劳动力资源对进出口竞争的影响：以纺织业为例【外文翻译】.doc

1、 外文翻译 原文 LABOR MARKET EFFECTS OF IMPORT COMPETITON: THEORY AND EVIDENCE FROM THE TEXITILE AND APPAREL INDUSTRIES Material Source: Atlantic Economic Journal Author: BEN S. SHIPPEN Abstract: Since the early 1980s, much attention has been given to the possibility of trade-related job losses and wage ef

2、fects in the textile and apparel industries. This paper uses aggregate time series data from the Annual Survey of Manufacturers (Bartlesman and Gray, 1996) with import price data from the Bureau of Labor Statistics (Alterman, 1991) for 1977-91 to test the effect of imports on employment and wages in

3、 textiles and apparel. Theoretical models suggest that import competition should be a factor in the determination of employment, and possibly wages, regardless of whether the country is represented as a price-setter or price-taker. The empirical analysis provides some support. Key words: Labor marke

4、t Import competition Wage Textile industry Introduction There is widespread speculation that international trade has a large negative effect on employment and wages especially in low wage sectors where foreign labor costs are particularly low. Since the early 1980s, much attention has been given to

5、the possibility of displaced domestic labor and wage losses in the textile and apparel industries. Historically, textiles and apparel have been important industries in the U.S. economy, in terms of both output and employment. As recently as 1973, they employed 2.3 million workers or more than 11 per

6、cent of the total number of workers in manufacturing. By 1993 however, employment had declined to about 1.6 million workers and 8.3 percent of manufacturing-a decline of more than 30 percent. Real wages for both industries remain well below the manufacturing average. In 1993 the average apparel wage

7、 was 41percent less, and the average textile wage was 26 percent less than average wages in other manufacturing industries. A popular explanation for employment declines and low wages is the presence of import competition. The purpose of this paper is to look at this popular notion by testing the ef

8、fect of imports on employment and wages in textiles and apparel with aggregate time series data. Such an examination is absent in literature. This paper is organized as follows. First, a theoretical model of textile and apparel supply and demand is presented to explain how import competition is expe

9、cted to affect workers. In the world market, the U.S. is modeled as both a price-setter and a price-taker. The theoretical models suggest that import competition could be a factor in determining employment regardless of whether the U.S. is represented as a price-setter or price-taker. However, this

10、would only be a determining factor of wages if the U.S. is a price-setter. Two empirical models are developed to test whether the U.S. textile and apparel market is a price-setter or price-taker. If the U.S. textile and apparel industries are price-takers, then a measure of import price should be su

11、fficient to measure the effect of import competition on wages and employment. However, if the U.S. textile and apparel market is a price-setter, then endogeneity bias is a possibility in the import price variable and corrective steps must be taken to model the import effects. The analysis here sugge

12、sts that the effect of import competition on employment and wages is small. Theoretical Analysis There are two limiting, theoretical possibilities regarding the way the U.S. production market affects the price in the world market: 1) Every domestic change in demand and supply causes a change in worl

13、d prices. 2) Domestic changes in demand and supply have no affect on world prices. The first possibility assumes that the U.S. is a large producer or consumer in the world market and is, thus, a price-setter. The second assumes that the U.S. is a small producer and a small consumer on the world mark

14、et and is, thus, a price-taker. The distinction between these two possibilities is important because it determines whether import price is endogenous or exogenous to shocks in the domestic market. Starting with the assumption that the U.S. is a price-setter in the world market, the import demand cur

15、ve is a function of excess demand. The excess demand curve is negatively related to price and, as the domestic demand and supply curves shift, the intercept of the demand for imports changes. Assuming perfect substitutability between domestics and imports as demand for domestic production increases,

16、 then excess demand for import production will also increase. However, if domestic supply increases, then excess demand for imports will decrease. The aggregate import supply curve represents the excess supply of goods in other countries. The supply schedule is positively related to price and reflec

17、ts changes in foreign supply and demand. Foreign incomes and cost conditions abroad are inversely related to the supply schedule while technological changes positively affect supply. The alternative model assumes that the U.S. is a price-taker. The excess demand curve in this model is the same as wi

18、th the U.S. as a price-setter, with the intercept of the demand curve in the import market starting at the level of equilibrium of demand and supply in the U.S. domestic market. Changes in domestic conditions will change both the domestic equilibrium point and the excess demand in the import market.

19、 The supply curve in the import market is perfectly elastic, however, indicating that although changes in domestic demand and supply affect the quantity of imports purchased, they do not affect price. Employment and Wage Models of Import Competition Two models are used to calculate import competitio

20、n effects on employment and wages in the textile and apparel industries, The first assumes that the U.S. market is a price-taker and, following the empirical framework of (Grossman， 1986), uses import price as the measure of international competition. The second model assumes that the U.S. is a pric

21、e-setter and, in an approach similar to (Revenga， 1992), uses two-stage least squares (2SLS) with weighted exchange rates as an instrument for import price. First, consider the structural labor demand schedule. Let itL be labor demand in industry i and year t. Then dln itL = 1 dln itW +d itZ + itw1

22、(1) where: 1 is the employment elasticity with respect to the wage; itW is the average industry wage; itZ is a vector of observable factors that shift the demand for labor in industry i and year t; is a vector of employment elasticity that corresponds to itZ ; and itw1 is an error term that is desig

23、ned to capture unobserved demand shocks. The vector of variables that shifts labor demand includes the measures of aggregate demand in the economy and direct measures of labor demand. The vector of demand shifters include measures of import competition, real gross national product (GNP), domestic pr

24、oduction, and an index of the cost of materials. Employment is expected to vary positively with GNP and the industry output. It could vary inversely or positively with the cost of materials depending on whether the output or substitution effect is dominant. The measure of import competition is an in

25、dex of import prices. Employment is expected to vary positively with import price as an increase in the import price increases the production in the U.S. Next, consider the labor supply to an industry: dln itL =c.dln itW +d itH + itw2 (2) where: itW is the industry wage at year t; c is the industry

26、supply elasticity with respect to the wage; itH is a vector of observed factors that shift labor supply; is a vector of industry supply elasticity; and itw2 represents unmeasured supply shocks. It is necessary to estimate the system in reduced form because wages and employment are simultaneously det

27、ermined. The equations written in reduced form are: dln itL = 1 d itZ + 2 d itH + itv (3) and dln itW = 1 d itZ + 2 d itH + itv (4) where itv and itv are error terms that reflect unmeasured shocks to labor demand and supply. For 2SLS estimation, a first-stage equation is required to estimate the end

28、ogenous import price variable used in the second-stage equation. The first-stage equation is: dln itP = 1 d itZ + 2 d itH + 3 d itI + it (5) where itP represents the average import price for industry i and year t, itI represents the weighted exchange rate variable for industry i and year t, and it r

29、epresents a random error term for industry i and year t. Predicted values of dln itP are then used to estimate (3) and (4) using 2SLS estimation. Data The data used are obtained from the National Bureau of Economic Research productivity and trade databases. These are compiled for the 1972 manufactur

30、ing industries at the standard industry codes four-digit level (450 total) for 1958-91. Other variables include value of shipments, cost of materials, production employment, production hours, and production payroll on an annual basis. These variables have been converted to the three-digit level to a

31、ccommodate the index of import price data. Variables that are not included in the National Bureau of Economic Research databases but are used in these equations are GNP, an index of import prices, and a weighted exchange rate. Constant GNP is drawn from the Economic Report of the President 1993 for

32、1972-91 while the import price variable is international price indices taken from the Bureau of Labor Statistics (Alterman， 1991) for selected three-digit manufacturing industries for 1977-91. The weighted exchange rate is from the Bureau of Census International Trade Division(Alterman， 1991) for th

33、e top five exporting countries to the U.S. in textiles and apparel in 1991, and the total weighted index is from the Economic Report of the President. Finally, implicit price deflators for value of shipments and the index of the cost of materials are also taken from the Economic Report of the Presid

34、ent. The Results of the Ordinary Least Squares (OLS) and 2SLS Models Table 1 presents the results of the import price specification. The employment elasticity of domestic output with respect to hours and employment in the apparel industry is positive and significant. Also in apparel, the coefficient

35、 on the alternative real wage is estimated to be -1.558 in the employment equation, suggesting that a one percent increase in the alternative wage (which, in turn, increases own wage) would reduce employment by 1.59 percent in these industries. In the textile industry, the coefficient on the alterna

36、tive wage in the employment equation is also significant. GNP and the index of the cost of materials in this model have no effect on employment or wages in either industry. TABLE 1 Reduced Form OLS Regressions with Import Price Variable Hours Employment Wage Apparel industry n=51 Import Price 0.442

37、(2.209) 0.359 (2.207) -0.053 (0.466) GNP 0.314 (0.707) 0.727 (2.011) 0.236 (0.941) Cost of Materials -0.095 (0.462) -0.046 (0.277) -0.099 (0.853) Domestic Output 0.512 (3.746) 0.443 (3.991) 0.137 (1.782) Alternative Wage -1.436 (2.745) -1.588 (3.739) 0.736 (2.500) Time Trend -0.002 (0.606) -0.003 (1

38、.388) 0.003 (1.755) 2R 0.3678 0.441 0.3193 Durbin-Watson 2.000 1.912 2.266 Textile Industry n = 38 Import Price 0.081 (0.887) 0.061 (0.593) -0.038 (0.810) GNP 0.258 (0.659) 0.311 (0.708) -0.091 (0.455) Cost of Materials -0.179 (0.713) -0.049 (0.174) 0.104 (0.809) Domestic Output 0.829 (10.588) 0.624

39、 (7.112) 0.053 (1.329) Alternative Wage -0.887 (1.847) -1.327 (2.465) 0.877 (3.558) Time Trend 0.000 (0.083) -0.003 (0.792) -0.001 (0.850) 2R 0.7936 0.6448 0.4827 Durbin-Watson 1.657 1.915 1.851 The coefficient on import prices in the apparel industry is positive and significant for the hours and em

40、ployment equations, indicating that an increase in import price has a positive effect on hours worked and overall production employment. The coefficient on the import price variable is small and insignificant in the apparel wage equation. This result, however, is not unexpected if apparel workers ar

41、e taken from a large labor market where apparel firms are wage-takers. On the other hand, the coefficient on import prices in the hours, employment, and wage equations in the textile industry is small and insignificant, suggesting that a change in the import price in this specification has little ef

42、fect on hours worked, people employed, or average wages of production workers. In this model, only changes in domestic output and the alternative wage have significant effects on hours, employment, or wages in the textile industry. Table 2 presents the results for the 2SLS model for the textile and

43、apparel industries. This estimation assumes that the U.S. is a price-setter and is designed to correct for endogeneity bias in the import price variable by estimating import price as a function of weighted exchange rates. This corrected import price variable should not be correlated with the depende

44、nt variables or with industry-specific domestic shocks. TABLE 2 2SLS Regressions with Import Price: Variable Hours Employment Wage Apparel industry n=56 Import Price 0.528 (1.033) 0.541 (1.168) 0.124(0.456) GNP 0.337 (0.794) 0.405 (1.052) -0.114(0.504) Cost of Materials 0.190 (0.833) 0.144 (0.694) -

45、0.038(0.311) Alternative Wage -1.0551(1.906) -0.568(1.131) 0.591(2.010) 2R 0.1162 0.0874 0.0135 Durbin-Watson 1.885 2.035 2.095 Textile Industry n = 42 Import Price 0.269 (0.828) 0.211 (0.726) 0.033(0.408) GNP 0.872 (1.283) 0.710 (1.168) 0.254(1.484) Cost of Materials 0.366 (0.804) 0.446 (1.092) -0.

46、028(0.247) Alternative Wage -1.500 (1.598) -0.369(0.440) 0.639(2.707) 2R 0.2357 0.1208 0.1463 Durbin-Watson 1.958 1.868 1.865 Domestic output remains the most significant variable in the employment equations of both industries, with positive coefficients in the two-stage estimations. The alternative

47、 wage variable in the employment equation continues to be positive in the textile industry and negative in apparel. GNP and the index of the cost of materials are insignificant in all equations for both industries. The coefficient on the import price variable in the 2SLS estimations increases from t

48、he previous OLS estimations for textiles and apparel. In the apparel industry, the coefficient in the hours and employment equations increased from 0.442 to 0.528 log points and from 0.359 to 0.541 log points, respectively. 4 Since using 2SLS models often results in a loss of efficiency compared to

49、OLS estimates, it is not surprising that these results change from being significant to insignificant at standard levels. The estimation would benefit with increased observations to determine if the problem is associated with the small sample, or if import prices truly have no effect. The estimate of import price also appears larger in the 2SLS equations of the textile industry, compared to the results of the OLS estimation, although the coefficients are insignificant. Conclusions The results from the two models are m