Trade and long term unemployment: aquantitativeassessment

Transcription

1 Trade and long term unemployment: aquantitativeassessment Céline Carrère Université de Genève and CEPR Anja Grujovic Université de Genève Frédéric Robert-Nicoud Université de Genève, SERC and CEPR July 2, 2014

2 Objectives Provide a theoretical framework that can explain rich effects of trade on unemployment: Multi-country, multi-sector gravity model with trade frictions (in the style of EK model of Costinot, Donaldson and Komunjer, 2012) Labour market frictions and equilibrium unemployment (as per DMP a la Helpman and Itskhoki, 2010) Calibrate the model to explore welfare and employment variations in a set of counter-factual exercises TTIP on US and EU 2/1

3 Results 1. Trade openness has an ambiguous effect on unemployment 2. Unemployment rises if trade liberalisation results in reallocation of labour into sectors with higher-than average labour market frictions. 3. Trade gains are only weakly correlated with employment effect of trade reforms 3/1

4 Previous empirical work Effect of trade openness on employment ambiguous: Negative: Mexico (Revenga, 1994); Chile (Levinsohn, 1999); Brazil (Menezes et al., 2011) Positive: cross-section (Dutt et al, 2010); Mauritius (Milner and Wright, 1998); China, India and Malaysia (Lee, 2005) No relationship: Colombia (Goldberg and Pavnick, 2005); India (Hasan et al, 2012) Ambiguous relationship: cross-section (Carrère et al, 2014) 4/1

5 Previous theoretical work Policy-made frictions: Brecher (1974) Davis (1998) Matching frictions: Davidson, Martin and Matusz (1999) Helpman and Itshkoki (2010) Carrère, Fugazza, Olarreaga, and Robert-Nicoud (2014) Typically 1 small-open economy, or 2 countries 5/1

6 Plan of talk Intro Theoretical framework: DMP meets EK Comparative statics: main result Estimation and calibration Conclusion 6/1

7 Trade setting - EK model Environment One primary factor of production, Labor Sectors k =1,...,K, each with a great many varieties, x. An arbitrary number of countries, denoted by i and j =1,...,I. Preferences CES across varieties, and Cobb-Douglas across sectors Technology TFP: z ik (x) are random variables such that: θ z P {z ik (x) > z} =1 exp z ik s vary by origin country i and sector k θ governs the scope for comparative advantage Iceberg trade costs Delivering 1 unit of a good to country j requires shipping τ ijk 1 units from country i (with τ iik = 1) z ik 7/1

8 Labour market setting - DMP-type search and matching frictions 1. Firms open vacancies and workers choose a sector 2. Firms search for workers and workers search for jobs 3. Job matches are realized and sector-specific wages w ik are determined. 4. Employment rate in sector k is: ik = wi P i 1 λ µ ik 5. Labour mobility across sectors and risk neutral workers imply ik w ik = w i 8/1

9 Lemma: Labour market frictions matter for patterns of trade Novelty of the model: Gravity-like equation where in addition to usual trade frictions we have labour-market frictions: θ tijk E ijk = α jky j, where T jk t ijk T jk τ ijk w λ z ik µ ik i =1 i P 1 λ i 1/θ t θ i jk Relative costs now take into account labour-market matching efficiencies: more efficient matching higher market shares. 9/1

10 Plan of talk Intro Theoretical framework: DMP meets EK Comparative statics: main result Estimation and calibration Conclusion 10 / 1

11 Comparative statics: employment and trade reforms How does employment rate react to trade reforms which impact trade structure? Two effects: ˆ i =(1 λ)ˆω i + 1 µ i K Êik Êi (µ ik µ i ) E ik E i k=1 =cov(êik,µ ik) Real-wage effect: as global market share grows, firms create more vacancies, employment rate increases in each sector, and real wage goes up. Reallocation effect: labour reallocates across sectors with different levels of labour market frictions. Effect on employment rate depends on the covariance: cov Êik,µ ik < 0 = ˆ i ˆωi < 0 =0 11 / 1

12 Plan of talk Intro Theoretical framework: DMP meets EK Comparative statics: main result Estimation and calibration Conclusion 12 / 1

13 Estimation 1: matching efficiencies Estimate (time-invariant) sector-specific labour market matching efficiencies µ ik Assuming µ ik = µ i µ k : ln it =(1 λ)lnω it +lnµ i +ln which can be estimated in two steps: 1st step: ln it = β ln ω it + FE i + FE t + ν it 2nd step: e ν it = k β k E ikt E it + it k E ikt E it µ k 13 / 1

14 Estimates of sector-specific matching efficiencies High matching efficiencies ISIC Rev 3 Sector description Matching eff. 12 Mining of uranium and thorium ores Other mining and quarrying Mining of metal ores Leather, leather products and footwear Forestry, logging and related service activities Mining of coal and lignite, extraction of peat Low matching efficiencies ISIC Rev 3 Sector description Matching eff. 30 Office, accounting and computing machinery Transport, storage and communications Extraction of crude petroleum and natural gas and related services Wholesale and retail trade - restaurants and hotels Community, social and personal services Finance, insurance, real estate and business services / 1

15 Estimation 2: gravity parameters Estimate sector-specific trade frictions: Gravity equation to estimate sector-specific trade costs, where we define τ ijk = D δ ij (1 + tariff ijk). ln E ijk = θ ln(1 + tariff ijk ) θδ ln(d ik ) θ ln(κ ik ) +ln (τ i jkκ i k) θ +ln(α jk E jk ) FE i ik FE jk Calibrate the model to explore employment variations in the example of TTIP 15 / 1

16 Calibration results: FTA general equilibrium Countries ωi FTA k sfta FTA ik µ k i Unempl. rate (u ω i 2008 k s2008 µ ik k 2008 ui 2008 i FTA ui 2008 ) u i i 2008 (in %) (in % ) US USA EU NLD BEL DNK IRL GBR FRA ITA AUT DEU SVN FIN CZE POL SWE EST HUN PRT ESP GRC Other NOR OECD JPN CHE CAN ISR ISL MEX / 1

17 Calibration results: example of reallocation in USA 17 / 1

18 Calibration results: Autarky general equilibrium Countries ωi A k sa ik µ A k i Unempl. rate (u ω i 2008 k s2008 µ ik k 2008 ui 2008 i A ui 2008 ) u i i 2008 (in %) (in % ) US USA EU IRL GBR DNK NLD FRA BEL GRC ESP DEU PRT SWE HUN POL ITA FIN EST AUT SVN CZE Other ISL OECD ISR NOR CHE CAN MEX / 1

19 Plan of talk Intro Theoretical framework: DMP meets EK Comparative statics: main result Estimation and calibration Conclusion 19 / 1

20 Conclusion We developed a multi-country, multi-sectoral gravity trade model with trade frictions in the wake of EK (2002) and labour market frictions as per DMP, to provide an analytical explanation for the ambiguous effect of trade on unemployment: We find that unemployment may rise following trade liberalisation if it results in reallocation of labour into sectors with lower-than-average labour matching efficiencies, and vice versa. We show that gains from trade are not necessarily a good proxy for unemployment, as unemployment may rise even when gains from trade exist. 20 / 1