«Professor Dwight M. Jaffee Fisher Center for Real Estate and Urban Economics Haas School of Business University of California Berkeley CA 94720-1900 ...»
International Trade and California Employment:
Some Statistical Tests
Professor Dwight M. Jaffee
Fisher Center for Real Estate and Urban Economics
Haas School of Business
University of California
Berkeley CA 94720-1900
(Second in a series of working papers on the general topics of Foreign Trade and California’s
Growth, by Dwight M. Jaffee, Cynthia A. Kroll, Ashok Deo Bardhan, Josh Kirschenbaum, and
David Howe) Working Paper: 98-259 February 1998 Acknowledgement I want to express thanks to my colleagues on this project, Cynthia A. Kroll, Ashok Deo Bardhan, Josh Kirschenbaum, and David Howe, as well as David Lyons and other participants at the Public Policy Institute of California, for their comments on an earlier draft of this paper.
International Trade and California Employment:
Some Statistical Tests Abstract This paper presents preliminary statistical tests of the effects of rising international trade on California employment. We found a strong positive relationship between exports and manufacturing employment and a weak, but positive, effect of imports on manufacturing employment. Since higher imports are normally expected to reduce domestic production, the latter result suggests that there exists an additional link between imports and domestic demand.
A particular mechanism for this additional link is that low-cost imports of a good, based on foreign production, may lead to an expansion in domestic demand for the good. Although manufacturing employment in that industry will probably still fall as a result of the low-cost imports, employment in the associated service sector for that industry is likely to rise. Indeed, the increase in service sector employment may dominate, leading to an overall increase in the industry’s employment. The paper uses California’s computer industry to illustrate and quantify this mechanism.
1 Introduction International trade for the United States has been expanding rapidly in recent years, and it seems likely that it will continue growing rapidly for the foreseeable future. Based on the available evidence, California’s international trade is expanding at an even faster pace. In this paper, we present preliminary statistical tests of the effects of rising international trade on California employment. A primary innovationis that we look at the effect of international trade on both manufacturing and service sector employment.
In the traditional analysis of the manufacturing sector, imports and exports, the two components of international trade, are presumed to have opposite effects on a region’s employment. Everything else the same, rising exports are expected to raise domestic employment. This occurs because, if domestic demand is fixed at a moment in time, then rising export demand creates rising domestic production and employment. 1 Comparably, everything else the same, rising imports are expected to decrease domestic employment. This occurs because, if domestic demand is fixed at a moment in time, then rising imports substitute for domestic production, which creates falling domestic employment.
In reality, the connections between international trade and domestic employment are much more complex. Compared with the simple description, everything else is unlikely to be the same. At a fundamental level, international trade may not have any long-run effect on the employment level, if wages, prices, and exchange rates are always adjusting to maintain full employment. Of course, there may still be a short-run impact of trade on employment because the economy’s equilibrium adjustment takes time. In the context of this study, there This assumes the economy is not operating at full capacity or full employment.
are three further reasons why international trade may influence employment even if total US
employment is fixed:
1) International trade may influence the distribution of employment among the states.
2) International trade may influence the level of employment across sectors and industries.
3) International trade may influence the employment share of production and nonproduction workers.
As an example of this third effect, international trade may switch the proportions of total employment between the manufacturing and service sectors of an economy. Specifically, in this paper, we consider the possibility that rising international trade creates rising service sector employment, even as it creates falling manufacturing sector employment. In fact, we find in the computer industry that international trade may actually be the source of rising total employment, even though it creates falling manufacturing sector employment!
As another example, international trade may alter the proportions of domestic employment between low skill and high skill jobs. Specifically, Bardhan and Howe (1998b) study the extent to which imports tend to reduce low skill jobs, while exports tend to expand high skill jobs. This, of course, is related to the switch between service sector and manufacturing sector employment created by international trade.
The organization of this paper is as follows:
In Section 2, we present regression tests of the basic relationship between international trade and California’s manufacturing employment. These tests rely only on US international trade data, since California import data do not yet exist, as described in detail in Jaffee (1998).
The results confirm a strong relationship between international trade and manufacturing sector employment, especially that rising exports are associated with rising employment.
In Section 3, we expand the analysis to include the effect of international trade on service sector employment. Since the international trade data for service industries is extremely limited, (recall the discussion in Jaffee (1998)), we cannot carry out a detailed statistical analysis that would be comparable to the analysis in Section 2 for the manufacturing sector. Instead, we focus on one important industry, the computer industry, to illustrate how international trade interacts with both manufacturing and service sector employment. We find that the impact of international trade on computer industry service employment is decidedly positive, so much so that the impact of international trade on total employment in the computer industry is likely to be positive, even if the effect of trade on manufacturing employment in the computer industry is negative. These results are reconfirmed by our detailed interview study of the computer industry, described in Bardhan and Howe (1998a).
Section 4 provides a summary of the paper’s conclusions.
2 Estimates of Trade Effects on California’s Manufacturing Employment In this section, we develop and test a simple model of the relationship between international trade and manufacturing employment in California. We evaluate the effects of
international trade at three different levels:
1. The influence of international trade on US production for each industry.
2. The influence of international trade on California production for each industry.
3. The influence of international trade on California employment for each industry.
We begin with the influence of international trade on US production.
A Model of The Influence of International Trade on Production At the US level, for each SIC code i and at each time t, a basic identity links domestic
demand Di with domestic production Pi, exports Xi, and imports Mi :
This states that US domestic demand for a good is satisfied by US production, plus imports,
minus exports. 2 It is more useful to solve the equation for production:
Equation (3) indicates that the growth rate of production is determined by the weighted growth rates of domestic demand, exports, and imports, where each weight represents the lagged share of that variable relative to production. In particular, for a given value of demand growth D i, changes in the weighted growth rates of exports and imports create comparable changes in production growth (positive for rising exports and negative for rising imports).
Rising exports or imports, of course, may also create changes in domestic demand, in which case the total effect of export and import growth on production growth may vary from that A similar relationship was used in Jaffee (1998), equation (3).
shown in equation (3). The magnitude of these total trade effects can be evaluated by eliminating the identity by removing the industry demand variable from equation (3).
Specifically, we replace industry demand variable D i with an aggregate measure (over all SIC codes) D. For each industry i, all of the effects of international trade will then be attributed to
the export and import variables. The estimated equation is thus :
The null hypothesis is that the correlation of both export and import growth with demand growth for industry i is zero. In this case, the expected coefficient estimates for equation (4) are α 0 = 0, α 1 = 1, α 2 = −1, α 3 0. On the other hand, for example, if export growth were positively correlated with demand growth for industry i, then we would expect α1 1, and if import growth were positively correlated with demand growth for industry, then we would expect α2 -1.
Estimates of The Influence of International Trade on Production Pooled time-series, cross-section, estimates based on equation (4) are shown in Table 1.
All variables are measured as annual percent changes, consistent with equation (4). The data cover the years 1989 to 1995, so the estimation period is 1990 to 1995 annually taking into account the lagged values of the variables. The equations are estimated with ordinary least squares. The “a” equations in the upper half of the table are based on the cross-section of 2digit SIC codes and the “b” equations in the lower half of the table are based on the crosssection of 3-digit SIC codes. We begin by looking at the results based on the 2-digit SIC code cross-section.
Results Based on the 2-Digit SIC Code Cross-section Equation (1a) in Table 1 provides estimates of the total effects of exports and imports on US production growth across 2-digit SIC codes. Export growth is the most significant factor, with a coefficient close to 1.0, which is consistent with the null hypothesis that export growth and domestic demand growth are uncorrelated. Import growth, in turn, actually receives a positive, although small and insignificant, coefficient, quite different from the expected coefficient of –1.0 under the null hypothesis. This results suggests that imports tend to have a positive correlation with the growth of demand across goods. This could arise, for example, if the market demand for goods rises in response to the expansion of low-cost imports. We will discuss this mechanism in more detail in the following section. Finally, the control variable for aggregate demand has a significant and positive coefficient as expected.
The R2 is 0.40, which is respectable for a percentage change specification estimated using pooled data. Overall, the equation indicates a significant and positive net effect of international trade on US production growth.
Equation (2a) in Table 1 provides comparable estimates of the trade determinants of California’s production growth across 2-digit SIC codes. The export and import variables use the same US data as in equation (1a), due to the lack of a full set of California trade data. The aggregate production growth variable is the total for California’s 2-digit industries. The estimated coefficients indicate that exports are a significant determinant of California’s production growth, and that imports have a positive effect that is actually larger and more significant than for the US. This could reflect the importance of high technology products for the California economy, based on a mechanism described in the next section. The R2 of equation (2a) is 0.27, a value lower than for equation (1), but still respectable. Overall, the equation indicates a significant net positive effect of international trade on production growth at the California level.
Equation (3a) in Table 1 provides comparable estimates of the trade determinants of California employment growth across 2-digit SIC codes. The right-hand-side variables are exactly the same as in equation (2a). The trade effects on California’s employment growth and its production growth should be similar, since production is the primary determinant of employment. The estimates bear this out. In fact, exports are an even more significant determinant of employment growth, while imports remain a significantly positive factor. The R2 is 0.21, slightly lower than the value for California’s production growth. Overall, the equation indicates a significant net positive effect of international trade on employment growth at the California level.
Results Based on the 3-Digit SIC Code Cross-section Estimates of the effects of international trade based on 3-digit SIC code data are presented in the “b” equations in the bottom half of Table 1 In each equation, we find very low R2 values, even just 0.04 for the California employment equation. This implies, not surprisingly, that factors other than international trade and aggregate demand are the most important determinants of the growth of employment in each specific industry in California.
Nevertheless, the coefficients for international trade and aggregate demand for the 3-digit SIC code data are statistically significant, even more so than for the estimates from the 2-digit SIC code data. Furthermore, the coefficient values for the 3-digit regression coefficients are similar to those estimated from the 2-digit data. Thus, these equations confirms that there is a significant net positive effect of international trade on production growth at both the US and California levels, and on employment growth at the California level.
There are 3 important qualifications to these results:
1. Imports and exports are based only on US data, because a full set of California international trade data are not currently available.