Declining US listings tick size

Declining US listings: What has the tick size got to do with it? Man D. Nguyen a and Talis J. Putnins a,b a b Universit...

0 downloads 39 Views 1MB Size
Declining US listings: What has the tick size got to do with it? Man D. Nguyen a and Talis J. Putnins a,b a b

University of Technology Sydney

Stockholm School of Economics in Riga

Abstract Nothing. Contrary to the US IPO Task Force’s conjecture, all of the evidence considered in this paper suggests that the reductions in the tick size in US stock markets in the late 1990s have little or nothing to do with the subsequent decline in the number of listed stocks and IPOs. We find that since the 1990s, liquidity has improved for firms in all size groups, including small stocks. Small stock valuations are not adversely affected by the tick size reductions. We find no evidence that company managers seek to undo the tick size changes through stock splits. Similarly, IPOs occur at a similar price level as previously and consequently the cross-sectional distribution of share prices (including mean and median prices) remains practically unchanged. Our findings suggest that increasing the tick size, as is currently being piloted by the US SEC, is unlikely to stimulate IPOs.

JEL classification: G10, G15, G30

Keywords: liquidity, stock market listing, IPO, delist, tick size

1

1. Introduction One major role of equity market is enabling companies to raise funds for their operations and investment. The US market is a successful example. However, the number of listed companies in the US has dropped significantly since 1996, signaling a problematic trend as fewer companies choose the stock market as a source of capital. Doidge, Karolyi, and Stulz (2017) show that the decline in the number of listed companies in the US is in part due to a lower number of IPOs and in part due to a higher number of companies delisting. The declining listings trend suggests the net benefit of being listed has dropped through time, and because the cost of listing has remained approximately the same, it is likely the benefits of being listed have fallen.1 Why the benefits of being listed have dropped is an important, unresolved question. This paper examines the role played by market microstructure, in particular the tick size (the minimum price increment in which stocks are quoted and traded) and market liquidity, as a potential driver of the decline in the number of US listed companies. The conjecture that reductions in US stock tick sizes in the late 1990s or, more generally, secondary market structure and market liquidity could be a driver of the decline in the number of listed companies has received much attention among policymakers in recent years. For example, the US IPO Task Force (2011) argues that changes in the US’s capital market structure toward a low-cost, frictionless environment has favored highly liquid, large cap stocks at the expense of their smaller cap counterparts. Particularly, lower spreads decrease market makers’ profit, making them less inclined to provide liquidity in small and infrequently traded stocks. The Report contends that the reductions in US tick sizes in the late 1990s (decimalization), which coincides with the turning point (peak) in the number of US stocks, has played an important role in changing market maker incentives and the liquidity of small stocks, with adverse effects on small stocks’ benefits of being listed. Motivated by these largely untested claims, the US Securities and Exchange Commission (SEC) in 2016 commenced a Tick Size Pilot. The Pilot program, scheduled to last two years, involves increasing the minimum tick size for selected stocks from one to five cents to investigate how the tick size affects market liquidity, volatility, and long-term growth in small firms.

1

Changes in compliance and regulatory costs cannot explain the decrease in the number of stocks listed in the US (Coates and Srinivasan, 2014; Hanley, 2017; Doidge, Karolyi, and Stulz, 2013; and Gao, Ritter, and Zhu, 2013).

2

While the US Tick Size Pilot will shed light on some of the immediate and direct effects of tick size changes, it has limited ability to draw conclusions about long-term impacts and less direct effects, such as capital raising and IPOs. This is because the Pilot involves a temporary change to tick sizes over a limited time horizon and affects only selected companies. Investors can switch to trading similar stocks, while brokers, liquidity providers, and investment bankers might not fully commit their resources to changing their activities or dealing with pilot stocks because the program is only temporary. Similarly, firms are unlikely to change how they raise capital on the basis of a temporary change and unlisted firms are unlikely to undertake an IPO because of the Pilot. We therefore take a different approach to examining the role of decimalization as a potential driver of the decline in listed companies—we revisit the tick size reductions of the late 1990s and their medium/long-run effects on listed companies, market liquidity, and IPOs. Across a wide range of different tests, we consistently find that decimalization is unlikely to have caused or even contributed to the subsequent decline in the number of listed stocks and number of IPOs. Our main results are as follow. First, we find that secondary market liquidity for US stocks, including smaller ones has improved significantly since the late 90s. Firms’ quoted spread has dropped by more than two-third, Amihud’s Illiquidity has dropped while turnover has improved noticeably after the reductions in tick size. Better liquidity reduces stocks’ liquidity premium, allowing firms to face lower costs of capital. Second, we document that small stock valuations are not harmed by changes in tick size. While there are several dimensions of liquidity (e.g. spreads, depth, and execution time), and many methods to measure these dimensions which could potentially provide contradict results and interpretations, valuations would capture the net effect of changes in liquidity on firm values. Using a difference-in-differences approach, we find that small stocks’ values, measured by Peters and Taylor’s Total Q, improved relative to their larger counterparts after the decimalization. Third, we do not find any evidence to support the conjecture that firms undo the tick size changes. If smaller tick sizes are detrimental, especially for smaller stocks, corporate managers are expected to lower their stock prices to reduce the relative tick size (tick size divided by stock price).2 In contrary to this view, the number of stock splits diminishes since the first change in

2

Angel (1997) argues that a larger relative tick size incentivizes dealers to make markets.

3

minimum tick size in 1997. IPO offer prices also do not decrease compared to pre decimalization level. Consequently, we show that cross-sectional stock prices remain practically unchanged. One concern is potential selection bias in the sample of US stocks. There are fewer firms listed on stock exchanges but these firms have become much bigger (Doidge, Karolyi, and Stulz, 2017). Many listed companies have chosen to delist or been acquired by others whilst small, private firms have become reluctant to list on exchange. This would create a sampling bias because today sample of stocks includes large liquid stocks and overlook small less liquid stocks. Fewer liquid firms which were included in the sample 10 or 20 years ago might have delisted, leaving more liquid firms (survivors) in the sample for recent years. The sample also potentially omits firms which might have listed but chose not to and seek capital elsewhere. As a result, overall market liquidity measures might be driven by large stocks and overstated. We developed a simple approach and find no evidence of selection bias, confirming the improvements of liquidity over the past decades. Our paper relates to three strands of literature. The first strand explores trends and drivers of the number of firms listed on stock markets through time. A recent study in this area is Doidge, Karolyi, and Stulz (2017) which documents a drop in the number by almost 50% from its peak in 1996. The paper argues that both decrease in the number of new lists and increase in the rate of delists contribute to the US listing gap. Many policies are put in place to assist small firms with IPO but the decline in the number of IPO persists. Other regulation burdens and compliance costs cannot explain these differences through time (Coates and Srinivasan, 2014; Hanley, 2017; Doidge, Karolyi, and Stulz, 2013; and Gao, Ritter, and Zhu, 2013). Doidge, Karolyi, and Stulz, (2013) conjectures that financial market globalization can partly explain this declining trend. However, there is no evidence that small US IPOs have migrated to other markets and small IPOS have increased in other countries (Lowry, Michaely, and Volkova, 2017). Thus, we take one step further and investigate whether decimalization is a driver of this decreasing trend in the number of US listed stocks. The second strand of literature studies the relation between secondary market liquidity and firms’ capital raising activities. Mauer and Senbet (1992) and Ellul and Pagano (2006) provide theoretical models and some evidence where greater expected secondary market illiquidity and adverse selection risk increase IPO underpricing. Stulz, Vagias, and Dijk (2013) show that market liquidity and equity issuance (IPO and SEO) are positively correlated. After controlling for other 4

firm and market factors, the paper documents that firms are more likely to utilize private than public equity issues and postpone public equity issues when market liquidity deteriorates. Other studies find that illiquidity influences expected returns on secondary market, which will aggravate firm’s cost of capital via liquidity premium (Amihud and Mendelson, 1986; Chen, Lesmond, and Wei, 2007; Bao, Pan, and Wang, 2011; Friewald, Jankowitsch, and Subrahmanyam, 2012; and Anthonisz and Putnins, 2017). Therefore, if stocks’ liquidity levels are low causing high cost of capital, it might be sensible for those companies to delist and seek capital privately. However, we show that stock liquidity has improved since decimalization, allowing companies to raise funds at cheaper rates. This would result in higher benefits and lower costs of being listed and hence increase firm valuations. We add to the area of literature which investigates the impacts of decreases in tick sizes (including decimalization) on stock liquidity. Overall, smaller tick sizes have decreased spreads, trade size and long run volatility. Quoted depth decreased after decimalization but cumulative depth did not change (Bessembinder, 2003; Bacidore, Battalio, and Jennings, 2003). However, it appears that execution speed for institutional orders has declined after decimalization, taking longer time to fill institutional orders (Chakravarty, Panchapagesan, and Wood, 2005). Stock prices’ short-term volatility increases while long-run volatility decreases. (Chakravarty, Wood, and Van Ness 2004; Ronen and Weaver, 2001). It is unclear whether decimalization has largely impacted stock positively since results on different dimensions of liquidity are mixed. Our paper provides policy implications. 2. Hypotheses Studying the impact of market liquidity on firm’s financing decision is important not only to firms but also to regulators and other market participants. Ljungqvist and Tag (2016) provide a political economy model of delisting and show that “delisting can inadvertently impose an externality on the economy by reducing citizen-investors’ exposure to corporate profits and thereby undermining popular support for business-friendly policies. By facilitating companies’ departures from the stock market, private equity firms can trigger a chain of events that may lead to long-term reductions in aggregate investment, productivity, and employment”. Firms face different costs of being listed on exchange. Direct costs consist of initial costs of going public such as IPO fee, underwriter hiring fees, due diligence and disclosure costs. In addition, the monetary and human capital cost of maintaining a listing is substantial. Public firms 5

must produce quarterly and annual reports. Insiders such as executives are required to disclose their transactions of securities in their companies, which are also, by laws, monitored by the companies. Many firms such as insurance companies and banks are regulated by many government organizations (for example the SEC, the Federal Reserve, Federal Deposit Insurance Corporation), as well as the exchange whose requirements can be duplicated, creating redundancies that increase compliance costs (Hanley, 2017). Indirect costs include costs of revealing competitive information and risk of becoming targets of M&A. If the costs outweigh the benefits, being listed might not be worthwhile for companies, causing them to delist. However, changes in compliance and regulatory costs cannot explain the decrease in the number of stocks listed in the US (Coates and Srinivasan, 2014; Hanley, 2017; Doidge, Karolyi, and Stulz, 2013; and Gao, Ritter, and Zhu, 2013). Weild and Kim (2010) and the IPO Task Force Report (2011) claim that decimalization and Reg NMS lead to a loss of liquidity and aftermarket support for new issue and small stocks. The papers contend that in today efficient market where spreads and commission are approaching zero, firms need to be large enough to attract research and investors and there is no incentive for institutions to make market and commit capital in smaller stocks. The demise of traditional market makers also exacerbates this problem as there is no guaranty for an orderly market in small stocks. Therefore, developments in market microstructure might beneficial for large stocks but harmful for small stocks, making secondary market less attractive as a source of fund. Based on the literature and the conjectures of the IPO Task Force (2011) we propose three hypotheses about the influence of decimalization on firms’ listing decisions. We postulate that changes in tick size are beneficial for large firms as opposed to smaller firms. Firms that experience lower liquidity as a result of decimalization are more likely to delist or become targets in M&A, causing a drop in the number of listed stock. Hypothesis 1: Secondary market liquidity for small firms has deteriorated due to decimalization and causes firms to delist. Next, we consider the effect of decimalization on companies’ valuation. Becoming a public company increases liquidity for firms’ equity by increasing their investor base. Better liquidity can improve company valuations via different channels. First, it improves corporate governance by enhancing price discovery which allows managers to learn from market reactions and make better decisions. Second, agency costs are reduced because more informative stock prices provide the market with signals of managerial performance. Liquid stocks allow firms to use more effective 6

stock-based remuneration and align shareholders’ and managers’ interests more closely. Finally, better price discovery and stock liquidity would diminish adverse selection risk and illiquidity premium, which then lower firms’ cost of capital. These improvements in valuation enable firms to increase their investments. Thus, low secondary market liquidity reduces most of these benefits and therefore making being listed a less attractive proposition from a firm’s perspective. If decimalization is harmful to small stocks, changes in valuations would be more severely affected for small firms compared to their large counterparts. Hypothesis 2: Decimalization decreases small firms’ valuation relative to large firms’ valuation. Angel(1997) proposes an optimal relative tick size hypothesis which contend that firms adjust their stock prices so that it maximizes their valuations. If a wide tick size is optimal, companies would undo the tick size change by lowering their stock prices so that the relative tick size (tick size divided by stock price) moves back to pre-decimalization level. Firms are able to do this through stock splits. Moreover, newly listed firms (IPO) would prefer to offer their stocks at a lower price range so that the relative tick size is at the optimal level. Consequently, we expect to see a lower average price of all stocks after decimalization. Hypothesis 3a: The number of stock splits increase post decimalization. Hypothesis 3b: IPO offer prices are lower post decimalization. Hypothesis 3c: Stock prices decrease post decimalization.

3. Data Our sample includes all stocks listed in the US. We count the number of domestic listed stocks using data obtained from the Center for Research in Securities Prices (CRSP). This excludes collective investment vehicles such as investment funds, REITs. We use daily data to calculate quarterly liquidity measures for stocks because many small stocks are occasionally traded and our accounting variables are reported every three months. Daily prices and volumes for stocks are collected from CRSP and accounting variables are obtained from COMPUSTAT database. Since our main interest is at firm level, we use PERMCO as our identity variable. In instances where there are multiple PERMNOs per PERMCO, we choose ones with higher market capitalization. We measure liquidity of all stocks using quoted spread, Amihud’s (2002) Illiquidity, and turnover. GDP deflator factor is obtained from DataStream. We winsorize the top and bottom one percent 7

of liquidity measures. IPO data is obtained from the SDC Platinum database. For liquidity metrics, our study period is from 1993 to 2016 due to the availability of spread figures in CRSP.

4. Empirical analysis 4.1 Listing counts Table 1 reports the number of listed stocks, delists, IPOs, and total amount raised from IPOs for selected years from 1975 to 2015. The number of listed stocks rises from 4,781 in 1975 to its top at 7,885 in 1997 and plummets to 3,742 in 2015. The number of IPOs and delists exhibit similar trends. Both figures peak in late 1990s and decrease dramatically over the next 20 years. Despite the fall in the number of listed stock, total market capitalization for all US stocks increases significantly by almost four folds over four decades, suggesting that the average US firm has become much bigger. We also examine firms that are listed in each exchange separately and the trends are similar. Thus, this is not an exchange specific phenomenon. [Insert table 1] 4.2 Change in US listing composition In this section, we document the change in the number of large and small listed stocks in the US. We form large, medium, and small firm groups based on their market capitalization. Starting in 1975, firms are sorted into deciles. Companies in the lowest 6 deciles are classified as small whereas firms in the top decile (decile ten) are large. Firms in decile seven, eight, and nine are medium firms. We obtain the thresholds for small and large firms and inflate them with GDP growth. As illustrated in Figure 1, the total number of US listed stocks was almost 5,000 in 1995. This number rose for the next twenty years and reached its peak in 1996 with about 8,000 stocks. During the next two decades, the number of listed firms decreased consistently to about 4,000, a half of its peak in 1996. Both an increase in the number of firms which delist and a decrease in the number of IPO contribute to this dramatic drop (Doidge et. al.,2017). Equally importantly is the changes in the number of large and small firms. By construction, the number of small firms was six times higher than the number of large firms. However, these two groups display two opposite trends over the forty-year period. The number of large stocks has increased steadily while that of small stocks has diminished by almost 70%. In 2016, the number of large stocks is more than double the number of small stocks and higher than its medium

8

counterparts. These trends demonstrate that while there are less firms being listed on stock exchanges, firms have become much larger. [Insert Figure 1] 4.3 Stock liquidity 4.3.1 Liquidity through time The IPO Task Force (2011) claims that developments in market microstructure toward a frictionless environment are harmful for small companies and causes less (small) firms choose to be listed. We investigate this claim by assessing stocks’ liquidity. Figure 2 shows the equally weighted average Amihud’s Illiquidity and quoted spreads for small, medium, and large firms, as well as firm quintiles sorted on capitalization. Overall, the metrics for all groups have decreased suggesting all stocks have become more liquid. Although illiquidity and quoted spread of small firms are higher compared to medium and large firms, they have undeniably improved since decimalization. [Insert Figure 2] Turnovers for the groups offer a slightly different picture. This metric for the whole market has improved since early 90s but there is a big gap between the groups. Benefits from market developments appear to be strongest for large firms. Turnover of medium firms exhibits a slight improvement whereas that of small firms is very volatile. It rose and almost leveled with turnover of large firms in 2005 but plummeted during periods of market turmoil such as the Dot Com bubble in 2000 and the Global Financial Crisis in 2008. We formally test these trends by regressing liquidity metrics on a time variable which is the number of quarter since January 1993. We also include interaction terms between firm groups and the time variable to examine the differences between firm groups’ liquidity changes through time. According to table 2, the coefficients of independent variables are statistically significantly negative for quoted spreads and illiquidity. Small stocks appear to benefit the most compared to their larger counterparts since the magnitude of the coefficients on the interaction terms for small stocks are higher than those on large stocks when we control for firm fixed effects. The coefficients from turnover regressions are also statistically significant. We also analyzed these liquidity metrics through time for quintiles of stocks sorted on market capitalization and the trends are similar. The outcomes indicate that market microstructure changes have enhanced both large and small stocks’ liquidity. 9

[Insert table 2] 4.3.2

Selection bias in liquidity measures As demonstrated above, the number of listed firms has dropped by half while listed firms

have become much larger over the past two decades. As a consequence, liquidity metrics in recent years show improvements perhaps not because of development in the market but due to the inclusion of large firms which are very liquid and the exclusion of delisting firms which are likely to be illiquid. This might give rise to potential sample selection issues causing liquidity measures to be overstated. We investigate this using the model discussed in Appendix A. In short, we hypothesize that if stock market developments have certainly enhanced stock liquidity, stocks that existed twenty years ago should exhibit better liquidity measures if they were to find themselves trading in today’s market environment. Table 3 reports the coefficients from regressing stocks’ quoted spreads, illiquidity, and turnover on their idiosyncratic characteristics. The estimated parameters of market capitalization and volatility are statistically significant throughout the whole period, suggesting them to be consistent strong factors of firms’ liquidity. Larger firms have lower quoted spreads, illiquidity, and higher turnover. Volatility appears not only to increase quoted spreads and illiquidity but also turnover. [Insert table 3] Other accounting variables such as dividend payout ratio, leverage ratio, and return on equity show significant relations to liquidity metrics but these relations are not consistent throughout the study period. As shown in table 3, the coefficients of ROE are strong for most years except 1999 and 2006. Similarly, there are periods, such as from 2003 to 2006, in which none of these accounting variables possesses explanatory power for illiquidity. We obtain the coefficients from above regressions to estimate liquidity for firms which were listed in 1993 as if they were trading from 1994 to 2016. If selection bias exists, a significant gap between the actual overall market liquidity and 1993-firms selection-corrected liquidity is expected. Figure 3 represents the actual and selection-corrected liquidity metrics from 1994 to 2016. Accordingly, actual and selection-corrected quoted spreads follow each other closely throughout the period. Selection-corrected quoted spreads are lower than the actual figures most of the time since 2002 except in 2008 and 2009. Amihud’s illiquidity and turnover figures exhibit similar trends. Overall, the actual and selection-corrected values are not statistically significantly 10

different from each other. This suggests that there is no evidence of selection bias in recent liquidity measures. Equity market microstructure developments including reductions in tick size have improved liquidity for stocks. [Insert Figure 3] 4.4 Valuation effect of decimalization There are several dimensions of liquidity such as costs, depth, and time to fulfill the orders. While it is possible to measure costs by several spread metrics, measuring depth and execution time is difficult due to data limitation. Investors with large orders such as mutual funds are interested in depth at not only top of the order book but also the next few levels. However, most databases only provide depth at the best bid and ask level, causing quantifying the total available liquidity inaccurate. Similarly, evaluating the duration taken to complete an order is challenging since order and trader IDs are not supplied. There are many methods of measuring stock liquidity which might yield counter results. For example, changes in market microstructure might decrease spreads and overall depth at the same time, making it hard to conclude on the overall outcome. One remedy is to investigate the valuation effects of changes in stocks’ liquidity as valuation of firms would reflect the overall impact of changes in liquidity. Secondary market liquidity influences a firms’ valuation via several channels. First, it affects firms’ cost of capital. Illiquid stocks have higher liquidity premium and thus higher required rate of return. Second, liquid stocks enhances price discovery, allowing managers to learn from market reactions and make better decisions. Last but not least, liquid stocks reduce agency costs through effective stock-based remuneration which aligns shareholders’ and managers’ interests closely. Therefore, changes in market design that result in better liquidity should improve firms’ valuation. The decrease in the number of listed stocks started since 1996, which coincides with the tick size reduction in 1997. Weild and Kim (2010) and the IPO Task Force (2011) argue that lower tick sizes cause a loss of aftermarket liquidity for small firms. If decimalization is harmful to small firms, a decrease in firm valuation for small firms relative to their larger counterparts is expected. We assess the valuation effects of reductions in tick sizes in a difference-in-differences setup. Different to a standard model in which there are a treated and a control groups, both groups in our analysis are exposed to the reduction in tick size. Our aim is measuring the difference in the impact of changes in trading tick size on valuation of companies with different relative tick11

to-price level. We sort firms into quartiles based on their relative tick size. We include the top quartile which contains firms with high relative tick sizes and the bottom quartile which contains firms with low relative tick size as our treated and control group, respectively, in our analysis. We perform the following regression: 𝑌𝑖,𝑡 = 𝛾0 + 𝛾1 𝐷ℎ𝑖𝑔ℎ + 𝛾2 𝐷𝑝𝑜𝑠𝑡 + 𝛾3 𝐷ℎ𝑖𝑔ℎ 𝐷𝑝𝑜𝑠𝑡 + 𝑒𝑖,𝑡

(1)

Where 𝑌𝑖,𝑡 is the firm 𝑖’s Peters and Taylor (2016) Total Q, which is an improved version of Tobin Q where intangible assets are included in the denominator, at time 𝑡.3 Total Q is the ratio of a firm’ market value of equity and debt on the replacement cost of its total assets. This effectively measures a firms’ market valuation; 𝐷ℎ𝑖𝑔ℎ is a dummy variable that takes a value of 1 if the firm has a high relative tick size (top quartile) and 0 if the firm has a low relative tick size (bottom quartile); 𝐷𝑝𝑜𝑠𝑡 is a dummy variable for the reduction in stocks’ tick sizes. Our pre-period is 1995 as the tick size reduction (1/6th change) started in the second quarter of 1997. We choose this period to avoid potential pre-participation by the market in valuing firms and hence biased estimates. Our post-period is 2001 when the minimum tick size is dropped to one cent. The parameter 𝛾3 is of interest as it measures the difference in the effect of decimalization on firms with high and low relative tick size. We discard firms with missing Tobin Q value in either 1995 or 2001. This leaves us with 1,464 observations each year. [Insert Table 3] Table 3 reports the results of the difference-in-differences regression (1). Panel A represents the results for groups sorted on relative tick size while Panel B reports the results for groups sorted on market capitalization. All coefficients are statistically significant at 1% level. It appears that firms with low relative tick size have higher Total Q. This is reasonable since a majority of these firms have high stock prices and large market capitalization. The negative coefficient on 𝐷𝑝𝑜𝑠𝑡 implies that decimalization decreases companies’ valuation. This is perhaps partly due to the drop in overall market prices because of the dot com bubble in 2000. Interestingly, the coefficient on the interaction term 𝐷𝑙𝑜𝑤 𝐷𝑝𝑜𝑠𝑡 suggests that the reduction in valuation of firms with high relative tick size, which are mostly small firms with low prices, is smaller than that of low relative tick size firms. The average Total Q of low relative tick size firms falls by 1.304 while that of high relative tick size firms decreases by only 0.103 (-1.304 + 1.201) after decimalization.

3

Total Q for firms are obtained from Peters and Taylor Total Q database provided by WRDS.

12

This indicates that reductions in tick size are more harmful for high price stocks. We also perform this test for large and small firms and find similar results. The reduction in valuation of small companies is lower than the fall in valuation of large companies post decimalization. 4.5 Do corporate managers undo the tick size change? Angel (1997) argues that companies practice stock splits to achieve optimal relative tick size (tick size divided by stock price) because a larger relative tick size incentivizes dealers to make markets. A larger relative tick size also makes investors inclined to place limit orders to supply liquidity. If smaller tick size after decimalization is detrimental to stocks’ liquidity and hence the benefits of being listed, corporate managers would undo the tick size change by lowering stocks prices via stock splits as this would bring back the relative tick size to pre-change level. Additionally, new IPOs would select a lower price range for their offer prices after decimalization.

4.5.1 Trends in the number of stock splits. We first examine the number of stock splits through time. Angel (1997) and Schultz (2000) argue that stock splits increase percentage quoted spreads which encourage brokers to endorse the stocks. Therefore, if decimalization reduces percentage quoted spreads causing brokers less inclined to endorse the stocks, managers would attempt to split stocks to undo this reduction. Figure 4 exhibits the loss in popularity of stock splits since 1998. 4 The number of splits fluctuated around more than 100 per quarter before 1999 but diminished thereafter. From 2008 to 2016, the number of splits level around 10 per quarter. This repudiates the hypothesis that managers undo the tick size changes by reducing stock prices via stock splits. [Insert Figure 4] 4.5.2 Trends in IPO offer prices. Figure 5 represents the trend in the average IPO offer prices from 1990 to 2016. We exclude REITs, depositories, close-ended funds, units, and penny stocks with offer prices lower than $1. The average IPO offer prices show an upward trend during the 90s but remain stable since 1998. This is consistent with a recent study by Lowry, Michaely, and Volkova (2017). Firms do not appear to lower their offer prices to reduce the relative tick size, implying decimalization might not be as detrimental as the IPO Task Force (2011) claims.

4

We remove reverse splits for the purpose of this analysis. Reverse splits make up around 7% of total number of splits in the last two decades. However, inclusion of reverse splits changes the results negligibly.

13

[Insert Figure 5] 4.5.3 Net trends in stock prices. We examine the average stock prices time series variation. Figure 6 presents the average prices of stock deciles sorted on market capitalization. In contrast to the proposition that firms lower stock prices to undo the tick size changes, the average prices in all deciles increase through time. The average prices for deciles in 2005 are statistically significantly greater than those in 1995. The results indicate that firms do not lower their prices to revert to the pre-decimalization relative tick size level. [Insert Figure 6]

4.6 Is the delisting probability related to relative tick size? If the reductions in tick size is harmful to stocks and eradicate listing benefits relative to the costs of listings, a negative relation between relative tick size and the probability of delisting is expected. We perform a logit analysis to measure this relation. Logit analysis allows us to access the probability of a dichotomous outcome using explanatory variables. In our analysis, the dependent variable takes a value of 1 if the firm delists and 0 otherwise, and the explanatory variable is firms’ relative tick size in previous year. Year fixed effects are considered. We also performed the analysis in which the dependent variable takes a value of 1 if the firm delists for reason rather than mergers (i.e. liquidation, or not meeting exchange/regulatory requirements) and the results are similar. [Insert Table 4]

Table 4 reports results from logit regressions which indicate a positive relationship. The higher the relative tick size, the higher is the probability that the firm will delist. We acknowledge that stock prices decline prior to delisting, causing an increase in relative tick size. Thus, we run similar regressions using one, two, and three lag of relative tick size as explanatory variable and attain similar outcomes. The results reject the hypothesis that small relative tick size increases firms’ probability of delisting.

4.7 Discussion

14

All results from above tests indicate that reductions in tick size are not harmful to small firms. Liquidity measured by percentage quoted spreads and Amihud’s illiquidity has improved significantly for all stocks, including small ones. Companies do not appear to undo the tick size changes by lowering their stock prices via stock splits, nor do new IPOs firms lower their offer prices. Small market capitalization firms’ valuation did not decrease as much as their larger counterparts after decimalization. This indicates the increase in the benefits of being listed for firms due to improved liquidity. Our results are consistent with recent studies on the US 2016 Tick size pilot in which minimum tick sizes for selected stocks increase from one cent to five cents. Song and Yao (2017) document negative impacts of the pilot. Accordingly, the increase in tick size increases quoted spreads, realized spreads, and price impact while deteriorates liquidity, trading volume, and price efficiency. Prices for pilot firms also dropped, suggesting a decrease in their valuation. Similarly, Hansen, Li, Lunde, and Patton (2017) find that although quoted depth at the national best bid and offer prices increase, volatility increases by around 16% and traded volume decreases for pilot stocks. Griffith and Roseman (2017) obtain similar results and contend that the increase in tick size fails to improve market quality for small stocks. These results regarding the impact of the tick size on market quality are more limited compared to our long sample because the effects might not show up in a pilot for various reasons, mainly due to the fact that the Tick size pilot is temporary. Since it is implemented for only two years, market participants might not adjust their behaviors or trading strategies. Investors might avoid these stocks by switching to other similar stocks that are not involved in the pilot. Hansen, Li, Lunde, and Patton (2017) find a substitution effect where trading in stocks within the control group is influence because investors substitute these stocks for stocks in the test groups. In addition market makers are aware that the tick size will drop after the pilot and might not want to commit their resources in providing liquidity and research reports as the pilot anticipated. 5. Conclusion In this paper, we investigate how reductions in minimum tick size influence the number of listed stocks. Using different tests, we find no evidence of the tick size being a driver of the drop in the number of US listing. Reductions in tick size improve firms’ liquidity measured by quoted spreads, Amihud’s illiquidity, and turnover. Improved secondary market liquidity would lower the

15

cost of secondary offerings, making it cheaper for firms to raise capital from equity market. This would incentivize firms to be listed. Our paper provides policy implications on recent debate on the effectiveness of the tick size pilot. Consistent with other studies on the subject, our results indicate that the pilot might not be able to achieve its intention to bring back the number of IPOs to the level in late 1990s. An increase in tick size will not help in improving secondary market liquidity and would be detrimental for small cap stocks. The paper also contributes to the literature on the drop of the number of listed stocks in the US. Our results show that the tick size has nothing to do with the decline in the US listings and future research can explore other revenues.

16

Appendix A: Selection bias Recent samples of stocks might exhibit selection bias due to a large drop in the number of listed firms. As illustrated in Figure 1, the number of small stocks has dropped significantly while the number of large stocks shows a noticeable increase. The number of small stocks was three times higher than the number of large stocks in 1993 but has become less than the number of large stocks in 2016. The number of medium firms also increased significantly and make up forty percent of the total number of firms in 2016. These changes in stock sample composition might inflate both equally and value weighted liquidity measures in recent years. We develop a simple approach to address this issue. If stock market developments have certainly enhanced stock liquidity, stocks that existed twenty years ago should exhibit better liquidity measures if they were to find themselves trading in today’s market environment. We run the following cross-sectional regression to examine how firms’ characteristics influence their stock liquidity in 2016: 𝑌𝑖 = 𝛼𝑖 + 𝛽𝐶 𝐶𝑖 + 𝑒𝑖

(2)

where 𝑌𝑖 is firm 𝑖’s average liquidity measures such as quoted spread, Amihud’s illiquidity, and turnover; 𝐶𝑖 is a vector that contains firm’s specific characteristics and 𝑒𝑗,𝑡 is the error term. Appendix B contains firm characteristic variables that influence its stocks’ liquidity. We obtain and apply the coefficients on the 1993 sample using the following: ̂𝐶 𝐶𝑗 + 𝑒𝑖 ̂𝑗 = 𝛼𝑖 + 𝛽 𝑌

(3)

̂ Where 𝑌̂ 𝑗,𝑡 is the estimated liquidity of firm 𝑗 in 2016; 𝛽𝑖𝑑𝑖𝑜 is a vector containing estimated parameters from equation (1). The estimated 2016 liquidity of the 1993 sample’s stocks are compared to the actual liquidity of the 2016 sample’s stocks. If selection bias exists, these two figures would be statistically significantly different from each other. We extend this analysis and estimate liquidity of stocks in the 1993 sample in every year between 1994 and 2016. We are interested in what would be the overall market liquidity if all firms that existed in 1993 were in these years. ̂ 𝑌̂ 𝑗,𝑡 = 𝛼𝑗 + 𝛽𝐶,𝑡 𝐶𝑗,𝑡 + 𝑒𝑗,𝑡

(4)

We obtain one set of coefficients 𝛽̂ 𝐶,𝑡 for each year from 1994 to 2016. This becomes a panel analysis where 𝑡 denotes the year in which the coefficients are estimated and 𝑗 represents firms that existed in 1993.

17

Appendix B: Variables 𝑖 and 𝑡 stand for firm and quarter, respectively. All liquidity metrics are winsorized at 1% level for each stock and for each date. Variable

Description

Data source

Spread

Average quarterly quoted spread using daily observations calculated CRSP as: 1

𝑆𝑝𝑟𝑒𝑎𝑑𝑖,𝑡 = 𝐷 ∑𝐷 𝑑=1 where 𝑚𝑑

=

𝐴𝑠𝑘𝑑 −𝐵𝑖𝑑𝑑 𝑚𝑑

𝐴𝑠𝑘𝑑 +𝐵𝑖𝑑𝑑 2

; 𝐴𝑠𝑘𝑑 and 𝐵𝑖𝑑𝑑 are at the end of trading

day 𝑑. Unit is decimal. ILLIQ

Average quarterly Amihud’s (2002) illiquidity using daily

CRSP

observations calculated as: 𝐼𝐿𝐿𝐼𝑄𝑖,𝑡 = 𝑙𝑜𝑔 [1 +

105 𝐷

∑𝐷 𝑑=1

|𝑟𝑖,𝑡 | $𝑉𝑜𝑙𝑢𝑚𝑒𝑖,𝑡

]

Where 𝑟𝑖,𝑡 and $𝑉𝑜𝑙𝑢𝑚𝑒𝑖,𝑡 are daily return and traded dollar volume, respectively, for stock 𝑖 during day 𝑑 of quarter t. Turnover

𝑇𝑢𝑟𝑛𝑜𝑣𝑒𝑟𝑖,𝑡 =

𝑣𝑜𝑙𝑢𝑚𝑒𝑖,𝑡 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 𝑠ℎ𝑎𝑟𝑒𝑠𝑖,𝑡

CRSP

where 𝑣𝑜𝑙𝑢𝑚𝑒𝑖,𝑡 is the total number of traded 𝑖 shares in quarter 𝑡 MktCap

Average market capitalization of stock calculated using daily prices

CRSP

and the number of outstanding shares. Volatility

Standard deviation of daily stock returns

MB

𝑇𝑜𝑡𝑎𝑙 𝑏𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘ℎ𝑜𝑙𝑑𝑒𝑟𝑠 ′𝑒𝑞𝑢𝑖𝑡𝑦 Compustat 𝑂𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔 𝑠ℎ𝑎𝑟𝑒𝑠𝑖,𝑡 × 𝑆𝑡𝑜𝑐𝑘 𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑒𝑛𝑑 𝑜𝑓 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑦𝑒𝑎𝑟

Leverage ROE DY

CRSP

𝑇𝑜𝑡𝑎𝑙 𝑏𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑙𝑜𝑛𝑔 𝑡𝑒𝑟𝑚 𝑑𝑒𝑏 𝑎𝑛𝑑 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑑𝑒𝑏𝑡 𝑇𝑜𝑡𝑎𝑙 𝑏𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘ℎ𝑜𝑙𝑑𝑒𝑟𝑠 ′𝑒𝑞𝑢𝑖𝑡𝑦

Compustat

𝑁𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑇𝑜𝑡𝑎𝑙 𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘ℎ𝑜𝑙𝑑𝑒𝑟𝑠 ′𝑒𝑞𝑢𝑖𝑡𝑦

Compustat

𝑇𝑜𝑡𝑎𝑙 𝑑𝑖𝑣𝑖𝑑𝑒𝑛𝑑 𝑓𝑜𝑟 𝑡ℎ𝑒 𝑦𝑒𝑎𝑟 𝑆𝑡𝑜𝑐𝑘 𝑝𝑟𝑖𝑐𝑒 𝑎𝑡 𝑒𝑛𝑑 𝑜𝑓 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑦𝑒𝑎𝑟

Compustat

18

Appendix C: Regression of quoted spread on stock’s characteristics. This table reports the results from regressing firms’ liquidity measures (quoted spreads, illiquidity, and turnover) on firms’ market capitalization (mktCap), stock return volatility (volatility), Market-to-Book ratio (MB), return on equity (ROE), dividend yield (DY), and leverage (lev) computed using COMPUSTAT annual data. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1993 to 2016 available from CRSP. ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels, respectively.

Year

Intercept

mktCap (x10,000)

Volatility

MB (x1000)

ROE

DY

leverage

(x1000)

(x1000)

𝑅2

Panel A: Quoted spread 1994

1.292***

-1.255***

108.648***

-3.584

-0.262***

-29.089

7.710*

0.579

1995

0.978***

-1.264***

119.877***

0.037

-0.361***

-0.131

2.837**

0.614

1996

1.068***

-1.074***

111.363***

-16.513***

0.014

11.218

35.033***

0.611

1997

0.175***

-0.682***

121.017***

-7.798***

0.125***

33.616*

13.338***

0.604

1998

0.322***

-0.446***

105.845***

-4.511***

-0.078*

-2.298

8.135**

0.548

1999

1.119***

-0.292***

63.682***

-9.152***

-0.043

-23.95

4.392*

0.357

2000

1.363***

-0.188***

53.532***

-10.206***

-0.165***

-17.142

-1.077

0.254

2001

1.628***

-0.197***

37.721***

-8.526***

-0.903***

-0.318

34.694***

0.166

2002

-0.101

-0.141***

64.482***

-7.391***

-0.011***

15.238*

10.074***

0.433

2003

-0.338***

-0.155***

64.677***

0.144

-0.018***

8.748

0.196

0.494

2004

0.016

-0.097***

45.120***

-0.443

-0.073***

0.323

0.037

0.442

2005

0.076**

-0.091***

29.933***

-0.35

-0.057*

6.826**

0.527

0.232

2006

-0.003

-0.077***

30.487***

-0.171

-0.026

5.022*

0.372

0.256

2007

0.001

-0.058***

24.927***

-2.882***

-0.112***

3.352*

5.417**

0.243

2008

0.083***

-0.045***

21.589***

-3.149***

-0.145***

0.683

2.469**

0.191

2009

-0.561***

-0.084***

42.489***

-2.024*

-0.084**

-6.740**

2.252

0.258

2010

0.726***

-0.171***

22.883***

0.192

-0.019***

-3.204**

0.598

0.202

2011

-0.445***

-0.048**

43.751***

-2.613**

0.039**

1.456

1.424

0.314

2012

-0.280***

-0.056***

32.850***

-1.568**

-0.107***

1.941

4.746**

0.211

2013

-0.377***

-0.030*

45.114***

-2.150**

-0.064***

1.046

1.333

0.301

2014

-0.053

-0.034***

29.702***

-2.189**

-0.032***

-0.215

2.764

0.226

2015

0.244***

-0.045***

12.354***

-0.422

-0.137***

1.542

0.122

0.119

2016

0.124***

-0.040***

19.014***

-0.338

-0.125***

0.648

-0.292

0.18

19

Table 3 continued. Year

Intercept

mktCap (x10,000)

Volatility

MB

ROE

(x1000)

DY

leverage

(x1000)

(x1000)

𝑅2

Panel B: Amihud's illiquidity 1994

0.271***

-0.202***

7.093***

-0.694*

-0.011*

-9.384*

1.826***

0.257

1995

0.308***

-0.234***

7.823***

-0.125

0.022*

-9.711*

0.392*

0.218

1996

0.271***

-0.167***

7.509***

-3.382***

0.031***

-7.797*

6.355***

0.218

1997

0.173***

-0.118***

8.075***

-1.220***

0.020***

-2.129

1.541**

0.193

1998

0.134***

-0.059***

7.999***

-0.598***

0.022***

-8.974***

0.674

0.217

1999

0.175***

-0.039***

5.837***

-1.420***

0.006

-10.141***

0.679*

0.178

2000

0.187***

-0.026***

5.318***

-1.408***

-0.001

-9.767***

0.124

0.133

2001

0.312***

-0.033***

2.230***

-1.252***

-0.027**

-3.152***

5.402***

0.041

2002

0.255***

-0.037***

5.846***

-1.629***

-0.001*

-5.000***

2.405***

0.135

2003

0.177***

-0.046***

7.732***

-0.194

0

-2.157

0.673

0.178

2004

0.162***

-0.038***

6.930***

-0.126

-0.003

-0.766

0.012

0.162

2005

0.119***

-0.033***

5.710***

-0.137

-0.004

1.226

0.215

0.069

2006

0.085***

-0.033***

6.425***

-0.096

0

1.494

0.236

0.083

2007

0.103***

-0.030***

5.128***

-1.426***

-0.019*

1.239

3.709***

0.059

2008

0.085***

-0.020***

4.962***

-1.547***

-0.037**

0.159

1.361**

0.068

2009

0.001

-0.026***

6.849***

-0.486*

-0.029***

-1.450**

0.555

0.166

2010

0.226***

-0.044***

3.479***

0.057

-0.002*

-0.671**

0.136

0.138

2011

-0.072***

-0.017***

10.090***

-0.723**

0.017***

0.349

0.348

0.233

2012

-0.024

-0.021***

8.344***

-0.563**

-0.027***

0.476

1.655***

0.134

2013

-0.062***

-0.011**

11.307***

-0.630**

-0.008

0.165

0.326

0.213

2014

-0.021*

-0.010***

8.680***

-0.770**

0.003

-0.091

1.105

0.169

2015

0.055***

-0.012***

3.921***

-0.209

-0.023*

0.395

0.055

0.09

2016

0.044***

-0.012***

4.854***

-0.168

-0.028***

0.191

-0.058

0.12

20

Table 3 continued. Year

Intercept

mktCap (x10,000)

Volatility

MB (x1000)

ROE

DY (x1000)

leverage (x1000)

𝑅2

Panel C: Turnover 1994

1.073***

0.109

0.006

-0.6

0.023*

-52.818***

-2.254*

0.01

1995

0.935***

0.274***

0.46

0.23

0.003

-65.832***

-0.523

0.009

1996

1.057***

0.281***

2.134***

7.593***

0.051**

-64.794***

-17.310***

0.017

1997

0.965***

0.007

7.514***

3.687***

0.054***

-10.479**

-4.339**

0.031

1998

0.995***

0.102***

6.797***

2.399***

0.053**

-50.533***

-3.270**

0.029

1999

0.744***

0.097***

11.610***

5.693***

0.160***

-45.021***

-1.797

0.056

2000

0.495***

0.058**

21.002***

7.795***

0.102***

-23.858**

-1.801

0.105

2001

0.141***

0.075***

27.391***

3.901***

0.428***

-9.165***

-12.903***

0.179

2002

1.045***

0.098***

3.947***

6.809***

0.004*

-22.517***

-5.520***

0.018

2003

1.147***

0.118***

2.347***

0.702

0.003

-20.123***

-2.039

0.006

2004

1.213***

0.062**

8.507***

0.882**

0.023***

-8.342**

-0.098**

0.02

2005

-0.205**

0.159***

68.449***

0.434

0.316***

-17.837**

-0.635

0.136

2006

0.464***

0.160***

47.646***

0.39

0.320***

-20.118***

0.404

0.085

2007

0.888***

0.141***

36.045***

7.022***

0.221***

-17.393***

-8.116

0.08

2008

1.412***

0.097***

21.848***

6.802**

0.448***

-10.289***

-3.546

0.031

2009

2.267***

0.114***

-1.074

1.492

0.109***

3.318

-0.879

0.01

2010

2.155***

0.033

-0.073

-0.538

-0.001

9.724***

-0.251

0.017

2011

1.594***

0.161***

13.228***

3.522

-0.019

-7.541**

-0.595

0.021

2012

1.216***

0.145***

23.626***

2.527**

0.204***

-8.829**

-6.336**

0.038

2013

1.287***

0.072***

17.326***

2.431*

0.107***

-3.986

-0.146

0.025

2014

0.950***

0.051***

35.646***

2.903

0.048**

-1.468

-0.313

0.082

2015

1.031***

0.05

41.953***

0.302

-0.218

-7.068

-0.135

0.098

2016

0.919***

0.039*

37.727***

1.361

-0.287***

-2.551

-0.059

0.146

21

References Amihud, Y., and Mendelson, H., 1986. Asset pricing and the bid-ask spread. Journal of Financial Economics 17, 223–249. Angel, J.J., 1997. Tick size, share prices, and stock splits. Journal of Finance 52, 655 – 681. Bacidore, J., Battalio, R.H., and Jennings, R.H., 2003, Order submission strategies, liquidity supply, and trading in pennies on the New York Stock Exchange. Journal of Financial Markets 6, 337–362 Barth, M.E., Landsman, W.R., and Taylor, D.J, 2017. The JOBS Act and information uncertainty in IPO firms. Accounting Review (forthcoming). Bessembinder, H., 2003. Trade execution costs and market quality after decimalization. Journal of Financial and Quantitative Analysis 38, 747–777. Bharath, S.T., and Dittmar, A.K., 2010. Why do firms use private equity to opt out of public markets? Review of Financial Studies 23, 1771–1818. Biais, B., and Foucault, T., 2014. HFT and market quality. Bankers, Markets and Investors 128, 5–19. Boehmer, E., Fong, K.Y., and Wu, J.J., 2013. Algorithmic trading and changes in firms’ equity capital. Unpublished working paper. Chakravarty, S., Wood, R. A., and Van Ness, R. A., 2004. Decimals and liquidity: A study of the NYSE. Journal of Financial Research 27, 75–94. Coates, J.C., and Srinivasan, S., 2014. SOX after ten years: A multidisciplinary review. Accounting Horizons 28, 627–671. Dambra, M., Field, L.C., and Gustafson, M.T., 2015. The JOBS Act and IPO volume: Evidence that disclosure costs affect the IPO decision. Journal of Financial Economics 116, 121– 143. Davis, R.L., Maslar, D.A., and Roseman, B.S., 2017. Secondary market trading and the cost of new debt issuance. Unpublished working paper. Decker, R.A., Haltiwanger, J., Jarmin, R.S., and Miranda, J., 2016. Where has all the skewness gone? The decline in high-growth (young) firms in the US. European Economic Review 86, 4–23. Doidge, C., Karolyi, G.A., and Stulz, R.M., 2017. The US listing gap. Journal of Financial Economics 123, 464–487. 22

Ellul, A., and Pagano, M., 2006. IPO underpricing and after-market liquidity. Review of Financial Studies 19, 381–421. Gao, X., Ritter, J.R., and Zhu, Z., 2013. Where have all the IPOs gone? Journal of Financial and Quantitative Analysis 48, 1663–1692. Griffith, T., & Roseman, B. S., 2016. Making cents of tick sizes. Unpublished working paper. Grullon, G., Larkin, Y., and Michaely, R., 2015. The disappearance of public firms and the changing nature of US industries. Unpublished working paper. Gustafson, M.T., and Iliev, P., 2017. The effects of removing barriers to equity issuance. Journal of Financial Economics 124, 580–598. Hanley, K.W., 2017. The economics of primary markets. Unpublished working paper. Haslag, P.H. and Ringgenberg, M., 2015. The causal impact of market fragmentation on liquidity. Unpublished working paper. Hendershott, T., Jones, C.M., and Menkveld, A.J., 2011. Does algorithmic trading improve liquidity? Journal of Finance 66, 1–33. Hansen, P., Li, Y., Lunde, A., and Patton, A., 2017. Mind the gap: an early empirical analysis of SEC's “Tick size pilot program”. Unpublished working paper. Jones, C.M., 2013. What do we know about high-frequency trading? Unpublished working paper. Ljungqvist, A., Persson, L., and Tåg, J., 2016. Private equity’s unintended dark side: On the economic consequences of excessive delistings. NBER working paper no.21909. Lowry, M., Michaely, R., and Volkova, E., 2017. Initial public offering: a synthesis of the literature and directions for future research. Unpublished working paper. Massa, M., and Xu, M., 2013. The value of (stock) liquidity in the M&A market. Journal of Financial and Quantitative Analysis 48, 1463–1497. Mauboussin, M.J., Callahan, D., and Majd, D., 2017. The incredible shrinking universe of stocks. Unpublished working paper. Mauer, D.C., and Senbet, L.W., 1992. The effect of the secondary market on the pricing of initial public offerings: Theory and evidence. Journal of Financial and Quantitative Analysis 27, 55–79. Securities, U. S., & Exchange Commission. (2012). Report to congress on decimalization. Washington, DC.

23

Stulz, R.M., Vagias, D., and Van Dijk, M.A., 2013. Do firms issue more equity when markets are more liquid? NBER working paper no. 19229. Schultz, P., 2000. Stock splits, tick size, and sponsorship. Journal of Finance 55, 429-450. Song, S., & Yao, C., 2017. The real costs of natural experiments. Unpulished working paper. Weild, D., and Kim, E., 2010. Market structure is causing the IPO crisis—and more. Grant Thornton Capital Markets Series, Grant Thornton International.

24

Table 1: Listing, delisting, and IPO counts, total market capitalization, and IPO proceeds for selected years. This table reports the number of listed companies, total market capitalization, number of IPOs and the amount raised, and the number of delistings in the US from 1975 to 2015. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1975 to 2016 available from CRSP. . IPOs information is from SDC Platinum database. CRSP delist codes are used to segregate the number of delisting. Delisting - Mergers include firms which delisted due to mergers and acquisition. Delisting - Others include firms that delisted due to liquidation, or not meeting exchange/regulatory requirements, or voluntary. US population is obtained from the World Bank database.

Number

listing

Total

IPO

of

per

market

Year

listings

capita

cap ($ bil)

of IPOs

1975

4,781

22.137

556.944

1980

4,896

21.546

1985

6,144

1990

Number Proceeds

Delisting -

Delisting -

Delisting -

($ bil)

Total

Mergers

Others

5

0.176

176

90

86

975.468

142

1.370

290

185

105

25.823

170.899

305

5.389

542

264

278

6,071

24.320

273.768

192

4.720

511

195

316

1995

7,268

27.294

459.376

640

31.644

539

321

218

2000

6,982

24.744

149.623

575

64.478

860

578

282

2005

4,933

16.698

129.492

179

29.155

374

231

143

2010

4,090

13.221

119.683

120

32.248

319

194

125

2015

3,839

11.963

211.498

157

26.791

239

174

65

25

Table 2: Regressions of liquidity metrics through time. This table reports the results from regressions of liquidity metrics on time and interaction terms between time and dummy variables for firm groups. Time is the number of quarter since January 1993. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1993 to 2016 available from CRSP. GDP deflators are provided by the US Bureau of Economic Analysis. t-statistics are based on double clustered standard errors and reported in parentheses. ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels, respectively.

Quoted

Quoted

spread

ILLIQ

turnover

spread

ILLIQ

turnover

(x1,000)

(x1,000)

(x1,000)

(x1,000)

(x1,000)

(x1,000)

0.007

-1.78

0.000

56.781**

591.705***

274.190

(0.01)

(-0.24)

(0.00)

(40.26)

(43.73)

(25.99)

-0.435***

-2.192***

3.056***

-0.674***

-4.762***

3.413

(-19.9)

(-9.22)

(13.16)

(-23.43)

(-19.15)

(17.03)

9.8

1.7

0.9

18

4.5

1.7

0.037

-1.442

0.304

56.794***

591.035***

276.399***

(0.06)

(-0.20)

(0.05)

(40.27)

(43.69)

(25.97)

-0.513***

-3.228***

3.911***

-0.696***

-4.271***

1.734***

(-14.47)

(-6.41)

(9.47)

(-20.71)

(-8.38)

(3.80)

-0.503***

-2.965***

2.617***

-0.661***

-4.283***

2.047***

(-16.55)

(-8.26)

(8.42)

(-22.11)

(-12.09)

(7.11)

-0.411***

-1.915***

3.043***

-0.675***

-4.839***

3.644***

(-19.07)

(-8.15)

(12.30)

(-23.40)

(-19.14)

(17.89)

𝑅2

9.9

1.8

0.9

18.1

4.5

1.8

Stock fixed effects

Yes

Yes

Yes

No

No

No

Panel A Intercept

𝑡𝑖𝑚𝑒

𝑅2

Panel B Intercept

𝐷𝑠𝑚𝑎𝑙𝑙 × 𝑡𝑖𝑚𝑒

𝐷𝑚𝑒𝑑𝑖𝑢𝑚 × 𝑡𝑖𝑚𝑒

𝐷𝑏𝑖𝑔 × 𝑡𝑖𝑚𝑒

26

Table 3: Difference-in-differences test for valuation effect around decimalization. This table reports the results from the difference-in-differences regressions. Panel A represents results for groups formed based on relative tick size whereas Panel B represents results for groups formed based on market capitalization. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1993 to 2016 available from CRSP. Dependent variable is firm 𝑖’s Total Q obtained from Peters and Taylor Total Q database. Independent variables include

𝐷ℎ𝑖𝑔ℎ a dummy variable for firms with high relative tick size (Panel A) or 𝐷𝑠𝑚𝑎𝑙𝑙 a dummy variable for

firms with low market capitalization (Panel B), and 𝐷𝑝𝑜𝑠𝑡 a dummy variable for periods after decimalization (2001). ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels, respectively.

Parameter

Estimate

t-stats

Panel A: groups formed on relative tick size Intercept

2.790***

16.09

𝐷ℎ𝑖𝑔ℎ

-1.718***

-8.7

𝐷𝑝𝑜𝑠𝑡

-1.304***

-6.69

1.201***

5.04

𝐷ℎ𝑖𝑔ℎ × 𝐷𝑝𝑜𝑠𝑡

Panel B: groups formed on market capitalization Intercept

2.503***

15.99

𝐷𝑠𝑚𝑎𝑙𝑙

-1.548***

-8.3

𝐷𝑝𝑜𝑠𝑡

-1.017***

-5.68

𝐷𝑠𝑚𝑎𝑙𝑙 × 𝐷𝑝𝑜𝑠𝑡

0.758***

3.42

27

Table 4: Relation between relative tick size and probability of delist This table reports logit regressions estimated over the period from 1993 to 2012. The dependent variable equals one if a firm delisted in that year and zero if it remains being listed. Independent variable relative tick size is computed as tick size divided by the median stock price for the quarter. Model (1) considers year fixed effects. The sample includes all firms listed on AMEX, NASDAQ, and NYSE available from CRSP. We count a delisting as such in the year in which a firm drops out and CRSP provides the delist code. auROC reports the area under Receiver Operating Characteristic (ROC) curve. ***, **, and * indicate statistical significance at 1%, 5%, and 10% levels, respectively.

Intercept Relative tick size

(1)

(2)

-1.816***

-2.326***

5.05***

5.176***

0.601

0.594

Yes

No

auROC Year fixed effects

28

Figure 1: Number of small, medium, and big firms. This figure shows trends in the number of small, medium and big domestic firms in the US from 1975 to 2016. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1975 to 2016 available from CRSP. Firms are sorted into deciles at the start of 1975. The first six deciles set the level below which firms are classified as small. The top decile contains big firms. Firms with values in between the sixth and ninth decile level are medium firms. These thresholds are updated yearly using the GDP deflators provided by the US Bureau of Economic Analysis. The number of firm per capita is displayed on the secondary axis with unit is in million. Two vertical lines indicate the time when minimum tick size is reduced to 16th and 1 cent.

29

Panel A1: Quoted spread – groups

Panel A2: Quoted spread – quintiles

30

Panel B1: Amihud’s illiquidity – groups.

Panel B2: Amihud’s illiquidity – quintiles.

31

Panel C1: Turnover – groups

Panel C2: Turnover - quintiles

Figure 2: Liquidity measures for US stocks. Thí figure shows the equally weighted average percentage quoted spreads, Amihud’s illiquidity, and turnover for small, medium, big and all domestic firms in the US, as well as for firms in quintile sorted by market capitalization and formed quarterly. Quintile 1 contains smallest firms while quintile 5 contains largest firms. Thresholds for small and big firm groups are obtained by forming deciles sorted on market capitalization in the first quarter of 1975 where decile one contains smallest firms and decile ten contains biggest firms. Threshold for small firms is the cut-off point

32

between decile six and seven whereas threshold for big firms is the cut-off point between decile nine and ten. Thresholds are inflated with US deflator downloaded from DataStream. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1975 to 2016 available from CRSP. At firm level, quoted spreads are measured relative to mid-quote in decimal; turnovers are calculated using the total traded dollar volume over the quarter divided by the product of price and the number of outstanding stock; and Amihud’s illiquidity is calculated using quarterly return and volume observations. Two vertical lines indicate the time when minimum tick size is reduced to 16th and 1 cent.

33

Panel A: Quoted spread

Panel B: Amihud’s illiquidity

34

Panel C: Turnover

Figure 3: Actual and selection-corrected liquidity measures This figure represents the actual and selection-corrected equally weighted average liquidity metrics (quoted spread, Amihud’s illiquidity, and turnover) for US listed stocks. Actual spreads are calculated for all firms for that year whereas selection-corrected quoted spreads are estimated for all listed firms on NYSE, AMEX, and NASDAQ which existed in 1993. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1993 to 2016 available from CRSP. At firm level, quoted spreads are measured relative to mid-quote in decimal; turnovers are calculated using the total traded dollar volume over the quarter divided by the product of price and the number of outstanding stock; and Amihud’s illiquidity is calculated using quarterly return and volume observations. Two vertical lines indicate the time when minimum tick size is reduced to 16th and 1 cent.

35

Figure 4: Number of stock splits This figure represents the total number of stock splits per year and the yearly average number of stock split per firm from 1975 to 2016. Only ordinary stocks splits, where CRSP’s share code equals 10 or 11 and distribution code equals 5523, are included. Reverse splits are excluded.

36

Figure 5: Average US IPO offer price This figure represents the average IPO offer price issued in the US from 1990 to 2016. Data is obtained from SDC Platinum. All REITs, ADR, Closed-end Funds, and Units are excluded. Penny stocks whose offer prices of less than $1 are also removed. Two vertical lines indicate the time when minimum tick size is reduced to 16th and 1 cent.

37

Figure 6: Average price for US stocks by market capitalization deciles. This figure represents the quarterly median prices of US stock quintiles from 1975 to 2016, sorted on market capitalization, respectively. The sample includes all firms listed on AMEX, NASDAQ, and NYSE from 1975 to 2016 available from CRSP. Stock prices are obtained from CRSP. Two vertical lines indicate the time when minimum tick size is reduced to 16th and 1 cent.

38