[Note: This section is intended to follow Kelly Bradfield’s examination of the 2002 Trade Promotion Act’s mandate that language supportive of America’s IT sector be included in foreign trade agreements negotiated by the Bush administration.]

Counting Widgets in a Globalized, Digitalized World

The preceding discussion brings forward the well-documented story of how the federal government and especially the Defense Advanced Research Program Agency (DARPA) nurtured the development of the Information Technology sector in the United States.(1) By enlisting the nation’s diplomatic leverage to support of the trading aspirations of American IT companies, the Trade Promotion Authority Act of 2002 has already shown itself to be a vital asset in advancing the interests of U.S. e-commerce.

Determining just how successful these efforts to promote free trade in digital goods and services are is apt to prove exceedingly difficult, however. The Internet, arguably DARPA’s central accomplishment, has become as a powerful enabler of international trade. But its emergence poses some fundamental methodological challenges for international trade economists and especially for those government agencies worldwide which are responsible for monitoring foreign trade and publishing international trade statistics.

An ever-widening range of products and services are being digitized and shipped to market via the Internet, a medium essentially devoid of the customs barriers at which governments have historically collected information on exports and imports. Government statisticians have been struggling to devise ways of quantifying trade in products that not only elude these customary monitoring mechanisms but which frequently defy easy classification. For example, a packaged CD recording of Elgar’s Enigma Variations clearly qualifies as a good, but the same music downloaded from a website to a portable media player is generally regarded as a service.(2) And that is an important distinction because, at least in the United States, the agency which is primarily responsible for collecting data on merchandise trade (the Foreign Trade Division of the Census Bureau) is not the same agency that has the lead role in collecting data on trade in services (the Bureau of Economic Analysis).(3) Moreover, while merchandise trade statistics are principally derived from documents filed in conjunction with the actual movement of goods across our borders, service export data are derived largely from periodic surveys of companies known to be engaged in providing services abroad.

For economists, the growing volume of business being conducted via the Internet raises serious questions about the very concept of foreign trade. By providing a frictionless and essentially cost-free means of delivery, the Internet forces trade analysts to rethink the relevance of distance and time to the concept of comparative advantage. There are important political implications as well. Unless the means are found to track the more ethereal forms of commerce taking place via the Internet and to produce accurate, detailed statistics on a timely basis, one result could be to further sever the ongoing national debate over the virtues of free trade and globalization from any empirical grounding.

The Rise of Digital Trade

The emergence of the Internet has not merely revolutionized communications; it has enabled the rapid evolution of a global economy. As broadband networks continue to be extended into virtually every populated corner of the globe, companies will have superior tools to organize their multilateral commercial activities, coordinate intricate business relationships, and manage complex supply chains.

At the same time, the Internet is fast becoming the preferred means of delivering digitized products and services to customers and clients throughout the world. The Internet has thus joined steamships, railroads, aircraft, trucks, and pipelines as a primary mode of transport. Apple Computer’s financial reports even allude to “an evolving digital lifestyle” in which the Internet permits the integration of a growing array of digital devices such as Apple’s iPods and iPhones, along with personal computers, digital cameras, televisions, PDAs, game-playing devices.(4) It is a notion well-understood if not completely appreciated by anyone who has tracked the pitch-by-pitch progress of a World Series game on a BlackBerry while riding a subway in Shanghai. A

s a mode of transport, the Internet’s advantages are manifest. It allows goods and services to be traded in a manner that is essentially frictionless, with few – if any – of the costs, logistical impediments, shipping delays, and adverse environmental consequences typically associated with more conventional methods of bringing products to market. Not surprisingly, musical recordings, films and videos, books, newspapers and magazines, databases, and reference materials have all migrated to the Internet. More than that, the Internet is fast becoming the preferred repository of data once stored on business and personal computers. One result is that unimaginably vast quantities of data have become mobile, accessible from multiple locations through a multiplicity of electronic devices.

Although an increasingly wide array of products is being digitized, this chapter is principally concerned with crossborder trade in computer software. America’s software and information industries play a vital role in the nation’s economy and even greater role in the economies of several states and regions. According to a 2008 study conducted by Content First, LLC for the Software and Information Industry Association, these industries grew more than three times faster than the overall U.S. economy in 2005, with growth of 10.8 percent compared with 3.2 percent for U.S. Gross Domestic Product. In the preceding year, these industries grew 11.1 percent compared with 3.9 percent for GDP. Revenues generated by the nation’s software and information industries reached $564 billion by 2005, up by more than 10 percent since the beginning of the decade.(5)

According to the SIIA, the U.S. software and information industries employed more than 2.7 million Americans in 2006.(6) Net employment by these industries grew by 17 percent between 1997 and 2006, adding more than 400,000 jobs. The Bureau of Labor Statistics predicts there will be more than two million job openings in software and information occupations between 2006 and 2016.5 For example, the demand for computer software engineers, just one of the key occupations in these industries, will increase by 450,000, with the total number of jobs in this occupation reaching nearly 1.2 million by 2016.

In California alone, there were more than 310,000 people employed by various segments of the software industry in 2004, according to the California Employment Development Department. By 2014, the number of jobs in the software companies is expected to swell by nearly 115,000.(7)

Software Trade Data Published by the U.S. Government

The U.S. Government publishes statistics on software trade based on information collected by the Census Bureau and the Bureau of Economic Analysis (BEA).,i.(8) The Census Bureau’s software export data are derived from two sources: Service Annual Surveys and Shippers Export Declarations (which, by law, must be filed in conjunction with any export shipment valued in excess of $2,500). For its part, BEA conducts Quarterly Surveys of Transactions in Selected Services and Intangible Assets with Foreign Persons, which specifically requests information on receipts and payments for general-use computer software. BEA’s survey data are published periodically in the Survey of Current Business. Although the quarterly surveys also cover receipts and payments for customized software, the requested information may be reported in one of a number of service categories.

The Census Bureau’s Service Annual Survey (SAS) covers Software Publishers,(9) which are businesses engaged in the production and distribution of software including software design, installation, and support as well as royalties and license fees for intellectual property rights. The SAS includes revenue from the sale of personal, business, or mainframe computer software to a customer or a client located outside of the United States. It also includes services performed for unaffiliated and affiliated foreign firms (i.e., foreign parent firms, subsidiaries, branches, etc.); software as a service and packaged software exported on media (goods transactions).

However, the SAS does not include revenues from other services, such as custom application design and development, IT technical consulting services, re-sale of computer hardware and software, and IT-related training services. The SAS also excludes services provided to domestic subsidiaries of foreign firms; custom software services purchased by foreign residents, but performed in the United States. The SAS does not breakout exports for major countries of destination, and it does not cover imports.

SAS Software Publishers data double count to some extent BEA’s data, but are not directly comparable. The SAS classifies transactions according to the primary industry of the firm providing the service (or good), not by type of service (or good) as BEA does.

By contrast, BEA classifies services by type of service according to the International Monetary Fund’s balance-of–payments guidelines, which do not conform to the North American Industrial Classification System (NAICS) classifications and definitions used by Census. Affiliated general use software transactions are included in BEA’s aggregate data on affiliated royalty and license fee transactions, but the general-use software transactions cannot be separately identified. Unaffiliated general-use software transactions include receipts and payments for rights to distribute general-use software, and rights to reproduce or use general-use computer software that was electronically transmitted or made from a master copy. Included are licensing fees for reproducing copies of general use software for local area network computer systems.

Electronically transmitted software includes purchased software accessed or downloaded from the Internet. The BEA data exclude the value of pre-packaged general-use software that is not intended for use on a server in a LAN environment that was physically shipped to or from the United States and included in merchandise trade statistics. BEA also excludes fees for custom software and programming services as well as the compensation earned by individual U.S. residents who work abroad for less than one year writing software for foreign residents, and who are paid by foreign residents.

BEA reports software sales to foreign persons by U.S. MNCs through their affiliates. It also reports software sales to U.S. persons by foreign MNCs through their affiliates. These data are classified by the primary industry (software publishers) of the affiliate rather than by type of service. As a result, these sales do not include software sold through affiliates that are classified in several other industries.BEA reports software sales to foreign persons by U.S. MNCs through their affiliates. It also reports software sales to U.S. persons by foreign MNCs through their affiliates.

These data are classified by the primary industry (software publishers) of the affiliate rather than by type of service. As a result, these sales do not include software sold through affiliates that are classified in several other industries. BEA data do not include the value of custom software (customized technical data) that is exported and imported on media since the shipper is only required to report the value of the media. Nor does it include the value of software embedded or pre-loaded on hardware that is exported and imported. The Census/ITC does provide breakouts by major region and countries of destination for exports and imports as well.

Limitations of the Data

BEA’s Quarterly Survey of Transactions in Selected Services and Intangible Assets with Foreign Persons (BE-125) (10) is required only from U.S. firms that had (a) receipts from foreign persons regardless of whether unaffiliated and affiliated in any one of the types of services and intangible assets listed on the survey that exceeded $6,000,000 in the previous fiscal year or that are expected to exceed that amount in the current fiscal year; or (b) payments to unaffiliated and affiliated foreign persons in any of the types of services or intangible assets that exceeded $4,000,000 in the previous fiscal year or that are expected to exceed that amount in the current fiscal year. Needless to say, a $6 million reporting threshold for outbound service transactions can mask a fairly large number of transactions involving numerous smaller software companies dealing in highly specialized or niche markets.

Sampling, of course, presents its own problems. Especially in the case if less mature industries where new start-ups (and flare-outs) are common and where firms merge or acquire others with unusual alacrity, maintaining a statistically satisfactory sampling of the relevant businesses is itself a major challenge. Perhaps most crucially, the need to rely on surveys to measure fast-growing trade in services (not to mention those goods which have assumed the appearance of services) invariably results in appreciable delays in reporting.

For companies exceeding the threshold reporting levels, the BE-125 survey requires information on receipts from transactions involving “rights related to general-use computer software.” However, data on receipts derived from transactions involving customized software programs could conceivably be reported under one of a number of categories including “computer and data processing services,” “management services” or “financial services.” For example, a company providing the customized software needed to coordinate the operations of a large hotel or hospital might justifiably report its receipts under “management services” while a similar company specializing in banking software could reasonably report its receipts under “financial services.” In both cases, however, the service provider is a purveyor of customized computer software. Unfortunately, the BE-125 survey instrument may not reveal this to be the case because a company may elect to identify itself not as being in the “Computer systems design and related services” sector but in some other professional service category.

Similarly, payments and receipts for the use of intangible assets and proprietary rights (such as patents, copyrights, and industrial processes and designs) and, through licensing agreements, the use of produced originals or prototypes (such as manuscripts, computer programs, and sound recordings) appear under Royalties and License Fees in BEA’s periodic reports on U.S. International Transactions in Goods and Services.(11) BEA provides a more detailed set of figures for general-use computer software. However, to protect the confidentiality interests of private companies, the likelihood that data may be suppressed increases with the degree of disaggregation. Customized software is more apt to be reported in ways that are not helpful to an analysis of trade in computer software. As the U.S. Government acknowledges, “the total value of these services may be scattered across several categories of cross-border trade and sales by affiliates.”(12) In addition, computer-related services may be embedded in goods that are exported to foreign markets, or they may be delivered in ways that result in entries in the U.S. international transactions accounts under income rather than under trade in goods and services.

Not surprisingly, the relevant statistics are not lacking in critics. Laments one recent study commissioned by the U.S. Software and Information Industry Association: “reported statistics should be viewed as conservative estimates of total trade in software and information services for the total overseas market. Methods and data for tracking international trade have lagged behind the fast-evolving software and information products and services that are sold in digital (often online), rather than physical, form.”(13)

Surveys may not be entirely accurate or complete because they are subject to the vagaries of respondents’ willingness to provide complete and accurate information. (Anecdotal evidence suggests that companies are frequently reluctant to cooperate enthusiastically with BEA surveys out of a widespread concern that any attention from a government agency is merely a prelude to the imposition of new taxes, regulations or both.) Moreover, to preserve confidentiality, relevant data are occasionally suppressed.

The results of those surveys are invariably published well after the period being surveyed.

Statistical sources do not capture the full scale of the global software and information market. Some sales go unrecorded because of the difficulties described above of tracking trade in intangible products and services. Also, the total value of computer services for foreign consumers may be scattered across several categories of cross-border trade and sales by affiliates. In addition, some computer-related services may be embedded in goods that are exported to foreign markets. The officially reported statistics should, therefore, be viewed as conservative estimates of total trade in software and information services.

(The data collection problem is not limited to the United States. According to one official with EUROSTAT: “Exports and imports of downloaded digital products are not shown separately in the trade in services statistics within the balance of payments framework. As a result, they are not published as such. In principle, they should be classified to relevant service categories, e.g. trade in downloaded software to computer services, music, video etc to audiovisual and related services, financial services associated with e-commerce to financial services etc.”(14) )

The implications go well beyond an academic debate over numbers and the categories into which they should properly fit. Existing methods of monitoring international trade in computer software products have failed to keep pace with constantly evolving business practices. Official data on software exports increasingly capture only portions of the overall circumstances of trade as they are conducted in the contemporary global economy. This under-reporting ultimately distorts to public policy debates over trade policy and the impact of globalization.

To a large extent, those debates are informed by the monthly report on “U.S. International Trade in Goods and Services” which is jointly released by the U.S. Census Bureau’s Foreign Trade Division and the U.S. Bureau of Economic Analysis.(15) The report provides data on the nation’s international trade during the month which had ended approximately five to six weeks prior to each monthly report’s release date. This report instantly becomes grist for newspaper reporters and columnists, television news producers and, increasingly, bloggers. The figures in this report engender headlines, usually about much or how little the nation’s trade deficit has increased. The report’s numbers also influence political debate in Washington as well as the behavior of financial institutions on Wall Street, the decisions of international investors, and the currency interventions of central banks worldwide. It is a closely watched and immensely influential report. Yet it is deficient its failure to capture the full extent of trade involving digitized goods and services conducted through the medium of the Internet.

A still unresolved issue is whether products that were formerly delivered in a tangible format but are now delivered electronically in digitized form should be considered goods, services or some hybrid? The distinctions are important in the regulation of international trade because goods fall under the General Agreement on Tariffs and Trade (GATT) while services are covered by the General Agreement on Trade in Services (GATTS).

Nonetheless, the disaggregated data, both on direct sales through U.S. affiliates abroad and on cross-border trade, indicate a strong and growing market for U.S. software and information products overseas: Global sales generated by overseas affiliates of U.S. software publishers; newspaper, book and database publishers; and firms providing computer system design and related services total $60 billion dollars. An additional $19 billion is generated through cross-border exports of computer and information services, receipts for royalties and license fees for computer software and sales of prepackaged software goods. Yet those figures were for 2004, the latest year for which comprehensive data are available from BEA).

U.S. software publishing industry sales through overseas affiliates totaled almost $11 billion based on preliminary 2005 statistics, the latest year for which data are available for this industry. These sales rose by 33 percent from $8 billion in 2000. The newspaper, periodical, book and database publishing industry segments29 sold nearly $7 billion through overseas affiliates in 2005, up by 24 percent from $5.5 billion in 2000. The computer systems design and related services industry segment30 overseas affiliates’ sales reached an estimated $44 billion in 2004, the last year for which data were released. According to SIIA, the BEA did not provide data for computer system design and related services sales through foreign affiliates in 2005 to avoid disclosure of individual company data.

The American software and information industries also export their products and services directly from the United States. Cross-border exports represent an important and strategic part of overseas trade generated by the industries. Direct exports are composed of cross-border sales of computer and information services, royalties and license fees for general-use computer software and sales of prepackaged, or boxed, software. These direct export sales reached almost $19 billion in 2006, representing an increase of more than 30 percent from $14.3 billion in 2000.

Rather than exporting from the United States, many U.S. software and information firms sell their services directly in foreign markets through their affiliates abroad. In fact, such direct sales represent a substantial share of international trade in software and information. A local presence is often advantageous because an overseas subsidiary is better positioned than the parent company located in the United States to design and distribute software and information services tailored to local market conditions and requirements. In 2004, total overseas sales (all U.S. industries) through affiliates amounted to $483 billion. The U.S. software and information industries represented 13 percent of this total, establishing these industries as important drivers of continued U.S. economic growth.

The American software and information industries also export their products and services directly from the United States. Cross-border exports represent a smaller, but important, part of overseas trade generated by the U.S. software and information industries. These direct export sales reached almost $19 billion in 2006, representing a jump of more than 30 percent from the $14.3 billion in 2000.

Looking Ahead

New business models are evolving that take advantage of a steadily evolving and increasingly robust IT infrastructure. These novel structures will enable higher degrees of business relationships and collaborations that will operate with less and less consideration of the physical distance between participants. Paradigms are shifting from those based on local partnerships or clusters to those which seek to leverage global connections and resources. Increasingly specialized firms offering both traditional and new kinds of services are proliferating as a result of the low-cost or no-cost connectivity offered by broadband Internet access.

Email, instant messaging, voice over Internet protocols, video teleconferencing in high-definition and high-fidelity are all obliterating national boundaries and point toward a time when the traditional metric and accountings will have diminishing relevance in describing economic reality.

According to the Wall Street Journal, “many tech-industry watchers say online software will be the way most businesses prefer to buy software in the future.”(16) With online software, there is nothing to install on a computer. Instead, online software operates much like email accounts with Google or Yahoo, where the operating software and relevant data files are maintained on a distant server, unconnected except through Internet protocols to the user’s computer. Similarly, “cloud computing,” where data is managed and stored in distant servers, is swiftly migrating to the consumer market as computer manufacturers seek to meet growing demand for small portable computing units.(17)

Yet traditional metrics continue to inform political behavior. America’s conversation about international trade and globalization tend very much to be centered on the monthly publication of trade deficits followed by the period release of current account numbers. Yet, as we have seen, those clues about the nation’s economic well-being are becoming more and more attenuated.