Globalization and Unemployment

Transcription

1 Globalization and Unemployment Priya Ranjan University of California, Irvine 27th June, 2013 Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

2 unemployment one of the big economic problems, trade economists have generally tended to abstract away from it most trade models are full employment models with fully exible wages outside economics profession, people believe that one of the important e ects of trade is the destruction of jobs, leading to signi cant unemployment. such reports are common in the various popular forms of the news media which completely ignore the creation of new jobs as a result of international trade a search of news articles in the New York Times since 1990, revealed a total of 275 articles on NAFTA as the primary subject. Out of this, 147 articles talked about job destruction in the US as a consequence of NAFTA. Most of the statements made in the House and the Senate during the NAFTA debate were about NAFTA s impact on jobs. source: Matusz and Davidson (Book, 2004) in sharp contrast, there is not a listing for unemployment in the index of the 4000 pages long Handbook of International Economics, primarily used by academic economists Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

3 Recent quotes on Globalization and Jobs But o shoring from West to East has also contributed to job losses in rich countries, especially for the less skilled, yet increasingly for the middle classes too... In a survey by NBC News and the Wall Street Journal in 2010, 86% of Americans polled said that o shoring of jobs by local rms to low-wage locations was a leading cause of their country s economic problems." Barack Obama s presidential campaign last year repeatedly claimed that his rival, Mitt Romney, had sent thousands of jobs overseas when he was working in private equity. Mr Romney, in turn, attacked Chrysler, a car rm, for planning to make Jeeps in China. France s new Socialist government has appointed a minister, Arnaud Montebourg, to resist delocalisation. Germany s chancellor, Angela Merkel, worries publicly about whether the country will still make cars in 20 years time." Source: The Economist (Jan 19th, 2013) Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

4 Key Questions How does globalization (trade and o shoring) a ect unemployment? What are the gains from trade in the presence of labor market frictions? Which labor market policies / institutions are optimal in the face of globalization? Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

5 Plan of the talk Discuss theoretical literature on globalization and unemployment link between trade in goods and unemployment o shoring of inputs/tasks and unemployment Policy Implications Discuss empirical research on globalization and employment/unemployment cross-country studies intra-country studies inter-state studies inter-industry studies; intra-industry studies local labor market impacts Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

6 Why search and matching models of unemployment key paradigm in macro-labor (2010 Nobel to DMP) Mechanisms empirically identi ed Good ex post model t Can be combined with various trade models Can be combined with various wage setting mechanisms Minimum wages Individual/collective bargaining Wage posting Fair/e ciency wages Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

7 A simple model of trade and unemployment Product Market single non-tradable nal good Z produced using two potentially tradable inputs X and Y Z = AX 1 α Y α α α (1 α) 1 α input prices p x and p y ; unit cost of producing Z : c(p x, p y ) = 1 A (p x ) 1 α (p y ) α Z is chosen as the numeraire ) p z = 1, and therefore, c(p x, p y ) = 1 A (p x ) 1 α (p y ) α = 1 relative demand for the two inputs a function of the relative price : X d Y d = (1 αp x α)p y Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

8 Product Market X uses only labor while Y uses both capital and labor production conducted in one worker rms with the following output x = 1; y = hk φ ; 0 < φ < 1 aggregate production in each sector given by X = (1 u x )L x ; Y = h(1 u y )L y k φ L i : total number of workers a liated with sector-i u i : unemployment rate in sector i aggregate labor constraint: L x + L y = L capital market clearing condition: K = k(1 u y )L y relative supply of the two intermediate goods is X S Y S = (1 u x )L x h(1 u y )L y k φ Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

9 Labor Market (static version of Pissarides model) θ i : vacancy rate (number of vacancies divided by labor force, L i ) θ i also market tightness (vacancy/searchers) constant-returns-to-scale matching function M i (θ i L i, L i ) = m i (θ i L i ) 1 µ L µ i = m i θ 1 µ i L i ; 0 < µ < 1 M i L i = m i θ 1 µ i : job nding rate for a worker in sector-i M i θ i L i = m i θ µ i : rate at which vacant jobs are lled in sector-i unemployment rate in each sector is given by: u i = 1 m i θ 1 µ i c i : cost of posting a vacancy in sector-i (in terms of numeraire) results depend on the modeling of vacancy cost b : unemployment bene t Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

10 Wage Bargaining/Free Entry wage determined by Nash bargaining w i = arg max (w i b) β 1 J i β w i J i : value of a job for a rm in sector-i J x = p x w x ; J y = p y hk φ rk w y Nash bargaining implies the following wage bargaining equations w x = (1 β) b + βp x ; w y = (1 β) b + β p y hk φ rk free entry implies zero pro t from vacancy creation p x w x = c x θ µ x m x ; p y hk φ rk w y = c y θ µ y m y above equations known as job creation condition in the literature Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

11 Free mobility of workers (no arbitrage) zero pro t and wage bargaining equations imply w x = b + β c x θ x µ 1 β m x w y = b + β c y θ y µ 1 β m y a worker should be indi erent between searching in either sector expected returns from searching in either sector must be equalized m x θ 1 µ x w x + (1 m x θ 1 µ x )b = m y θ 1 µ y w y + (1 m y θ 1 µ y )b above yields the no arbitrage condition c x θ x = c y θ y Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

12 Autarky Equilibrium start with a p x p y ; 1 A (p x ) 1 α (p y ) α = 1 determines p x, p y wage bargaining equation w x = (1 β) b + βp x determines w x ; job creation condition, p x w x = c x θ x µ m x, determines θ x no arbitrage condition c x θ x = c y θ y determines θ y wage bargaining equation, job creation condition, and the condition for optimal choice of capital, φp y hk φ 1 = r determine w y, k, and r L y is determined from market clearing condition for capital: L y = K k(1 u y ) relative supply determined for each p x p y easy to verify that relative supply upward sloping intersection with downward sloping relative demand determines autarky equilibrium Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

13 Comparative Statics in Autarky Equilibrium Result 1: An increase in the relative price of the labor intensive good increases the real wage of workers, reduces their rate of unemployment, and reduces the real reward of capital. Stolper-Samuelson type result Intuition? Result 2: If countries have identical technologies and identical labor market parameters, then the capital abundant country has a comparative advantage in the capital intensive good and the labor abundant country has a comparative advantage in the labor intensive good. Heckscher-Ohlin theorem Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

14 International Trade capital abundant country comparative advantage in Y ) # p x p y ) from result 1 " u x and " u y impact on aggregate unemployment: u = L x L u x + L y L u y # L x and " L y increases) impact on aggregate unemployment ambiguous Key Result: trade a ects unemployment through two channels: changes in the sectoral unemployment rates ( u x and u y ) and intersectoral reallocation of labor ( L x and L y ). Caveat: result above holds in a diversi ed equilibrium if both sectors are active, no arbitrage condition ensures sectoral unemployment rates move in the same direction if the economy becomes completely specialized in the export sector, then unemployment in the export sector decreases compared to autarky Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

15 Extensions pure re-allocation e ect of trade on unemployment vacancy costs for both sectors and the unemployment bene ts in terms of good X Now the wage bargaining and job creation equations for sector X are w x = (1 β)bp x + βp x p x w x = p x c x θ µ x m x θ x becomes independent of p x so does θ y by no arbitrage condition trade does not a ect sectoral unemployment rates trade still a ects aggregate unemployment through sectoral re-allocation of workers Helpman and Itskhoki (ReStud, 2010) model has this feature Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

16 Extensions vacancy costs for both sectors and the unemployment bene ts in terms of good Y same qualitative results as when vacancy cost in terms of numeraire suppose c x = c y ; m x = m y since sectoral unemployment rates identical, reallocation of resources induced by trade has no a ect on unemployment Key insight: trade a ects aggregate unemployment through both sectoral unemployment changes and re-allocation only in special cases does it a ect unemployment through only one channel Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

17 Imperfect mobility of labor simplify the production side by assuming φ = 0 labor is the only factor of production imperfect intersectoral mobility of workers as in Mitra and Ranjan (JIE, 2010) workers have idiosyncratic preferences for working in a particular sector per-period utility (or disutility) to individual-j of ε j i from being part of the labor force in sector-i in addition to utility derived from consumption de ne ϕ j ε j x ε j y ; G (ϕ) distribution function If ϕ j > 0, then ϕ j is the cost to worker j of moving from sector X to sector Y if ϕ j < 0, it is costly for worker j to move from sector Y to sector X ϕ j = 0, 8j, captures perfect mobility Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

18 with Imperfect Mobility of Labor equilibrium sectoral choice is determined by a cuto value of ϕ denoted by bϕ where bϕ(θ x, θ y ) = β 1 β (c y θ y c x θ x ) above replaces the no arbitrage condition If ϕ j bϕ(θ x, θ y ) then search in sector-x If ϕ j < bϕ(θ x, θ y ) then search in sector-y a fraction 1 G (bϕ) of workers are a liated with sector X, while the remaining fraction G (bϕ) are a liated with sector Y L x = (1 G (bϕ))l; L y = G (bϕ)l easy to verify relative supply of X increasing in relative price of X due to increased attractiveness of X sector labor market not everyone moves due to costly mobility Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

19 Imperfect mobility and trade suppose the country has a comparative advantage in X trade liberalization )" p x p y )" p x, # p y )" θ x, # θ y )# bϕ )" L x = (1 G (bϕ))l, # L y = G (bϕ)l key result: with imperfect mobility trade leads to opposite movements in sectoral unemployment rates in HO model with perfect mobility sectoral unemployment rates moved together impact on aggregate unemployment ambiguous since sectoral unemployment rates move in opposite directions greater degree of intersectoral mobility implies greater inter-sectoral re-allocation more workers move to sector where unemployment declines with perfect labor mobility all workers move to the sector where unemployment declines (pure Ricardian case) special case: c x = c y and m x = m y )sectoral unemployment rates identical)trade leads to an unambiguous decrease in aggregate unemployment Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

20 Other models of trade and unemployment Davidson and Matusz (Book, 2004) main focus on the role of e ciency in job search, rate of job destruction, and rate of job turnover in determination of comparative advantage Moore and Ranjan (EJ, 2005) 2 sector, 2 factor model with unemployment for both skilled and unskilled workers trade liberalization reduces unemployment of skilled and increases unemployment of unskilled in skill abundant countries; opposite in skill-scarce countries skill-biased technological change can potentially reduce the unemployment of both Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

21 Trade and unemployment in Armington Model Felbermayr, Larch & Lechthaler (2013, AEJ Policy) two countries: fh, F g, specialized in speci c, imperfectly substitutable input good iceberg trade costs, τ nal, non-tradeable good produced using two inputs with a CES production function P H = P (p H, τp F ), for consumption and investment labor only factor of production matching function same as described earlier wage determined by Nash bargaining vacancy cost C j = p ζ j P j 1 ζ cj, ζ 2 (0, 1), j 2 fh, F g key result: lower τ decreases unemployment in both H and F intuition: lower τ ) lower P j )lower C j )lower p j C j creation more pro table )vacancy similar to the pure Ricardian case described earlier Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

22 Trade and unemployment in Melitz model Felbermayr et al. (2011, JET ) an increase in average productivity can be analyzed in a similar fashion as a decrease in P j in Felbermayr et al. (2013, AEJ Policy) in response to a # in τ trade liberalization pushes unemployment # and wages " intuition: rms intensify their search e ort because their productivity increases in comparison to the recruitment costs Helpman-Itskhoki (2010, ReStud) 2 sector model with one sector producing di erentiated goods a la Melitz (ETRICA, 2003) # in τ doesn t a ect sectoral employment because vacancy cost for both sectors in terms of homogeneous good # in τ a ects aggregate unemployment due to composition e ect Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

23 O shoring o shoring: sourcing of inputs (goods and services) from foreign countries which enables the fragmentation of production process includes the procurement of inputs both from a foreign a liate and a non-a liate quantitative importance of o shoring: index of outsourcing abroad of goods and services in year 2000: 81% in Belgium, 69% in Netherlands, 61% in both Denmark and Sweden, 43% in Finland (OECD (2007) 39 percent of the top 500 European rms had engaged in o shoring of services (UNCTAD 2004, 153) Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

24 A simple Model of o shoring and Unemployment Production Function L: Domestic Labor M: O shored input Z = A(L σ σ 1 + M σ σ 1 ) σγ σ 1 ; 0 < γ < 1 σ : elasticity of substitution between domestic labor and foreign produced inputs γ captures the diminishing returns which makes rm size determinate Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

25 Labor market either random matching with wage bargaining as described earlier or competitive search a la Moen (JPE, 1997) rms post vacancies and wages to attract workers higher wages attract more applicants (longer queue)=) vacancies lled faster impact of o shoring on unemployment and wages similar across these alternative speci cations policy implications di erent because no search externality in the competitive search framework Hosios condition for e cient search always satis ed use competitive search in the analysis below assume all rms have identical matching functions Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

26 Government Budget Constraint b : unemployment bene ts t : payroll tax on employed workers T : tax on pro ts L : aggregate amount of labor u : unemployment rate balanced budget constraint: t(1 u)l + T = ulb Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

27 Worker s Utility workers risk averse; creates a role for social protection utility function is given by U(x); U 0 > 0, U 00 < 0 x = w t when they are employed; x = b + z when unemployed, where z is the value of leisure/home production. expected utility of a worker searching for a job in sub-market-i/ rm-i: mθ 1 µ i U(w i t) + 1 mθ 1 µ i U(b + z) W : highest utility that a worker can expect from a job at another rm/sub-market a worker applies for a job at rm-i (sub-market-i) if mθ 1 µ i U(w i t) + 1 mθ 1 µ i above satis ed with equality in equilibrium U(b + z) W Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

28 O shoring Equilibrium rms maximize pro ts given by n Π = Max A(L σ 1 L,M,v,w subject to the constraint σ + M σ σ 1 o σγ ) σ 1 wl cv φm mθ 1 µ U(w t) + 1 mθ 1 µ U(b + z) W L = vmθ µ (number of vacancy times the probability of vacancy being lled) c : vacancy cost φ : cost of o shored input labor market equilibrium condition L = L(1 u) = Lmθ 1 µ Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

29 O shoring Equilibrium without intervention Suppose b = t = T = 0 impact of a decrease in φ (the cost of o shoring) Result 1: When σ > 1 1 γ, o shoring increases unemployment and reduces wages, thereby, reducing the welfare of workers. When σ < 1 1 γ, the opposite happens. Irrespective of the value of σ, pro ts always increase. Therefore, o shoring is Pareto improving when σ < 1 1 γ, but not so when σ > 1 1 γ. reconciles di ering results in the literature (Grossman, Rossi-Hansberg (AER, 2008), Mitra and Ranjan (JIE, 2010), Keuschnigg and Ribi (JIE,2009)) intuition: two opposing e ects of o shoring: productivity e ect and job relocation e ect Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

30 o shoring and welfare in the absence of govt. intervention assuming that rms (or the recipients of pro ts) are risk neutral, aggregate welfare given by SW Π + LW Result 2: When σ > 1 1 γ, a decrease in the o shoring cost not only increases pro ts and reduces the welfare of workers, but can reduce the aggregate social welfare as well. This possibility is more likely the more risk averse are the workers. Show using Numerical Simulation Policy Implication: in the absence of social protection, there may be a case for restricting o shoring Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

31 Figure 1: Offshoring and Welfare without social protection Figure 1a: offshoringand profits Profits offshoring cost ρ= 3 Figure 1d: offshoringand profits Profits offshoring cost ρ= 1.5

32 Impact of o shoring on the role of government Result 3: When σ < 1 1 γ, o shoring leads to a reduction in unemployment and hence allows for a reduction in the payroll taxes to provide a given level of unemployment bene ts. When σ > 1 1 γ, however, o shoring requires an increase in the payroll taxes to meet a given level of unemployment bene ts. Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

33 Potential Pareto Improvement Pareto improving policy intervention when σ > 1 1 γ Result 4: When unemployment bene ts are sub-optimal from the point of view of workers welfare, it is possible to increase unemployment bene ts to o set the adverse impact of o shoring on workers without reducing pro ts compared to the pro ts before the o shoring shock. Result 5: O shoring can be Pareto improving if the government reduces payroll taxes and increases the taxes on pro ts by enough to keep the domestic unemployment and net of tax wages unchanged. Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

34 endogenous job destruction Suppose α 2 [0, 1] random productivity of a worker, revealed after the worker is hired rms have a cuto rule: keep workers with productivity above λ and re others average productivity of retained workers is 1 1 R 1 (1+λ) λ λ αdα = 2 pay a wage of w to retained workers incur ring cost of f for each red worker f w : severance payment f t : administrative cost of ring rm s maximization problem given by ( σ λ σ Max A( L + M σ 1 σγ σ ) σ 1 wl L,M,w,λ 2 subject to the constraint (1 λ)u(w) + λu(f w + z) W ) λ (f w + f t ) (1 λ) L φm Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

35 O shoring with endogenous job destruction Result 1: When σ > 1 1 γ, o shoring increases unemployment and reduces wages, thereby reducing the welfare of workers. When σ < 1 1 γ, the opposite happens. Irrespective of the value of σ, pro ts always increase. Therefore, o shoring is Pareto improving when σ < 1 1 γ, but not so when σ > 1 1 γ. O shoring increases (decreases) unemployment by increasing (decreasing) the ring of workers when σ > 1 1 γ (σ < 1 1 γ ) Result 2: When σ > 1 1 γ, a decrease in the o shoring cost not only increases pro ts and reduces the welfare of workers, but can reduce the aggregate social welfare as well. This possibility is more likely the more risk averse are the workers. Result 3: When severance payments are sub-optimal from the point of view of workers welfare, it is possible to increase severance payments to o set the adverse impact of o shoring on workers without reducing pro ts compared to the pro ts before the o shoring shock. Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

36 O shoring and the role of labor market institutions Ranjan, JIE 2013 wages determined through collective/individual bargaining in a model with search frictions o shored input and domestic labor good substitutes (σ > 1 1 γ case) if wages bargained individually, then o shoring increases unemployment key result with collective bargaining relationship between o shoring and unemployment becomes non-monotonic starting from a high cost of o shoring, a decrease in the cost of o shoring reduces unemployment rst and then increases it intuition? calibration exercise for Sweden predicts that a decrease in the cost of o shoring, starting from the present level, would reduce unemployment Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

37 Groizard, Ranjan, and Rodriguez-Lopez (working paper, 2013) rms use a continuum of inputs to assemble goods inputs combined using a constant elasticity of substitution (CES) production function (generalization of Grossman and Rossi-Hansberg) heterogeneous o shoring cost à la Grossman and Rossi-Hansberg rm heterogeneity and free entry-exit à la Melitz 2 country extension with rms exporting Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

38 Groizard, Ranjan, and Rodriguez-Lopez (2013) relationship between o shoring and employment complex several channels of in uence at both intensive and extensive margins some positive some negative elasticity of demand for products and elasticity of substitution between inputs play a crucial role higher elasticity of demand and lower elasticity of substitution create a positive productivity e ect of o shoring on jobs rm level employment e ect of o shoring related to rm productivity important to recognize the link between o shoring and exporting o shoring driven exporting a potential source of job creation o shoring more likely to lead to net job creation with exporting possibilities Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

39 Dutt, Mitra and Ranjan (JIE, 2009) cross-sectional estimating equation unemployment i = α 0 + α 1 TR i + X i β + ɛ i TR i : extent of trade restrictions in country i X i controls: labor laws, macroeconomic distortions, and country size test using cross-country data averaged over use six di erent measures of trade policy Results: positive relationship between unemployment and measures of trade restrictions variation in protection or openness can account for 4-24% of the cross-country variation in unemployment rates 1% point increase in the average tari rate increases unemployment by about 0.35% points Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

40 Dutt, Mitra and Ranjan (JIE, 2009) dynamic speci cation (to capture short run e ects) u it = a 1 u it 1 + b 0 L it + b 1 L it b 4 L it 4 + X it B+ (η i + υ it ) L it k liberalization dummy in country i at time (t k) ; X it vector of controls Result using panel data from : positive and signi cant coe cient on contemporaneous liberalization dummy rise in unemployment in the year of liberalization trade liberalization leads to short-term spike in unemployment rates (" u about 0.6% points) negative and signi cant coe cients at lags 1 and 2 over 2-3 years, employment recovers, and the rise in unemployment reverses Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

41 Dutt, Mitra and Ranjan (JIE, 2009) Heckscher-Ohlin Speci cation u i = γ 0 + γ 1 TR i + γ 2 TR i (K /L) i + γ 3 (K /L) i + X i δ + ν i Result TR i : trade restrictions in country i, (K /L) i : capital-labor ratio for the year 1990 U i = γ (TR ) i 1 + γ 2 (K /L) i prediction of the H-O model γ 1 > 0 and γ 2 < 0 such that γ 1 + γ 2 (K /L) i? 0 as (K /L) i 7 (K /L) where (K /L) = γ 1 /γ 2 is the turning point capital-labor ratio determined endogenously from the data little support for the H-O proposition (γ 2 insigni cant in most cases) conjecture: H-O e ects get dominated by Ricardian-type productivity-related e ects of trade Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

42 Felbermayr, Pratt and Schmerer (EER, 2011) use following speci cation: u it = S s=1 ρ s u it s + θ 1 RealOpenness it + X it δ + υ i + υ t + ε it Real openness: (X + M)/GDP at PPP exchange rate controls: labor market institutions, product market regulation (PMR), population, output gap Result: θ 1 ranges from 0.09 to 0.05 (highly signi cant).09 to.05 percentage point decrease in unemployment from a 1 percentage point change in real openness Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

43 Felbermayr, Pratt and Schmerer (EER, 2011) Heckscher-Ohlin speci cation separately look at the unemployment of high skilled and low skilled Result signi cant for high skilled unemployment: u high = 0.2 Openness 0.34 L low L high Openness. L low L high high-skill unemployment decreases with openness in high-skill abundant countries but increases with openness in high-skill scarce countries consistent with the predictions Moore and Ranjan (EJ, 2005) trade reduces unemployment of skilled labor in a skill-abundant country Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

44 Intra-Country Study of Trade and Unemployment Hasan, Mitra, Ranjan and Ahsan (JDE, 2012) use state level data from India create employment-weighted protection measures at state level Results: no signi cant relationship between trade liberalization and unemployment trade liberalization reduces unemployment in states with with exible labor market Inter-industry analysis lower trade protection associated with lower probabilities of becoming unemployed trade liberalization reduces industry-speci c unemployment but with a lag when all lags simultaneously introduced, rst lag negative, second and third lags positive trade liberalization increases unemployment on impact (one year after), but reduces unemployment over 2-3 years Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

45 Intra-Country Study of Trade and Unemployment Autor, Dorn, and Hanson et al. (working paper, 2012) study the e ect of rising Chinese import competition between 1990 and 2007 on local U.S. labor markets exploit cross-market variation in import exposure stemming from initial di erences in industry specialization Results similar to Hasan et al. (JDE, 2012) rising exposure increases unemployment, lowers labor force participation, and reduces wages in local labor markets explains one-quarter of the contemporaneous aggregate decline in U.S. manufacturing employment Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

46 Other related empirical work Davidson and Matusz (RIE, 2005) use rm level data from the U.S. negative correlation between job destruction and net exports across sectors equivalent to a positive correlation between net imports and job destruction Biscourp and Kramarz (JIE, 2007) use French rm level data exporting has a positive impact on job growth in French rms while importing has a negative e ect on job growth Hummels et al. (working paper, 2011) employer-employee data from Denmark exporting positively associated with employment but o shoring negatively associated with employment Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

47 Empirical evidence on o shoring and employment Gorg (Book chapter, 2011) surveys this literature Gorg and Hanley (IREF, 2005) using plant-level data for the electronics industry from Ireland nd that o shoring reduces employment Ibsen et al. (working paper, 2010) using Danish data nd that the import of intermediate goods is positively related with employment growth consistent with a strong productivity e ect generated by imported inputs Amiti and Wei (World Economy, 2009) service o shoring by U.S. rms has a small negative employment e ect when industries de ned narrowly e ect disappears at a higher level of aggregation (96 industries) Amiti and Wei (Economic Policy, 2005) using U.K. data nd that o shoring of services does not negatively a ect either the manufacturing employment or the services employment Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

48 empirical evidence on o shoring and employment Hizen and Swaim (NIER, 2007) using data from 17 countries nd that o shoring within the same industry has no impact on employment but inter-industry o shoring a ects employment positively Wagner (RWE, 2011) using rm-level data from Germany does not nd any statistically signi cant e ect of o shoring on employment, but he does nd a strong positive e ect of o shoring on rm productivity Harrison and Mcmillan (ReStat, 2011) when the tasks performed by the subsidiary of a multinational are complementary to the tasks performed at home, o shoring leads to more job creation in the United States consistent with strong productivity e ect in the case of input complementarity in Groizard, Ranjan, and Rodriguez-Lopez (2013) Groizard, Ranjan, and Rodriguez-Lopez (2013) use establishment level employment data from California Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48

49 Concluding Remarks growing theoretical and empirical literature on globalization and unemployment theoretical literature identi es channels through which globalization a ects unemployment productivity channel, endowment channel changes in sectoral unemployment rates, inter-sectoral allocation of labor o shoring channel (productivity e ect, job relocation e ect etc.) very little policy oriented research empirical literature cross-country evidence points towards bene cial e ects of trade liberalization on unemployment evidence of short term spike followed by long term decline in unemployment evidence on the relationship between o shoring and employment/unemployment mixed more work needs to be done Ranjan (University of California, Irvine) Globalization and Unemployment 27th June, / 48