PUBLIC INFRASTRUCTURE CAPITAL AND ECONOMIC ACTIVITY-CROSS-SECTIONAL EVIDENCE FROM CANADIAN PROVINCES

In the last decade there has been extensive interest in statistical attempts to develop quantitative measures of the linkage between public infrastructure capital and economic activity. These empirical analyses date largely from the initial work by Aschauer (1989a, 1989b, 1989c). Aschauer looked at time-series data for the U.S., as well as data from across countries, and showed a striking correlation between the growth of economic activity and the size of public infrastructure capital, and this seemed to be correlated with the slowdown in national productivity growth. Aschauer's results indicated a very high marginal productivity of public infrastructure capital, thereby suggesting that an expansion of public infrastructure would significantly improve economic performance. A number of authors have re-examined or further explored this link between infrastructure spending and economic activity. Some sought to refine the analysis and separate the effects of different types of public infrastructure, such as highway investments compared to others. Other authors employed different databases, different industries and different countries' data. Some authors were able to show similar results to Aschauer. But other authors found the results to be questionable, particularly as more sophisticated econometric analysis was undertaken. This study follows the approach of Munnell (1990b) and Eisner (1991), Evans and Karras (1994), Holtz-Eakin (1994) and Moomaw, Mullen and Williams (1995) in using combined cross-section and time series aggregate regional data to test the relationship. Data are for the 10 Canadian provinces, from 1961-1997. An aggregate production function is postulated relating provincial GDP to measures of labor and private capital inputs, along with various measures of public infrastructure capital (the complete results are in Waters, 1999). Simple regression analysis produces results similar to those of Aschauer and Munnell, i.e., that public infrastructure capital appears to be significant and the marginal product of public capital is larger than that of private capital. The paper explores the implications of various econometric specifications including corrections for serial correlation, cross-province correlations, incorporating fixed province- and time-effects and province-specific time trends to avoid non-stationarity problems. The analysis shows that estimates of a link between various measures of public infrastructure and GDP are quite sensitive to the specific formulation chosen.

Language

  • English

Media Info

  • Features: References; Tables;
  • Pagination: p. 29-47

Subject/Index Terms

Filing Info

  • Accession Number: 00804770
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jan 10 2001 12:00AM