In the following, we intend to explain how we have constructed independent variables to represent the hypotheses for the cross-sectional analysis. The hypotheses represent three distinct themes to set the scope for our analysis, namely 1) lockup characteristics, 2) IPO characteristics, and 3) lockup period characteristics.
When combined, these will ensure an adequate assessment of the variation in the dependent variable, namely the CAR.
4.5.1 Independent variables on lockup characteristics
Hypothesis 2A addresses the lockup duration for each IPO in the sample. Bradley et al. (2001) and Field and Hanka (2001) treat this variable by grouping their observations according to lockup durations that are either less than (or equal to) 180 calendar days, or longer than 180 calendar days. Due to the more heterogeneous nature of Nordic lockups, we assume that Nordic lockup durations may deviate from the standardized structure of either 180 or 360 days. Therefore, to capture the full effect of this potential variation, we suggest a discrete measurement of calendar days as the best suited approach for lockup duration.
40 Hypothesis 2B concerns the portion of locked up shares at the time of IPO. To establish a variable that is comparable among all firms, we use a relative measurement of locked up shares as a percentage of total shares offered to the public in connection with the IPO (i.e. the “free-float”). The free-float represents the sum of newly issued shares and existing shares offered. We exclude the impact of whether an over-allotment option is fully or partly exercised, as this potential effect concerns the post-IPO period and is not known for certain at the time of IPO.
Hypothesis 2C considers whether a lockup provision restricts a PEVC-shareholder. We employ a dummy variable to distinguish between PEVC-restrictive lockups and those that do not restrict PEVC-shareholders, by assigning the value of 1 to the former and 0 to the latter. This approach is in line with that of Brau et al.
(2004), Field and Hanka (2001) and Hoque (2011). We have not distinguished between PE- and VC-shareholders since there exists no common definition for the specific stages of financing. This logic is emphasized by BVCA and PWC (2014), with the focal notion being that venture capital itself can be an investment strategy within the scope of PE. Furthermore, we are aware that PEVC-restrictive lockups may also include other shareholders that are non-PEVC. Such instances can potentially induce ambiguous effects, however, it cannot be avoided due to the fundamental structure of the lockup. We will return to this issue when discussing the output of the analysis.
4.5.2 Independent variables on IPO characteristics
Hypothesis 3A concerns the first-day return of an IPO, for which there exist varying approaches for constructing an optimal measurement. A noteworthy methodology is employed by Tolia and Yip (2003) who apply the concept of “Hot vs. Cold” IPOs, where an IPO is defined as either “Cold”, “Cool”, “Hot” or “Extra Hot”, dependent on the magnitude of its first-day return. However, the bounds in the “Hot vs. Cold” approach are determined subjectively and do not follow a common framework. Therefore, we will disregard the categorical variable framework of Tolia and Yip (2003), and instead construct the variable as a realized percentage return. This return will be based on the adjusted closing price on the first trading day relative to the offer price, for which we collect data from S&P Capital IQ.
Hypothesis 3B addresses the role of underwriter reputation. There is no common method for measuring the reputation of an underwriter, as it can be based on different characteristics such as size, regional focus, or experience (either in terms of volume or value). Yung and Zender (2010) based their assessment of underwriter reputation on a publication by Carter and Manaster (1990), in which they rank highly reputable underwriters as those with low variation in type of issued firms. In line with Yung and Zender (2010), we will also measure underwriter reputation according to rankings. However, we will instead base our rankings on value, in order to reflect each underwriter’s overall involvement with IPOs. These rankings are collected as league tables from
41 Thomson One’s database and are based on total IPO proceeds, with full credit to each underwriter (see Appendix 1 for league table criteria). For each year in the sample period, the highest ranked underwriter will be assigned a value of 1, the second ranked underwriter will be assigned a value of 2, and so forth. Underwriters that are not in the top-25 league table for a given year are assigned a value of 26. We will apply rolling calendar-year windows and consider an underwriter’s rank value from the calendar-year prior to the IPO, so it reflects the concurrent market sentiment. Furthermore, if multiple underwriters are affiliated with an IPO, we will consider the average rank of all involved underwriters. In this manner, an IPO with multiple high-ranked underwriters will be better positioned than those that include less reputable underwriters.
4.5.3 Independent variables on lockup period characteristics
For the independent variables on lockup period characteristics, we shift focus to the post-IPO period up until lockup expiration (i.e. the lockup period).
Hypothesis 4A considers the volatility of each stock price during the lockup period. For this variable, Doran et al. (2014) constructed an estimate based on idiosyncratic volatility, defined as the standard deviation of a stock’s daily residual returns according to the Fama and French three-factor model (Fama and French, 1993).
However, we suggest that one extends this measurement to also include the systematic risk for a stock. This ensures representation of the total uncertainty that is related to both firm-specific risk and overall market uncertainty for a given period. The independent variable on volatility is therefore defined as the standard deviation of a stock price within the estimation window (-60, -5). By using the defined estimation window, one ensures complete comparability with CAR, which is based on beta- and alpha-estimates for the same estimation window. Our approach for measuring volatility is in line with the methodology of Harper, Johnston, and Madura (2005) and Nowak (2015) and is based on stock prices that are attained from S&P Capital IQ.
Hypothesis 4B addresses purchases and sales made by insiders prior to lockup expiration. As previously mentioned, we have collected data from FactSet as it has been deemed the most comprehensive and streamlined provider of such information. The data represents the total insider position at the end of each month. The measurement of lockup period insider trading is thus constructed by the percentage change from the first end-of-month insider position post-IPO, to the last end-end-of-month insider position before lockup expiration26. Our estimate of insider trades includes all registered insiders (i.e. also those that are not restricted by a specific lockup provision), as the focal notion is that any insider trades prior to lockup expiration provides signals to the market regarding the true value of the firm.
26 Due to data availability, this measurement has not been based on the estimation window.
42 Lastly, Hypothesis 4C concerns a stock’s price-run leading up to lockup expiration. Since a price-run reflects a series of consecutive price movements in the same direction, we construct this variable by calculating a stock’s cumulative daily stock return during the estimation window (-60, -5). For example, a hypothetical stock price that increases by equal increments from 10 (𝑡 = −60) to 20 (𝑡 = −6), and then drops to 10 on the final day in the estimation window (𝑡 = −5), would yield an actual return of 0% and a cumulative return of approx.
20%. Hence, the cumulative return allows us to account for series of consecutive price movements and thereby attain an optimal representation of a stock’s price-run.