How to write research papers on Labor Economic Modelling

Transcription

1 How to write research papers on Labor Economic Modelling Research Methods in Labor Economics and Human Resource Management Faculty of Economics Chulalongkorn University Kampon Adireksombat, Ph.D. EIC Economic Intelligence Center Siam Commercial Bank Outline What should we put in a research paper? Example 1: The Effects of the 1993 Earned Income Tax Credit Expansion on the Labor Supply of Unmarried Women Example 2: The Effects of Decentralized Minimum Wage Setting on Actual Wages in Thailand: Unconditional Quantile Regression Approach Some research ideas using Thai data 1

2 1. What should we put in a research paper? No definite rules, need to make sure that readers will understand what we write Introduction Research questions*** Motivation Brief summary of what we are writing in this paper Institutional details Background of what we want to study Literature review Review what previous work has done Data Empirical approach Summary statistics Regression specification Results Show your results Discuss what you find Conclusion/policy implication 2 Example 1: The Effects of the 1993 Earned Income Tax Credit Expansion on the Labor Supply of Unmarried Women Author: Kampon Adireksombat Published in Public Finance Review (January 2010) Introduction Earned Income Tax Credit (EITC): a refundable income tax credit targets low- and middle-income working families in the United States. the tax credit is paid as a lump sum along with the annual tax return working as an earnings subsidy the biggest group of recipients are unmarried women with children (single mothers) EITC expansion in 1993 Substantially increased the credit available to unmarried women with two or more children (2+group) relative to those with one child (1 child group) and those with no children (no children group) Motivation and research question Check whether the 2+ group increased their labor supply relative to 1 child and No children groups 3

3 Example 1: Institutional details (1) Maximum credits in 2005 dollars dollars $ Year No Children One Child Two Children 4 Example 1 Institutional details (2) (current prices) 5

4 Example 1: Literature review Discuss what previous works examine and find Examples of previous works Dickert, Houser, and Scholz (1995) Eissa and Liebman (1996) Meyer and Rosenbaum (2001) Hotz, Mullin, and Scholz (2006) Contribution Check whether the 2+ group increased their labor supply relative to 1 child and No children groups (different identification strategy) Use national level data 6 Example 1: Data Data: US Current Population Survey Sample: Widowed, divorce, separated or never-married women, aged Period: and

5 Example 1: Empirical Approach: simple check labor force participation 8 Example 1: Empirical Approach: simple check total hours worked 9

6 Example 1: Empirical Approach: simple check hours worked by those already in the labor force 10 Example 1: Empirical Approach: regression equation for labor force participation 11

7 Example 1: Empirical Approach: regression equation for annual hours worked 12 Example 1: Result (1) 13

8 Example 1: Result (2) 14 Example 1: Result (3) 15

9 Example 1: Result (4) 16 Example 1: Conclusion/Policy implication EITC expansion in 1993 Effective policy - Increase labor supply of unmarried women with two or more children - Well-targeted, lower education women increase more labor supply than higher education women In 2009, change the threshold to - No children (max credit = USD 457) - One child (max credit = USD 3,043) - Two children (max credit = USD 5,028) - Three and more children (max credit = USD 5,657) 17

10 Example 2 The Effects of Decentralized Minimum Wage Setting on Actual Wages in Thailand: Unconditional Quantile Regression Approach [Work in progress] Kampon Adireksombat Economic Intelligence Unit Siam Commercial Bank John Giles Development Research Group World Bank 18 In Brief Research Question: What is the effect of the minimum wage on actual wages? Are there any differential effects on the actual wages of workers in formal and informal sectors? Sample Group: Female and male workers aged years old Sample Period: Data: Thai Labor Force Survey Empirical Method: Kernel Function and Unconditional Quantile Regression 19

11 Motivation: Dual Sector Model In the classical dual sector model: an increase in the minimum wage in the formal sector results in a decrease in wages in informal sector Negative relationship between a minimum wage and the actual wages in informal sectors. The Todaro ( AER 1969) model predicts that increases in minimum wages in the urban area can indirectly raise rural wages. Positive relationship between a minimum wage and the actual wages in informal sectors. More recently, McIntyre(2004) s model predicts that an increase in minimum wages can increase wages in both sectors. Ambiguous theoretical prediction 20 Minimum Wage in Thailand 1973-first applied 1998-decentralized to provincial levels following a recommendation by the International Labour Organization (ILO) As a result, there is substantial cross-provincial and cross-time variation in minimum wages 21

12 Definition A minimum wage rate was defined as a wage rate which an employee deserves and is sufficient for an employee s living (Office of National Wage Committee 1996). Unlike other minimum wages in other developing countries, where minimum wages are defined as the payment sufficient for the worker and his family members to dwell in the society (for example, Brazil (Starr 1982)), a Thai minimum wage rate is defined for an employee only. makes the minimum wage and individually actual wage more comparable due to consistent definitions. 22 Minimum Wage (THB) by Region Bangkok Central North Thai Bahts Northeast South 2008 Graphs by Region Year 23

13 Infaltion Rate by Region Overall Bangkok Central Inflation Rate North Northeast South Graphs by Region Year 24 Real Minimum Wage (in 2007 THB) by Region Bangkok Central North Thai Bahts Northeast South Graphs by Region Year 25

14 Decentralized Minimum Wage Setting Under the 1998 Labour Protection Act, minimum wage policy is decentralized into 2 levels the basic level, set by the National Minimum Wage Committee the provincial level, set by the Provincial Minimum Wage Sub committees, The provincial minimum wage must not be below the basic level. In case that no minimum wage is set in a province, the basic level is mandatory enforced. Increased variation in the nominal minimum wage rates from 3 levels in 2001 to 21 levels in Three institutions of the Thai minimum wagefixing machinery National Wage Committee (NWC) : a tripartite committee Provincial Subcommittee on Minimum Wage (PSMW) : a tripartite committee Subcommittee on Technical Affairs and Review (STAR): NWC appoints 11 members to form the STAR 27

15 The Minimum wage setting procedure Each PSMW makes a recommendation on its provincial minimum wage to NWC NWC sends a recommendation to STAR for a technical review STAR submits a review back to NWC for a final consideration and approval 28 Labor Force Survey Data LFS includes detailed information on demographic variables (age, gender, region, marital status, education) and employment and unemployment characteristics (work status, hours worked, salary per month, occupation, industry) , carried out three rounds annually (on February, May and August) , the fourth round (on November) was added From 2001, conducted monthly 29

16 Sample Male and female workers at working age (15-60 years old) With no earning information in LBF, the self-employed workers were excluded from the sample. The resulting sample is 103,418 observations. 9-year data from 1999 (when the LPA fully went into effect) to 2007 To capture the full effects of minimum wage changes given the time that is needed to adjust production process to economize on lowskilled labor. To keep the sample consistent and avoid the issue of repetitive sample, we use the first round data for and February data for Previous Studies A large body of literature on the effect of minimum wage on actual wages in emerging and developing countries focus on Latin America and Caribbean countries For example, Bell 1997; Fajnzylber 2001; Lemos 2002; Strobl and Walsh 2003; Maloney and Núñez 2004; Gindling and Terrell They find consistent results that increases in minimum wages have a positive effects on the wages of formal sector workers A pioneer study on the effect of minimum wage on actual wages in Asia is Rama (2001). Using 1993 aggregate data at the provincial level from Indonesia, Rama finds that doubling the real minimum wage results in an increase in average wages by 5 to 15 percent 31

17 Contribution Decreasing real minimum wage Stable inflation rate Definition of minimum and actual wages are more comparable Exogenous cross-provincial and cross-time variation in minimum wages Unconditional Quantile Regression (Firpo, Fortin, and Lemieux (Econometrica 2009)) 32 How To Define Formal and Informal Sectors? I follow Ginding and Terrell (2005) by using firm size and location. Under the 1998 Labor Protection Act (LPA), a firm with 10 employees or more must file a work regulation, including working conditions with the Ministry of Labor and is required to make a contribution to the Compensation Fund. Therefore, I define workers who work for a firm with 10 or more workers as those in the formal sector and workers who work in a smaller firm as those in the informal sector. To define formality by urban/rural dichotomy, I categorize a firm located in municipal area as an urban firm and a firm located in non-municipal area as a rural firm. This urban/rural definition is consistent with Thai local administration. 33

18 How To Define Formal and Informal Sectors? As a result there are four sectors in my study Urban Large (Formal) Urban Small (Informal) Rural Large (Informal) Rural Small (Informal) 34 Empirical Method Kernel density function of log real wage and analyze whether there are spikes in a distribution of actual wage around the minimum wage rates. Unconditional Quantile Regression (Firpo, Fortin, and Lemieux (Econometrica 2009)) To test whether there are differential effects of minimum wages across sectors, I estimate these equations separately for each sector. 35

19 Kernel : Urban Large Enterprises 1999 Kernel : Rural Large Enterprises Source: Source: Labor data force from survey National of National Statistic Office Statistical and Minitstry Office and of Labor Ministry of Labor Source: Source: Labor data force from survey National of Statistic National Office Statistical and Minitstry Office and of Labor Ministry of Labor Kernel : Urban Small Enterprises Kernel : Rural Small Enterprises Source: Source: Labor data force from survey National of National Statistic Statistical Office and Office Minitstry and of Ministry Labor of Labor Source: 1999 Labor data from force National survey of Statistic National Office Statistical and Minitstry Office and of Labor Ministry of Labor 36 Kernel : Urban Large Enterprises 2007 Kernel : Rural Large Enterprises Source: Source: Labor data force from survey National of Statistic National Office Statistical and Minitstry Office and of Labor Ministry of Labor Kernel : Urban Small Enterprises Source: Source: Labor data force from survey National of National Statistic Office Statistical and Minitstry Office and of Labor Ministry of Labor Kernel : Rural Small Enterprises Source: Source: Labor data force from survey National of National Statistic Statistical Office and Office Minitstry and of Ministry Labor of Labor Source: Labor data force from National survey of Statistic National Office Statistical and Minitstry Office of and Labor Ministry of Labor 37

20 Kernel : No Education Kernel : Less than Elementary Source: Labor data force from National survey of Statistic National Office Statistical and Office Minitstry and of Ministry Labor of Labor Source: Source: data Labor from force National survey Statistic of National Office Statistical and Minitstry Office of and Labor Ministry of Labor Kernel : Elementary Kernel : Lower Secondary Source: 2007 Labor data force from National survey of Statistic National Office Statistical and Office Minitstry and of Ministry Labor of Labor Source: 2007 data Labor from force National survey Statistic of National Office Statistical and Minitstry Office of and Labor Ministry of Labor Kernel : Upper Secondary Kernel : Diploma Source: 2007 Labor data from force National survey Statistic of National Office Statistical and Minitstry Office of and Labor Ministry of Labor Source: Source: Labor data force from survey National of National Statistic Statistical Office and Office Minitstry and Ministry of Labor of Labor Kernel : College Source: Source: Labor data force from survey National of National Statistic Statistical Office Office and Minitstry and Ministry of Labor of Labor 39

21 Average Proportion of Workers Earning at and below the Minimum Wage Proportion Daily Wage Proportion Daily Wage of Minimum wage below 0.95 of Minimum wage All Urban large firms Urban small firms Rural large firms Rural small firms No Education Urban large firms Urban small firms Rural large firms Rural small firms Less than elementay school Urban large firms Urban small firms Rural large firms Rural small firms Elementary school Urban large firms Urban small firms Rural large firms Rural small firms Lower secondary school Urban large firms Urban small firms Rural large firms Rural small firms Upper secondary school Urban large firms Urban small firms Rural large firms Rural small firms Diploma Urban large firms Urban small firms Rural large firms Rural small firms College Urban large firms Urban small firms Rural large firms Rural small firms Note: Author's calculation from Labor Force Survey data. 40 Summary statistics by education level and sector No Education Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Less than Elementary School Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Elementary School Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Lower Secondary School Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Upper Secondary School Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Diploma Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked College Urban Large Urban Small Rural Large Rural Small Real Daily Wage (in 2007 THB) Years of work experience Male Average weekly hours worked Note: Data are from Labor Force Survey, National Statistical Office of Thialand. Means are weighted with sample weights. Standard Deviations are in parenthesis. Sample includes female and male workers aged from years old. 41

22 Estimation Method In Thailand, the effect of minimum wage on actual wages is likely to be different at different points of the wage distribution We need estimation methods that go beyond the mean Conditional quantile regression Limitation: do not average up to their unconditional population counterparts. As a result, cannot be used to estimate the impact of an explanatory variable on the corresponding unconditional quantile. In other words, cannot be used to answer a question as simple as what is the effect on the actual wages of increasing minimum wage by one dollar, holding everything else constant? 42 Unconditional Quantile Regression Proposed by Firpo, Fortin, and Lemieux (Econometrica 2009) to estimate the impact of changes in the explanatory variables on the unconditional quantiles of the outcome variable. First, running a regression of the Recentered Influence Function (RIF) of the unconditional quantile dependent variable on the explanatory variables. Second, running OLS regression of the dependent variable on covariates [Stata command: Rifreg] As a result, we yield RIF-OLS estimate, allowing us to estimate the marginal effects of changes in the distribution of an independent variable on a given quantile of the unconditional distribution of Y, all else equal. 43

23 Regression Equation ln(w it) = β 0 + β 1 ln(mw ipt ) + X it β j + β j Occupation dummies + β k Industry dummies + β t Year dummies + β p Province dummies + β r Region dummies + ε it Note: X it is a vector of demographic characteristics, including age, marital status, education, and potential work experience 44 Elasticities of Real Daily Wage wrt. Real Minimum Wage At each quantile Quantile Baseline (General CPI) Urban Large Urban Small Rural Large Rural Small *** 0.882** 0.293* (0.294) (0.437) (0.172) (0.345) *** 1.390*** 0.463** 0.877*** (0.154) (0.215) (0.184) (0.116) ** * 1.208* ** 0.512** * 1.128** * (0.153) (0.213) (0.145) (0.118) *** 0.734*** 0.767*** (0.217) (0.286) (0.179) (0.217) *** 0.578*** 0.858*** (0.272) (0.208) (0.142) (0.178) * 0.770* ** ** * (0.236) (0.258) (0.138) (0.220) *** (0.258) (0.239) (0.223) (0.177) ** *** (0.281) (0.307) (0.393) (0.232) *** (0.317) (0.417) (0.635) (0.240) Observations 47,665 21,087 20,049 14,617 Notes: Estimated from the Labor Force Survey sample weighted data. Robust standard errors are in parentheses. * (**, ***) signifies statistical significance at the 1 (5,10) percent level. All regressions include experience, province, region, occupation, industry and year dummies. 45

24 Quantile General CPI with Bootstrap Std. errors Urban Large Urban Small Rural Large Rural Small *** 0.882* (0.324) (0.459) (0.178) (0.354) *** 1.390*** 0.463** 0.877*** (0.165) (0.194) (0.217) (0.153) *** 1.208*** 0.512*** 1.128*** (0.147) (0.237) (0.129) (0.142) ** 0.734*** 0.767** (0.232) (0.332) (0.189) (0.305) *** 0.578*** 0.858*** (0.253) (0.211) (0.157) (0.200) * 0.770*** *** (0.252) (0.266) (0.148) (0.240) ** (0.282) (0.245) (0.231) (0.197) * *** (0.290) (0.423) (0.404) (0.246) *** (0.309) (0.540) (0.708) (0.250) Quantile Lag of real minimum wage and General CPI Urban Large Urban Small Rural Large Rural Small *** 1.094** (0.315) (0.450) (0.181) (0.347) *** 1.355*** 0.516*** 0.518*** (0.160) (0.232) (0.200) (0.117) ** 0.887*** *** (0.160) (0.227) (0.157) (0.117) *** 1.209*** (0.225) (0.298) (0.189) (0.225) *** 0.554*** 0.856*** (0.282) (0.218) (0.151) (0.185) ** 0.560** *** (0.250) (0.267) (0.144) (0.226) * (0.270) (0.240) (0.234) (0.187) *** ** (0.295) (0.307) (0.425) (0.235) (0.331) (0.425) (0.685) (0.245) 46 Conclusion What is the effect of the minimum wage on actual wages? A decline in real minimum wage are associated with a decline in the actual wages, especially for workers the lower quantile of wage distribution Are there any differential effects on the actual wages of workers in formal and informal sectors? Yes, the effects are more economically and statistically significant in informal sectors, especially workers in small firms 47

25 Limitation and Future Research Include self-employed and foreign workers into the sample Compare results from pre- and post-decentralization periods Effects on employment and income 48 Some research ideas using Thai data Data sources Bank of Thailand National Economic and Social Development Board National Statistical Office 49

26 Average actual wages in many provinces are lower than the existing minimum wages Actual daily wages and provincial minimum wages in 2010* Unit: THB Proposed new minimum wage: country-wide THB Average actual daily wage Existing provicial minimumwage Phayao Sisaket Maehongson Nontaburi Bangkok Note: Source: Actual daily wages are calculated from 2010 Labor Force Survey (LFS) data, using only the first quarter data (most updated data available) SCB EIC analysis based on data from Labor force survey (National Statistical Office) and Ministry of Labor 50 Sluggish labor force growth in the next decade Philippines Malaysia Vietnam Indonesia Singapore Thailand Korea Unit: % compound annual growth rate 0.4% 1.0% 1.6% 1.6% 2.6% 2.3% 2.5% 2010F 2020F Unit: % compound annual growth rate 2.1% 1.6% 0.9% 1.1% - 0.4% 0.2% - 0.4% Source: SCB EIC analysis based on data from International Data Base (IDB) of US Census Bureau; National Economic and Social Development Board (NESDB) 51

27 Low formal workforce Unit: Million persons 38 Labor force 17 Wage 21 Non-wage 9 Monthly wage 8 Daily wage and others Note: Source: Employees are those employed by private and public sector. Non-employees include own account workers, unpaid family workers, employers and unemployed persons. Unemployed persons include seasonal inactive labor force as well. SCB EIC analysis based on data from Labor Force Survey (National Statistical Office) 52 While GDP has grown, wages have not Unit: Index 2001= Real GDP Real wage F Note: Source: 2010 GDP is from SCB EIC forecast. Wage data are calculated from Labor force survey data, using full-year data, with the exception of 2010 data, using only the first quarter data. SCB EIC analysis based on data from Labor Force Survey (National Statistical Office); National Economic and Social Development Boards 53