Friday, November 12, 2004

Flannery and Rangan: more proof of the trade-off model

Have you had your fill of capital structure papers yet? Of course you have not! You can never have too many! So here is yet another paper on capital structure (with a similar finding) is to be presented at the AFA meetings.

It is by Flannery and Rangan. In the paper entitled Partial Adjustment toward Target Capital Structures they show that firms do have targets for their capital structure and do behave in a manner that attempts to get their leverage ratios back to this target.

The paper begins by describing the pecking order and the trade-off theory of capital structure. As they correctly point out, in its strictest sense, the pecking order theory implies that firms do not have a target capital structure—a conclusion which seemingly flies in the face both surveys of executives (example Graham and Harvey 2001) and mounting evidence.

In the authors’ words:

“In a pecking order world, observed leverage reflects primarily a firm’s historic profitability and investment opportunities. Firms have no strong preference about their leverage ratios, and, a fortiori, no strong inclination to reverse leverage changes caused by financing needs or earnings growth.” (I confess I had to look up a fortiori to be sure ;) )

Flannery and Rangan proceed to investigate changes to firms’ capital structure that result from stock price changes but do so in a world where it is not costless to move back towards the targeted capital structure. (The inclusion of both market price induced changes and transaction costs into their adjustment model is i^3—insightful, interesting and important)

The authors examine an enormous data set of “all firms present in the Compustat Industrial Annual tapes between the years 1965 and 2001.” For each of these firms they winsorize the data (which is largely in the form of ratios) at 1% and 99% to reduce the impact of outliers and data errors. (BTW winsorize just means to set the values in the outer tails equal to the 1st and 99th percentiles. -this was corrected from initial posting)

Using a detailed regression analysis, the paper finds that firms do move back to their previous target. That is under levered firms lever up, while over leveraged firms reduce their leverage. Moreover, when the distance away from this target is studied, firms furthest away from the target move the most back towards it. (see figure 1). This is consistent with a world where there are transaction costs of rebalancing capital structures since for smaller deviations away from the target, it may not be worthwhile to make the costly adjustment.

In their conclusion, the authors summarize the findings:

<>“We find strong evidence that nonfinancial firms identified and pursued a target capital ratio during the last 35 years of the twentieth century. The evidence is equally strong across size classes and time periods. As earlier researchers have found, target debt ratios depend on well-accepted firm characteristics. Firms that are under- or over-leveraged by this measure soon adjust their book debt ratios to offset the observed gap. Unlike some recent studies, we estimate that firms return relatively quickly to their target leverage ratios when they are shocked away. The mean sample firm acts to close its (market) leverage gap at the rate of more than 30% per year. One might dispute whether a 30% annual adjustment speed is “slow” or “rapid”, but it is far from zero for U.S. nonfinancial firms.

Our results strongly support the tradeoff theory of firm capital structure and the relevance of costly (partial) adjustment toward target leverage. Indicators of the pecking order and market timing (à la Baker and Wurgler [2002]) theories carry statistically significant coefficients, but their economic effects are swamped by movements toward a leverage target that reflects firm-specific characteristics."
Very Interesting! Or as I said above, i^3! :)

Alternative sites/cites:

From SSRN: Flannery, Mark Jeffrey and Rangan, Kasturi P., "Partial Adjustment Toward Target Capital Structures" (May 3, 2004).

From the AFA program:

And from Mark Flannery’s web site:

BTW I just emailed both authors and have been told that modified version will be out within a few weeks. I will update the above links.


Anonymous said...

>For each of these firms they winsorize the data (which >is largely in the form of ratios) at 1% and 99% to >reduce the impact of outliers and data errors. (BTW >winsorize just means to drop those firms that are in >the 1st and 99th percentiles.)

Winsorizing is a process of replacing values in the extreme tails with the value of the x'th percentile, in this case 1% and 99%. This is good because you are not throwing away data, and the process leaves the order statistics unaffected, but dampens the efect of outliers.

Anonymous said...

Yes, I was planning to make the same comments about 'winsorize'.

Curiously, who wrote this blog? Since this website is called '', I would suppose this concept should be understood.

PS: I am just a Finance student.