Oil Price Uncertainty and IPOs

We examine the impact of oil price uncertainty on IPO volume in the oil and gas sector. By using the implied volatility of oil options, a forward-looking uncertainty measure, we identify the effect of uncertainty on the going-public decision. Oil price uncertainty exhibits a strong negative relation to IPO volume. A one standard deviation decrease in the implied volatility results in a 29% increase in the number of quarterly IPOs. The effect is concentrated among the price-sensitive upstream producers. We further report that uncertainty positively impacts the IPO withdrawal decision and increases the value of postponing the offering.


INTRODUCTION
Oil price uncertainty can affect both firms' investment and financing decisions.The impact of oil price uncertainty on investments in the oil and gas sector is well documented (Kellogg, 2014;Elder, 2019;Dossani and Elder, 2022).The general consensus is that elevated uncertainty causes firms to defer investments in anticipation of better market conditions.Surprisingly, the relationship between oil price uncertainty and the financing of oil and gas firms has been neglected in the literature.To fill this gap, we study the impact of oil price uncertainty on the most important financing event in the firm's life-cycle-the initial public offering (IPO).Through the IPO firms raise substantial amounts of funds that have real effects on investment, employment and growth in the oil and gas sector.Therefore, it is important to understand how and why firms change their IPO decisions in response to oil price uncertainty.
Studying the oil and gas sector has several advantages over examining an aggregated market-wide sample.First, oil and gas firms' discount rates are directly linked to oil price uncertainty (Christoffersen and Pan, 2018), where especially upstream producers' cash flows are highly sensitive to oil price changes (Doshi et al., 2018).Second, we are able to construct a forward-looking measure of uncertainty by using the implied volatility of oil options.Being derived from options prices, implied volatility reflects the forward-looking price uncertainty assessments of the heterogeneous agents trading in futures markets (Singleton, 2014).Third, based on prior work [see, e.g., Kel-logg (2014); Elder (2019); Dossani and Elder (2022)], we can assume that our oil price uncertainty measure is, to a large extent, exogenously determined.This setting allows us to clearly identify the role of uncertainty in the IPO process.Christoffersen and Pan (2018) offer a direct channel for oil price uncertainty to affect the cost of capital.They show that oil price uncertainty leads to a higher risk premium in the equity markets.Higher discount rates negatively impact valuations, and the lower valuations discourage firms to conduct an IPO.Due to the high sensitivity of oil and gas sector firms to the risk premium associated with oil, the negative relation between uncertainty and IPO volume should be more prominent in this sector.Real options theory offers a second explanation for why IPO volume is negatively related to uncertainty.In a real options framework, the option value of waiting increases with uncertainty as long as the investment is irreversible to some extent (Bernanke, 1983;Brennan and Schwartz, 1985;Majd and Pindyck, 1987;Pindyck, 1991;Davis and Cairns, 2017;Elder, 2019).This also holds true for an IPO (Pastor and Veronesi, 2005), the decision to go public is partially reversible until the issue date.After the company is taken public it is difficult to reverse the decision.Hence, IPO volume decreases with uncertainty through the cost of capital and real options channels.Therefore, our main Hypothesis (1) states that oil price uncertainty negatively impacts IPO volume.
Our main tests explore how implied oil volatility impacts the time-series variation of IPO volume.We measure IPO volume both in terms of the number of IPOs and the total proceeds raised.We find strong support for our main hypothesis by using a sample of 450 completed oil and gas IPOs during the time-period 1/1/1987-12/31/2019.A one standard deviation decrease in implied volatility corresponds to 29% (22%) increase in the number of IPOs (proceeds).Since oil price uncertainty can be linked to other macro-economic factors driving the going-public decision, we further study oil and gas IPO volume in relation to market-wide IPO volume.This measure allows us to identify the oil and gas-specific variation in IPO volume.Again, we find strong support for our main hypothesis.Oil and gas sector IPOs are negatively affected by oil price uncertainty.We create placebo tests exploring how our parsimonious model captures IPO volume in the less oil price-sensitive Hi-tech sector.The placebo tests show that oil price uncertainty is not a direct driver of relative IPO volume among Hi-tech firms.Since the beginning of our sample period, the oil industry has undergone significant changes.The fracking revolution in the post-2012 era nearly doubled the U.S. oil production (Gilje et al., 2016), which increased the number of oil and gas IPOs.Therefore, we explore the impact of oil price uncertainty on oil and gas IPOs before and after the fracking revolution and report similar findings for the two time periods.
To further identify the effect of oil price uncertainty on IPO volume, we study upstream firms in isolation.Upstream firms differ from the others in the oil and gas sector due to their sensitivity to crude oil prices (Kumar and Rabinovitch, 2013).Upstream producers engage in exploration and production, whereas downstream refiners focus on refining and marketing.Upstream producers sell their output on the physical market; their cash flows are highly sensitive to the underlying commodity price.In contrast, other firms, such as refiners, see different dynamics since they can take advantage of the crack spread, the difference between the prices of refined products and the cost of crude oil as an input (Suenaga and Smith, 2011).This built-in margin allows downstream firms to transfer part of the price variation to their customers.The economic impact of crude oil price uncertainty on IPO activity differs between upstream and other oil and gas sector firms.Therefore, Hypothesis (2) states that upstream firms' IPO volume should be more adversely affected by the level of uncertainty.
To test Hypothesis (2) we split our sample into upstream and other (midstream and downstream) oil and gas sector firms, exploring the effect of uncertainty on IPO volume in the two groups.
Our findings lend support to Hypothesis (2), upstream firms are highly sensitive while other firms IPO activity is unrelated to oil price uncertainty.Our findings are not surprising.Upstream firms' cash flows are theoretically more linked to oil price fluctuations, as their discount rates are highly sensitive to the risk premium associated with oil.The elevated discount rates and the resulting lower valuations make upstream firms refrain from going-public during times of high uncertainty.The same applies to the real options channel, where upstream firms' higher sensitivity to oil price uncertainty increases the value of deferring the IPO.
The effect of oil price uncertainty on IPOs does not only affect the IPO at the pre-filing stage but also after the filing of the IPO.Elevated uncertainty during this stage of the IPO process can cause a firm to either choose to postpone the issue or reverse it completely.Firms may prefer to postpone their IPO in anticipation of more favorable market conditions (Pastor and Veronesi, 2005).This mechanism leads us to develop Hypothesis (3a), stating that the time between the filing and IPO date increases with uncertainty.Two not mutually exclusive reasons can cause a firm's complete IPO withdrawal.First, through a DCF framework, increased uncertainty during the IPO process lowers valuation through increased cost of equity capital.This makes the IPO less favorable for the incumbent shareholders ultimately leading to a complete withdrawal.Second, Busaba et al. (2001) and Busaba (2006) argue that the IPO filing is coupled with a valuable option.At the time of the IPO filing, the issuer does not know the exact market valuation, so the offer price is uncertain.The firm will exercise this option if the market valuation exceeds the incumbent shareholders reservation value of the firm.When the market valuation is below the reservation value, the firm will withdraw its offering, and the option expires worthless.Hence, Hypothesis (3b) states that firms are more likely to reverse their IPO decision following increased oil price uncertainty.
To test Hypothesis (3a), we study the impact of oil price uncertainty on the number of days between the filing and the issue date of the IPO (window length).The findings support Hypothesis (3a), window length is positively linked to oil price uncertainty.This suggests that uncertainty also increases the value of the option to postpone the issue.Managers respond to elevated uncertainty by deferring the listing in anticipation of better market conditions.To test Hypothesis (3b), we explore the impact of oil price uncertainty on IPO withdrawals.Our findings reveal that uncertainty increases the likelihood of withdrawals in the oil and gas sector.
In sum we show that oil price uncertainty has a real impact on the going-public decision.However, the IPO decision is affected by several factors such as stock market, political and macro-economic uncertainty as well as the level of the WTI and debt market conditions.We effectively control for all these factors and still report a strong negative link between the oil price uncertainty and IPO volume.Even though we control for all these factors, omitted variables can cause biases.We test for the severity of a potential omitted variable bias (OVB) by using the Oster (2019) test, and report that our results are not severely affected by OVB To the best of our knowledge, we are the first to explore the role of oil price uncertainty on financing decisions.We contribute to two strands of literature.First, we contribute to the vast literature on how oil price uncertainty affects corporate decision-making.The effect of oil price uncertainty in a macro-economic context is either focused on oil price uncertainty as an output predictor (Aye et al, 2014;Jo, 2014;Gao et al, 2022;Rahman and Serletis, 2011) or aggregated investments (Elder and Serletis, 2010).Moreover, a vast amount of literature examines how oil price uncertainty affects firm-level investment decisions.The empirical literature focuses on the impact of oil price uncertainty on firm-level capital expenditure (Henriques and Sadorsky, 2011;Phan et al, 2019;Maghyereh and Abdoh, 2020), and reports a negative relation.Kettunen et al. (2011) and Chronopoulos et al. (2016) theoretically explore the link in real options settings.Kellogg (2014), Doshi et al, (2018), Dossani and Elder (2022) instead solely focus on investments among upstream firms.Several studies examine the impact of oil price uncertainty on M&A activity (Fan, 2000;Barrows et al, 2020).The consensus is that oil price uncertainty negatively affects both investment and output.Differing from all the above studies, our focus is on the firm-level financing decisions instead of output and investments.
Second, we contribute to the broader literature on how uncertainty affects the going-public decision.Prior work focuses on the effect of macro-economic (Lowry, 2003;Thanh, 2020), political (Colak, 2017;Luo et al., 2017), COVID-19 induced (Baig and Chen, 2021), and other uncertainty (Barth et al., 2017;Crain et al., 2021).Unlike prior work, we offer an uncertainty measure directly impacting the firm's cash flows.Furthermore, we differ from the above papers by conducting a single sector study.Conducting a within industry study allows us to use a forward-looking industry-specific measure of uncertainty directly linked to the output and input prices.Working with a homogenous set of firms similarly affected by oil price uncertainty facilitates identification.Jin and Jorion (2006) argue that potential endogeneity concerns are alleviated in a single industry setting.Furthermore, Rajan and Servaes (1997), Pagano et al. (1998) and Ritter (1984) all argue for a substantial industry heterogeneity in the time-series of IPO volume.Hence, understanding the role of uncertainty at the industry level also adds to the understanding of aggregated IPO volume.
After the introduction, the remainder of the paper is structured as follows: part two discusses data and empirical setting; part three describes the empirical findings; part four concludes the study.

DATA
We consider all U.S. IPOs in the Thomson Financial Securities Data Company (SDC) database between 1/1/1987 and 12/31/2019.To be considered an oil and gas IPO, firms must have one of the following 4-digit SIC codes: 1300-1399, 2911, 2990, 5171, 5172, 2911, 2992, 3533, 4612, 4613, 4619, 4922, 4923, 4924, 4925, 5541, 5983 and 6792.After the initial screening, we end up with 562 IPO fillings, where 112 represent withdrawn IPOs, resulting in a final sample of 450 completed IPOs raising on average $89 million in 2019 dollars.We further distinguish between upstream IPOs (SIC codes: 1300-1399) and other oil IPOs (See, Appendix A2 for a more detailed overview of the IPO classification).The other oil and gas IPO group mainly consists of downstream firms (e.g.refineries) and a few midstream operators (e.g. trading companies).Our final IPO sample includes 308 upstream and 142 other oil and gas IPOs.
We use two IPO volume measures as main dependent variables, the number of quarterly IPOs (#IPOs) and the natural logarithm of inflation-adjusted aggregated quarterly proceeds [ln(Proceeds)].In additional measures, we control for market-wide changes in IPO volume by scaling the dependent variables by the total number of IPOs (#Oil IPO share) and by market-wide proceeds ($Oil IPO share).We further distinguish between upstream and other oil and gas sector IPOs by creating two variables (#Upstream and #Other) to capture the number of upstream and other oil and gas IPOs.In auxiliary tests, we examine the role of uncertainty on the number of days between filing and issuance (Window Length), and the effect on IPO withdrawals (Withdrawn IPO).
As a measure of oil price uncertainty (Implied Oil Volatility), we use the implied volatility of West Texas Intermediate (WTI) crude oil contracts calculated based on at-the-money second nearby options, which the Commodity Research Group provides.The implied volatility measure is calculated using the Black and Scholes (1973) option pricing model modified for commodities.Once the front-month option expires, the options are rolled over into the second month.Implied volatility of the WTI crude oil contract is collected from 1/1/1987 and ending at 12/31/2019.Our measure of oil price uncertainty is slightly different from CBOE's traditional oil VIX (OVX).Differing from us, CBOE value weighs the quoted volatility of numerous "out-of-the-money" options to calculate OVX.One significant advantage of our measure is that it has been available since 1987, while OVX started in 2007.To ensure the reader that our measure is closely related to the more granular OVX we have plotted the series over time in Figure 1.As shown in the figure, the two oil price uncertainty measures exhibit a high pairwise correlation (0.98).Implied oil price volatility potentially has advantages over other uncertainty measures.Alquist et al. (2013) suggest that implied volatility is a better measure of oil price uncertainty relative to standard measures such as GARCH and historical volatility.Kellogg (2014) similarly reports that implied volatility better predicts investments in the upstream oil sector compared to the other measures.On the other hand, Dossani and Elder (2022) show that GARCH and implied volatility yield similar results.Hence, using implied oil volatility derived from option prices should at least be as good proxy for oil price uncertainty as econometric uncertainty measures.
To capture the valuation impact of uncertainty through the cost of the capital channel we control for variables affecting the other main valuation determinant-cash flows.Therefore, we control for the WTI index (WTI Level).While implied oil volatility captures uncertainty about the cash flows of the IPO firms, the level of WTI is an important determinant of the cash flow levels (Kumar and Rabinovitch, 2013).Hence, we capture both cash flow and discount rate dynamics by including both the WTI and implied oil price volatility.To capture if the oil futures market is in contango or backwardation, we include the WTI Roll Yield.We measure WTI Roll Yield as the difference between the price of the front contract and the 6 th contract.A positive roll means backwardation and a negative indicates contango.To consider the lack of liquidity during the last day of the contract, we roll the position from the front to the second nearest contract and from the 6th to the 7th contract.
We include additional control variables in line with prior IPO prediction models analyzing the number of quarterly IPOs [see, e.g., Lowry (2003)].Our control variables capture market-wide uncertainty and other market-wide factors identified in the literature to drive IPO volume, including the Fed Funds Rate, GDP Growth, Equity Uncertainty, Political Uncertainty and an IPO market heat indicator (IPO Market Heat).All control variables are defined in Appendix A1.To assess the impact of oil price uncertainty on IPO volume, we aggregate our dependent variables [#IPOs and ln(Proceeds)] at the quarterly level.Table 1 shows that the average quarter has 3.39 oil and gas IPOs raising an average proceeds of $110 million.Lowry (2003) predicts IPO volume across all industries and finds that the number of IPOs are non-stationary [cannot reject the Dickey-Fuller test].We evaluate the stationarity of the variables using the augmented Dickey-Fuller test, as described by Elder and Kennedy (2001).We include four lags in our calculation following Schwert's (1989) rule of thumb.In contrast to overall IPOs, #Oil IPOs is stationary; thus, we can reject the augmented Dickey-Fuller test at the 1% level (ADF stat = -3.73),i.e. oil IPOs revert towards a normal quarterly volume.Our second main dependent variable [ln(Proceeds)] exhibits similar properties, rejecting the null of the unit root test at the 1% level (ADF stat = -4.93).
We further examine the properties of the main independent variable (Implied Oil Volatility).By rejecting the null of the Dickey-Fuller test at the 1% level, we can conclude that our main independent variable is stationary.Furthermore, as all independent variables are stationary, enter the model with a one-quarter lag and oil price uncertainty is exogenously determined, our tests should not exhibit reverse causality issues.
Table 1 also shows the characteristics and properties of the dependent and control variables used in auxiliary tests.Oil and gas IPOs represent 4% of the total IPOs in terms of the number of IPOs (#IPO share) and 6% of the total proceeds ($IPO share).Upstream IPOs (2.33) are more prevalent in the sample compared to downstream and midstream IPOs (#Other = 1.07).The average time between filing and issue date is 81.15 days.All dependent variables reject the null of the Dickey-Fuller test at conventional levels, meaning that our variables are stationary.Table 1 further shows the characteristics and properties of our control variables.All control variables except the level of the WTI index, Fed Funds Rate and Political Uncertainty are stationary.Therefore, we use the return of the WTI (WTI Return) and the first difference of the federal funds rate (Δ Fed Funds Rate) and Political Uncertainty (Δ Political Uncertainty) in our analysis.
Table 2 reports the pairwise correlations between the variables.Among the control variables the highest correlation is found between WTI Roll Yield and WTI Return (0.48).Except the high correlation between the GDP Growth and Implied Oil Volatility, the second highest correlation is found between GDP Growth and Implied Oil Volatility (-0.47).

Empirical design
To test our main hypothesis, that oil price uncertainty negatively impacts IPO volume in the oil and gas sector, we perform several reduced-form estimations.In our baseline specifications (1a and 1b), we explain the number of IPOs (#IPOs) and the total proceeds [ln(Proceeds)] raised with oil price uncertainty (Implied Oil Volatility) plus a vector of control variables.Therefore, our main specifications take the following form: where t is a time index at the quarterly level.The vector of control variables consists of: WTI Return, Δ Fed Funds Rate, GDP Growth, Equity Uncertainty, Δ Political Uncertainty, Hot IPO Market.We estimate three versions of the models: 1.) we only include the oil VIX to test for the univariate relation (excluding controls); 2.) we include the control variables; 3.) we acknowledge the seasonality of IPO volume and include quarter indicators.All models are estimated using Newey and West (1987) standard errors with 4 lags.Since overall IPO volume is highly cyclical and correlated with several macro-economic factors [See, e.g., Lowry, (2003); Ritter (1984)], our findings may be driven by underlying factors that affect both oil price uncertainty and overall IPO volume.To ensure that we do not solely capture a general IPO effect, we scale our dependent variables by market-wide IPO numbers.This allows us to capture the oil and gas sector IPO variation distinct from market-wide fluctuations in IPO volume.We scale #Oil IPOs and total proceeds by the total number of U.S. IPOs and aggregated proceeds to create #Oil IPO Share and $Oil IPO Share.We estimate the following models: Oil IPO Share Implied OilVolatility Oil IPO share Implied OilVolatility To further identify the effect of oil price uncertainty on IPO volume, we split our sample into upstream and other oil and gas sector firms.Upstream firms differ from other type of firms in the oil and gas sector due to their sensitivity to crude oil prices (Kumar and Rabinovitch, 2013).Therefore, the economic impact of crude oil price uncertainty on IPO activity differs between upstream and other oil and gas sector firms.Upstream IPO volume should be more adversely affected by increased oil price uncertainty.We create variables for the number of upstream IPOs (#Upstream) and the number of other oil and gas sector IPOs (#Other) and estimate the following models:  In auxiliary tests, we examine how uncertainty affects the time between the IPOs filing date and the issue date (Window Length).We argue that elevated uncertainty causes firms to postpone their IPOs, which would manifest itself through a positive relation between Oil VIX and Window Length.In many cases increased uncertainty not only forces the firm to postpone the issue but also causes a withdrawal of the filed IPO.To study the effect of oil price uncertainty on withdrawals, we use a sample of all withdrawn IPOs at the firm level.We then create an indicator variable taking the value of one if the withdrawal is in the oil and gas sector.By using Probit models, we test if changes in the Oil VIX between filing and the withdrawal date can explain the withdrawal likelihood.We further acknowledge that several of our dependent variables (#Oil IPOs, #Upstream and #Other) are not continuous variables but instead of count data type.Therefore, we re-estimate all our models using Poisson regressions allowing for a one lag autocorrelation (Schwartz et al, 2006).

EMPIRICAL FINDINGS
In the first set of tests, we examine the univariate relation between oil price uncertainty and IPOs, and how the dynamics between oil price and oil price uncertainty affect the going-public decision.Panel A of Table 3 1 shows univariate differences in IPO volume [#Oil IPOs and ln(Proceeds)], where we split the sample at the 50 th percentile into high and low uncertainty quarters.Our tests reveal that both #Oil IPOs and ln(Proceeds) are significantly higher during low uncertainty quarters.The magnitudes of the differences are large, the average number of oil IPOs is 4.00 during low compared to 2.716 in high uncertainty quarters (p<0.05), which corresponds to a 68% higher IPO volume.We report similar differences in the amount of proceeds raised (p<0.01), the numbers given in natural logarithms (low uncertainty 5.374; high uncertainty 3.989) correspond to $215.8 million raised during low compared to $54.0 million raised during high uncertainty quarters.Our univariate findings suggest that uncertainty is an important factor in determining IPO volume.Figure 2 shows the time-series relation between Implied Oil Price Volatility and the number of quarterly IPOs.From the graph, it becomes clear that the two series correlate negatively, i.e. the number of IPOs is higher during low uncertainty quarters.Furthermore, both series appear to be stationary without an evident time trend.
Panel B of Table 3 explores the dynamics between oil price uncertainty, oil price and IPO volume.We double sort quarters at the 50 th percentile in terms of the WTI and Oil VIX levels.We examine the differences in IPO volume between the four groups.Our findings suggest that oil price uncertainty has a greater impact on IPO volume during high oil price quarters.This holds for both the number of IPOs and the proceeds raised.We find similar evidence when analyzing differences between high and low oil price quarters, oil price only affects IPO volume during quarters of low oil price uncertainty.Quarters with low uncertainty coupled with high oil prices exhibit the highest IPO volume.Interestingly, all other quarters exhibit similar IPO volume (3; 2.4; 2.9 for #Oil IPOs), except the low uncertainty and high oil price quarters where the number of oil IPOs is 5.233.The pattern is similar when examining the proceeds raised in the IPO.Our findings highlight the important dynamics between the oil price and uncertainty in determining the going-public decision.
Next, we test our baseline models (1a) and (1b) to evaluate Hypothesis (1), which states that higher oil price uncertainty is linked to lower IPO volume.Columns (1) to (3) of Table 4 explore the link between oil price uncertainty and the number of quarterly oil IPOs (#Oil IPOs).In line with our expectations, all three models show a negative relation between Implied Oil Price Volatility and #Oil IPOs (p<0.01).The economic magnitude is large in all three specifications, a one standard deviation decrease in the Implied Oil Price Volatility (13.59) results in an increase in the number of quarterly Oil IPOs between 0.88 and 0.98.This corresponds to a 26% to 29% increase in the number of quarterly IPOs relative to the quarterly average (3.39).In columns (4) to (6) we instead test for the impact of oil price uncertainty on quarterly proceeds.Similarly, we report a strong negative relation between Oil VIX and ln(Proceeds).The economic magnitude is again substantial, a one standard deviation decrease in the Implied Oil Price Volatility represents a 0.94 to 1.05 increase in log proceeds, corresponding to 20% to 22% relative to the mean ln(Proceeds).This corresponds to $22.2-$24.8 million raised during the quarter in dollar values.Only Equity Uncertainty and Δ Political Uncer- tainty exhibit consistent negative and positive relations to IPO volume among the control variables.In models ( 5) and ( 6), the WTI return affects the IPO volume positively.Our findings confirm Hypothesis (1) that oil price uncertainty negatively impacts IPO volume.
Next, we aim to ensure that we identify an oil and gas sector effect.Since oil price uncertainty is correlated with other macro-economic factors that potentially can drive market-wide IPO volume, we create #Oil IPO Share and $Oil IPO Share.By scaling our dependent variables, we distinguish oil and gas sector-specific IPO volume variation from market-wide IPO volume.If IPO volume in the oil and gas sector is more affected by oil price uncertainty relative to market IPOs, we expect a negative Implied Oil Price Volatility coefficient.
Columns (1) to (3) in Table 5 report the impact of Implied Oil Price Volatility on #Oil IPO Share.The estimation in column (1) do not show any distinct impact of Implied Oil Price Volatility on #Oil IPO Share.This suggests that without accurate controls, the oil and gas sector IPOs follow the same pattern as market-wide IPOs.After including standard controls in column (2), we observe a distinct negative relation of oil price uncertainty on the proportion of oil IPOs (p<0.01).When further adjusting for the seasonality of listings, the effect becomes even more pronounced (p<0.01).Models (4) to (6) all show a significant negative impact of Implied Oil Price Volatility on the oil and gas sector share of total proceeds raised in IPOs during the given quarter.Our controls reveal that the share of oil and gas IPOs are more affected by WTI return and Equity Market Uncertainty compared to normal IPOs.Furthermore, oil IPOs are less sensitive to GDP Growth compared to overall IPOs and are less likely to be clustered within hot issue markets.Overall, our findings in Table 5 further confirm Hypothesis (1), highlighting a significant negative impact of oil price uncertainty on IPO volume in the oil and gas sector.We construct a placebo test to ensure that our findings are not driven by factors correlated with oil price uncertainty.In our tests we have chosen a sector that by definition should have a lower exposure to oil price uncertainty compared to the oil and gas sector-the Hi-tech sector.We identify Hi-tech companies following Loughran and Ritter (2004).To observe if Hi-tech IPOs are affected by Implied Oil Price Volatility we aggregate both the number of Hi-tech IPOs scaled by the market-wide IPOs (#Hi-tech IPO share) and Hi-tech proceeds scaled by total proceeds ($Hi-tech IPO share).Our findings in Appendix A3 do not show any direct link between Implied Oil Price Volatility and the share of Hi-tech IPOs.Both the coefficients of Implied Oil Price Volatility and the corresponding t-stat is close to zero in all specifications except in column (4).
In the next set of tests we examine Hypothesis (2), stating that upstream IPOs should be more sensitive to oil price uncertainty.We create two additional dependent variables #Upstream and #Other (#Other includes IPOs by midstream and downstream firms) to test the hypothesis.The rationale behind our tests is that downstream and midstream firms do not face the same oil price sensitivity since they can pass on price fluctuations to their customers (i.e. through the crack spread), while upstream producers' cash flows are more sensitive to oil price fluctuations (Doshi et al, 2018;Kumar and Rabinovitch, 2013).
Columns (1) to (3) of Table 6 show that upstream IPO volume is highly sensitive to oil price uncertainty.Implied Oil Price Volatility is negative and statistically significant at the 1% level in all three specifications.In line with our expectations, other oil and gas IPO firms in columns (4) to (6) experience low sensitivity to oil price uncertainty.We only find a weak negative effect in the univariate setting of column (4).As predicted by Hypothesis (2), we find that upstream producers' going-public decision is more sensitive to oil price uncertainty relative to downstream and midstream firms.This table shows time-series regression at the quarterly level.The dependent variables are the number of upstream Oil IPOs during the given quarter (#Upstream) in columns (1) to ( 3) and the number of other Oil IPOs during the given quarter (#Other)in columns ( 4) to (6).All independent variables enter the model with one lag.The t-stats are based on Newey-West standard errors using four lags.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Next we test Hypothesis (3a), stating that high levels of uncertainty increase the value of the real option to postpone the listing.To test our hypothesis we create a variable (Window Length), which is the number of days between the filing date and issue date averaged across firms at the quarterly level.Oil VIX is measured with one lag prior to the filing date to capture uncertainty at the time of the filing.Hence, if managers take into account uncertainty we expect that an elevated Implied Oil Price Volatility prolongs the time-window between filing and listing.
Column (1) of Table 7 reports no effect of Implied Oil Price Volatility on the Window Length in a univariate setting.However, after including standard controls in column (2) and quarter fixed effects in column (3), we find positive and significant relationships between Implied Oil Price Volatility and Window Length.Our results indicate that elevated uncertainty causes managers to postpone the listing.This result is in line with real options models [see, e.g., Pastor and Veronesi, (2005)], where higher uncertainty increases the value of the waiting option, causing managers to defer listing in anticipation of better market conditions.
Next we test Hypothesis (3b), stating that a higher level of oil price uncertainty increases the likelihood of a withdrawn oil and gas IPO.We test Hypothesis (3b) at the firm level and collect a sample of all withdrawn IPOs in the U.S. during 1/1/1987-12/31/2019.Our dependent variable is an indicator taking the value of one if the withdrawn IPO is in the oil and gas sector and zero for withdrawn IPOs in other sectors.Since, our control group consists of all other withdrawn IPOs, we effectively control for macro-economic factors causing firms to withdraw their IPOs.The main independent variable is the difference in the Implied Oil Price Volatility between the filing date and the withdrawal date (Implied Oil Volatility Diff.).We expect oil and gas sector withdrawals to be positively related to Implied Oil Volatility Diff., indicating that changes in oil price uncertainty impacts the listing completion.In columns (1) to (3) of Table 8, we estimate the oil and gas IPO withdrawal likelihood using Probit models with standard errors clustered on year.In columns (4) to (6) we further include year indicators to control for additional market-wide factors.Models (3) and ( 6) also include a control for the filed IPO amount.All Implied Oil Volatility Diff.coefficients are positive and statistically significant regardless of the model specification.In support of Hypothesis (3b), our findings show that changes in oil price uncertainty have real effects on the decision to complete an IPO after the filing.

Discussion and Robustness
This study examines how oil price uncertainty affects IPO volume in the oil and gas sector.In our reduced-form estimations, we make the strong assumptions that implied oil price volatility is exogenous with respect to IPO volume.Since, this is a strong assumption, we take several steps to alleviate this concern.Endogeneity stems from either reverse causality or confounding factors.First, reverse causality is a lesser problem since oil price uncertainty is likely not driven by IPO volume.The determinants of the oil price and oil price uncertainty are vastly discussed in the literature (Kilian, 2009;2010;2014;Baumeister and Kilian, 2016;Pierru et al., 2018;Baumeister and Hamilton, 2019).They reveal that other factors such as OPEC decisions (Pierru et al, 2018) and macro-economic indicators (Barsky and Kilian, 2001;2004) affect oil price uncertainty and are not driven by firm-level financing decisions.As a further step, we use lagged explanatory variables to alleviate concerns about potential simultaneity and reverse causality.
A greater concern relates to omitted variables in our main specifications.We address this problem in three ways.First, as oil price uncertainty can be correlated with macro-economic un-certainty (Barsky and Kilian, 2001;2004), we include a wide range of controls motivated by prior studies (e.g., Lowry et al., 2003) and add WTI return along with equity and political uncertainty measures.Second, we scale our dependent variable by the market-wide number of IPOs, allowing us to distinguish between oil and gas sector-specific IPO variation from overall IPO activity.This is important since oil price uncertainty is correlated with several factors likely also to impact the going-public decision (e.g.GPD, political uncertainty etc.).Third, even though we do not use excluded instruments in our analysis due to the exogeneity of oil price uncertainty, we test for the severity of the potential omitted variable bias using Oster (2019) methodology of coefficient stability. 1We conduct the Oster's (2019) partial identification test in our main specification (Column 2 of Table 4).Our results indicate that the beta coefficient estimates stay within negative bounds after using the Oster (2019) proposed of 1.3* as the key input to the formula.The calculated estimations suggest that the observables are at least as informative as the non-observables in our models.
The oil and gas sector has undergone significant changes since the sample began in 1987.One of the major revolutions came with fracking in 2012 (Gilje et al., 2013), which led to a near doubling of oil production in the U.S. As seen in Figure 2, a spike in oil and gas IPOs coincides with the fracking revolution.This motivates us to study if our results are driven by the Post-2012 or Pre-2012 time periods.We replicate our analysis from Table 4 by dividing our sample into two sub-samples (before and after the fracking revolution).The outcome of the analysis (Appendix A4) does not alter our previous interpretation, Implied Oil Price Volatility has a negative and robust im-1.The Oster (2019) test is widely used as a test of the coefficient stability in the economics literature (Acharya et al., 2019;Heimer et al., 2019;Ruhm, 2019)  We further acknowledge that several of our dependent variables (#Oil IPOs, #Upstream and #Other) are not continuous variables but instead of count data type.Therefore, we re-estimate all our models using Poisson regressions, allowing for a one lag autocorrelation using the model of Schwartz et al., (2006).Our findings in Appendix A5 confirm previous results, oil price uncertainty has a negative impact on total oil and gas and upstream IPO volume, while downstream and midstream IPOs do not exhibit any significant sensitivity to oil price uncertainty.

CONCLUSIONS
We examine the impact of oil price uncertainty on IPO volume in the oil and gas sector.Our uncertainty measure (Implied Oil Volatility) potentially offers advantages over other measures.The measure is forward-looking and exogenously determined, allowing for a cleaner identification in our tests.The theoretical argument for firms to refrain from their IPO stems from the cost of the capital and real options channels.Even though the two channels are not mutually exclusive, they provide different predictions on how uncertainty affects the likelihood of going-public.The cost of capital channel postulates that increased uncertainty has an adverse impact on the firm's valuation, and a lower valuation makes the cost of going-public outweigh the benefits.The real options channel postulates that the value of postponing the offering increases, given that the decision is not completely reversible.
We develop three hypotheses.First, we test the general effect of oil price uncertainty on the going-public decision, using four different measures of IPO volume.Second, we study upstream producers in isolation, this set of firms are theoretically more sensitive to changes in uncertainty.Third, we explore how uncertainty affects the time between filing and IPO and how it affects the likelihood of a withdrawn IPO.
We test our predictions using a sample of 450 completed oil and gas IPOs during the time-period 1/1/1987-12/31/2019.We find strong support for our main hypothesis, a one standard deviation decrease in implied volatility corresponds to an increase of 29% (22%) in the number of IPOs (proceeds).Concurring with Hypothesis (2), the negative relation between oil price uncertainty and IPOs is concentrated among the price-sensitive upstream producers.We further report that oil price uncertainty increases the time between filing and issue, and increases the IPO withdrawal likelihood.This table re-estimate our main results using Poisson models to take into account of that our dependent variables are of count data type.The dependent variable in columns (1) to (3) is the number of IPOs in the oil and gas sector during the given quarter.In columns (4) to (6) the dependent variable is the number of upstream IPOs and in columns ( 7) to ( 9) the number of other IPOs according to our classification in Appendix A2.The t-stats are reported in parentheses.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Figure 1 :
Figure 1: Implied oil volatility and the OVX

Figure 2 :
Figure 2: Implied oil volatility and the number of oil and gas IPOs

Table 1 : Descriptive Statistics
*This table shows descriptive statistics at the quarterly level.The fourth column shows augmented Dickey and Fuller unit root tests.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 2 : Correlation Matrix
This table shows the pairwise correlation between all included variables.Variable definitions can be found in Appendix A1.

Table 3 : Univariate differences
This table shows univariate differences in the IPO volume measures[# Oil IPOs; ln(Proceeds)].We split the quarters on the median of the level of Implied Oil Volatility and the level of WTI.***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 4 : Baseline results
This table shows time-series regression at the quarterly level.The dependent variables are the number of IPOs in the oil and gas sector during the given quarter (#Oil IPOs) in columns (1) to (3) and the total quarterly proceeds raised by IPO firms in the oil and gas [ln(proceeds)] in columns (4) to (6).All independent variables enter the model with one lag.The t-stats are based on Newey-West standard errors using four lags.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 5 : Oil IPOs relative to all IPOs
This table shows time-series regression at the quarterly level.The dependent variables are the number of Oil IPOs during the given quarter scaled by the total number of IPOs (#IPO share) in columns (1) to (3) and the total proceeds raised by oil IPOs in relation to total proceeds raised during the given quarter ($IPO share) in columns (4) to (6).All independent variables enter the model with one lag.The t-stats are based on Newey-West standard errors using four lags.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 7 : Time between the filing and issue of the IPO
This table shows how oil price uncertainty affects the time between the filing and issue date (Window Length).The oil price uncertainty is measured at the beginning of the filing quarter.All independent variables enter the model with one lag.The t-stats are based on Newey-West standard errors using four lags.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Table 8 : Likelihood of withdrawal
This table shows Probit models estimating the likelihood of an IPO withdrawal is conducted by an oil and gas firm in sample including all withdrawals.The main independent variable is Oil VIX difference which is the difference between the Oil VIX at the withdrawn month minus the Oil VIX at the filing month.The t-stats are calculated from heteroscedasticity robust standard errors clustered on year.All variables are defined in Appendix A1. ***, **, * denote statistical significance at the 1%, 5%, and 10% levels, respectively.pact on IPO volume in both sub-samples.The Hot IPO Market indicator falls out from the Post-2012 sample due to low overall IPO activity during this period.