Dong Lee, Hyun-Han Shin and René Stulz

paper finds that in the mid 1990's the relationship btw an industry’s q and its funding rate changed. prior to 1996 the "high-funded" industries generally had a higher q than the "low-funded" industries after 1996 that relationship inverted, "high-funded" industries had a lower q then "low-funded" industries.

"....Assuming that a proxy for an industry’s Tobin’s q measures the industry’s growth opportunities, we find that, until the middle of the 1990s, the industries with the highest funding rates were industries with better growth opportunities relative to the lowest funding rate industries, as expected for functionally efficient capital markets. After the middle of the 1990s, capital no longer flows more to the industries with the best growth opportunities. From the middle of the 1990s until 2014, industries with the lowest funding rates have better growth opportunities than the industries with the highest funding rates. We show that this change is entirely due to high q firms that invest less after the middle of the 1990s and use the resulting excess cash flow to repurchase shares. Without these firms, there is a positive correlation between an industry’s funding rate and its growth opportunities throughout the sample period rather than only before the middle of the 1990s.We reach these conclusions using Tobin’s q as a measure of industry growth opportunities, which is defined as the ratio of the market value of the assets of the industry divided by their book value. As we rely on the use of medians and non-parametric tests, however, our conclusions are less sensitive to Tobin’s q mismeasurement than typical parametric approaches. Nevertheless, we confirm that our conclusions are robust using Tobin’s q estimates that account for intangibles and are not an artifact of measurement error in Tobin’s q......The median q of the industries in the top funding rate quintile is 1.33. This median q is not significantly different from the median q of 1.29 for the industries in the bottom funding rate quintile. We expect the greater inflow of resources to be associated with higher capital expenditures and higher asset growth for the most funded industries in comparison to the least funded industries. We find that this is the case. The differences in capital expenditures and asset growth between the two quintiles of industries are large. For capital expenditures, the median ratio of capital expenditures to assets is 66% higher for the most highly funded industries compared to the least funded industries. Asset growth of the most highly funded industries is more than three times the asset growth of the least funded industries......Given the change in regime that appears to take place in the middle of the 1990s, we split our sample into two sub-periods. The first sub-period is from 1971 to the end of 1996 and the second from the end of 1996 to 2014. With this split, we find that in the first sub-period the allocation of capital is exactly what one would expect from the efficient capital allocation hypothesis. During this sub-period, the median Tobin’s q of the industries that are most highly funded is 17% higher than the Tobin’s q of the least funded industries (1.30 versus 1.11) and the difference is significant at the 1% level. A significant q difference emerges between the two groups of industries three years before the year when these industries are classified as most-funded and least-funded. The q difference disappears in year 2 after the selection year. Consequently, the evidence for q is also supportive of the convergence hypothesis before the end of 1996 After 1996, the pattern is opposite. The median Tobin’s q of the industries that are most- funded is 15% lower than the Tobin’s q of the least funded industries (1.37 versus 1.58). The difference is significant at the 5% level and is extremely persistent as it is significant for five years after industries are assigned to the funding rate quintiles. Further, the difference five years after assignment is greater than in the year of assignment, which directly contradicts the convergence hypothesis.....In summary, the relation between an industry’s q and its funding rate changes drastically around the middle of the 1990s. In the 25 sample years before 1996, the high-funded industries (top funding rate quintile) tend to have a higher q than the low-funded industries (bottom funding rate quintile). Further, there is a positive correlation between industry q and the funding rate across all sample industries. After 1996, the opposite occurs: the high-funded industries have a lower q than the low-funded industries and the cross-industry correlation between industry q and the industry funding rate is negative.....With functionally efficient capital markets, we expect capital to flow more to the industries with the best growth opportunities. As a result, these industries should invest more and see their assets grow more relative to industries with the worst growth opportunities. We find that industries that receive more funds have a higher industry Tobin’s q until the mid-1990s, but not since then. Since industries with a higher funding rate grow more, there is a negative correlation not only between an industry’s funding rate and industry q but also between capital expenditures and industry q since the mid-1990s. We show that capital no longer flows more to the industries with the best growth opportunities because, since the middle of the 1990s, firms in high q industries increasingly repurchase shares rather than raise more funding from the capital markets...."

Dong Lee, Hyun-Han Shin and René Stulz, "Why does capital no longer flow more to the industries with the best growth opportunities?," Ohio State University, December 2016, https://papers.ssrn.com/sol3/papers.cfm