Thursday, December 29, 2011

Information Content of Dividends Hypothesis

Do corporate dividend changes predict future profitability and cash flows of firms? According to John Lintner's information content of dividends (ICD) hypothesis, this question should be answered affirmatively.

Statistical Analysis

Different statistical analysis have been carried out since Lintner's first presentation of the ICD hypothesis in 1956. The main findings of these analysis can be summarized as follows:

- There is no statistically significant relationship between both above cited variables. (Watts – Journal of Business 1973, 191 et seq.)
- The ICD hypothesis is not reliable because management tends to increase dividends on the basis of overoptimistic forecasts about future earnings. (De Angelo, Skinner – Journal of Financial Economics 1996, 341 et seq.)
- The ICD hypothesis is only reliable as far as the year before and after a dividend increase is concerned. (Benartzi, Michaely, Thaler – Journal of Finance 1997, 1007 et seq.)
- The validity of the ICD hypothesis depends on firm characteristics such as corporate governance structure (How is the ownership structure shaped? Who are the shareholders and which kind of dividend policy do they expect?) and growth stage (The ICD hypothesis namely applies to low-growth firms because such firms have simply less other investment opportunities.). (Choi, Joo, Park – Accounting and Finance 2011, 869 et seq.)
- Negative dividend changes are somewhat predictive of future earnings decreases whereas positive dividend changes provide less information about future earnings. (Choi, Joo, Park – Accounting and Finance 2011, 869 et seq.)


Reading about the above described ICD hypothesis discussion, I am wondering about 3 main issues:

- First and foremost, can we lead this discussion without taking into account the legal form of the firm, the practical implementation of such legal form, and the way dividends are decided upon by respective organs? For example, many jurisdictions provide for the shareholders’ meeting to decide upon a distribution of dividends. However, the processes of such shareholders' decision making as well as the required quorum might deviate substantially among various jurisdictions. In addition, I would expect a decision and (its motivation) to distribute dividends differentiate considerably in a small company where shareholders and top management are identical as opposed to a multinational corporation where stocks are listed and widespread among individuals.

- What about the applicable tax regime for dividends? For example, a higher dividend policy might be linked to a favorable tax regime rather than the company's past and/or future profitability.

- Finally, is this discussion not an example of a confusion between correlation and causality? Is a firm's future profitability higher because of higher dividends in the past? Isn't it more likely that past dividends got higher because of higher profits? Didn't those higher profits simply mean that the company was doing well in the past and, therefore, continued to do well in the future?