How Do Households Adjust to Trade Liberalization? Evidence from China s WTO Accession

Transcription

1 How Do Households Adjust to Trade Liberalization? Evidence from China s WTO Accession Mi Dai Beijing Normal University Wei Huang National University of Singapore Yifan Zhang Chinese University of Hong Kong Oct. 24, 2017

2 Motivation A large wave of trade liberalization in developing countries over the last few decades. Trade liberalization is associated with substantial adjustments in the labor market. Regions exposed to larger tariff cuts experienced negative labor market consequences, in particular wage losses (Topalova, 2010; Kovak, 2013; Dix-Carneiro and Kovak, 2017). But the existing literature of trade liberalization focuses almost exclusively on wage and employment effects. We emphasize the role of households in insuring individuals against the labor market risks induced by trade liberalization.

3 This Paper We find a series of behavioral response of households to smooth out consumption reduction in trade liberalization episodes. Labor supply changes Parental co-residence Changes in savings Consumption smoothing through both the earnings channel (labor supply changes) and expenditure channel (parental co-residence, savings) We find no evidence of Income transfer Household borrowing/lending We show these responses are quantitatively important in weakening the negative impacts of trade liberalization on household consumption

4 This Paper We investigate household adjustments to tariff cuts due to WTO entry in urban China Why China? WTO accession in December 2001: exogenous tariff changes A comprehensive household survey data with rich information - e.g. wages, working status, household structure, household income, and consumption records Methodology: "local labor market approach" Identification is based on the variation of tariff changes across industries, and the variation of pre-wto industry mix of employment across Chinese cities.

5 Main Findings Regions that initially specialized in industries facing larger tariff cuts experienced more significant wage declines relative to other regions. Household respond to this income shock in several ways: 1. Labor supply increases Labor supply increased at both intensive and extensive margin Labor supply increases are strongest for females and for the elderly * consistent with "added worker effects" in labor literature Employment transition from tradable to non-tradable sector, particularly for males. 2. More young adults co-reside with parents. 3. Save less Back of envelope calculation suggests that consumption reduction due to trade liberalization would be 30%-50% larger if households had not taken these behaviors (still in progress).

6 Literature Local labor market effects of trade liberalization Topalova (2010); Edmond, Pavcnik and Topalova (2010); Kovak (2013); Dix-Carneiro and Kovak (2014, 2017); Hakobyan and McLaren (2015) - We document novel margins of adjustments at household level - These adjustments offset the negative labor market consequences of trade liberalization The impact of China s WTO entry On other countries: Autor et al. (2013); Pierce and Schott (2016); Bloom et al. (2016); Che et al. (2016) On China: Han et al. (2012); Yu (2015); Lu and Yu (2015); Fan et al. (2015); Brandt et al. (2017) Household response to income shocks (Kochar,1999; Kaplan,2012; Gorbachev, 2015; Blundell et al.,2016)

7 Empirical Strategy We estimate the following equation at either individual or household level Yit = α + βtariffc, t 1 + γd( cityc, yeart, ageit, genderit, educit ) + εit Yit: Individual/household level outcomes Tariffc,t-1: Regional tariff level of prefecture city c. D: city*gender; gender*year*age; gender*education of an individual (for individual regressions) or household head (for household regressions) Standard errors are clustered at the city level.

8 Empirical Strategy Measurement of regional tariff: Tariff = λ τ c, t jc, j, t j λ jc, is industry j s share in city c s tradable sector employment between (pre-wto years). Identification relies on two sources of variation: Variation in the degree of tariff reduction across industries Variation in the pre-wto industry mix of employment across cities Check for robustness using alternative weighing schemes Use year 2001 s employment weights Weights adjusted for labor share (Kovak, 2013)

9 Instrumental Variable Estimation Tariff reduction may be subject to political economy considerations (Grossman and Helpman, 1994). But this is unlikely in the context of China s WTO accession There was very little policy discretion in the extent of trade liberalization in each industry. Regardless of the initial level, after the WTO entry, the post-wto tariffs converged to a relatively uniform level. As a robustness check, we follow Brandt et al. (2017) and use maximum allowable tariff rate as an IV for actual tariff rate. These rates were basically determined in 1999.

10 Initial Tariff v.s. Tariff Change tariff tariff city tariff 2001

11 Data Urban Household Survey (UHS) from NBS, Covering 179 cities in 18 provinces; Repeated cross-section Beijing, Liaoning, Shanghai, Jiangsu, Zhejiang, Shandong and Guangdong are coastal provinces; Shanxi, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Chongqing, Sichuan, Yunnan, Gansu and Xinjiang are inland provinces In 2007, our sample includes 131,000 individuals in 179 prefecturelevel cities Variables: - Labor market outcomes (wage, working status) - Household member characteristics - Household balance sheet (income and expenditure) - Consumption records

12 Data Annual Survey of Industrial Firms from NBS All state-owned and above scale non-state-owned firms Calculate total employment by city and 4-digit industry Used to construct employment weights in regional tariff Tariff data from Chinese Customs Chinese population census data in 2000 and 2005 from NBS

13 Summary Statistics (1) (2) (3) Panel A: Individual level variables Variable Full sample Age < Retire age Age >= Retire age Working (Yes = 1) (0.45) (0.36) (0.37) Working at tradable sector (Yes = 1) (0.38) (0.41) (0.11) Working at nontradable sector (Yes = 1) (0.50) (0.48) (0.36) Working hours (Monthly) (91.9) (80.9) (52.1) Log(wage) (1.06) (0.95) (1.78) Observations 591, , ,440 Panel B: Household level variables Variable Full sample HH head age < 50 HH head age > 50 Household size (0.83) (0.63) (1.01) Parental coresidence (0.46) (0.37) (0.50) Household income per capita (86.1) (85.0) (86.8) Consumption per capita (55.4) (54.4) (56.6) Saving rate (0.25) (0.25) (0.26) Observations 251, , ,228

14 Evolution of Regional Tariff,

15 Tariff Change by Region,

16 Regional Tariff Change City name Tariff 1998 Tariff 2007 Tariff change 5 cities with largest tariff cut Major industry in Shi yan Manufacture of Transport Equipment Hao zhou Manufacture of Beverages Zhou kou Manufacture of Textile Fu yang Manufacture of Beverages Pu ning Manufacture of Textile 5 cities with smallest tariff cut Jin chang Smelting and Pressing of Non-ferrous Metals Chang zhi Mining and Washing of Coal He gang Mining and Washing of Coal Pan zhi hua Smelting and Pressing of Ferrous Metals Qi tai he Mining and Washing of Coal

17 Wage Growth v.s. Tariff Change

18 Labor Supply Change v.s. Tariff Change

19 Co-residence Change v.s. Tariff Change

20 Household Income Change v.s. Tariff Change

21 Consumption Change v.s. Tariff Change

22 Wage Effects Dep. Var: (1) (2) (3) (4) (5) (6) Log (Wage) OLS 2SLS Sample Full sample Tradable Non-tradable Full sample Tradable Non-tradable Tariff 1.76*** 2.22*** 1.45*** 2.67*** 3.05*** 2.47*** (0.47) (0.77) (0.42) (0.74) (0.90) (0.72) Observations 379,389 95, , ,389 95, ,225 R-squared Basic controls Yes Yes Yes Yes Yes Yes Clusters Regional tariff cut reduced regional wages (relative to other regions). Larger effect in tradable sector. Smaller but significant effect in non-tradable sector.

23 Wage Effects: Mechanisms & Cross-validation Using Firm-level data in China s industrial sector, we find: 1. In regions with larger tariff cut, firms pay lower wages, have lower profit, lower investment, and lower domestic sales. 2. Such effect does not exist during the pre-wto period ( ). 3. Similar conclusions hold using industry-level tariff as regressor. Import competition induced by tariff reduction has an short-run adverse effect on firms, transmitting to lower wages for workers.

24 Wage Effects: Mechanisms & Cross-validation (1) (2) (3) (4) Dep. Var. log wage log investment log dom. sales log profit Period Panel A: Regional tariff Tariff_t *** 1.08* *** (0.25) (0.59) (0.35) (0.83) firm FE Yes Yes Yes Yes year FE Yes Yes Yes Yes Observations 1,434, ,156 1,318,575 1,132,034 R-squared Panel B: Industry tariff Tariff_t *** 0.46*** 0.16* 0.63*** (0.05) (0.17) (0.10) (0.15) Firm FE Yes Yes Yes Yes year FE Yes Yes Yes Yes Observations 1,436, ,396 1,310,617 1,105,894 R-squared

25 Labor Supply Adjustments It is extensively documented in the labor literature that individual/household labor supply respond to income shocks (Gorbachev, 2015; Blundell et al.,2016). However, insufficient attention has been paid to the relationship between trade liberalization and labor supply changes (Arkolakis, 2014) We investigate how labor supply respond to trade liberalization. Extensive margin: labor participation (work or not) Intensive margin: working hours We also study the employment transition from tradable to non-tradable sector.

26 Labor Supply Adjustments Dep. Var. (1) (2) (3) (4) Working or not (Yes=1) Working at tradable sector (Yes = 1) Working at nontradable sectors log(working hours) Working people Panel A: OLS Tariff -0.42*** 0.43** -0.85*** -1.42** (0.14) (0.18) (0.21) (0.67) Observations 591, , , ,160 R-squared Panel B: 2SLS Tariff -0.59*** *** (0.18) (0.21) (0.24) (0.72) Observations 591, , , ,160 R-squared Controls in both panels Basic controls Yes Yes Yes Yes Clusters

27 Labor Supply Adjustments 1. Regional tariff cut leads to increases in regional labor supply, both at extensive margin (labor participation) and intensive margin (working hours). Consistent with the literature documenting increased labor supply to offset adverse household income shocks 2. Contraction of regional employment in the tradable sector. reallocation away from tradable sector or net worker exit in tradable sector 3. Employment expansion in the non-tradable sector. Reallocation towards non-tradable sector or net worker entry in nontradable sector

28 Labor Market Adjustments: Heterogeneity Dep. Var.: (1) (2) (3) (4) (5) (6) (7) (8) Working Working at tradable sector Working at nontradable sector Log wage Male Female Male Female Male Female Male Female Panel A: OLS *** ** ** *** ** ** -1.40*** 1.9*** 1.7*** *** 2.7* *** -1.08*** *** -1.10*** - - Panel B: 2SLS *** *** 2.4** *** 0.57* ** -0.81* 2.4*** 1.7* ** 0.89*** *** -1.22*** 1.9** 2.0** ** *** *** 5.7*** *** -0.61** *** -0.65** - -

29 Labor Market Adjustments: Heterogeneity 1. Stronger wage decline for men, but stronger labor supply increases for women. Consistent with the added worker effect in the labor literature (Gorbachev, 2015; Blundell et al.,2016). 2. More job churning (reallocation from tradable to non-tradable sector) for men, but more net entry for women. 3. Labor supply increased for the elderly/retired.

30 Income Transfer The UHS provides information of transfer income Transfer in Public transfer: income from retirement pension, disaster relief funds, regular donation and compensation, medical fees, Private transfer: supporting income between households, and income from nonusual-residing members of households Transfer out

31 Income Transfer (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Have public transfer income log(public transfer income) Have private transfer income log(private transfer income) Have any transfer out log(transfer out) Tariff 0.42** 0.39** 1.57** 1.22* (0.17) (0.16) (0.67) (0.31) (0.30) (0.84) (0.85) (0.11) (0.10) (0.92) (0.90) Basic Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Household Structure No Yes No Yes No Yes No Yes No Yes No Yes Observations 218, ,819 97,367 97, , ,819 78,493 78, , , , ,773 R-squared Clusters Transfer income from the government decreased in response to tariff reduction. Governments in developing countries lack trade-adjustment assistance programs (Shelton, 2007).

32 Parental Co-residence and Household Size The option to co-reside with the parents provides an important insurance against labor market risks (Kaplan, 2012). Faced with negative income shocks, young adults are more likely to live with their parents for expenditure sharing Did the income shocks induced by trade liberalization affect people s parental co-residence decision? Construct two variables to reflect the co-residence decision: Log household size (# family members aged above 20) Co-residence dummy (=1 if parents and adult children live together)

33 Parental Co-residence and Household Size (1) (2) (3) (4) (5) (6) Dep. Var. log(hh size) Coresidence (Yes = 1) HH head HH head HH head Sample Full sample Full sample Age<50 Age>=50 Age<50 HH head Age>=50 Panel A: OLS Tariff -0.27** ** -0.50** ** (0.13) (0.11) (0.25) (0.24) (0.18) (0.50) Observations 251, , , , , ,228 R-squared Panel B: 2SLS Tariff -0.35** -0.24* -0.50* * (0.15) (0.13) (0.29) (0.26) (0.20) (0.48) Observations 251, , , , , ,228 R-squared Controls in both panels Basic controls Yes Yes Yes Yes Yes Yes Clusters Household size increased because more young adults co-reside with parents. Effect only present for households with HH age above 50, suggesting young adults move to reside with their parents, not vice versa.

34 Effects on Household Income, Consumption, and Savings We just showed that households change behaviors to insure against income shocks induced by trade reform. Smoothing earnings: Labor supply/employment shifts Smoothing expenditure: Parental co-residence/ savings How effective are those behaviors? We investigate the effect of tariff cut on Household income per capita Household consumption per capita

35 Effects on Household Income, Consumption, and Saving (1) (2) (3) (4) (5) (6) Dep. Var. Log household income/capita Log consumption/capita Saving rate Panel A: OLS Tariff 1.17*** 1.05*** 1.03*** 0.90*** (0.36) (0.36) (0.33) (0.32) (0.12) (0.12) Observations 251, , , , , ,492 R-squared Panel B: 2SLS Tariff 1.58*** 1.42*** 1.08** 0.91* 0.31** 0.32** (0.55) (0.54) (0.51) (0.50) (0.15) (0.15) Observations 251, , , , , ,492 R-squared Controls in both panels Basic controls Yes Yes Yes Yes Yes Yes Household structure No Yes No Yes No Yes Clusters Reduction in household income and consumption per capita Wage effects > household income effects > consumption effects Decline in savings

36 Effects on Household Borrowing and Lending (1) (2) (3) (4) (5) (6) (7) (8) Borrow (Yes = 1) Log(Borrowed) Lend (Yes = 1) log(lended) Tariff (0.41) (0.41) (1.40) (1.40) (0.35) (0.35) (1.18) (1.17) Household Structure No Yes No Yes No Yes No Yes Observations 251, , , , , , , ,648 R-squared Clusters No evidence of the effects on household borrowing and lending.

37 Robustness Checks: Controlling for Other Confounding Policy Changes Non-tariff barriers Drawing on annual circulars of the Ministry of Foreign Trade and Economic Cooperation and the Ministry of Commerce We construct a city-level measure of import license control as the share of products produced in this city that are under import license control. FDI liberalization Based on FDI restriction data from Catalogue for the Guidance of Foreign Investment Industries issued by the Ministry of Commerce of China We construct city-level FDI restriction measure as the share of industries that are either prohibited or restricted in the Catalogue

38 Robustness Checks: Controlling for Other Confounding Policy Changes Export expansion Tariff uncertainty reduction resulting from the US granting permanent normal trade relations (PNTR) to China after China s WTO entry has substantially increased Chinese exports (Handely and Limao, 2017; Pierce and Schott, 2016). Following the recent literature (Facchini et al., 2017; Erten and Leight, 2017), we construct regional level tariff uncertainty measures to capture the export effects. Minimum wage adjustments We collect the minimum wage from all the cities after 1998 from City Statistical Yearbooks.

39 More Robustness Checks Alternative measures of regional tariff Effective rate of protection 2001 weights Kovak s measure Allowing tariff to vary over time Alternative sample Consistent cities 2002 afterwards only Migration issue Using the 2000 and 2005 population census, we create a dummy variable for new migrants who moved to current place after 2002 Only keep the individuals who lived in current city before 2002

40 Robustness Checks (1) (2) (3) (4) (5) (6) Dep. Var.: log(wage) working log(hh size) coresidence log(hh income/capita) log(hh consumption/capita) Panel A: Control for confounding policies Import license 1.26** -0.41*** * 0.92** 0.74** (0.55) (0.15) (0.15) (0.32) (0.41) (0.37) FDI 1.77*** -0.36** -0.27** -0.49** 1.18*** 1.04*** (0.48) (0.16) (0.13) (0.24) (0.36) (0.33) Exports 2.00*** -0.28* -0.24* -0.40* 1.27*** 1.05*** (0.49) (0.16) (0.21) (0.39) (0.36) Minimum wage 1.78*** -0.37** -0.26** -0.44** 1.21*** 1.02*** (0.47) (0.16) (0.12) (0.22) (0.36) (0.33) Panel B: Alternative regional tariff measure ERP 0.46*** -0.10** -0.06* -0.12* 0.30*** 0.28*** (0.14) (0.05) (0.04) (0.07) (0.11) (0.10) Labor-share adj. 1.36*** -0.25* -0.24** -0.41* 0.95*** 0.82*** (0.44) (0.13) (0.11) (0.23) (0.31) (0.28) 2001 weights 1.66*** -0.39** -0.29** -0.52** 1.05*** 0.92*** (0.45) (0.16) (0.12) (0.23) (0.35) (0.33) actual tariff 1.55*** -0.37** -0.24** -0.43* 1.00*** 0.79** (0.41) (0.15) (0.12) (0.24) (0.33) (0.31) Panel C: Alternative sample Consistent cities 2.94*** -0.70*** -0.39** -0.62* 1.09** 0.44 (1.06) (0.26) (0.19) (0.35) (0.54) (0.30) 2002 and years afer 2.58*** -0.39* -0.24** -0.37** 0.66** 1.06** (0.76) (0.21) (0.11) (0.18) (0.30) (0.49) Living here -0.42*** 1.77*** -0.26** -0.48** 1.19*** 1.04*** before 2002 (0.14) (0.47) (0.13) (0.25) (0.36) (0.33)

41 Pre-trends Examination & Placebo Tests Unbiased estimation of the DID framework requires that the time trends of outcome variables in regions with larger tariff cuts would be parallel with those in other regions if China had not lowered tariffs. Conduct two sets of pre-trends examination / placebo tests. 1. Plot pre-wto ( ) outcome against post-wto tariff change. We expect no significant correlation 2. Split all cities into large tariff cut regions v.s. small tariff cut regions, plot the outcome difference of the two groups in each year during

42 Placebo Tests The aforementioned effects didn t play out during the pre-wto period (i.e )

43 Placebo Tests The aforementioned effects didn t play out during the pre-wto period (i.e )

44 Pre-trends Examination

45 Conclusions We investigate the impacts of tariff reduction after the entry into WTO on local labor market outcomes and household behaviors in urban China Regions that initially specialized in industries facing larger tariff cuts experienced more significant wage declines relative to other regions. Document a series of behavioral response to the trade-induced income shocks. Labor supply changes/employment transition Parental Co-residence Savings Household serves as an important insurance against the labor market risks induced by trade liberalization.