IPO as an Optimal Stopping Time: A Structural Estimation

Abstract

Going public is an important milestone for a firm. There are significant benefits and costs associated with the decision of being listed. On one hand the firm can raise the required capital for investment and firm growth through the IPO, make the firm more visible, transfer the risk to shareholders, and relax the borrowing constraints. It has significant implications for the product market competition too. The amount of cash generated through IPO if invested may also affect the future product market competition and the firm's profitability. On the other hand, going public is associated with significant amount of fixed costs and costs associated being under increased public scrutiny. The IPO process and the uncertainty associated are further complicated by the adverse selection problem associated with a firm taken public. The investors are asymmetrically informed about the future prospects of the firm relative to the management. The informational superiority of the firm leads to the standard lemons problem. The firm has to underprice the issue to give enough incentives to issuer to invest in the IPO. This is an indirect cost and lowers the IPO proceeds. The firm would therefore want to signal about its future prospects through its recent past performances. Hence going public decision and its implications are inherently dynamic in nature. The (management of the) firm thus wants to time the IPO decision well. Many of the associated costs and benefits are hidden. In this paper we formulate a dynamic programming based structural model of the decision to go public and estimate these hidden parameters using data from Indian IPOs. We model the going public decision as an optimal stopping time problem for a firm in a dynamic setting. Each period the management of the utility maximizing firm decides whether to list the firm or not. The source of management's consumption is the profit it generates through its profit maximizing investment decisions for the firm. Each period the management decides (a discrete choice) whether to take the firm public or not. Given its going public decision it chooses the level of investment (continuous choice) and consumes the rest. If the firm decides to go public, the firm makes the static (continuous choice) decision of how much of the firm to sell (dilution) and how much of the IPO proceed to reinvest. Although these decisions are static, its implications are dynamic in nature. The dilution decision is an outcome of the trade off between the IPO proceeds and the claim of the management on the future profits and private benefits of control of the firm. The IPO proceed reinvest decision is an equilibrium outcome of the trade off between the cash out prospect and future profitability improvement (hence improved managerial income) of the firm. We formulate a dynamic programming model and characterize the equilibrium. We outline a dynamic programming based structural estimation procedure to estimate the model parameters. The model and the estimation procedure is applicable for any general IPO process. We take the model to data from the Indian IPOs taken public during the period 1999-2006. We have data for all the IPOs during this period. For each firm we have data three years prior to the firm decides to go public and all the years after it went public. Above that we have data on firms which never went public in the entire sample period. The data set is unique in the sense that we can track the firm before it went public as well as after. We first perform a few descriptive and reduced form analysis to test the basic predictions of our model. For example the structural model equilibrium predicts a monotonic threshold policy of going public decision, monotonic to the underlying state variables. We run standard panel data logistic regression to confirm the predictions. Going public is also expected to improve the profitability of the firm. Two similar types of firms, who would otherwise both like to go public if they can afford the associated costs of going public. We match the sample of going public firms with that of the firm which did not go public. We employ the matching estimation procedure and estimate the average improvement in sales, EBIT etc. due to the treatment of going public decision. The treatment effect predicts significant improvement in profitability in the matched sample. We expect that the firm would like to signal its future profitability and alleviate the adverse selection cost through its recent past performance before going public and advertise the same in the IPO prospectus. The risk factor chapter in the IPO prospectus is one such place. We use a novel technique to read the IPO prospectuses and its risk factor chapter and classify the risks associated with the future prospect of the firm as outlined in the risk factor chapter. This chapter is written by the firm itself and can be seen as an equilibrium response to mitigate the adverse selection costs. We regress the IPO proceeds and the underpricing on different types of risks and find a negative relation to firm specific risks impacting IPO proceeds. Since there is inherent selection and endogeneity bias associated with the going public decision we employ Heckman selection estimation to estimate such regression. The structural estimation brings out the fixed costs associated with the process of going public. This fixed cost includes the underwriting fees as well as opportunity cost associated with the going public decision. The high fixed partly explains why otherwise similar firms do not go public despite its obvious benefits. Other fundamental parameters include the unobserved quality of the projects in hand. The structural estimation procedure helps us conduct counterfactual experiments: for example what if the fixed costs were reduced by an online going public decision and its impact on the going public decisions. What impact does it have on transmission of information etc. To the best of our knowledge this is the first paper which analyzes going public decision from a structural view point and estimates it.