly, the minimum wage increase isfound to have a negative effect on employment in the formal sector, and a positiveeffect on self-employment.

It should be noted that this finding should be interpreted with caution. As pre-sented in section 3, there is a proportion of workers paid below the minimum wage,and the kernel density analysis does not show clear evidence on the compliance ofthe minimum wage in Vietnam. If the minimum wage is not strongly binding, thedifference-in-differences estimators might not be able to capture the whole effect ofthe higher minimum wage on formal sector employment. The difference-in-differ-ence estimators may simply capture the differential transition propensity betweenworkers above and workers below the minimum wage.



Tab

le4.

Theim

pacto

ntheminim

um

wag

eincrease

onem

ploym

ent

Outcom

esan

dmatch

ing

schem

es20

0420

06Difference-in-

differences

Treated

Match

edCon

trol

Difference

Treated

Match

edCon

trol

Difference

(1)

(2)

(3)=(1)�

(2)

(4)

(5)

(6)=(4)�

(5)

(7)=(6)�

(5)

Hav

ejob(%

)Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

97.0***

95.8***

1.2

1.2

(0.0)

(0.0)

(0.0)

(1.5)

(2.2)

(2.5)

(2.5)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

97.0***

95.9***

1.1

1.1

(0.0)

(0.0)

(0.0)

(1.5)

(1.8)

(2.3)

(2.3)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

97.0***

95.5***

1.5

1.5

(0.0)

(0.0)

(0.0)

(1.5)

(1.4)

(2.0)

(2.0)

Hav

eform

alsector

jobs

(%)

Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

58.3***

69.4***

�11.2*

�11.2*

(0.0)

(0.0)

(0.0)

(4.6)

(4.9)

(6.8)

(6.8)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

58.3***

69.2***

�10.9*

�10.9*

(0.0)

(0.0)

(0.0)

(4.6)

(4.3)

(6.3)

(6.3)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

58.3***

70.0***

�11.8**

�11.8**

(0.0)

(0.0)

(0.0)

(4.6)

(3.2)

(5.7)

(5.7)

Self-em

ploy

ed(%

)Five

nearestn

eigh

bours

match

ing

0.0

0.0

0.0

27.2***

15.3***

12.0*

12.0*

(0.0)

(0.0)

(0.0)

(5.0)

(4.0)

(6.5)

(6.5)

Kerne

lmatch

ing:

band

width

=0.01

0.0

0.0

0.0

27.2***

15.7***

11.5*

11.5*

(0.0)

(0.0)

(0.0)

(5.0)

(3.4)

(6.0)

(6.0)

Kerne

lmatch

ing:

band

width

=0.05

0.0

0.0

0.0

27.2***

15.4***

11.8**

11.8**

(0.0)

(0.0)

(0.0)

(5.0)

(2.7)

(5.5)

(5.5)

Notes:T

hetreatedareworke

rswho

worke

din

theform

alsector

andha

dmon

thly

wag

esbe

low

350,00

0VND

in20

04.T

hematch

edcontrolisalso

worke

rsin

theform

alsector

in2004

andha

dmon

thly

wag

esfrom

350,000to

650,000VND

in2004.T

hetreatedan

dmatch

edgrou

psha

vebe

enmatch

edba

sedon

theclosen

essof

theprop

ensity

score.

Colum

ns(1)an

d(2)repo

rtthemeanou

tcom

esof

thetreatm

entgrou

pan

dthematch

edcontrolg

roup

in20

04resp

ectiv

ely.

Colum

n(3)isthedifferen

cein

themeanou

tcom

ebe

tweenthetreatm

enta

ndcontrolg

roup

sin

2004.S

imila

rly,

columns

(4)a

nd(5)p

resent

themeanou

tcom

esof

thetreatm

entg

roup

andthematch

edcontrolg

roup

in2006

resp

ectiv

ely.

Colum

n(6)isthediffer-

ence

betw

eencolumns

(4)a

nd(5).Colum

n(7)isthedifferen

cebe

tweencolumns

(3)a

nd(6).The

differen

ce-in

-differen

cesestim

ator

ispresen

tedby

equa

tion(C

.1)inApp

endix

C.S

tand

arderrors

inbracke

t(Stan

darderrors

arecalculated

usingbo

otstrapwith

500replications.S

tand

arderrors

are

also

correctedforsamplingweigh

tsan

dclus

tercorrelation).*sign

ificant

at10

%;**significant

at5%

;***

sign

ificant

at1%

.So

urce:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.


600 Nguyen

To examine the above issue, we divide the treatment group who had monthlywages below 350,000 VND into groups: one with wages below 250,000 VND, andanother with wage from 250,000 to 350,000 VND. If formal workers who had verylow wages are more likely to have moved to the informal sector, there would be sig-nificant differences in employment between the two groups. However, Table 5 doesnot show a significant difference in employment pattern after the minimum wageincrease.

A problem in the estimation in Table 5 is the low number of observations. Thus,in Table 6 we examine the difference in employment between workers who hadmonthly wages from 350,000 to 500,000 VND in 2004 and those who had monthlywages higher than 500,000 and lower than 650,000 VND. In other words, the controlgroup is divided into one with lower wages and another with higher wages. Again,we do not find significant differences in employment between the two groups. Thus,we expect that our estimates presented in Table 4 might reflect the effect of mini-mum wages.

6. Conclusions

Since the year 1993, there have been nine adjustments of the minimum monthlywage in Vietnam. All of these adjustments are increases in the minimum wage. Themain reason for the minimum wage increase is to compensate for high inflation andto the increased welfare of low-wage workers. However, this positive effect can bemitigated if the minimum wage increases also result in unemployment. This articleis the first attempt to measure the impact of the minimum wage increase on employ-ment, wages and expenditures of workers who are below the minimum wage andworking in the formal sector, that is state and private enterprises/organizations inVietnam.

Using data from VHLSSs 2004 and 2006, the article found that there was a largeproportion of workers receiving wages below the minimum wage. The proportionof workers below the minimum wage in the formal private sector was 3.7 and 3 per-cent in 2004 and 2006, respectively. Meanwhile, the corresponding numbers for thestate sectors are 5.1 and 4.2 percent. In the informal sector (that is households areemployers), the proportion of workers below the minimum wage in this group was6.5 and 6.8 percent in 2004 and 2006, respectively.

Next, the study measures impacts of the minimum wage increase in 2005 onemployment, monthly wages and consumption expenditure of the workers in theformal sector and having wages below 350,000 differences with propensity scorematching. It is found that the impact on overall employment is very small andnot statistically significant. However, the minimum wage increase has a negativeand statistically significant effect on employment in the formal sectors. Workerswith low wages can lose their job in the formal sector and become self-employedbecause of the minimum wage increase.



Tab

le5.

Difference-in-differencesof

employm

entb

etweenworkersbelow

andab

ove250,000

VND

Outcom

esan

dmatch

ingschem

es20

0420

06Diff-in-diff

Treated

Match

edCon

trol

Difference

Treated

Match

edCon

trol

Difference

(1)

(2)

(3)=(1)�

(2)

(4)

(5)

(6)=(4)�

(5)

(7)=(6)�

(5)

Hav

ejob(%

)Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

97.4***

97.9***

�0.5

�0.5

(0.0)

(0.0)

(0.0)

(2.5)

(2.9)

(4.9)

(4.9)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

97.4***

97.6***

�0.2

�0.2

(0.0)

(0.0)

(0.0)

(2.5)

(3.1)

(6.5)

(6.5)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

97.4***

98.2***

�0.9

�0.9

(0.0)

(0.0)

(0.0)

(2.5)

(2.8)

(5.2)

(5.2)

Hav

eform

alsector

jobs

(%)

Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

52.6***

50.5***

2.1

2.1

(0.0)

(0.0)

(0.0)

(8.2)

(6.3)

(8.2)

(8.2)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

52.6***

48.9***

3.7

3.7

(0.0)

(0.0)

(0.0)

(8.2)

(6.7)

(9.2)

(9.2)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

52.6***

48.0***

4.6

4.6

(0.0)

(0.0)

(0.0)

(8.2)

(6.4)

(10.1)

(10.1)

Self-em

ploy

ed(%

)Five

nearestn

eigh

bours

match

ing

0.0

0.0

0.0

28.9***

35.3***

�6.3

�6.3

(0.0)

(0.0)

(0.0)

(7.5)

(8.3)

(11.3)

(11.3)

Kerne

lmatch

ing:

band

width

=0.01

0.0

0.0

0.0

28.9***

36.3***

�7.3

�7.3

(0.0)

(0.0)

(0.0)

(7.5)

(8.8)

(14.0)

(14.0)

Kerne

lmatch

ing:

band

width

=0.05

0.0

0.0

0.0

28.9***

37.7***

�8.7

�8.7

(0.0)

(0.0)

(0.0)

(7.5)

(8.1)

(13.7)

(13.7)

Notes:T

hetreatedareworke

rswho

worke

din

theform

alsector

andha

dmon

thly

wag

esbe

low

250,00

0VND

in20

04.T

hematch

edcontrolisalso

worke

rsin

theform

alsector

in20

04who

hadmon

thly

wag

esfrom

250,00

0to

350,00

0VND

in20

04.T

hetreatedan

dmatch

edgrou

psha

vebe

enmatch

edba

sedon

theclosen

essof

theprop

ensity

score.

Colum

ns(1)an

d(2)repo

rtthemeanou

tcom

esof

thetreatm

entgrou

pan

dthematch

edcontrolg

roup

in2004,respe

ctively.

Colum

n(3)isthedifferen

cein

themeanou

tcom

ebe

tweenthetreatm

enta

ndcontrolg

roup

sin

2004.S

imila

rly,

columns

(4)a

nd(5)p

resent

themeanou

tcom

esof

thetreatm

entg

roup

andthematch

edcontrolg

roup

in20

06,respe

ctively.

Colum

n(6)isthedif-

ferenc

ebe

tweencolumns

(4)a

nd(5).Colum

n(7)isthedifferen

cebe

tweencolumns

(3)a

nd(6).Stan

darderrors

inpa

renthe

ses(Stand

arderrors

are

calculated

usingbo

otstrapwith

500replications.S

tand

arderrors

arealso


tsan

dclus

tercorrelation).*

sign

ificant

at10

%;**significant

at5%

;***

sign

ificant

at1%

.So

urce:E

stim

ationfrom

pane

ldataof

VHLSS

s20

04an

d20

06.


602 Nguyen

Tab

le6.

Difference-in-differencesof

employm

entbetweenworkersbelow

andab

ove500,000VND

Outcom

esan

dmatch

ing

schem

es20

0420

06Difference-in-

differences

Treated

Match

edCon

trol

Difference

Treated

Match

edCon

trol

Difference

(1)

(2)

(3)=(1)�

(2)

(4)

(5)

(6)=(4)�

(5)

(7)=(6)�

(5)

Hav

ejob(%

)Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

96.8***

95.2***

2.6

2.6

(0.0)

(0.0)

(0.0)

(1.4)

(1.6)

(2.6)

(2.6)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

96.8***

95.8***

1.0

1.0

(0.0)

(0.0)

(0.0)

(1.4)

(1.7)

(2.4)

(2.4)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

96.8***

95.8***

1.1

1.1

(0.0)

(0.0)

(0.0)

(1.4)

(1.6)

(2.3)

(2.3)

Hav

eform

alsector

jobs

(%)

Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

68.6***

71.1***

�2.4

�2.4

(0.0)

(0.0)

(0.0)

(4.0)

(3.8)

(5.4)

(5.4)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

68.6***

71.9***

�3.2

�3.2

(0.0)

(0.0)

(0.0)

(4.0)

(4.3)

(5.6)

(5.6)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

68.6***

72.1***

�3.4

�3.4

(0.0)

(0.0)

(0.0)

(4.0)

(4.1)

(5.5)

(5.5)

Self-em

ploy

ed(%

)Five

nearestn

eigh

bours

match

ing

0.0

0.0

0.0

18.7

14.9

3.8

3.8

(0.0)

(0.0)

(0.0)

(3.2)

(4.1)

(5.4)

(5.4)

Kerne

lmatch

ing:

band

width

=0.01

0.0

0.0

0.0

18.7

12.1

6.6

6.6

(0.0)

(0.0)

(0.0)

(3.2)

(3.6)

(5.5)

(5.5)

Kerne

lmatch

ing:

band

width

=0.05

0.0

0.0

0.0

18.7

12.9

5.8

5.8

(0.0)

(0.0)

(0.0)

(3.2)

(3.4)

(5.5)

(5.5)

Notes:T

hetreatedareworke

rswho

worke

din

theform

alsector

andha

dmon

thly

wag

esfrom

350,00

0to

500,00

0VND

in20

04.T

hematch

edcon-

trol

isalso

worke

rsin

theform

alsector

in20

04who

hadmon

thly

wag

eshigh

erthan

500,00

0an

dlower

than

650,00

0VND

in20

04.T

hetreatedan

dmatch

edgrou

psha

vebe

enmatch

edba

sedon

theclosen

essof

theprop

ensity

score.Colum

ns(1)a

nd(2)rep

ortthe

meanou

tcom

esof

thetreatm

ent

grou

pan

dthematch

edcontrolg

roup

in2004,respe

ctively.

Colum

n(3)isthedifferen

cein

themeanou

tcom

ebe

tweenthetreatm

entan

dcontrol

grou

psin

2004.S

imila

rly,

columns

(4)an

d(5)presen

tthemeanou

tcom

esof

thetreatm

entgrou

pan

dthematch

edcontrolg

roup

in20

06,respe

c-tiv

ely.

Colum

n(6)is

thedifferen

cebe

tweencolumns

(4)an

d(5).Colum

n(7)is

thedifferen

cebe

tweencolumns

(3)an

d(6).Stan

darderrors

inpa

renthe

ses(Stand

arderrors

arecalculated

usingbo

otstrapwith

500replications.Stan

dard

errors

arealso


tsan

dclus

tercorrelation).*sign

ificant

at10

%;**significant

at5%

;***

sign

ificant

at1%

.So

urce:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.



The study’s findings imply that the informal sector can still play an importantrole in generating employment in a developing country. There is no effective unem-ployment insurance in Vietnam. Thus, in the event of job loss, people have tobecome self-employed or to find a wage job in the informal sector quickly. Socialprotection policies such as unemployment insurance will be helpful for the unem-ployed to ensure their living and find another job in the formal sector.

Finally, it should be noted that the study relies on the assumption that there areno spill-over and numeraire effects of minimum wage increases. This assumptionmight be strong. Allowing spill-over and numeraire effects of minimum wageincreases is out of scope of this study, but certainly important for further studies.

References

Abadie, A. and Imbens, G. W. (2008). ‘On the failure of the bootstrap for matching estimators’Econometrica, 76(6) pp. 1,537–1,557

Abowd, J. M., Francis, K. and David, N. M. (1999). ‘Minimum wages and employment inFrance and the United States’, NBER Technical Working Paper No. 6996, Cambridge, MA:National Bureau of Economic Research.

Arango, C. A. and Pach�on, A. (2004), ‘Minimum wages in Colombia: Holding the middle witha bite on the poor’, mimeo, The Colombian Central Bank, Colombia.

Bao Moi (2009). ‘T�ang lương tối thiểu nen mừng hay lo?’, B�ao Mới (New Newspaper). Availableonline at: http://www.baomoi.com/Tang-luong-toi-thieu-nen-mung-hay-lo/47/2615949.epi. Accessed: 8/4/2009.

Bazen, S. (2000). ‘The impact of the regulation of low wages on inequality and labour marketadjustment: A comparative analysis’, Oxford Review of Economic Policy, 16(1), pp. 57–69.

Bell, L. (1997). ‘The impact of minimum wages in Mexico and Colombia’, Journal of Labor Eco-nomics, 15(3), pp. 102–135.

Brown, C. (1999). ‘Minimum wages, employment and the distribution of income’, in Ashenfel-ter, O. and Card, D. (eds.), Handbook of Labor Economics, Vol. 3, Amsterdam: ElsevierScience, 1999, pp. 2,101–2,163.

Brown, C., Gilroy, C. and Kohen, A. (1982). ‘The effect of the minimum wage on employmentand unemployment’, Journal of Economic Literature, 20, pp. 487–528.

Burkhauser, R. V., Couch, K. A. and Wittenburg, D. C. (2000). ‘A reassessment of the new eco-nomics of the minimum wage literature with monthly data from the current populationsurvey’, Journal of Labor Economics, 18(4), pp. 653–680.

Campolieti, M., Fang, T. and Gunderson, M. (2005). ‘Minimum wage impacts on youthemployment transitions, 1993–1999’, Canadian Journal of Economics, 18(1), pp. 81–104.

Card, D. (1992). ‘Do minimum wages reduce employment? A case study of California, 1987–1989’, Industrial and Labor Relations Review, 46(1), pp. 38–54.

Card,D. andKrueger,A. (1994). ‘Minimumwages and employment:A case study of the fast- foodindustry inNewJerseyandPennsylvania’,AmericanEconomicReview, 84(5),pp. 772–793.

Card, D. and Krueger, A. (1995). Myth and Measurement: The New Economics of the MinimumWage, Princeton, NJ: Princeton University Press.


604 Nguyen

Card, D. and Krueger, A. (2000). ‘Minimum wages and employment: A case study of the fast-food industry in New Jersey and Pennsylvania: Reply’, American Economic Review, 90(5),pp.1397–1420.

Cochran, W. G. (1968). ‘The effectiveness of adjustment by subclassification in removing biasin observational studies’, Biometrics, 24, pp. 295–313.

Cochran, W. G. and Chambers, S. P. (1965). ‘The planning of observational studies of humanpopulation’, Journal of the Royal Statistical Society, 128(2), pp. 234–266.

Cunningham, W. V. (2007). Minimum Wages and Social Policy Lessons from Developing Countries,Washington, DC: The World Bank.

Dan Tri (2009a). ‘Tăng lương sẽ không tác động nhiều đến tăng giá’, Báo Dân Trí (People Intel-lectual Newspapers), Hanoi: Vietnam, dated 8/4/2009: Hanoi, Vietnam: Association of Edu-cation Encouragement of Vietnam.

Dan Tri (2009b). ‘Mừng ít, lo nhiều chuyê n tăng lương’, Báo Dan Trí (People Intellectual Newspa-pers), Hanoi: Vietnam, dated 8/4/2009.

Deaton, A. (1997). The Analysis of Household Surveys, Baltimore, MD: The Johns HopkinsUniversity Press.

Dehejia, R. and Wahba, S. (1998). ‘Propensity score matching methods for non-experimentalcausal studies’, NBER Technical Working Paper No. 6829, Cambridge, MA: National Bureauof Economic Research.

Dickens, R., Machin, S., Manning, A., Metcalf, D., Wadsworth, J. and Woodland, S. (1995).‘The effect of minimum wages on UK agriculture’, Journal of Agricultural Economics, 46(1),pp. 1–19.

Dickens, R., Machin, S. and Manning, A. (1999). ‘The effects of minimum wages on employ-ment: Theory and evidence from Britain’, Journal of Labor Economics, 17(1), pp. 1–22.

Dinardo, J., Fortin, N. and Lemieux, T. (1996). ‘Labor market institutions and the distributionof wages, 1973–1992: A semiparametric approach’, Econometrica, 64, pp. 1,001–1,044.

Dowrick, S. and Quiggin, J. (2003). ‘A survey of the literature on minimum wage’, Unpub-lished Manuscript, Australian National University.

Duy-Tuan (2009). ‘Bộ trưởng L��DTB&XH: Nghe thì cứ tưởng là tăng lương…’, VTC News, 8/4/2009.

Fajnzylber, P. (2002). ‘Minimum wage effects throughout the wage distribution: Evidencefrom Brazil’s formal and informal sectors’, Universidade Federal de Minas Gerais, Depart-ment of Economics and CEDEPLAR, Belo Horizante, Brazil.

Freeman,R.B. (1994). ‘Minimumwages–again!’ International Journal ofManpower, 15(2),pp. 8–25.Gindling, T. and Terrell, K. (2004). ‘The effects of multiple minimum wages throughout

the labor market’, Discussion Paper No. 1159, Bonn, Germany: Institute for the Studyof Labor.

Gindling, T. H. and Terrell, K. (2005). ‘The effect of minimum wages on actual wages in for-mal and informal sectors in Costa Rica’, World Development, 33(11), pp. 1,905–1,921.

Government of Vietnam (2005). ‘Government decree on minimum wage’, Decree No. 118/2005/N��D-CP, September 15, 2005.

Government of Vietnam (2006). ‘Government decree on minimum wage’, Decree No. 94/2006/N��D-CP, September 7, 2006.

Government of Vietnam (2008). ‘Government decree on regional minimum wage for labor inenterprises, collective, farm, households and other organizations in Vietnam’, Decree No.110/2008/ND-CP, October 10, 2008.



Government of Vietnam (2009). ‘Government decree on minimum wage’, No. 33/2009/N��D-CP, April 6, 2009.

Gunderson, M. (2005). ‘Minimum wages in Canada: Theory, evidence and policy’, MinimumWage No. 512, Federal Labour Standards Review Commission, Canada.

Ha, A. (2013). ‘Tạo chính sách bình đẳng cho lao động ở khu vực kinh tế phi chính thức’, (Poli-cies to increase equality for labor in the informal sector), People Council Newspaper,Hanoi, Vietnam. Available online at: http://daibieunhandan.vn/default.aspx?tabid=74&NewsId=275462. Accessed: 21/3/2013.

Hahn, J., Todd, P. and van der Klaauw, W. (2001). ‘Identification and estimation of treatmenteffects with a regression-discontinuity design’, Econometrica, 69(1), pp. 201–209.

Hamermesh, D. (1986). ‘The demand for labor in the long run’, in Handbook of Labor Economics,vol. 1 Ashenfelter, O. and Layard, R (eds.), North-Holland: Amsterdam.

Harrison, A. and Scorse, J. (2005). ‘Moving up or moving out? Anti-sweatshop activists andlabor market outcomes’, NBER Technical Working Paper No. 10492, Cambridge, MA:National Bureau of Economic Research.

Heckman, J. and Pag�es, C. (2003), ‘Law and employment: Lessons from Latin America andthe Caribbean’, NBER Working Paper 10129, Cambridge, MA: NBER. Available online at:http://www.nber.org/papers/w10129.

Heckman, J., Ichimura, H., Todd, P. (1997). ‘Matching as an econometric evaluation estima-tors: Evidence from evaluating a job training programme’, Review of Economic Studies, 64(4), pp. 605–654.

Heckman, J., Lalonde, R. and Smith, J. (1999). ‘The economics and econometrics of active labormarket programs’, in Ashenfelter, A. and Card, D. (eds.), Handbook of Labor Economics, Vol.3, Amsterdam: Elsevier Science, pp. 1,865–2,097.

Katz, L. F. and Krueger, A. B. (1992). ‘The effect of the minimum wage on the fast food indus-try’, Industrial and Labor Relations Review, 46(1), pp. 6–21.

Lemos, S. (2004). ‘Minimum wage policy and employment effects: Evidence from Brazil’,Economia, 5(1), pp. 219–266.

Maloney, W. F. and Nu~nez, J. (2001). ‘Measuring the impact of minimum wages: Evidence fromLatin America’, in Heckman, J. J. and Pag�es, C. (eds.), Law and Employment: Lessons from LatinAmerican and the Caribbean,Chicago, IL:UniversityofChicagoPress,pp. 109–130.

Mangan, J. and Johnston, J. (1999). ‘Minimum wages, training wages and youth employment’,International Journal of Social Economics, 26(1), pp. 415–429.

Montenegro, C. E. and Pag�es, C. (2004). ‘‘Who benefits from labor market regulations? Chile,1960–1998’, in Heckman, J. and Pag�es, C. (eds.), Law and Employment: Lessons from LatinAmerica and the Caribbean, Chicago, IL: University of Chicago Press, pp. 401–434.

Neri, M., Gustavo, G. and Camargo, J. M. (2000). ‘Efeitos Informais do Salario Minimo ePobreza’, Texto Para Discussão No. 724, IPEA, Rio de Janeiro.

Neumark, D. and Wascher, W. L. (1992). ‘Employment effects of minimum and subminimumwages: Panel data on state minimum wage laws’, Industrial and Labor Relations Review, 46(1),pp. 55–81.

Neumark, D. and Wascher, W. L. (1994). ‘Employment effects of minimum and subminimumwages: Reply to Card, Katz, and Krueger’, Industrial and Labor Relations Review, 47(3),pp.497–512.

Neumark, D. and Wascher, W. L. (1995). ‘Minimum wage effects on school and work transi-tions of teenagers’, American Economic Review Papers and Proceedings, 85(2), pp. 244–249.


606 Nguyen

Neumark, D. and Wascher, W. L. (2000). ‘The effect of New Jersey’s minimum wage increaseon fast-food employment: A reevaluation using payroll records’, American EconomicReview, 90(5), pp. 1,362–1,396.

Neumark, D. andWascher,W. L. (2002). ‘State-level estimates of minimumwage effects: New evi-dence and interpretations from disequilibrium models’, Journal of Human Resources, 37(1), pp.35–62.

Neumark, D. and Wascher, W. L. (2003). ‘Minimum wages and skill acquisition: Another lookat schooling effects’, Economics of Education Review, 22(1), pp. 1–10.

Neumark, D. and Wascher, W. L. (2007). ‘Minimum wages and employment’, IZA DP No.2570, Bonn, Germany: Institute for the Study of Labor.

Neumark, D., Schweitzer, M. and Wascher, W. (1998). ‘The effects of minimum wagesthroughout the wage distribution’, NBER Working Paper No. 7519, Cambridge, MA:National Bureau of Economic Research.

Pham, T. (2013). ‘Công nhân mãi sống mòn với lương tối thiểu’, Dan Tri Newspaper. Availableonline at: http://dantri.com.vn/xa-hoi/cong-nhan-mai-song-mon-voi-luong-toi-thieu-718866.htm. Accessed: 13/4/2013.

Pierre, G. (2012). ‘Recent labor market performance in Vietnam through a gender lens’, PolicyResearch Working Paper 6056, Washington, DC: The World Bank.

Rama, M. (2001). ‘The consequences of doubling the minimum wage: The case of Indonesia’,Industrial and Labor Relations Review, 54(4), pp. 864–881.

Reich, M. and Hall, P. (2001). ‘A small raise for the bottom: The impact of the 1996 1998California minimum wage increases’, in Lincoln, J. and Ong, P. (eds.), The State of CaliforniaLabor, 2001, Los Angeles, CA: University of California Institute for Labor and Employ-ment, pp. 123–148.

Rosenbaum, P. R. and Rubin, R. (1983). ‘The central role of the propensity score in observa-tional studies for causal effects’, Biometrika, 70(1), pp. 41–55.

Rubin, D. B. (1977). ‘Assignment to a treatment group on the basis of a covariate’, Journal ofEducational Statistics, 2(1), pp. 1–26.

Rubin, D. (1979). ‘Using multivariate sampling and regression adjustment to control bias inobservational studies’, Journal of the American Statistical Association, 74(366), pp. 318–328.

Rubin, D. (1980). ‘Bias reduction using Mahalanobis-metric matching’, Biometrics, 36(2), pp.293–298.

Singell, L. D. and Terborg, J. R. (2006). ‘Employment effects of two Northwest minimum wageinitiatives’, Economic Inquiry, 45(1), pp. 40–55.

Smith, J. and Todd, P. (2005). ‘Does matching overcome LaLonde’s critique of nonexperimen-tal estimators?’ Journal of Econometrics, 125(1–2), pp. 305–353.

Thai-Uyen (2009). ‘T�ang lương tối thiểu, lương hưu’, B�ao Thanh Nien (Youth Newspapers), dated06/04/2009.

Turner, M. D. and Demiralp, B. (2001). ‘Do higher minimum wages harm minority and innercity teens?’ Review of Black Political Economy, 28(4), pp. 95–121.

Van der Klaauw, W. (2002). ‘Estimating the effect of financial aid offers on college enrollment:A regression-discontinuity approach’, International Economic Review, 43(4), pp. 1,249–1,287.



Appendix A: Propensity score matching estimators

The control group is constructed by matching each participant i (workers exposed tothe minimum wage increase) in the treatment group with one or more non-partici-pants (workers not exposed to the minimum wage increase) j whose propensityscore is closest to the propensity score of the participant i. For a participant i, denotenic as the number of non-participants j who are matched with this participant, and w(i,j) the weight attached to the outcome of each non-participant. These weights arenon-negative and sum up to 1, that is,

Pnicj¼1

wði; jÞ ¼ 1. The estimator of ATT is givenby the following:

dATT ¼ 1np

Xnpi¼1

Y20061i �

Xnicj¼1

wði; jÞY20060j

24

35�

Xnpi¼1

Y20040i �

Xnicj¼1

wði; jÞY20040j

24

35

8<:

9=; ð1AÞ

where np is the number of the participants in the data sample. Y20061i and Y2006

0j are theobserved outcomes of participant i and matched non-participant j in 2006 (after theminimum wage increase in 2005), respectively. Y2004

0i and Y20040j are the observed out-

comes of participant i and non-participant j in 2004 (before the minimum wageincrease in 2005), respectively. Equation (1A) can be written as follows:

dATT ¼ 1np

Xnpi¼1

Y20061i � 1

np

Xnpi¼1

Xnicj¼1

wði; jÞY20060j

24

35

� 1np

Xnpi¼1

Y20040i � 1

np

Xnpi¼1

Xnicj¼1

wði; jÞY20040j

24

35: ð2AÞ

The terms in Equation (2A) correspond with the means of treated and matched con-trols in 2004 and 2006 in Table 4.

Participants are matched with non-participants based on the closeness of thevalue of d(i,j) (where d(i,j) is the distance between the propensity score of participanti and that of non-participant j). There are several ways to estimate weights, w(i,j), forall matches. N-nearest neighbours matching gives each matched non-participantequal weight w(i,j) = 1/n. We can also assign different weights to different non-par-ticipants depending on the distance between their propensity score and participants’propensity score (Heckman et al., 1997; Smith and Todd, 2005). The kernel matchingmethod matches a participant with one or many non-participants depending on akernel function and a selected bandwidth. In this article, we used an Epanechnikovkernel with bandwidth of 0.05.


608 Nguyen

Appendix B: Estimation of propensity score

Table B1. Mean and standard deviation of variables in the 2004 VHLSS

Variables Type Mean SD

Married (yes = 1) Binary 0.629 0.483Age Discrete 32.700 11.200Age squared Discrete 1,194.500 795.400Sex (male = 1, female = 0) Binary 0.578 0.494Educational degree (yes = 1)Less than secondary education Binary 0.269 0.444With secondary degree Binary 0.432 0.496With technical degree or post-secondary Binary 0.299 0.458

Main occupation (yes = 1)Agriculture/forestry/fishery Binary 0.063 0.243Unskilled workers Binary 0.253 0.435

Household variablesEthnic minorities (yes = 1) Binary 0.088 0.283Area of annual crop land per capita (1,000 m2) Continuous 0.532 0.978Aquaculture surface per capita (1,000 m2) Continuous 0.026 0.253

Regional dummy variablesRed River Delta Binary 0.248 0.433North East Binary 0.177 0.382North West Binary 0.029 0.167North Central Coast Binary 0.100 0.300South Central Coast Binary 0.130 0.337Central Highlands Binary 0.033 0.178South East Binary 0.122 0.328Mekong River Delta Binary 0.161 0.368Urban areas (yes = 1) Binary 0.316 0.465Number of observations 463

Notes: The sample used to compute estimates in this table includes labourers in the formal sector aged from15 to 60 years old and monthly wages in 2004 below 650,000 VND. Mean and standard deviations are also cor-rected for sampling weights.Source: Estimation from panel data of VHLSSs 2004 and 2006.



Table B2. Logit regressions of the propensity score

Explanatory variables Large model Small model

Married 0.1966(0.3257)

Age �0.1489* �0.1144*(0.0773) (0.0691)

Age squared 0.0021** 0.0017*(0.0010) (0.0010)

Sex (male = 1, female = 0) �0.5442** �0.5112**(0.2589) (0.2534)

Lower or upper secondary 0.3212(0.3169)

Technical degree or post-secondary �0.0909(0.3571)

Agriculture/forestry/fishery �0.2780(0.6532)

Unskilled workers �0.3458(0.3146)

Ethnic minorities (yes = 1) 1.9255*** 1.7718***(0.4700) (0.4335)

Area of annual crop land per capita (1,000 m2) �0.4069* �0.4055**(0.2174) (0.1918)

Aquaculture surface per capita (1,000 m2) 0.1741(0.4145)

Red River Delta Omitted

North East �0.6520 �0.6020*(0.4064) (0.3586)

North West �0.4134(0.8444)

North Central Coast 0.2509(0.3922)

South Central Coast �0.7200* �0.7487**(0.4140) (0.3761)

Central Highlands �1.2979 �1.3914*(0.8739) (0.7903)

South East �1.0070** �1.0144**(0.5081) (0.4522)


610 Nguyen

Table B2 (Continued)

Explanatory variables Large model Small model

Mekong River Delta �0.3851(0.3931)

Urban areas (yes = 1) 0.0283(0.2852)

Constant 1.7397 1.1782(1.2852) (1.1093)

Observations 463 463R2 0.08 0.06

Notes: The treatment group includes workers having monthly wages below 350,000 VND, and the controlgroup includes workers having wages equal to or above 350,000 VND. Thus, the dependent variable is adummy variable indicating ‘monthly wage below 350,000 VND’. Robust standard errors in parentheses. Stan-dard errors are also corrected for sampling weights and cluster correlation. *significant at 10%; **significantat 5%; ***significant at 1%.Source: Estimation from panel data of VHLSSs 2004 and 2006.

Figure B1. Propensity score of the treatment and control groups

Small modelLarge model

0 .2 .4 .6 .8

Propensity score

Untreated Treated

0 .2 .4 .6 .8

Propensity score

Untreated Treated.



Appendix

C:R

obustanalysis

Tab

leC1.

Theim

pacto

nem

ploym

ent:sm

allm

odel

toestimatethepropen

sity

score

Outcom

esan

dmatch

ing

schem

es20

0420

06Difference-in-

differences

Treated

Match

edCon

trol

Difference

Treated

Match

edCon

trol

Difference

(1)

(2)

(3)=(1)�

(2)

(4)

(5)

(6)=(4)�

(5)

(7)=(6)�

(5)

Hav

ejob(%

)Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

97.0***

96.1***

0.9*

0.9*

(0.0)

(0.0)

(0.0)

(1.5)

(2.3)

(1.8)

(1.8)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

97.0***

96.0***

1.0*

1.0*

(0.0)

(0.0)

(0.0)

(1.5)

(2.3)

(1.8)

(1.8)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

97.0***

96.1***

0.9*

0.9*

(0.0)

(0.0)

(0.0)

(1.5)

(1.7)

(1.9)

(1.9)

Hav

eform

alsector

jobs

(%)

Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

58.3***

67.9***

�9.6*

�9.6*

(0.0)

(0.0)

(0.0)

(4.6)

(5.3)

(5.6)

(5.6)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

58.3***

67.9***

�9.6*

�9.6*

(0.0)

(0.0)

(0.0)

(4.6)

(4.5)

(0.1)

(5.5)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

58.3***

68.9***

�10.6**

�10.6**

(0.0)

(0.0)

(0.0)

(4.6)

(3.6)

(5.2)

(5.2)

Self-em

ploy

ed(%

)Five

nearestn

eigh

bours

match

ing

0.0

0.0

0.0

27.2***

14.5***

12.7***

12.7*

(0.0)

(0.0)

(0.0)

(5.0)

(4.4)

(6.9)

(6.5)

Kerne

lmatch

ing:

band

width

=0.01

0.0

0.0

0.0

27.2***

14.8***

12.4***

12.4*

(0.0)

(0.0)

(0.0)

(5.0)

(3.8)

(6.4)

(6.4)

Kerne

lmatch

ing:

band

width

=0.05

0.0

0.0

0.0

27.2***

14.7***

12.6***

12.6**

(0.0)

(0.0)

(0.0)

(5.0)

(2.8)

(5.8)

(5.8)

Notes:T

hose

treatedareworke

rswho

worke

din

theform

alsector

andha

dmon

thly

wag

esbe

low

350,00

0VNDin

2004.T

hematch

edcontrolisalso

worke

rsin

theform

alsector

in20

04who

hadmon

thly

wag

esfrom

350,00

0to

650,00

0VNDin

2004

.Treated

andmatch

edgrou

psha

vebe

enmatch

edba

sedon

theclosen

essof

theprop

ensity

score.Colum

ns(1)a

nd(2)rep

ortthe

meanou

tcom

esof

thetreatm

entg

roup

andthematch

edcontrolg

roup

,resp

ectiv

ely,

in20

04.C

olum

n(3)isthedifferen

cein

themeanou

tcom

ebe

tweenthetreatm

enta

ndcontrolg

roup

sin

2004.Sim

ilarly,

columns

(4)a

nd(5)p

resent

themeanou

tcom

esof

thetreatm

entg

roup

andthematch

edcontrolg

roup

in2006,respe

ctively.

Colum

n(6)isthedifferen

cebe

tweencol-

umns

(4)a

nd(5).Colum

n(7)isthedifferen

cebe

tweencolumns

(3)a

nd(6).The

differen

ce-in

-differen

cesestim

ator

ispresen

tedby

equa

tion(1C)in

App

endix

C.Stand

arderrors

inpa

renthe

ses(Stand

arderrors

arecalculated

usingbo

otstrapwith

500replications.Stand

arderrors

arealso

corrected

forsam

plingweigh

tsan

dclus

terc

orrelatio

n).*sign

ificant

at10

%;**significant

at5%

;***

sign

ificant

at1%

.So

urce:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.


612 Nguyen

Tab

leC2.

Theim

pacto

nem

ploym

ent:control

grou

pshav

ingmon

thly

wag

esbetween350,000an

d550,000VND

Outcom

esan

dmatch

ing

schem

es20

0420

06Difference-in-

differences

Treated

Match

edCon

trol

Difference

Treated

Match

edCon

trol

Difference

(1)

(2)

(3)=(1)�

(2)

(4)

(5)

(6)=(4)�

(5)

(7)=(6)�

(5)

Hav

ejob(%

)Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

97.0***

96.4***

0.6*

0.6*

(0.0)

(0.0)

(0.0)

(1.5)

(2.6)

(1.4)

(1.4)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

97.0***

96.4***

0.6*

0.6*

(0.0)

(0.0)

(0.0)

(1.5)

(2.6)

(1.4)

(1.4)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

97.0***

96.6***

0.4*

0.4*

(0.0)

(0.0)

(0.0)

(1.5)

(2.1)

(1.3)

(1.3)

Hav

eform

alsector

jobs

(%)

Five

nearestn

eigh

bours

match

ing

100.0

100.0

0.0

58.3***

69.3***

�11.0*

�11.0*

(0.0)

(0.0)

(0.0)

(4.6)

(6.1)

(6.6)

(6.6)

Kerne

lmatch

ing:

band

width

=0.01

100.0

100.0

0.0

58.3***

70.2***

�11.9*

�11.9*

(0.0)

(0.0)

(0.0)

(4.6)

(5.5)

(6.6)

(6.6)

Kerne

lmatch

ing:

band

width

=0.05

100.0

100.0

0.0

58.3***

68.9***

�10.6*

�10.6*

(0.0)

(0.0)

(0.0)

(4.6)

(4.7)

(6.1)

(6.1)

Self-em

ploy

ed(%

)Five

nearestn

eigh

bours

match

ing

0.0

0.0

0.0

27.2***

15.7***

11.5*

11.5*

(0.0)

(0.0)

(0.0)

(5.0)

(4.7)

(6.4)

(6.4)

Kerne

lmatch

ing:

band

width

=0.01

0.0

0.0

0.0

27.2***

15.6***

11.6*

11.6*

(0.0)

(0.0)

(0.0)

(5.0)

(4.4)

(6.2)

(6.2)

Kerne

lmatch

ing:

band

width

=0.05

0.0

0.0

0.0

27.2***

16.2***

11.0*

11.0*

(0.0)

(0.0)

(0.0)

(5.0)

(3.9)

(6.2)

(6.2)

Notes:T

hetreatedareworke

rswho

worke

din

theform

alsector

andha

dmon

thly

wag

esbe

low

350,00

0VND

in20

04.T

hematch

edcontrolisalso

worke

rsin

theform

alsector

in20

04who

hadmon

thly

wag

esfrom

350,00

0to

550,00

0VND

in20

04(the

controlg

roup

includ

es20

9worke

rs).The

treatedan

dmatch

edgrou

psha

vebe

enmatch

edba

sedon

theclosen

essof

theprop

ensity

score.

Colum

ns(1)a

nd(2)rep

ortthe

meanou

tcom

esof

thetreatm

entg

roup

andthematch

edcontrolg

roup

in20

04,respe

ctively.

Colum

n(3)isthedifferen

cein

themeanou

tcom

ebe

tweenthetreatm

ent

andcontrolg

roup

sin

2004

.Sim

ilarly,

columns

(4)an

d(5)presen

tthemeanou

tcom

esof

thetreatm

entgrou

pan

dthematch

edcontrolg

roup

in20

06,respe

ctively.

Colum

n(6)isthedifferen

cebe

tweencolumns

(4)an

d(5).Colum

n(7)isthedifferen

cebe

tweencolumns

(3)a

nd(6).The

differ-

ence-in

-differen

cesestim

ator

ispresen

tedby

equa

tion(1C)in

App

endix

C.S

tand

arderrors

inpa

renthe

ses(Stand

arderrors

arecalculated

using

bootstrapwith

500replications.S

tand

arderrors

arealso


tsan

dclus

tercorrelation).*

sign

ificant

at10

%;*

*significant

at5%

;***sign

ificant

at1%

.So

urce:E

stim

ationfrom

pane

ldataof

VHLSS

s2004

and2006.



Table C3. Difference-in-differences regressions: marginal effectsExplanatory variables Dependent variable

is ‘employed’Dependent variable is

‘employed in theformal sector’

Dependent variableis ‘self-employed’

Below minimum wages(yes = 1) 9 Year 2006

0.00006 �0.12932* 0.14110***(0.00055) (0.06734) (0.04931)

Married 0.00589 0.08199 0.06302(0.00936) (0.06204) (0.03910)

Age 0.00039 0.02294 �0.00492(0.00052) (0.01485) (0.01003)

Age squared �0.00001 �0.00038* 0.00009(0.00001) (0.00020) (0.00013)

Sex (male = 1, female = 0) 0.00004 �0.05179 �0.06377*(0.00045) (0.05034) (0.03506)

Lower or upper secondary �0.00189 0.10794* �0.10480***(0.00245) (0.06281) (0.03894)

Technical degree or post-secondary

�0.00219 0.19653*** �0.11405***(0.00387) (0.06310) (0.03430)

Agriculture/forestry/fishery �0.00106 �0.14032 0.10447(0.00336) (0.11835) (0.09013)

Unskilled workers �0.00584 �0.15918*** �0.03215(0.00803) (0.06129) (0.03462)

Area of annual crop landper capita (1,000 m2)

0.00433 0.03003 0.02326(0.00568) (0.02636) (0.01434)

Red River Delta OmittedNorth East �0.00034 �0.01844 0.06791

(0.00134) (0.07018) (0.06506)North Central Coast �0.00054 �0.00943 �0.01287

(0.00141) (0.09638) (0.06279)South Central Coast 0.00020 0.04376 �0.02882

(0.00064) (0.08522) (0.05664)Central Highlands �0.11766 �0.11191 0.05241

(0.16875) (0.20907) (0.11940)South East 0.00037 0.03202 �0.07232*

(0.00059) (0.08594) (0.04366)Mekong River Delta 0.00057 0.05701 �0.00362

(0.00088) (0.06802) (0.05327)Urban areas (yes = 1) 0.00036 0.08392* �0.06808**

(0.00061) (0.04907) (0.03275)Observations 463 463 463R2 0.44 0.08 0.11

Notes: This table presents the difference-in-differences regression. The standard difference-in-differenceregression is

Y¼b0þTDb1þTb2þDb3þXb4þe; ð1CÞ

where Y is employment, T is a year dummy, with a one for 2006 and zero for 2004, D is dummy variable indi-cating monthly wage below 350,000 VND (or wages below 650,000 VND in the second sample), X is the vectorof control variables. The difference-in-difference estimator is the coefficient of interaction between T and D,that is, b1. However, since the treatment and control groups have the same outcomes in 2004, the variables TDand T in difference-in-differences are dropped. When the dependent variable dummy, a probit model is used.


614 Nguyen

We report the marginal effects of explanatory variables in this table, since the probit model is non-linear. Thesample includes labourers in the formal sector in 2004 aged 15 to 60 and monthly wages below 650,000 VND.The treatment group is workers with monthly wages below 350,000 VND. Robust standard errors in parenth-eses. Standard errors are also corrected for sampling weights and cluster correlation. *significant at 10%;**significant at 5%; ***significant at 1%.Source: Estimation from panel data of VHLSSs 2004 and 2006.



Documents

The impact of minimum wages on employment of low-wage workers