The World Bank Economic Review, 39(1), 2025, 191–210 https://doi.org10.1093/wber/lhae022 Article Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Minimum Wage Policy and Poverty in Indonesia Nurina Merdikawati and Ridho Al Izzati Abstract This paper investigates whether minimum wage policy played any significant role in poverty reduction in Java Island, Indonesia, between 2002 and 2014. Its identification strategy exploits variation in minimum wages over time within pairs of geographically proximate districts. The study finds that the minimum wage has a distributional impact on wage workers just below the 20th percentile up to those in the middle of the wage distribution, with no overall loss of employment. However, the minimum wage policy has no distributional impact on per capita household expenditure, and a limited effect on changes in poverty status. JEL classification: I32, J38, O1 Keywords: minimum wage, poverty, Indonesia 1. Introduction Minimum wage has been a prominent labor market policy instrument in both developed and developing countries. The policy has been popular as a tool to improve low-wage workers’ circumstances. Beyond its main purpose to increase low-wage workers’ earnings, policy debates also exist, on the extent to which the minimum wage can be an effective antipoverty program. Its proponents argue that the minimum wage policy increases low-income families’ wages, helping them escape poverty. Employment losses as a result of minimum wage increases are negligible, and those who gain from the policy far exceed those who lose. Opponents disagree, arguing that employment losses can be sizeable, with vulnerable workers losing their jobs and suffering an overall decline in their income.1 Changes in poverty thus hinge upon the labor market impact of minimum wage. Analyzing the distributional impact of minimum wage is particularly relevant in this context; however, there are relatively fewer studies focusing on this matter compared to the large literature on the employ- ment effects of minimum wage (Dube 2019). The literature is even scarcer in developing countries, with only a few studies examining the distributional impact of minimum wage such as Yamada (2016) in the Nurina Merdikawati (corresponding author) is a research fellow at Australian National University, Canberra, Australia; her e-mail address is nurina.merdikawati@anu.edu.au. Ridho Al Izzati is a researcher at the SMERU Research Institute, Jakarta, Indonesia; his e-mail address is rizzati@smeru.or.id. This paper has benefitted from comments from Sarah Dong, Robert Breunig, Xin Meng, Chris Manning, Diana Contreras Suárez, Asep Suryahadi, and seminar participants at the Crawford School of Public Policy, the 2021 Asian and Australasian Society of Labor Economics, and the 2021 Australasian Development Economics Workshop. The authors thank Joseph Marshan for assistance with various aspects of the dataset. All remaining errors are the authors’. A supplementary online appendix for this article can be found at The World Bank Economic Review website. 1 See Belman and Wolfson (2014, 2016) for recent reviews on the employment impact of minimum wage in developed and developing countries, respectively. C The Author(s) 2024. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (https://creativecommons.org/licenses/by- nc- nd/4.0/), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact journals.permissions@oup.com 192 Merdikawati and Izzati context of Indonesia, and Neumark, Cunningham, and Siga (2006) in Brazil. Findings from the distribu- tional effects of the minimum wage would then inform the pathway through which the minimum wage may have any poverty-reducing effects. This paper examines the impact of district minimum wages on poverty in Indonesia from 2002 to 2014. As part of the first-stage analysis, it investigates the distributional impact of district minimum wages on Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 workers’ wages and earnings and also their employment before investigating their distributional impact on per capita household expenditure and their subsequent poverty outcome. This paper uses a cross-sectional individual and household dataset for those living in Java Island dis- tricts. During this period, the average district minimum wages in Java Island increased by 67 percent in real terms, while the poverty rate decreased from 17 percent to 10 percent, and this paper asks whether minimum wage policy has played any role in reducing the poverty rate. To answer this research question, this paper relies on variation in minimum wages across districts and over time, reflecting the context of decentralization since the early 2000s. The district minimum wage–setting process during this period fol- lowed complex negotiations between tripartite committees at district and provincial levels, consisting of representatives from local governments, labor unions, and employers, where the provincial leaders made the final decisions on minimum wage changes. This study considers the district minimum wage–setting process as endogenous, as local leaders are likely to take into account the socioeconomic conditions of their localities when making decisions on annual district minimum wage increments. To address endogeneity concerns, this paper creates a number of unique district-pairs between two adjacent districts, and exploits the variation in minimum wages within each of these district-pairs. With this approach, it assumes that adjacent districts tend to share similar trends and common shocks, thereby providing more reliable control. This applies to analysis of the overall impact of minimum wage on all respondents in the dataset, those disaggregated by their poverty status, and also the distributional analysis using the recentered influence function (RIF) of the unconditional quantile regression (UQR) following Firpo, Fortin, and Lemieux (2009). With this identification strategy, this paper allows local trends to vary over time, provided they are shared within each district-pair. This identification strategy is akin to other methods which restrict varia- tion to geographically proximate areas such as difference-in-spatial differences of Magruder (2013) and Kim and Williams (2021), and spatial fixed effects introduced in Jones et al. (2022), Conley and Udry (2010), Magruder (2012), and Goldstein and Udry (2008). This paper finds that district minimum wages have a positive wage effect for wage workers in the 15th percentile of the wage distribution. This positive effect lingers all the way to those in the middle of the wage distribution. When the sample is disaggregated into poor and nonpoor wage workers, this paper finds that both groups report a positive wage effect of similar magnitude. The wage gain is about 2 percent due to 10 percent increment in the district minimum wages. Wage increases are higher, between 3 percent and 4 percent, for wage workers in the 15th to 30th percentile of the wage distribution. However, this study finds no distributional impact on earnings of nonwage workers,2 as their estimates are less precisely measured, especially for those at the bottom of the distribution; as such none of them are statistically significant. Nonwage workers’ earnings estimates only turn positive and statistically signifi- cant among the nonpoor subgroup. With regard to the employment effect, the findings generally reveal a null effect. However, there is a 0.06 percentage point decline in the probability of being employed among workers belonging to poor households. When the paper further analyses the distributional impact on per capita household expenditure, it also finds zero effect across all percentiles. The coefficient estimates for those in the bottom 20th percentile are close to 0.20, however, the sizeable standard errors render them not statistically significant. In the further 2 Nonwage workers include own-account workers, self-employed with temporary workers, self-employed with permanent workers, casual workers, and family/unpaid workers. The World Bank Economic Review 193 analyses of poverty impact, this paper finds that the increase in district minimum wages does not affect the individual’s poverty status. Overall, the result of the zero poverty-reduction effect of the minimum wage is robust to politicization of the minimum wage, heterogeneous district macroeconomic and labor market conditions, and different ways to construct poverty lines. The paper also shows that migration and commuting are unlikely to Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 confound the analysis. This paper adds to the literature on the welfare impact of minimum wage in a developing country context where labor market segmentation is common. Previous literature analyzing the minimum wage impact on poverty mostly focuses on Latin American and Caribbean countries (Arango and Pachón 2004; Kristensen and Cunningham 2006; Neumark, Cunningham, and Siga 2006; Gindling and Terrell 2010; Alaniz, Gindling, and Terrell 2011; Ham 2018). This paper’s analysis, which starts with the distributional impact on wages and earnings of individual workers and also their employment outcomes, provides a more systematic and compelling assessment of whether the minimum wage may have any effect on per capita household expenditure and changes in poverty outcomes. Its results of zero poverty effect of minimum wage align with that of Yamada (2016), despite different identification strategies and periods of coverage, with this paper evaluating a much more substantial increase in minimum wage changes, both in nominal and real terms. This paper also contributes to the extensive debate on the employment impacts of minimum wage in developing countries generally, and more specifically in Indonesia. Previous studies on Indonesia relied on observations during periods of limited minimum wage variation at the district level (Rama 2001; Suryahadi et al. 2003; Alatas and Cameron 2008; Harrison and Scorse 2010; Comola and Mello 2011; Purnagunawan 2011; Magruder 2013; Del Carpio et al. 2015; Hohberg and Lay 2015). Thus, this paper provides the most recent estimates on the labor market impact of minimum wage increases in the era post decentralization in Indonesia, with rich variation in minimum wages across districts and over time. Moreover, this paper reports labor market outcomes for different subgroups of workers based on their respective per capita household expenditure in relation to the district poverty line. With this approach, the paper provides more insights into whether the minimum wage may have differential effects on workers residing in poor and nonpoor households. The paper also applies a methodology that is more aligned with recent advances in the minimum wage literature, and arguably provides more credible causal estimates. Its identification strategy, which uses variation in minimum wages within each district-pair, would allow each district to have differential trends, so long as those trends are shared among adjacent districts. This method is applied to address the endogeneity of minimum-wage setting, which has rarely been discussed in previous literature on the impact of minimum wage in Indonesia or other developing countries. The remainder of the paper is organized as follows. Section 2 provides background on minimum wage in Indonesia. Section 3 discusses the data and research design. Section 4 presents empirical findings, and section 5 provides a further discussion. Section 6 concludes. 2. Institutional Context Minimum wage has been part of labor market policy in Indonesia since the 1970s. In the 1990s, minimum wages were mostly set at the provincial level and issued by the Ministry of Manpower. In the Ministry’s regulation No. 1/1999, it stipulated that several factors must be considered in minimum-wage setting. These factors included workers’ needs, consumer price indices, firms’ capacity, growth, and sustainability, typical wages in surrounding areas, labor market conditions, economic growth, and per capita income. The Manpower Law of 2003 further reiterated that minimum wages should be set not only based on workers’ needs for decent living, but also taking into account its impact on productivity and economic growth. 194 Merdikawati and Izzati Following decentralization in the early 2000s, there has been much more variation in minimum wages, as district-level governments were given more responsibility for setting their own minimum wages. The district minimum wages were proposed by the district heads to their respective provincial governors after deliberation with tripartite councils consisting of representatives from the district governments, employ- ers’ associations, and workers’ unions. The provincial governors then sought advice from the tripartite Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 provincial wage council before issuing a decree announcing the upcoming changes in the provincial and district minimum wages in their areas of jurisdiction. Typically, provincial minimum wages are set at the lowest district minimum wage in the respective province. District and provincial minimum wages are set on a monthly wage basis. The minimum wage decree sets the monthly minimum wage for one particular year. The district mini- mum wages are revised every year with the provincial governor announcing the newly adjusted minimum wages in November or December of the preceding year. The revised minimum wages become effective in the following year. In the earlier regulatory documents, there was no specific mention to whom the minimum wage law would not apply. However, its de facto implementation pointed to limited legal coverage to wage workers. Meanwhile, nonwage workers, including the self-employed, were often not covered by minimum wage law (Bird and Manning 2008). Only in the Job Creation Law of 2020 were micro and small enterprises3 exempted from the minimum wage law. Based on the above description of the minimum-wage setting in Indonesia, it is likely that district minimum wages would be endogenously determined. Local economic and labor market conditions were likely to be taken into account when determining district minimum wages during the period between 2002 and 2014. 2.1. Trends and Variation in District Minimum Wages The paper plots the trends of the highest and lowest nominal district minimum wages in each respective province on Java Island,4 where its economic output contributes to more than 50 percent of Indonesia’s GDP. Figure 1 shows six panels illustrating trends between 2002 and 2014. For the province of DKI Jakarta (fig. 1a), there is only one minimum wage set at the provincial level. Meanwhile, minimum wages started to vary at the district level in DI Yogyakarta in 2012 (fig. 1d). In West Java (fig. 1b) and East Java (fig. 1e), there were considerable gaps between the highest and lowest district minimum wages in each respective province. Based on fig. 1, 2013 marked the steepest increase, observed in almost all provinces. However, in general, district minimum wages followed a smooth trend. Overall, between 2002 and 2014, the average nominal district minimum wages in Java Island increased almost fourfold and grew by 67 percent in real terms.5 Figure 2 shows the variation in nominal minimum wages across districts on Java Island, between 2002 and 2014. The two maps are color-coded according to the value of district minimum wages by quantile groups. They show that nearby districts tend to have similar minimum wage rates. This suggests that neighboring districts may share similar labor market characteristics, which, in turn, lead to comparable minimum wages. This further motivates the identification strategy later on as the paper attempts to pro- vide a causal impact of the district minimum wages by exploiting variation in minimum wages within nearby districts. 3 Micro and small businesses are either own-account workers or employing up to 19 workers. Their legal status is often in the form of individual businesses or sole proprietorships, of which they can be categorized as informal sector enterprises (Rothenberg et al. 2016). 4 There are 6 provinces and 119 districts on Java Island, while in total, there were 34 provinces and 511 districts in Indonesia by 2014. 5 This is considered a significant increase in the minimum wage, as also experienced in Brazil between 2003 and 2012 where the real minimum wage increased by 62 percent (Saltiel and Urzúa 2022). The World Bank Economic Review 195 Figure 1. Trends of Nominal Minimum Wage from 2002 to 2014 (a) DKI Jakarta, (b) West Java, (c) Central Java, (d) DI Yogyakarta, (e) East Java (f) Banten. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Source: Authors’ analysis using district minimum wages data obtained from the respective provincial governors’ decrees. Note: The above figures show the trends of the minimum wage for the provinces of (a) DKI Jakarta, (b) West Java, (c) Central Java, (d) DI Yogyakarta, (e) East Java, and (f) Banten. Each figure plots the lowest (dotted line) and highest (solid line) district minimum wage (based on 2014) in the respective province. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Merdikawati and Izzati Figure 1 – continued 196 The World Bank Economic Review 197 Figure 2. Variation of Monthly Nominal Minimum Wage across Districts (in Rupiah) (a) 2002. (b) 2014. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Source: Authors’ analysis using the respective provincial governors’ decrees on district minimum wages on Java Island. The shapefile for the map is from Statistics Indonesia, sourced via the Humanitarian Data Exchange platform. 2.2. Wage Distribution Relative to Minimum Wage Figure 3 shows the earnings density for wage and nonwage workers residing on Java Island using the dataset from the National Socioeconomic Survey (Susenas), collected by Statistics Indonesia (BPS), in 2014. The paper did not unpack the nonwage workers into their own category, as information about their earnings tends to be patchy with a lot of missing data (as shown in Table S1.2 in the supplementary online appendix available at The World Bank Economic Review website). Wage workers are categorized as formal sector workers by BPS, along with self-employed with per- manent workers. Family/unpaid workers are categorized as informal sector, while the remaining types of employment can be categorized into formal or informal sector depending on the workers’ types of occu- pation (BPS 2009). This study presents the analysis by grouping them as wage and nonwage workers, due to the extensive missing values for the latter category, as mentioned previously. In 2014, 18.5 percent of employment were own-account workers, 12.2 percent were self-employed with temporary workers, 3.3 percent were self-employed with permanent workers, 40.5 percent were wage workers, 10.7 percent were casual workers, and 14.8 percent were family/unpaid workers. Figure 3a plots the wage density by calculating the difference between the respective individual’s log monthly wage and log district monthly minimum wages for wage workers. In fig. 3b, the difference is between individuals’ log monthly earnings and their respective district monthly minimum wages for non- wage workers. All panels show that compliance with the minimum wages is not complete, as workers are 198 Merdikawati and Izzati Figure 3. Wage Density for Different Types of Workers in 2014 (a) Wage workers, (b) Nonwage workers Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), and governors’ decrees on district minimum wages in provinces on Java Island. Note: Each figure plots the wage density by subtracting the corresponding log monthly wage/earnings of workers residing on Java Island from their respective log district minimum wage. paid below the stipulated district minimum wages. Nevertheless, in fig. 3a, there is a visible spike at zero where the log wages are equal to the log minimum wages. This indicates that district minimum wages are binding, at least, for some employers of wage workers. Based on this, this study hypothesizes that the minimum wage policy will show a positive wage effect among wage workers. For nonwage workers, the bunching at zero is not that visible as shown in fig. 3b. The extent of non- compliance to district minimum wages is also more substantial among nonwage workers. Table S2.1 further shows the earnings density of different types of nonwage workers, where it finds that the bunch- ing in most cases is to the left of their respective district minimum wages, except for the self-employed with permanent workers. This shows the varieties of employment types under the category of nonwage employment, where those who are self-employed with permanent workers typically run more established businesses and receive higher earnings than the other types of nonwage workers. Overall, the paper finds that district minimum wages are unlikely to be binding for nonwage work- ers. Hence, the earnings of nonwage workers may not be affected that much by district minimum wage changes. 3. Research Design and Data Sources In the following section, the paper discusses identification strategy, similarity of control groups, datasets construction, and descriptive statistics. 3.1. Identification Strategy The study estimates the impact of the district minimum wages using individual-level data from Susenas. The classic two-way (district and time) fixed effects regression specification is as follows: yidt = β (MWd (i )t ) + γ Xit + μd (i ) + δt + εidt (1) In equation (1), yidt is the dependent variable for individual i in district d in year t , MWd (i )t is the log of district minimum wages, Xit is a vector of individual-level covariates (age, square in age, and dummies for gender, marital status, education, family size, and number of children), and εidt is the error term. equation (1) controls for district fixed effect, μd (i ) , and time fixed effect, δt . The World Bank Economic Review 199 District minimum wage policies in Indonesia are unlikely to be randomly distributed, as explained in section 2. The minimum wage setting is more likely to be correlated with the underlying socioeconomic factors in the respective localities. The endogeneity concerns come from potential omitted variable bias as a wide range of time-varying factors are taken into account in determining the minimum wage change. Given this, the common trends assumption required by equation (1) is unlikely to be satisfied. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 As mentioned in section 2, and illustrated in fig. 2, district minimum wages among nearby districts tend to be in similar ranges, as shown by similar shades. Despite their proximity, their minimum wages may differ, whether by a small or large magnitude. This is likely to be correlated with their initial condition, or factors that do not change over time, such as their status as city (kotamadya) or regency (kabupaten), or the fact that they belong to two different provinces, all of which are already taken into account in the district fixed effect, μd (i ) . Although nearby districts may have different minimum wage levels at a particular time, this paper argues that they are likely to experience similar local trends and shocks, and so they provide a more suitable control group. The paper adopts this approach into the identification strategy, by taking into account district-pairs-year fixed effects, to address the possible bias from the endogeneity of district min- imum wages. By doing this, the study allows local trends to vary over time, as long as those trends are shared among the district-pairs. As a next step, the paper constructs a unique district-pairs6 in the dataset, and in the preferred speci- fication, it controls for this district-pairs-year fixed effects in the following equation (2): yidt = β (MWd (i )t ) + γ Xit + μd (i ) + τpair(i )t + εidt (2) Here, τpair(i )t represents district-pairs-year fixed effects. This approach is similar in concept to difference-in-spatial-differences used in Magruder (2013) and Kim and Williams (2021), and spatial fixed effect estimators introduced in Jones et al. (2022), Conley and Udry (2010), Magruder (2012), and Goldstein and Uzdry (2008). For further ease of exposition, the district-pairs-year fixed effects is the de-meaning process that for any variable, zidt , z ¯ idt is defined as the mean of z at time t of the pair formed by the district d , yidt − y ¯ it ) + (εidt − ε ¯ idt = β (MWd (i )t − (MW d (i )t ) + γ (Xit − X ¯idt ). By using equation (2), the variation is restricted to proximate districts, which are more likely to share common shocks. In this situation, conditional on covariates and district fixed effects, the district minimum wages are uncorrelated with the residual outcome within each district-pair. This is different from the assumption of common trends across districts that have to be justified in equation (1). For estimating the impact of district minimum wages on individual wage for wage workers, individual earnings for nonwage workers, and per capita household expenditure distribution, the paper uses the re- centered influence function (RIF) of unconditional quantile regression (UQR) following Firpo, Fortin, and Lemieux (2009). In this context, the paper controls for factors related to individual characteristics such as education level, but it would not condition the distributional statistics on those factors (Dube 2019). As such, this is a different approach from conditional quantile regression which defines the quantiles based on those factors. The RIF specification for the ν th quantile, Qv is as follows: RIF (yidt , Qv ) = βv (MWd (i )t ) + γv Xit + μvd (i ) + τvpair(i )t + εvidt (3) The RIF specification in equation (3) controls for a list of covariates, but does not define the quantiles based on them. The main independent variable of interest, βv is the minimum wage effect at the ν th quantile of the outcome variables. All regressions are weighted by the individual weights provided in Susenas, 6 Out of 114 districts on Java Island, the paper constructs 57 unique district pairings, as shown in table S3.1 of the supplementary online appendix. With this unique pairing, if a district shares its border with multiple districts, it only considers the pair with the shortest distance from the centroid, if possible. The results are robust to alternative sets of pairings (results are available upon request). 200 Merdikawati and Izzati unless otherwise specified. The standard errors are clustered by district, which is the unit at which the policy variable varies. 3.2. Similarity of Control Groups Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Following Dube, Lester, and Reich (2016), the paper calculates the mean absolute differences between nonadjacent and adjacent district-pairs in the selected covariates. It considers macroeconomic covariates such as GDP growth and unemployment rate, and also other covariates representing demographic char- acteristics of the districts such as log population, average population age, share of male population, share of primary school graduates or less, share of senior high school graduates or above, and urbanization rate. Data on population and GDP growth is from the World Bank’s Indonesia Database for Policy and Economic Research (INDO-DAPOER), while those on demographic characteristics are calculated from Susenas, and unemployment rate is calculated from Indonesia’s Labor Force Survey (Sakernas). The paper uses the period from 2002 to 2014, which is similar to the main dataset. For each time period (year), it calculates the absolute differences in levels and one-year changes of these variables between (1) a district and its nonadjacent pair and (2) a district and its adjacent district-pair, respectively. The adjacent district-pairs are the main pairs used in the main estimation based on the closest centroid distance (57 pairs). Meanwhile, for nonadjacent district-pairs, each of the 114 districts in 6 provinces is paired with every possible out-of-province district, for a total of 119,988 pairings (district-year).7 Table 1 shows the results for these variables at levels and one-year change. In all covariates, the mean absolute differences are larger for nonadjacent pairs, and in most cases, the gaps are statistically significant. The average percentage gap in absolute differences for the eight variables is about 39 percent for level and 7 percent for one-year change. The gaps are substantially higher for levels of average population age and unemployment rate, and for one-year change, the gaps are higher for log population and GDP growth. In general, the paper concludes that the closest centroid distance district-pair offers a control group with similarity (or balance) in observed covariates. These local district controls may reduce bias from other omitted confounders. 3.3. Dataset Construction The paper uses individual-level cross-sectional data8 from Susenas, collected by BPS, from 2002 to 2014. The Susenas dataset9 provides detailed information on individual and household-level characteristics. The determinant of poverty in Indonesia is not household income since wages and earnings data are more difficult to collect and standardize because most of the people work in the informal sector. Instead, it is 7 Following Dube, Lester, and Reich (2016), this study further collapses the dataset to the initial district-pair-period level. Then, it calculates means of the absolute differences in covariates between districts within pairs. The study clusters the standard error multidimensionally for each of the two districts in the pair. 8 The panel data series is available in Susenas, albeit with a small number of observations and often not representative at the district level. Another dataset with a panel data feature is the Indonesia Family Life Survey (IFLS). Though the sample is representative at the national level, it is not equipped with representative sampling at the district level. The decision to use the cross-sectional dataset from Susenas is not without its shortcomings, as it is not possible to control for individual fixed effects. Nevertheless, Susenas has rich information on individual characteristics that can be used as control variables. Additionally, for the research question analyzing the impact of district minimum wages on poverty, the source of potential omitted variable bias, if any, is more likely from the time invariant and/or time-varying district variables as discussed in section 3, instead of due to individual specific variables. 9 The paper relies mostly on the core datasets in the July rounds for 2002 to 2010. These datasets capture more than 250,000 households in their sample size, which are representative up to the district level. For 2011 to 2014, Susenas was fielded quarterly. In order to have the dataset that is representative at the district level, the paper pooled the observations from the March, June, September, and December rounds. The World Bank Economic Review 201 Table 1. Mean Absolute Differences in Covariates between Nonadjacent and Adjacent District-Pairs Covariates (1) (2) (3) (4) Nonadjacent pairs Adjacent pairs Gap Percentage gap Level Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Log population 0.810∗∗∗ 0.746∗∗∗ 0.065 8.7 (0.036) (0.068) (0.058) Urbanization rate 0.342∗∗∗ 0.285∗∗∗ 0.056∗∗ 19.6 (0.007) (0.022) (0.022) Share of primary school graduates or less 0.183∗∗∗ 0.131∗∗∗ 0.051∗∗∗ 38.9 (0.005) (0.009) (0.010) Share of senior high school graduates or above 0.158∗∗∗ 0.117∗∗∗ 0.040∗∗∗ 34.2 (0.005) (0.009) (0.009) Share of male population 0.014∗∗∗ 0.011∗∗∗ 0.003∗∗∗ 27.3 (0.000) (0.000) (0.000) Average age 1.761∗∗∗ 0.822∗∗∗ 0.935∗∗∗ 113.7 (0.053) (0.043) (0.068) GDP growth 0.019∗∗∗ 0.015∗∗∗ 0.004∗∗∗ 26.7 (0.001) (0.001) (0.001) Unemployment rate 0.039∗∗∗ 0.028∗∗∗ 0.012∗∗∗ 42.9 (0.001) (0.001) (0.001) One year difference Log population 0.021∗∗∗ 0.016∗∗∗ 0.004∗∗∗ 25.0 (0.001) (0.001) (0.001) Urbanization rate 0.014∗∗∗ 0.013∗∗∗ 0.001∗∗ 7.7 (0.000) (0.001) (0.001) Share of primary school graduates or less 0.035∗∗∗ 0.036∗∗∗ -0.000 0.0 (0.000) (0.001) (0.001) Share of senior high school graduates or above 0.033∗∗∗ 0.032∗∗∗ 0.001 3.1 (0.001) (0.001) (0.001) Share of male population 0.010∗∗∗ 0.010∗∗∗ 0.000 0.0 (0.000) (0.000) (0.000) Average age 0.513∗∗∗ 0.484∗∗∗ 0.029∗∗ 6.0 (0.008) (0.013) (0.012) GDP growth 0.022∗∗∗ 0.020∗∗∗ 0.002 10.0 (0.001) (0.002) (0.001) Unemployment rate 0.025∗∗∗ 0.024∗∗∗ 0.001 4.2 (0.001) (0.001) (0.001) Source: Authors’ analysis using data from the World Bank’s Indonesia Database for Policy and Economic Research (INDO-DAPOER), Indonesia’s National Socioe- conomic Survey (Susenas), and Labor Force Survey (Sakernas). Note: The reported estimates are from testing the differences in mean absolute values between nonadjacent and adjacent district-pairs in the selected covariates. Gap is a test of the difference in mean absolute value for each covariate, between nonadjacent and adjacent district-pairs. Percentage gap divides the gap value by the mean for adjacent pairs. ∗∗∗ Statistically significant at 1 percent, ∗∗ 5 percent, ∗ 10 percent. based on whether per capita household expenditure is below the stipulated poverty line. The paper uses the household expenditure from the Susenas core module, whenever available.10 The main sample is limited to those individuals aged 15–64 who reside in districts located on Java Island. This choice is due to the availability of district minimum wages, measured as monthly wages, which were obtained from each provincial governor’s decree.11 Data on district minimum wages outside Java 10 The exceptions are for 2008 (July round), and 2011–2014 (pooled rounds), whose household expenditure is obtained from the Susenas consumption module. 11 The authors thank Chris Manning and Raden M. Purnagunawan for sharing their dataset, which was used to fill any gaps in information from the decrees. 202 Merdikawati and Izzati can be more challenging to compile, as their governors’ offices tend not to make their decrees accessible online.12 The district minimum wages are arguably the effective minimum wages on Java Island after decentralization in the early 2000s. The data on district minimum wages used in the regression analysis are, indeed, the statutory minimum wages stipulated in their respective governor’s decrees. Individual labor market outcomes are available from Susenas collected by BPS. Susenas only provides Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 information on monthly wages for wage workers in 2002–2007 and 2011–2014, and monthly net earn- ings for nonwage workers in 2007,2011, 2013, and 2014. Only wages and earnings from the main source of employment are reported in Susenas. As mentioned in section 2, missing data is more prevalent for earnings of nonwage workers, but less so for wage workers. Wages and earnings can be received in cash or as goods.13 For labor market outcomes, the paper also presents results based on individuals’ poverty status, whether they are poor or nonpoor. Indonesia’s official poverty rate is calculated annually using Suse- nas based on its consumption module. Susenas is designed to assess the poverty rate only at the provincial level in urban and rural areas, respectively. BPS uses a basic need approach to determine the food and nonfood poverty lines and bases them on 52 food items and 36 nonfood items, respectively (BPS 2014). Theoretically, poverty status is based on the position of individuals’ per capita household expenditure in relation to their respective district poverty line. If individuals’ per capita household expenditure is below the district poverty line, they are categorized as poor. If their per capita household expenditure is equal or above the district poverty line, they are nonpoor. However, BPS relies on indirect measures to estimate the poverty rate for each district despite the overall Susenas sampling being designed to be representative at the district level. The paper applies the cumulative distribution function (CDF) method to replicate BPS’s poverty rate at the district level. It uses CDF of the per capita household expenditure for each district to rank the households and separate the sample by using the poverty rate for each district as cut off in the CDF to determine the household poverty status.14 3.4. Descriptive Statistics Table 2 shows summary statistics of the main outcomes (individual and household level). The total em- ployment rate increased from 2002 to 2014 from 61.4 percent to 67.9 percent. The increase is higher in nonpoor households than in poor households. From the same sample, the study separates the workers by their status, namely wage workers and nonwage workers. From table 2, 38 percent (0.234/0.614) and 62 percent (0.380/0.614) of the population are categorized as wage and nonwage workers in 2002. The pattern is still similar in 2014 with a significant increase in the employment rate of wage workers (0.288). A nonpoor individual is more likely to be employed as a wage worker, while the poor worker is more likely to work as nonwage workers. Nonpoor individuals are also more likely to earn higher nominal monthly wages (or earnings) than poor individuals for both types of workers (wage and nonwage workers). From 2002 to 2014, nominal monthly wages for wage workers increased by more than 200 percent. The increase was almost identi- cal between poor and nonpoor individuals. Meanwhile, earnings from nonwage workers also increased 12 Nevertheless, there was no district-level minimum wage in most Outer Island provinces during the early years. In those provinces, minimum wages mostly varied at the provincial level. 13 The phrase “net earnings for nonwage workers” refers to the regular compensation or income received from one’s main job, whether in cash or goods. For self-employed individuals, net earnings are calculated as profit after deducting all expenses from revenue. When earnings are received for a period less than a month, they are equivalently calculated on a monthly basis. 14 Since BPS uses the indirect measure, applying the district poverty line directly to determine poverty status may lead to an inconsistent trend in poverty rate over the year. The World Bank Economic Review 203 Table 2. Summary Statistics Variable 2002 2014 (1) (2) (3) (4) (5) (6) Obs. Mean Std. Dev Obs. Mean Std. Dev. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Individual level Total employment rate All 222,344 0.614 0.487 223,782 0.679 0.467 Poor 35,918 0.633 0.482 203,107 0.650 0.477 Nonpoor 186,426 0.610 0.488 20,675 0.682 0.466 Wage workers employment rate All 222,344 0.234 0.423 223,782 0.288 0.453 Poor 35,918 0.156 0.363 203,107 0.172 0.378 Nonpoor 186,426 0.248 0.432 20,675 0.299 0.458 Nonwage workers employment rate All 222,344 0.380 0.485 223,782 0.391 0.488 Poor 35,918 0.476 0.499 203,107 0.478 0.500 Nonpoor 186,426 0.362 0.481 20,675 0.383 0.486 Monthly wages for wage workers (thousands of rupiah) All 52,191 653.284 1,060.375 61,076 2,034.02 2,411.15 Poor 5,661 298.165 233.541 3,478 937.50 601.84 Nonpoor 46,530 694.716 1,110.311 57,598 2,097.24 2,460.61 Monthly earnings for nonwage workers (thousands of rupiah) 2007 2014 All 32,262 732.004 830.699 78,612 1,369.807 2,364.758 Poor 4,538 435.704 308.386 8,315 691.974 748.604 Nonpoor 27,724 780.562 877.965 70,297 1,451.279 2,476.896 Household level 2002 2014 Poverty status 86,783 0.171 0.377 93,192 0.104 0.305 Monthly household per capita 86,783 205.327 264.065 93,192 789.312 944.472 expenditure (thousands of rupiah) District level Statutory monthly district minimum 106 357.162 92.563 113 1,373.396 474.259 wage (thousands of rupiah) Source: Authors’ analysis using data from Indonesia’s National Socioeconomic Survey (Susenas), Statistics Indonesia (BPS), and the respective provincial governors’ decrees on district minimum wages. Note: Sample means are reported for all individuals in districts located on Java Island. Data on earnings for nonwage workers are available for selected years starting in 2007. Monthly wages and earnings are in nominal rupiah. between 2007 and 2014, with a larger increase for nonpoor individuals (86 percent) compared to poor individuals (59 percent). For the period between 2002 and 2014, the poverty rate decreased from 17.1 percent to 10.4 percent, while the average nominal per capita household expenditure increased substantially, almost fourfold. For the same period for the district level variables, the average nominal statutory district minimum wages also increased by almost fourfold. 4. Empirical Findings The following section provides empirical results of the impact of district minimum wages on labor market outcomes and poverty. 204 Merdikawati and Izzati Figure 4. Distributional effect of Minimum Wage (a) On monthly wages of wage workers, (b) On monthly earnings of nonwage workers Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), and the respective provincial governors’ decrees on district minimum wages. Note: The figure reports coefficients associated with the log of district minimum wage on the log of monthly wages of wage workers (fig. 4a) and the log of monthly earnings of nonwage workers (fig. 4b). The results are based on 19 regressions, where for each regression the specification follows equation (3). All regressions control for district and district-pair-year fixed effects, and individual characteristics comprising gender, marital status, educational attainment, size of household, number of children, age, and age squared. The dark shaded area represents 90 percent, and the light shaded area 95 percent district-cluster-robust confidence intervals. 4.1. Labor Market Impact of District Minimum Wages Before estimating the impact of minimum wage on poverty in Indonesia, the paper first focuses on the labor market impact of minimum wage. The minimum wage policy may have a poverty-reducing effect if, for example, the policy has a strong effect on the income of workers at the bottom of the distribution. The following figures plot the distributional effect of minimum wage on monthly wages of wage workers (fig. 4a) and monthly earnings of nonwage workers (fig. 4b) using the RIF specification in equation (3). Both figures plot the results for 19 RIF regressions for the impact of minimum wage on workers in the 5th percentile up to those in the 95th percentile of wages and earnings of wage workers and nonwage workers, respectively. Figure 4a shows that the minimum wage has a statistically significant positive wage effect for wage workers in the 15th percentile of the wage distribution. The effect dissipates and disappears from the 50th percentile onwards. The effect is not statistically significant for workers at the bottom of the wage distri- The World Bank Economic Review 205 Table 3. The Effect of Minimum Wage on Workers’ Earnings (1) (2) (3) All Poor Nonpoor (a) Log of monthly wage (Wage workers) Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Log of district minimum wage 0.203∗∗∗ 0.184∗ 0.198∗∗∗ (0.051) (0.108) (0.055) Observations 545,006 47,298 497,708 Mean dependent variable 13.61 12.90 13.67 (b) Log of monthly earnings (Nonwage workers) Log of district minimum wage 0.127 −0.235 0.172∗∗ (0.085) (0.239) (0.085) Observations 241,202 27,942 213,260 Mean dependent variable 13.48 12.98 13.55 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), Statistics Indonesia (BPS), and the respective provincial governors’ decrees on district minimum wages. Note: The table reports coefficients associated with the log of district minimum wage on the log of monthly wage for wage workers (row (a)), and log of monthly earnings for nonwage workers (row (b)). District-cluster-robust standard errors are in parentheses. All estimates control for district and district-pair-year fixed effects, and individual characteristics comprising gender, marital status, educational attainment, size of household, number of children, age, and age square. Table 3a shows the results for wage workers whose observations are only available from 2002–2007 and 2011–2014. Table 3b shows the results for nonwage workers which are only reported in 2007,2011, 2013, and 2014. ∗∗∗ Statistically significant at 1 percent, ∗∗ 5 percent, ∗ 10 percent. bution, or those below the 15th percentile. Because Java Island has a high ratio of district minimum wages to average wages, it is not surprising that the district minimum wages have a positive and statistically significant effect on those in the middle of the wage distribution. It was 0.56 in 2002 and increased to 0.74 in 2014. Furthermore, district minimum wages reached the level of median wage in 2014. Meanwhile, fig. 4b shows no statistically significant effects of minimum wage across the distribution of monthly earnings of nonwage workers, as all coefficient estimates hover around zero. The standard errors of estimates for nonwage workers at the bottom of the distribution are also very wide, as data on nonwage earnings tend to be noisy. There are caveats to the wage and earnings data the paper uses in both estimates. However, these are the more reliable and relevant data sources for the analysis, as detailed in section 3. Table 3 presents the overall wage and earnings effects of the district minimum wages. The paper also provides estimates for the subgroups of workers belonging to poor and nonpoor households. In Indone- sia’s context, poverty measurement is based on per capita household expenditure as income is difficult to measure precisely as majority of the population work in the informal sector. The groupings of poor and nonpoor are based on whether each worker belongs to a household whose monthly expenditure per capita is below or equal to and above their respective district poverty line. Table 3 shows that the minimum wage has a positive wage effect on all wage workers, poor and nonpoor alike. The elasticity estimates among the poor and nonpoor are relatively similar, implying that a 10 percent increase in the statutory district minimum wages increases the monthly wage of wage workers by about 2 percent. This aligns with the findings from the distributional wage impact in fig. 4a. This further shows that the wage impact is higher and more statistically significant for those in the 15th to 30th percentile of the wage distribution. 206 Merdikawati and Izzati Table 4. The effect of Minimum Wage on Workers’ Employment (1) (2) (3) All Poor Nonpoor (a) Probability of being employed Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Log of district minimum wage −0.005 −0.055∗∗ 0.004 (0.015) (0.027) (0.016) Observations 2,934,258 378,812 2,555,446 Mean dependent variable 0.648 0.638 0.649 (b) Probability of being employed as wage workers Log of district minimum wage 0.008 0.020 0.008 (0.017) (0.034) (0.016) Observations 2,934,258 378,812 2,555,446 Mean dependent variable 0.255 0.161 0.268 (c) Probability of being employed as nonwage workers Log of district minimum wage −0.013 −0.075 −0.003 (0.020) (0.047) (0.019) Observations 2,934,258 378,812 2,555,446 Mean dependent variable 0.393 0.478 0.381 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), Statistics Indonesia (BPS), and the respective provincial governors’ decrees on district minimum wages. Note: The table reports coefficients associated with log of district minimum wage on each of respective dependent variable. The dependent variable in each row (a), (b), and (c) is a dummy variable equal to 1 if the individual is employed, employed as a wage worker, and employed as a nonwage worker, respectively, and 0 otherwise. District-cluster-robust standard errors are in parentheses. All estimates control for district and district-pair-year fixed effects, and individual characteristics comprising gender, marital status, educational attainment, size of household, number of children, age, and age square. ∗∗∗ Statistically significant at 1 percent, ∗∗ 5 percent, ∗ 10 percent. Meanwhile, the study finds a null effect for minimum wage impact on nonwage workers’ earnings. However, it detects a positive earnings effect among the nonpoor nonwage workers,15 suggesting that the “lighthouse effect” is in place, but the effect is zero for poor nonwage workers. The paper acknowledges its limitations where it cannot provide the distributional analysis of household income and instead, only reporting on the distributional analysis of individuals’ wages and earnings. When the dataset is constructed at the household level, the paper finds that more than half of the households in the sample have incomplete information on their household income. This is related to the prevalence of missing data on wages and earnings as detailed in table S1.1 and S1.2 of the supplementary online appendix, where missing values are more common for nonwage workers’ earnings. In section S4 of the supplementary online appendix, the paper shows the distributional effect of minimum wage on nonzero household combined reported income. With less than half of observations, it finds the null effect across all percentiles except of those at the 15th percentile. Table 4 estimates the employment impact of district minimum wages. The study finds no impact on employment in general. However, there is a negative employment impact for the poor with a 1 per- 15 This paper does not explore potential mechanisms through which the minimum wage may increase the earnings of nonwage workers. One possibility is that increases in wages among wage workers may lead to increases in demand for goods and services produced by nonwage workers, which in turn may raise nonwage workers’ earnings. The World Bank Economic Review 207 Figure 5. Distributional Effect of Minimum Wage on per Capita Household Expenditure Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), and the respective provincial governors’ decrees on district minimum wages. Note: The figure reports coefficients associated with the log of district minimum wage on the log of per capita household expenditure. The results are based on 19 regressions where for each regression, the specification follows equation (3). All regressions control for district and district-pair-year fixed effects, and individual characteristics comprising gender, marital status, educational attainment, size of household, number of children, age, and age square.d The dark shaded area represents 90 percent, and the light shaded area 95 percent district-cluster-robust confidence intervals. cent increase in district minimum wages reducing the probability of being employed by 0.06 percentage points. 4.2. Impact of the Minimum Wage on Poverty This study reveals that the minimum wage has a positive wage impact on poor wage workers, but there is also a possibility of job loss with the poor workers reporting negative probability of being employed. The following section examines whether the minimum wage has any effect on the welfare of households, based on their per capita household expenditures. In Indonesia, poverty status is often measured by household per capita expenditure, where the poor are usually at the bottom 20th of the distribution. Figure 5 plots the distributional effect of minimum wage on per capita household expenditure using the RIF specification in equation (3). The results show zero minimum wage effect across per capita household expenditure distribution. The elasticity that measures the impact of percentage changes in the district min- imum wages on per capita household expenditure is close to 0.20 at the 20th to 30th percentile, however, with sizeable standard errors, none of them are statistically significant. Table 5 further estimates the overall impact of household per capita expenditure, and, as expected, the paper finds that there is no statistically significant effect. When the impact is disaggregated among poor and nonpoor households, null results remain. The paper further estimates the minimum wage impact on household poverty status, where it also finds zero effect. 5. Discussions Despite promising minimum wage impact on the labor market outcomes of poor workers as shown in section 4, the study finds that minimum wage has a limited poverty-reducing impact. The results align with Yamada (2016), who evaluated the impact of provincial minimum wages in Indonesia between 1993 and 2000. For the present paper, it is possible that the zero effects of minimum wage on the poverty rate is because the wage gains from the increase in the minimum wage are limited to wage workers, while the majority 208 Merdikawati and Izzati Table 5. The Effect of Minimum Wage on Household per Capita Expenditure and Poverty Status Log household per capita expenditure (1) (2) (3) (4) All Poor Nonpoor Poverty status Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 Log of district minimum wage 0.050 0.062 0.057 0.001 (0.051) (0.056) (0.052) (0.012) Observations 1,183,224 137,332 1,045,892 1,183,224 Mean dependent variable 12.73 11.83 12.85 0.113 Source: Authors’ analysis using individual-level data from Indonesia’s National Socioeconomic Survey (Susenas), Statistics Indonesia (BPS), and the respective provincial governors’ decrees on district minimum wages. Note: The table reports coefficients associated with the log of district minimum wage on each of respective dependent variable. For poverty status, the analysis assigns a dummy of 1 for households with per capita expenditure below the district poverty line and 0 otherwise. District-cluster-robust standard errors are in parentheses. All estimates control for district and district-pair-year fixed effects, and the characteristics of the heads of households comprising gender, marital status, educational attainment, size of household, number of children, age, and age square. ∗∗∗ Statistically significant at 1 percent, ∗∗ 5 percent, ∗ 10 percent. of workers in the poor households rely on their earnings as nonwage workers, as seen in table 2. Even among the wage workers, the compliance rates to the minimum wage law are far from perfect as shown in fig. 3. Moreover, wage workers belonging to households in the bottom 20th percentile of household per capita expenditure tend to work for firms not paying the monthly stipulated district minimum wages, as seen in table S5.1 of the supplementary online appendix. Additionally, wage workers earning around the district minimum wages may often live in more affluent households (section S5 of the supplementary online appendix).16 The share of females as minimum wage workers17 who live in more affluent households is also higher; as such they are more likely to be the secondary earners in the household. Minimum wage in Indonesia is also set based on the basic needs of a single worker instead of the needs of a family, while poverty measurement in Indonesia is based on per capita household expenditure. Thus, the wage gains from a single worker in the household may have a limited effect on boosting the per capita household expenditure to a level that exceeds their respective district poverty line. The paper tests the robustness of the estimates to the inclusion of other covariates, such as taking into account politicization of district minimum wages, and economic and labor market characteristics at the district level. The results are robust and hardly change after including those factors in the estimation, as shown in section S6.1 and S6.2 of the supplementary online appendix. The paper also further analyzes whether migration and commuting may confound the treatment effect of which it finds that they are unlikely to confound the coefficient estimates (section S6.3 of the supplementary online appendix). As an additional robustness test, the paper re-runs the regressions for the subgroups of poor and nonpoor respondents using a variety of different ways to measure the poverty line, and the results are generally robust.18 16 This is not surprising as the ratio of district minimum wage to average wage in Indonesia is considered high, and the ratio increased from 0.56 in 2002 to 0.74 in 2014. Meanwhile, the district minimum wage reached the level of the median wage in 2014. The district minimum wage was also set way above the poverty line, with the ratio of 3.50 in 2002 and 4.61 in 2014. 17 The paper defines minimum wage workers as those wage workers who earn between 0.9 and 1.1 times their respective district minimum wage. 18 The paper considers other ways of constructing poverty lines that may change the composition of poor and nonpoor respondents. They are as follows: (1) adopting the $2.15 PPP global extreme/absolute poverty line from the World Bank; (2) calculating the average of the district poverty line in 2002 for Java Island and then inflating it yearly using the consumer price index; and (3) determining the poverty line as the 20th percentile in 2002 for Java Island and then inflating it yearly using the consumer price index. Results are available upon request. The World Bank Economic Review 209 6. Conclusion Minimum wage policy in developing countries is often advocated as a labor market instrument to protect low-wage workers. However, it is less clear to what extent the policy is well targeted at improving the living standard of the poor and eventually reducing the poverty. This issue has to be studied empirically. Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 This paper seeks to understand whether the minimum wage has any poverty-reducing impact in In- donesia, a nation that has been actively implementing minimum wage policy since the late 1980s and at the same time has been succeeding in further reducing its poverty rate. The study exploits minimum wage variation across districts and over time on Java Island in Indonesia. In the distributional impact analysis, the study finds a positive wage effect for wage workers in the 15th percentile of the wage distribution up to those in the middle of the wage distribution, while there is no distributional effect for nonwage workers. Even with a relatively limited employment loss, the positive wage effect among wage workers fails to increase overall per capita household expenditure, which is used as a proxy to determine poverty status in Indonesia’s context. As such, it is not surprising that the paper finds no changes in poverty status. Overall, the poverty-reducing effect of the minimum wage policy in Indonesia seems to be limited. The positive wage effect is only found among wage workers, while the majority of the poor work as nonwage workers where the minimum wage is less likely to be effective. Even among the poor wage workers who benefit from the district minimum wage increase, the minimum wage hikes are unlikely to be substantial enough to lift poor workers’ average per capita household expenditure to rise above the poverty line and eventually show a statistically significant poverty-reducing effect. The results suggest the limits of minimum wage policy as an effective poverty-reduction tool. Other factors may be at play in reducing poverty in Indonesia. The minimum wage policy has constantly been revised, and starting in 2016, increments have been tied to national economic growth and inflation. Pre- sumably, the objective is to have more predictable increases in the annual minimum wage (Dong and Manning 2017) compared to the previous regime, where there were instances of sudden substantial in- creases, potentially linked to political motives. Further revision of the minimum wage policy was intro- duced in October 2020 through the Job Creation Law. This raises questions on how the newly enacted law may affect labor market outcomes and to what extent labor market policy can have any welfare effect in the country, which warrants further research on this topic. Conflict of Interest None. Data Availability Statement The data underlying this article will be shared on reasonable request to the corresponding author. References Alaniz, E., T.H. Gindling, and K. Terrell. 2011. “The Impact of Minimum Wages on Wages, Work and Poverty in Nicaragua.” Labour Economics 18(SUPPL. 1): S45–59. Alatas, V., and L.A. Cameron. 2008. “The Impact of Minimum Wages on Employment in a Low-Income Country: A Quasi-Natural Experiment in Indonesia.” Industrial and Labor Relations Review 61(2): 201–23. Arango, C.A., and A. Pachón. 2004. “Minimum Wages in Colombia: Holding the Middle with a Bite on the Poor.” Working Paper No. 280. Central Bank of Colombia. Bogota, Colombia. Belman, D., and P.J. Wolfson. 2014. “What Does the Minimum Wage Do?” W.E. Upjohn Institute for Employment Research. Kalamazoo, MI, USA. Belman, D., and P.J. Wolfson. 2016. “What Does the Minimum Wage Do in Developing Countries? A Review of Studies and Methodologies.” Conditions of Work and Employment Series No. 62. ILO. Geneva, Switzerland. 210 Merdikawati and Izzati Bird, K., and C. Manning. 2008. “Minimum Wages and Poverty in a Developing Country: Simulations from Indonesia’s Household Survey.” World Development 36(5): 916–33. BPS. 2009. 2006 Economic Census: Analysis of Labor Force. Jakarta: BPS. BPS. 2014. 2014 Calculation and Analysis of Macro Poverty in Indonesia. Jakarta: BPS. Comola, M., and L.D. Mello. 2011. “How Does Decentralized Minimum Wage Setting Affect Employment and In- Downloaded from https://academic.oup.com/wber/article/39/1/191/7674037 by WORLDBANK THIRDPARTY user on 05 February 2025 formality? The Case of Indonesia.” Review of Income and Wealth 57(SUPPL. 1): S79–99. Conley, T.G., and C.R. Udry. 2010. “Learning about a New Technology: Pineapple in Ghana.” American Economic Review 100(1): 35–69. Del Carpio, X., H. Nguyen, L. Pabon, and L.C. Wang. 2015. “Do Minimum Wages Affect Employment? Evidence from the Manufacturing Sector in Indonesia.” IZA Journal of Labor and Development 4(17):1-30. Dong, S.X., and C. Manning. 2017. “Labour-Market Developments at a Time of Heightened Uncertainty.” Bulletin of Indonesian Economic Studies 53(1): 1–25. Dube, A. 2019. “Minimum Wages and the Distribution of Family Incomes.” American Economic Journal: Applied Economics 11(4): 268–304. Dube, A., T.W. Lester, and M. Reich. 2016. “Minimum Wage Shocks, Employment Flows, and Labor Market Fric- tions.” Journal of Labor Economics 34(3): 663–704. Firpo, S., N.M. Fortin, and T. Lemieux. 2009. “Unconditional Quantile Regressions.” Econometrica 77(3): 953–73. Gindling, T.H., and K. Terrell. 2010. “Minimum Wages, Globalization, and Poverty in Honduras.” World Development 38(6): 908–18. Goldstein, M., and C. Udry. 2008. “The Profits of Power: Land Rights and Agricultural Investment in Ghana.” Journal of Political Economy 116(6): 981–1022. Ham, A. 2018. “The Consequences of Legal Minimum Wages in Honduras.” World Development 102: 135–57. Harrison, A., and J. Scorse. 2010. “Multinationals and Anti-Sweatshop Activism.” American Economic Review 100(1): 247–73. Hohberg, M., and J. Lay. 2015. “The Impact of Minimum Wages on Informal and Formal Labor Market Outcomes: Evidence from Indonesia.” IZA Journal of Labor and Development 4(14): 1-25. Jones, M., F. Kondylis, J. Loeser, and J. Magruder. 2022. “Factor Market Failures and the Adoption of Irrigation in Rwanda.” American Economic Review 112(7): 2316–52. Kim, J.H., and B.D. Williams. 2021. “Minimum Wage and Women’s Decision-Making Power within Households: Evidence from Indonesia.” Economic Development and Cultural Change 70(1): 359–414. Kristensen, N., and W. Cunningham. 2006. “Do Minimum Wages in Latin America and the Caribbean Matter? Evi- dence from 19 Countries.” Policy Research Working Paper No. 3870. World Bank. Washington, DC, USA. Magruder, J.R. 2012. “High Unemployment yet Few Small Firms: The Role of Centralized Bargaining in South Africa.” American Economic Journal: Applied Economics 4(3): 138–66. Magruder, J.R. 2013. “Can Minimum Wages Cause a Big Push? Evidence from Indonesia.” Journal of Development Economics 100(1): 48–62. Neumark, D., W. Cunningham, and L. Siga. 2006. “The Effects of the Minimum Wage in Brazil on the Distribution of Family Incomes: 1996–2001.” Journal of Development Economics 80(1): 136–59. Purnagunawan, R.M. 2011. “Minimum Wages and Labour Market Outcomes in Decentralized Indonesia: A Case Study from Java.” PhD thesis. Canberra, Australia: Australian National University. Rama, M. 2001. “The Consequences of Doubling the Minimum Wage: The Case of Indonesia.” Industrial and Labor Relations Review 54(4): 864–81. Rothenberg, A.D., A. Gaduh, N.E. Burger, C. Chazali, I. Tjandraningsih, R. Radikun, C. Sutera, and S. Weilant. 2016. “Rethinking Indonesia’s Informal Sector.” World Development 80: 96–113. Saltiel, F., and S. Urzúa. 2022. “Does an Increasing Minimum Wage Reduce Formal Sector Employment? Evidence from Brazil.” Economic Development and Cultural Change 70(4): 1403–37. Suryahadi, A., W. Widyanti, D. Perwira, and S. Sumarto. 2003. “Minimum Wage Policy and Its Impact on Employment in the Urban Formal Sector.” Bulletin of Indonesian Economic Studies 39(1): 29–50. Yamada, K. 2016. “Tracing the Impact of Large Minimum Wage Changes on Household Welfare in Indonesia.” European Economic Review 87: 287–303.