May 2000 Briefing Paper
China and the States
Booming trade deficit with China will accelerate job destruction in next decade with losses in every state
by Robert E. Scott
In April, the Clinton Administration published several hundred pages of state-by-state “opportunity reports” purporting to show that “the passage of PNTR [Permanent Normal Trade Relations with China]…would open new export and employment opportunities in all 50 states” (White House 2000). These reports were issued in an attempt to persuade Congress to approve a recently negotiated trade deal with China to ease its entry into the World Trade Organization (WTO). These state reports, however, are embarrassingly shallow-they offer no serious analysis and were obviously put together cookie-cutter fashion, with much of the information and claims of great benefits repeated word for word from one state report to the next (China Trade Relations Working Group 2000). These reports not only fail to provide a single estimate of the jobs to be gained in any of the states, they also totally disregard the role of imports in trade. It should go without saying that exports and imports are the two fundamental components of trade. In order to discern whether trade has contributed to economic growth or detracted from it, it is necessary to look at both of these components. Ignoring the impact of imports is like trying to balance a checkbook by counting only the deposits and ignoring the withdrawals. If PNTR for China is approved by Congress, a projection of trade trends over the next decade shows that the trade deficit will expand, resulting in sizeable job losses in every state and in virtually every sector of the economy. These projections are based on the U.S. International Trade Commission’s (USITC) own China-U.S. trade model and clearly demonstrate that if PNTR passes, the already huge trade deficit with China will only get worse. The existing trade deficit with China is the product of current U.S. trade policies: while China runs a huge trade surplus with the United States, it has a sizeable trade deficit with the rest of the world, and with the European Union and Japan in particular. By giving China PNTR status, Congress will be giving up America’s most effective tool for changing those policies-bilateral trade negotiations designed to address the unique trade problems with China. Without the ability to negotiate in this way, the deficit with China will surely grow. As the trade deficit grows, job losses will continue to mount. The job loss forecasts in this report are actually conservative and understated, as they were arrived at using the USITC’s methodology. In fact, the USITC’s analysis makes some optimistic assumptions: (1) that China will comply with all the terms of the agreement; (2) that China will not devalue its currency; and (3) that the inevitable increase in U.S. foreign investment will not lead to an increase in U.S. imports from China. Under the protections offered by this proposed trade agreement, U.S. firms clearly intend to greatly expand foreign direct investments in China, just as they did in Mexico after the NAFTA agreement was completed. U.S. affiliate plants in China will rapidly expand their exports to the United States and reduce U.S. exports to China, which will worsen the U.S. trade deficit, as was the case with Mexico after NAFTA took effect (Burke 2000).
If Congress approves this trade pact and grants China PNTR:
The absolute level of the U.S. trade deficit with China will increase by at least 80% between 1999 and 2010, resulting in the elimination of 872,091 jobs during the next decade, even if U.S. exports to China grow more rapidly than imports from that country.
Every state will suffer significant net job losses over the next decade, as U.S. trade deficits expand. Major losses will be experienced in California (84,294 jobs lost), Texas (50,409 lost), and Pennsylvania (45,824 lost). Eleven states will lose more than 28,500 jobs in this period.
Every industry in the United States will lose jobs due to increased trade deficits, including agriculture and other natural resources (5,095 jobs lost), services (58,391 lost), and most of all manufacturing (742,201 lost, representing 85% of total jobs destroyed).
U.S. imports from China more than tripled in real terms between 1992 and 1999 (increasing 202%), while U.S. exports to China increased by only 69% in the same period, as shown in the top half of Table 1. The U.S. trade deficit of $16.5 billion with China in 1992 increased 256% to $58.7 billion in 1999 (all trade flows are reported in inflation-adjusted 1987 dollars). 
The USITC estimated that the China-WTO deal will increase U.S. exports to China by an average of 10.1% and U.S. imports from China by 6.4% (USITC 1999, Table ES-4, xix).  A central weakness of the USITC study is that it fails to consider the underlying dynamic trends in U.S.-China trade or to state how those trends will be affected by China’s entry into the WTO. Job losses caused by growing trade deficits with China are reported in the last column of Table 1. The United States lost 683,231 jobs due to growing trade deficits with China between 1992 and 1999. The China-WTO deal under consideration will eliminate at least 872,091 jobs between 1999 and 2010, even with the USITC’s highly optimistic assumptions. If China devalues its currency or fails to live up to the terms of the agreement, job losses will be even larger.
The net total for jobs lost to growing trade deficits under the China-WTO deal reflects even higher levels of job turnover in exporting and import-competing industries (bottom half of Table 1). The expansion of exports to China is forecast to create 276,221 jobs between 1999 and 2010, while the growth in imports will eliminate 1,148,313 domestic jobs. All workers losing jobs, and most of those who move from one job to another, will experience substantial costs that usually include temporary and permanent reductions in wages and benefits.
When the economy is at or near full employment, as it was in 1999, then jobs destroyed by trade deficits will not always result in an increase in the overall level of unemployment. In this case, the most important effect of trade on the economy may be its role in moving labor out of one industry and into another. A total of 11.3 million workers – about 8.8% of the labor force – have either gained or lost a job opportunity due to trade between 1992 and 1999 (Scott 2000c). Thus, a significant share of the labor force was exposed to trade-related adjustment in the past seven years, which helps explain the rise in economic insecurity expressed by workers in this era, despite high levels of economic growth and historically low levels of unemployment.
will suffer job losses
State-level employment in each of the 183 industries is used to estimate the job gains or losses of the China-WTO deal in each of the 50 states. Every state in the United States will suffer significant job losses from the China-WTO deal over the next decade, as shown in Figure 1.
Job losses for each state are reported in Table 2. Major losses will be experienced over the next decade in California (84,294 jobs lost), Texas (50,409 lost), and Pennsylvania (45,824 lost). Eleven states will lose more than 28,500 jobs in this period. Other states that will experience significant job losses include New York (58,699 jobs lost), North Carolina (47,151 lost), Tennessee (38,098 lost), Ohio (34,687 lost), and Massachusetts (28,501 lost). A number of farm states will also be hard hit, including Illinois (38,082 jobs lost), Indiana (30,324 lost), Minnesota (17,833 lost), and Kansas (2,573 lost). Job losses will be felt in every sector of the economy
The China-WTO deal will eliminate jobs in every sector of the economy over the next decade, as shown in Table 3. These results include both the direct effects of trade on an industry, as well as the indirect effects on workers in industries that provide goods and services to exporting and import-competing industries (see Appendix 1). Thus, jobs will be lost in agriculture, mining, and manufacturing, all of which are sectors that directly compete with imports from China. However, large numbers of jobs will also be lost in construction, transportation, finance, and other services. Each of these sectors provides goods or services that are used in the production of traded goods.
The largest effects of China’s accession to the WTO will be felt in the manufacturing sector, with 85.1% of the total jobs lost due to rising trade deficits. The rise in imports from China will be responsible for 81.2% of the jobs lost in manufacturing. Interestingly, only 68.7% of the jobs gained through increased exports will be located in manufacturing. This is not, however, the result of higher wages paid in export-related industries (see Appendix 2). Rather, services production requires much higher levels of non-labor input than the manufacturing sector. The surprising loss of jobs in agricultural industries
Overall employment in agriculture, forestry, and fishery industries is estimated to decline by about 5,000 jobs (as shown in Table 3), despite the USITC’s forecast of an improved trade balance for this sector.
The indirect effects of the massive $47 billion growth in the total U.S. trade deficit with China between 1999 and 2010 will offset the direct gains from the $440 million increase in agricultural exports to China. Thus, farmers will be hurt by the China-WTO deal as well. Agri-business interests are the only sectors of the farm community that are sure to benefit from the China-WTO deal, since they profit from the growth in both exports to and imports of farm products from China. Furthermore, if the actual impacts of the deal are worse than forecast by the USITC, then job losses in the agricultural sector could accelerate, as they did following the completion of the NAFTA agreement. Conclusion
The United States deserves a better deal. This proposed trade pact does not solve the current U.S. trade problem with China and will, in fact, make matters worse. By abandoning Congress’ annual review of trade relations with China, the United States is forever sacrificing all other means for dealing with this problem, leaving the weak WTO dispute resolution system as the only mechanism for addressing trade problems. The Clinton Administration’s proposal will eliminate the opportunity for direct, bilateral negotiations over trade. It should be clear that the United States would be better served by a deal that doesn’t trade away the best means of shaping a balanced trade relationship with China.
May 2000 The author thanks Jung Wook Lee and Nick Trebat for administrative and research assistance, and Jeff Faux and Christian Weller for comments on earlier drafts.
This study uses the model developed in Rothstein and Scott (1997a and 1997b), which solves at least three problems prevalent in previous research on the employment impacts of trade, including:
the tendency to look only at the effects of exports and ignore imports;
the failure to adjust trade data for inflation;
the application of a single employment multiplier to all industries, despite differences in labor productivity and utilization.
The model developed here assumes that real U.S. imports in 1999 grow at the rates estimated by the USITC, with exports increasing 10.1% per year and imports growing 6.4% per year (USITC 1999, Table ES-4, xix).  These estimates do not take into account the rapid underlying expansion of U.S. imports, exports, and the bilateral trade deficit with China. The USITC model assumes that the two economies were in equilibrium prior to the accession agreement, and that they will move to a new equilibrium in a short period of time after China joins the WTO. This report, however, places the USITC forecast in a dynamic context by assuming that exports and imports continue to grow each year at the rates estimated, indefinitely, or until this new equilibrium growth path is disturbed by some other economic “event,” such as a currency crisis in China or the United States. This approach has been criticized by some academics, but these critics often have simply made up new estimates of the U.S.-China trade deficit that have no relationship to official U.S. government statistics. In fact, the dynamic model used in this report is extremely favorable to the government’s case,  illustrated by comparing the growth rates shown in the fourth column of Table 1. This study assumes that U.S. trade with China increases each year at the overall rates forecast by the USITC in each of 183 industries. The resulting trade flows in 1999 and 2010 are shown in the bottom half of Table 1. Exports would increase by 188% in the next decade, as shown in the fourth column, an expansion that is 2.7 times larger than the increase that actually occurred between 1992 and 1999. U.S. imports from China will expand by only 98% in the next decade (see the bottom half of Table 1), after increasing 202% in the past seven years alone. Thus, the USITC’s forecast assumes that the growth rate of exports will nearly triple, while import growth will fall by more than half. If these results were to be achieved, the China-WTO agreement might be viewed as a truly remarkable milestone. Unfortunately, recent U.S. history with the NAFTA and WTO agreements suggests that these estimates widely overstate the actual benefits of the accession agreement. For example, the United States had a small surplus in its trade with Mexico in 1993, the year before NAFTA took effect.
This surplus turned into a deficit of $22.8 billion last year, which has resulted in the closure of thousands of U.S. factories and the loss of hundreds of thousands of jobs in this country (Scott 1999b and 2000b). The NAFTA experience notwithstanding, this report uses the USITC’s estimates of the effects of the WTO agreement on U.S.-China trade to estimate the employment effects of that agreement, which is effectively a best-case scenario for the trade proposal.  The model used here is based on the Bureau of Labor Statistics’ 183-sector employment requirements table, which was derived from the 1987 U.S. input-output table and adjusted to 1993 price and productivity levels (BLS 1996). A three-step process was used to estimate the effects of trade on the United States, by industry, and then to allocate those job losses to each of the 50 states. Step 1: Conversion of trade data to constant dollars
We used three-digit-industry, SIC-based trade data (Bureau of the Census 2000). These were deflated to constant, 1987 dollars with industry-specific, chain-weighted price indices (BLS Office of Employment Projections 1999) that are based on the BLS producer price index (PPI). The BLS price indices for 1999 will not be released until sometime in the summer of 2000. 1998 and 1999 estimates for the annual rates of growth of the PPI (BLS 2000) at the three-digit SIC level were combined with a concordance of SIC and BLS three-digit industries, and used to construct an estimate of the PPI for each BLS industry in 1999. The constructed BLS series was then used to deflate current dollar trade flows with China for 1992 and 1999. Step 2: Estimating the national employment impacts of changing trade flows
The BLS 183-sector employment requirements table was used to estimate the employment impacts of trade for each of the 183 industries in the BLS model, and for groups of one-digit industries.  Results of these estimates are presented in Tables 1 and 3. Specifically, we defined E as the employment-requirements [183 x 183] matrix and let T be a [9 x 183] matrix of trade flows. The columns of T represent exports, imports, and the trade balance in period 1, period 2, and the changes in each of these variables over the period. Then job gains or losses for all industries, for each period and trade flow, matrix J, are determined by the following equation:
J = ET
J is also a [9 x 183] matrix, and the elements of each column are the sum of the direct and indirect job losses or gains in that industry. Thus, for example, element J1, 1 is the employment loss or gain in industry 1 due to exports in all 183 industries in period 1. Trade flows in 2010 were estimated, as described in the text, by assuming that exports increase by 10.1% per year and that imports expand by 6.4% per year, given real trade flows in each of the 183 BLS industries in 1999. Step 3: Estimating the employment effects of trade at the state level
State-level employment effects were calculated by allocating imports and exports to the states on the basis of their shares of three-digit, industry-level employment (BLS 1997).  Results of these estimates are presented in Figure 1 and Table 2. An ES202 file (UI data) was used to construct S, a [50 x 183] matrix, the elements of which are the U.S. employment shares of each of the 183 BLS industries in each of the 50 states. Job gains or losses for each state, for each period, and flow, L are determined by the following equation:
L = SJ
In other words, job losses in each state L are estimated by allocating job losses in each industry, at the national level J, to the respective industries in each state S. The elements of L, which is a [50 x 9] matrix, are thus the sum of the job losses in each of the 50 states in all industries, for each of the nine trade flows in Table 1 (e.g., exports, imports, and the trade balance in 1992 and 1999 and the change in each variable in this period). The results of L for the change in the trade balance between 1999 and 2010, for each state, are summarized in Figure 1 and Table 2. This is the last column of matrix L – the others are not reported here but are available from the author upon request to firstname.lastname@example.org.
A prior EPI study of the effects of the WTO-deal on U.S. employment (Scott 2000a) used a “multiplier” relationship that assumed that $1 billion of U.S. exports would generate 13,000 jobs in the United States, following the earlier work of Gary Hufbauer at the Institute for International Economics.  Hufbauer has retracted his earlier private forecast and now claims that only 6,800 U.S. manufacturing jobs are created “per billion dollars of additional output” (Hufbauer and Rosen 2000, 7).  Hufbauer’s new estimate of the employment multiplier is directly contradicted by the “domestic employment requirement matrix” published each year by the U.S. Bureau of Labor Statistics (Tables 1, 2, and 3 in this report are based on actual U.S. trade data for 1999 and the BLS employment-requirements matrix). These calculations, as described in Appendix 1, also show that U.S. exports generated 11,200 jobs per billion dollars of exports (measured in current 1999 dollars), including 7,700 manufacturing jobs. However, imports that are displaced by goods from China displaced 14,300 jobs per billion dollars, and the overall trade deficit resulted in a net loss of 15,000 jobs per billion.
Employment multipliers for exports are smaller than those for imports because manufacturing made up only 69% of U.S. jobs supported by exports to China in 1999, and 81% of U.S. import-jobs from that country. Services are the next largest source of employment from both exports and imports. Employment multipliers for exports are not lower because wages are higher in services (see Mishel et al. 1999, 129). Rather, the labor content of services is lower simply because profits, interest, and rent payments are a larger share of the costs of doing business in these sectors. Simply put, on average, exports generate less employment than competing imports, and the wages paid in those export industries are no better than in the more numerous import jobs that are displaced. This is especially true at the margin in those industries where exports and imports are expanding most rapidly (Scott, Schmitt, and Lee 1999) and where wages in import-competing industries (such
as motor vehicles and electronics) are actually higher than in rapidly growing export sectors (such as agricultural products).
Hufbauer and Rosen (2000, 10) claim that the actual trade deficit in 1999 was $43 billion rather than $68 billion and that “the deficit numbers used by the opponents…are distorted because they miscount Chinese trade routed through Hong Kong” and because trade in services is not included in the deficit figures used. Hufbauer and Rosen’s claims, however, are simply inconsistent with the facts. In 1999 the United States had a surplus of $2.1 billion in goods trade with Hong Kong (Bureau of the Census 2000), and a surplus of $1.6 billion in services trade with China. These figures explain less than $4 billion of the claimed $25 billion “miscount” of China’s trade.
Hufbauer and Rosen’s claim that official U.S. trade data are “distorted” is particularly suspect. If the Clinton Administration’s own trade estimates are unreliable, then why are its estimates of the benefits of the China-WTO deal any more reliable? This report continues to rely on official, government estimates of U.S.-China trade flows. Inclusion of trade with Hong Kong and services trade with China would not significantly affect the results of this study.
1. U.S. exports to China in current 1999 dollars were $13.1 billion, and imports from China were $81.8 billion, resulting in a trade deficit of $68.7 billion in 1999. [RETURN TO TEXT]
2. These results include all static and dynamic effects on U.S.-China trade only. [RETURN TO TEXT]
3. The USITC report also includes forecasts of export growth for selected 2-, 3-, and 4-digit industries. However, these estimates are incomplete, for the purposes of this study, which requires 3-digit growth forecasts for all components of U.S. trade. Furthermore, no forecast is presented of the expected growth of imports by industry from China. Therefore, the average rates of growth of exports and imports that were forecast by the USITC are applied to each 3-digit industry for consistency of treatment. [RETURN TO TEXT]
4. Hufbauer and Rosen (2000, 7) claim that “the EPI calculations are built on an absurd extrapolation of an exaggerated bilateral deficit.” This rather hyperbolic charge is based on numerous assertions that are at odds with official U.S. government trade statistics. See Appendix 2 for further discussion of the Hufbauer and Rosen critique of Scott (2000a). [RETURN TO TEXT]
5. See Scott (2000a) for a further discussion of three major reasons why the USITC has probably overestimated the benefits of the China Accession Agreement, including: (1) China’s history of violating the letter and spirit of every trade agreement it has signed with the United States for at least the past two decades; (2) exclusion of services from these estimates, especially since services are likely to facilitate foreign direct investment in China that will further increase the U.S. trade deficit (see Burke 2000); and (3) the implicit assumption that China will not devalue its currency, notwithstanding the fact that a 30% devaluation by China (last done in 1994) would more than offset the value of any and all tariff reductions included in the agreement (see forthcoming paper by Palley, AFL-CIO). [RETURN TO TEXT]
6. Three-digit PPI estimates were unavailable for a number of commodities, including farm products; lumber and wood products; metal mining; coal, oil, and natural gas; and non-metallic minerals. Two-digit PPI data were used to approximate price changes in these industries, and for a few manufacturing industries for which 3-digit PPI date were not available. [RETURN TO TEXT]
7. The most recent version of the BLS employment requirements table is available from the Bureau of Labor Statistics (1999). The documentation file for this matrix, “BLS 1998 Domestic Employment Requirements Matrix (DOMREQ98.DAT),” notes that “the entries in each row of the 1998 domestic employment requirements table show industry employment supported directly and indirectly per $1 million (1992 prices) of sales of goods to final users for the commodity shown at the head of each column.” An earlier version of this matrix (based on 1987 prices) is used here to maintain comparability with state employment shares (see Appendix 1). [RETURN TO TEXT]
8. Other studies (see California State World Trade Commission, 1996, which found 47,600 jobs created in California from increased trade with Canada alone) have allocated all employment effects to the state of the exporting company. This is problematic because the production-along with any attendant job effects-need not have taken place in the exporter’s state. If a California dealer buys cars from Chrysler and sells them to Mexico, these studies will find job creation in California. However, the cars are not made in California; the employment effects should instead be attributed to Michigan and other states with high levels of auto industry production. Likewise, if the same firm buys auto parts from Mexico, the loss of employment will occur in auto industry states, not in California. [RETURN TO TEXT]
9. These employment estimates assume that each $1 billion of exports generates 13,000 jobs in the domestic economy, following Hufbauer (1999), and that $1 billion of imports eliminates 13,000 jobs. [RETURN TO TEXT]
10. Hufbauer and Rosen do not provide input-output based evidence in support of this assertion (Hufbauer and Rosen 2000, 7). Rather, they present a completely ad hoc regression model (Hufbauer and Rosen 2000, Appendix C), based on aggregate data for U.S. manufacturing exports for 39 quarters between 1992 and 1999. The BLS employment-requirem
ent matrix used in this study, which is based on long-established procedures and widely accepted national accounting frameworks, uses data for U.S. trade in 183 3-digit industries (based on the 1992 SIC classification). Given these differences in technique, it is somewhat surprising to find that Hufbauer and Rosen’s estimate falls within 13% of the BLS estimate for 1999 cited here. The Institute for International Economics ignored the non-manufacturing jobs associated with exports in its recent study; correcting for this would add another 3,500 jobs per billion of 1999 dollars to the manufacturing total derived here. [RETURN TO TEXT]
Bronfenbrenner, Kate. 1997. “We’ll Close! Plant Closings, Plant-Closing Threats, Union Organizing and NAFTA.” Multinational Monitor. March, pp. 8-13.
Bureau of the Census. 2000. Unpublished data from “Special Compilation of U.S. Trade Statistics.” Available in machine readable form. Washington, D.C.: U.S. Department of Commerce.
Bureau of Labor Statistics. 2000. Selective Access link to Producer Price Index Revision-Current Series. < http://stats.bls.gov/sahome.html >
Bureau of Labor Statistics. 1999. Domestic Employment Requirement Matrix. < ftp://ftp.bls.gov/pub/special.requests/ep/ >
Bureau of Labor Statistics. 1997. ES202 Establishment Census. Washington, D.C.: U.S. Department of Labor.
Bureau of Labor Statistics, Office of Employment Projections. 1996. Employment Outlook: 1994-2005 Macroeconomic Data, Demand Time Series and Input Output Tables. Washington, D.C.: U.S. Department of Labor.
Bureau of Labor Statistics, Office of Employment Projections. 1999. Unpublished data from upcoming Employment Projections. Washington, D.C.: U.S. Department of Labor.
Burke, James. 2000. China, U.S. Multinationals, and the Trade Deficit: U.S. Firms Build Export-oriented Production Base in China’s Low-wage, Low Labor-protection Economy. Briefing Paper. Washington, D.C.: Economic Policy Institute.
California State World Trade Commission. 1996. “A Preliminary Assessment of the Agreement’s Impact on California.” Sacramento, Calif.: California State World Trade Commission.
China Trade Relations Working Group. 2000. “State by State Opportunity Reports.” < http://www.chinapntr.gov./statebystate8.htm >
Hufbauer, Gary. 1999. Personal correspondence regarding “Analysis on Implications of WTO Entry for China’s Economy” from the China Business Council (May 13).
Hufbauer, Gary, and Daniel H. Rosen. 2000. American Access to China’s Market: The Congressional Vote on PNTR. International Economics Policy Briefs. Washington, D.C.: Institute for International Economics.
Mishel, Lawrence, Jared Bernstein, and John Schmitt. 1999. The State of Working America: 1998-99. Ithaca, N.Y.: ILR Press, an imprint of Cornell University Press.
Rothstein, Jesse, and Robert E. Scott. 1997a. NAFTA’s Casualties: Employment Effects on Men, Women, and Minorities. Issue Brief #120. Washington, D.C.: Economic Policy Institute.
Rothstein, Jesse, and Robert E. Scott. 1997b. NAFTA and the States: Job Destruction Is Widespread. Issue Brief #119. Washington, D.C.: Economic Policy Institute.
Scott, Robert E, Thea Lee, and John Schmitt. 1997. Trading away good jobs:An Examination of Employment and Wages in the U.S., 1979-94. Briefing Paper. Washington, D.C.: Economic Policy Institute.
Scott, Robert E. 1999. NAFTA’s Pain Deepens: Job Destruction Accelerates in 1999 With Losses in Every State. Briefing Paper. Washington, D.C.: Economic Policy Institute.
Scott, Robert. E. 2000a. The High Cost of the China-WTO Deal: Administration’s Own Analysis Suggests Spiraling Deficit, Job Losses. Issue Brief #137. Washington, D.C.: Economic Policy Institute.
Scott, Robert E. 2000b. NAFTA Imports Lead Growth in U.S. Trade Deficit. Trade Fax. Washington, D.C.: Economic Policy Institute (February 18).
Scott, Robert E. 2000c. The Facts About Trade and Job Creation. Issue Brief #139. Washington, D.C.: Economic Policy Institute (March 24).
U.S. International Trade Commission (USITC). 1999. Assessment of the Economic Effects on the United States of China’s Accession to the WTO. Investigation No.332-403, Publication 3229. Washington, D.C: USITC.
The White House. 2000. “Commerce Secretary Daley Releases State Export Opportunity Reports Highlighting Benefits of Expanded Trade with China.” Press Release (April 12). http://www.whitehouse.gov . >