Underpricing in the IPO market with sentiment traders: An experimental study

How to cite

Nakaizumi, Takuya; Kawagoe, Toshiji; Taguchi, Satoshi; Ogawa, Kazuhito; Takahashi, Hidenori. 2024. "Underpricing in the IPO market with sentiment traders: An experimental study." AEA RCT Registry. AEARCTR-0014907. https://doi.org/10.1257/rct.14907-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Takuya Nakaizumi* , Toshiji Kawagoe , Satoshi Taguchi , Kazuhito Ogawa , Hidenori Takahashi

* Primary investigator

Registry ID
AEARCTR-0014907
DOI
https://doi.org/10.1257/rct.14907-1.0
First registered
Last updated
Status
On going
Trial dates
2024-02-17 – 2025-01-12
Countries
Japan (Osaka), Thailand (Bangkok)
JEL codes
G40
Secondary IDs
JSPS KAKENHI Grant Numbers JP20K01643, 19K01562,21H04979
Registry record
http://www.socialscienceregistry.org/trials/14907

Abstract

In this study, we experimentally investigate the cause of underpricing and underperformance in the IPO market. Two well-known anomalies of IPOs are high first-day returns and subsequent long-term underperformance. To address this issue, we investigate the IPO market model of Ljungqvist et al. (2006) in the laboratory. In this model, a fraction of sentiment traders enter the market in each period. They irrationally believe that the IPO price is higher than the fundamental value. Other traders forecast IPO prices rationally, i.e., they know the probability P that the IPO price will fall to the fundamental value. The issuing company and an underwriter take the behavior of these traders into account. The underwriter decides in each period how many shares to sell in the IPO market. Ljungqvist et al. (2006) show that underpricing and underperformance occur even when the underwriter has no bargaining power and the firm can set the price as high as the IPO market price under a certain condition. In our pilot experiment, the firm exogenously sells the stock to subjects playing the role of the underwriter. The behavior of rational and sentimental investors is exogenously determined by the computer according to the theory. We have two treatment variables, the probability P that the IPO price will fall to the fundamental value and the number of initial shares Q. The higher P, the more likely it seems that underpricing will occur. However, this is not the case in our pilot experiment. Rather, underpricing is more likely to occur when P is large. The higher P is, the more risk-averse subjects (underwriters) might be expected to sell the shares in earlier periods. However, the subjects (underwriters) are more likely to sell the stock in later periods when P is large in order to keep the price high and preserve enough revenue in the earlier period. We found that the observed behavior maximizes the subjects'(underwriter's) profit.

Intervention

Ljungqvist, et al [2006] explore the model including sentiment investors and derive both underpricing and underperformance results of IPO pricing endogenously. In their model, there are some sentiment investors who have excessive expectations and buy the stocks at a higher price. This initially causes underpricing, and then leads to further declines as they revise their naive estimates, which leads to underperformance.

At first glance, the underpricing results may be theoretically obvious because sentiment investors buy stocks at higher prices than their fundamental value. But their important theoretical contribution is that underpricing still occurs even if the underwriter has no bargaining power, which means that the IPO price could be equal to the stock price, which is not underpricing, even if sentiment investors buy at a higher price than the fundamental value. Although sentiment investors tend to buy the stock more than the fundamental price, and the price could be higher than the fundamental value, such an expectation could change when the bubble crashes, and the price could go down to the fundamental value thereafter. Considering the possibility of bubble crash, the IPO company must set the IPO price equal to the expected value, which could be lower than the stock price.

If there is no crash, the IPO company can sell the stock at the same price as the market price and there is no underpricing. If crashes occur, on the other hand, the IPO company must sell the stock at the expected value, which is lower than the initial prices, so underpricing occurs.

This is the main logic of Ljungqvist, et al [2006] in which underpricing occurs or not. And it is mainly based on the amount of initial shares that the IPO company sells to the underwriters at the beginning. If the amount of initial stock is smaller than the demand of sentiment investors of the market, there is no uncertainty and underpricing could not occur. If the amount of initial stock is larger, on the other hand, the underwriter should sell the remaining stock at the second stage of the market, even if the bubble could crash and the price could go down to the fundamental value, because selling all the amount at the first stage will make rational investors buy stock and the price will go down drastically. So the underwriter has to wait for the remaining shares to be sold in the next phase. Then underpricing occurs.

We verify the results through experiments. In the experiment, IPO company is exogenous, and we hire the subject assigning the role of underwriter who sells the stock in the market. As Ljungqvist, et al [2006] assumed, the subject has two chances to sell the stock. Namely, the market opens twice and they can sell the stock twice. The investors, whether sentimental or rational, are price takers and exogenous. And the price is set automatically.

From their proposition, if initial stock amount is smaller, underprice does not occur, while large, it occurs. And we use two conditions: small amount and large. In addition, the probability of crash is considered important. So, other condition is higher probability or lower probability that bubble will continue. Thus, we introduce 4 conditions. In the experiment, both IPO company and investor are assumed to be exogenous. And we hire the subjects who assign the role of underwriter. In the experiment,

Timeline of Ljungqvist, et al [2006] model and our experiment is as follows
1. The IPO company (exogenous) offers the number of shares to be issued and sells them to the underwriter (endogenous) at the price equal to the expected average price of the market.
2. Underwriter (subject) sells shares in the open market (open for two periods) in order to maximize their profit. The price of the market at the first stage as well as the second state is determined by the total initial stock and crash probability,
3. After selling the stock, the underwriters' profit is revealed and they get the reward.

Experimental design

First we gather the subjects mainly from University students, both Kansai University and King Mongkut's Institute of Technology Ladkrabang University based on the procedure admitted by research ethics committee. After arriving at the experiment venue is as follows.
1. Explanation and collection of the consent form and withdrawal of consent form.
2. Explanation of the experiment (about 20 minutes) , practice and confirmation test using Ztree to enter the values specified by the experimenter and check the screen display (you cannot proceed until you get the correct answer, about 15 minutes)
3. Main experiment (about 60 minutes) (Note that Ztree does not support Thai, so the experiment is conducted by displaying the English screen and the Thai PDF file while Ztree support Japanese, so there are no additional documents in Japan.)
4. Completing the receipt and paying the honorarium
5. Dissolution
The experiment supervisor will strictly manage the personal information obtained, and after 10 years of publication, all documents and other materials will be destroyed.

Design details

Randomization method
we gather the subjects that are ordered randomly.
Randomization unit
both Kansai University in Japan and King Mongkut's Institute of Technology Ladkrabang in Bangkok.
Clusters
4 times 3, total 12 clusters each clusters contain 25-45 subjects.
Observations
12 times around 40 subjects means around 500 subjects, mainly university stundets of both Kansai University in Japan and King Mongkut's Institute of Technology Ladkrabang in Bangkok.
Arms
500 students.
Primary outcomes
Four conditions are outlined for the experiments: low and high quantities of shares, paired with low and high probabilities of continuing market bubbles.

Cited by

No citation data available for this registration yet.

Keywords