Model Data Results References

Intra-Industry Trade and North-North Migration: How TTIP Would Change the Patterns

Mario Larch1 Steffen Sirries1

1University of Bayreuth

ETSG 2014

1 / 38 Model Data Results References Funding and Notes on the Project

This project is funded by DFG within the Project “Gravity with Unemployment” part of a bigger research agenda: Heid and Larch (2012) for unemployment, Heid (2014) for informality up to now without labor market imperfections work in progress → input is very welcome!

2 / 38 Model Data Results References Why trade and migration? I

Traditional view: H-O-type trade models with migration: Trade substitutes for migration (Mundell (1957)). Signing free trade agreements lowers the pressure of migration. Expected (bilateral) correlation between trade and migration → negative! Gaston and Nelson (2013) BUT...

3 / 38 Model Data Results References Why trade and migration? II 15 10 5 0 TRADEFLOWS in logs (normalized) −5 −5 0 5 10 15 20 MIGRATION STOCKS in logs (normalized)

4 / 38 Model Data Results References Why trade and migration? III 2 0 −2 Residuals of trade gravity −4 −5 0 5 Residuals of migration gravity

corr(ˆetrade , eˆmigr ) = 0.2304 5 / 38 Model Data Results References Why trade and migration? IV

There might be more than just a spurious correlation between trade and migration. New trade models with intra-industry trade and north-north migration → complementarity of trade and migration (e.g OECD). This is not new (Head and Ries (1998)): Migrants also bring along preferential access to the markets of the country of origin. We think simpler: Migrants are not just a factor but consumers and though they bring their demand! Migrants have idiosyncratic shocks which drive the migration decision!

6 / 38 Model Data Results References Why “A Quantitative Framework”?I

One of the core issues in empirical international trade is the quantification of the welfare gains from trade liberalization. Workhorse model: gravity equation for trade flows in new quantitative trade models (Eaton and Kortum (2002) (Econometrica), Anderson and van Wincoop (2003) (AER), Santos Silva and Tenreyro (2006) (RevStat), Anderson and Yotov (2010) (AER), Waugh (2010) (AER), Fieler (2011) (Econometrica), Arkolakis, Costinot, and Rodr´ıguez-Clare (2012) (AER)). General equilibrium effects are natural to trade shocks, as are third country effects → need for structural estimation. All frameworks in the literature so far assume labor to be immobile (to the best of our knowledge).

7 / 38 Model Data Results References Why “A Quantitative Framework”?II

To the best of our knowledge, there is no structurally estimable model with trade AND migration. (Redding (2012), di Giovanni, Levchenko, and Ortega (2012), Behrens, Mion, Murata, and S¨udekum(2011)). The interdependence might be interesting especially for (ex-ante) welfare analysis of migration- and trade-policy evaluations.

“The exact relationship between trade and migration is potentially important. For example, if the relationship were a complementary one, and if policy makers were to view further reductions in trade barriers as improbable, then in order to facilitate trade, liberalizing migration might make economic sense.” (Gaston and Nelson (2013))

8 / 38 Model Data Results References What we do! I

Research Question How does welfare analysis of trade liberalization change when we allow for labor to be mobile? Put differently: 1 What happens when we add a recent migration model (and a production side) to Anderson and van Wincoop (2003)? 2 For policy makers: Should we also negotiate for labor market integration within FTAs?

9 / 38 Model Data Results References What we do! II

We develop a multi-country, GE structurally estimable trade model which explicitly accounts for migration. Therefore we use one of the most simple new trade frameworks for intra-industry trade. We add a very recent and simple modeling of the migration decision. So, we propose a framework which enables us to disentangle trade and migration cost reductions in counterfactual analysis (examples: TTIP, EU single market, ...).

10 / 38 Model Data Results References What we don’t do!

... for the moment ... unemployment, heterogeneous migrants, assimilation, return migration (dynamics), refugee migration, remittances, heterogeneous firms or sectors, ...

11 / 38 Model Data Results References What we find! I

From theory (link between trade and migration): real wage differences influence migration decision → increased immigration increases size (respectively nominal GDP) of a country → changes nominal wages and price indices → equilibrium changes

12 / 38 Model Data Results References What we find! II

From empirics: Analysis of a hypothetical Transatlantic Trade and Investment Partnership; once with migration and once without. Take home messages: Signing TTIP would have positive welfare effects for signers. Accounting for migration for such an evaluation matters a lot in terms of welfare!

13 / 38 Model Data Results References Trade I

We start with a standard trade model: Multi-country perfect competition model (Armington (1969)) with differentiated goods at the country-level and simplest rationale for trade iceberg type-trade costs.

14 / 38 Model Data Results References Trade II

Representative consumer in country j. The quantity of purchased goods from country i is given by cij , leading to the following utility function:

σ " N # σ−1 1−σ σ−1 X σ σ Uj = βi cij . (1) i=1

With iceberg trade costs tij > 1, profit maximization implies that pij = pi tij , where pi is the factory gate price of the good in country i.

15 / 38 Model Data Results References Trade III

The representative consumer maximizes Equation (1) subject to the PN budget constraint Yj = i=1 pi tij cij . The value of aggregate sales of goods from country i to country j can then be expressed as

 1−σ βi pi tij Xij = pi tij cij = Yj , (2) Pj

and Pj is the standard CES price index. In general equilibrium, total sales correspond to nominal income, i.e., PN Yi = j=1 Xij . Assuming labor to be the only factor of production and full employment, GDP is given by total factor income, i.e., Yi = wi Li . ⇒ nothing new here!

16 / 38 Model Data Results References Migration I

Migration is a choice from a menu of locations with individual/idiosyncratic utility from migration. bilateral migration costs are equal to all migrants from i to j and we end up with McFadden (1974) like discrete choice equation. ⇒ comparable to Anderson (2011)

17 / 38 Model Data Results References Migration II

So we start with the log utility from migration from i to j for a worker h

∗ ∗ Vij = ln(Ui ) − ln(Uj ) + εijh − ln(δij ), (3)

∗ ∗ where Uj and Ui are the indirect utility functions taken from the trade system, εjih captures idiosyncratic, individual specific utility from migration and δji are migration costs. The migration flow from i to j is given by

Mij = G(Vij )Ni , (4)

where Ni is the number of natives in i and G(Vij ) gives the proportion of migrants from i to j. When εijh is assumed to be i.i.d. this proportion is given by the probability as in McFadden (1974)

exp(Vij ) G(Vij ) = P , (5) k exp Vik

18 / 38 Model Data Results References Migration III

w since U∗ = j and U∗ = wi for one worker, we get j Pj i Pi

wi wj Vij = ln( ) − ln( ) + εijh − ln(δij ). (6) Pi Pj

Inserting (6) in (4) gives

wj Pi 1 wj Pj δij Mij = Ni (7) P ( wk Pi 1 ) k wi Pk δij wj 1 Pj δij = Ni . (8) P ( wk 1 ) k Pk δik

19 / 38 Model Data Results References Linking Trade and Migration

The simple channel is: migration is driven by real wage differences → migration changes demand/size of a country → changes wages (nominal) and price indices → again changes real wage differences

How we go on now: 1 numerical example 2 estimate structural parameters (estimation stage) 3 plug parameters and observed variables into the model, solve and extract results 4 re-set the trade and/or migration cost vectors (counterfactual) 5 do 3 again and compare the results

20 / 38 Model Data Results References Linking Trade and Migration (numerically) I

3 countries (A, B, C), same size, same GDP, σ = 5, no trade costs, no migration costs

(1−σ) (1−σ) Country Yi Ni tij δij wi Li Xii Mii βi Pi wi /Pi A 100 100 1 1 1 100 33,33 33,33 1 0,76 1,3158 B 100 100 1 1 1 100 33,33 33,33 1 0,76 1,3158 C 100 100 1 1 1 100 33,33 33,33 1 0,76 1,3158

21 / 38 Model Data Results References Linking Trade and Migration (numerically) II

Introducing trade costs: (1−σ) (1−σ) (1−σ) (1−σ) tAB = tBA = 0, 5; tBC = tCB = 0, 5; (1−σ) (1−σ) tAC = tCA = 0, 4. → B is less remote. No migration at all (infinite migration costs)

(1−σ) (1−σ) Country Yi Ni tij δij wi Li Xii Mii βi Pi wi /Pi A 100 100 0,4 0 1 100 53,106 100 fix 0,758 1,3193 B 100,67 100 0,5 0 1,0067 100 49,563 100 fix 0,75 1,3423 C 100 100 0,4 0 1 100 53,106 100 fix 0,758 1,3193

22 / 38 Model Data Results References Linking Trade and Migration (numerically) III

Allowing for free migration: trade costs as before (1−σ) free migration δij = 1

(1−σ) (1−σ) Country Yi Ni tij δij wi Li Xii Mii βi Pi wi /Pi A 100 100 0,4 1 1,006 99,425 52,948 33,142 fix 0,762 1,3202 B 102,32 100 0,5 1 1,012 101,15 50,573 33,717 fix 0,753 1,344 C 100 100 0,4 1 1,006 99,425 52,948 33,142 fix 0,762 1,3202

23 / 38 Model Data Results References Data

We just keep the OECD sample of two bilateral data sets CEPII bilateral trade data from Head, Mayer, and Ries (2010) with many geographical information. “Where on Earth is everybody”-data from Ozden,¨ Parsons, Schiff, and Walmsley (2011). International bilateral migration stock data (censuses and official registers).

24 / 38 Model Data Results References Estimation stage and fit of the model I

We estimate a standard trade gravity with importer and exporter fixed effects (Head, Mayer, and Ries (2010) JIE) and a standard migration gravity with importer and exporter fixed effects (Grogger and Hanson (2011) JDE). Both gravities are estimated via PPML. With the estimated structural parameters for the trade and migration costs and observed values for N and GDP we go into the model.

25 / 38 Model Data Results References Estimation stage and fit of the model II

We estimate the following gravity model:

xij  1−σ 1−σ  zij ≡ = exp k − (1 − σ) ln tij − ln Πi − ln Pj + εij , (9) yi yj

where zij are normalized trade flows. Trade costs are specified as

1−σ tij = exp(β1PTAij + β2 ln DISTij + β3CONTIGij + β4LANGij ), (10)

and control for Πi and Pj using importer and exporter fixed-effects.

26 / 38 Model Data Results References Estimation stage and fit of the model III

We estimate the following gravity model:

xij  1−σ 1−σ  mij ≡ = exp k − (1 − σ) ln δij − ln Ωi − ln Wj + εij , (11) yi yj

where mij are normalized migration flows. Migration costs are specified as

1−σ δij = exp(γ1PTAij + γ2 ln DISTij + γ3CONTIGij + γ4LANGij ), (12)

and control for Ωi and Wj using importer and exporter fixed-effects as we control for the multilateral resistance terms in the trade gravity.

27 / 38 Model Data Results References Estimation stage and fit of the model IV 15 10 5 tradeflow in logs model 0 −5 0 5 10 15 tradeflow in logs observed

28 / 38 Model Data Results References Estimation stage and fit of the model V 20 15 10 migration in logs model 5 0 0 5 10 15 migration in logs observed

29 / 38 Model Data Results References Counterfactual Analysis I

What would happen if TTIP would be signed? That means, we will have a PTA dummy equal to one between the USA and the EU in addition to existing dummies equal to one in the PTA vector. First step: We look at the inception of TTIP at all, i.e. we change the PTA vector in the trade and migration cost estimation. Second step: We do the same in a framework without migration, i.e. Anderson and van Wincoop (2003). Other, more differentiated counterfactuals possible and of highest interest: EU single market, other reduction of migration cost, ...

30 / 38 Model Data Results References Counterfactual Analysis II

Table: Worldwide changes from inception of TTIP with migration

world GDP change world trade change world migration change .0056011 .096451 .0464708

31 / 38 Model Data Results References Counterfactual Analysis IV

Table: Selected results of welfare changes by country

Country rel. welfare change Model without migration abs. difference Percent difference MEX -.0022877 -.0012785 -.0010092 .7894172 TUR -.0011883 -.0029055 .0017172 -.5910235 CHL -.0011581 -.0010562 -.0001019 .0965121 ...... JPN -.000161 -.0004006 .0002396 -.5980565 ...... DEU .0078152 .0033957 .0044195 1.301525 ...... USA .0103697 .0046371 .0057326 1.236233 ...... AUT .0138356 .0077528 .0060827 .7845843 ...... PRT .0208332 .0129 .0079331 .6149693 SVK .0208737 .0123877 .008486 .6850356 EST .0593799 .0446894 .0146905 .3287251 averages .0108304 .0060021 .0048283 .3080928 32 / 38 Model Data Results References Conclusion

Quantitative framework for international trade and migration. Demand effect of migrants contributes to welfare. Signing TTIP would have positive effects for signers. Migration matters for a welfare analysis of trade cost reductions.

33 / 38 Model Data Results References

Thank’s a lot for your invitation, your attention, and your comments!

34 / 38 Model Data Results References ReferencesI

Anderson, J. E. (2011): “The Gravity Model,” Annual Review of Economics, 3, 133–160. Anderson, J. E., and E. van Wincoop (2003): “Gravity with Gravitas: A Solution to the Border Puzzle,” American Economic Review, 93(1), 170–192. Anderson, J. E., and Y. V. Yotov (2010): “The Changing Incidence of Geography,” American Economic Review, 100(5), 2157–86. Arkolakis, C., A. Costinot, and A. Rodr´ıguez-Clare (2012): “New Trade Models, Same Old Gains?,” American Economic Review, 102(1), 94–130. Armington, P. S. (1969): “A Theory of Demand for Products Distinguished by Place of Production,” Staff Papers - International Monetary Fund, 16(1), 159–178. Behrens, K., G. Mion, Y. Murata, and J. Sudekum¨ (2011): “Spatial frictions,” CEPR Discussion Papers. di Giovanni, J., A. Levchenko, and F. Ortega (2012): “A Global View of Cross-Border Migration,” .

35 / 38 Model Data Results References ReferencesII

Eaton, J., and S. Kortum (2002): “TECHNOLOGY, GEOGRAPHY, AND TRADE,” Econometrica, 70(5), 1741–1779. Fieler, A. C. (2011): “Nonhomotheticity and Bilateral Trade: Evidence and a Quantitative Explanation,” Econometrica, 79(4), 1069–1101. Gaston, N., and D. R. Nelson (2013): “Bridging Trade Theory and Labour Econometrics: the Effects of International Migration,” Journal of Economic Surveys, 27(1), 98–139. Grogger, J., and G. H. Hanson (2011): “Income maximization and the selection and sorting of international migrants,” Journal of Development Economics, 95(1), 42–57. Head, K., T. Mayer, and J. Ries (2010): “The erosion of colonial trade linkages after independence,” Journal of International Economics, 81(1), 1 – 14. Head, K., and J. Ries (1998): “Immigration and Trade Creation: Econometric Evidence from Canada,” Canadian Journal of Economics, 31(1), 47 – 62.

36 / 38 Model Data Results References ReferencesIII

Heid, B. S. (2014): “Essays on International Trade and Development,” . Heid, B. S., and M. Larch (2012): “International Trade and Unemployment: A Quantitative Framework,” . McFadden, D. L. (1974): “Conditional Logit Analysis of Qualitative Choice Behavior,” Frontiers in Econometrics, pp. 105–143. Mundell, R. A. (1957): “International Trade and Factor Mobility,” American Economic Review, 47(3), 321–335. ΩOzden,¨ Parsons, Schiff, and Walmsley Ozden,¨ c., C. R. Parsons, M. Schiff, and T. L. Walmsley (2011): “Where on Earth is Everybody? The Evolution of Global Bilateral Migration 1960-2000,” World Bank Economic Review, 25(1), 12 – 56. Redding, S. J. (2012): “Goods Trade, Factor Mobility and Welfare,” . Santos Silva, J. M. C., and S. Tenreyro (2006): “The Log of Gravity,” The Review of Economics and Statistics, 88(4), 641–658. Waugh, M. E. (2010): “International Trade and Income Differences,” American Economic Review, 100(December), 2093–2124.

37 / 38 Model Data Results References

Intra-Industry Trade and North-North Migration: How TTIP Would Change the Patterns

Mario Larch1 Steffen Sirries1

1University of Bayreuth

ETSG 2014

38 / 38