More on Finance and Growth: More Finance, More Growth?

Abstract

T en years ago the literature on the relation between finance and growth was set on its modern course by the publication of King and Levine’s (1993) influential paper. Much of the following work in this area was done by Ross Levine and his coauthors. Thus, none better than Levine himself could summarize the progress over the past decade, in the struggle to move from a correlation between financial development and economic development (Goldsmith, 1969) to establishing a causal relation between finance and growth. Levine emphasizes advances along two dimensions. First, in the measures of financial development. Goldsmith (1969) relied on the ratio of the value of financial intermediary assets to gross domestic product (GDP) as his only measure of financial development. Levine and coauthors have used many different variables, e.g., the liquid liability to GDP ratio, the credit in the private sector to GDP ratio, and the level of stock market turnover. Rajan and Zingales (1998) have even used the quality of accounting standards as a measure of a firm’s ability to raise funds. Nevertheless, as I will discuss momentarily, this first area is probably where less progress has been made. The second and more important dimension emphasized by Levine’s survey is in the attempt to establish causality. This is the area where most innovations have taken place. Their first step was to use the time dimension to identify the cause-effect relation (King and Levine, 1993), relying on the old “post hoc ergo propter hoc” argument. Levine and coauthors have subsequently enriched this approach using dynamic panel estimation, and further progress has been made in the use of instrumental variables (Rajan and Zingales, 1998, and Levine, 1998 and 1999). In both cases they use the La Porta et al. (1998) measures of legal origin as instrumental variables. I will discuss later whether and when these can be considered good instruments. A third step in trying to establish causality, which is not adequately surveyed by Levine, is the “natural experiment” approach. In a very clever paper, Jayaratne and Strahan (1996) use the banking deregulation across U.S. states as an exogenous change in financial development. This omission, justified on the basis of a decision not to focus on within-country studies, is the only shortcoming in Levine’s survey. Personally, I trust much more the natural experiment approach than the more sophisticated, but less robust, dynamic panel estimation techniques. The final step in the quest for a causal link, amply summarized by Levine, is to look in more detail at the mechanism through which finance spurs growth (see, e.g., Rajan and Zingales, 1998, and Demirguc-Kunt and Maksimovic, 1998). In spite of this minor quibble, Levine’s survey does an excellent job of summarizing the progress made in the past decade. In 1993 many people doubted that there was a relation between finance and growth; now very few do. Since Levine has documented so well what has been done, my role as a discussant is to describe what remains to be done. I will focus, thus, on the weak links in the quest for a reliable relation between finance and growth that policymakers can use in their decisions. I focus on six such weak links.