PROBLEMS OF IDENTIFYING AND MEASURING INTERMEDIATE (PRODUCER) SERVICES IN THE COMPILATION AND USE OF INPUT‐OUTPUT TABLES*
论文详细讨论了国家投入产出表编制中中间服务的统计问题，如公司-机构统计差异、总部和研发支出处理，并提出基于微观产业组织研究的公司-机构数据协调方法，利用加拿大官方数据验证可行性。
The paper is mainly concerned with statistical problems relating to intermediate services that arise in the construction of national input‐output (I–O) tables. Though these problems are sometimes discussed in the literature, their precise nature is usually not spelled out in any detail and this is done in the paper. The problems are closely related to the company‐establishment statistical dichotomy permeating the ultimate sources and allocation of intermediate services. Important examples can be found regarding the statistical treatment of head offices, research and development expenditures, and international trade of intermediate services. Presently used procedures for Canadian and U.S. I–O compilation show evidence of statistical inconsistencies and lack an appropriate framework to utilize full information. The paper suggests a possible approach for reconciling company and establishment data based on industrial organization linkage studies at the microlevel. Considerable empirical support is offered, using various official Canadian statistical publications, to show that the suggested approach is both feasible and has desirable properties. The paper goes on to argue that the contemporary information technology revolution has profound implications for I–O compilation and use with special reference to intermediate services. Four major implications are explained in the context of the growing microelectronics technological change and related literature. Some basic suggestions are put forward with regard to joint‐cost allocation and inter temporal comparisons problems with respect to I–O compilation. It also appears that some fundamental rethinking of commonly accepted standard industrial classification conventions may be called for in the near future if I–O tables are to remain relevant and viable. The paper thus features a somewhat broader view of I–O statistical problems than usual and attempts to show that this view is potentially appropriate to questions of economic policy formulation.