A. IPO Sample

Data on IPO filings conducted in the U.S. markets are from the LSEG SDC New Issues database from 1982 to 2021. We deploy the following data filters that are common in the empirical IPO literature (see Purnanandam and Swaminathan (Reference Purnanandam and Swaminathan2004), Liu and Ritter (Reference Liu and Ritter2011)). We exclude unit offers, closed-end funds, REITs, ADRs, limited partnerships, spin-offs, issues of noncommon shares, IPOs with an offer price less than $5, and IPOs in the financial industry with SIC codes between 6000 and 6999. Additionally, SDC provides corporate address information for each IPO, which we use to identify firm locations. Other IPO characteristics, including VC backing, listing exchange, underwriter market share, negative income status, offer price, and shares outstanding, also come from SDC.

We obtain data on the number of years since the company was founded (IPO age) from Jay Ritter’s website.Footnote ¹⁰ Information on stock prices and returns comes from the CRSP database, and accounting data come from the Compustat database. For the firm’s accounting data, we use the nearest quarter reported before its IPO when available.Footnote ¹¹ To be included in our sample, we also require nonmissing asset, market-to-book, and asset tangibility data. These filters result in a sample of 7,456 IPO filings. Figure 1 presents the geographic dispersion of these IPOs.

FIGURE 1 Total Number of IPOs by State

Figure 1 shows the total number of IPOs, sorted by the state where the firm is headquartered. The sample period is 1982–2021.

B. Identifying Intangible-Intensive Firms

The main hypothesis of this article is that increased REC will increase the propensity of intangible-intensive firms to go public. We identify a firm as intangible-intensive if it offers technology services according to the SDC. The SDC’s classification of offering these services encompasses any of the following categories: biotechnology, communications, computer equipment, electronics, and general technology. In practice, this classification is used at the time of the IPO to categorize banking league tables and identify companies with business models based on providing technological services. This measure is suitable for our purposes for several reasons. First, it identifies firms whose core business relies on intangible assets, which the conventional standard industrial classification system cannot capture. Second, as this measure is based on the firm’s business model, it is an ex ante measure of the firm’s intangible intensiveness and not specific to the timing of the IPO. Third, it is readily accessible and is not subject to the researcher’s judgment.

We also adopt several alternative approaches to classify a firm as intangible-intensive. We use an indicator for whether the firm is in the bottom tercile of asset tangibility compared to the sample of all IPO firms. Tangibility is defined as the firm’s net property, plant, and equipment (PP&E) divided by book assets. The assumption is that firms with low asset tangibility are more likely to be intangible-intensive.

Next, we create an indicator for firms that report negative income at the time of the IPO. This measure captures the notion that intangible-intensive firms are likely to initially lack positive income. The last proxy is an indicator that the firm is less than 5 years old at the time of IPO, which we consider a young firm. Intangible-intensive firms are generally younger because their business operations involve recently developed technologies and cutting-edge processes. Firm age also broadly reflects the financing wedge that may arise due to information asymmetry between insiders and outside investors. Consequently, REC is likely to be important for younger IPOs because proximity will help investors learn directly or indirectly from insiders or other stakeholders.Footnote ¹²

For each measure of intangible intensiveness, we need a baseline measure of the total number of private firms in the state and year that could potentially go public. We obtain this information from the U.S. Census Bureau’s Business Dynamics Statistics (BDS) data.Footnote ¹³ Because we divide the IPO sample into different categories (e.g., intangible-intensive vs. nonintangible-intensive, negative-income vs. nonnegative-income, younger vs. older firms), private firms also must be split into these same groups. We take the following approach to form these groups because most of these characteristics are not observable for private firms.

For our primary group, intangible intensiveness, we calculate the percentage of IPOs designated as intangible-intensive by SDC within each 2-digit NAICS industry code. We use these percentages to estimate the number of private firms that are intangible-intensive within each 2-digit NAICS industry code in the BDS data. For example, for the Professional, Scientific, and Technical Services industry (NAICS code 54), 72% of IPOs in this group are intangible-intensive. Therefore, we assume that 72% of all private firms in this industry are intangible-intensive. We apply this procedure across all industries for each state and year.Footnote ¹⁴

We repeat the same procedure for the low tangibility and negative income classifications. One exception to this approach is our classification of younger versus older private firms. In this case, we have state-level data on private firms by age. We therefore group private firms by age directly rather than imputing their sample from the IPO data. Across all our measures, we restrict the sample to firms 20 years or younger, as the majority of firms that choose to IPO are less than 20 years old. As a robustness exercise, we also use an expanded sample that includes all firms.

C. REC

Apart from identifying intangible intensiveness among IPO firms, our main hypothesis requires a measure of the supply of capital in the public equity market in a region. We quantify local capital by measuring the amount of REC, the capital and resources controlled by institutional investment managers in the firm’s state. These investors (e.g., fund managers and broker/dealers) can allocate these funds for investment opportunities such as IPOs. To measure REC, we use equity holdings data from 13F regulatory filings from LSEG (formerly Thomson).Footnote ¹⁵ At the end of each year, we sum the total dollar amount of reported equity holdings of all institutional investors located in each state.

Calculating REC requires accurate investor location data. For institutional investor locations, we collect all 13F SEC filings in the online Edgar system from 1999 to 2021 and extract investment managers’ contact information and business addresses. We begin in 1999 because reporting in the Edgar system is relatively sparse before then. Using these business addresses, we first correct misspellings and errors, and then identify the state where each manager is located. We can identify the managers’ location relatively precisely because investors file 13Fs as frequently as each calendar quarter. For manager locations before 1999, we use Nelson’s Directory of Investment Managers. By combining these two data sources, we can identify the manager locations of 99% of Thomson 13F equity assets reported during our sample period.

As REC is designed to identify capital that can serve an information discovery role, indexed capital would fall outside of this category. We therefore exclude the Big Three index funds (i.e., BlackRock, Vanguard, and State Street Global Advisors) from our measure of REC.

As a final step, we exclude institutional investor holdings of firms headquartered in the investor’s state from REC. This adjustment negates the possibility that a positive local economic shock would induce a positive correlation between local equity capital and local IPOs.Footnote ¹⁶ Given that the standard Compustat database provides only the most recent headquarters location, we identify accurate historical firm headquarters locations as follows: First, we begin with the corrected locations from Bai, Fairhurst, and Serfling (Reference Bai, Fairhurst and Serfling2020), which are based on historical SEC filings and hand-collected data from the Mergent Manuals (via Moody’s Manuals and Dun & Bradstreet’s Million Dollar Directory). For firms that are not included in this data, we obtain corporate address information from the point-in-time (PIT) Compustat database.Footnote ¹⁷ As a robustness test, we use only the PIT Compustat headquarters locations and find nearly identical results.

A central assumption of our hypothesis is that when the marginal investor in an IPO is located in the same region as the firm, they benefit from an informational advantage in evaluating local businesses. We define a region using state boundaries because they are parsimonious and allow for sufficient IPOs in a given year. In Section IV.D, we conduct the analysis using MSAs to define regions. The main inferences remain unchanged.

D. Other Measures of Capital

As an alternative to the REC measure, we use Ln(REC), External Pensions. This variable is a subset of REC based on three major U.S. state public pension systems: the California Public Employees’ Retirement System (CALPERS), the Florida State Retirement System (FRS), and the South Carolina Retirement System (SCRS). As these systems are funded by a combination of state government and employee contributions, the source of capital is specific to each state. They publicly report their public-equity allocations, including the external managers they employ and the net asset value invested with each. Using data for 1987–2017, we hand-match managers to states and exclude any assets managed within a pension’s home state.Footnote ¹⁸ The measure thus captures capital invested locally but sourced from out of state, which we use in Section III.C to address identification concerns about local economic shocks affecting our results.

We also include a measure of RVC in some of our tests. For this measure, we use the assets under management (AUM) by venture capitalists in a particular state and year to capture local VC activity. From the National Venture Capital Association (NVCA) yearbooks, the data span 1981–2021. We use this measure to confirm that our results with REC are not being driven by correlated activity in the VC sector.

E. Additional State-Level Data

Besides the different measures of capital, we include additional state-level economic controls. State-level nominal GDP data come from the Bureau of Economic Analysis (BEA). Bank deposits are from the FDIC’s Summary of Deposits (SOD) data.Footnote ¹⁹ The total amount of individual tax revenue collections reported in each state comes from the IRS.Footnote ²⁰

F. Sample Summary Statistics

Panel A of Table 1 contains the descriptive statistics of the IPO-level sample. Table A.1 in the Supplementary Material lists the variable definitions and data sources. The first part of the table reports the characteristics of the IPOs. The median IPO firm is 8 years old. Using our main measure of intangible-intensive firms ( $ \unicode{x1D7D9} $ (Intangible)), 54% of the IPOs are classified as such. On average, 42% of the sample has VC backing before their IPO. We note that while intangible-intensive IPOs often have VC backing, the correlation between the two variables is only 0.45. The percentages of IPOs for the alternative intangible-intensive measures are as follows: low tangibility, negative income, and young firms at 34%, 45%, and 33%, respectively.

TABLE 1 Summary Statistics: IPO-Level Sample

In terms of other characteristics of IPOs, 16% are listed on the NYSE or AMEX, and the remaining 84% are listed on NASDAQ, which is similar to the statistics reported in previous studies. The average IPO firm has a post-issue asset market-to-book ratio of 2.4 or an equity market-to-book ratio of 2.77. IPO firms have, on average, a 0.91% BHAR over the subsequent 6 months. The one-year BHAR for IPO firms is −6.4%, which is consistent with prior findings documenting the long-run underperformance of IPOs (Ritter and Welch (Reference Ritter and Welch2002)).

Panel A of Table 1 also reports the summary statistics for the state-level REC, RVC, and related measures of capital.Footnote ²¹ The average amount of REC is $435.25 billion compared to $18.27 billion for RVC. The table also includes statistics for nominal state-level GDP, bank deposits, and income. For the number of private firms, the average IPO firm comes from a pool of 43.50 thousand firms that are 20 years or younger. Using the expanded sample of firms (regardless of age), the average pool size is 127.59 thousand firms.

In Panel B of Table 1, we separate and compare IPOs using our main measure of intangible-intensive IPOs. We observe that our different proxies for intangible-intensive firms are positively correlated, as these firms are more likely to have negative income and are younger. These firms are more likely to list on NASDAQ and have VC backing. They tend to be smaller, but still generate similar IPO proceeds. Intangible-intensive firms go public more frequently in states with higher REC, RVC, and higher state GDP and income.

Panel A of Table 2 presents the number of intangible-intensive IPOs by Fama–French 49 industry classification. Although they often appear in expected fields (e.g., computer software and hardware, electronic equipment, business services), intangible-intensive IPOs are distributed across numerous other industries. These include pharmaceutical products, medical equipment, communication, health care, and even retail. Indeed, 34 of the nonfinancial industry classifications have at least one intangible-intensive IPO during our sample. Panel B of Table 2 presents the number of intangible-intensive IPOs by state. California, Massachusetts, and New York have the highest number of such IPOs. Nonetheless, we observe broad geographical coverage: except for Hawaii, West Virginia, and Wyoming, all other states in our sample have at least one intangible-intensive IPO.

TABLE 2 Distribution of Firms with Intangible-Intensive IPOs

A. The Frequency of IPOs

To test the main hypothesis that REC affects the propensity of regional firms to go public, we design the following fixed effects Poisson regression:

(1) $$ {\displaystyle \begin{array}{l}\mathrm{No}.\ {\mathrm{IPOs}}_{ist}={\beta}_1\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\times Ln{(\mathrm{REC})}_{st-1}+{\beta}_2\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip9em +\hskip2px {\beta}_3Ln{(\mathrm{REC})}_{st-1}+{\beta}_4{\mathrm{State}\ \mathrm{Controls}}_{st-1}+{\beta}_5\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip9em \times {\mathrm{State}\ \mathrm{Controls}}_{st-1}+{\beta}_6Ln{(\mathrm{No}.\ \mathrm{of}\ \mathrm{Firms})}_{ist}+{\delta}_s+{\gamma}_t{\varepsilon}_{ist}.\end{array}} $$

For each state ( $ s $ ) and year ( $ t $ ), we have two categories of IPO: intangible-intensive IPOs ( $ i=1 $ ) and other IPOs ( $ i=0 $ ). The data structure is such that for each state and year, there are two observations: the number of intangible-intensive IPOs and the number of other IPOs. For example, in 2000, Minnesota had five intangible-intensive IPOs and two other IPOs, constituting two distinct observations. The data structure in our sample would be as follows:

The main independent variable of interest is the interaction between the natural logarithm of REC at the state level and the intangible-intensive IPO indicator. Panel A of Table 3 provides the summary statistics for this state-level data structure.Footnote ²² The average number of IPOs per state-year for our different measures of intangible intensiveness is presented in Panel B of Table 3.

TABLE 3 Summary Statistics: State-Level Sample

To estimate a propensity to go public, the model includes a measure for the total number of firms in the state and year to provide a baseline supply of firms. For the number of firms, we include estimates of both the total number of intangible-intensive firms and other firms in a state. Continuing our example for Minnesota in 2000, we estimate an underlying population of 10,353 intangible-intensive firms and 26,241 other firms. The measure we use, Ln(No. of Firms), takes the natural logarithm of the number of firms to reduce the influence of extreme values. See Section II.B for additional details.

We also include state-level deposits as a proxy for local banking capital, as well as state-level GDP and income, to account for other regional economic factors that may influence both REC levels and IPO activity. In some specifications, we interact these economic variables with the intangible-intensive indicator to ensure that the indicator’s interaction with REC is not capturing the effect of one of these potentially correlated economic factors.

We include state fixed effects in the specifications. Consequently, identification comes from within-state changes in REC over time rather than from cross-state differences in REC. As certain states have higher levels of REC and IPO activity due to size and economic conditions, this fixed effect separates the effect of REC on the occurrence of IPOs from these factors. We also include year fixed effects to remove macroeconomic shocks (e.g., the dot-com bubble) that alter both REC and the frequency of IPOs. In later specifications, we use state-by-year fixed effects that control for any local economic shock affecting both REC and IPO activity in 1 year. This approach ensures that the effect of REC on intangible-intensive IPOs does not come from some other confounding factors. We cluster our standard errors at the state and year levels.

Table 4 presents the results from estimating equation (1). In column 1, we find that higher REC is associated with more IPOs, but the effect is not statistically significant. In column 2, we examine whether the effects differ by the intangible intensiveness of the firm, using our main intangible measure. We find that REC has a positive and statistically significant effect (at the 1% level) on intangible-intensive IPOs, both relative to nonintangible-intensive IPOs ( $ {\beta}_1 $ ) and in absolute terms ( $ {\beta}_1+{\beta}_3 $ ). We find similar estimates in column 3, where we interact the intangible measure with other state-level economic variables. The effect of REC is economically meaningful. Using column 3, a 10% increase in REC leads to 3.9% more intangible-intensive IPOs relative to nonintangible-intensive IPOs. This increase in REC leads to 2.4% more intangible-intensive IPOs in absolute terms. Given that there are 2.04 intangible-intensive IPOs on average in each state-year (see Table 3) and 51 states, including DC, this equates to 2.5 additional IPOs nationwide each year.Footnote ²³

TABLE 4 Incidences of IPOs

Column 4 repeats the specification from column 3 but introduces state-by-year fixed effects. The relative effect of REC on intangible-intensive IPOs ( $ {\beta}_1 $ ) remains similar to column 3, confirming that increases in REC spur additional intangible-intensive firms to go public. Unlike earlier specifications, we cannot quantify the effect of REC on intangible-intensive IPOs in absolute terms, but rather its effect only relative to nonintangible-intensive IPOs.Footnote ²⁴ For a 10% increase in REC, we document a 3.5% increase in intangible-intensive IPOs relative to nonintangible-intensive IPOs.Footnote ²⁵

Columns 5 through 7 repeat the specification in column 4 but replace $ \unicode{x1D7D9} $ (Intangible) with alternative measures of intangible intensiveness. These measures include $ \unicode{x1D7D9} $ (Low Tangibility) (column 5), $ \unicode{x1D7D9} $ (Negative Income) (column 6), and $ \unicode{x1D7D9} $ (Young Firm) (column 7). We interact each of these measures with REC and the other state-level control variables, and recompute the baseline number of firms in each state for each of these measures (see Section II.B). In all 3 columns, we find that the effect of an increase in REC is statistically significant. A 10% increase in REC is associated with 2.5% more low tangibility IPOs (column 5), 2.8% more negative income IPOs (column 6), and 0.9% more young firm IPOs (column 7).Footnote ²⁶ For each measure, the calculated change is relative to the other category of IPO firms (e.g., low tangibility relative to nonlow tangibility).

In terms of other control variables, the state deposits variable is generally statistically insignificant (except in column 3). This result suggests that local banking capital does not meaningfully affect incidents of resident-firm IPOs. State-level GDP and income also do not generally predict IPO activity. In contrast, states with more private firms tend to have more IPOs.

To confirm the robustness of our main findings, we perform the following tests. First, we relax the age restriction that private firms must be 20 years or younger for the construction of Ln(No. of Firms). The results are presented in Table A.2 in the Supplementary Material. Second, we use an alternate version of REC that is calculated using firm locations from Compustat PIT data instead of data from Bai et al. (Reference Bai, Fairhurst and Serfling2020) (see Table A.3 in the Supplementary Material). Third, we restrict our REC measure to assets managed by institutions that have invested in a new IPO in the last 4 quarters (see Table A.4 in the Supplementary Material). In all cases, the main results are unchanged.

Fourth, in Table A.5 in the Supplementary Material, we repeat our analysis for each of the 4 decades of our sample. The effect of REC on the decision to go public remains statistically significant and economically important in all 4 decades of our sample. Although the estimated magnitudes are largest in the period that includes the 2000 tech bubble, our findings are not solely driven by this event or other specific market episodes. Additionally, Figure A.1 in the Supplementary Material plots the coefficient estimates on REC using 5-year rolling windows and does not show any extreme outlier periods.

In summary, the results support our main hypothesis that REC has a positive and economically meaningful effect on IPO activity for intangible-intensive and informationally opaque firms. This effect is separate from other economic conditions. To the extent that these firms rely on financing from local sources of equity capital, the agglomeration effect captured by REC is a novel and important factor in firms’ IPO decisions.

B. The Role of VC

Although VC is a potentially important funding source for intangible-intensive private firms, it is not omnipresent: 37% of the 4,049 intangible-intensive IPOs in our sample lack VC backing. Nevertheless, it remains possible that our measure, REC, is capturing regional differences in VC activity. To address this concern, we conduct additional tests in Table 5, where we explicitly control for the RVC.Footnote ²⁷

TABLE 5 Incidences of IPOs, Regional Venture Capital

Column 1 of Panel A of Table 5 shows the equivalent analysis of column 4 from Table 4 for the subsample with nonmissing RVC data. We observe similar parameter estimates. Column 2 replaces REC with RVC in the interaction term with intangible-intensive firms. Comparing columns 1 and 2, a 10% increase in REC leads to a 3.6% increase in intangible-intensive IPOs relative to other IPOs. A 10% increase in RVC is associated with a 2.7% increase in intangible-intensive IPOs relative to other IPOs.Footnote ²⁸ To confirm that REC is not simply a proxy for the presence of VC, column 3 includes both sources of funding. We find that the positive and significant effect of REC on IPO activity is robust to the inclusion of RVC.

As a further test, in column 4, we exclude any IPOs that had VC backing from the analysis. This specification addresses the concern that pre-IPO firms with institutional investments drive the results (Kwon et al. (Reference Kwon, Lowry and Qian2020), Chernenko et al. (Reference Chernenko, Lerner and Zeng2021), and Huang et al. (Reference Huang, Mao, Wang and Zhou2021)). We continue to find a positive and statistically significant effect of REC on intangible-intensive IPOs, reinforcing the conclusion that the REC effect is not driven by VC availability.

Separate from REC, the effect of VC is not obvious because regional VC can push IPO activity in opposite directions. More RVC may spur intangible-intensive startups and thus increase IPOs, but it can also help such firms remain private longer, reducing IPOs. The positive RVC coefficient indicates that the former channel dominates. One reason may be that RVC does not capture other late-stage funding that keeps firms private (e.g., private equity, endowments, pensions, sovereign wealth funds). In any case, both RVC and REC impact IPO decisions.

Another potential concern is that our results are concentrated in specific states that are hubs of intangible-intensive IPO activity and abundant VC funding. There are high numbers of intangible-intensive IPOs in California, Massachusetts, and New York (Panel B of Table 2). These states also have the most VC funding. Although the use of state fixed effects removes any persistent differences across states in Table 4, and state-by-year fixed effects further remove any local economic shocks that affect both REC and IPO activity, these states could still be the main driver of our results. In Panel B of Table 5, we repeat the analysis of column 3 of Panel A of this table but exclude California, Massachusetts, and New York (columns 1, 2, and 3, respectively). In column 4, we exclude all three states from the sample. REC remains statistically significant across all specifications. The results indicate that the effects of agglomeration captured by REC on the propensity of intangible-intensive firms to go public exist outside of these specific states.

C. Shocks to REC

The results so far are consistent with our hypothesis that the increase in intangible-intensive IPO frequency is driven by the increase in REC. However, it is plausible that changes in REC reflect underlying omitted regional factors that are correlated with IPO activity. To mitigate this concern, all of our previous specifications exclude local equity holdings from the REC measure. Further, these specifications include state-level macroeconomic variables, state fixed effects, and year fixed effects. In certain regressions, we use state-by-year fixed effects to absorb localized shocks that firms in the same state and year might contemporaneously experience. The significant effect of REC on intangible-intensive IPOs persists in all these instances.

Nevertheless, we implement an alternative identification strategy that uses a subset of REC originating from nonlocal sources of capital. As this capital comes from outside of the state, it is less likely to be related to underlying local economic conditions. This version of REC rules out the possibility that REC is only serving as a proxy for local economic conditions rather than its distinct impact on IPOs.

To do so, we obtain the investment allocations of three major state public pension systems (California Public Employees’ Retirement System, Florida State Retirement System, and South Carolina Retirement System). Our key assumption is that out-of-state pension systems do not choose a particular manager based on the specific economic conditions of that manager’s state. As pension flows are uninformed and not correlated with future return performance (Goyal and Wahal (Reference Goyal and Wahal2008)), they are unlikely to anticipate future regional economic growth and subsequent IPO activity. In support of this assumption, we verify in Table A.6 in the Supplementary Material that changes in pension fund allocations are not significantly related to past, present, or future GDP growth.

Additionally, these out-of-state allocations serve as a shock to the amount of deployable capital that local managers can invest in IPOs. As there is a significant turnover in these allocations, they are more likely to capture new flows in REC rather than static investment holdings. Indeed, we find that pension funds choose four new managers per year on average and investment allocations persist for about 5 years.

The geographic coverage of this measure of capital is significant, despite stemming from only three pension funds. Presented in Figure 2, these pensions have allocations in 23 different states. Within these states, the amount of out-of-state pension capital is also economically meaningful. In 13 of these states, the pension amounts account for 1%–6% of the state’s total REC.

FIGURE 2 Pension Allocation by State

Figure 2 shows the average allocation of out-of-state pension funds for each state. The sample period for the allocations is 1987–2017. Any in-state funds (California for CALPERS, Florida for FRS, and South Carolina for SCRS) are excluded. States in gray do not receive any pension flows from CALPERS, FRS, or SCRS.

In Table 6, we estimate the same specifications as in Table 4, but use the external pension-based REC measure instead of the original REC measure. Across all these specifications, we find 0.8% to 1.5% more intangible-intensive IPOs relative to nonintangible-intensive IPOs for a 10% increase in pension-based REC.Footnote ²⁹ All of these effects are statistically significant. Overall, these findings imply that the REC effect we document is not attributable to omitted local economic factors.

TABLE 6 Incidences of IPOs, External Pensions

A. REC and Acquisitions

Our study primarily focuses on the decision to go public, but private firms have other pathways to raise capital or exit, such as selling the business to another firm. We expand our framework to include acquisitions to assess whether REC has a unique role in helping firms go public. As REC does not play a direct role in merger and acquisition (M&A) activity, we should expect REC to have different effects for M&As versus IPOs.

In this expanded framework, we include the number of intangible-intensive or nonintangible-intensive acquisitions at the state-year level. We estimate the following augmented version of equation (1), where observations are indexed by intangible intensiveness ( $ i $ ), acquisition status ( $ a $ ), state ( $ s $ ), and year ( $ t $ ):

(2) $$ {\displaystyle \begin{array}{l}\mathrm{No}.\hskip.35em \mathrm{IPOs}\ {\mathrm{or}\ \mathrm{Acquisitions}}_{iast}={\beta}_1\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\times \mathrm{ln}{(\mathrm{REC})}_{st-1}\\ {}\hskip19.12em \times \unicode{x1D7D9}{(\mathrm{Acquisition})}_{ast}+{\beta}_2\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip19.12em \times \mathrm{ln}{(\mathrm{REC})}_{st-1}+{\beta}_3\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip19.12em +\hskip.35em {\beta}_4\unicode{x1D7D9}{(\mathrm{Acquisition})}_{ast}\times \mathrm{ln}{(\mathrm{REC})}_{st-1}\\ {}\hskip19.12em +\hskip.35em {\beta}_5\unicode{x1D7D9}{(\mathrm{Acquisition})}_{ast}+{\beta}_6\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip19.12em \times \unicode{x1D7D9}{(\mathrm{Acquisition})}_{ast}+{\beta}_7\mathrm{ln}{(\mathrm{REC})}_{st-1}\\ {}\hskip19.12em +\hskip.35em {\beta}_8{\mathrm{State}\ \mathrm{Controls}}_{st-1}+{\beta}_9\unicode{x1D7D9}{(\mathrm{Intangible})}_{ist}\\ {}\hskip19.12em \times {\mathrm{State}\ \mathrm{Controls}}_{st-1}+{\beta}_{10}\unicode{x1D7D9}{(\mathrm{Acquisition})}_{ast}\\ {}\hskip19.12em \times {\mathrm{State}\ \mathrm{Controls}}_{st-1}+{\beta}_{11}\mathrm{ln}{(\mathrm{No}.\ \mathrm{of}\ \mathrm{Firms})}_{ist}\\ {}\hskip19.12em +\hskip.35em {\delta}_s+{\gamma}_t+{\varepsilon}_{iast}.\end{array}} $$

For each state and year, there are four observations: the number of intangible-intensive firms that IPO ( $ \unicode{x1D7D9} $ (Intangible) $ {}_{ist}=1 $ and $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=0 $ ), the number of intangible-intensive firms that are acquired ( $ \unicode{x1D7D9} $ (Intangible) $ {}_{ist}=1 $ and $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=1 $ ), the number of nonintangible-intensive firms that IPO ( $ \unicode{x1D7D9} $ (Intangible) $ {}_{ist}=0 $ and $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=0 $ ), and the number of nonintangible-intensive firms that are acquired ( $ \unicode{x1D7D9} $ (Intangible) $ {}_{ist}=0 $ and $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=1 $ ). The outcome variable is either the number of IPOs (if $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=0 $ ) or the number of acquisitions (if $ \unicode{x1D7D9} $ (Acquisition) $ {}_{ast}=1 $ ).

To identify acquisitions at the state level, we obtain M&A data from SDC. Similar to our IPO sample, this data set includes the location of target firms and the technology services variable that we use for our intangible-intensive measure. We restrict the sample to deals involving private target firms acquired by public or private acquirers.Footnote ³⁰ On average, there are 15.7 intangible-intensive acquisitions in a given state and year and 40.4 nonintangible-intensive acquisitions.

Table 7 shows coefficient estimates from equation (2). Column 1 provides a baseline specification for acquisitions and IPOs, accounting for state economic factors and differences in the underlying populations of intangible-intensive and nonintangible-intensive firms. Column 2 examines the role of REC on these different outcomes. First, we find that REC retains a positive and economically significant effect on intangible-intensive IPOs, consistent with our main finding in Table 4. We also observe a negative effect of REC on intangible-intensive acquisitions relative to intangible-intensive IPOs. For intangible-intensive firms, the negative estimate for the triple interaction indicates that REC increases the incidence of going public relative to being acquired. The results are similar using state-year fixed effects in column 3. Overall, even when expanding the set of possible transition paths for private standalone firms, we still find that REC has the greatest impact on intangible-intensive IPOs.

TABLE 7 Incidences of IPOs and Acquisitions

B. Local Retail Equity Capital

Beyond REC, we have examined the effect of VC and, to an extent, bank capital (through state deposits) on the firm’s IPO decision. Local retail equity capital could also affect a firm’s propensity to go public. As measuring retail capital is difficult, we use the total dividend income reported by individuals in each state as a proxy for this source of funding. We assume that states with higher reported dividend income, after controlling for aggregate personal income, have more retail equity ownership and a greater likelihood of influencing local firms’ IPO decisions. These data have been used by Lin (Reference Lin2019) to measure the stock market participation of local residents.

In Table A.7 in the Supplementary Material, we repeat our main specifications from Table 4 with the inclusion of dividend income and its interaction with $ \unicode{x1D7D9} $ (Intangible). We find that dividend income is positively associated with intangible-intensive IPOs and is statistically significant at the 10% level in one specification. The effect of REC on intangible-intensive IPOs, however, remains statistically significant and of the same economic magnitude as our main results, even after controlling for local retail equity capital.

C. Geographic Dispersion of Business Operations

If REC facilitates the IPOs of intangible-intensive firms due to its proximity to the firms’ headquarters, we might expect weaker results for firms with more geographically dispersed operations. To test this implication, we use the approach in García and Norli (Reference García and Norli2012) and Bernile, Kumar, and Sulaeman (Reference Bernile, Kumar and Sulaeman2015) to quantify the geographic diversification of firm operations through their 10-Ks.

Specifically, we denote firms with 5% or less of their 10-K language devoted to their headquarters’ state as Geographically Diversified based on the sample of operation locations in Bernile et al. (Reference Bernile, Kumar and Sulaeman2015). We chose 5% because it separates firms that are the most diversified outside of their headquarters’ state. For these firms, the proximity of capital to the headquarters may matter less for going public. Similar to the acquisitions analysis in Section IV.A, we create four state-year level groups, based on whether firms are geographically diversified and whether they are intangible-intensive. Using this data, we estimate a similar specification to equation (2), where observations are indexed by intangible intensiveness, geographic diversification, state, and year. This sample is more limited, covering 2,294 IPOs from 1993 to 2007 (compared to 7,456 IPOs from 1982 to 2021 in our main sample). Nonetheless, we find that the effect of REC on intangible-intensive IPOs is weaker for firms with more geographically diversified economic activity, consistent with spatial proximity being a less important factor in the IPO decisions of these firms. The results are presented in Table A.8 in the Supplementary Material.

D. Alternative Geographic Boundaries

Throughout this article, we use state boundaries to calculate REC, which allows the full set of institutional investors and IPOs to be included in the analysis. As a robustness test, we reconstruct the REC measure and private firm counts at the more granular MSA level. 5,396 of the IPOs occur in these MSAs. Estimating similar specifications at the MSA level to those in Table 4, we continue to find support for our main hypothesis that increased REC increases the propensity of intangible-intensive firms to go public. The results are provided in Table A.9 in the Supplementary Material.

A. REC and Local Institutional Ownership

Thus far, we document that increases in REC lead to more intangible-intensive IPOs. If these additional IPOs are enabled by the spatial proximity of institutional investors serving an information or certification role, it should be evident in the ownership of these IPO firms.

To test this implication, we construct a new variable, Local IO, defined as the percentage of a firm’s shares held by institutional investors located in the same state, measured 1 year after the IPO. In contrast to previous tests, we perform this analysis at the firm level rather than the state level. We estimate the following linear regression for IPO firm $ f $ in state $ s $ and year $ t $ :

(3) $$ {\displaystyle \begin{array}{l}\mathrm{Local}\ {\mathrm{IO}}_{fst}={\beta}_1\unicode{x1D7D9}{(\mathrm{Intangible})}_{ft}\times Ln{(\mathrm{REC})}_{st-1}+{\beta}_2\unicode{x1D7D9}{(\mathrm{Intangible})}_{ft}\\ {}\hskip8.5em +\hskip.35em {\beta}_3Ln{(\mathrm{REC})}_{st-1}+{\beta}_4{\mathrm{Firm}\ \mathrm{Controls}}_{ft}\\ {}\hskip8.5em +\hskip.35em {\beta}_5{\mathrm{Industry}\ \mathrm{Returns}}_{ind,t-1}+{\delta}_s+{\gamma}_t+{\alpha}_{ind}+{\varepsilon}_{fst}.\end{array}} $$

Firm-level controls include indicators for whether the firm’s IPO is listed on the NYSE or AMEX, whether it has VC backing, the IPO underwriter’s market share, and the firm’s age and size at IPO.Footnote ³¹ In addition to year and state fixed effects, we include industry fixed effects $ \left({\alpha}_{ind}\right) $ based on the Fama–French 49 industry classification in some specifications. We also add lagged returns of the firm’s specific industry to control for time-varying changes in industry performance.

Table 8 presents the results. In column 1, we find that more REC is associated with higher levels of local ownership for both local nonintangible-intensive and intangible-intensive IPOs. For nonintangible-intensive IPOs, a 10% increase in REC is associated with 10 basis points more local institutional ownership.Footnote ³² This change translates to a 2.72% relative increase in local institutional ownership compared to the sample mean of 3.68%. For intangible-intensive IPOs, a 10% increase in REC increases local ownership by 3.80% (relative to the sample mean).Footnote ³³ The estimates remain similar when including industry fixed effects (column 2). Next, we add state fixed effects (column 3) or state-by-year fixed effects (column 4). Although the estimated magnitudes of REC on local ownership of intangible-intensive IPOs are smaller, they remain statistically significant.

TABLE 8 Local Institutional Ownership

Table 8 shows a significantly positive relation between REC and local institutional ownership over a relatively short horizon (1 year). Next, we examine this effect over a longer horizon (up to 5 years). Although local institutional investors are likely to benefit from their information advantage in the short term, it is less clear whether these benefits persist over time. As the firm’s operations geographically expand, the firm’s public information environment becomes richer, potentially eroding the relative information advantage of local institutions.Footnote ³⁴

For this test, we calculate local institutional ownership for up to 5 years post-IPO. Using the specification from column 1 of Table 8 on this longer-horizon ownership, in Figure 3, we plot the estimated coefficient for $ \unicode{x1D7D9} $ (Intangible) $ \times $ Ln(REC) for each year.Footnote ³⁵ The estimated effect of REC on local institutional ownership in intangible-intensive IPOs persists for at least 5 years post-IPO.

FIGURE 3 Effect of REC on Local Institutional Ownership over Time

Figure 3 shows the estimated coefficient for $ \unicode{x1D7D9} $ (Intangible) $ \times $ Ln(REC), measuring the effect of REC on the local institutional ownership for each year in the 5 years following the firm’s IPO. The specifications include all other control variables and fixed effects as in column 1 of Table 8. The 95% confidence intervals for each estimate are also provided.

Overall, the results suggest that this local advantage is durable. The persistence in local institutional ownership, despite likely geographic expansion, may be attributable to several factors. First, the information advantages gained during initial due diligence could be sticky and result from enduring relationships forged with management. Second, geographic expansion is relatively modest, such that their information advantage persists. Third, corporate headquarters could continue to hold strategic importance, facilitating continued access to serendipitous sources and information sharing for local investors.

B. REC and IPO Valuations of Intangible-Intensive Firms

In this section, we consider the valuation of intangible-intensive IPO firms. We expect these firms that go public due to the availability of REC to have incrementally higher valuations. To test this prediction, we estimate the following linear regression for IPO firm $ f $ in state $ s $ and year $ t $ :

(4) $$ {\mathrm{Valuation}}_{fst}={\displaystyle \begin{array}{l}{\beta}_1\unicode{x1D7D9}{(\mathrm{Intangible})}_{ft}\times Ln{(\mathrm{REC})}_{st-1}+{\beta}_2\unicode{x1D7D9}{(\mathrm{Intangible})}_{ft}\\ {}+\hskip2px {\beta}_3Ln{(\mathrm{REC})}_{st-1}+{\beta}_4\mathrm{Firm}\ {\mathrm{Controls}}_{ft}\\ {}+\hskip2px {\beta}_5{\mathrm{Industry}\ \mathrm{Returns}}_{ind,t-1}+{\delta}_s+{\gamma}_t+{\alpha}_{ind}+{\varepsilon}_{fst}.\end{array}} $$

To measure valuation, we use two ratios: Market-to-Book (Assets) and Market-to-Book (Equity). Firm controls and fixed effects are the same as in the local institutional ownership analysis in Section V.A.

The results are presented in Table 9. Using either valuation measure, we find that intangible-intensive firms that IPO in areas with higher REC are valued more than other IPO firms. The difference in valuation is statistically significant in all specifications. We also observe that the standalone effect of REC on valuation is not statistically different from zero. This nonresult is consistent with the view that the role of REC concentrates in intangible-intensive firms.

TABLE 9 Valuation

C. Investor Optimism and Post-IPO Performance

One alternative explanation for the positive relation between the frequency of intangible-intensive IPOs and REC is elevated investor sentiment. Investors may become overly optimistic about firms in certain regions, and managers may exploit this opportunity to raise capital. To test for investor optimism, in Table 10, we use the BHAR at the 6- and 12-month horizons. If local investment is unrelated to firm fundamentals and driven by investor sentiment, these stocks should eventually underperform. We do not find such evidence. At the 6-month horizon (columns 1–2) or the 12-month horizon (columns 3–4), REC is not negatively associated with BHARs for intangible-intensive firms. Overall, the evidence suggests that more frequent IPOs resulting from higher REC do not lead to systematic overvaluations.

TABLE 10 Post-IPO Performance