ECONOMIC DEVELOPMENT QUARTERLY / November 2001
Keeping this context in mind, our survey was structured to capture basic information across a number of impact categories, including trade, investment, total sales, and employment. With regard to trade, potential NAFTA effects include cheaper imports of raw materials or intermediate inputs, increased import competition, new export development, or a mix of all three. We initially expected that the relative importance of any specific trade effect would show up in terms of broader aspects of company performance (e.g., sales or employment). Finally, we included an investment variable to assess the extent of NAFTA-related capital out-migration (i.e., outflows of foreign direct investment [FDI] to either Canada or Mexico). This variable was included in light of NAFTA’s liberalized rules regarding FDI within the trade bloc. The question thus remains, Does NAFTA deserve the attention it regularly attracts in local academic, media, and policy circles?
SURVEY METHOD AND EMPIRICAL RESULTS
As a first step toward answering this question, a postal survey of 200 WNY industrial establish- ments was conducted by CUSTAC in August of 1999. The survey was stratified in two ways. First, the region’s 100 largest manufacturing establishments were surveyed (complete coverage). Sec- ond, a random sample of 100 small-to-medium-size enterprises (SMEs) was surveyed. We defined these SMEs as single-plant firms with fewer than 100 employees.
The survey instrument was designed to yield categorical data across the various impact dimen- sions noted earlier. A categorical approach was adopted for several reasons. First, preliminary attempts to obtain ratio or ordinal data proved fruitless, in that pretests revealed that few firms could separate the impact of NAFTA from other factors (e.g., national economic conditions). For instance, telephone interviews with the chief executive officers (CEOs) of five of WNY’s largest manufacturing companies revealed that exports to Mexico had increased appreciably since 1994 but that pre-NAFTA growth rates had also been strong. All five of these CEOs suggested that we rephrase our questions to capture general impressions rather than exact impact figures. Second, our pretests revealed that questions based on ordinal scales were also hard to answer in that many respondents stated that they simply could not rank the impact of NAFTA. For example, several of the SMEs that we talked to noted that they export to and import from Canada on a sporadic basis and that the task of assigning an impact rating was difficult in light of factors such as the relatively weak value of the Canadian dollar in the post-NAFTA period (among other things). On the basis of these pretests, then, our final survey instrument was reduced to a crude device that listed only four response classes across each impact category: positive, negative, no impact, and impossible to tell.
. . . our survey results are suggestive rather than conclusive.
Of the 200 questionnaires that were mailed, a total of 70 valid returns were received (giving a response rate of 35%). The response rate for the random sample of SMEs was 28%, compared to 42% for the larger establishments. One possible reason for the relatively low response rate for the survey as a whole is that most industrial firms in the WNY area have not been directly affected by NAFTA. Follow-up telephone calls to nonrespondents revealed a consistent set of reasons for nonparticipation in the survey (e.g., “we neither import nor export,” “we have no import competi- tion,” “NAFTA is not relevant to our business,” etc.). Put another way, it would appear that the sur- vey had little saliency to a substantial number of firms. This said, roughly half of the nonrespondents refused to comment on their reasons for nonparticipation. From the outset, then, it should be recognized that our survey results are suggestive rather than conclusive. After all, there are no published data that can be mustered to compare respondents versus nonrespondents in terms of trade or NAFTA-related variables. All that we can say with confidence is that respondents and nonrespondents differed little in terms of size (employment), industry focus (durable vs. nondura- ble goods), and age (number of years in business).
The final sample contained firms distributed across the following industries: fabricated struc- tural metal (11), electronics (9), auto parts (7), food processing (4), chemicals (6), textiles and apparel (8), scientific instruments (7), plastics and ceramics (6), machine tools (5), furniture (4), and paper products (3). This distribution closely matches WNY’s industrial profile in terms of manufacturing employment. Despite our low response rate, the sample at least resembles the big- ger picture in terms of the structure of the manufacturing sector.